seminal work in the 1960s, using the percutaneous needle biopsy technique to excise small samples of human skeletal muscle, made it possible to conduct invasive studies of metabolism and determine the impact of training, diet, and other manipulations on selected biochemical, metabolic, histological, and contractile characteristics (for review, see Ref. 41). Several studies identified muscle glycogen as a major determinant of endurance exercise capacity (10, 12, 80) and an inability to continue exercise when the glycogen stores were restricted (43). Furthermore, several days of diet-exercise manipulation resulted in “supercompensated” muscle glycogen levels that, in turn, translated into significant improvements in performance of a “real-life” endurance event (54). Since then, our knowledge about muscle glycogen has expanded to include roles such as fuel sensor, regulator of intracellular signaling pathways promoting exercise training adaptation, and mediator of the osmotic characteristics of the muscle cell (38, 39, 50, 65, 81). Show
Current sport nutrition guidelines recognize that glycogen availability can be strategically manipulated to promote outcomes ranging from enhanced training adaptation to optimal performance. Indeed, the reader is directed to recent reviews regarding strategies to enhance the cellular response to an exercise stimulus through training with low carbohydrate (CHO) availability (6, 38). The aim of the current minireview, however, is to revisit scenarios in which a performance benefit is associated with matching muscle glycogen stores to the fuel requirements of training or competition. We highlight recent advances in our understanding of the optimal nutritional strategies to promote rapid and effective restoration of this important muscle substrate and describe some of the molecular signals by which glucose transport is increased in the exercised muscle after strenuous exercise. The reader is also referred to previous comprehensive reviews on these topics (13, 50, 52). Competitive endurance athletes undertake a prodigious volume of training with a substantial amount of exercise performed at intensities that are close to or faster than race pace (115). As such, preparation for and competition in endurance exercise events lasting up to 3 h are dependent on CHO-based fuels (muscle and liver glycogen, blood glucose and blood muscle, and liver lactate) to sustain high rates of muscle energy production (16, 57, 75, 106). However, the body’s reserves of CHO are not as plentiful as those of lipids or proteins, so an important goal of the athlete's daily diet is to provide the trained musculature with the substrates necessary to fuel the training program that supports optimal adaptation and recovery. Rates of postexercise glycogen synthesis have been investigated using a variety of exercise protocols and dietary regimens. Depletion of muscle glycogen provides a strong drive for its own resynthesis (116). Indeed, even in the absence of postexercise CHO intake, glycogen synthesis occurs at rates of 1–2 mmol·kg wet wt of muscle−1·h−1 through gluconeogenesis (61), or, particularly in the case of high-intensity exercise, lactate (44). However, postexercise CHO ingestion is the most important determinant of muscle (and liver) glycogen synthesis, with the highest rates of resynthesis (typically within the range of 5–10 mmol·kg wet wt−1·h−1) observed when large amounts of CHO are consumed soon after the completion of the exercise bout, and then continued throughout recovery. Several factors contribute to the enhanced synthesis rates during the first 2 h after exercise: these include activation of glycogen synthase by glycogen depletion (83) as well as exercise-induced increases in insulin sensitivity (87) and permeability of the muscle cell membrane to glucose. Nevertheless, with a mean glycogen storage rate of 5–6 mmol·kg wet wt−1·h−1, 20–24 h of recovery are normally required for normalization of muscle glycogen levels following extreme exercise depletion (30). This scenario provides a challenge to athletes who undertake multiple sessions of training in a 24-h period (e.g., swimmers, rowers, or distance runners) or competition (e.g., tournament tennis, cycling tour) with <12–15 h recovery from the first session, after which muscle glycogen content is likely to be reduced by at least 50% (102). Glucose, fructose, and galactose are the primary monosaccharides in the human diet having an energy value of 15.7 kJ/g and producing ~38 mol of ATP/mol monosaccharide. The most important monosaccharide for muscle metabolism is glucose, which is phosphorylated to glucose 6-phosphate by the enzyme hexokinase and either directed toward glycolysis or glycogen synthesis. Glycogen synthase catalyzes the incorporation of UDP-glucose through α-1–4-glycosidic linkages into the expanding glycogen polymer, with branching enzyme catalyzing formation of α-1,6-branchpoints (31). The many branching points formed by the α-1,6 bonds (approximately every 8–12 glucose units) on the glycogen molecule provide multiple sites for the addition of glucose residues during glycogen synthesis (glycogenesis), or glycogen breakdown during exercise (through glycogenolysis). Until the discovery of the protein glycogenin as the mechanism for glycogen biogenesis (101), the source of the first glycogen molecule that acted as a primer in glycogen synthesis was not known. Glycogenin is located at the core of the glycogen molecules and is characterized by autocatalytic activity that enables it to transfer glucose residues from UDP-glucose to itself (3). Before glycogenin is able to synthesize a glycogen molecule, it must form a 1:1 complex with glycogen synthase (101). Glycogenin then initiates granule formation by the addition of 7–11 glucose residues to a single tyrosine residue on the protein, which serves as a substrate for glycogen synthase. The branching enzyme and glycogen synthase then act in concert to catalyze the formation of two distinct pools of glycogen: proglycogen (PG) and macroglyocgen (MG) (59, 60). In the initial stages of glycogen formation, the PG granules grow by the addition of glucose residues forming the larger mature MG. PG and MG contain the same amount of protein but differ in the number of glycogen units and also in their rates of degradation and synthesis (1, 3, 95). It appears that PG is more sensitive to dietary CHO and is synthesized more rapidly following exercise-induced glycogen depletion, reaching a plateau after 24 h (1). The synthesis of MG is a relatively slower process, persisting for 48 h postexercise (1). The different rates of synthesis of the PG and MG granules explain, in part, the biphasic pattern of postexercise glycogen storage (52) and demonstrate that the amount of glycogenin has a direct influence on how much glycogen the muscle cell can store. Factors that influence glycogenin concentrations are largely unexplored and required investigation. In the period after glycogen-lowering exercise, glycogen synthesis is a key priority for the previously contracted muscles, and glycogen synthase activity and glucose transport are increased dramatically to meet this obligatory requirement. Indeed, an enhanced metabolic action of insulin in skeletal muscle (glucose transport, glycogen synthase activity, glycogen synthesis) is observed after glycogen-depleting exercise (85), which can persist for up to 48 h (67). It is this enhanced insulin sensitivity in skeletal muscle that, in large part, contributes to the restoration and, depending on the degree of prior glycogen depletion, even a “supercompensation” of muscle glycogen stores. While the molecular mechanisms involved in postexercise increased insulin sensitivity are not fully understood (50), the magnitude of postexercise glycogen depletion has been strongly linked to the enhanced metabolic action of insulin in this period (85). Glycogen stores in human muscle (and liver) vary and are largely determined by the training status of the individual and their habitual CHO intake (42). The resting muscle glycogen content of an untrained person consuming a mixed diet is ~80–85 mmol/kg of muscle wet wt and somewhat higher at ~120 mmol/kg wet wt for individuals undertaking regular endurance-type exercise training (12). After exhaustive glycogen-depleting exercise and with 36–48 h of a high (>8 g/kg body mass)-CHO diet, muscle glycogen content can be supercompensated (11), reaching 200 mmol/kg wet wt (97). Because 1 g of glycogen is stored in muscle with 3–5 g of water (76, 98), an athlete’s body mass typically increases 1–2% after several days of “CHO loading” (12). Whereas skeletal muscle glycogen stores provide between 300 and 700 g of glycogen (depending on the active musculature), a smaller amount of glycogen is stored in the liver, providing ~100–120 g glycogen in an average 75-kg male. Despite the relative small amounts of glycogen stored in the liver, it is the only endogenous source of glucose that directly regulates blood glucose homeostasis. Indeed, in the absence of exogenous CHO ingestion, hypoglycemia will occur when liver glycogen stores become depleted. However, when CHO is ingested during exercise, liver glycogen is typically maintained (17, 34). Few studies have determined the impact of CHO ingestion on postexercise repletion of liver glycogen (33) and brain glycogen (64), and these are beyond the scope of the present review. Recently, the role and regulation of muscle glycogen have been specified to be dependent on its subcellular localization (74). With the use of transmission electron microscopy, studies undertaken in the 1970s and 1980s revealed both fiber-type differences and a localization-dependent utilization of glycogen during exercise. A quantitative approach (62) has identified three distinct subcellular locations of glycogen: 1) intermyofibrillar glycogen in which glycogen particles are located between the myofibrils next to sarcoplasmic reticulum and mitochondria, 2) intramyofibrillar glycogen where glycogen particles are located within the myofibrils between the contractile filaments, and 3) subsarcolemmal glycogen whereby glycogen particles are located from the outermost myofibril to the surface membrane. The implications of these distinct pools of glycogen for glycogen resynthesis, muscle function, and fatigue resistance are of key interest but require further investigation before practical recommendations can be made to exploit this knowledge. The remainder of this review will focus on factors that influence muscle glycogen synthesis and strategies that can be used by athletes to enhance muscle glycogen storage, with particular relevance to scenarios in which conditions for glycogen storage are suboptimal (brief time periods between exercise sessions and/or the inability to consume adequate CHO intake). Under most conditions, dietary CHO represents the main substrate for muscle glycogen synthesis, with factors such as the quantity, timing, and type of CHO intake markedly influencing the rate of muscle glycogen storage. Synthesizing data from a range of studies that have monitored glycogen storage over 24 h following exercise-induced depletion, including two dose-response studies (19, 28), a “glycogen storage threshold” appears to occur at a daily CHO intake of ~7–10 g/kg body mass (24). Specific attention has been focused on the early (0–4 h) phase of recovery because of the slightly higher muscle glycogen synthesis rates during this time, as well as the practical issues of the multiday exercise programs undertaken by athletes. Initial guidelines recommended that athletes consume 50 g (~1 g/kg body mass) of CHO every 2 h during the early period of recovery, based on observations of similar rates of postexercise glycogen storage following CHO intakes of 0.7 and 1.4 g/kg body mass (15), or 1.5 g and 3.0 g/kg body mass (48) at such intervals. However, more recent work (33, 82, 109, 111) has reported 30–50% higher rates of glycogen synthesis (10–11 mmol·kg wet wt−1·h−1) over the first 4 h of recovery with larger CHO intakes (e.g., >1 g·kg−1·h−1), at least when CHO is consumed as repeated small feedings. Thus, when immediate postexercise refueling is a priority, current guidelines promote larger intakes of CHO in patterns of frequent consumption. The popular concept of a “window of opportunity” for postexercise refueling was created by a well-publicized study (47) that reported that immediate intake of CHO after prolonged exercise resulted in higher rates of glycogen storage (7.7 mmol·kg wet wt−1·h−1) during the first 2 h of recovery than when this same feeding was delayed after 2 h (~4.4 mmol·kg wet wt−1·h−1). Although these data show more effective glycogen synthesis during early postexercise recovery, the key finding of that study was that glycogen synthesis rates remained very low until CHO feeding was initiated. Thus, immediate provision of CHO to the muscle cell should be seen as a strategy to initiate effective refueling rather than to simply take advantage of a period of moderately enhanced glycogen synthesis. This has significance when there is only 4–8 h of recovery between exercise sessions, but a longer (>8 h) recovery time (78) may compensate for a delay in the initial feeding. Indeed, the negative feedback loop from glycogen concentrations on its own synthesis (116) may contribute to the equalization of muscle glycogen content over time. The frequency of intake of the recommended amounts of CHO (e.g., large meals vs. a series of snacks) does not affect glycogen storage in longer-term recovery, despite marked differences in blood glucose and insulin responses (21, 28). This is in apparent conflict to the observations of higher rates of muscle glycogen synthesis during the first 4–6 h of recovery when large amounts of CHO are fed at 15- to 30-min intervals (51, 109, 111). One theory to explain this “paradox” is that the maintenance of blood glucose and insulin profiles is most important during the first hours of recovery and perhaps when total CHO intake is suboptimal. However, during longer periods of recovery, or when total CHO intake is above this “threshold,” manipulations of plasma substrates and hormones within physiological ranges do not confer any additional benefit. Early studies of single nutrient feedings showed glucose and sucrose to be more effective than fructose in restoring muscle glycogen after exercise (15). This confirmed the hypothesis that glycogen synthesis is more effective with dietary CHO sources that elicit higher blood glucose and insulin responses. However, the results of the first studies of food-derived CHO were inconsistent (28, 88) because of the misuse of the structural classification of “simple” or “complex” to predict the glycemic impact of CHO-rich foods. The subsequent use of published glycemic index (GI) foods to construct postexercise diets found that glycogen storage was increased during 24 h of recovery with a CHO-rich meal based on high-GI foods compared with an identical amount of CHO eaten in the form of low-GI foods (22). However, the magnitude of increase in glycogen storage (~30%) was substantially greater than the difference in 24-h blood glucose and insulin profiles, particularly because the immediate postexercise meal produced a large glycemic and insulinemic response, independent of the GI of the CHO consumed. Other studies have confirmed greater gut glucose release and greater hepatic glucose output in response to meals immediately postexercise, favoring an increase in muscle glucose uptake and glycogen storage (91). The malabsorption of some very-low-GI CHO-rich foods was postulated to account for less efficient glycogen storage by reducing the effective amount of CHO consumed; this is supported by observations of lower postexercise glycogen storage from a poorly digestible high-amylose starch mixture compared with intake of glucose, maltodextrins, and a high amylopectin starch (53). Finally, a drink containing a special glucose polymer of high molecular weight and low osmolarity was found to enhance glycogen synthesis in the first 2 h of recovery, although this effect disappeared thereafter (82). This benefit was attributed to a faster rate of gastric emptying (58) and may point to the benefits of foods that are rapidly digested and emptied when more rapid glycogen restoration is needed. Nevertheless, in other studies, solid and liquid forms of CHO-rich foods have been found to be equally effective in providing substrate for muscle glycogen synthesis over 2–24 h (55, 84). Indeed, direct comparison with intravenous administration of matched concentrations of glucose in one investigation showed that gastric emptying of foods/drinks was not the rate-limiting process for glycogen synthesis. A separate study, which found that intravenous delivery of supraphysiological concentrations of glucose and insulin can increase rates of postexercise glycogen synthesis over 8 h to levels achieved by glycogen supercompensation protocols (37), is largely of theoretical interest only since its use contravenes antidoping rules in sport. Although dietary CHO intake has the most robust effect on muscle glycogen synthesis, rates of glycogen storage may be manipulated by other nutrients or nutrition-related factors. Outcomes of this knowledge can be used to increase glycogen storage by employing strategies to increase muscle glycogen synthesis rates when conditions are suboptimal (e.g., when total CHO intake is below targets set for maximal synthesis rates or when the refueling period is limited) or by avoiding factors that can interfere with optimal muscle glycogen synthesis. There is increasing awareness that suboptimal intake of energy in relation to exercise energy expenditure (termed relative energy deficiency in sport) results in an impairment of energy-requiring activities involved in body maintenance and health such as protein synthesis, bone turnover, or hormone pulsatility (69). It is intuitive that glycogen storage could be decreased in the face of inadequate energy intake, either by a downregulation of the energetics of glycogen synthesis or the reduced availability of glucose for storage because of demands for immediate oxidation. Indeed, there is evidence that the relationship between dietary CHO and glycogen storage is underpinned by total energy intake. For example, glycogen supercompensation protocols were reported to be less effective in female than male athletes (103), but this finding was later reinterpreted as an outcome of the relatively lower energy intake in the female cohort (104). In the latter study, female subjects showed a substantial enhancement of muscle glycogen storage associated with increased dietary CHO intake only after total energy intake was also increased (104). It should be noted that these studies involved a 4-day glycogen-loading protocol and did not collect data that would explain the mechanism of energy-related glycogen storage changes. Therefore, we are left to speculate whether this is an acute issue related to alternate fates for exogenous CHO when energy intake is suboptimal and/or a more chronic suppression of glycogen synthesis in the face of low energy availability. The coingestion of other macronutrients, either present in CHO-rich foods or consumed at the same meal, may directly influence muscle glycogen restoration independent of their effect on energy intake. Factors that may directly or indirectly affect glycogen storage include the provision of gluconeogenic substrates, as well as effects on digestion, insulin secretion, or the satiety of meals. Protein has received most attention, since an insulinotropic amino acid and/or protein mixture can augment postprandial insulin release and stimulate both glucose uptake and glycogen synthase activity in skeletal muscle tissue (26, 113), thus further accelerating muscle glycogen synthesis. Indeed there is evidence that this occurs when amino acids and/or protein are coingested with CHO below the threshold for glycogen storage (e.g., 0.5–0.8 g CHO·kg−1·h−1) (9, 45, 46, 111, 112, 117). However, as discussed by Betts and Williams (13), when CHO intake is adequate (e.g., >1 g·kg−1·h−1), the coingestion of protein has no further effect on glycogen synthesis (8, 51, 109). Protein intakes of around 0.3–0.4 g/kg appear to maximize this effect (13); this is also considered the optimal amount to promote muscle protein synthesis goals (68). The effects of coingesting fat with CHO-rich meals on postexercise glycogen storage have not been systematically investigated. In the only available study involving endurance sport, the addition of fat and protein (0.4 g/kg and 0.3 g/kg body mass per meal, respectively) to a diet containing adequate CHO to achieve maximal glycogen storage over 24 h of refueling failed to increase rates of glycogen synthesis despite markedly different responses in blood glucose and free fatty acid concentrations (19). The consumption of large amounts of alcohol is of interest since this practice often occurs in the postcompetition period, particularly in team sports. Separate studies of 8 and 24 h recovery from glycogen-depleting exercise in well-trained cyclists who consumed ~120 g alcohol (equal to 12 standard drinks) have been undertaken (20). Muscle glycogen storage was reduced during both recovery periods when alcohol displaced an energy-matched amount of CHO from a standard recovery diet. Evidence for a direct effect of elevated blood alcohol concentrations on muscle glycogen synthesis was unclear, but it appeared that, if an immediate impairment of glycogen synthesis existed, it might be compensated by adequate CHO intake and longer recovery time (20). A range of other dietary substances has been studied in relation to their potential to accelerate the rates of muscle glycogen storage or increase glycogen storage from a given amount of CHO, through mechanisms including increased muscle glucose uptake and insulin sensitivity as well as an enhancement of cellular signaling events. With regard to the latter issue, short-term supplementation with creatine monohydrate to increase muscle total creatine content has been shown to upregulate the mRNA content of select genes and proteins involved in a range of cellular activities, including glycogen synthesis, with the suggested mechanism being a change in cellular osmolarity (93). Table 1 summarizes studies of glycogen storage in relation to exercise which prior or simultaneous creatine supplementation has been undertaken and includes investigations in which an increase in glycogen storage has been observed in muscle that has been creatine loaded (32, 71, 77, 90, 100). Although it is not a universal finding, Sewell and colleagues (94) postulated that the glycogen-depleting or “muscle-sensitising” effect of exercise is needed to achieve the stimulatory effect of creatine loading on postexercise glycogen loading. Recently, Roberts et al. (88) reported a greater increase in postexercise muscle glycogen storage following creatine (20 g/day) supplementation in addition to a high-CHO diet. The greater postexercise increase in muscle glycogen became evident as early as 24 h after exercise and was maintained following 6 days of postexercise recovery on a CHO-rich diet. Although the mechanism(s) underlying this observation remains to be elucidated, it seems evident that creatine supplementation can further augment muscle glycogen storage. However, it remains to be established whether this effect occurs in highly trained athletes. Furthermore, the practical implications of any benefits of creatine use to refueling in endurance athletes should be weighed against the 1–2% gain in body mass that is associated with creatine loading.
Here it should also be noted that changes in muscle water content secondary to the whole body fluid changes experienced by athletes (i.e., hyperhydration and, more commonly, dehydration) could also alter glycogen synthesis due to changes in cell osmolarity and cell volume. This has not been systematically addressed, although an early study investigated the effect of dehydration on glycogen synthesis, based on the hypothesis that the binding of water to glycogen might make cellular hydration a permissive factor in muscle glycogen storage (72). This study found that dehydration equivalent to loss of ~5% body mass or 8% body water did not interfere with glycogen storage during 15 h following cycling exercise, although muscle water content was lower than in the trial involving euhydrated recovery. Further investigation is warranted (72). Other dietary constituents with purported effects on insulin sensitivity and glucose tolerance have been investigated in relation to muscle glycogen storage in various trained and untrained human populations. Studies have shown varying effects of caffeine use on muscle glycogen storage in trained individuals. In one investigation, intake of caffeine (8 mg/kg) with CHO (1 g·kg−1·h−1) resulted in substantially higher rates of muscle glycogen storage over 4 h of recovery (79). However, another study (7) found no difference in muscle glycogen synthesis when an hourly caffeine intake of 1.7 mg·kg−1·h−1 was added to large CHO feedings (1.2 g·kg−1·h−1) for a postexercise recovery period of 6 h. There is no apparent explanation for the discrepancy in these findings, and the practicality of using caffeine as a postexercise refueling aid must also be questioned in view of its interruption to sleep patterns. Isolated studies (Table 1) have reported enhancement of muscle glycogen storage following the use of the insulin mimetic fenugreek [containing the unique amino acid 4-hydroxyleucine, conjugated linoleic acid and hydoxycitric acid (found in Garcinia Cambogia fruit)]. However, these findings have not been replicated. For example, although muscle glycogen synthesis during 4 h of recovery was found to be enhanced when an extract isolated from fenugreek was added to a high dose of dextrose (92), a subsequent investigation from the same group failed to find any refueling advantages after 4 or 15 h of postexercise recovery when this product was consumed in combination with CHO (99). Therefore, it would be premature to consider these ingredients as an aid to accelerate muscle glycogen recovery for competitive athletes. The effects of muscle damage from the prior exercise bout need to be considered in the context of refueling. In particular, rates of glycogen synthesis are impaired after muscle-damaging eccentric contractions and/or impact injuries because of reductions in GLUT4 translocation (5) and reduced glucose uptake (4). Early laboratory-based work from Costill and colleagues reported that isolated eccentric exercise (29) or exhaustive running (14) was associated with reduced rates of muscle glycogen restoration during 24 and 72 h of postexercise recovery, with a time course suggesting that this phenomenon did not occur in the early phase (0–6 h) of recovery but was associated with later recovery (114). Although these findings are generally attributed to damage to muscle fibers and local inflammation, glycogen synthesis in damaged muscles might be partially overcome by increased amounts of CHO intake during the first 24 h after exercise (29). Of course, few studies have followed the time course of muscle glycogen recovery after real-life sporting activities. Several investigations of recovery from competitive soccer have reported a delay in glycogen restoration following football matches (36, 49, 56) such that it remained below resting levels after 24 h of recovery in both type 1 and type 2 fibers and after as much as 48 h of recovery in type 2 fibers despite relative high CHO intakes (36). Although these findings are generally attributed to the eccentric component of the movement patterns in soccer (sudden changes in direction and speed) and direct contact between players, an intervention within one study also found rates of glycogen storage below rates normally associated with recovery from cycling exercise when simulated soccer activities of different duration were undertaken with the removal of the body contact and a reduction in eccentric movements (36). Therefore, further observations of muscle glycogen recovery following competitive sports events is warranted, including the investigation of mechanisms that could explain attenuated muscle synthesis rates. Because athletes frequently undertake specialized activities after competition or key training sessions to promote various aspects of recovery, it is of interest to consider how such practices might interact with glycogen storage goals. For example, therapies that alter local muscle temperature to alleviate symptoms of exercise-induced muscle damage appear to have some effect on factors that are important in muscle glycogen synthesis although the overall effect is unclear. In one study, intermittent application of ice reduced net glycogen storage over 4 h of recovery compared with a control leg (108), whereas, in a companion study by the same laboratory, the application of heat was associated with greater refueling (100). Alterations in blood flow to the muscle secondary to temperature changes were presumed to play a role in these findings, although a reduction in muscle enzyme activities was also suspected to be a factor in explaining the outcomes of ice therapy. However, another study of cold-water immersion following exercise failed to find evidence of impaired glycogen storage during the recovery period (35). Therefore, the benefits of postexercise application of cold or heat on muscle glycogen repletion following exercise remains to be addressed in future research. Strategies to achieve glycogen supercompensation have slowly evolved since the first description of this phenomenon in the pioneering studies of Bergström and coworkers (2, 10–12, 43). These researchers (using themselves as subjects) showed that several days of a low-CHO diet followed by a similar period of high-CHO intake resulted in a localized doubling of muscle glycogen concentrations in muscle that had been previously depleted of glycogen through exercise. From this finding emanated the “classical” 7-day model of CHO loading, involving a 3- to 4-day “depletion” phase of hard training and low-CHO intake, finishing with a 3- to 4-day “loading” phase of high-CHO eating and exercise taper. A subsequent field study (54) and documented implementation by successful athletes illustrated its benefits to performance of distance running and cemented CHO loading in the practice and language of sports nutrition for endurance sports (18). Surprisingly, there have been few refinements of this potentially valuable technique, despite the fact that it was derived from observations on active but essentially untrained individuals. These increments in knowledge are shown in Fig. 1.  Fig. 1.Evolution of knowledge regarding protocols for carbohydrate (CHO) loading, as shown by diet and training manipulations in the 7 days before an endurance event. The “classical” loading protocol for glycogen supercompensation was developed by Bergström et al. (10) in untrained active individuals and confirmed in well-trained individuals by Sherman and colleagues (97). A “modified” protocol of high-CHO intake and exercise taper, deleting the depletion phase, was found to be similarly successful in athletes in the latter study (96). More recent work suggests that the supercompensation occurs in 24–48 h of taper and high-CHO intake in well-trained individuals (25). BM, body mass. A decade later, Sherman and colleagues showed that well-trained runners were able to supercompensate muscle glycogen stores with 3 days of taper and a high-CHO intake, regardless of whether this was preceded by a depletion phase or a more typical diet and training preparation (97). This “modified” and more practical CHO loading protocol avoids the fatigue and complexity of extreme diet and training requirements associated with the previous depletion phase. A more recent update on the time course of glycogen storage found that it increased significantly from ~90 mmol to ~180 mmol/kg wet wt with 24 h of rest and high-CHO intake and thereafter remained stable despite another 2 days of the same conditions (25). Although the authors concluded that this was an “improved 1-day CHO loading protocol” (25), the true loading phase from the last training session was ~36 h. In essence, the study provides a midpoint to the glycogen storage observations of Sherman and colleagues (97) and suggests that supercompensation is probably achieved within 36–48 h of the last exercise session, at least when the athlete rests and consumes adequate CHO intake. Of course, it is not always desirable for athletes to achieve total inactivity in the days before competition, since even in a taper some stimulus is required to maintain previously acquired training adaptations (70). An athlete’s ability to repeat glycogen supercompensation protocols has also been examined. Well-trained cyclists who undertook two consecutive periods of exercise depletion, followed by 48 h of high-CHO intake (12 g·kg−1·day−1) and rest, were found to elevate their glycogen stores above resting levels on the first occasion but not the next (66). Further studies are needed to confirm this finding and determine why glycogen storage is attenuated with repeated CHO loading. Current sports nutrition guidelines no longer promote a universal message of “high-CHO intakes at all time” or the need to maximize muscle glycogen storage. Indeed CHO requirements may be low on days or for athletes where a light/moderate training load has only a modest requirement for glycogen utilization or replacement (23). Intakes may be similarly low when there is a deliberate decision to undertake exercise with low glycogen stores to induce a greater skeletal muscle adaptive response (6), and there may even be benefits from deliberately withholding CHO after a high-quality training session to minimize glycogen restoration and extend the period during which adaptive responses are elevated (63). Nevertheless, there are numerous real-life scenarios in which athletes want to optimize muscle glycogen storage, either by accelerating the rates of glycogen synthesis, by promoting greater storage from a given amount of dietary CHO, or by increasing the total muscle glycogen pool. These include supercompensating muscle glycogen stores before an endurance/ultraendurance event (e.g., preparation for a marathon), normalizing muscle glycogen for shorter games/events within the weekly training microcycle (e.g., weekly or biweekly soccer game), rapidly restoring muscle glycogen between two events or key training sessions held <8 h apart (two matches within a tennis tournament or a swimmer’s twice daily workouts), and maximizing muscle glycogen storage from a diet in which energy intake is restricted (an athlete on a weight loss program, restrained eater, or an athlete in a weight-making sport). Current sports nutrition guidelines for muscle glycogen storage, summarized in Table 2, provide recommendations for both short-term (e.g., 0–6 h after glycogen-depleting exercise) and longer-term (12–48 h) refueling (23, 105). Although these strategies provide useful practices for many athletes, they are biased toward conditions in which the athlete is able to consume large/optimal amounts of CHO. A range of questions that can extend our current knowledge on muscle glycogen synthesis in more practical ways is provided in Table 3.
No conflicts of interest, financial or otherwise, are declared by the authors. L.M.B., L.J.v.L., and J.A.H. approved final version of manuscript. REFERENCES
Page 2sustained physical exercise leads inexorably to a reduced capacity to produce voluntary force. Although multiple processes contribute to this “muscle fatigue,” it is ultimately manifest as impaired muscle function and/or a reduction in the capacity of the central nervous system to activate muscles. The term “peripheral fatigue” is typically used to describe force reductions due to processes distal to the neuromuscular junction, whereas those due to processes within motoneurons and the central nervous system are commonly known as “central fatigue.” The physiology of fatigue has been studied for well over a century (see Ref. 28 for a comprehensive historical review), and recent reviews have summarized current understanding of various aspects of fatigue (2, 13, 23, 62, 84, 92, 93). As part of this Highlighted Topic series, we consider the factors that determine the recovery of voluntary force-generating capacity after various types of exercise. In attempting to document the mechanisms of fatigue and recovery, an important consideration is that sustained exercise affects physiological processes throughout the neuromuscular system. Critically, alterations in these underlying processes may either contribute to, or compensate for, fatigue (see sections below for examples and details). Although some progress has been made in documenting interrelationships between exercise characteristics, physiological responses, and impaired force-generating capacity, much remains to be learned. Given this, and the constraints of the review format, we consider primarily recovery of neuromuscular performance in terms of voluntary or artificially evoked forces measured in intact humans. We attempt to link these functional measures of fatigue with the likely underlying processes where possible and highlight areas in which a lack of available evidence prevents this. As a consequence of our focus on work that we believe provides the clearest inferences regarding the mechanisms of recovery, a number of interesting and important issues are omitted. For example, we do not consider the effects of low-intensity exercise, nutrition, or other interventions on the recovery process. Some of these issues are dealt with in other papers in the Highlighted Topic series. Because of the task-dependent nature of fatigue (23), the review is structured according to sections that each consider the classes of exercise in which recovery has been documented. We initially consider maximal voluntary contractions at a single joint, because central fatigue is most easily studied in this type of task. In particular, we extrapolate recent insights into the mechanisms of central fatigue in these tasks to the postexercise recovery period. We then consider what general implications can be drawn about recovery from apparent differences in fatigue and recovery between maximal and sustained submaximal contractions at a single joint. Finally, we consider recovery from everyday exercise, such as running and cycling, which involve large muscle masses, and consequently challenge systemic homeostasis. A reduction in the maximal force that a person can produce during a isometric maximal voluntary contraction (MVC) provides the most straightforward demonstration of fatigue. Accordingly, tasks involving a sustained MVC provide a convenient model to study fatigue and recovery, because there is a continuous measure of fatigue during the protocol (i.e., instantaneous MVC force) and because recovery can easily be tracked with the same apparatus used to induce fatigue (i.e., without the requirement to reposition the subject or switch between tasks). Isometric conditions are also convenient for measurement of forces evoked artificially by electrical or magnetic stimulation of motor nerves, descending tracts, or the motor cortex. Evoked forces at rest can provide information about fatigue and recovery of the muscle fibers, whereas force responses to stimulation that is “superimposed” upon voluntary contractions can reveal the extent to which voluntary neural drive is sufficient to generate the maximal force of which the muscles are capable. It is important to acknowledge that, although measures obtained during MVC provide valid and easily interpreted information about neuromuscular function, such measures may not be ideally sensitive to some physiological changes that are important for exercise performance. Examples that demonstrate this point include observations of exacerbated force declines during low-frequency motor unit firing (16, 22, 83, 95, 102) and observations that some interventions such as hyperthermia (98) or prior locomotor exercise (83) have much greater effects on sustained than brief MVC performance. Despite these limitations, in many cases measurements obtained during MVC provide the best available evidence regarding muscle force-generating capacity and the capacity of the central nervous system to drive muscles, and the current review will focus extensively on work that exploits these measures. Maximal voluntary force declines rapidly and progressively during a sustained MVC, typically falling to below 50% of baseline within 1–2 min. There is also a rapid but partial recovery of voluntary force over the first few minutes after cessation of this type of exercise, with the largest component occurring within 15–30 s. This suggests that reperfusion of the exercising muscles is a key factor in initial recovery, a conclusion supported by the observation that recovery is delayed if the muscles are held ischemic. Further recovery of MVC force is much slower and may reach only ~80% by 4–5 min postexercise (see Fig. 1A and Refs. 29, 50, 98). Fig. 1.Percentage change in MVC (A, D), twitch amplitude (B, E), and voluntary activation (C, F) measured at various times after the cessation of fatiguing exercise. A–C show data from isometric contractions; D–F show data from locomotor exercise. Values are taken from text or estimated from figures across a number of papers. Some papers contribute multiple points and others only one. No distinction was made between voluntary activation measured with motor nerve or motor cortex stimulation. Note that the time postexercise in minutes is on a logarithmic scale for the locomotor exercise only. Papers contributing to the graphs for isometric contractions of 2–3 min (open circles) are (29, 32, 48, 51, 98). For 2–3 min contractions of hand muscles (green triangles), contributors are (50, 76). For contractions of >20 min (solid circles), contributing papers are (88, 89, 102, 106). Apart from the hand muscles, data are from elbow flexors and knee extensors. Papers contributing to the locomotor exercise graphs for exercise of <12 min (open circles) are (33, 35, 86, 96, 97). Contributing papers for exercise of 40–90 min (gray triangles) are (16, 74, 75, 83). For exercise of >5 h (solid circles), papers are (58, 65, 72, 94, 95). Exercise of <12 min represents only knee extensors, whereas both longer durations also contain some plantarflexor data. Also of note, the exercise of >5 h was running, 40–90 min included cycling, running, and a soccer game, and exercise of <12 min included cycling and one study of running. The size of superimposed twitches evoked by stimulation of motor nerves or the motor cortex also increases within 15–30 s of sustained MVC, indicating that part of the voluntary force reduction is due to suboptimal output from the motor cortex (29, 40, 41, 48, 50, 51, 91, 98). This failure of voluntary activation has been estimated to account for ~25% (100) of the total force reduction during sustained maximal contractions, but voluntary activation usually completely recovers to prefatigue levels within ~30 s of exercise termination (see Fig. 1C and Refs. 29, 48, 50, 51, 98). The dissociation in the time course of recovery between MVC and voluntary activation implies that the sustained impairments in voluntary force production originate predominantly within the muscle fibers. Further support for this conclusion derives from observations of prolonged, incomplete recovery of electrically evoked twitches and tetani after repeated isometric contractions to the limit of tolerance (22). A detailed coverage of what is currently known about the physiological processes that accompany sustained exercise is beyond the scope of this paper, but see Taylor et al. (92) and Allen et al. (2) for fuller accounts of central and peripheral fatigue mechanisms, respectively. Here, we provide a brief overview (see Fig. 2 for a summary) and emphasize that the time courses of change in these processes need not reflect that of the functional recovery in voluntary force. This is because physiological responses to sustained exercise may either contribute to, or compensate for fatigue, and recovery of voluntary force is ultimately determined by the interplay of such underlying processes. For example, during sustained maximal contractions, both the excitatory and the inhibitory (silent period) responses of motor cortex output cells to transcranial magnetic stimulation increase. These changes suggest extra cortical excitability, which should improve motor output, but also extra cortical inhibition, which might contribute to fatigue. At the same time the extent to which voluntary output from the cortex can harness the full capacity of muscles decreases (i.e., there is supraspinal fatigue; e.g., Refs. 29, 40, 41, 91). Stimulation during intermittent MVCs with different duty cycles shows that these three effects have different time courses of development and return to baseline, with the silent period returning to baseline in ~10 s, the excitatory response to cortical stimulation in 15–30 s, and supraspinal fatigue in ~1 min (91). Although the factors that underlie a failure to harness the full capacity of cortical outputs to drive motoneurons appropriately for maximal voluntary force generation are not known, a role for feedback from group III and IV muscle afferents is likely. Fig. 2.Schematic illustration highlighting some key neuromuscular responses to exercise. Various physiological processes that are affected by fatigue are identified in bold lettering, with the approximate time course of their recovery (or return to baseline) indicated in parentheses. Additional information about the mechanism by which physiological changes impair function is provided where possible. MN, motoneuron. When firing of metabolically sensitive muscle afferents is prolonged after a fatiguing contraction by preventing blood flow to the muscle, supraspinal fatigue continues until blood flow is allowed to resume (29, 51). Moreover, firing of afferents from the fatigued muscle affects voluntary activation of other muscles in the same limb (50, 51). In contrast, the excitatory and inhibitory responses elicited by stimulation of the motor cortex typically return to pre-exercise values despite the occlusion (29, 51). This suggests that muscle afferent firing may limit drives to the motor cortex (and other descending) output cells during maximal effort, without apparent direct actions on motor cortical cells. However, debate continues on the actions of group III and IV afferents on motor cortical excitability because responses evoked by stimulation of the cortex and measured in the muscle are influenced by both cortical and spinal excitability. Hence interpretation of changes in responses to cortical stimulation is not clear cut (49). Indeed, it is possible that supraspinal fatigue could occur despite relatively stable outputs from supraspinal centers. Here, central fatigue would be generated by changes in input-output properties of the motoneuron pool, such that a similar set of cortical outputs that are untapped by volition and available to artificial stimulation would have a proportionally greater effect on muscle force. In contrast to the uncertainty regarding supraspinal contributions to fatigue, it is clear that central fatigue must be affected by the motoneuron pool itself (see Refs. 28, 57, 92 for reviews). Changes at this level can arise from tonic and phasic reflex inputs and other inputs associated with the exercise as well as changes in intrinsic properties of the motoneurons. Superimposed on such changes are neuromodulatory effects, produced for example, by descending monoaminergic drives. Although these changes are the focus of current work in human and animal studies, it is not simple to link their effect to a precise aspect of motoneuronal or spinal behavior or to determine their effect on motor output in a voluntary contraction. However, two things are clear. First, changes in the excitability of the motoneuron pool must be compensated by changes in descending drive to keep motoneuronal output constant. Hence a reduction in excitability (through inhibition or disfacilitation, see below) would necessitate greater drive. Such a reduction would likely produce a greater subjective effort for the same submaximal motor output. Second, the changes documented so far at a motoneuronal level have a range of time courses, ranging from milliseconds to minutes. Some examples are given briefly below. Inputs from group III/IV muscle afferents can act at segmental sites to modify excitability of the motoneurons and at supraspinal sites to affect the level of drive to the motoneuron pools (13, 18, 28, 57, 85). Existence of these effects has long been studied with circulatory occlusion (e.g., Ref. 14). Not surprisingly, restoration of muscle blood flow and removal of K+ and other metabolites rapidly attenuates the central effects of group III/IV muscle afferent firing with recovery of voluntary activation in ~30 s (48, 50, 51). More recently, lumbar intrathecal injection of fentanyl has been used to reduce group III/IV inputs to the central nervous system and attenuate an inhibition of voluntary motor output (e.g., Refs. 3, 5). Two approaches illustrate depression of motoneuronal “excitability” after voluntary isometric exercise. First, during relaxation after contraction, the propensity of the motoneurons to discharge a recurrent action potential (termed an F wave) is depressed for several minutes after a 2-min MVC (e.g., Refs. 52, 81). This depression occurs in hand and leg muscles and is less for weaker contractions (53). Although in simple terms this can be considered a depression in intrinsic motoneuronal behavior (i.e., because it is seen in an evoked response that does not require synaptic activation), one constraint is that the measurement is dominated by changes in large high-threshold motoneurons in the pool (e.g., Refs. 24). Second, evidence for profound change at a spinal level comes from the use of high-intensity conditioning TMS during an MVC to interrupt descending voluntary drive and allow the motoneurons to be tested during artificial “relaxation.” Testing is done with a cervicomedullary stimulus, which produces a muscle response by activation of corticospinal axons. Studying motoneuron behavior in the absence of volitional activity greatly simplifies the range of factors that are at play and makes it possible to determine mechanisms. After 15 s of an MVC of elbow flexors, the corticospinal response is virtually abolished (61). This spinal inhibition affecting the corticomotoneuronal path takes 2–3 min to recover after the end of the fatiguing MVC. The phenomenon also occurs during and after submaximal contractions and preferentially affects the motoneurons active in the contraction (59). Finally, although detailed consideration of the intramuscular processes that determine recovery from exercise are beyond our scope (refer to Ref. 2), characteristics of evoked forces illustrate some general principles. For example, reductions in evoked twitch magnitude and tetanic forces evoked by low-frequency stimulation are consistently greater than declines in MVC or high-frequency tetanic force (16, 22, 83, 95, 102). The time course of recovery of forces evoked by high-frequency stimulation is also much more rapid than that of recovery of low-frequency stimulation forces (or twitches). Force produced by high-frequency stimulation returns near to baseline within 20 min, even after a prolonged series of contractions to the limit of endurance in the presence of ischemia, whereas low-frequency force impairments can persist for more than 24 h (22). Differential fatigue and recovery effects as a function of motoneuron firing frequency likely follow from the sigmoidal shape of the Ca2+ force relation (2) and may reflect alterations in release or reuptake of Ca2+ from the sarcoplasmic reticulum or reduced Ca2+ sensitivity at the contractile apparatus. Note that single twitches create conditions that lie close to the origin of the Ca2+ force relation. By contrast, the rapid partial restitution of high-frequency force in the first seconds of recovery probably follows from muscle reperfusion, with clearance of K+ allowing repolarization of the t-tubule membranes likely to play a major role (2). Moreover, the general principle that responses to sustained exercise can either contribute to, or compensate for, fatigue holds for peripheral as well as central processes. For example, exercise can cause a slowing in the contractile properties of muscle, such that a lower rate of muscle fiber action potentials is required to generate a fully fused tetanus (39, 100, 103). This type of effect would partially compensate centrally mediated declines in motoneuronal firing rates, although the presence of central fatigue underscores the fact that, despite this partial compensation, voluntary drive is insufficient to generate the maximum evocable muscle force. The contrast between the fatigue responses in maximal and sustained or intermittent submaximal contractions can inform understanding of the factors that determine recovery after exercise (93). Although central fatigue cannot be measured using peripheral or cortical stimulation during the submaximal task, it can be documented during maximal efforts inserted within the main task. Furthermore, although not a direct measure of central fatigue, perceived effort increases out of proportion to the level of EMG. This is best seen in a sustained contraction in which the participant holds a submaximal target EMG level. In such contractions, the alteration in the EMG to force relationship produced by peripheral fatigue results in reduced force output. However, participants report that progressively more effort is required to do the task; which is to produce the same EMG (e.g., Ref. 60). This suggests that central mechanisms also influence performance during submaximal tasks (56, 80, 81). A key distinction between maximal and submaximal tasks is that additional motor units are progressively recruited as fatigue develops during sustained low-force contractions (1, 19, 30). By contrast, it is likely that all available motor units are recruited at high rates at the beginning of a sustained MVC and firing rates progressively decline with fatigue and may eventually cease in some high-threshold units (e.g., Ref. 71). Thus, for a given contraction duration, less fatigue occurs in high-threshold units for submaximal than maximal contractions. This may be related to the observation that central fatigue contributes proportionally more to the total force reduction during sustained submaximal than maximal contractions. For example, impaired voluntary activation accounts for ~65% of the reduced MVC during 70 min of elbow flexion at 5% MVC (88), ~40% of the MVC drop during 43 min of contraction at 15% MVC (89), but only ~25% of the force drop for a 2 min MVC (100). There is likely to be some maintenance of muscle perfusion during submaximal contractions, depending on the target force, and duty cycle when contractions are intermittent, which should reduce the accumulation of metabolites that leads to both firing of the subset of group III and IV afferents that are sensitive to noxious stimuli and t-tubule depolarization by K+. Accordingly, resting twitches evoked by motor nerve stimulation do not recover appreciably within 20–30 min after sustained, weak contractions of the elbow flexors (88, 89) (see Fig. 1B), suggesting that mechanisms of peripheral fatigue in such conditions relate mainly to impaired intracellular Ca2+ handling or sensitivity. Despite slow recovery of evoked twitch forces, MVC force typically shows rapid, but partial, recovery within the first few minutes after termination of sustained submaximal contractions. Voluntary activation measured by motor nerve or motor cortical stimulation has a correspondingly rapid initial recovery component but may not return to prefatigue levels until 20–30 min postexercise (46, 47, 88, 89, 104, 105) (Fig. 1C). Perceived effort, measured during brief efforts, takes ~5 min to recover fully (80) but has not often been documented. Thus, although the initial, partial restoration of voluntary force after sustained low force contractions is likely due to central fatigue recovery, impaired voluntary activation persists for longer after submaximal contractions sustained for 6–70 min than after maximal contractions sustained for up to 2 min. The mechanism underlying this delayed central recovery is not known. Sustained contractions at a single joint are a convenient model to study fatigue and involve physical demands that are similar to some activities of daily living (e.g., holding a bag of groceries). However, there is uncertainty about the degree to which the processes that constitute fatigue in such tasks also apply to activities such as walking, running, and cycling, which typically require higher rates of energy use, and consequently greater cardiovascular and ventilatory demands. There is an extensive literature on the physiological responses to fatiguing locomotor exercise (see Refs. 37, 62, 66, 84 for reviews), but direct measurement of muscle fatigue is challenging in such tasks because it is difficult to measure force-generating capacity during and immediately after exercise: there is typically some delay required to couple subjects to a myograph and initiate neuromuscular recording and stimulation. Nonetheless, muscle fatigue has been documented after running (55, 58, 65, 78, 80, 95, 101), cycling (4, 5, 7–12, 15, 20, 33, 35, 38, 43, 44, 54, 56, 67, 68, 79, 83, 85, 94, 96, 97), and skiing (63) of durations ranging from a few minutes to multiday ultraendurance events (see also Refs. 13, 62 for review). Care is needed in interpretation of this literature, however, because time and logistical constraints sometimes prevent satisfaction of criteria necessary to ensure valid measurements (see Refs. 28, 99). Note also that general trends in recovery time courses are more difficult to identify from the available data on locomotor exercise (Fig. 1). During locomotor exercise at a constant power output, sense of effort and EMG amplitude increase progressively over time (9–11, 35, 97), suggesting that fatigue accumulates throughout exercise. Although it is difficult to measure muscle fatigue directly within the first 1–2 min postexercise, evidence from a rhythmic “locomotor-like” knee flexion/extension task suggests that there is rapid, but partial, recovery over tens of seconds that is typical of sustained maximal and submaximal isometric contractions (26, 27, 67). However voluntary force capacity is still reduced from baseline at 1–3 min after termination of fatiguing locomotor exercise, and this is due to both peripheral and central fatigue (7, 33, 35, 38, 83, 96, 97). As for single joint isometric contractions, the relative contribution of impaired muscle function and voluntary activation to muscle fatigue probably depends upon the duration and intensity of exercise (17, 96, 97), with peripheral fatigue contributing relatively more to MVC reduction after short, high-intensity exercise, and central fatigue contributing relatively more during longer-duration, moderate-intensity exercise (see Fig. 1, D–F). The extent of central fatigue development may depend more on exercise duration than intensity, because longer-duration trials resulted in greater voluntary activation reductions than short-duration trials when exercise was self-paced and involved a high-intensity “end-spurt” (97). Both central and peripheral fatigue can persist for well over 30 min after prolonged locomotor exercise, with extreme endurance events lasting many hours or days reportedly resulting in the longest-lasting impairments (72, 79, 80, 83, 95). Repeated sprints and sports such as tennis and soccer also induce prolonged central and peripheral fatigue (31, 34, 42, 66, 69, 70, 75). However, systematic attempts to document the time course of recovery as a function of exercise duration and/or intensity have not been made. Recovery is further complicated for running, which involves eccentric contractions that induce muscle damage, because damage induces long-lasting impairments in evoked muscle forces and voluntary activation (73). Despite this, it appears that the determinants of peripheral fatigue recovery may be similar for single joint isometric contractions and locomotor exercise. Prolonged reductions in twitch forces and forces evoked by low-frequency stimulation occur in both (28, 56, 62, 83, 102), as do reductions in Ca2+ ATP-ase activity and Ca2+ uptake into the sarcoplasmic reticulum (16, 22, 32, 102). By contrast, central fatigue persists longer after locomotor exercise than after high-force isometric contractions. The factors that underlie persistent central fatigue after sustained low-force isometric contractions might be important, but there is also the possibility that factors associated with homeostatic regulation of body temperature (33, 68, 77, 98), systemic oxygen or carbon dioxide concentrations (6, 11, 12, 21, 36, 45, 64, 77, 82, 87), and metabolism (90) contribute. One potential candidate is reduced serotonergic or other neuromodulatory inputs to the motoneuron pool. For example, 35–40 min of fatiguing locomotion reduces firing rates of serotonergic neurons in the medullary raphe nuclei of cats (25), and spontaneous firing takes ~45 min to recover to baseline rates. The time course and mechanisms of recovery after fatiguing exercise are highly dependent on the characteristics of the preceding exercise bout. We have summarized some key points in Table 1, but considerable work remains before a complete description of the recovery processes will be possible. It is currently unclear what factors underlie the prolonged central fatigue that can accompany long-duration single joint and locomotor exercise. Work also remains to document how the time course of neuromuscular recovery is affected by exercise intensity and duration in locomotor exercise. Better understanding of the factors that modulate recovery from muscle fatigue may be of practical use to enhance rehabilitation and sports performance.
GRANTSThe Australian Research Council and the National Health and Medical Research Council of Australia supported this work. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONST.J.C. and J.L.T. prepared figures; T.J.C., J.L.T., and S.C.G. drafted manuscript; T.J.C., J.L.T., and S.C.G. edited and revised manuscript; T.J.C., J.L.T., and S.C.G. approved final version of manuscript. Page 3the immune system is integral to the body's defense against infection. It also influences other physiological systems and processes, including tissue repair, metabolism, thermoregulation, sleep/fatigue, and mental health. Over the past 40 years, exercise immunology has developed into its own discipline based on the recognition that the immune system mediates many exercise effects and that stress responses mediated through the nervous and endocrine systems play a key role in determining exercise-induced immune changes (84). A classic paradigm in exercise immunology is that an “open window” of immunodepression can occur during recovery from intense exercise. In particular, this paradigm proposes that after intense exercise, some immune variables (e.g., lymphocyte and natural killer cell numbers and antibody production) transiently decrease below preexercise levels. As a result of this immunodepression, microbial agents, especially viruses, may invade the host or reactivate from a latent state, leading to infection and illness (87). If exercise is repeated again while the immune system is still depressed, this could lead to a greater degree of immunodepression and potentially a longer window of opportunity for infection (87). Exercise-induced fatigue exists on a continuum. Repeated bouts of intense exercise on the same day or over several days may cause acute fatigue, as indicated by an inability to maintain exercise workloads (64). An athlete who trains intensely for 1−2 wk may experience a state of “functional overreaching,” which is associated with a temporary performance decrement, followed by improved performance. Intense training over an extended period without sufficient balance between training and recovery may lead to “nonfunctional overreaching” (NFOR; 64). This condition is typically characterized by persistent fatigue, performance decrement, muscle soreness, and psychological and hormonal disturbances that can last for weeks or months. Depending on the time needed to recover from NFOR, an athlete may be diagnosed (retrospectively) as experiencing “overtraining syndrome” (64). Recognition of the link between excessive training and risk of illness has stimulated interest in nutritional supplements and physical therapies to counteract immunodepression and to restore immune function after exercise training. In this mini-review, we update the current state of knowledge about the temporal changes in the immune system following exercise; how repeated bouts of exercise on the same day, extended periods of intense training, and sleep disruption influence the immune system; and the efficacy of various strategies for restoring immune function after exercise. A single exercise bout causes profound changes in the number and composition of blood leukocytes that may persist long into exercise recovery. All major leukocyte subpopulations tend to increase in number during exercise as a result of hemodynamic shear stress and/or catecholamines acting on leukocyte β2-adrenergic receptors (126). The postexercise recovery period is marked by opposite effects on blood neutrophil and lymphocyte numbers. Neutrophil number (and, consequently, the total leukocyte count) often continues to increase long into the recovery period (up to 6 h after exercise cessation), particularly if the exercise bout is prolonged (>2 h; 86). This sustained “neutrophilia” is characterized by an increased presence of immature, less differentiated, precursor neutrophils in the blood (117), most likely in response to the increased plasma levels of soluble agents including glucocorticoids, growth hormone, and cytokines such as IL-6 and granulocyte colony-stimulating factor, which mobilize myeloid cells from the bone marrow (117). Although this neutrophilia following prolonged exercise is akin to that observed during bacterial infection (>7.0 × 106/ml), 24 h of recovery are usually sufficient for neutrophil number to return to normal (126). A delayed monocytosis is sometimes observed within 1−2 h after very prolonged exercise, but monocyte number typically returns to the resting level within 6 h after exercise cessation (126). By contrast, lymphocyte number decreases rapidly after exercise. Following prolonged and/or high-intensity exercise in particular, lymphocyte number commonly decreases to below the preexercise value within as little as 30 min (126). This “lymphopenia” can often reach levels typical of clinical lymphopenia (<1.0 × 106/ml), but the lymphocyte count is usually restored to both the resting and clinically normal level within 4−6 h of recovery (126). After prolonged bouts of exercise (e.g., 2-h cycling), natural killer (NK) cells (which account for most of the exercise-induced lymphocytosis) may be ∼40% lower than the baseline value for up to 7 days after exercise (104). Exercise-induced lymphopenia reflects the preferential movement of lymphocyte subtypes with potent effector functions (e.g., NK cells, γδ T cells, and CD8+ T cells) out of the blood. Even within these subsets, there is a preferential egress of discrete subtypes of highly differentiated NK cells, γδ T cells, and CD8+ T cells with phenotypes associated with tissue-migrating potential and effector capabilities (107). The rapid lymphopenia observed during the early stage of exercise recovery was initially of concern, particularly because early studies reported large rates of lymphocyte apoptosis (programmed cell death) after exhaustive exercise (62). However, these findings have not been substantiated. Subsequent studies have reported lymphocyte apoptosis of the order of 0−2% after exercise, even though the blood lymphocyte count was up to 30−40% lower than at rest (66, 105). Lymphocytes and monocytes leave the blood in large numbers during exercise recovery under the influence of glucocorticoids. Lymphocyte subtypes that preferentially egress the peripheral blood during exercise recovery also have phenotypes consistent with tissue migration (e.g., expression of surface adhesion molecules and chemokine receptors; 108). These lymphocytes most likely translocate to peripheral sites of potential antigen encounter, such as the lungs or the gut (48). The skin has long been considered a likely destination for effector lymphocytes in response to exercise and stress in general (24). However, recent evidence indicates that CD8+ T cells and NK cells mobilized by exercise do not express cutaneous homing receptors on their surface (121). Exercise appears to “prime” effector T cells, thereby allowing them to transmigrate to the peripheral tissues that require enhanced immune surveillance following physical stress (54). Compared with the resting condition, the percentage of circulating lymphocytes expressing effector cytokines is lower following prolonged exercise (115), but it is unknown whether this decline reflects impairment at the individual cell level or preferential movement of effector T cells into peripheral tissues (e.g., lungs and gut). Recent evidence showing that exercise redeploys T cells that are specific to latent herpesviruses such as cytomegalovirus (CMV) and Epstein-Barr virus (EBV; 111, 112) suggests that this response may be a countermeasure against stress-induced viral reactivation (107). Exercise may also mobilize “older” functionally exhausted/senescent lymphocytes to undergo apoptosis in the tissues and allow new “recruits” to take their place (106, 107). Monocytes mobilized by exercise are likely to infiltrate skeletal muscle and differentiate into tissue-resident macrophages that facilitate repair and regeneration, particularly following arduous bouts of exercise that cause significant skeletal muscle damage (85). Monocytes with effector phenotypes are also preferentially redeployed after exercise. The CD14+/CD16+ “proinflammatory” monocytes are preferentially mobilized over their CD14+/CD16− counterparts (109). Monocyte expression of pathogen recognition receptors [e.g., toll-like receptors (TLRs)] tends to decrease in response to moderate-intensity exercise (109). Conversely, prolonged, intense exercise (60-km cycling time trial) increases TLR2 and TLR4 expression on monocytes, which may indicate a heightened proinflammatory state (11). A recent study showed that acute exercise mobilizes angiogenic T cells, which may facilitate vascular remodeling during exercise recovery (53). Exercise is also known to mobilize hematopoietic stem cells, which may participate in skeletal muscle repair and regeneration after exercise (25, 49). It has been suggested that exercise may have a role as an adjuvant to mobilize stem cells in donors for hematopoietic stem cell transplantation (25). In addition to cellular redeployment, the recovery phase of exercise, especially following very prolonged bouts of endurance-based exercise, is marked by striking alterations in the functional capacity of several blood leukocyte populations. Neutrophil bactericidal activity is greatly influenced by the intensity and duration of exercise. For example, 1 h of cycling at 50 vs. 80% of V̇o2max increases and reduces neutrophil oxidative burst activity, respectively (94). During the early stages of recovery after exercise, neutrophil bactericidal activity continues to increase after 40 min to 1 h of moderate-intensity exercise, whereas it remains impaired after exhaustive or prolonged exercise (86). NK cell cytotoxicity after exercise bouts of relatively short duration tends to remain unchanged on a per cell basis during recovery (81) but may decline after very prolonged bouts (33). T cell proliferation in response to mitogen stimulation typically decreases both during and after exercise, regardless of exercise modality, intensity, or duration (126). Prolonged exercise may also reduce T cell homing and migration (8), lipopolysaccharide-induced cytokine secretion by monocytes (113), and the percentage of T cells producing effector cytokines in response to mitogen stimulation (115). Thus the general trend during exercise recovery is that short bouts of moderate-intensity exercise have little effect on (or might even enhance) cellular immune function, whereas prolonged bouts (>1.5 h) of heavy exertion appear to reduce the normal functioning of all major immune cell subtypes. These effects may leave athletes susceptible to illness during recovery from competition or heavy training (87). The repeated-bout paradigm (87) proposes that compared with a single bout of exercise, repeated exercise bouts on the same day (27, 73, 96, 102) or over several days (43) cause different changes in circulating cell counts, lymphocyte proliferation, and NK cytotoxicity. Subsequent research has investigated changes in other immune responses to repeated exercise bouts on the same day with short vs. long recovery and intensified training over weeks or months. Figure 1 summarizes the evidence for changes in the immune system after repeated exercise bouts and days to months of intense training. Fig 1.Evidence heat map comparing differences in immune responses to two vs. one exercise bout on the same day (A), short vs. long recovery between bouts on the same day (B), consecutive days of exercise (C), and weeks (D) or months (E) of intensified training. Numbers represent the number of studies demonstrating an increase/greater change (red), no difference (green), or decrease/smaller change (blue) compared with the first bout of exercise, long recovery, before training, or healthy athletes (refer to text). ELA, elastase; MPO, myeloperoxidase; NK, natural killer; LPS, lipopolysaccharide; MCP1, monocyte chemoattractant protein 1. Studies on the effects of one vs. two bouts of exercise on a single day have included physically active (56–58, 63, 96, 102), highly trained (23, 88), or elite (12, 73, 98, 99) participants. The exercises involved cycling (12, 27, 56–58, 63, 96, 98, 99, 102), running (23, 88), or rowing (73). The duration and intensity ranged from <15 min at maximal intensity (27, 73) up to 2 h at medium-high intensity (i.e., 60−75% V̇o2max; 56, 96). The recovery period between exercise bouts was most commonly 3−4 h but ranged from 45 min (102) to 12 h (88). Compared with the initial bout of exercise, the following immune variables typically show either a greater relative change or a higher absolute value after a second bout of exercise: total leukocyte count (57, 63, 99, 102), neutrophil count (23, 73, 96, 99, 102), oxidative burst (per neutrophil; 12), elastase release (per neutrophil; 57), CD4+ T cell count (73, 99, 102), whole blood IL-8 production (23), and NK cell activation (represented by CD69 expression; 99). Conversely, lymphocyte proliferation (96, 102) and whole blood IL-6 production (23) are typically lower following a second bout compared with the first bout of exercise. Three studies on the effects of recovery duration between two bouts of exercise included highly trained or elite athletes who cycled for 65 min at 75% V̇o2max, twice each day, with either 3 or 6 h between the exercise bouts (12, 97, 98). A short recovery period (i.e., 3 h) induces either a greater relative increase or higher absolute values for neutrophil count (12, 97), oxidative burst activity per neutrophil (12), and CD8+ T cell and NK cell counts (97) after the second bout of exercise. Exercise-induced changes in lymphocyte (97), monocyte, and eosinophil (12) counts; absolute oxidative burst activity (12); NK cytotoxicity (97); and plasma concentrations of IL-6 and IL-1ra (98) do not differ after a short vs. long recovery period. Studies detailing how the immune system responds to exercise repeated on consecutive days or every second day have included untrained or physically active participants (43, 116, 118) or well-trained or elite athletes (60, 73, 77, 79). Exercise included 3 × 6-min maximal rowing, repeated twice over 2 days (73), and 1−3-h cycling (77, 79, 118) or running (60) at 50−70% V̇o2max repeated over 3 days, every second day for 3 days (43), or daily for 7 days (116). Exercise-induced changes in plasma cytokine and elastase concentrations and cytokine mRNA expression in leukocytes and muscle diminish over consecutive days (77, 118). By contrast, changes in total leukocyte, neutrophil, and monocyte counts (73, 116, 118), lymphocyte proliferation (79), neutrophil chemotaxis (116), leukocyte IL-1ra mRNA expression (77), plasma myeloperoxidase concentration, and salivary IgA secretion rate (79) do not change over time. Changes in lymphocyte subsets (43, 73), oxidative burst activity (79, 116, 118), salivary IgA secretion rate (60, 77), and NK cell count and cytotoxicity (73, 79) over consecutive days are more variable. Studies on short periods (2−4 wk) of functional overreaching have reported decreases in resting neutrophil degranulation (95), lymphocyte proliferation, and antibody production (124). Neutrophil count, plasma cytokine concentrations, CD4-to-CD8 T cell ratio, and salivary IgA concentration are more variable (or do not change) in response to functional overreaching (39, 95, 124). Athletes who exhibit signs of nonfunctional overreaching and/or frequent upper respiratory illness present with lower salivary IgA concentration (26, 35, 60); lower cytokine production by monocytes, neutrophils, and dendritic cells (67); and a greater number of activated (CD25+) lymphocytes (29). Changes in differential blood cell counts, lymphocyte subsets, and NK cell count following extended periods of intensified training are variable (29, 35, 61). Studies of athletes exhibiting the hallmarks of overtraining syndrome (including illness) have not revealed any consistent or characteristic immune profile (30, 101). Sleep disturbances influence immunity via activation of the hypothalamic-pituitary-adrenal axis and the sympathetic nervous system (46). Chronic sleep disturbance and disruption to the normal circadian rhythm are associated with inflammation and desynchronization of rhythmic immune variables. These responses likely contribute to increased risk of infection, cardiovascular disease, and cancer in shift workers (21, 68). Despite evidence that athletes experience poor sleep patterns compared with nonathletes (16, 55), surprisingly little is known about how sleep disturbance influences the immune responses to exercise. Compared with normal sleep, a disrupted night's sleep appears to prime the immune system and enhance immunosurveillance by stimulating total lymphocytes, CD8+ T cells, NK cells, and γδ T cells to leave the blood and migrate to potential sites of infection during the early recovery period after exercise (45). By contrast, other studies indicate that a night without sleep does not influence leukocyte trafficking, neutrophil degranulation, or mucosal immunity at rest or after exercise (31, 93). Subtle immune changes have been observed after a night without sleep, including a shift toward a T helper 2 cytokine profile (46). It is uncertain whether these subtle immune modifications with acute sleep loss are clinically meaningful. When considering the potential effects of poor sleep on immunity in athletes, it is important to distinguish between acute and chronic sleep disturbance. Chronic sleep disturbance (12 nights, 50% sleep loss) increases the plasma inflammation markers C-reactive protein and IL-6 (38). However, intervening daytime naps can counter this apparent inflammatory response (103). Short sleep duration (<7 h/night for 7 days) decreases the response to hepatitis B vaccination and the likelihood of clinical protection (90). Similarly, a night of wakefulness after hepatitis A vaccination decreases the specific antibody response 2−4 mo later (52). People who experience poor-quality sleep and/or regular sleep deprivation also have a 4−5 times greater risk of developing the common cold (16, 91). Continued research efforts should be directed toward monitoring and improving sleep in athletes and understanding the implications for immune health. Research over the last 30 years has investigated whether nutritional strategies counteract exercise-induced immunodepression and systemic inflammation (32, 125). A comprehensive review of the literature on these is beyond the scope of this mini-review; other, more detailed reviews on this topic are available [e.g., Gleeson (32) and Walsh et al. (125)]. Here, we focus on the most effective nutritional strategies (primarily carbohydrate ingestion) for restoring systemic immune function in the first few hours after exercise (9, 10, 19, 51, 65, 76, 78, 82, 83) and over consecutive days (7). We also assess whether the timing of nutritional interventions (i.e., before, during, or after exercise) influences their effectiveness (50, 59). Carbohydrate supplementation during prolonged, intense exercise consistently attenuates exercise-induced increases in circulating cytokines (74, 125) and the redistribution of neutrophils (74, 76, 78), monocytes (76, 82), natural killer cells (78), and lymphocytes (51). The immunomodulatory effects of carbohydrates arise from better maintenance of blood glucose concentrations and blunted release of stress hormones such as catecholamines and glucocorticoids during and after exercise (51, 76, 78, 82, 83). Although the systemic release of IL-6 during exercise is related to muscle glycogen depletion (114), the precise mechanism by which carbohydrate supplementation reduces systemic IL-6 release from contracting muscle during exercise is not clear, because carbohydrate supplementation does not alter muscle glycogen content (75). In several studies, the immunomodulatory effects of carbohydrate supplementation were observed to “carry over” into the recovery period (i.e., ≥2 h postexercise; 51, 76, 78, 82, 83). Nieman et al. reported that carbohydrate supplementation during 2.5-h high-intensity running reduces the number of neutrophils (immediately and 1.5 h postexercise), monocytes (immediately and 6 h postexercise), and lymphocytes (immediately and 3 h postexercise) (76, 78). Extending carbohydrate ingestion to the postexercise recovery period also reduces neutrophil count (74) and blood granulocyte and monocyte phagocytosis 6 h postexercise (82). Lancaster et al. showed that carbohydrate consumption (30 and 60 g/h) during 2.5-h cycling minimized the suppression of CD4+ and CD8+ T lymphocytes, which express and produce IFNγ, during the 2 h following exercise (51). Considering that exercise-induced responses of the adaptive immune system are relatively slow (125), it is important to assess whether these effects are maintained over consecutive days of exercise. Carbohydrate ingestion before, during, and after two exercise bouts on 2 consecutive days attenuated the decrease in antigen-stimulated proliferative lymphocyte responses before exercise on the second day (7). Carbohydrate ingestion also enhanced lymphocyte proliferative responses to mitogen stimulation postexercise on the second day (7). These findings suggest that carbohydrates may help to diminish potential cumulative immunodepression over consecutive days of exercise. The immunomodulatory effects of carbohydrate may depend on the timing of carbohydrate intake. The ingestion of a glucose solution 15 min, but not 75 min, before 1-h high-intensity cycling prevented immunoendocrine perturbations (50). The lack of an effect of carbohydrates ingested 75 min preexercise was potentially associated with an insulin-induced decrease in the plasma glucose concentration before exercise, which, in turn, might have enhanced immunoendocrine responses (50). Carbohydrate ingestion during either the first or the second of two 90-min bouts of cycling on the same day better maintained plasma glucose and attenuated plasma stress hormone responses to the second bout (59). By contrast, carbohydrate ingestion during the 2-h recovery period between these exercise bouts had no such effects (59). These findings suggest beneficial effects of a timely carbohydrate supplementation (i.e., shortly before and/or during exercise) on immune responses to exercise. This may be particularly relevant with more prolonged and/or intense exercise protocols and when the recovery duration between two consecutive exercise bouts is short. Carbohydrate ingestion does not influence all aspects of the immune system. For example, carbohydrate supplementation does not alter the exercise-induced suppression of natural killer cell function (78) or salivary IgA secretion (18). Importantly, it remains unclear whether the immunomodulatory effects of carbohydrates have clinical relevance for resistance to illness or adaptation of the immune system to regular exercise stress (32, 125). Recent evidence indicates that carbohydrate supplementation during prolonged exercise blunts exercise-induced immune-endocrine perturbations but does not prevent the suppression of in vivo immunity (22). More research is required to examine the effects of carbohydrates (or other nutritional strategies) on in vivo immune function in response to acute and chronic exercise. Some studies have investigated the effects of dietary carbohydrate intake on immune responses to consecutive days of exercise intended to deplete muscle glycogen (9, 10, 34, 65). A higher carbohydrate intake consistently attenuated certain components of immunodepression well into the recovery period (i.e., ≥2 h postexercise) after the second exercise session (10, 34, 65). Athletic training often involves conditions of low carbohydrate availability, e.g., due to abbreviated recovery periods and/or as part of a “train low-compete high” training regime (41, 42). These investigations therefore have particular practical implications. Compared with a higher carbohydrate intake (8 g·kg−1·day−1), very low carbohydrate intake (0.5 g·kg−1·day−1) leads to greater perturbation in leukocyte subsets during recovery from exercise (65). These effects may be related to sustained elevation of plasma cortisol concentration (65). Bishop et al. observed that compared with a low-carbohydrate diet (1.1 g·kg−1·day−1), a high-carbohydrate diet (8.4 g·kg−1·day−1) for 3 days after glycogen-lowering cycling attenuated plasma cortisol and cytokine concentrations and circulating total leukocyte and neutrophil counts following subsequent exercise (10). Consuming a high-carbohydrate diet (8.5 g·kg−1·day−1) also reduces overreaching symptoms during 11 days of intense training, compared with moderate carbohydrate consumption (5.5 g·kg−1·day−1; 1). The periodic implementation of “train-low” strategies (e.g., by commencing training with low muscle glycogen stores) may further amplify metabolic adaptations in skeletal muscle (41, 42). When considering their dietary carbohydrate intake, athletes should aim to achieve a balance between minimizing immunodepression and maximizing metabolic adaptations in skeletal muscle. In view of the detrimental effects of low carbohydrate availability on the immune system, chronic carbohydrate restriction should be avoided during intense periods of training (32, 41). Additional research is warranted to better understand the effect of long-term training consisting of intermittent train-low sessions on immune function and susceptibility to illness. Recognizing the importance of protein for immunocompetence (15), there are benefits of postexercise protein ingestion (18, 19, 69) or a diet high in protein (128) on immune responses to exercise. On the basis of previous results indicating that exercise-induced lymphocyte trafficking was impaired during high-intensity training, Witard et al. examined whether a high-protein diet can restore these impaired immune responses (128). Consuming a high-protein diet (3 g·kg−1·day−1) helped to minimize exercise-induced changes in lymphocyte distribution during a period of intense training (128). Interestingly, an energy- and carbohydrate-matched normal protein diet (1.5 g·kg−1·day−1) failed to provide the same benefit (128). The high-protein diet was also associated with fewer self-reported upper respiratory illnesses (128). Another study demonstrated that protein and leucine supplementation for 1–3 h postexercise during 6 days of high-intensity training enhanced neutrophil respiratory burst activity after the last exercise session (69). Consuming a carbohydrate-protein solution immediately, but not 1 h, after exercise prevents a decrease in neutrophil degranulation during the postexercise recovery period (19). Recent research has shown that the timing, distribution, and amount of postexercise protein intake modulate the blood and tissue availability of protein/amino acids and adaptive responses of skeletal muscle (3, 42). Notably, amino acid-sensitive mammalian target of rapamycin (mTOR) signaling is also a key mechanism underlying leukocyte trafficking (110). More studies are therefore needed to examine whether different postexercise protein feeding patterns influence immune function during recovery from exercise. Except for carbohydrate supplementation, evidence for effective nutritional countermeasures to exercise-induced immune alterations is limited (32, 125). Among other types of nutritional supplements [e.g., probiotics and vitamin D; for reviews, see Gleeson (32) and Walsh et al. (125)], antioxidants and phytochemicals such as quercetin have been studied for their potential capacity to minimize immune perturbations, particularly during exercise recovery (75, 77, 79, 80). Some data point toward beneficial effects of quercetin supplementation on immune health after intense exercise (77, 79). Other findings suggest an increased need for nutritional antioxidants during the first 24 h of recovery from intense exercise lasting several hours such as an Ironman triathlon (71). However, taken together, the present literature is not sufficiently robust to recommend supplementation with phytochemicals or antioxidants to prevent immune suppression and illness in athletes and exercising individuals. Athletes often take high doses of antioxidant/phytochemical supplements in the belief that this will reduce their risk of illness (47). However, high doses of antioxidant/phytochemical supplements can interfere with training adaptations (42, 72). A natural diet rich in fruits, vegetables, whole grains, and nuts delivers antioxidants and phytochemicals in physiologically effective amounts that are most likely sufficient to help maintain immune function following exercise and during exercise training (32, 72, 125). In addition to nutritional interventions, other research has examined the efficacy of nonsteroidal anti-inflammatory drugs (NSAIDs) and various physical therapies for restoring immune function after exercise. Some studies (13, 20, 70, 89, 120, 127) have shown effects of NSAIDs, but other human studies have failed to demonstrate any effects of NSAIDs (122, 123), cryotherapy (44), compression garments (5, 37, 92), active recovery (2, 119), or other physical therapies (28) on immune responses during recovery from exercise (Fig. 2). Despite this lack of empirical evidence for the benefits of NSAIDs and physical therapies for restoring immune function after exercise, some of these treatments are associated with positive psychological outcomes and other effects not related to immune function after exercise (4, 17). Therefore, although the physiological effects of these physical treatments are not understood fully at the present time, they may confer some important benefits for athletes, which may involve the immune system, perhaps indirectly. Fig. 2.Summary of the effects of nonsteroidal anti-inflammatory drugs (NSAIDs) and physical therapies on immune changes during recovery from exercise. The redeployment of effector lymphocytes from blood to peripheral tissues is seen as an integral part of the physiological stress response and the immune system's response to prepare the body for potential injury by mobilizing its “troops” (cells) to increase immunosurveillance (24). Therefore immune integrity during exercise recovery may be characterized by the host's ability to redeploy effector lymphocytes effectively to the peripheral tissues. Lymphocyte redeployment may be impaired following very prolonged bouts of exercise or in athletes who are overreaching. The redeployment of CD8+ T cells both during and after exercise is significantly reduced after 1 wk of high-intensity training compared with normal training (129). Specifically, the egress of CD8+ T cells was 1.4-fold higher after normal compared with high-intensity training. Considering other evidence that stress-induced leukocyte redeployment is linked to poor clinical outcomes following surgery (100), immune cell redistribution and infection risk after exercise warrant further investigation. High-volume exercise training may impair the redeployment of viral-specific T cells and NK cells, thereby reducing antiviral “patrolling” during exercise recovery (Fig. 3). Herpes viruses such as EBV and CMV are highly prevalent in the population, and reactivation of these viruses from a latent state is indicative of systemic immunodepression (107). EBV viral DNA is present in saliva from athletes after even short periods of high-intensity training (36), but it is unknown whether this reflects training-induced impairment in the trafficking of virus-fighting lymphocytes. Fig. 3.Overreaching or heavy training is associated with impaired cellular immune function and latent viral reactivation. Reduction in lymphocyte trafficking and function during exercise recovery impairs immune surveillance, which may increase susceptibility to opportunistic infection. Lowered immunity may also allow previously acquired viruses to reactivate from a latent state, which may cause further immunodepression and increase susceptibility to infection. HSV-1, herpes simplex virus type 1. Cellular immune function in response to exercise is typically assessed in isolated cells ex vivo or at the cell population level in the blood compartment. This approach can make it difficult to interpret changes in immune cell function after exercise because of the massive alterations in the cellular composition of discrete leukocyte subtypes. On the one hand, it seems intuitive to interpret lower immune cell function measured in blood during the early stages of exercise recovery as indicative of immunodepression. On the other hand, it is equally possible that after exercise, the most functional immune cells (i.e., those with effector phenotypes and high tissue-migrating potential) are redeployed to other areas of the body where they are needed. If true, this suggests that systemic immunosurveillance may be enhanced during exercise recovery, despite an apparent depressed profile in the blood compartment. It is difficult to determine the biological significance of exercise-induced changes in immune cell function when assessed in vitro. This is because cell function is typically assessed relative to the total cell population (e.g., percentage of T cells responding to mitogen stimulation, number of target cells killed per NK cell, oxidative burst activity per neutrophil, etc.) without accounting for exercise-induced changes in the subset composition of these cell populations. For example, the proportion of NK cells expressing the activating receptor, NKG2C, is markedly elevated during exercise recovery (6). As a result, NK cell cytotoxic activity for the total NK cell population increases markedly when an NKG2C-sensitive target cell is used to assess NK cell function (6). Conversely, when an NKG2C-insensitive target cell line (K562) is used, NK cell killing is not affected by exercise (6). Therefore the proportional shifts in the composition of cell subtypes should be considered when interpreting exercise-induced changes in immune cell function at the total cell population level in vitro. Moreover, assessment of in vitro immune cell function using venous blood samples does not account for the complex interactions among immune cells and soluble factors within tissues (e.g., gut, lungs, and skin). However, some evidence suggests that reduced immune cell function in vitro may coincide with changes in vivo and rates of illness (14, 40). The validity of the original paradigm of cumulative immunodepression with repeated bouts of exercise (87) is somewhat difficult to assess. Months of intense training increase the incidence of illness in elite athletes (26, 30, 35). However, on the basis of these studies, we can only assume, but not assert, that increased incidence of illness results from an imbalance between training and recovery. Research that has systematically manipulated the balance between training and recovery has not identified any immune variables that are consistently depressed as a result of insufficient recovery after exercise. However, with one exception (79), these studies have not tracked the incidence of illness after repeated bouts of exercise. Reduced salivary IgA concentration and secretion rate (amount of IgA secreted over a fixed period) may predispose athletes to illness in the long term (26, 35). IgA binds microorganisms such as bacteria and viruses in the mucosa so that they can be destroyed by immune cells. However, short-term changes in salivary IgA concentration and secretion rate after repeated bouts of exercise are variable (56, 58, 60, 79). Salivary IgA concentration and secretion rate may decrease incrementally over longer periods. The repeated exercise models used in many of the studies described above induce only acute fatigue (64). Accordingly, smaller exercise-induced changes in immune variables following repeated bouts of exercise may actually represent positive adaptation of the immune system, as opposed to depression of immunity that may lead to illness. Pedersen et al. (87) suggested that there is a critical threshold for exercise intensity and duration that determines the risk of immunodepression after repeated bouts of exercise. However, no studies have systematically determined the effects of repeated exercise bouts of different intensity and duration. There are also no data on the effects of repeated bouts of anaerobic or resistance/strength exercise, or a combination of different types of exercise on the same day. The large gaps in Fig. 1 show that much remains to be learned about the effects of repeated bouts of exercise on the immune system. Among various nutritional interventions that have been studied to counteract immunodepression during exercise recovery, carbohydrate supplementation has proven the most effective. A balanced and well-diversified diet that meets the energy demands in athletes and exercising individuals is certainly a key component to maintain immune function in response to strenuous exercise and intense periods of training. Additional research is warranted to investigate how the timing and pattern in the ingestion of nutrients, particularly carbohydrates and protein/amino acids, influence recovery of the immune system after exercise. Sleep disturbances can depress immunity, increase inflammation, and promote adverse health outcomes in the general population. However, the limited data available on how sleep disturbances influence immune responses to exercise are inconsistent. Physical treatments that are used after exercise (e.g., hydrotherapy and massage) may enhance the athlete's sense of well-being and should be considered as adjunct therapies for maintaining immune health. GRANTSJ. M. Peake is supported by funding from the Centre of Excellence for Applied Sport Science Research at the Queensland Academy of Sport, Brisbane, Queensland, Australia. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSJ.M.P. and R.J.S. prepared figures; J.M.P., O.N., N.P.W., and R.J.S. drafted manuscript; J.M.P., O.N., N.P.W., and R.J.S. edited and revised manuscript; J.M.P., O.N., N.P.W., and R.J.S. approved final version of manuscript. REFERENCES
Page 4calcification and fibrosis of the main pulmonary arteries have been observed in a large proportion of young racehorses (1, 5). In humans, vascular calcification is now recognized as a major contributor to arterial stiffening (7), which is considered a major risk factor for several cardiovascular events (34). In fact, increased arterial stiffness has been associated with the development of systemic hypertension, microvascular disease in the brain and kidneys, and the development of pulmonary hypertension (11, 26, 33). Arterial stiffness increases with aging, hypertension, exposure to inflammation, or diseases like diabetes mellitus (13, 27, 30). In horses, arterial calcification has been described in cases of aortic rupture and aorto-pulmonary fistulation, resulting in sudden death (28). However, the importance of arterial calcification and stiffness in the progression of other equine cardiovascular diseases has not been studied. Elastic arteries like the aorta and the large pulmonary arteries function as conduits, as well as storing energy and blood during systole. As they distend to accommodate blood during systole, and then return to their presystolic form, they dampen the pressure created by the ventricular contraction and allow a steady blood flow to the peripheral tissues during the cardiac cycle (16, 34). A pressure and flow wave (the pulse wave) is created with each heartbeat and transmitted through the arterial tree to the capillary bed. As this wave moves peripherally, it faces the impedance (opposition to fluctuating flow) created by changes in the properties of the vascular wall or the vascular diameter (4). When the pulse wave encounters opposition, a fraction of the wave continues moving forward and the others are reflected back toward the heart. The multiple reflected waves add to each other, and the resulting wave reaches the main arteries (aorta or pulmonary) in late systole or early diastole. The speed at which this wave travels through the arteries is defined as the pulse wave velocity. In stiff arteries, however, because of the changes in mechanical properties of the arterial walls, the pulse wave velocity increases and the retrograde wave tends to arrive earlier. The early arrival means that, instead of arriving during the following diastole, the retrograde wave arrives during the following systole, adding-on to the systolic pressure wave being created. This change in timing will lead to an increase in systolic pressure, as well as to a decrease in the diastolic pressure (16), resulting in increased pulse pressure. On the other hand, the changes in elastic properties occurring in the large arteries lead to increased propagation of pulsatile blood flow deeper into the microcirculation, triggering inflammation, and vascular remodeling at the capillary level (33, 34). Carotid-to-femoral pulse wave velocity (PWV) (velocity of a pulse wave traveling between the human carotid and femoral arteries) is considered the gold standard method to determine aortic stiffness in humans (18). PWV can be determined using waveforms derived from blood pressure (15), arterial wall distension (36), or flow (6) recordings. In general, PWV (PWV = distance traveled by the pulse wave/time interval taken to travel said distance) can be measured in two ways: 1) by detecting a pulse wave in two locations (determining the interval of time needed for the wave to travel a distance) and then estimating the distance between the two locations where the wave was detected (e.g., carotid artery to femoral artery); or 2) by detecting two pulse waves consecutively in two different locations, while running an electrocardiogram to time the two waves detected (32). Considering the frequency with which arterial calcification was observed in racing horses and the common occurrence of diseases affecting the pulmonary vasculature (e.g., exercise-induced pulmonary hemorrhage), it is essential to develop a technique to assess arterial stiffness and its potential clinical implications. We hypothesized that using a dual-pressure sensor catheter, it is possible to collect arterial pressure data and determine PWV in the main pulmonary arteries of horses. Therefore, the main objective of this study was to develop a technique to determine the PWV in the main pulmonary arteries of horses. As a secondary objective, histological analysis of the pulmonary artery walls was performed to investigate a potential link between the presence of lesions and changes in PWV. MATERIALS AND METHODSTen mature horses, deemed clinically healthy on the basis of physical examination, were used. Horses with no abnormalities with cardiac, (i.e., increased heart rate, murmurs) or pulmonary auscultation (i.e., increased respiratory rate, crackles, or wheezes), fever, or ongoing medical treatments (e.g., systemic antimicrobials, corticosteroids, or hormonal therapies) were considered. This study was approved by the Animal Care Committee of the University of Guelph (Animal Utilization Protocol no. 3278) and conformed to the standards of the Canadian Council on Animal Care. Blood pressure waveforms were recorded simultaneously in two locations within the pulmonary artery (PA) to calculate the PWV. A custom-made catheter fitted with two pressure sensors with 5-cm distance separation, with the distal sensor placed 5 mm from the tip of the catheter [7 French (Fr) × 170 cm, Transonic Scisense, London, ON, Canada] was used to record the waveforms. In preparation for each trial, a 14 G × 13 cm catheter was placed in the left jugular vein for administration of sedatives. Hair was clipped over the cardiac silhouette on both sides of the thorax to facilitate ultrasound imaging of the heart and pulmonary artery trunk. Three small regions were clipped for electrocardiogram electrode placement (one over the right shoulder, one on the xyphoid region, and one on the pectoral muscles). The distal one-third of the right jugular groove was also clipped for placement of the pressure sensor catheter (PSC). For each trial, the horse was sedated and placed in stocks. The drugs used and the level of sedation were adjusted for each horse, according to its demeanor and behavior, as well as the optimization of the protocol at the time, as the procedures were simplified and the data collection became more efficient. Horses 1–5 were sedated with xylazine (0.5 mg/kg iv) and morphine boluses (0.07 mg/kg iv), and then maintained on a constant infusion rate of xylazine (1 mg·kg−1·h−1) and morphine (0.05 mg·kg−1·h−1 during PSC manipulation for data collection (Table 1). Horses 6, 9, and 10 were sedated only with xylazine boluses (0.3–0.4 mg/kg iv). Horses 7 and 8 were sedated with detomidine (0.01 mg/kg iv), and butorphanol (0.01–0.02 mg/kg iv) boluses for PSC placement, and remained sedated with xylazine boluses (0.3–0.4 mg/kg iv). In an attempt to minimize the potential effect of sedation, horses 6–10 were allowed a washout period of 30 min after the last sedation with xylazine, before data were collected.
The ECG electrodes were applied in order to measure cardiac rhythm and monitor rate during the entire procedure. The right jugular groove was aseptically prepared, and local anesthetic block with bupicavaine was performed (Hospira, Saint-Laurent, Quebéc, Canada). A stab incision was made using a no. 15 scalpel blade to facilitate placement of an introducer needle (14G × 5 cm) in the jugular vein. This needle was then exchanged over-the-wire (0.038′′ × 50 cm) for a tear-away introducer sheath (16 Fr × 13.5 cm, ref. 32365; Qosina, Edgewood, NY). Subsequently, a 9 Fr × 100-cm-long catheter introducer sheath (LCIS, Super Arrow-Flex model CL-07900; Teleflex, Markham, ON, Canada) was passed into the vein, and the tear-away introducer was removed. The LCIS was advanced into the pulmonary artery trunk via the right heart under transthoracic ultrasonographic guidance. Once the catheter was fully inserted into the pulmonary artery trunk, it was secured in place by suturing it to the skin. A Tuohy-Borst valve was attached to the LCIS using a custom-made adapter, and the catheter was flushed continuously with a heparinized saline solution (10,000 IU/l). The pressure sensor catheter was calibrated before each trial, according to the manufacturer’s recommendations. Briefly, a two-point calibration curve was made (0 mmHg and 100 mmHg) using the pressure control unit (SP200 Pressure System, Transonic Scisense, London, ON, Canada). Then, the catheter was submerged in 0.9% saline solution for 20 min at room temperature. Finally, each pressure sensor was held just below the meniscus of the saline solution (where the pressure is closest to 0 mmHg), and the offset (above or below 0 mmHg) was corrected by adjusting the output of each sensor to zero, using the pressure control unit. Once the LCIS was in place, the PSC was advanced into the pulmonary artery, until the proximal sensors were between 10 and 20 cm beyond the tip of the introducer sheath. Thoracic radiographs were taken to determine the location of the PSC within one of the main branches of the pulmonary artery. The catheter was manipulated if required, and radiographs were repeated until the PSC was in the desired location. Continuous ECG and pressure data were collected at maximum insertion of the PSC and then at 5-cm steps backward along the pulmonary artery, as the catheter was withdrawn into the heart. The pressure waveforms were recorded at a frequency of 2,000 Hz, and data were recorded for a minimum of 2 min in each location. The signals produced by the PSC and the ECG tracings were acquired simultaneously using a commercially available data acquisition system (PowerLab model no. ML870, and BioAmp model no. 136; ADInstruments, Colorado Springs, CO), and recorded with the LabChart 7 software (ADInstruments). The PSC and LCIS were removed, and the horses were humanely euthanized with an overdose of pentobarbital sodium immediately after or within 24 h of data collection. Minimum diastolic, peak systolic, and mean arterial pressures (MAP) were determined using the data pad automatic functions of LabChart 7. Pulse pressure (PP) was determined for each cycle by subtracting the diastolic pressure from the systolic pressure. Pressure data from 10 consecutive cardiac cycles were analyzed, and the mean of diastolic, systolic, mean arterial, and pulse pressures were determined for each position. Ten consecutive full cardiac cycles were used to calculate the pressure wave transit times, at each individual sampling location, for each horse using MathLab (The MathWorks, Natick, MA). The transit times were determined using the statistical phase offset (SPO) method (31). Briefly, this method uses a series of data points for the calculation of transit time. First, all raw data are smoothed using a 21-point moving average. Then, the difference between the two pulse pressure waveforms across a given interval is determined. The mean of these differences and the standard deviation are calculated. The distal waveform is then incremented a single time step, 0.0005 s for this study, and the difference average and SD calculations are repeated. These steps are repeated until the minimum SD of the pressure differences is detected, which occurs when the two waves overlap the most. The sum of time steps needed to reach the minimum SD will correspond to the transit time for the pulse wave between the two sensor locations. The PWV was determined using the formula: PWV = (Δd/Δt), where PWV is the pulse wave velocity, Δd is the distance between pressure sensors, and Δt is the transit time of the pulse wave. The step at which the proximal sensor passed the pulmonic valve into the right ventricle was determined by detecting a significant increase in pulse pressure and a distinctive pressure waveform. On the basis of this information, the location of the sensors in the PA at each measurement was estimated considering the 5-cm steps taken from distal to proximal PA. Complete circumferential ring samples of the pulmonary artery trunk (within 5 cm from the bifurcation) and main branches (up to 2 cm distal to the bifurcation) were collected, fixed in 10% buffered formalin and embedded in paraffin. Fixed tissues were cut (without decalcification) into sections of 5-μm thickness and stained with HE for routine histologic examination. Histological examination of all samples was performed by two of the authors (G. Silva and L. Arroyo) who were blinded to the results obtained for PWVs. Each slide was classified on a categorical scale: normal, when there were no obvious calcium deposits; elastin fiber disruption or accumulation of disorganized connective tissue; mild lesions, when there were focal calcium deposits and/or accumulation of connective tissue affecting less than 25% of the medial thickness; moderate lesions, when there were focal calcium deposits and/or connective tissue deposition across 25–50% of the medial thickness; and severe lesions, when there were focal calcium deposits and/or connective tissue deposition occupying more than 50% of the medial thickness. When the lesions were multifocal and diffuse along the tunica media, the lesion was also classified as severe. For each variable (MAP, systolic pressure, diastolic pressure, PP, PWV), the mean and SDs were calculated at each position of data collection. The average MAP, PP, and PWV of all horses were also calculated. The presence of outliers was investigated using the Grubbs test, set at 10% significance level. The data were log-transformed, and maximum and minimum values at each point were consecutively tested in two cycles to determine whether either should be considered an outlying value. After each cycle, values considered outliers were removed from the data. Statistical analysis was performed with SAS (Statistical Analysis Software, Singapore). The PWV, MAP, and PP measurements were analyzed in light of the histology results, and correlation statistical analysis was considered to investigate any associations between the PA PWV and the histological classification. However, considering the significant variability among the subjects (age, preconditioning, and sedation protocol), which was not controlled for in this study, correlation analysis was not determined. RESULTSSix mares, three geldings, and one stallion with a mean age of 10 yr (ranging from 3 to 20 yr) were included. Seven were Thoroughbred and three were Standardbred horses. The mean body weight was 481 kg (ranging from 348 kg to 646 kg) (Table 1). Catheterization of the main branches of the pulmonary artery was achieved and confirmed by thoracic radiographs in all horses (Fig. 1B). The LCIS reached the left pulmonary artery branch in 9/10 horses and the right pulmonary artery in one case, as confirmed by echocardiography. Catheter manipulation for optimal placement in one of the main PA branches was required in 3/10 cases. Fig. 1.A: Pulmonary artery trunk (horse 9) and main branches with calcified lesions on the endothelial surface (arrows). B: lateral thoracic radiograph of a horse (horse 2) showing the pressure sensor catheter (PSC) in the desired location within the pulmonary artery. The thin arrow shows the distal tip of the catheter introducer sheath, and the PSC can be seen as a thin straight white line (white arrow) fitted with two pressure sensors (black arrows). Cardiac arrhythmias were recorded in all horses (except horse 6) at some point during catheter placement and/or the data collection period. The following arrhythmias were observed: second degree atrioventricular blocks, atrial premature contractions, ventricular premature contractions, and ventricular tachycardia (Table 1). All arrhythmias resolved spontaneously once catheter manipulation was discontinued. ECG was not recorded in horse 6 due to an equipment malfunction. The mean (±SD) PWV was 2.3 ± 0.7 m/s in the proximal PA trunk and 1.1 ± 0.1 m/s 15–20 cm more distal in a main PA branch (Fig. 2A). The mean (±SD) of the mean arterial pressures was 30.1 ± 5.2 mmHg in the proximal PA trunk, and 22.0 ± 6.0 mmHg at the distal site (15 cm) in a main PA branch (Fig. 2B). The mean (±SD) pulse pressure was 15.0 ± 4.7 mmHg and 13.5 ± 3.3 mmHg, in the proximal PA trunk and distal main PA branch site (15 cm), respectively (Fig. 2C). Calculation of lower and upper limits for detection of outliers was performed. There were several transit time (TT) results within this interval that were considered outliers because they appeared to deviate significantly from the expected results. In horses 1, 2, 6, 7, 9, and 10, there were 1–4 TTs considered to be outliers and removed from the calculations. Horse 3 was removed from the final data set due to unreliable recorded data, including unexplained spikes or drops in pressure that resulted in the poor quality of the pulse waves available for PWV calculations. On further analysis of the data collected in this horse, it was concluded that the faulty data were the result of a poor connection between the PSC and the data acquisition system. Fig. 2.A: graphic representation of pulmonary artery pulse wave velocities for each horse and the average of all horses (colored lines) measured from proximal pulmonary artery (PPA) to distal (20–25 cm) at different depths within the pulmonary artery. Mean arterial pressures (B) and pulse pressures (C) measured for each horse and averaged for all horses (colored lines) measured with the proximal pressure sensor at the most proximal site in the pulmonary artery. Horses 1, 5, 9, and 10, aged 20, 3, 18, and 16 yr old, respectively, tended to have higher PWV, while horses 2, 3, 4, 6, 7, and 8, aged 7, 6, 4, 16, 5, and 5 yr old, respectively, tended to have lower PWV at the level of the PA trunk (Fig. 2A). Horses 4, 5, 9, and 10 had higher MAP, while horses 1, 2, 3, 6, 7, and 8 had lower MAP at most of the locations measured along the PA (Fig. 2B). The PP was higher than average in horses 3, 4, 5, 7, 9, and 10, whereas three horses (1, 2, 6) had PP below average in all locations measured along the PA. The PP of horse 8 fluctuated above and below average along the PA (Fig. 2C). On post mortem examination, macroscopic lesions were noted in the pulmonary artery trunk and/or proximal main branches of 4/10 horses. The lesions noted were raised plaques, firm on palpation, which did not diffuse through the endothelium. The size of the lesions varied from focal lesions of 1-mm diameter to diffuse 3 cm × 4 cm plaques (Fig. 1A). Upon histological analysis, three horses were classified with normal PA trunks, five with mild lesions, one with moderate lesions, and one had severe lesions. In the left pulmonary arteries, one horse was classified as normal, four horses had mild lesions, two had moderate lesions, and three had severe lesions. In the right pulmonary arteries, four horses were classified as normal, one had mild lesions, three had moderate lesions, and one had severe lesions (Table 1). Horses 5, 9, and 10 had moderate to severe histological lesions, as well as higher PWV, MAP, and PP in distal PA branches. The remaining horses had histological lesions that ranged from normal to moderate in the PA trunk and main branches. DISCUSSIONPulse wave velocity of the main pulmonary arteries was measured in sedated standing horses. A dual PSC was successfully placed deep within the lung vessels in all horses, and pressure waves were recorded for PWV calculations. Placement of a pressure sensor catheter within the PA via right heart catheterization (RHC) is a well-described procedure in horses and is considered to be the gold standard to determine pulmonary artery pressures (20, 22). An alternative to using pressure and geometric measurements of the artery to determine its stiffness is calculating the velocity of a pulse wave traveling through the vessel. This relation is explained by the Moens-Korteweg equation: PWV = √Eh/ρD, where E is the wall elastic modulus, h is the wall thickness, ρ is the blood density, and D is the vessel diameter. Considering that Eh is equivalent to stiffness, from this formula, it can be inferred that PWV increases proportionally with stiffness (16). Several techniques are available to determine PWV in superficial peripheral arteries. However, reports on pulmonary artery PWV are scarce even in humans and include either MRI studies (29) or RHC for arterial pressure measurements (15). Considering the size of a mature horse, the current techniques available to assess arterial stiffness are not suitable for horses, and therefore, right heart catheterization and arterial pressure measurement were the best alternative. Calculation of pulse wave transit times is essential to determine PWV. The foot-to-foot technique, in which a point of minimum pressure in the diastolic portion of the waveform is used (e.g., the point of minimum pressure immediately preceding systole), is the technique most commonly used by clinicians to determine pulse wave transit times (24). Our research group compared several methods to calculate PA-PWV in horses and determined SPO technique as a reliable and robust (31), and therefore, it was used to calculate pulse wave transit times in this study. The transient self-limiting cardiac arrhythmias noted during catheter placement and manipulations are not unexpected. These arrhythmias have been reported in horses with RHC (23) and commonly resolve spontaneously after catheter removal, and few cases require treatment (10). Since these arrhythmias were self-limiting and no complications were encountered with RHC in this study, this procedure appears to be safe and well tolerated by the horses. The average pulse wave velocity was higher at the proximal pulmonary artery trunk (2.3 m/s) and then gradually decreased (1.1 m/s) toward the distal vascular tree, as expected in a tubular structure with walls that deform under the pulse pressure (16). This pulmonary artery PWV in horses compared with those reported by Kopeć et al. (15) for healthy human adults (1.9–4.0 m/s). Reference values for PWV in the systemic or pulmonary circulation in healthy horses could not be found; therefore, our results require further confirmation in a larger population. The average MAP obtained (30 mmHg) was comparable to previous reports for PAP in horses at rest (8, 19). Similarly, the average pulse pressures recorded in the main pulmonary artery trunk (15 mmHg) were also consistent with those previously reported (21). This was considered an indication that the methods used to measure the PA pressures and to locate the PSC in the arterial trunk were accurate. In general, pulmonary artery stiffening is mediated by three different mechanisms: 1) modification of the intrinsic material properties of the arterial wall; 2) active contraction of the smooth muscle cells and myofibroblasts in the arterial wall; and 3) alteration of the stiffness of the wall due to the operating conditions (increases in blood pressure), which result in increased arterial dilation (16). Increased wall stiffness due to increased dilating pressure occurs because elastin and collagen support the wall at different stretch levels, and the two have different elasticities. Therefore, when the wall is at a low stretch level, it is mainly elastin exerting its effect, while at a higher stretch level, it is collagen, a stiffer structural protein, which supports the wall elasticity (16). The MAP is the most significant physiological variable affecting arterial stiffness (34). In humans, to prevent interference with arterial blood pressure, several conditions should be met for a more standardized and accurate measurement of PWV. The measurements should be performed in a quiet environment, after the patient has been resting in a supine position for at least 10 min, with no food or caffeine intake during the 3 h before the measurements, and being aware of “white coat syndrome” effects (35). The horses used in our study were not preconditioned to the research facility rooms (stocks, radiology room, stalls) or instrumentation, and some of the younger horses (horses 4, 5, 7, and 8) had been minimally handled before this study, which may have also influenced their level of stress. This subject variability, together with the optimization of the protocol and quicker data collection, led to the use of different sedation protocols. To minimize the potential effects of sedation in the arterial pressure, a shorter-acting sedative was used (xylazine bolus alone), and a washout period (20–30 min) was allowed before data collection in some horses. The use of shorter-acting sedation, followed by a washout period had the desired effect, and these horses were awake at the time of data collection. However, this study was not designed to evaluate the effect of the different sedation protocols in the MAP, and therefore, no conclusions can be drawn regarding which sedation protocol might interfere less with PWV measurements. Although this study was not designed to compare groups with potentially different arterial stiffness, we observed that most horses with PWV above the average were older than 10 yr (horses 1, 9, and 10). The aortic PWV of humans subjects of 20 yr of age is ~8 m/s, but increases to 13.5 m/s by the age of 80 (3). Aging may play a passive role as a mechanism of arterial stiffness due to the continuous accumulation of repeated mechanical stress cycles to which the elastin fibers are exposed throughout life, causing elastin fiber degradation (2, 9, 25). However, horse 5 (5 yr-old) had a high PWV (3.6 m/s), whereas horse 6 (16 yr old) had a low PWV (1.7 m/s), which suggests that other factors aside from age could affect vascular stiffness and, therefore, PWV in the PA of horses, as occurs in humans. The changes in intrinsic material properties of the arterial wall may be caused by abnormal deposition or cross-linkage of elastin or collagen (14, 16, 17), or eventually medial calcification (7), which likely occurs in the PA of horses. Histological analysis revealed that most horses (9/10) in this study had some degree of histological lesions in the pulmonary artery walls, with 7/10 animals having lesions classified as moderate to severe. On the basis of previous observations, this finding comes as no surprise (1, 12). The changes in intrinsic material properties of the arterial wall may be caused by abnormal deposition or cross-linkage of elastin or collagen (14, 16, 17), or eventually medial calcification (7), which may be observed in the PA of horses. Hypertension, ongoing systemic inflammation, hyperglycemia and hyper-insulinemia, and changes in the calcium phosphate homeostasis may contribute to the arterial wall remodeling observed in the human arterial walls (7, 13, 34). In our study, although all horses appeared healthy on general clinical examination, their previous medical history was unknown, and so the cause(s) of the arterial lesions observed could not be investigated. Interestingly, the only three horses with severe histological lesions (horses 5, 9, and 10; there was no PWV data for horse 3), were also the only horses that had simultaneously higher than average PWV and MAP in the proximal PA and higher PP in the distal PA branches. An association between increased PWV, MAP, and PP and more severe histological lesions is suspected in these cases, as increased arterial stiffness may lead to increased MAP and PP, and vice versa (11). However, considering the small sample and multiple potentially confounding factors that were not accounted for, this observation remains to be tested. This study described an effective technique for catheterization and data collection to determine PWV in the pulmonary arteries of standing horses. There were multiple limitations and variable factors encountered, or unaccounted for in the study design, and, therefore, no conclusions were drawn regarding our secondary objective. Nonetheless, these confounding factors were identified and should be addressed in future studies. In summary, PWV was measured for the first time in the pulmonary arteries of standing sedated horses, without major adverse events secondary to RHC. Because of confounding factors mentioned above, the suspected association between the presence of histological arterial lesions and changes in PWV could not be confirmed at this time. Nevertheless, this study may be the cornerstone to understanding the potential role that lesions like arterial medial calcification may play in the pathogenesis of pulmonary vascular diseases. GRANTSThe study was supported by Equine Guelph Grant 051647. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSG.T.dA.S., B.B.G., and L.G.A. conceived and designed research; G.T.dA.S., B.B.G., D.E.G., and L.G.A. performed experiments; G.T.dA.S. and M.M. analyzed data; G.T.dA.S. and L.G.A. interpreted results of experiments; G.T.dA.S., D.E.G., and L.G.A. prepared figures; G.T.dA.S. drafted manuscript; G.T.dA.S., B.B.G., D.E.G., L.V., M.L.O., J.R., and L.G.A. edited and revised manuscript; G.T.dA.S. and L.G.A. approved final version of manuscript. REFERENCES
Page 5vascular control across different organs subserves a range of primary requirements from thermoregulation in the skin to blood filtration in the kidney and support of cellular energetics in the heart, skeletal muscle, and brain. In each organ blood flow control must be regulated in accordance with the required function within an error tolerance presumably dictated by the extent of damage incurred to the organ or organism by hypoperfusion. The brain does not endure the up to 100-fold increases in oxygen demand (V̇o2) incurred by vigorously contracting muscle(s). However, it supports extremely high oxidative function at rest and this increases in response to the neuromuscular activation requirements of physical exercise (15, 25, 34, 36, 37, 43, 49, 52, 61). Unlike skeletal muscle, the brain is notably lacking in energy storage mechanisms (28, 58, 70) and increases its fractional oxygen extraction to ~50% rather than the 80–90% found in skeletal muscle (25, 51). Thus, once thought to be constant (41), there is now substantial evidence that cerebrovascular blood flow increases during exercise (15, 20, 25, 27, 32, 34, 36, 37, 43, 52, 61). Moreover, that increase is highest, and laterally symmetrical, for cycling at moderate exercise intensities but reduced for vigorous and maximal intensities where hyperventilation and hypocapnia induce cerebral arteriolar vasoconstriction (25, 31, 43, 49, 52, 61, 69). Given the brain’s lack of O2 stores, and intolerance to anoxia, that elevated blood flow and oxygen delivery must be rapidly and precisely matched to the brain’s metabolic requirements. Whereas quantitative steady-state responses of blood flow, ventilation, or V̇o2 can provide valuable information, resolution of the response kinetics to a change in demand from rest to exercise, for example, can uncover sentinel features of the underlying control mechanisms. Such analyses are technically and mathematically challenging but also extremely rewarding. For instance, kinetics analyses of pulmonary and muscle V̇o2 have demonstrated that, in health but not disease (e.g., heart failure, type II diabetes, and chronic obstructive pulmonary disease), the speed of the kinetics response at exercise onset is limited by mitochondrial function and not O2 delivery (5, 7, 51). At the arteriolar level, kinetics analyses of vasodilation (and vasoconstriction) have extended these findings and explain how O2 delivery temporally and quantitatively matches oxidative demands in healthy young but not aged muscles (6, 39, 40, 54). Similar analyses for ventilation have unveiled how the carotid body facilitates ventilatory dynamics and blood-gas and acid-base regulation across metabolic transitions (67). It is surprising therefore that cerebrovascular kinetics have not been studied to date. The exercise hyperemia of the brain has been demonstrated with anterior and middle cerebral artery blood velocity (MCAV) (15, 34, 36, 37, 43) and “steady-state” cerebral blood flow increasing by 10–30% following the onset of low- and moderate-intensity exercise (21, 35, 36). It is also thought that life-long physical activity results in greater cerebral blood flow to support brain tissue, possibly through angiogenesis and improved cerebrovascular responsiveness to demand (1, 4). In this regard, cerebral endothelial nitric oxide synthase levels are increased following an exercise training program in rodents (48, 71) supporting that chronic exercise elevates cerebrovascular control as seen for the coronary and skeletal muscle vasculatures (42). It is not known whether such training improves either the speed or amplitude of the hyperemic response of brain blood flow in response to physical exercise or alternatively whether that response is altered or compromised by age and or diseases such as Alzheimer’s, stroke, diabetes, or heart failure, in part, because kinetics analyses of cerebrovascular blood flow have not been conducted. The MCA was chosen as the vessel of interest for assessing the kinetics of cerebrovascular regulation since previous work has focused on the MCA during exercise (1, 21, 35, 36, 55, 62). Here we report the use of a novel analysis method for assessing the kinetics of cerebrovascular regulation to provide the first characterization of the dynamic cerebrovascular [specifically, the middle cerebral artery blood mean velocity, mean MCAV (termed MCAv throughout)] response to exercise. Specifically, we tested the hypotheses that 1) following the onset of exercise MCAV increases exponentially; 2) the parameters of this response (amplitude, time constant, and mean response time) are not systematically different between right and left MCA, and 3) the kinetics profile and kinetics parameters may be substantially altered by age and disease (stroke). Specifically, we demonstrate the putative application of this technique for characterizing differences across age and disease states with the future purpose of identifying and testing therapeutic strategies for better preserving, or recovering, cerebrovascular and, potentially, brain function. MATERIALS AND METHODSTen healthy young adults with low cardiac risk (50) and three older adults without cardiovascular disease but considered high cardiac risk, in part, because of their age (50) were recruited to participate in the study. One additional individual, 88 days postischemic stroke, was recruited as a case example of known cerebrovascular injury. No individuals were considered to be competitive athletes (Table 1 contains participant demographics). The Kansas State University (KU) Medical Center Human Subjects Committee approved all experimental procedures, which complied with the Declaration of Helsinki. Institutionally approved written informed consent was obtained from each individual before participation in the study. We did not directly assess hormone level, but premenopausal females exercised around the early follicular phase of the menstrual cycle (21, 60). Female participants over 65 were considered postmenopausal.
Participants were screened either over the phone or in person. Inclusion criteria were 1) 20–85 yr of age, 2) ability to perform repeated bouts of moderate-intensity exercise, and 3) transportation to KU Medical Center for testing. Exclusion criteria were 1) inability of study staff to acquire a signal of the MCA using transcranial Doppler ultrasound; 2) inability to perform the alternating leg movements on the seated recumbent stepper (T5XR NuStep, Ann Arbor, MI); 3) diagnosis of Parkinson’s disease, mild cognitive impairment, Alzheimer’s disease, or multiple sclerosis; and 4) pulmonary disease or dependency on supplemental oxygen. Before reporting to the laboratory at KU Medical Center, participants were asked to abstain from the following before testing: food for 2 h before testing (21), caffeine for at least 6 h, and vigorous exercise for 12 h. The laboratory room for the experimental session was dimly lit, quiet and temperature maintained between 22 and 24°C (11, 12). External stimuli were kept to a minimum. After written informed consent was obtained, resting heart rate (HR) was taken by a handheld device (Tuffsat Ohmeda; GE Healthcare, Chicago, IL). Then, the participant was familiarized with the equipment and procedures. All participants were instructed to breathe only through their nose during the experiments. A nasal cannula was placed in the participants’ nares, and, if needed, adjustments to the position of the nasal cannula were made to ensure optimal end-tidal carbon dioxide (PETCO2 in mmHg) reading (BCI Capnocheck Sleep 9004; Smiths Medical, Dublin, OH). The nasal cannula remained in place at rest and during exercise familiarization to allow the individual to practice breathing through their nose. During testing, participants were observed closely to ensure breathing was exclusively through their nose. Participants practiced the reciprocal motion of the recumbent stepper at the prescribed rate of 120 steps/min. The recumbent stepper was used for this study since it is the modality of choice for older adults (44) and is often used with those after stroke due to motor and balance impairments (8–10, 13). Next, a target work rate was identified by setting the resistance to 40 W and then increasing at a rate of 10 W every 30 s until their target HR for moderate-intensity exercise was achieved and maintained for 1 min. Moderate-intensity exercise was defined as 45–55% of HR reserve calculated using the Karvonen formula and age-predicted maximum heart rate of 220 age (50). Work rates estimated to be in the upper region of the moderate-intensity domain were specifically selected because we wished to evoke the greatest increase in MCAV (25, 31, 43, 49, 52, 61, 69). The lactic acidosis and consequent hyperventilation associated with vigorous and maximal intensity exercise were avoided because hypocapnia induces cerebral vasoconstriction and reduced cerebral blood flow (69). After the target work rate was found, study staff then completed the American College of Sports Medicine cardiac risk stratification (50), nonexercise V̇o2max estimate questionnaire (38), participant demographics, and information pertaining to past medical history and physical activity participation (~20 min) allowing the HR to return to within five beats of resting. Three laboratory members conducted the experimental sessions. One individual read instructions to the participant using a standardized script, the second team member monitored beat-to-beat blood pressure (Finometer; Finapres Medical Systems, Amsterdam, The Netherlands) and PETCO2, and the third team member set up and monitored the transcranial Doppler ultrasound (TCD) (Multigon Industries, Yonkers, NY), electrocardiogram (ECG) (Cardiocard; Nasiff Associates, Central Square, NY), and data acquired through an analog-to-digital data acquisition unit (NI-USB-6212; National Instruments) and custom-written software operating in MATLAB (v2014a; The Mathworks, Natick, MA). Participants were seated in the semirecumbent stepper. HR was monitored continuously using V5 on the ECG. The left arm was placed on a padded table and was adjusted to ensure the arm remained at the level of the right atrium (21) and continuously monitored to ensure minimal movement during the experiment. Beat-to-beat blood pressure was acquired from the left middle finger using a finger photoplethysmograph (Finometer PRO; Finapres Medical Systems). Right arm brachial artery blood pressure was assessed with the arm at heart level using an automated sphygmomanometer with microphone (Tango M2; Suntech, Morrisville, NC). This allowed for comparison between devices to ensure accurate blood pressure measures before data collection (21). Middle cerebral artery velocity (MCAV) was measured using TCD at rest and during exercise. With the use of an adjustable headband, 2-MHz probes with ultrasonic gel were placed over the cranial temporal bone window (3). The MCA was accurately identified using practice standards for probe positioning and orientation, depth selection, and flow direction (3). Depth and gain settings were adjusted to ensure optimal signal strength and then the probes were fixed in place. During the initial set up, the participants sat quietly for 20 min and were reminded to keep their arms and hands relaxed, to breathe through their nose, and to face forward. The recording period started with 90 s of rest. After 60 s of rest, the participant was informed that exercise would begin shortly. At 84 s, a visual countdown was provided for the participant to begin exercise. We chose to standardize exercise initiation while minimizing the ramp-up time to target intensity. All subjects began exercising at 60% of their target work rate for the subsequent moderate-intensity exercise. We increased the watts at 10-s intervals using one-third the difference of the starting and target watts until the target power was achieved resulting in attainment of the target work rate 30 s into the transition. Subjects maintained this work rate for 6 min and then cooled down for 2 min. The participants rested quietly until the HR reached plus or minus five beats of their resting HR before subsequent exercise bouts. Subjects completed three rest-to-exercise transitions, which were temporally aligned to the start of exercise and then averaged. This procedure improves the signal-to-noise ratio and thus better reveals the underlying physiological response (68). Test-retest reliability for the baseline (BL) and criterion kinetics measurements of response amplitude (Amp), time delay (TD), time constant (τ), and mean response time (MRT, TD + τ) were established for all eight younger participants across the three transitions. All variables were sampled at 500 Hz. To analyze, the data were divided by R-to-R cardiac interval. For each cardiac cycle, mean finger arterial pressure calculated as area under the pressure curve (MAP in mmHg), mean left and right MCAV (cm/s), PETCO2, and HR (beats/min) were calculated. Data with R-to-R intervals greater than 5 Hz or changes in peak blood flow velocity greater than 10 cm/s in a single cardiac cycle were considered not physiologically real and censored. Acquisitions with more than 15% of data points censored were discarded. MCAV, MAP, and PETCO2 were then interpolated to 0.5 Hz using shape-preserving, piecewise cubic interpolation. Kinetics analyses were conducted for the left and right MCA during exercise using 3-s time-binned mean values over the entire exercise bout with a monoexponential model: MCAV(t)=BL+Amp(1−e−(t−TD)/τ) where MCAV(t) is the MCAV at any point in time, BL is the baseline before the onset of exercise, Amp is the peak amplitude of the response, TD is the time delay proceeding the increase in MCAV, and τ is the time constant. Mean response time (MRT) was calculated as the sum of the model derived τ and TD. The total exercising MCAV response (Tot) was calculated as the sum of BL and Amp. Time-to-63% of the steady-state response was assessed as a model-independent measure of the response. Specifically, this measurement provides a nonbiased check of the model fitting without making any assumptions regarding the temporal profile of change.All curve fitting and statistical analyses were performed using a commercially available software package (SigmaPlot 12.5; Systat Software, San Jose, CA). Differences in resting values, exercise responses, and kinetic parameters were analyzed using Student’s paired t-tests. Differences between younger and older participants were analyzed using Student’s unpaired t-tests. Normality was verified via the Shapiro-Wilk normality test. Differences were considered significant when P < 0.05. The limits of agreement between left and right MCAV variables were investigated using the Bland-Altman method (14). Data are presented as means ± SE unless otherwise noted. RESULTSEight healthy young participants (6 female, 2 male) were included in the data analysis. Two subjects were excluded from kinetics analysis because a valid MCAV signal was not acquired from both the left and right MCA. The work rate for the exercise transitions was 104 ± 5 (range: 85–125) W. Consistent with moderate-intensity exercise (45–55% of HR reserve) from rest to exercise, HR increased from 86 ± 5 to 129 ± 6 beats/min (P < 0.01) and MAP increased 8 ± 4 mmHg (PETCO2was increased from rest 36 ± 1 to exercise 42 ± 2 mmHg (P < 0.01). The test retest reliability (n = 8) was high with coefficients of variation for the BL, Amp, and MRT of 3, 14, and 12%, respectively. MCAV increased from rest to exercise in all young subjects with a mean Amp of 15.5 ± 3.1 (range: 9.0–35.7) cm/s for the left and 13.4 ± 2.1 (range: 7.4–25.6) cm/s for the right MCA (P = 0.21) (Table 2). As seen from the exemplar presented in Fig. 1, MCAV increased in a close-to-exponential pattern being well-fit by a delay + exponential function. This notion was supported by inspection of each individual response, analysis of the residuals (Fig. 1, line at bottom), and the high r2 of the fits themselves: left MCAV: 0.85 ± 0.03 (range: 0.73–0.97); right MCAV: 0.82 ± 0.04 (range: 0.6–0.95). In addition, the coefficient of variation of the actual time-to-63% of the steady-state response with the MRT calculated by the model was a mere 5 and 11% for the left (82.7 ± 4.1 vs. 83.3 ± 5.2 s; P = 0.78) and right MCA (81.9 ± 6.0 vs. 82.1 ± 6.5 s; P = 0.98), respectively, again indicating good model fit to the response. The delay was highly variable among subjects being 21.0–76.6 s for the left and 12.1–64.1 s for the right MCA (Table 2). Similarly, τ varied from 13.6–52.8 s for the left and 12.8–53.8 s for the right MCA. No MCAV parameter (i.e., BL, Amp, MRT, etc.) was significantly correlated with work rate or ΔPETCO2 (data not shown).
Fig. 1.Typical middle cerebral artery velocity (MCAV) at rest and response following the onset of moderate-intensity exercise (dashed vertical line, time 0). Notice the close fit (solid curve at top) to the time delay + exponential model as supported by the high correlation coefficient and residuals profile (solid line at bottom). Results from subject 2. Figure 2 presents the best (top) and worst (bottom) model fits from our group of young healthy subjects to emphasize that, despite some variability among subjects, the technique is tenable. Fig. 2.Middle cerebral artery velocity (MCAV) at rest and response following the onset of moderate-intensity exercise (dashed vertical line, time 0). Solid symbols are left MCA, and hollow symbols are right MCA. Top: illustrates subject 10 in whom there was excellent agreement between left and right. Bottom: represents poorest agreement among the young participants, subject 7. MRT, mean response time. Overall left and right MCAV correlation was 0.819 (P < 0.05) with a coefficient of variation of 7.6%. As evident in Table 2 there was a close correspondence of the mean values between the left and right MCA for BL, Amp, TD, τ, or MRT although the coefficient of variation did differ among the parameters (i.e., 10.8% BL, 22.8% Amp, 22.1% τ, 8.7% TD, and 7.6% MRT). Figure 3 demonstrates the close correlation for the overall response (MRT, r = 0.82, P < 0.05) with this conclusion supported by the Bland-Altman plot (Fig. 3, bottom). In addition when the absolute exercising total MCAV was examined during exercise there was no difference between left (84.8 ± 4.8 cm/s) and right (77.8 ± 4.7 cm/s, P = 0.07) with a coefficient of variation of 11.1%. Fig. 3.Top: correlation between left and right middle cerebral artery velocity (MCAV) mean response time (MRT) following the onset of moderate-intensity exercise (n = 8). Overall left and right MCAV correlation was 0.819 (P < 0.05) with a coefficient of variation of 7.6%. Bottom: Bland-Altman plot demonstrating the generally close correspondence between left and right MRTs. Overall left and right MCAV correlation was 0.819 (P < 0.05) with a coefficient of variation of 7.6%. Older adult participants (2 female, 1 male) were considered high cardiovascular risk due to reporting existing signs/symptoms of cardiac disease [ankle edema (n = 2), orthopnea (n = 1)] and the following risk factors: age, sedentary lifestyle, hypertension, and dyslipidemia (50). The work rate for the exercise transitions was 88 ± 7 (range: 70–90) W and trended toward being significantly lower than younger participants (P = 0.06). In stark contrast to the young healthy participants, the older adults had a lower MCAV BL in both the left (43.1 ± 5.2 cm/s, P = 0.002) and right MCA (44.7 ± 7.0 cm/s, P = 0.03) as well as a markedly slowed MRT for the left (117.9 ± 10.3 s, P = 0.009) but not the right MCA (99.4 ± 7.1, P = 0.17). Despite the lower work rate, BL, and slowed kinetics for these three older subjects, the Amp was not reduced significantly in either the left (10.4 ± 2.6 cm/s, P = 0.37) or right MCA (11.8 ± 2.4 cm/s, P = 0.67) compared with the younger participants. An older subject is represented by the open circles in Fig. 4. Fig. 4.Typical middle cerebral artery velocity (MCAV) at rest and response following the onset of moderate-intensity exercise (dashed vertical line, time 0). Solid circles are from a representative young healthy subject, subject 2. Hollow circles are from an older healthy subject (subject 20). Hollow squares are from stroke patient (subject 201) using the ipsilateral MCA. By comparison, note very slow mean response time (MRT) and low amplitude of response in the older subject and absence of any response in the stroke patient. The patient had a right ischemic stroke in the MCA territory. A dramatic response for the right MCAV was the extremely low BL (37.9 cm/s) and absence of any increase in MCAV with exercise (Fig. 4). The signal for the left MCA was not attainable. DISCUSSIONThe principal novel findings of the present investigation are aligned closely with our hypotheses. Namely, following the onset of moderate-intensity exercise, MCAV in young healthy individuals increased, after a time delay, in a close-to-exponential profile to achieve an elevated steady-state within 2–3 min. The MCAV values for BL, the response parameters (Amp, τ, MRT, TD, and Tot), and exercising steady-state were not systematically different between right and left MCA with coefficients of variation between 8 and 23% and high correlation coefficients (up to 0.82 for MRT). The three older healthy individuals exhibited lower baseline and total amplitude of MCAV response following the onset of exercise and notably slower MCAV kinetics, with τ and MRT being well outside the range of their younger healthy counterparts (i.e., 2–4 SD longer than the young average, Table 2). In marked contrast to the healthy individuals, the stroke patient exhibited a baseline that was markedly lower than the healthy young or aged subjects and displayed absolutely no increase in MCAV during exercise. Given the current emphasis on cerebrovascular health in brain aging and disease, understanding the speed of these kinetics responses may have important clinical implications. There is mounting evidence that vascular disease, poor cerebral endothelial function, and cardiac risk factors (hypertension, hyperlipidemia) contribute to vascular dementia and Alzheimer’s disease (17, 19, 26, 53, 59, 65, 66). We may find that resolution of the kinetics response to a change in physical demand such as from rest to exercise can help clarify the interconnection between cerebrovascular health and brain function/dysfunction. The present investigation suggests that there are not systematic differences in response kinetics or Amp between right and left MCA in healthy adults. This finding supports that either the left or right MCA can be used as an indicator of the kinetics response in both arteries during bilateral lower extremity exercise although this notion should be tested in larger and possibly more diverse populations. However, we do not know whether systematic lateral differences exist in the presence of neurologic injury such as stroke. Understanding whether true differences exist in the kinetics response between the ipsi- and contralateral vessels is an important future direction. Herein, we present what we hope will prove to be a sensitive technique for assessing the kinetics of cerebrovascular regulation with exercise (or other interventions). It is anticipated that determination of MCAV kinetics will be especially valuable for investigating cerebrovascular function in those at risk for, or suffering from, neurologic disease. Detecting improvements in cerebrovascular function following therapeutic or exercise interventions especially as these relate to response kinetics as well as amplitude would be a powerful capability for refining/assessing such interventions objectively. That said we recognize important limitations that must be considered. First, we are unable to measure changes in MCA diameter. The assumption of constant MCA diameter is critical for MCAV to be used as a direct proxy for cerebral blood flow in the absence of direct diameter measures. Prior work has not permitted consensus as to whether MCA diameter changes with exercise or not (16, 29). Whereas there are some reports that MCA diameter is invariant under hypercapnic, hypocapnic, and orthostatic challenges (2, 18, 22, 56, 62–64), there is also evidence for MCA diameter changing dynamically with motor activity (23) and visual stimulation (22, 30), although fluctuations may be very modest in larger vessels such as the MCA (22, 30). It is also thought that cerebral vessels may undergo regular oscillation at rest (24). If so, a dynamic challenge such as exercise might be important for improving the signal-to-noise ratio for assessment of cerebrovascular function. Second, the values presented in this investigation represent a small population and therefore should not be considered to span the normal range for cerebrovascular responses in young healthy adults. We provide cardiovascular risk and estimated V̇o2max as descriptive measures of general cardiovascular health. A much broader range of ages and diseases must be characterized with this technique to approach generalizability, as appropriate. Additionally, the inability to demonstrate significantly lower response Amp in the older subject cohort was undoubtedly due to the small subject number and consequently low statistical power. Third, the kinetics parameters reported here must, to some degree, be dependent on the extant experimental conditions and not necessarily indicative of the capacity of the system to adapt to all moderate-intensity exercise challenges. For example, to minimize movement artifact we chose to increase work rate over 30 s rather than measuring the response to an immediate step increase to the intended work rate. It is likely that our kinetics parameters may have been different had we chosen a different forcing function. However, we specifically selected an exercise modality and test that does not compromise cranial stability and thus signal fidelity making them suitable for young healthy subjects as well as their older counterparts and, crucially, many stroke patients. Fourth, changes in MAP and also the partial pressure of arterial CO2 (Pco2) can impact MCAV via alterations of driving pressure and also downstream vascular (arteriolar) resistance, respectively. For the moderate-intensity exercise used herein, MAP increased 8 mmHg (P = 0.08) during exercise. The correlation between ΔMCAV and ΔMAP was not significant with <40% of the MCAV variance being potentially explained by MAP. With respect to PETCO2, it is important to note that PETCO2 is not the same as arterial Pco2 and the increase from 36 to 42 mmHg (P < 0.01) is expected to arise from the altered breathing pattern and rate of CO2 evolution rather than altered arterial Pco2. Indeed, for moderate-intensity exercise a substantial amount of literature supports that humans regulate their arterial Pco2 ~40 mmHg (33). As for MAP, there was no significant correlation between changes in PETCO2 and MCAV. For our experimental purposes, it was important to track PETCO2 to ensure that the subjects did not hyperventilate (i.e., drive down PETCO2) due to excitement, nervousness or some other nonspecific ventilatory responses, which would have induced cerebral arteriolar vasoconstriction. As the PETCO2-arterial Pco2 difference is inversely related to breathing frequency and directly related to tidal volume and CO2 output, for the exercise intensity herein the work of Jones et al. (33) supports that an increase in PETCO2 of 3–5 mmHg would be expected in the absence of altered arterial CO2. Current work in healthy adults considers that moderate-intensity exercise may be beneficial for motor learning (57) while other studies report high or vigorous intensity results in improved performance (46, 47). If the premise for improving motor learning is related to increased cerebral blood flow, we need to consider what exercise intensity, duration, and frequency evoke an optimal cerebral blood flow response for healthy individuals and those with brain-related pathology (45). As mentioned previously, there is evidence that moderate exercise intensities do increase cerebral blood flow whereas cerebral arteriolar vasoconstriction reduces blood flow at higher exercise intensities due to the arterial acidemia and consequent peripheral chemoreceptor-mediated hyperventilation and hypocapnia (25, 31, 43, 49, 52, 61, 69). Future work should include the MCAV kinetics response in addition to Amp and consider how they relate across specific age groups and chronic health conditions and whether there might be gender-related differences. In conclusion, we have described a method for assessing the kinetics of cerebrovascular response to exercise. We present evidence for symmetrical vascular responses in young healthy subjects and preliminary data supporting notable differences in the MCAV response kinetics parameters related to age and cerebrovascular damage. With the growing interest in exercise benefits for the brain, ecologically valid methods for measuring cerebrovascular response kinetics across the transition from rest to exercise may become an established investigative technique. Future work should explore the reliability, sensitivity, and specificity of this technique across populations of interest with a view to determining the potential efficacy of therapeutic interventions designed to improve cerebrovascular health and neurological function. GRANTSS. A. Billinger was supported in part by National Institute of Child Health and Human Development (NICHD) Grant K01-HD-067318. J. F. Sisante and S. J. Kwapiszeski were supported in part by NICHD Grant T32-HD-057850. E. D. Vidoni received partial support from the University of Kansas Alzheimer's Disease Center (P30AG035982). D. C. Poole and J. C. Craig were supported, in part, through National Heart, Lung, and Blood Institute Grant HL-2-108328. The Georgia Holland Research in Exercise and Cardiovascular Health (REACH) laboratory space was supported by the Georgia Holland Endowment Fund. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSS.A.B., S.J.K., E.D.V., and R.M. conceived and designed research; S.A.B., J.C.C., S.J.K., J.-F.V.S., E.D.V., R.M., and D.C.P. analyzed data; S.A.B., J.C.C., R.M., and D.C.P. interpreted results of experiments; S.A.B., J.C.C., S.J.K., E.D.V., and D.C.P. drafted manuscript; S.A.B., J.C.C., J.-F.V.S., E.D.V., and D.C.P. edited and revised manuscript; S.A.B. approved final version of manuscript; J.C.C., E.D.V., and D.C.P. prepared figures; S.J.K. and J.-F.V.S. performed experiments. We thank Anna E. Mattlage, Shayla Murphy, and Kelsy Schoen for assistance with script writing and data collection. Thank you to the individuals who supported the Walk Across Kansas and raised funds to purchase equipment used in this project. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). REFERENCES
Page 6charles sherrington was the first to trace the source of the stretch reflex to muscle spindles and subsequently describe reflexes as a simple expression of the interactive action of the nervous system (48). Since then, reflex excitability has been probed by delivering transient tendon taps or electrical nerve stimuli and observing the muscle response (for reviews, see Refs. 42, 51, 53). Reflex examinations have enhanced our understanding of the nervous system and have proven beneficial for diagnosing and monitoring rehabilitation efforts for many disorders associated with spasticity or other reflex abnormalities, including stroke, polyneuropathies, spinal cord injury, and cerebral palsy (21, 34, 39, 45, 52). There are, however, some limitations associated with the assessment of reflex pathways using transient stimuli (mechanical or electrical). Generally, many stimuli need to be delivered to obtain a reliable average response (24, 40); this can be somewhat inefficient and multiple abrupt stimuli permit the influence of anticipation and habituation effects (19). Furthermore, due to the large and relatively synchronized volleys of Ia afferent input generated by a tap or nerve stimulation (6), sufficient recovery time must be left between stimuli to allow postactivation depression to dissipate (10, 25). Limitations of previous methods become particularly salient when studying reflex actions in postural muscles during tasks in which they are actively engaged such as standing and walking. During standing, additional recovery time likely must be left between stimuli to allow for recovery of balance following the postural perturbation of each stimulus. Triceps surae reflex responses are also dependent on postural sway, with increased H reflex responses observed during anterior sway and decreased responses during posterior sway (50). If sway is not accounted for during stimulus delivery, results averaged over a limited number of samples could be skewed or highly variable; this observation is supported by the finding that individuals with greater postural sway demonstrate poor reliability in reflex amplitude (44). Suprathreshold white noise stimulation methods have proven effective for assessing the frequency characteristics of human vestibular reflexes in posturally active muscles (stochastic vestibular stimulation; Ref. 13) and could be adapted to mechanical tendon stimulation to circumvent some of the limitations of traditional tendon tap methods. In general, noisy stimuli delivered to a neural system can be used to assess connectivity through estimates of the amount of frequency variability in ongoing neural activity that can be explained by the frequency content of the stimulus (18, 28). A suprathreshold noisy tendon vibration (NTV) methodology, along with linear systems analysis, could provide researchers and clinicians with a unique tool to unobtrusively examine reflex excitability in posturally active muscles and gain insight into the frequency characteristics of stretch reflex coupling. Previous research has probed the frequency characteristics of stretch reflexes by applying continuous sinusoidal stretches (at discrete frequencies between 10 and 50 Hz) to upper limb muscles and subsequently measuring the magnitude of EMG modulation and phase lag between the mechanical stretch and muscle response (36). The modulation strength and estimated phase delays provided insight into the operation of reflex circuitry at different frequencies, as well as provided an arguably superior measure of the delays inherent in the reflex pathway (36). The delays measured from the phase estimates take into account the envelope of the response and therefore are more representative of the “average response time of the average unit” (36), whereas the latency measured to the onset of a response evoked by transient stimuli is likely representative of only the fastest conducting axons. Sinusoidal stimuli can also shed light on motorneuron evoked response properties, such as where their response(s) lie within a given stretch cycle and how susceptible they are to coupling with stimuli close to their firing rate to produce a sharp increase in reflex gain (carrier resonance effect; Ref. 37). Although the pure sinusoidal stimulus method imparts numerous benefits, there are several notable drawbacks that include 1) response contamination from voluntary tracking of the predictable stimulus (9), 2) the development of movement illusions or tonic vibration reflex, and 3) the experimental time necessary to test a series of frequencies individually. Stretch reflexes have also been examined using large amplitude, low-frequency (<25 Hz) pseudorandom joint perturbations (27, 31–33). In particular, low-frequency ankle joint perturbations in humans lying prone have been used to identify the ongoing relationship between ankle movement velocity and muscle activity (31, 32). Using pseudorandom stimuli, these authors identified some key differences in reflex organization between the triceps surae and tibialis anterior (31, 32). Other researchers who have used a pseudorandom muscle stretching protocol also chose to apply primarily low-frequency stimuli that are relevant to voluntary movement and within motorneuron firing rate limits (8, 27, 31–33). Human muscle spindles, however, have the capacity to respond and entrain to higher frequency stimuli (exceeding 100 Hz; Ref. 15), and strong reflex EMG modulation can be observed during high-frequency sinusoidal stretching (50 Hz; Refs. 35, 36). In addition, high-frequency components are present in impulses naturally experienced by the ankle joint, for example, during a trip; therefore, these high-frequency components have functional relevance. Thus we focused our experiment on responses to a broadband noisy mechanical stimulus that contained power up to the highest frequency that could evoke a discernable reflex response. Similar to spinal stretch reflexes, transient electrical or mechanical stimuli are typically used to evoke cortical potentials to study the ascending transmission of sensory information (14, 17, 38). Thus these methods of evoking cortical activity are subject to similar limitations as tendon tap or H reflexes, such as the potential influence of anticipation, habituation, and postural interference. Estimates of coherence between noisy peripheral sensory input and somatosensory cortex activity may have the additional benefit of providing a useful alternative approach to probe the frequency characteristics of somatosensory-evoked cortical potentials. The primary objective of our experiment was to explore the use of Achilles NTV to assess the frequency characteristics of triceps surae reflexes and sensorimotor cortex-evoked potentials during standing. We also aimed to examine the scaling of reflex responses to different vibration amplitudes in the soleus and medial and lateral gastrocnemius muscles. Finally, we aimed to establish trial durations that evoke consistent responses and identify whether stretch reflexes evoked by noisy mechanical stimuli are subject to habituation over 2 min of stimulus exposure. METHODSEight healthy young adults (age = 27 ± 5.3 yr, 4 male) free of musculoskeletal and neurological disorders participated. Participants provided written informed consent and all procedures were approved by the University of British Columbia Research Ethics Board. Participants stood on a force plate (OR6–7; AMTI) with their stance width normalized to foot length and their gaze directed onto a visual target positioned at eye level ~3 m ahead. A 3-cm diameter probe, attached to a linear motor (model MT-160; Labworks), was positioned against the right Achilles tendon. The linear motor was secured onto two near-frictionless linear slides and was pulled forward onto the tendon by a weighted pulley system (Fig. 1); this setup was decoupled from the force plate and was able to maintain a constant ~1 N preload force on the tendon. A force transducer (model 31; Honeywell) was placed in line with the probe and motor piston and an accelerometer (model 220–010; X Tronics) was secured to the back of the motor piston. Acceleration and force signals were differentially amplified ( ×1 and ×100, respectively) and low-pass analogue filtered at 600 Hz (Brownlee model 440; AutoMate Scientific). All motor command signals were generated using LabVIEW 11 software and output at 5 kHz from a PXI-6225 multifunctional data acquisition board (running with a PXI-8106 real-time controller in a PXI-1031 chassis). Analogue voltage commands were sent to a motor amplifier (PA-141; Labworks) for open-loop control of tendon stimulation. Fig. 1.Experimental setup showing the linear motor on frictionless slides pulled forward using a weighted pulley system (A) and a zoomed in view of the probe positioned against the Achilles tendon of a participant standing on a force plate (B). Sample profile of the noisy vibration acceleration over time and the acceleration power spectrum (C). SOL, soleus muscle; MGas, medial gastrocnemius muscle; LGas, lateral gastrocnemius muscle. Electromyography (EMG) was recorded from the soleus (SOL), medial gastrocnemius (MGas), and lateral gastrocnemius (LGas) muscles using surface electrodes positioned over the muscle bellies in bipolar arrangement (amplified ×2,000, 10-Hz highpass and 1,000-Hz lowpass filter; NeuroLog NL824 preamplifier and NL820 Isolator; Digitimer). The ground electrode was placed on the lateral malleolus. Electroencephalography (EEG) was recorded across the sensorimotor cortex using scalp ring electrodes placed over Cz (active), Fpz’ (reference), and the right mastoid process (ground) (amplified ×20,000; 1-Hz highpass and 1,000-Hz lowpass filter; GRASS P511 AC amplifier; Astromed). EEG electrode impedance levels were maintained below 5 kΩ. Ocular and facial muscle artifacts were identified and displayed to participants, and participants were subsequently instructed to minimize these behaviors during trials. Forces and moments from the force plate were amplified (×1,000–4,000) and sampled at 100 Hz, while motor voltage commands, probe acceleration and force, surface EMG, and EEG signals were sampled at 2 kHz (Power 1401 A/D board and Spike2 software; Cambridge Electronic Design). Participants completed a 2-min quiet stance trial and mean foot center of pressure (COP) positions were calculated in the mediolateral and anteroposterior directions to determine their neutral position. Participants began subsequent trials at their neutral position, and in the case that their COP drifted, experimenters provided verbal feedback to guide them back to neutral. Four 2-min trials of NTV were conducted where a white noise signal low-pass filtered at 100 Hz was delivered to the right Achilles tendon. In the recorded probe acceleration, power below 10 Hz and above 115 Hz was less than or equal to −13 dB (ref. peak plateau of power spectrum); a sample recorded acceleration profile and power spectrum are displayed in Fig. 1C. This stimulus bandwidth was chosen based on pilot data that demonstrated the maximum frequency that could contain significant coherence was ~90 Hz even when larger stimulus bandwidths were delivered (e.g., 10–300 Hz). The NTV was delivered at four different amplitudes (vibration root mean square accelerations: 5, 10, 15, and 20 m/s2); these amplitudes were also chosen based on pilot data that suggested these amplitudes would fall approximately within a steep, ascending portion of the reflex recruitment curve. Two additional trials of 20 tendon taps (30 Hz raised-cosine bell curve pulses at 25 m/s2; 8- to 12-s interstimulus interval) were conducted to compare the temporal characteristics of the responses elicited by noisy stimulation to the responses elicited by taps. The tendon tap and NTV trials were presented in block-randomized order. Forces and moments from the force plate were digitally low pass filtered at 10 Hz (5th order dual pass Butterworth filter) and COP was calculated from moments (Mx and My) and vertical force (Fz). For tendon stimulation and quiet stance trials, the frequency spectra of COP in the anteroposterior and mediolateral directions was calculated (frequency resolution 0.0122 Hz) and mean power frequency (MPF) were determined as: MPF=∑j=1nfjPj/∑j=1nPj where f is frequency and P is power.EMG, EEG, force, and acceleration data were digitally low pass filtered at 1,000 Hz (5th order dual pass Butterworth filter). For the tendon tap trials, EMG data were full wave rectified and EMG and EEG signals were trigger averaged to the tap stimulus onset within a window from 20 ms preceding to 300 ms following the stimulus. COP was also trigger averaged within a window from 0.5 s preceding to 4.5 s following the tap. For noisy stimulation trials, EMG data were full wave rectified and coherence analysis was performed using the NeuroSpec2.0 software package developed by Rosenberg, Halliday, and colleagues (20, 47) for MATLAB (Mathworks). Our approach was similar to that of previous research conducted to establish the time and frequency characteristics of vestibular responses elicited by stochastic stimuli (11–13). To determine the strength of the linear association between two signals in the frequency domain, coherence functions were calculated between probe acceleration (input signal) and rectified surface EMG of each muscle and EEG (output signals) (13, 20, 47). Coherence was calculated as the magnitude of the input-output signal cross spectra squared divided by the product of the input and output autospectra (13). Thus coherence values provide normative estimates of the frequency coupling strength between two signals. To identify temporal characteristics of coherent frequencies, cross covariance was calculated using the inverse Fourier transform of the input-output signal cross spectra and normalized by the product of the vector norms of the input and output signals. Therefore, cross covariance values are bounded by −1 and +1 and provide an estimate of the signal coupling strength in the time domain (12). Our convention was that acceleration toward the tendon and increased (rectified) EMG were assigned positive polarities. For example, a positive correlation would represent acceleration into the tendon is associated with increased EMG, or acceleration away from the tendon is associated with decreased EMG. As described by Halliday et al. (20), 95% confidence limits for coherence (positive threshold) and cross covariance (positive and negative thresholds) were constructed under the hypothesis of independence between the two signals. Values exceeding these limits provide evidence of a significant linear relationship between the stimulus and response. Phase was also estimated to infer the temporal relationship between the NTV and EMG at all frequencies containing significant coherence (1). To analyze pooled responses, data were concatenated across participants and stimulus amplitudes to compare between the three muscles; this yielded a total of 3,712 disjoint sections (1.024 s/segments; frequency resolution = 0.9765 Hz). Data were also concatenated across participants and muscles for comparisons between stimulus amplitudes; this yielded a total of 2,784 disjoint segments (1.024 s/segments; frequency resolution = 0.9765 Hz). For analysis of individual 2-min trials, data were sectioned into 116 segments to obtain a frequency resolution of 0.9765 Hz (1.024 s/segments). The number of segments used in the analysis is an important factor in determining the 95% confidence limits constructed around coherence and cross-covariance traces. In addition, we divided each 2-min trial in various ways to answer two questions: 1) what is the minimum trial duration required to obtain reliable reflex measures, and 2) does the reflex response habituate over the trial? To answer the first question, 10-s portions of data (9–10 segments) were incrementally added and normative error in peak-to-peak cross covariance was calculated for each duration as: error=∑(|C120|−|Cn|)2C120×100% where C120 is cross covariance for the full trial duration and Cn is cross covariance for different trial lengths between 10–110 s in 10-s increments (4). Since our segment size was 1.024 s, each 10-s addition of data increased the number of segments by either 9 or 10 segments because incomplete segments were removed from the analysis. Signal to noise ratios were also calculated for each trial duration as peak-to-peak cross covariance divided by the width between the 95% confidence limits. To answer the second question, coherence and cross covariance were calculated and compared between the first ~40 s (39 segments, 39.936 s data used) and the last ~40 s of the trial. Comparisons of different trial durations were conducted for the SOL muscle in response to the 10 m/s2 RMS acceleration noisy stimulus. We elected to probe trial duration and habituation effects in SOL since it is the most commonly studied lower limb muscle for reflex testing, and we chose the medium-low level NTV based on results that showed this subtle stimulus level evoked strong responses across all participants.To examine if the tendon stimulation affected the frequency content of postural sway, we performed a one-way repeated-measures ANOVA (5 levels: 4 vibration amplitudes and no vibration) on anteroposterior and mediolateral COP MPF. To examine how reflex responses scaled with stimulus amplitude, we conducted a two-way (muscle × NTV amplitude) repeated-measures ANOVA on peak-to-peak cross covariance. Significant ANOVA effects were followed up with Fisher least significant difference post hoc comparisons. Pooled subject data were concatenated across stimulus amplitudes and a χ2 extended difference of coherence (DOC) test was performed to determine whether the independent coherence estimates significantly differed between muscles (1). Similarly, pooled subject data were concatenated across muscles and a χ2 extended DOC test was performed to determine if coherence significantly differed between stimulus amplitudes. Significant main effects were followed up with pairwise DOC tests between successive stimulus amplitudes and between each muscle combination. Finally, to determine if the SOL muscle response to NTV habituated throughout the trial, we compared peak-to-peak cross covariance in the first 40 s of the trial to last 40 s of the trial using a paired t-test. Effects were considered significant at an α-level of 0.05, all error bars demonstrate standard error (n = 8). RESULTSNo participants reported illusory movements in response to the NTV (even when asked to close their eyes) nor did they report any noticeable interference with standing balance. Participants naturally maintained their COP around their neutral position and verbal feedback to correct postural drift was only necessary for two subjects. MPF of COP in both the anteroposterior and mediolateral directions were not affected by the noisy vibration (P = 0.583 and 0.773, respectively; Fig. 2). Perturbations were observed in participants’ COP traces in response to tendon taps; triggered-average anteroposterior COP demonstrated that the taps evoked a directional postural response that required several seconds for recovery (Fig. 2C). Meanwhile, the noisy stimulus did not produce any noticeable change in COP relative to quiet stance. Fig. 2.Mean power frequency of center of pressure (COP) in the anteroposterior (A) and mediolateral (B) directions across the four stimulus amplitudes and quiet stance (0 stimulus) trials. Representative traces of anteroposterior (AP) COP during the noisy tendon vibration, quiet stance, and tendon tap trials, as well as a representative trace of AP COP trigger-averaged to the tap stimulus (C). In all participants, significant (exceeding 95% confidence limits) muscle responses were observed in the SOL and MGas EMG in the frequency (coherence) and time (cross covariance) domains for all NTV stimulus amplitudes. Responses were characterized by a significant coherence band generally within ~10–80 Hz (Fig. 3), with the peak response observed at ~40 Hz in all muscles (SOL 36 ± 4 Hz; MGas 39 ± 5 Hz; LGas 47 ± 9 Hz). Background activity in the LGas muscle during standing was low or absent in the majority of participants, and significant reflex responses were not always observed in LGas EMG, particularly at the lower levels of stimulation (absent in 50% of participants). For all three muscles, the slope of the phase estimate was generally linear, indicating a fixed reflex delay across frequencies (Fig. 4). The magnitude of the slope corresponded to approximately a 40-ms delay. There was, however, a small upward deflection in the phase estimates, accompanied by a reduction in the coherence strength, at ~20 Hz. Fig. 3.Representative traces of stimulus triggered-average muscular and cortical responses to tendon taps (A); data are shown for both unrectified and rectified EMG. Representative traces of coherence and cross covariance between the stimulus acceleration and triceps surae EMG and sensorimotor cortex EEG for the 10 m/s2 noise stimulus (B). Horizontal lines indicate 95% confidence intervals. SOL, soleus muscle; MGas, medial gastrocnemius muscle; LGas, lateral gastrocnemius muscle. Fig. 4.Coherence (A) and phase estimates (B) for each muscle for data pooled across participants and stimulus amplitudes. SOL, soleus muscle; MGas, medial gastrocnemius muscle; LGas, lateral gastrocnemius muscle. The NTV-EMG cross covariance displayed a short-latency biphasic profile with a positive peak followed by a trough. For the SOL muscle, the first peak in the cross covariance occurred at a lag of 38.1 ± 0.8 ms (range = 34 to 42 ms), and similar lag times were observed for the MGas (38.5 ± 1.9 ms) and LGas (38.4 ± 3.1 ms) muscles. Lag times observed in the cross covariance generally corresponded to the latencies of the first peak stimulus-triggered average response to tendon taps (SOL = 41.3 ± 2.4 ms; MGas = 40.5 ± 2.5 ms; LGas = 41.3.7 ± 1.6 ms; Fig. 3). Peak-to-peak cross covariance positively scaled with NTV amplitude (Fig. 5); statistically there was a significant main effect of stimulus amplitude on peak-to-peak cross covariance [F(3,21) = 13.135, P < 0.001], main effect of muscle [F(2,14) = 7.464, P = 0.006], and muscle × stimulus amplitude interaction [F(6,42) = 2.464, P = 0.039]. These results indicate that overall the SOL muscle had stronger coupling with the noisy stimulus and scaled more with increases in the stimulus amplitude. In contrast, LGas demonstrated the weakest coupling and the shallowest rate of increase with stimulus amplitude. Post hoc comparisons revealed significant overall differences in peak-to-peak cross covariance between SOL and LGas at all four stimulus amplitudes (5, 10, 15, and 20 m/s2; P range 0.009–0.016) and between LGas and MGas at the 5 m/s2 (P = 0.041) and 10 m/s2 (P = 0.025) stimulus amplitudes. Fig. 5.Noisy tendon stimulation results for the triceps surae muscles showing increases in mean peak-to-peak cross covariance (A) in response to increases in tendon stimulation amplitude. Sample cross covariance traces (B) and coherence traces (C) from the SOL muscle of one participant showing the increases with stimulus amplitude. SOL, soleus muscle; MGas, medial gastrocnemius muscle; LGas, lateral gastrocnemius muscle. Results for pooled data revealed a general increase in coherence with increases in stimulus amplitude (Fig. 6A). The χ2 extended DOC test demonstrated a significant effect of the stimulus amplitude on coherence at frequencies between ~10 and 60 Hz (Fig. 6B). Pairwise DOC tests indicated significant increases in coherence between the 5 and 10 m/s2 stimulus amplitudes at frequencies ~20–40 Hz, and significant increases in coherence between the 15 and 20 m/s2 stimulus amplitudes at frequencies ~10–30 Hz. Pooled data also revealed generally stronger NTV-EMG coherence for the SOL and MGas muscle compared with the LGas (Fig. 6C). The χ2 extended DOC test indicated a significant effect of the muscle on coherence at frequencies between ~20–70 Hz (Fig. 6D). Pairwise DOC tests demonstrated significantly higher coherence in SOL compared with both MGas and LGas at frequencies ~30–50 Hz and significantly higher coherence in MGas compared with LGas at frequencies ~50–70 Hz. Fig. 6.Results from data concatenated across participants and muscles to demonstrate overall coherence at each stimulus amplitude (A) as well as pooled difference of coherence (DOC) results across stimulus amplitudes, and pairwise DOC results between successive stimulus amplitudes (B). Results from data concatenated across stimulus amplitudes to demonstrate overall coherence for each muscle (C) as well as pooled DOC results across muscles, and pairwise DOC results between each muscle (D). SOL, soleus muscle; MGas, medial gastrocnemius muscle; LGas, lateral gastrocnemius muscle. One participant’s EEG data were excluded due to facial muscle and blink artifacts. Clear stimulus triggered-average-evoked potentials were observed in the EEG recording across the sensorimotor cortex in all remaining seven participants in response to tendon taps. However, significant NTV-EEG coherence was only observed in five out of the seven participants, and coherence was often absent during the lower stimulus amplitudes (5 and 10 m/s2). Significant coherence was observed within a frequency range of ~40–70 Hz (within the γ-band), with the peak coherence located at 54 ± 8 Hz (Fig. 3). The peak EEG response occurred at a slightly longer lag relative to muscle responses, with the trough observed at 50 ± 6 ms and peak observed at 53 ± 6 ms; these latencies generally correspond to early event related potentials from tendon taps in our experiment (peak 47.8 ± 4.5 ms; trough 56.2 ± 8.7) and previous experiments (14). EEG responses were subtle relative to muscle responses and characterized by multiple peaks and troughs. Interestingly, there was no prominent EEG activity in the cross covariance at longer lag times (e.g., ~100 ms and later) that would correspond to later stages of sensory processing. Two-minute trials of the 10 m/s2 NTV were subdivided into 10-s additive sections to determine the minimum trial duration necessary to obtain reasonable reflex estimates in the SOL muscle. With ~10 s of data collection (9 segments), mean normative error in the peak-to-peak cross covariance was high (~40%) and there was a large amount of between-participant variability in normative error (Fig. 7). As expected, confidence intervals and background noise decreased as the trial duration (and number of segments used in the analysis) increased. The decline in the mean and variability of normative error with each addition of ~10 s of data began to plateau at the ~40-s data length, signifying the reflex responses measured with ~40 s of data collection approximated responses measured with 2 min of data collection (10% difference). In addition, the signal to noise ratio increased with each addition of data segments, and this increase was steeper between 10 and 40 s (Fig. 7). The signal to noise ratio approximately doubled between ~10 and ~40 s of data, while further increases in data length up to ~120 s only resulted in an increase in the signal to noise ratio by another 1.7-fold. Fig. 7.Signal to noise ratios (A) and normative error in peak-to-peak cross covariance between each data length and the full 120-s trial length (B). Mean peak-to-peak cross covariance calculated from the first 40 s vs. the last 40 s of the trial (C). There was no evidence that the peak-to-peak cross-covariance amplitude differed between the beginning and end of the trial (P = 0.417; Fig. 7C), indicating no significant habituation to the NTV over 2 min of stimulus exposure. DISCUSSIONThe primary objective of our experiment was to examine the frequency responses of triceps surae EMG, and sensorimotor cortex EEG, evoked by noisy (10–115 Hz) Achilles tendon vibration in standing participants. Our results showed surface EMG was significantly coherent with NTV across a broad frequency range, with strong responses observed in the SOL and MGas muscles. Our results also demonstrate that reasonable SOL coherence estimates can be obtained with minimal perturbation to standing balance, and without habituation effects. We observed significant coherence between the NTV and surface EMG that extended from ~10–80 Hz. This bandwidth falls within the muscle spindle vibration sensitivity range (15), but exceeds the upper limit of human individual motor unit firing rate capabilities (maximal soleus motor unit rates ~20 Hz; Ref. 29). In spite of known homosynaptic postactivation depression and motorneuron afterhyperpolarization effects, high-frequency sine wave vibration bursts (3 cycles at 100 Hz) have previously been shown to generate three distinct reflex responses in SOL surface EMG (16). EMG responses to high-frequency vibration likely reflect that high-frequency spindle input can effectively shift the firing probability of individual triceps surae motor units. The profile of the reflex response evoked by NTV in the time domain exhibited a short latency peak followed by a trough; the lag time between the NTV and muscle response (~38 ms) approximately corresponds to the latency of the T reflex in the lower limb observed in our experiment (~41 ms) as well as in previous experiments (24, 40, 55). Similar to our tap-evoked responses, there was no evidence of responses in the cross covariance at longer delays that could parallel the medium or long latency responses observed in ramp-and-hold stretch reflexes (56). Therefore, we believe our NTV method can be used to characterize the functional short latency reflex coupling between type Ia spindle afferents and lower limb motorneurons. It has previously been shown that an adapted trigger-averaging technique used with sinusoidal stimuli allowed for the identification of a shorter reflex onset latency compared with cross-correlation lag times (30). However, there was no difference between the trigger-averaged latency and cross-correlation lag time measured to the peak of the response (30), as was done in our experiment. This suggests that trigger averaging to either transient or sinusoidal stimuli could provide clearer identification of the earliest reflex onset latency (mediated by the fastest conducting axons) compared with correlational techniques. However, trigger-averaging techniques do not provide important information about the frequency characteristics of responses, which are obtained through estimates of coherence, gain, and phase using the framework developed by Halliday and colleagues (1, 20, 47). The slopes of the phase estimates for each muscle were linear and suggested a relatively conserved stimulus-response delay of ~40 ms across frequencies containing significant coherence. There was, however, a slight upward deflection in the phase slope at ~20 Hz, accompanied by a discontinuity in the coherence plot at ~20 Hz. This 20-Hz phenomenon was observed in all three muscles, although it was more prominent at higher stimulus amplitudes and in SOL. This pattern of a decrease in EMG modulation with sinusoidal stimuli around 20–25 Hz, accompanied by an upward deflection in the phase estimate, has previously been observed in the flexor carpi radialis (FCR) muscle (35). Matthews (35) suggested that this pattern could arise from the interference between two different latency reflex responses (which traveled through pathways that impart different delays), since the sum of the two responses would create a phase intermediate between them. The appearance of a longer latency interfering response in the FCR was suggested to have a functional role in stabilizing the system to tremor or clonus (35). The interference pattern that we observed in the triceps surae muscles at ~20 Hz requires more detailed investigation. The amount of EMG that could be explained by the tendon vibration increased as we increased the amplitude of the NTV, and the overall reflex response and amplitude scaling were strongest in the SOL and weakest in the LGas muscle. DOC tests also demonstrated stronger NTV coupling in the SOL muscle ~30–50 Hz compared with MGas and LGas. Response differences within the triceps surae group could reflect differences in their spindle density, where SOL houses a higher overall number (~400) and density (0.94 spindles/g) of muscle spindles compared with the gastrocnemius muscles (~150 spindles total, 0.4 spindles/g; Refs. 2, 54). SOL muscle spindles might also experience a higher stimulus intensity compared with gastrocnemius spindles due to dampening of the vibration as it travels through tissue. Reflex coupling strengths could also reflect differences in their relative contributions to standing balance, where SOL generally provides the majority of plantarflexion torque while LGas remains relatively silent (23). In addition, the higher proportion of slow twitch fibers in the SOL might favor stronger reflex coupling since animal studies have shown higher efficacy of Ia input to low-threshold motorneurons in a way that accentuates orderly recruitment (22). Across the four stimulus intensities, ~17–34% of the SOL EMG variability could be explained by the tendon stimulation. Despite this strong reflex coupling, there was no notable interference with posture, illusions of forward sway (i.e., illusory muscle lengthening) or presence of a tonic vibration reflex (TVR). Compared with sine wave vibration, the frequency variability of our noisy stimulus (10–115 Hz) seems to impede the generation of illusory movement and TVRs. We speculate that the frequency variability precludes the stimulus from producing a ramping reflex contraction. The absence of any illusory movement or TVR suggests NTV is a more suitable method for the continuous assessment of reflex excitability. Additionally, tendon taps produce a strong unidirectional postural response, whereas noisy stimulation produces a subtle and more bidirectional postural response; thus NTV circumvents some of the limitations of traditional tendon tap methods. The absence of low-frequency content in the NTV likely causes less interference with postural sway, similar to observations from stochastic vestibular stimulation (12). With the use of the RMS 10 m/s2 NTV, reasonably accurate reflex excitability estimates could be obtained with a minimum of ~40 s of data collection; the response amplitudes measured with ~40 s of data collection were 10% different from those measured with 2 min of data collection. T-reflex responses between successive stimuli are inherently variable (24, 40); therefore, our method of sampling the ongoing association between the stimulus and muscle activity over a sufficient window (e.g., 40–60 s) has advantages over traditional approaches (e.g., T and H reflex) that only provide discrete snapshots of reflex excitability. There was no indication that reflex responses to NTV habituated with continuous stimulus exposure over 2 min. Therefore, measurements obtained using this method are not affected by habituation within the intensity levels and durations necessary for reflex testing. Although tendon tap-evoked potentials were present in all participants throughout the experiment, NTV-cortical coherence was absent in two out of seven participants, and in two of the remaining participants it was only prominent during the highest stimulus amplitude (20 m/s2). When significant coherence was present, NTV-EEG peak coherence values were small and scaled with the stimulus amplitude from r2 ~0.04 to 0.07. Our success rate (71%) in extracting NTV-EEG coherence is similar to the success rate previously reported for extracting corticomuscular coherence during voluntary muscle contraction (50–75%; Refs. 41, 43, 46). When present, triceps surae corticomuscular coherence values are also modest (r2 < 0.06; Refs. 41, 43). The absence of strong NTV-EEG coherence could be due to nonlinearities between the stimulus and evoked cortical activity and the absence of low-frequency power to associate with the longer latency cortical-evoked responses. We observed NTV-EEG coherence within the frequency range of ~40–80 Hz, with the peak located at 54 Hz on average. This range corresponds to the γ-band oscillations recorded over the somatosensory cortex induced by mechanical or electrical nerve stimuli (3, 26). Specifically, magnetoencephalography (MEG) recordings have shown that tactile stimulation of the finger evokes γ-oscillations (concentrated around 60–90 Hz) over the sensorimotor cortex ~40–100 ms after stimulus presentation (3). Attention directed toward the spatial features of the stimulus further enhanced these γ-oscillations, which source analysis suggested originated from the primary somatosensory cortex (3). It has been suggested that γ-band oscillations in the somatosensory cortex reflect early stages of processing functionally relevant sensory information and that it is crucial in communicating with other somatosensory areas for higher level processing. The profile of the NTV-cortical cross covariance was oscillatory, and the initial trough lagged the NTV by ~50 ms. This short delay is generally in alignment with the latency of early event related potentials recorded over the somatosensory cortex in response to mechanical stimuli (5, 14, 49). This NTV-cortical lag time specifically approximates the latencies of the Achilles tendon tap-evoked potentials observed in this experiment as well as in previous experiments (14). There are some important considerations with regards to the assessment of reflexes using stimuli applied to the tendon. Tendon tap reflex responses (and likely NTV responses by extension) might not reflect the strength of direct monosynaptic connections between triceps surae spindles and motorneurons for several reasons. First, in addition to targeting triceps surae spindles, tendon taps have been shown to evoke multiple spikes in Ia afferent fibers innervating extensor hallucis longus, tibialis posterior, and intrinsic muscles of the foot (6). Tendon taps have also been shown to alter the discharge of some type II spindle afferents as well as Golgi tendon organ and skin afferents (6). Despite this, however, it appears to be the input from type Ia spindle afferents that accounts for the motor response (6). Second, the rise time of composite excitatory postsynaptic potentials is broad enough to permit time for oligosynaptic pathways to contribute to the response; thus it should be considered that these methods might not strictly assess monosynaptic reflex strength (7). It should also be noted that similar limitations regarding the purity of the afferent stimulus and spinal connections tested are present with direct nerve stimulation (H reflex; Refs. 6, 7). In addition, direct nerve stimulation bypasses natural mechanotransduction and generates very artificial, synchronized nerve impulses. Our noisy stimulation methodology has the advantage of mechanically stimulating receptors themselves within a physiological range and subsequently providing temporal and frequency information about somatosensory projections to muscle and to the cortex. Further exploration of tendon tap reflex responses of single motor units during standing using frequency and probability based measures (e.g., peristimulus time histograms and peri-stimulus frequencygrams), along with coherence between sensory and motor spike trains, is necessary to more fully understand the characteristics of the pathways that contribute to tendon tap and noisy vibration reflexes in humans while standing. Our findings indicate that noisy vibration of the Achilles tendon is an effective novel approach to study somatosensory reflexes in lower limb muscles during standing. Additionally, NTV-muscular coherence can shed light on short latency communication between sensory receptors and the motorneuron pool across frequencies. Our findings also show promise for the use of NTV to concomitantly assess somatosensory related cortical activity, although this requires further investigation. These NTV methods could enhance researchers’ and clinicians’ ability to assess reflexes in posturally active muscles efficiently and with minimal interference with standing balance. GRANTSThis work funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) grants (to M. G. Carpenter, J.-S. Blouin, and J. T. Inglis). J.-S. Blouin also received support from the Canadian Institutes of Health Research-Canadian Chiropractic Research Foundation and Michael Smith Foundation for Health Research. R. M. Peters received salary support from NSERC funding granted to J. T. Inglis, and R. L. Mildren also received financial support through an NSERC doctoral research award. A. J. Hill received internal funding from a Work Learn International Undergraduate Summer Research Award. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSR.L.M. and A.J.H. performed experiments; R.L.M. analyzed data; R.L.M., R.M.P., J.-S.B., M.G.C., and J.T.I. interpreted results of experiments; R.L.M. and A.J.H. prepared figures; R.L.M. drafted manuscript; R.L.M., R.M.P., A.J.H., J.-S.B., M.G.C., and J.T.I. edited and revised manuscript; R.L.M., R.M.P., A.J.H., J.-S.B., M.G.C., and J.T.I. approved final version of manuscript. We thank Brian Horslen for assistance with EEG recordings, Martin Zaback for participation in piloting and comments on the manuscript, and Geoffrey McKendry for assistance with data collection. REFERENCES
Page 7AbstractSolubilities of respiratory gasses in water, saline, and plasma decrease with rising temperatures and solute concentrations. Henry’s Law, C = α·P, states that the equilibrium concentration of a dissolved gas is solubility times partial pressure. Solubilities in the water of a solution depend on temperature and the content of other solutes. Blood temperatures may differ more than 20°C between skin and heart, and an erythrocyte will undergo that range as blood circulates. The concentrations of O2 and CO2 are the driving forces for diffusion, exchanges, and for reactions. We provide an equation for O2 and CO2 solubilities, α, that allows for continuous changes in temperature, T, and solution density, ρ, in dynamically changing states: a(T, ρsol, ρω, γ, β)=[1−γ⋅(ρsol−ρωρω)β]⋅(C1⋅e−k1T+C2⋅e−k2T) This two-exponential expression with a density scalar γ, and a density exponent β, accounts for solubility changes due to density changes of an aqueous solution. It fits experimental data on solubilities in water, saline, and plasma over temperatures from 20 to 40°C, and for plasma densities, ρsol up to 1.020 g/ml with ~0.3% error. The amounts of additional bound O2 (to Hb) and CO2 (bicarbonate and carbamino) depend on the concentrations in the local water space and the reaction parameters. During exercise, solubility changes are large; both ρsol and T change rapidly with spatial position and with time. In exercise hemoconcentration plasma, ρsol exceeds 1.02, whereas T may range over 20°C. The six parameters for O2 and the six for CO2 are constants, so solubilities are calculable continuously as T and ρsol change. NEW & NOTEWORTHY Solubilities for oxygen and carbon dioxide are dependent on the density of the solution, on temperature, and on the partial pressure. We provide a brief equation suitable for hand calculators or mathematical modeling, accounting for these factors over a wide range of temperatures and solution densities for use in rapidly changing conditions, such as extreme exercise or osmotic transients, with better than 0.5% accuracy. for the application in biochemical and physiological studies, we set out to provide accurate equations for gas solubility applicable to a wide range of temperatures. The foundation for quantification and modeling of gas transport depend on the principles that determine gas solubility in liquids and defines gas concentrations in the body. Henry’s law states that gas dissolves in a solvent in proportion to the partial pressure of gas at equilibrium. O2 and CO2 solubilities decrease with increasing temperature and are also diminished by the presence of other solutes, ions, polymers, or proteins (12). Solute concentration, not partial pressure, drives reactions and defines diffusion gradients, so it is quantitatively important to account for changes in gas solubility as temperature increases between the inflow and outflow of heat-producing organs to define reaction rates and binding equilibria. For O2 and CO2, the binding and unbinding reactions with Hb are further complicated by the Bohr and Haldane effects, where increasing Pco2 augments O2 release in tissue and vice versa in the lung (5, 7). Alveolar temperatures are lower than heart, brain, and liver temperatures, and, as arterial blood enters these highly metabolizing organs, the rising temperature depresses gas solubility as the blood warms (1). General equations that hold over temperatures from 15°C in skin to 44°C are needed for estimating solubilities during bypass operations with hypothermia, during dialysis where hematocrit and protein concentrations oscillate, and during intense exercise with hemoconcentration of 10–15%. The need is for the solubility in the water of solutions containing both small solutes and proteins. Hb binding of O2 and proteins forming carbamino groups with CO2 merely remove the gases from solution by the chemical reaction and are not to be accounted for as solubility, which is unaffected. We took data from the literature, assessing it as best we could for accuracy of methods and reproducibility of observations, and especially for large numbers of observations over wide ranges of temperature. We found high-quality data for O2 and CO2 in saline and water. Data in plasma were not of equivalent quality; for example, protein concentrations were rarely measured. The plasma data were sampled from humans, and “normal human” concentrations were assumed to be all the same. The output of this study is algebraic equations describing the solubilities of O2 and CO2 in pure water, saline, and acidified plasma across a wide range of temperatures. These will aid investigators using temperature-dependent reactions and exchange. In capillary beds, gases and heat exchange between blood and tissue, where the concentrations and temperatures are varied over micrometer distances. For the mathematical modeling of gas transport, it is essential to account for the changes in solubilities and partial pressures continuously. With changing metabolism and consequent osmotic transients, one needs also to account for the effects of changing blood density, as well as temperature, in defining the solubility. Reactions using or producing O2 and CO2 must be accounted for separately but simultaneously. METHODSWe selected data on O2 and CO2 solubility at temperatures relevant to human physiology, accounting for normal and hypo- and hyperthermic conditions. We selected data we judged to be carefully obtained, rejecting data in studies with inconsistencies in observations or analyses. A prime source was the solubility data collection in Respiration and Circulation (2), referencing several sources. Investigations utilizing different techniques or methods that were inconsistent were not included (9, 12). Only data obtained by directly measuring gas solubility were used for our analyses, and we ignored interpolated estimates and observations that appeared anomalous, a somewhat arbitrary process. Because the original data sets were imperfect in various ways and are sometimes conflicting, we advocate that new experiments be done under a uniform set of laboratory conditions to improve the accuracy of the data, the range of solution densities, and the types of solutions, and then reassess and presumably reparametrize our equations or write more accurate ones. The key thing is to provide solubility equations accounting for plasma density and temperature in space and time simultaneously, while blood circulates and the gases exchange. The selected solubilities in water, 0.9% sodium chloride, plasma, and serum for O2 are in Table 1 and for CO2 are in Table 2. All measurements and estimates of gas solubility in this paper have standard units of milliliters of gas per milliliter of solvent at 1 atm pressure. The plasma samples were acidified to pH ~3.5 to block carbamino binding (11, 26).
We used simple smooth equations to fit the data on O2 and CO2 solubilities vs. temperature in various solutions. First, we fitted the data in water, saline, and plasma separately, obtaining excellent fits to the data. On observing that the curves for the three solution types were almost parallel to each other, we obtained the best fits for exactly parallel equations related to each other by a scalar relative to water. Such a constraint can only result in poorer fits to the data, so the minimized sums of squares are inevitably larger, but, as the results will show, the fits are still rather good. We then sought a physical basis for this good result in terms of some characteristic measurement of the particular solutions. Molar concentrations did not appeal, because of the diverse sizes of molecules in plasma. Density scaling became the next hypothesis. If the three solutions could be related by density, and if the fitted curves were indeed parallel, then one equation could serve as the descriptor for the solubilities of a gas over the whole temperature range in the three solutions given an experimental measurement of density. Such an equation would be still very simple, if the ratios of solution densities remained constant over the temperature range, and, since the solutions were all mainly water, this seemed like an hypothesis worth evaluating. The worst case would be that the equation would become more complicated if the density ratios changed with temperature. The tabulated solubilities are monotonic with temperature and concave upward. Linear equations and polynomials are unsuitable as descriptors. We used exponential equations and found that two exponentials sufficed for the ranges covered. The solubility α(Τ) (ml gas/ml solution at 1 atm pressure) at temperature T (°C), is described by: α(T)=C1⋅e−k1T+C2⋅e−k2T(1) where C1 and C2 are scalar coefficients (ml gas/ml solution at 1 atm pressure), and k1 and k2 are fixed coefficients (1/°C). This function integrates the information in data set on solubility vs. temperature in a medium of constant composition.Graphical plots of the data suggested that the solubility-temperature curves were more or less parallel for solutions of differing composition. This leads to a second, more compact way of describing the data, namely assuming a scalar relationship between pairs of solutions, scaling in proportion to the solubility in pure water by a scalar factor φ, 0 < φ < 1. α(T, φ)=φ⋅(C1⋅e−k1T+C2⋅e−k2T)(2) This hypothesis is tested by our analysis. Equation 2 represents a potentially more powerful expression, describing two or more data sets. Here “more powerful” implies that the equation parameters are better constrained by virtue of using more data points to define the parameters of the equation, for 3 solutions, the 4 parameters of the 2 exponentials plus 2 φ values, for a total of 6, instead of the 12 parameters using Eq. 1 to fit each data set. The two φ values are φsalt and φpl for saline and plasma, respectively.The possible presence of scalar relationships between solubilities in water, saline, and plasma provokes one to hypothesize a basis for values of these scalars. We propose that solution density can explain differences in solubility for different solutions and can improve the modeling by incorporating a measurable parameter to replace an empirical value for φ. This idea is based on the observation that O2 and CO2 solubilities diminish from water to saline to plasma, in order of increasing density of the solutions. On this basis, the third quantitative hypothesis can be expressed: a(T, ρsol, ρw, γ, β)=[1−γ⋅(ρsol−ρwρw)β]⋅(C1⋅e−k1T+C2⋅e−k2T)(3) where ρsol is the density of the solution, ρw is the density of water at the same temperature, γ is a scalar describing a proportionality due to the difference in solution density from that of water, and β is an exponent defining the nonlinearity in the influence of density on solubility. Solution densities for water, 0.99332 g/ml, and saline, 0.99933 g/ml, were obtained at 37°C by Horlocker and Wedel (16); Hinghofer-Szalkay and Greenleaf (13) found ρ = 1.020 g/ml for human plasma. This affirmed the values found by Van Slyke et al. (27), whose observations of plasma-specific gravities (the ratio of plasma density to water) averaging 1.0268 ± 0.01 recalculate by dividing by water’s density of 0.99332, giving ρ = 1.01996 for plasma. The hypothesis that γ might be a constant over a range of solutions is to assume that charge effects might be negligible and that the nature of the space-occupying solutes is unimportant. As Van Slyke et al. (27) pointed out, this is not exactly true; if the globulin fraction of protein were increased with constant total protein, the ρ of 1.020 g/ml would increase a little.Equation 3 is based on the density ratios, the γ, being independent of temperature even though volumes and viscosities do change with temperature. From our analyses of the data of Sharqawy et al. (22) and Iqbal and Verrall (17), the density ratios for saline to water and for plasma to water over the broad temperature ranges are almost constant: over the range from 20°C to 40°C the scalar term in parentheses in Eq. 3 ranged ±0.075% for saline and ±0.1% for plasma compared with those at 30°C. Thus error in Eq. 3 due to inconstancy of the density ratios was <0.1% over the 20°C range of physiological temperatures. In experimental studies, errors in measurement of T and ρ will be greater than this and will dominate the error. To put these hypotheses differently, Eq. 1 expresses the temperature effect on solubility with a pair of exponentials, purely descriptive for each gas and each solution. Equation 2 tests the idea that solubilities in water can predict the solubilities in another solution over the whole range of temperatures if the scale ratio φ is determined. Equation 3 tests the broader idea that solution density is a unique determinant of gas solubility, presumably describing an excluded volume of water, reducing gas solubility. We used the JSim data analysis system (6) to fit the data with models, the two-exponential equations with their adjustable curvature. The equations were fitted to the data in Tables 1 and 2, adjusting the parameters to minimize the sum of squares of the differences between model and data. Of the eight optimization routines in JSim, SENSOP and NL2SOL gave secure convergence in fitting both single data sets and multiple data sets simultaneously. In fitting model 1 (Eq. 1), each curve was fit individually. In fitting model 2 (Eq. 2), the three curves (for O2 or CO2) were fitted simultaneously, adjusting the coefficients, the Ci, the exponents, the ki, that apply to all three curves, and the scalars for saline and plasma, a total of five parameters for each gas. In fitting model 3 (Eq. 3), the three curves (for O2 or CO2) were fitted simultaneously, adjusting the coefficients, the Ci, the exponents, the ki, and the density coefficient, γ, and density exponent, β, a total of six parameters. The root mean sum of squares of error (RMSE) was calculated for each model fit, comparing the fit to the observed data, and then normalizing the RMSE to normal RMSE by dividing the RMSE by the average solubility. When multiple data sets were fit simultaneously, models 2 and 3, the “overall” RMSE was calculated using all the data points fitted for a given gas. RESULTSThe fitting of model 1, Eq. 1, to the individual data sets are shown in Fig. 1, and the parameter are given in Table 3. These parameters give accurate estimations for gas solubilities in water and saline over the temperature ranges tabulated. The data for plasma are not as consistent as the water and saline data in that they show discernable experimental error, and the RMSE is threefold greater for O2 in plasma than in water. The CO2 data were somewhat noisier than for O2, but the fits are good. Neither Eq. 1 nor Eq. 2 can handle temperature changes dynamically, but neither can handle non-homeostatic conditions when solution density changes. Equation 3 handles hemoconcentration or dilution: this would presumably be represented by changes in γ, rather than β, but, in the absence of highly accurate data, one cannot affirm that β does not change with a change in protein concentration. Further data are needed on this point. Fig. 1.Model 1 data sets fitted individually for O2 and CO2 solubilities in water, saline, and plasma vs. temperature over the range from 10 to 50°C. Data are from Table 1. The best fitting Eq. 1 and RMSEs are reported in Table 3. Top: O2 solubilities. Water data (⧫, ◇) are from Hodgman (15) and Benson et al. (4), respectively. Saline data (▲) are from Sendroy et al. (21). Human plasma data (○) are from Christoforides et al. (8). Bottom: CO2 solubilities. Water data (⧫) are from Hodgman (15); saline data (▲) are from Van Slyke et al. (26). O2 plasma data (○) are from Bartels and Wrbitzky (3).
Model 2, Eq. 2, assuming a constant proportionality φ in solubilities over the 40° temperature range from water to saline or plasma for analyzing the same data, uses the parameters shown in Table 4. Six parameters for each gas describe the data. The two φ values are the scalars in Eq. 2, φsalt and φpl, for saline and plasma, respectively. This is the most constrained of the three models, but it fails to consider density or water fraction. Because the fitting of the water data depends also on the fitting of parallel curves for saline and plasma, the water parameters must differ from those for the individual fits to the water data in Table 3. These data show that both O2 and CO2 solubilities in plasma are ~10% less than in water over the range of temperatures, and that, in saline, the CO2 solubility is much less reduced by the salt than is O2 solubility. The reason for the relatively higher CO2 than O2 solubility is not clear: there was no buffer, only NaCl. The fits to the data by model 2 are good, slightly better than for model 3, but not quite as good as for model 1: the fittedness, appropriately, is worse than for the least constrained model and better than for the most constrained model.
Model 3, Eq. 3 tests the idea of basing the saline and plasma model curves on the water curve and the densities of the solutions compared with water density. The data sources for the solubilities did not provide the densities, so we used 1.020 g/ml, the average plasma densities reported by Van Slyke et al. (27) and Hinghofer-Szalkay and Greenleaf (13). Six free parameters are optimized to fit the three sets of data, fewer than for model 1, but the same as for model 2, and are reported in Table 5. The density scaling parameter γ, combined with the density exponent β, allows almost as much freedom in positioning the curves as did using individual scalars of model 2 (Eq. 2), so the normalized sums of squares are similar to those for model 2. The estimates for γ for O2, 0.59, and CO2, 2.26, are dissimilar, as are the β values. The overall density scalars in the eighth column of Table 5, the parenthetic phrase in Eq. 3, are similar to those from model 2, but, due to the constraint by the known densities, are not identical. There is actually some positive covariance between γ and β, so this overlap is, in effect, a reduction in flexibility in the fitting.
Fits of the density model 3 are shown in Fig. 2. The two sets of O2 solubility data are slightly different, but systematic. For the analysis, we took the water densities, ρw, from Sharqawy et al. (22). The saline densities, ρs, from Sharqawy et al. (22) were validated by van Lopik et al. (24). Plasma densities, ρpl, are from Van Slyke et al. (27) and Hinghofer-Szalkay and Greenleaf (13). Plasma densities in normal humans have an almost 1% SD, and in abnormal humans range from 1.009 to 1.027, as taken from the specific gravities of Van Slyke et al. (27), times 0.99332, water density. The plasma data determine the γ, the scaling from one density to another. The β modifies the nonlinear curvature in the density relationships among water, saline, and plasma: β is <1.0, having a plateauing effect on dα/dρ. Fig. 2.Fitting of model 3, density-defined parallel curves, to O2 and CO2 solubilities in water, saline, and plasma vs. temperatures in the physiological range. Data are from Table 1 and are the same data as in Fig. 1, omitting the data at temperatures <25°C and >45°C. The model fits use Eq. 3 with the parameters reported in Table 5. Top: O2 solubilities. Densities at 37°C were water 0.99332 g/ml, 0.9% saline 0.99933 g/ml, and human plasma 1.020 g/ml. Bottom: CO2 solubilities. The higher β for CO2 than for O2 reflects the fact that the gap between water and saline curves for CO2 is smaller than the gap between water and saline of the O2 curves. DISCUSSIONThe observed solubilities listed in Tables 1 and 2 were all obtained with similar methods: gas determination in each solutions was measured manometrically with a Van Slyke apparatus (25). Error is <2% in manometric methods when triply repeated estimations are made. Using equations to interpolate through a whole set of observations reduces the error further, probably to the order of 1/2%. Benson et al. (4) describes more accurate methods for equilibrating gas and fluids and mass spectrometry for the fluid concentrations to estimate gas solubilities in water with accuracy of 0.01–0.02%. For O2 solubilities in water, the results are indistinguishable from those in our selected data sets and provide further confidence in our data selection process. The amount of dissolved O2 in the blood is commonly taken to be 3 ml/l whole blood; it is at 37°C more correctly 2.816 ml/l whole blood, a 5.5% error. Such error is not acceptable for scientific studies or exercise models. Exercise can hemoconcentrate blood ~10–15% and permit differences between skin and core temperatures, rising as high as 12°C (18). Hemoconcentration alone drops gas solubility by at least the amount of water displaced, and temperature increasing at exercising muscle can reduce solubility at the working muscle by as much as 25% compared with the cooler peripheral tissues. Because temperatures vary widely between cooled skin, lung alveoli, and metabolically warmed organs, we need accurate solubilities for quantitative analysis of metabolic reactions and fluxes. The model equations describing solubilities as functions of temperature and solution density serve two purposes. First, they provide an integrated description that smooths through data that contain experimental error, reducing the error when estimating the solubility at a particular temperature. Second, they summarize the data to give quantitative measures that can be computed for non-steady-state conditions, such as the increase in temperature that occurs when blood passes through metabolizing tissues. Equations that estimate the solubility of respiratory gases are necessary to calculate the gas’s molar concentration, the driving force for diffusional exchanges and reaction rates. Undoubtedly, our Eq. 3 will be superseded, perhaps by more comprehensive equations accounting for more refined understanding of the relationship between solutes, temperature, density, and other factors influencing gas solubility, and the parameters must be revised as better data are acquired. Gas solubilities are highest in pure water, decreasing with increasing concentrations of other solutes Binding or chemical buffering, for example, O2 binding to Hb, or CO2 being buffered as bicarbonate, augments their total blood content for the same gas tension. Other nonspecific binding can occur, as reported by Power (19) for O2 and CO2 in lung tissue. In experimental conditions in which the Hb hemeporphyrin site for O2 was ostensibly blocked by NaNO2 or ferricyanide, Power noted that as much as 17% of the O2 is loosely bound in blood. Christoforides and Hedley-White (9) found apparent increases in O2 solubility with increasing concentrations of seemingly NaNO2-blocked Hb. We think it is likely that the apparent increase in O2 solubility in their studies is really due to the binding of the O2 to the heme site, despite the partial block, rather than nonspecific protein binding. The influences of temperature on gas solubility are needed, particularly in mathematical models in exercise physiology. Metabolism in tissues produces heat, reducing gas solubility, thus augmenting the Bohr effects of CO2 and pH in fostering increased O2 release in tissues. At rest, the liver, brain, and heart have the highest heat-generating rates per gram, while during whole body exercise, skeletal muscle becomes the primary heat generator (20). The effect of temperature on the oxyhemoglobin saturation curve is greater than the effects of pH, Pco2, and 2,3-diphospho-d-glycerate. Increases in each of these increase the P50, the Po2 at which the hemoglobin binding sites for O2 are half filled, as does a rise in temperature. The summed effects are significant and make up the basis for the Bohr and Haldane effects (23). Gas solubilities are dependent also on the concentrations of ions and solutes in the blood. At 37°C, 9 g of NaCl/l H2O depresses O2 solubility by 5% and CO2 by 3%, as seen in Figs. 1 and 2. In plasma, O2 and CO2 solubilities are diminished by ~10%. These numbers represent the diminution in aqueous diffusive transport between bloodstream and cell fluid. The large plasma proteins, albumins and globulins, displace water and increase viscosity, as well as leave less space for gas to diffuse (14). The oxyhemoglobin and carboxyhemoglobin saturation curves are driven by the partial pressures and are not changed by plasma solution densities. The need for a single equation for the solubility of a gas over a large range of temperatures and solution densities (or other factors) is for accurate accounting of the circulatory exchanges in different tissues in normal and exercising humans. By accounting for both density and temperature, we can account for changes in affinities of Hb for O2 and CO2 along the length of a capillary in a metabolizing organ using the Hb binding equations of Dash et al. (10). These account for the changing gas tensions (O2 and CO2) simultaneously with temperature and pH, which change markedly in single capillary passage. Having a single equation for both temperature and density effects also allows for the consideration of solubilities in transient situations: a provocative example is sudden strenuous exercise in which the heart and skeletal muscle, and probably the brain, heat rapidly by 2–4°, and the circulating blood is concentrated by 10–15% by virtue of osmotic forces pulling water into the muscle cells. Blood density and muscle temperatures change rapidly before the skin warms. In computational modeling, the concentrations, temperatures, blood densities, and blood and tissue osmolarities need to be calculated at every time step and at each space step along the capillary. The hemoconcentration during exercise is a measure of the density and can be used to provide relatively accurate, continuously changing calculated solubilities and gas tensions in space and time. The parameterized versions of Eq. 3 for the O2 and CO2 solubilities in practical applications, using the values in Table 5, are: aO2=[1−0.59⋅(ρsol−ρwρw)0.497]⋅(0.0328⋅e−0.0398T+0.0163)(4a) aCO2=[1−2.256⋅(ρsol−ρwρw)0.877]⋅(1.052⋅e−0.0571T+0.6821⋅e−0.012T)(4b) with ρw = 0.99332 g/ml and ρsol = 0.99933 g/ml for saline or 1.020 g/ml for plasma at 37°C. The plasma value is a normal population average, but can vary from 1.017 to 1.034 (27). The 1.020 is van Slyke’s specific gravity of 1.0268 multiplied by water density of 0.99332 g/ml.The equations are for the solubility of the gases in the water of the solution. Additional O2 is bound to Hb within erythrocytes, and additional CO2 is bound as bicarbonate or as carbamino groups. All of these are separate calculations to account for their total masses. Since normally over 99% of O2 is bound, it is the dissolved O2 that is the poor neglected cousin. However, for CO2, it is the opposite: the carbamino content can be high enough that the total CO2 curve summing the carbamino-CO2 bound to plasma protein and the dissolved CO2 is higher than the curve for pure water; in this case, the bound CO2 more than makes up for the excluded CO2. The CO2 data reported here were all on plasma acidification to prevent the carbamino binding; thus there is a need for sets of data to characterize the total CO2, summing dissolved carbamino and bicarbonate CO2 at varied temperatures. We recommend using Eqs. 4a and 4b for O2 and CO2 solubilities somewhat tentatively: they are based on approximations to density effects and are based on data obtained from a variety of laboratories and experimental measures. We are particularly insecure about the plasma equation parameters since few investigators have reported the composition of plasma or serum when measuring solubility or density at different temperatures. Their exact composition is unknown. Consequently, we recommend new solubility studies in which plasma protein composition and densities are measured; this will improve the accuracy of the plasma equations. Another approach is to replace the accounting for solution density with a scalar based on the water fraction of the solution. The water fraction of plasma is ~0.94, but is heavily dependent on protein levels. An equation like Eq. 3 can be used similarly for water fraction instead of density, replacing density by water mass per unit volume, using values of water = 1.0 g/ml, saline = 0.991 g/ml, and plasma = 0.935 g/ml. Coefficients for O2 become C1, 0.0324; C2, 0.0165; k1, 0.0402; k2, 0; γ, 0.292; and β, 0.395, and for CO2 become C1, 1.167; C2, 0.543; k1, 0.0509; k2, 0.0092; γ, 0.871; and β, 0.8063. This allows for a second method for estimating solution solubility where there is limited information provided on the solvent. The advantage over density might be that it may be simpler to measure water content than density. Either density or water fraction will work. In either case, more and more accurate data are needed for plasma and plasma constituents. Our observations leave us with the practical equations for gas solubilities as functions of temperature and density, Eqs. 4a and 4b for O2 and CO2 and two recommendations. First, to improve the accuracy of gas solubility data in physiological solutions and to improve the understanding of solution density effects vs. the ionic or other chemical influence on solubility, new experiments should be undertaken. Second, when making measurements of gas solubility, the densities and the exact composition of the solution should be described in detail. These equations for solubilities as functions of temperature and plasma density or water content are practical for quantitative analysis of gaseous exchange in continuously changing states and in integrative modeling or analyzing such data. GRANTSThis research was supported by National Heart, Lung, and Blood Institute Grant NHLBI T15 088516 and U01-HL122199 and National Institute of Biomedical Imaging and Bioengineering Grant EB08407. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSK.M.C. performed experiments; K.M.C. and J.B.B. analyzed data; K.M.C. and J.B.B. interpreted results of experiments; K.M.C. and J.B.B. prepared figures; K.M.C. and J.B.B. drafted manuscript; K.M.C. and J.B.B. edited and revised manuscript; K.M.C. approved final version of manuscript; J.B.B. conceived and designed research. The authors thank Erik Butterworth for development of JSim and its archival forms in XMML; Lucian Smith for development of translators to and from JSim to SBML and CellML; and Bartholomew Jardine for coding and curation and for making the models available to readers at www.physiome.org. The models and the Simulation Analysis System JSim are free to be downloaded and run on any Linux, Macintosh OSX, or Windows platform. REFERENCES
Page 8AbstractRemote ischemic conditioning has been convincingly shown to render the myocardium resistant to a subsequent more severe sustained episode of ischemia. Compared with other organs, little is known regarding the effect of transient liver ischemic conditioning. We proposed the existence of cardioprotection induced by remote liver conditioning. Male Sprague-Dawley rats were divided into sham-operated control (no further hepatic intervention) and remote liver ischemic conditioning groups. For liver ischemic conditioning, three cycles of 5 min of liver ischemia-reperfusion stimuli were conducted before-(liver preconditioning), post-myocardial ischemia (liver postconditioning), or in combination of both (liver preconditioning + liver postconditioning). Rats were exposed to 45 min of left anterior descending coronary artery occlusion, followed by 3 h of reperfusion thereafter. ECG and hemodynamics were measured throughout the experiment. The coronary artery was reoccluded at the end of reperfusion for infarct size determination. Blood samples were taken for serum lactate dehydrogenase and creatine kinase-MB test. Heart tissues were taken for apoptosis measurements and Western blotting. Our data demonstrate that liver ischemic preconditioning, postconditioning, or a combination of both, offered strong cardioprotection, as evidenced by reduction in infarct size and cardiac tissue damage, recovery of cardiac function, and inhibition of apoptosis after ischemia-reperfusion. Moreover, liver ischemic conditioning increased cardiac (not hepatic) glycogen synthase kinase-3β (GSK-3β) phosphorylation. Accordingly, inhibition of GSK-3β mimicked the cardioprotective action of liver conditioning. These results demonstrate that remote liver ischemic conditioning protected the heart against ischemia and reperfusion injury via GSK-3β-dependent cell-survival signaling pathway. NEW & NOTEWORTHY Remote ischemic conditioning protects hearts against ischemia and reperfusion (I/R) injury. However, it is unclear whether ischemic conditioning of visceral organs such as the liver, the largest metabolic organ in the body, can produce cardioprotection. This is the first study to show the cardioprotective effect of remote liver ischemic conditioning in a rat model of myocardial I/R injury. We also, for the first time, demonstrated these protective properties are associated with glycogen synthase kinase-3β-dependent cell-survival signaling pathway. local cardiac preconditioning or postconditioning has been shown to have direct cardioprotective effects against myocardial ischemia and reperfusion (I/R) injury. Similarly, remote ischemic pre- or postconditioning is a phenomenon whereby brief episodes of short periods of I/R stimuli executed in distant tissues or organs may exert protective effects in the heart and render the myocardium more tolerant to a subsequent sustained episode of ischemia or reperfusion injury. Studies have shown that this remote cardioprotective trigger can be used in distant vascular beds (lower and upper limbs; cerebral, renal, mesenteric, intestinal, renal, and abdominal arteries) (18). However, it is unclear whether brief ischemic conditioning of visceral organs such as the liver, the largest metabolic organ in the body, can produce cardioprotection. A few reports have shown that liver preconditioning could protect remote organs, such as lung (27) or kidney (2), and prevent the incidence of arrhythmias in vivo (14) or ex vivo (24). Based on these previous studies, we, therefore, hypothesized that remote liver conditioning may also protect the heart against I/R injury in vivo. Furthermore, the timing of liver ischemic conditioning treatment, i.e., pre- vs. postconditioning, has not been evaluated. Thus the primary aim of this study was to evaluate whether liver ischemic conditioning protects the heart, using metrics including infarct size and myocardial damage serum markers. The potential mechanisms involved in the cardioprotection afforded by remote ischemic conditioning are not known. It has been proposed that either endogenous neural or humoral agent(s) traveled from the remote organ to the heart, or production of anti-apoptosis and anti-inflammatory factors in the remote organ, mediated cardioprotection (10). Ultimately, the pro-survival reperfusion injury salvage kinase (RISK) pathway is activated in the conditioned heart (9). The glycogen synthase kinase-3β (GSK-3β) signaling molecules are key components of the prosurvival RISK pathway (31). Many studies have demonstrated that GSK-3β phosphorylation promotes survival of conditioned cardiac myocytes (11, 22, 31, 33). In addition, emerging evidence has indicated that the phosphorylation of GSK-3β at Ser9 results in the inhibition of GSK-3β activity, and that this inhibition of GSK-3β enhances cell survival and limits infarct size post-I/R injury (6). These findings let us hypothesize that remote liver ischemic conditioning protect the heart against I/R injury via inhibiting GSK-3β activity, thus promoting myocardial survival. Therefore, with the use of an in vivo rat model, this investigation aimed to determine whether remote liver ischemia conditions the heart against I/R injury, and whether this effect is associated with GSK-3β-dependent cell-survival signaling pathways. MATERIAL AND METHODSThe experimental procedures and protocols were approved by the Institutional Animal Care and Use Committee of Sichuan University (Sichuan, China, approval no. 2015035A) and in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH publication, 8th edition, 2011). Male Sprague-Dawley rats (7–8 wk old, 250–300 g) purchased from Chengdu Dashuo Experimental Animal Research Center (Chengdu, China), were housed at 20–25°C under a 12:12-h light-dark cycle. Humidity was maintained at 60 ± 5% before experiments. Rats were anesthetized with pentobarbital sodium (50 mg/kg ip). After establishing adequate anesthesia, they were incubated and ventilated with a rodent ventilator (Taimeng, Chengdu, China). A standard limb lead II configuration electrocardiographic system was applied for heart rate and ST segment monitoring (Biolap 420F, Taimeng). The animals were then subjected to instrumentation for hemodynamics and surgical interventions (see below) under constant monitoring. Baseline measurement of the hemodynamics was conducted 10 min after the animals were stabilized. The experimental protocol is summarized in Fig. 1. Rats that died of anesthesia or during surgery, or did not reach 3 h of reperfusion stage, were excluded. There were two experimental series. Experiment 1 (a total of 115 rats, among which 10 died during the ligation period or during the reperfusion period because of acute heart failure) had an overall survival rate of 91.3%. Following the interventions, hearts were collected at 40 min for protein phosphorylation analysis and at 3 h for infarction and apoptosis image studies post-cardiac reperfusion. In experiment 2, only liver I/R stimuli protocol was executed without cardiac I/R to examine hepatic damage. A total of 40 rats were used. Liver sections and samples were collected at the same time points as those in experiment 1. Rats were randomly assigned to a sham-operated group (sham; hepatic arterial and venous trunk were exposed without intervention, chests were opened without coronary artery ligation), control group (CON; no further hepatic intervention), or remote liver ischemic conditioning groups of preconditioning (RPre), postconditioning (RPost), or a combination of preconditioning and postconditioning (RPre+RPost). In addition, two groups with inhibitor (CON + inhibitor and RPre+RPost + inhibitor) were included in experiment 1. Fig. 1.Experimental protocols. In experiment 1, all hearts were subjected to 45-min ligation of the left anterior descending coronary artery (LAD), followed by 3 h of reperfusion (R), except for the sham-operated ones. For remote liver ischemic preconditioning (RPre), three cycles of 5 min of liver ischemia with 5-min intermittent reperfusions were conducted before myocardial ischemia. For remote liver ischemic postconditioning (RPost), the above three cycles of liver I/R stimulus were induced after myocardial ischemia (at the onset of myocardial reperfusion). For the combination of remote liver ischemic preconditioning and postconditioning (RPre+RPost), the liver conditioning stimulus was induced before myocardial ischemia and at the onset of myocardial reperfusion. The GSK-3β inhibitor SB-216763 was applied 5 min before reperfusion. Thirty minutes of acute memory phase were allowed before being followed by a 45-min LAD occlusion and a subsequent 180 min of reperfusion (I/R). Arrows indicate the time points at which tissue samples were harvested. In experiment 2, hepatic stimuli were conducted without myocardial I/R intervention. Liver was taken at the same time points as in experiment 1. To produce liver conditioning, laparotomy was performed via a midline abdominal incision. The hepatic arterial and venous trunk, as well as the portal vein, were identified and isolated. Three cycles of 5 min of liver ischemia with 5-min intermittent reperfusions, i.e., clamping the vessel with a microvascular clip to induce ischemia (ischemia period), and releasing the clip to initiate reperfusion (reperfusion period), were conducted pre- (RPre) or post-myocardial ischemia (RPost) or a combination of both (RPre+RPost). To produce cardiac I/R, a median thoracotomy was performed to expose the heart. Myocardial I/R was conducted by ligation of left anterior descending coronary artery (LAD) approximately halfway between the base and the apex for a period of 45 min, followed by release of the suture and reperfusion for 3 h. Successful coronary artery occlusion was verified by the presence of regional dyskinesia and epicardial cyanosis in the ischemic zone. Reperfusion was verified by visual observation of an epicardial hyperemic response. A heating blanket was used for body temperature maintenance. For the sham-operated animals, LAD was separated but not occluded. To determine if the cardioprotective effect of liver conditioning is associated with MAPKs signaling cascade, SB-216763 (0.6 mg/kg, GSK-3β inhibitor; Sigma-Aldrich, St. Louis, MO) was administrated 5 min before myocardial reperfusion via the femoral vein. For euthanasia, an overdose of pentobarbital sodium (200 mg/kg ip) was given at the end of the experiment. A 20-G heparin-filled catheter (Spacelabs Medical, Redmond, WA) was inserted from the right carotid artery to the left ventricle and connected to a pressure transducer (Biolap 420F, Taimeng) connected for hemodynamic parameters measurement. Left ventricular systolic pressure (LVSP), left ventricular end-diastolic pressure (LVEDP), and maximum rate of increase/decrease in left ventricular pressure (±dP/dtmax) were recorded throughout the experiment. At the end of the experiment, LAD was reoccluded and the heart was perfused with 1% Evan’s blue (Sigma-Aldrich). The left ventricular area at risk (AAR) was identified from the normal myocardium, which was stained blue. Hearts were then frozen at −20°C for 30 min before being cut into 2-mm-thick slices parallel to the atrioventricular groove. Heart slices were then incubated at 37°C with 1% triphenyltertrazolium chloride (Sigma-Aldrich) in 0.1 M phosphate buffer (pH 7.4) for 20 min. Infarcted tissue within the AAR area was stained white. After fixation in 4% paraformaldehyde for 24 h, the infarcted and the noninfarcted tissue were separated and weighed. Infarct size was expressed as a percentage of the AAR for each group. At the end of cardiac reperfusion, whole blood was collected and centrifuged to obtain plasma (4,000 rpm for 10 min at 4°C). Serum levels of lactate dehydrogenase (LDH) and creatine kinase-MB (CK-MB) were quantified using a Mindray BS-120 Chemistry Analyzer (Mindray Medical, Shenzhen, China). All samples were measured in duplicate. A parallel experiment was carried out for immunohistochemistry and terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining. Left ventricular AAR regions were identified from normal area by the use of Evans blue dye. AAR areas were then separated, rinsed in saline, dried, embedded in 10% phosphate-buffered formalin, and cut into 5-μm sections parallel to the atrioventricular groove. Serial sections of transverse myocardial and liver slices were deparaffinized in xylene and isopropanol and then were mounted on glass slides for immunohistochemical analysis and apoptosis measurements. The TUNEL method was used for determination of apoptotic cells. The heart or liver sections were stained using the In Situ Cell Death Detection Kit (Roche Diagnostics, Indianapolis, IN), according to the manufacturer’s instructions. TUNEL-positive nuclei were stained red, and TUNEL-negative nuclei were stained blue. More than 10 different microscopic fields per heart section were chosen at random and were evaluated blindly. The apoptotic index was calculated as a ratio of TUNEL-positive nuclei to the total nuclei population. Images were obtained and viewed using an inverted microscope (Olympus A/S, Ballerup, Denmark) and were analyzed with Image-pro plus (Media Cybernetics, Carlsbad, CA). Primary antibodies, including anti-Bcl-2 and anti-Bax antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), were used, followed by incubation with second antibodies (Santa Cruz Biotechnology) of either biotinylated goat anti-rabbit IgG (Bcl-2) or goat anti-mouse IgG (Bax). Cardiac tissues were stained with 3,3-diaminobenzene (Beijing Zhongshan Golden Bridge Biotechnology) solution, and all sections were counterstained with hematoxylin to visualize cell nuclei. Phosphate buffer solution was used as a negative control. Images were captured by an inverted microscope (CAST system, Olympus A/S) and analyzed with Image-pro plus software (Media Cybernetics). Ten views were randomly selected per slide, and positive expression was characterized as brown staining in the cytoplasm of cells. Statistical value was calculated by the ratio of optical density of area positively stained to mean optical density, i.e., positive expressive index. The left ventricular AAR region (including the infarct zone) isolated from the normal area was homogenized and prepared as previously described (12, 13, 15, 16). Briefly, the bicinchoninic acid method (Pierce, Rockford, IL) was adopted for protein concentration determination. Fifteen micrograms of protein were loaded per lane and resolved on a 10% SDS-PAGE gel before being transferred onto nitrocellulose membranes (VWR, Batavia, IL). Primary antibodies raised against phosphorylated GSK-3β (Ser9) (p-GSK-3β Ser9, rabbit, 1:1,000; Cell Signaling), total GSK-3β, and cleaved caspase-3 and caspase-3 (rabbit, 1:1,000; Cell Signaling) were used, followed by incubation with horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (Bio-Rad, Hercules, CA). An Amersham Imager 600 system (GE Healthcare, Little Chalfont, UK) was used for detection of signals. Band densities were analyzed by ImageJ Data Acquisition software (National Institutes of Health, Bethesda, MD). Phosphorylation signal densities were normalized to total protein signal densities. All samples were run in duplicate on separate gels. All values are expressed as means ± SE. Statistical analyses were performed using Graphpad Prism 5 software (NIH) or SPSS 13.0 software for Windows (SPSS Inc., Chicago, IL). The statistical test for hemodynamics over time was performed by two-way repeated-measures ANOVA. Other comparisons of two means (by an unpaired Student’s t-test), or several means (by one-way ANOVA) were performed, assuming statistical significance with P values < 0.05. RESULTSWe first examined if the live conditioning protocol (i.e., 3 episodes of 5-min ischemia followed by reperfusion) produce hepatic injury. Hematoxylin and eosin (Fig. 2A) and TUNEL (Fig. 2B) staining on liver sections revealed that these hepatic I/R interventions did not induce hepatic cell death. No cytoplasmic vacuolation, edema, hemorrhage, or steatosis was found in liver samples. Thus our liver conditioning protocol did not cause harm to the liver. After liver conditioning, myocardial ischemia followed by reperfusion was carried out. The myocardial infarct size in RPre (41.2 ± 1.7%, P < 0.05), RPost (37.8 ± 2.2%, P < 0.05), and RPre+RPost rats (28.6 ± 2.1%, P < 0.001) was significantly lower than that of rats in the CON group (48.9 ± 3.5%) (Fig. 2, C and D). The ratio of the AAR to the left ventricle was similar among groups (P > 0.05, Fig. 2D). Moreover, the reduction in heart infarct size was greater in rats subjected to a combination of treatments, RPre+RPost, compared with rats subjected to only RPre (P < 0.01) or RPost (P < 0.05). To explore the effect of liver ischemic conditioning on apoptosis, TUNEL staining was applied on heart sections after 3 h of reperfusion. TUNEL-positive nuclei were prevalent in the left ventricular AAR. However, liver ischemic conditioning caused a marked reduction of apoptotic nuclei (RPre: 26.4 ± 1.1%, RPost: 22.3 ± 1.9%, RPre+RPost: 18.7 ± 1.6%) compared with the CON group (32.6 ± 1.6%; P < 0.05, P < 0.01, and P < 0.001, respectively). In addition, RPre+RPost treatment effectively decreased apoptosis compared with RPre alone, following cardiac reperfusion (P < 0.05, Fig. 2, E and F). Fig. 2.Liver conditioning ameliorated myocardial damage post reperfusion. A: hepatic ischemia and reperfusion cycles did not cause liver damage. Representative (of n = 5 rats/group) hematoxylin-and-eosin-stained micrographs of liver section are shown. Scale bars, 100 μm. B: representative images of terminal transferase dUTP nick-end labeling (TUNEL)-stained liver sections. Nuclei were counterstained with 3,3-diaminobenzene (n = 4–5). RPre, remote liver ischemic preconditioning; RPost, remote liver ischemic postconditioning; RPre+RPost, the combination of remote liver ischemic preconditioning and postconditioning. Scale bars, 10 μm. C: representative sections of triphenyltetrazolium chloride (TTC)-stained heart subjected to 45-min myocardial ischemia followed by 3-h reperfusion. CON, control; RPre, remote liver ischemia preconditioning; RPost, remote liver ischemia postconditioning; RPre+RPost, the combination of remote liver ischemia preconditioning with remote liver ischemic postconditioning. D: quantification of myocardial infarct size expressed as a percentage of left ventricular (LV) area at risk (AAR) (top) and AAR expressed as a percentage of LV area (bottom). Values are means ± SE; n = 5 each group. *P < 0.05 and ***P < 0.001 compared with CON; ##P < 0.01 compared with RPre; and †P < 0.05 compared with RPost (by one-way ANOVA). E: representative images of TUNEL-stained heart sections. Myocardial apoptosis was determined by measurement of TUNEL-positive cardiomyocyte nuclei in the AAR of myocardium obtained from liver-conditioned and CON rats after coronary artery reperfusion injury. TUNEL-positive (red) cardiomyocytes were identified as apoptotic cells. Positive cells were not detected in the nonischemic zone. Arrows denote TUNEL-positive nucleus. Scale bars = 10 μm. F: bar graph showing the TUNEL-positive nuclei expressed as a percentage of total nuclei in the heart AAR sections. Values are means ± SE; each group, n = 4–6. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with CON; #P < 0.05 compared with RPre (by one-way ANOVA). G: serum levels of lactate dehydrogenase (LDH) in rats subjected to 45 min of left anterior descending artery occlusion followed by 3 h of reperfusion. Values are means ± SE; n = 11–13 per group. **P < 0.01 and ***P < 0.001 vs. sham; #P < 0.05 vs. CON. H: post-reperfusion injury mean serum levels of creatine kinase MB (CK-MB) for CON and liver ischemic conditioned rats. Values are means ± SE; n = 7–12 per group. ***P < 0.001 vs. sham; #P < 0.05 and ##P < 0.01 vs. CON. Table 1 illustrates the time course of hemodynamics during the experiment. Under baseline conditions, there was no statistical difference in systemic hemodynamics among the groups before the coronary occlusion (P > 0.05). Reperfusion caused a reduction of systolic function, as indicated by LVSP and dP/dtmax, and depression of diastolic function as indicated by LVEDP and −dP/dtmax, compared with the respective baseline values (P < 0.05, P < 0.01, or P < 0.001) in CON, RPre, and RPost groups, respectively. Repeated two-way ANOVA showed a statistically significant difference among CON, RPre, RPost, and RPre+RPost groups in LVSP (P = 0.002), dP/dtmax (P = 0.044), and −dP/dtmax (P = 0.008) over the period of measurement. The recovery of cardiac function, including LVSP, dP/dtmax, and −dP/dtmax, was greater in the RPre+RPost hearts than in the CON group or in hearts treated with a single liver conditioning stimulus (P < 0.05 or P < 0.01). Significant interactions between group assignment and time were observed for LVSP (P = 0.002) and LVEDP (P = 0.024).
Significant differences were detected among the experimental groups in LDH and CK-MB activities (LDH: P < 0.0001; CK-MB: P = 0.005). Myocardial reperfusion caused a significant elevation of LDH level in CON, RPre, RPost, and RPre+RPost groups (all P < 0.01 vs. sham), with the highest in CON rats, indicating the protective effect of liver conditioning against myocardial reperfusion injury (RPre, RPost, and RPre+RPost vs. CON, P < 0.05, Fig. 2G). In addition, although serum CK-MB level increased significantly after I/R injury in the CON rats compared with the sham group (P < 0.001), significantly lower CK-MB level was observed in rats with liver ischemic conditioning (P < 0.05 or P < 0.01 vs. CON, Fig. 2H). Immunohistochemistry indicated the expression of Bcl-2 and Bax proteins in cardiomyocytes of rats. As shown in Fig. 3, A–C, liver ischemic conditioning significantly upregulated the expression of Bcl-2 protein (P < 0.05, all vs. CON) and downregulated the expression of Bax proteins (P < 0.05, RPost vs. CON or P < 0.01, RPre+RPost vs. CON, Fig. 3C). Importantly, there was a dramatic protective effect of the combined treatment of RPre+RPost against reperfusion injury, as reflected by a greater Bcl-2-to-Bax ratio than that of the CON group (P < 0.05, Fig. 3D). Expression of cleaved caspase-3 and total caspase-3 protein in the myocardium was further determined by Western blotting (Fig. 3, E and F), Compared with the sham-operated group, reperfusion injury caused significant elevation of cleaved caspase-3 (Fig. 3E) or caspase-3 (Fig. 3F) protein expression (P < 0.01 or P < 0.001); however, this elevation was further attenuated with liver conditioning treatments (P < 0.001 vs. CON). Fig. 3.Myocardial apoptotic protein expression. A: representative immunostainings of Bcl-2 and Bax protein in the cytoplasm of the left ventricular myocytes isolated after 3 h of reperfusion. CON, control; RPre, remote liver ischemia preconditioning; RPost, remote liver ischemia postconditioning; RPre+RPost, the combination of remote liver ischemia preconditioning with remote liver ischemic postconditioning. Densitometric analysis of Bcl-2 (B), Bax (C), and Bcl-2/Bax (D) protein expression in rat left ventricles post-I/R is shown. Values are means ± SE; n = 4–6 per group. PEI, positive expressive index. *P < 0.05, **P < 0.01, and ***P < 0.001 vs. sham; #P < 0.05 and ##P < 0.01 vs. CON; †P < 0.05 vs. RPre. Representative Western blots (top) and quantification (bottom) of cleaved caspase-3 (E) and caspase-3 (F) protein band density (normalized to GAPDH) in sham, CON, and liver-conditioned rat left ventricles are shown. Values are means ± SE; n = 4–5 per group. *P < 0.05, **P < 0.01, and ***P < 0.001 vs. sham; ##P < 0.01 and ###P < 0.001 vs. CON. Importantly, remote organ ischemic conditioning stimulus could exert potent cardioprotection via activation of cell survival signaling pathways, such as GSK-3β signaling molecules. We determined levels of GSK-3β in all groups. Liver conditioning alone did not alter local liver GSK-3β phosphorylation (Fig. 4A). However, the ratio of phosphorylated (p) to total (t) GSK-3β in liver-conditioned hearts was almost double that of CON hearts post-cardiac I/R (P < 0.01, Fig. 4B), while total GSK-3β protein levels were similar between CON and liver-conditioned ischemic ventricles. We did not see any difference in cardiac GSK-3β phosphorylation levels among liver conditioning groups (P > 0.05) after cardiac I/R injury. Fig. 4.Liver ischemic conditioning stimulates ventricular GSK-3β phosphorylation. Representative immunoblots (top) and densitometric analysis (bottom) of phosphorylated GSK-3β (Ser9) (p-GSK-3β) and total GSK-3β in rat livers (A) and ventricles (B) are shown. Sham animals did not undergo liver stimulus. Sham, sham-operated group; RPre, remote liver ischemic preconditioning; RPost, remote liver ischemic postconditioning; RPre+RPost, the combination of remote liver ischemic preconditioning and postconditioning. All band densities were normalized to sham group. Values are means ± SE; n = 5 in each group. NS, no significant difference. *P < 0.05 and **P < 0.01 vs. sham; ##P < 0.01 vs. CON (by one-way ANOVA). Inhibition of GSK-3β has been reported to exert cardioprotection against reperfusion injury and reduce infarct size (31). To investigate whether GSK-3β pathway plays a role in the liver ischemic conditioning-induced cardioprotection in our experiments, we administrated GSK-3β inhibitor SB-216763 after the LAD occlusion 5 min before commence of reperfusion. We selected the RPre+RPost group to administrate SB-216763, since this group exhibited the most potent liver conditioning effect. Consistent with previous studies, inhibition of GSK-3β mimicked the effects of liver ischemic conditioning: the inhibitor-treated CON group had similar reduction in heart infarct size as the RPre+RPost-treated group. However, inhibition of GSK-3β in RPre+RPost did not produce additional infarct size reduction compared with this observed in RPre+RPost alone (Fig. 5, A and B). Meanwhile, SB-216763 decreased serum concentration of LDH and CK-MB after I/R injury, to levels similar to that of liver-conditioned rats without inhibitors (P < 0.05, P < 0.01, and P < 0.001 vs. CON, Fig. 5, C and D). Consistently, inhibitor-treated CON hearts significantly reduced the number of TUNEL-positive stained apoptotic nuclei in the AAR compared with nontreated CON (P < 0.01 or P < 0.001). GSK-3β inhibition did not further enhance the anti-apoptotic activity in RPre+RPost hearts compared with RPre+RPost alone (P > 0.05, Fig. 5E). Pharmacological inhibition of GSK-3β also mimicked the effects of liver conditioning on ventricular GSK-3β phosphorylation (Fig. 5F). SB-216763 did not further increase GSK-3β phosphorylation or cardioprotection in liver-conditioned rats. However, SB-216763 induced similar degrees of GSK-3β phosphorylation and cardioprotection in CON rats as in liver-conditioned rats, supporting the vital role of GSK-3β phosphorylation (inactivation) in liver conditioning-induced cardioprotection. Fig. 5.The protective effect of pharmacological inhibitors on I/R injury. A: representative sections of TTC-stained heart subjected to 45-min myocardial ischemia followed by 3-h reperfusion. CON, control; RPre+RPost, the combination of remote liver ischemia preconditioning with remote liver ischemic postconditioning. B: quantification of myocardial infarct size expressed as a percentage of left ventricular area at risk. Values are means ± SE; n = 4–5 each group. Values for CON and RPre+RPost rats are repeated from Fig. 2C for comparison. **P < 0.01 and ***P < 0.001 compared with CON (by one-way ANOVA). C: mean serum levels of LDH of CON and liver ischemic conditioned rats after I/R injury with or without SB-216763 or U-0126. Values are means ± SE; n = 7–13. Values for CON and RPre+RPost rats are repeated from Fig. 2G for comparison. **P < 0.01 and ***P < 0.001 compared with CON (by one-way ANOVA). D: mean serum levels of CK-MB of CON and liver ischemic conditioned rats after I/R injury with or without SB-216763 or U-0126. Values are means ± SE; n = 7–12. Values for CON and RPre+RPost rats are repeated from Fig. 2H for comparison. *P < 0.05 and **P < 0.01 compared with CON (by one-way ANOVA). E, left: apoptotic nuclei detected by TUNEL technique in the area of myocardium at risk. Arrow indicates TUNEL-positive nuclei (red). Scale bars, 10 µm. C, control; R, the combination of remote liver ischemia preconditioning with remote liver ischemic postconditioning; S, SB-216763. Right: graph showing the averaged percentage of TUNEL-positive cells in the ischemic regions of LVs. Values are means ± SE; n = 4–6, each group. Values for CON and RPre+RPost rats are repeated from Fig. 2, E and F for comparison. **P < 0.01 and ***P < 0.001 compared with CON (by one-way ANOVA). F: representative Western blots (top) and densitometric analysis (bottom) of phosphorylated GSK-3β (Ser9) (p-GSK-3β) and total GSK-3β in rat ventricles with (+) or without (−) SB-216763 after 180 min of cardiac I/R injury. Sham, sham-operated group; RPre, remote liver ischemic preconditioning; RPost, remote liver ischemic postconditioning; RPre+RPost, the combination of remote liver ischemic preconditioning and postconditioning. All band densities were normalized to sham group. Values are means ± SE; n = 5 in each group. ***P < 0.001 vs. sham; ###P < 0.001 vs. CON (by one-way ANOVA). The effect of inhibitor on hemodynamics was also investigated. At baseline, there were no differences among groups in the hemodynamic parameters before intervention. Remote liver conditioning with and without SB-216763 significantly improved hemodynamics in LVSP (P = 0.001), dP/dtmax (P = 0.019), and −dP/dtmax (P = 0.005), compared with CON over the period of measurement. We also observed significant group × time interactions for LVSP (P = 0.004) and dP/dtmax (P = 0.045). SB-216763 similarly improved hemodynamics in both CON and RPre+RPost-treated groups (P > 0.05). Thus inhibition of GSK-3β activity in CON resulted in similar hemodynamic enhancement as in RPre+RPost-treated hearts without inhibitor, i.e., enhanced cardiac function compared with non-inhibitor-treated CON hearts (Table 2).
DISCUSSIONTo summarize, our data demonstrate that liver ischemic conditioning offers strong cardioprotection, as seen by significant reduction in infarct size, restoration of cardiac function, inhibition of apoptosis, and alleviation of cardiac tissue damage after myocardial reperfusion injury. Moreover, these data suggest that this cardioprotection is associated with GSK-3β-dependent cell-survival signaling pathway. Remote ischemic conditioning (7), brief episode of regional ischemia followed by reperfusion in remote organs or limbs (other than hearts), has been convincingly shown to render the myocardium resistant to a subsequent sustained episode of ischemia. Preclinical studies have demonstrated that the short cycles of I/R stimulus imposed on limbs or distant organs attenuated the extent of myocardial injury (30). These are effective when conducted before myocardial ischemia [remote ischemic preconditioning (29)], or during cardiac reperfusion [remote ischemic postconditioning (1)]. Compared with other organs, little is known regarding the liver ischemic conditioning, which may have other beneficial effects, such as reducing inflammation or promoting hepatic regeneration (28). Several studies clearly demonstrated that liver preconditioning decreased the incidence of arrhythmias in vivo (14) and ex vivo (24). Here, we further demonstrated the existence of remote liver conditioning-induced cardioprotection. We showed that either liver ischemic preconditioning (RPre), postconditioning (RPost), or a combination of both treatments (RPre+RPost) attenuated myocardial damage post-reperfusion insult in vivo. Moreover, we found that the RPre+RPost combination decreased infarct size to a larger degree than RPre or RPost alone, which was in agreement with studies on limb conditioning when both strategies were used in combination for pulmonary protection in cardiac surgery patients with CPB (21). Although numerous animal experiments and clinical trials in patients with acute ST-elevation myocardial infarction or percutaneous coronary interventions confirmed the beneficial effect of remote ischemic conditioning (3, 30), recent reports revealed that it failed to improve clinical outcomes following cardiac surgery (8, 20). The apparent discrepancy may result from different study designs, protocols, and primary end point used, which can produce considerable variability in outcomes. Importantly, these negative results were derived from clinical trials of cardiac surgery, during which numerous inherent cardioprotective regimens were used, such as anesthetics, cardioplegia, or hypothermia; thus their influence on the interpretation of available data cannot be ruled out. It is clear that anesthetics have strong cardioprotective effects, including volatile anesthetics and propofol, whose presence during the surgery may largely conceal the existing protection effect of endogenous ischemic conditioning (16, 18, 35). Therefore, we chose sodium pentobarbital as the anesthetic to avoid any potential ambiguity. Pentobarbital may have effects on cardiac function preservation, depending on various surgery strategies or species; for example, one study showed that it could alter cardiac hemodynamics in mice (34). In contrast, no infarct size limitation was found in rabbits (5). One could argue that the use of pentobarbital may potentially eliminate the cardiac protective phenomenon observed in the present study; however, we and others (19) showed that remote ischemic stimulus conferred powerful cardioprotection compared with CON hearts in the presence of identical sodium pentobarbital in each group. Thus it is unlikely that pentobarbital would generate differences in parameters measured among groups. The mechanisms of liver ischemic conditioning-induced cardiac protection have not been completely defined. Apoptosis plays a crucial role in the pathogenesis of myocardial I/R injury. It is triggered by apoptotic signals and is dominated by numbers of regulating genes during reperfusion injury (17). Although liver conditioning-induced cardioprotection against I/R injury occurs concomitantly with increases in Bcl-2 protein expression and decreases in Bax protein expression, thus resulting in an increased Bcl-2-to-Bax ratio, it appears that these liver ischemic stimuli failed to alter cleaved or total caspase-3 proteins expression. As such, the mechanisms of remote liver conditioning contrast to those of local preconditioning or postconditioning, for which there is substantial evidence showing that caspase-3 plays a key role in mediating apoptotic cell death in the final degradation phase (11). Multiple signaling pathways were previously shown to participate in infarct reduction by remote conditioning (18). We demonstrated in the present study that GSK-3β, a key component of the RISK pathway, is involved and functionally significant in producing myocardial protection induced by liver ischemic conditioning in a rat model of I/R injury. GSK-3β regulates multiple biological events. Inactivation of GSK-3β through the phosphorylation at Ser9 sites leads to the inhibition of GSK-3β activity (4). Recent studies have shown that local ischemic conditioning or remote ischemic conditioning-induced GSK-3β phosphorylation and inactivation was cardioprotective (32, 33). Its phosphorylation increased the threshold of mitochondrial permeability transition pore opening, which promoted survival of cardiac myocytes (26). Our present results confirm and expand previous findings by demonstrating that liver ischemic treatment before or post-myocardial ischemia dramatically increased phosphorylation of ventricular GSK-3β (Ser9) post-reperfusion compared with nontreated rats (CON). Furthermore, pharmacological inhibition of GSK activity increases GSK phosphorylation and is, therefore, cardioprotective in terms of reducing infarct size or alleviating myocardial damage. It has been reported that the pharmacological GSK-3β inhibition (via Ser9 phosphorylation) could prevent mitochondrial permeability transition pore opening, thus shifting myocytes toward survival (22, 25). It is not surprising that, in the present study, we found infarct size reduction on administration of GSK inhibitor SB-216763. However, while all of these findings strongly suggest the favorable role of GSK inhibition in the myocardium, few contradictory studies showed that inhibition of GSK-3β failed to minimize infarction, or GSK-3β (Ser9) phosphorylation level did not change after reperfusion (23). This controversial debate may reflect species differences or experimental protocol variability. We acknowledge several limitations of this study. First, we used only one liver ischemic intervention protocol, i.e., three cycles of 5-min ischemia/5-min reperfusion. Thus it remains unknown whether other liver stimuli strategies can produce cardioprotection. Second, we did not use all three liver ischemic conditioned hearts for immunohistochemical analysis. Rather, we used only RPre+RPost-treated hearts, to prove the concept that liver conditioning can inhibit apoptosis in the myocardium. Thus it remains unknown whether RPre or RPost alone can affect apoptotic protein expression after I/R injury. Third, in the present study, we clearly showed that liver conditioning significantly reduced apoptotic index in ventricles compared with matched control ones. However, the current TUNEL technique is not capable of differentiating cell types; therefore, other cells infiltrated into the myocardium during I/R injury may possibly contribute to the population of apoptotic cells. In addition, the mechanisms by which remote liver conditioning enhance cytoprotective actions seem to be related to MAPK signaling pathways, since we found liver ischemic treatment elicited activation of RISKs, such as GSK-3β; large-scale identification and characterization of each signaling molecule, as well as their interactions, would be of great interest in future research on this topic. In conclusion, liver conditioning exerts strong cardioprotective effects, as seen by reduction in infarct size and recovery of cardiac performance. This liver conditioning-mediated cardioprotection is associated with GSK-3β-dependent signaling pathways. GRANTSThis study was supported by grant no. 81670300 (to Z. Hu) from the National Natural Science Foundation of China. G. W. Abbott was supported by US National Heart, Lung, and Blood Institute Grant HL-079275. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSS.Y., G.W.A., W.D.G., J.L., C.L., and Z.H. conceived and designed research; S.Y. and Z.H. performed experiments; S.Y., J.L., C.L., and Z.H. analyzed data; S.Y., G.W.A., W.D.G., J.L., C.L., and Z.H. interpreted results of experiments; S.Y., G.W.A., and Z.H. prepared figures; S.Y., G.W.A., W.D.G., J.L., C.L., and Z.H. approved final version of manuscript; G.W.A., W.D.G., and Z.H. edited and revised manuscript; Z.H. drafted manuscript. REFERENCES
Page 9AbstractDetermining the presence of thoracoabdominal asynchrony in chronic obstructive pulmonary disease (COPD) patients is clinically relevant, but there is no consensus on the optimal parameters for performing this analysis. We assessed 22 COPD patients (FEV1 40 ± 10% predicted) and 13 healthy controls during rest and exercise with optoelectronic plethysmography (70% maximum workload) on a cycle ergometer. Thoracoabdominal asynchrony was calculated by using phase angle and phase shift parameters following a three-compartment model involving the upper and lower rib cages and abdomen. Patients were classified as having thoracoabdominal asynchrony (TAA+) or not (TAA−) based on control values (mean ± 2 SDs). The chest wall volume and compartmental contribution were also measured. Thoracoabdominal asynchrony was observed in the lower rib cage. The phase angle detected more TAA+ patients at rest (15 vs. 7 patients) and during exercise (14 vs. 8 patients) compared with the phase shift. TAA+ patients also presented a lower chest wall volume, lower rib cage contribution, and higher abdominal contribution to chest wall volume compared with the control and TAA− patients. Thoracoabdominal asynchrony was more detectable during rest and exercise using the phase angle parameter, and it was observed in the lower rib cage compartment, reducing the chest wall volume during exercise in patients with COPD. NEW & NOTEWORTHY This study contributes to advance the knowledge over the previous lack of consensus on the assessment of thoracoabdominal asynchrony. We rigorously evaluated the related features that interfere in the measurement of the asynchrony (measurement tool, chest wall model and calculation parameter). Our results suggest that phase angle detects more suitably thoracoabdominal asynchrony that occurs on the lower ribcage and leads to a reduction in the chest wall volume during exercise in COPD patients. in healthy individuals, inspiration is characterized by an expansion of the rib cage and the abdominal wall, while during expiration, the thorax and abdomen return to their resting positions due to the compliance of the lungs and the chest wall. Thus, during the respiratory cycle, there should be coordination between thoracoabdominal compartments, and the failure of this coordination is called thoracoabdominal asynchrony (TAA) (12). TAA has been observed in patients with chronic obstructive pulmonary disease (COPD) (1, 23), either at rest or during voluntary efforts (23) and seems to be associated with airflow obstruction (8) and hyperinflation (1), leading to limitations in exercise capacity (6). TAA has been reported using magnetometers (8) and respiratory inductive plethysmography (13). Both record dimensional changes of the thorax and abdomen using a two-compartment model that includes the rib cage and abdomen (23). Optoelectronic plethysmography (OEP) has become the most used technique to assess TAA (4, 20). In contrast to magnetometers and respiratory inductive plethysmography, OEP usually adopts a three-compartment model involving the upper (URC) and lower rib cage (LRC) and the abdomen (ABD). This model divides the rib cage into two compartments because there is evidence that respiratory muscles influence each rib cage compartment in a different fashion. For instance, the contraction of the diaphragm has a higher impact on the movement of the LRC, whereas the movement of the URC is more dependent on the activity of accessory and inspiratory rib cage muscles (1, 27). There are at least six distinct parameters used to estimate TAA (11, 21), but phase angle (PA) and phase shift (PS) appear to be the most commonly used and have been compared in different situations. Sackner et al. (23) showed that PS distinguished TAA more consistently than did PA in COPD patients during quiet and controlled breathing (23). Rusconi et al. (22) found a great discrepancy between PA and PS in detecting TAA in infants after induced airway obstruction (22). Prisk et al. (21) compared six parameters in simulated signals and breathing patterns from rhesus monkeys and showed that PA was better at assessing TAA because PS demonstrated the largest sensitivity to noise-related errors. These studies were performed using the two-compartment model. In 2009 (11), the first study to compare asynchrony with the three-compartment model was published, and the authors suggested that the three-compartment model better estimates TAA in COPD patients. Since that investigation, most studies using OEP to assess TAA have adopted the three-compartment model (14, 20). Despite the contributions noted above, the optimal parameters to assess TAA and the relevance of the three-compartment model remain unclear. This study aimed to compare PA and PS in the quantification and detection of TAA in COPD patients during rest and exercise using the three-compartment model and to evaluate the relevance of this model in the analysis of TAA and respiratory kinematics. METHODSThis cross-sectional study included 35 subjects (22 with severe or very severe COPD and 13 controls). COPD patients [forced expiratory volume in the first second (FEV1)/forced vital capacity (FVC) ratio <70%] (10) aged between 50 and 70 yr were recruited from a tertiary university hospital outpatient obstructive disease clinic. Patients with a body mass index (BMI) between 20 and 30 kg/m2, not oxygen-dependent, clinically stable (no worsening of respiratory symptoms or emergency/unscheduled visits at least for 30 days) and receiving optimized medical treatment (bronchodilator with or without inhaled corticosteroid) were included. A control group with 13 non-COPD, age-matched, nonphysically active subjects (no physical activity performed more than twice a week in the last 90 days) was also included, with normal lung function (FEV1/FVC >0.7) and without health problems, composed of mostly university or hospital employees. The University Hospital Ethics Committee approved the study, and all subjects signed their informed consent. Participants visited the hospital on two nonconsecutive days within a 2-wk interval. During the first visit, the subjects performed body plethysmography and cardiopulmonary exercise tests. During the second visit, thoracoabdominal kinematics was assessed via OEP at rest and during exercise on a cycle ergometer (70% maximum workload obtained in the cardiopulmonary exercise testing). Body plethysmography was performed as previously described (1085 Elite D; Med Graphics, St. Paul, MN) (26), and the parameters evaluated included VEF1, FVC, total lung capacity (TLC), inspiratory capacity (IC), and residual volume (RV). Air trapping was considered when the ratio RV/TLC exceeded 50%, following previous descriptions (6, 29). Predicted values of lung function were adjusted to the Brazilian population (7). The cardiopulmonary exercise test was performed using an integrated system (CardiO2 System; Med Graphics) following established recommendations (3) on a cycle ergometer (Corival 400; Lode, Groningen, The Netherlands), with a fixed speed of 60 rotations per minute and through a ramp protocol with a progressive increase of 5 W/min for COPD patients and 20 W/min for healthy controls. The subjects wore a nose-clip, and pulmonary exchange was measured using a breath-by-breath automated gas analysis system through a pneumotachograph (CPX/D; Med Graphics). The peak oxygen consumption (V̇o2peak), heart rate, blood pressure, subject’s perception of effort, and maximum workload were measured. The test was interrupted when the patient reached physical exhaustion, as determined by the following factors: reaching the patient’s maximum heart rate (<10% of the maximal predicted) or reaching a plateau or peak V̇o2 independent of the increased workload or a respiratory coefficient ≥1.10, with the subject unable to maintain either speed or load during the test (28). The optoelectronic plethysmography system was used (OEP System; BTS, Milan, Italy) to perform this analysis. OEP assesses respiratory kinematics by recording the movement of the chest wall using 89 retro-reflective markers with a sampling frequency of 100 Hz (18). Additionally, OEP estimates the total volume of the chest wall and the following three compartments: URC, LRC, and ABD. The total volume of the chest wall is equivalent to the sum of the compartmental volumes. During the OEP evaluation, we also calculated the inspiratory and expiratory times and minute ventilation. The method used by OEP to estimate chest wall volumes has been previously described (1, 4). Thoracoabdominal asynchrony (TAA) was quantified by OEP at two distinct times, at rest (Rest) and during exercise (70% maximum workload (70% Wpeak), obtained in the cardiopulmonary exercise testing) (Exercise). First, the subjects were instructed to comfortably sit on the cycle ergometer, and the OEP data were collected during this period. After 1 min of rest, subjects were instructed to pedal without load for 3 min followed by 2 min of progressively increased load, until obtaining the target workload (70% Wpeak), which was still maintained for 1 min, followed by 2 min of recovery. At each moment of data capture (at rest and exercise), six homogeneous and consecutive respiratory cycles were analyzed. One average cycle was extracted from the volume signals equivalent to the chest wall and from each of the three compartments (URC, LRC, and ABD). The average cycle was used to estimate the volume of each compartment and its percentage contribution to the total chest wall volume. In addition, the average cycles were used to apply the parameters of PA and PS. Both parameters were calculated among all compartments: URC vs. LRC, URC vs. ABD, and LRC vs. ABD. The PA was calculated on the basis of the Lissajous figure (Fig. 1) (12, 20). Briefly, the volume signals of the compartments were plotted against each other. By convention, the superior compartment was plotted on the vertical axis. The URC was plotted on the y-axis when compared with the LRC and ABD, respectively (URC vs. LRC and URC vs. ABD), whereas the LRC was plotted on the y-axis when compared with ABD (LRC vs. ABD). The opening angle of the figure indicates the level of asynchrony between the compartments. To calculate the opening angle of the figure, the PA parameter considers two distances: first, the distance of a line in the middle of the curve (50% of the y-axis) and parallel to the x-axis (line segment m) was calculated, followed by calculation of the total distance of the curve on the x-axis (line segment s). PA was calculated according to the equation θ = sin−1 (m/s) and could vary from 0° to 180°, where an angle of 0° represents perfect synchrony between the compartments and 180° indicates total asynchrony. In the present study, a positive angle means that the motion of the superior compartment (drawn on the y-axis) is followed by the motion of the inferior compartment (x-axis), and a negative angle indicates the opposite, as previously described (20). Fig. 1.Parameters of the phase angle (PA) and phase shift (PS) calculation. In the PA, line segment m indicates the volume displaced by the abdomen at 50% of the upper rib cage volume, while line segment s represents the total volume displaced by the abdomen. In the PS, a and b are the events where the signals changed direction. The difference between b and a is the time interval during which the compartments moved in opposite directions. This phase difference was based on paradoxical motion (11, 21). The PS parameter represented the percentage of the respiratory cycle in which the compartments were moving in opposite directions. PS used a derivative method to scan sample-by-sample variations in each compartmental signal, enabling the parameter to detect the exact moment (event), in which there was any change in the direction of the signals during the respiratory cycle (i.e., when one compartment turned from expanding to retracting). After these events were detected in both compartments, it was possible to compare them and to calculate the time interval during which they were moving in opposite directions. The ratio between this time interval and the total time of the respiratory cycle produced the value of PS; 0% represented perfect synchrony between the compartments, and 100% indicated total asynchrony. To compare PS with PA, we used a linear transformation to express PS in terms of degrees. A percentage of 0% was equivalent to 0°, and 100% corresponded to 180°. The signal convention adopted in the calculation of PA was maintained; a positive angle meant that the motion of the superior compartment was leading. The methods to calculate PA and PS were developed using a numerical computing environment (MatLab; The Mathworks, Natick, MA). Figure 1 shows a representative graph of both parameters. In the present study, COPD patients were classified as presenting thoracoabdominal asynchrony (TAA+) or not (TAA−) using the 95% confidence interval of the reference values of PA and PS obtained in healthy subjects (control group, CG). Patients were classified into the TAA+ group if they presented PA or PS values (the analyses were performed separately) outside of the 95% confidence interval (mean ± 2 SD) obtained in the control group, in at least one of the three comparative analyses (URC vs. LRS, URC vs. ABD, or LRC vs. ABD). The data are presented as means ± SD, unless otherwise reported. The Kolmogorov-Smirnov test verified the normality of the data. To compare the values of PA and PS and respiratory parameters among the groups (CG, TAA+, and TAA−), we used two-way ANOVA with post hoc tests based on the Holm-Sidak method. The t-test (parametric data) or Mann-Whitney U-test (nonparametric data) was used to compare baseline characteristics between healthy individuals and COPD patients. The categorical variables were compared by the χ2- or Fisher test. The level of significance was set to 5% (P < 0.05). The statistical procedures were implemented in statistical software (Sigma-Plot 12.1, Systat Software, San Jose, CA). RESULTSThe baseline characteristics of the subjects are presented in Table 1. Most COPD patients (n = 14; 64%) were diagnosed with severe airflow limitation (FEV1 <50% of predicted), and 12 (55%) had air trapping (RV/TLC >50%). Interestingly, 11/12 (92%) patients presenting air trapping also presented thoracoabdominal asynchrony (TAA) when assessed by the phase angle (PA), and 9/12 (75%) patients when assessed by the phase shift (PS).
Table 2 shows the range of PA and PS in the control group, and the number of COPD patients presenting TAA. Considering the comparison among all three compartments—the URC, LRC, and ABD—PA classified more patients as presenting TAA (TAA+) compared with PS, either at rest (15 vs. 7, respectively, P < 0.05) or during exercise (14 vs. 8). Additionally, every patient classified as TAA+ using PS was also classified as TAA+ using the parameter of PA both at rest and during exercise.
Table 3 presents the comparison of the demographic and anthropometric data, as well as the lung function values of COPD patients either presenting thoracoabdominal asynchrony (TAA+) or not (TAA−) by both PA and PS parameters. There were no differences between TAA+ and TAA− patients in the demographic and anthropometric characteristics or in the level of airway obstruction (FEV1 and FEV1/CVF). However, TAA+ patients evaluated by the parameter of PS presented higher values of RV and TLC than did TAA− patients.
Figures 2, A and B and 3, A and B compare the presence of TAA (TAA+) using PS and PA between URC vs. LRC and between LRC vs. ABD, respectively, at rest (Fig. 2, A and B) or during exercise (Fig. 3, A and B). No TAA was detected between URC vs. ABD at rest or during exercise using PA or using PS (data not shown). In contrast, TAA was observed when we compared LRC vs. URC and ABD. Between URC vs. LRC, the TAA+ group presented values of PA and PS higher than those of TAA− patients and controls at rest (Fig. 2, A and B) and during exercise (Fig. 3, A and B). Interestingly, similar results were obtained in the analysis of LRC vs. ABD compartments; however, here, the values in the TAA+ group were negative in both the PA and PS parameters, and values closer to 0 were obtained in the TAA− group and controls at rest (Fig. 2, A and B) and during exercise (Fig. 3, A and B). Fig. 2.Thoracoabdominal asynchrony at rest between the upper (URC) and lower (LRC) rib cages (A) and between the LRC and abdomen (ABD) (B). The results are presented as box plots, where the center line represents the median, and box limits indicate the 25th (lower limit) and 75th percentiles (upper limit). The lines above and below the box limits represent the largest and smallest values, respectively. On the basis of the signal convention, positive values imply which compartment leads the motion. The URC leads the motion in Fig. 2A, and ABD leads the motion in Fig. 2B. Control Group (CG); n = 13 for the PA and PS. TAA−, patients without asynchrony, according to the group distribution with PA (n = 7) and PS (n = 15); TAA+, patients with asynchrony (n = 15 for PA, and n = 7 for PS); °, degrees; *P < 0.05 compared with CG and TAA− using the same parameter; #P < 0.05 between PA and PS. P values were obtained from a two-way ANOVA. Fig. 3.Thoracoabdominal asynchrony during exercise between the upper (URC) and lower (LRC) rib cages (A) and between the LRC and the abdomen (ABD) (B). The results are presented as box plots, where the center line represents the median and box limits indicate the 25th (lower limit) and 75th (upper limit) percentiles. The lines above and below the box limits represent the largest and smallest values, respectively. On the basis of the signal convention, positive values indicate which compartment leads the motion. The URC leads the motion in A, and the ABD leads the motion in B. CG, n = 13 for the PA and the PS; TAA−, patients without asynchrony according to group distribution with PA (n = 8) and PS (n = 14); TAA+, patients with asynchrony (n = 14 for PA, and n = 8 for PS). *P < 0.05 compared with CG and TAA− using the same parameter. &P < 0.05 compared with CG using the same parameter. #P < 0.05 between PA and PS. P values were obtained from a two-way ANOVA. Finally, the TAA+ patients evaluated by PA or PS presented similar tidal volumes at rest compared with TAA− and the controls. However, during exercise, the TAA+ patients presented a lower chest wall volume than did the control individuals (Table 4), but this was not observed using the PS parameter (Table 5). In addition, the TAA+ patients also showed different patterns of thoracoabdominal contribution, such as a lower contribution at the LRC and a higher contribution at the ABD, compared with the TAA− patients and controls. Still, during exercise, the presence of asynchrony caused the TAA+ patients to develop a lower minute ventilation and air flow than did the controls (Table 4 and 5).
DISCUSSIONOur results suggest that the PA is better able to detect TAA during rest and exercise in COPD patients than is the PS. In addition, TAA was observed only when the LRC compartment was included in the analysis with either PA or PS. Finally, using the PA, we observed that COPD patients presenting TAA also presented a reduction in chest wall ventilation during exercise. Our results demonstrate that the PA was better able than the PS to classify patients with TAA (Table 2). Most of the COPD patients (68%) presented asynchrony when assessed with the PA. In fact, we observed that the PA not only detected more patients with TAA (TAA+) but also revealed that all patients classified as TAA+ using the PS were among those classified as TAA+ using the PA. The difference in detecting TAA by using PA versus PS might be a consequence of the distinct working principles of these parameters. The PA calculates the opening angle of the Lissajous figure (12, 20) (Fig. 1), which provides an overall analysis of the movement of each compartment and does not consider sample-by-sample signal variations. In contrast, the parameter PS applies a derivative method to detect changes in the volume signals that allows calculating the shift between thoracoabdominal compartments (21). Because TAA is essentially a time analysis issue regarding the lag between compartments (12), we hypothesize that the PS detected fewer patients with TAA because it measures the total time in asynchrony between the compartments, including the effect of electrical noise and movement artifacts (16, 21). This could potentially increase the range of values, even in healthy individuals, among all compartments when compared with the PA (Table 2). In contrast, the PA could have detected a higher number of COPD patients with TAA because its measurement is based on the motion of one compartment plotted against the other (Fig. 1). Thus, the distance used to calculate the opening angle of the figure does not consider sample-by-sample variations and minimizes the interference of artifacts during data collection. Consequently, it seems more likely that PS underestimates the presence of asynchrony rather than PA overestimating asynchrony, and it is reasonable to conclude that PA detects more TAA in COPD patients than PS does. Finally, it is important to note that the PS calculation is ultimately a double-normalized value (one is a normalization of time, the other being a normalization to an expression of degrees, for comparison with PA values). While this approach is clearly important for enabling an acceptable comparison between PA and PS values, normalizations can induce nonlinearities that may ultimately contribute to increased variation in the PS calculation, and its inability to track TAA with the same fidelity as PA. We only detected asynchrony when the analysis of LRC was involved. This observation contrasts with previous studies showing that TAA occurs between the upper rib cage and abdominal compartments (8, 13, 23). A possible explanation may be the fact that previous studies assessed TAA using equipment that only allowed an approach following the bicompartmental model (i.e., magnetometer and respiratory inductive plethysmography), and the authors performed the measurements in the upper rib cage and abdomen. In contrast, we quantified TAA using a three-compartment model that considered the upper and lower rib cage separately, and TAA in COPD patients was observed in the LRC. Distinct muscular and pressure forces influence the upper and the lower rib cage (1, 27). For instance, when COPD patients increase ventilation during exercise, there is greater activity of the accessory muscles (i.e., sternocleidomastoid and scalene), causing the upper rib cage to move mostly along the anterior-posterior axis (1, 20). Likewise, because of the action of the diaphragm during inspiration and the abdominal muscles during expiration, the displacement of the abdominal compartment in COPD largely occurs along the same axis (anterior-posterior) (20, 25). On the other hand, the lower rib cage moves mainly along the lateral-lateral axis, influenced by the joint action of the external intercostal muscles and the diaphragm between the xiphoid process and lower costal margin, leading to the “bucket handle movement” of the ribs (8, 9). Interestingly, our results reinforce the hypothesis that the three-compartment model better represents the kinematics of the chest wall because we observed that TAA+ patients had a decreased contribution of the LRC compared with controls and TAA− patients (Table 4). On the basis of our findings, it seems reasonable to conclude that COPD patients are more prone to develop TAA in the lower rib cage. These findings may have clinical implications because the analysis of TAA using the three-compartment model would allow us to detect TAA in clinical situations, such as during moderate exercise or daily physical activities, and not only during severe effort, which leads to a paradoxical respiration, as previously described using the two-compartment model (24). We also observed that TAA+ patients presented with reduced chest wall volume and minute ventilation during exercise along with lower air flow compared with controls (Table 4). In a review of the literature, Hammer and Newth hypothesized that TAA may decrease the tidal volume because the opposite movement of the compartments might lead to ventilatory inefficiency (12); however, this has never been reported. Our results support this hypothesis; additionally, we quantified a volume reduction caused by the presence of TAA in COPD patients, which occurred mostly in the lower rib cage (Table 5). In our patients, thoracoabdominal asynchrony was mainly observed in the lower rib cage that presented a lack of coordination related to the upper rib cage and the abdomen. Asynchrony between upper and lower rib cages has been previously reported and is known as the Hoover sign, a paradoxical indrawing of the lower rib cage (1). This distortion occurs at least partly because the action of external and internal intercostals in COPD patients may have distinct effects in every rib cage compartment (15). For example, the external intercostals play a key role in the expansion of the upper rib cage during inspiration; however, they can have an expiratory action in the lower rib cage (15). In addition, the parasternal muscles can change their usual expiratory action (as internal intercostals), into an inspiratory action in the lower rib cage (15). Moreover, COPD patients exhibit a flattening of the diaphragm dome as, for example, during air trapping (15, 19). These mechanical alterations of the respiratory muscles and the chest wall lead to asynchrony of the lower rib cage either with the upper rib cage and/or with the abdominal compartment. Nevertheless, we observed that the lower rib cage remains almost static during breathing and has a very poor contribution to the chest wall volume. We hypothesize that this lack of displacement of the lower rib cage in TAA+ COPD patients may reflect a compensatory pattern to stabilize the chest wall and enable, despite the mechanical disadvantage, a coordinated function between the upper rib cage and abdominal compartments. In our study, 11 of the 12 COPD patients (92%) who presented air trapping (RV/TLC >50%) (6, 29) were also classified as TAA+ during rest and/or exercise, reinforcing previous studies and suggesting an association between air trapping and TAA (1, 5, 24). The presence of air trapping and asynchrony in COPD patients has been shown to lead to compensative adaptations in the chest wall and the diaphragm. Examples of these adaptations are an increase in the recruitment of the diaphragm (15), which moves in a narrower and lowered space of the lower chest wall (28), and a reduced abdominal pressure associated with the inward movement of the lower rib cage during asynchronous breathing (15). These adaptations have at least two effects: 1) increasing both the abdominal volume during inspiration and the strength and endurance of the diaphragm (15, 17) and 2) decreasing the mechanical efficiency of the respiratory muscles (e.g., external intercostals and parasternals), limiting the range of action of these muscles and restricting the expansion of the rib cage (15, 17). Altogether, these adaptations support our findings showing that TAA+ patients present a higher contribution of the abdomen than the rib cage compartment (Table 4). Considering that the three-compartment model establishes the lower rib cage between the xiphoid process and lower costal margin (27), an action below this region could cause the diaphragm to influence the lower rib cage less and to reduce its motion. Consequently, this should also increase the abdominal contribution because the diaphragm would act below the costal margin and directly move the abdominal compartment. Interestingly, our results also show that TAA+ patients exhibited a greater abdominal contribution to the chest wall volume compared with controls and TAA− patients (Tables 4 and 5). These results are supported by Priori et al. (20), who also showed that the abdominal volume is associated with the diaphragmatic displacement of the zone of apposition in TAA+ COPD patients. In summary, our results suggest that the presence of TAA in COPD patients leads to a ventilatory inefficiency that occurs mainly in the lower rib cage. Our study has some limitations. First, the OEP assessment was performed on the cycle ergometer, and to avoid obstruction of the cameras, the patients were asked to support their arms using sticks while pedalling. However, this is the only method to perform OEP during exercise and has been reported in other studies (1, 25) and described by the manufacturer of the OEP system (2). Second, we included males and females, and the women’s breasts may have interfered with the OEP evaluation. In addition, respiratory movements might change according to sex, and our control group included 38% of men, whereas in COPD patients, this percentage was 59%. However, the statistical analysis showed that the proportion of both sexes was similar between the COPD patients and control subjects. Third, we did not assess the respiratory muscle activity via electromyography because it was not an aim of this study. Nevertheless, future researchers may conduct this assessment to advance knowledge in this area. In conclusion, the parameter PA was better able to detect thoracoabdominal asynchrony at rest and during exercise than was PS. Thoracoabdominal asynchrony occurred in the lower rib cage and reduced the chest wall volume in COPD patients during exercise. The three-compartment model seems more appropriate for detecting asynchrony. GRANTSThis study was supported by São Paulo Research Foundation Grants 2010/50120-4 and Conselho Nacional de Pesquisa Grant 311443/2014-1. This research was conducted as part of D. Cano Porras’s Masters of Science degree, which was supported by the Comissão de Aperfeiçoamento de Pessoal do Nível Superior. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSD.C.P., A.C.L., C.C.M.dS., and D.M.P. performed experiments; D.C.P., A.C.L., C.C.M.dS., D.M.P., H.T.M., and C.R.F.C. analyzed data; D.C.P., A.C.L., C.C.M.dS., D.M.P., H.T.M., and C.R.F.C. interpreted results of experiments; D.C.P., H.T.M., and C.R.F.C. prepared figures; D.C.P., A.C.L., C.C.M.dS., D.M.P., R.S., H.T.M., and C.R.F.C. drafted manuscript; D.C.P., A.C.L., C.C.M.dS., D.M.P., R.S., H.T.M., and C.R.F.C. edited and revised manuscript; D.C.P., A.C.L., C.C.M.dS., D.M.P., R.S., H.T.M., and C.R.F.C. approved final version of manuscript. REFERENCES
Page 10although it is very well established that muscle wasting is a common finding in a variety of prolonged hypoxemic conditions, including chronic obstructive pulmonary diseases (15), congestive heart failure (41), and exposure to high altitudes (22), the deleterious effects of intermittent hypoxia (IH), a key clinical manifestation of obstructive sleep apnea (33), on skeletal muscle protein metabolism is still unclear. Skeletal muscle mass is maintained by a fine balance between protein synthesis and protein breakdown (31, 36). The three main proteolytic processes involved in the control of muscle protein metabolism in mammals are the lysosomal process, the Ca2+-dependent process, and the ubiquitin (Ub)-proteasome system (UPS). The acid hydrolases in lysosomes degrade the majority of extracellular and membrane proteins taken up by endocytosis as well as cytoplasmic proteins and organelles from autophagy (54). The Ca2+-dependent process contains at least two ubiquitous enzymes: calpain 1 or µ-calpain (low Ca2+-requiring form) and calpain 2 or m-calpain (high Ca2+-requiring form). Fagan et al. (18) were the first to demonstrate that intracellular proteins can be degraded by calpains in isolated chick skeletal muscle depleted of ATP in vitro. Other studies have shown that pharmacological inhibition of calpain prevents loss of cell function and/or integrity of membranes associated with tissue ischemia (32). Although this in vitro evidence (32) has indicated that calpains play a main role in hypoxic injury, the precise contribution of each proteolytic system to the enhanced breakdown rates observed in skeletal muscles in in vivo conditions remains to be elucidated. The UPS is the main intracellular pathway for protein degradation (48) that is activated in different systemic wasting conditions, including insulin resistance (67), diabetes (46), and treatment with excessive glucocorticoids (66). Among the UPS components are the muscle specific Ub-protein ligases (E3s), atrogin-1/MAFbx (muscle atrophy F-box), and MuRF-1 (muscle ring finger protein), whose induction precedes the loss of muscle weight (6, 26). For this reason, these Ub-ligases are up to now the best markers for muscle atrophy (11). Like atrogin-1 and MuRF1, genes associated with autophagy including LC3b (microtubule-associated protein 1 Light Chain 3) and GABARAPl1 (GABA receptor-associated protein), homolog to yeast Atg8, are also activated in catabolic situations, including hypoxia (4), and are considered atrophy-related genes or atrogenes (40, 51). In vitro studies have shown that atrogin-1 level was significantly increased in L6 myotubes after 24 h in hypoxic conditions, whereas MuRF1 level did not change (9). More recently, hypoxia in vivo for 4 days was shown to activate atrogin-1 and MuRF1 mRNAs in mice (61), while atrogin-1 mRNA was decreased in humans (13). However, the effect of a few hours of IH on the UPS activity has never been addressed. This is certainly important, because impairment of muscle protein metabolism is rapidly observed in critically ill patients hospitalized with diseases related to hypoxia such as sepsis and acute respiratory disorders (34). During stress conditions, several behavior, visceral, and endocrine changes are activated as part of homeostatic control (1). Previous studies from our (71) and other (39) laboratories have shown that chronic IH in rats activates the hypothalamic-pituitary-adrenal axis, resulting in increases in circulating corticosterone. Moreover, it was demonstrated that elevated plasma glucocorticoids are associated with the pathogenesis of insulin resistance in obese mice subjected to chronic IH (45). Thus we hypothesized that acute IH (AIH) would alter protein homeostasis in rat skeletal muscles through a glucocorticoid-dependent mechanism. Another attractive possibility is that the effects of AIH on protein metabolism are linked to disturbances of mitochondrial morphology, a dynamic process that is governed by well-ordered fusion and fission processes (37). Key proteins that promote mitochondrial fission include Dynamin1-like (DNM1L) and Fission 1 (Fis-1), whereas proteins regulating mitochondrial fusion include optic atrophy 1 (Opa1) and the large GTPases mitofusins 1 and 2 (MFN1/2) (16). Our approach was to utilize an established rodent model of short-term (8 h) IH (21) and investigate the rate of protein turnover (protein synthesis and degradation), the activity of three proteolytic systems (lysosomal, Ca2+ dependent, and UPS), and the genes involved in muscle atrophy, autophagy, and mitochondrial dynamics in soleus, a slow-twitch skeletal muscle, and in extensor digitorum longus (EDL), a typical fast-twitch skeletal muscle. We also determined whether the deleterious response on protein metabolism under hypoxic conditions would be abrogated by plasma depletion of glucocorticoids induced by adrenalectomy (ADX). We found that AIH induces insulin resistance associated with activation of the UPS, the autophagic-lysosomal process, and Ca2+-dependent proteolysis in both muscles through a glucocorticoid-dependent mechanism. METHODSThe incubation procedure required intact muscles of a sufficient thinness to allow for adequate diffusion of metabolites and oxygen (26); thus fed 4-wk-old male Wistar rats (~90 g) were used for all experiments. The following four experimental groups were used: 1) sham-operated rats maintained under normoxic conditions; 2) adrenalectomized rats maintained under normoxic conditions; 3) sham-operated rats exposed to AIH; and 4) adrenalectomized rats exposed to AIH. To avoid interference of food changes on the effects of hypoxia on proteolysis, a separate group of animals were maintained inside normoxia/hypoxia chambers without food. To investigate the in vivo effect of catecholamines secreted by adrenal medulla on proteolytic activities, another group of animals were adrenodemedullated and exposed or not to AIH. All rats were euthanized by cervical dislocation, with the exception of a group of animals that were euthanized by decapitation for metabolic and hormone measurements. Animals were obtained from the Animal Care Facility of the University of São Paulo at Ribeirão Preto. Skeletal muscles were harvested from rats after death according to the Ethical Principles in Animal Research adopted by the Brazilian Council for Animal Experimentation (CONCEA) and protocols approved by the Institutional Ethical Commission on Ethics in Animal Research (CEUA: 157/2008, 134/2011, and 091/2013), Ribeirão Preto Medical School, University of São Paulo. The term “AIH” in this work designates an experimental procedure whereby short periods of hypoxia [fraction of inspired O2 (FIO2) of 6% over 40 s] are interspersed with periods of normoxia (FIO2 of 20.8% over 9 min) during the 8 h of the light period. For this, all groups were housed in collective cages and maintained inside Plexiglas chambers (volume of 210 liters) equipped with gas injectors as well as sensors of O2, CO2, humidity, and temperature. Rats submitted to AIH were exposed to an experimental protocol of 5 min of normoxia (FIO2 of 20.8%) followed by 4 min of pure N2 injection to reduce FIO2 from 20.8% to 6%, remaining at this level for 40 s. After this hypoxic period, pure O2 was injected into the chamber to return FIO2 back to 20.8%. This 9-min cycle was repeated over 8 h (from 9:30 AM to 5:30 PM). The injections of N2 and O2 (White Martins, Sertãozinho, Brazil) into the chamber were regulated by a solenoid valve system whose opening-closing control was operated by a computerized system (Oxycycler, Biospherix).The slope of oxygen changes in the chamber used in the present study has already been published (72). In an identical chamber in the same room, the control group of rats were exposed to a FIO2 of 20.8% for 8 h. The control rats were also exposed to a similar valve noise due to the frequent injection of O2 to maintain FIO2 at 20.8%. In both AIH and control chambers, the gas injections were performed at the upper level of the chamber in order to avoid direct jets of gas impacting on the animals, which could cause stress. Rats were euthanized at 5:30 PM, immediately after the last cycle of hypoxia. Two separate groups of fed animals were also used: one euthanized at 11:30 AM (i.e., 2 h after the last cycle of hypoxia) and the other euthanized on the following day at 8:30 AM (i.e., 15 h after the last cycle of hypoxia). ADX was performed bilaterally under xylazine-ketamine anesthesia (10 and 85 mg/kg body wt ip, respectively) 8 days before use of the rats in experiments. For this, a 2-cm dorsal midline skin incision was made at the level of the 13th rib. The adrenal glands, located cranial and medial to the kidney, were removed. Sham-operated rats were subjected to the same procedure without adrenal gland extirpation. Because aldosterone is an adrenal hormone that exerts a key role in renal sodium resorption (30), after surgery 1% saline was provided ad libitum to the ADX rats and water was given to sham-operated animals. Adrenodemedullation (ADMX) was performed bilaterally under xylazine-ketamine anesthesia (10 and 85 mg/kg body wt ip, respectively) 10 days before use of the rats in experiments. Each adrenal medulla was squeezed through a nick made on its capsula. The animals did not require saline in their drinking water after surgery. Soleus and EDL muscles were rapidly and carefully dissected, avoiding damage to the muscles. The muscles were maintained on inert supports with pins and incubated at 37°C in Krebs-Ringer bicarbonate buffer, pH 7.4, equilibrated with 95% O2-5% CO2, containing glucose (5 mM), and in the presence of cycloheximide (0.5 mM) to prevent protein synthesis and the reincorporation of tyrosine back into proteins. Tissues were preincubated for 1 h and then incubated for 2 h in fresh medium of identical composition. The rates of overall proteolysis and of the different proteolytic systems were determined by measuring the rate of tyrosine release in the incubation medium. Because muscles cannot synthesize or degrade tyrosine, its release reflects the rate of protein breakdown. Tyrosine was assayed as previously described (65). It is well established that insulin and branched-chain amino acids inhibit the uptake and hydrolysis of cytosolic proteins within enlarged lysosomes (31). Thus, to measure the contribution of the lysosomal apparatus, muscles from two limbs of each animal were incubated in the absence or presence of insulin (1 U/ml) and branched-chain amino acids (leucine, 170 μM; isoleucine, 100 μM; valine, 200 μM) to inhibit the lysosomal process (31). A second protocol was used to quantitate the contribution of the lysosomal apparatus by incubating muscles with or without methylamine, a weak base that blocks lysosomal acidification. The difference of tyrosine release between the two muscles reflects the proteolytic lysosomal activity. To test the activity of Ca2+-dependent proteolysis, muscles from one limb were incubated in a Ca2+-free medium that contained, in addition to insulin and branched-chain amino acids, E-64 (50 μM) and leupeptin (25 μM), inhibitors of thiol proteases (calpains). The contralateral muscles were incubated in the presence of Ca2+ and the absence of E-64 and leupeptin. The difference between the two muscles estimates the activity of the Ca2+-dependent proteolytic process. For measurement of UPS activity, muscles from contralateral limbs were incubated under conditions that prevent activation of the lysosomal [10 mM methylamine and 1 U/ml insulin and branched-chain amino acids (170 μM leucine, 100 μM isoleucine, and 200 μM valine)] and Ca2+-dependent (Ca2+-free medium with cysteine-protease inhibitors, including 50 μM leupeptin) proteolytic systems. Muscles from left limbs were incubated in addition with the proteasome inhibitor MG132 (20 μM). UPS activity was calculated by the difference of tyrosine release between left and right muscles. Soleus and EDL muscles were harvested and incubated in the presence or absence of insulin (0.1 U/ml), as described above, for 2 h at 37°C in Krebs-Ringer bicarbonate buffer, pH 7.4, equilibrated with 95% O2-5% CO2, containing glucose (5 mM), l-[U-14C]tyrosine (0.05 μCi/ml), and all 20 amino acids at concentrations similar to those of rat plasma (53). At the end of this period the specific activity of acid-soluble tyrosine (intracellular tyrosine pool) in each muscle was estimated by measuring in this pool the radioactivity and the concentration of tyrosine. After measurement of the radioactivity incorporated into protein of the same muscle, the rate of protein synthesis was calculated with the specific activity of the intracellular pool of tyrosine, assuming that there was no recycling of the label during the incubation period. Soleus and EDL muscles were incubated in Krebs-Ringer buffer (pH 7.4, 5 mM glucose) in a specific flask to avoid gas exchanges with the environment. After one 1-h preincubation, the medium was replaced by the same buffer with the addition of [U-14C]glucose (0.05 µCi/µmol) and muscles were incubated for 2 h at 37°C while shaking gently. Hyamine, used to trap the CO2 produced, was injected in a container inside the flask, and sulfur acid (6 N) was added to the medium to stop the reaction. The liquid in the container with CO2 was removed after 2 h from the flask, and the radioactivity was counted (TRI-CARB 2100TR, Packard BioScience). Soleus and EDL muscles were harvested and immediately frozen in liquid nitrogen. RNA was isolated with TRIzol (Invitrogen, Carlsbad, CA) from individual skeletal muscles (n = 7). Reverse transcription into cDNA was performed with 2 µg of total cellular RNA, 20 pmol of oligo(dT) primer (Invitrogen), and Advantage ImProm-II reverse transcriptase (Promega, Madison, WI). Atrogin-1, MuRF1, GABARAPl1, LC3b, FIS1, FIS2, MFN1, and MFN2 mRNA levels were determined by real-time reverse transcription polymerase chain reaction (RT-qPCR) in soleus and EDL muscles with the SuperScript III Platinum SYBR Green One-Step RT-qPCR Kit with ROX (Invitrogen) and the ABI7000 sequence detection system (Applied Biosystems, Foster City, CA). The following settings were used: stage 1 (reverse transcription), 48°C for 30 min; stage 2 (denaturation), 95°C for 10 min; and stage 3 (PCR), 95°C for 15 s and 60°C for 60 s for 40 cycles. The qPCR primers used were as follows: Atrogin-1 (forward 5′-CTT TCA ACA GAC TGG ACT TCT CGA-3′; reverse 5′-CAG CTC CAA CAG CCT TAC TAC GT-3′), MuRF1 (forward 5′-TCG ACA TCT ACA AGC AGG AA-3′; reverse 5′-CTG TCC TTG GAA GAT GTC TT-3′), GABARAPl1 (forward 5′-CTT TCC CCT TGT TTA CCC TCC AT-3′; reverse 5′-CCC AAT GTC AAC CCC TTC-3′), LC3b (forward 5′-TTT GTA AGG GCG GTT CTG AC-3′; reverse 5′-CAG GTA GCA GGA AGC AGA GG-3′), Fis-1 (forward 5′-GCA CGC AGT TTG AAT ACG CC-3′; reverse 5′-CTG CTC CTC TTT GCT ACC TTT GG-3′), DNM1L (forward 5′-AGA ATA TTC AAG ACA GCG TCC CAA AG-3′; reverse 5′-CGC TGT GCC ATG TCC TCG GAT TC-3′), MFN1 (forward 5′-CCT TGT ACA TCG ATT CCT GGG TTC-3′; reverse 5′-CCT GGG CTG CAT TAT CTG GTG-3′), MFN2 (forward 5′-GAT GTC ACC ACG GAG CTG GA-3′; reverse 5′-AGA GAC GCT CAC TCA CTT TG-3′), β-actin (forward 5′-CAC CCG CGA GTA CAA CCT TC-3′; reverse 5′-CCC ATA CCC ACC ATC ACA CC-3′), and cyclophilin (forward 5′-AAG GAC TTC ATG ATC CAG GG-3′; reverse 5′-TGA CAT CCT TCA GTG GCT TG-3′). Fluorescence data were analyzed by SDS 1.7 software (Applied Biosystems). The relative quantification of mRNA levels was plotted as the fold increase in comparison to the values of the control group. Transcripts of interest were normalized to levels of cyclophilin B, a housekeeping gene that did not appreciably vary in previous studies (27, 55) and in the present study, and target transcript mRNA levels were calculated with the standard curve method (35). Ub-protein conjugates, μ-calpain, m-calpain, and caspase-3 were measured by Western blotting analysis. For total proteins (μ-calpain and caspase-3), soleus muscles were harvested (n = 8) and homogenized in 20 mM Tris·HCl and 5 mM EDTA, pH 7.5, in the presence of 1 mg/ml leupeptin, 100 mg/ml 4-(2-aminoethyl) benzenesulfonyl fluoride, and 5 µg/ml aprotinin. The homogenate was centrifuged at 14,000 g at 4°C for 20 min, retaining the supernatant, which was stored at −80°C. For myofibrillar protein fractions (Ub-protein conjugates), EDL muscles were harvested and homogenized in 5 mM Tris·HCl, pH 8.0, 5 mM EDTA, 5 mM DTT, 50 µM MG132, in the presence of 1 mg/ml leupeptin, 100 mg/ml 4-(2-aminoethyl)benzenesulfonyl fluoride, and 5 µg/ml aprotinin. Myofibrillar proteins were pelleted by centrifugation at 1,500 g for 5 min at 4°C. Pellets were washed three times in the same buffer containing 1% Triton X-100 and resuspended in 8 M urea, 50 mM Tris·HCl, pH 7.5. Supernatants were stored at −80°C. Protein content (total proteins and myofibrillar protein fractions) were determined in duplicate by the Lowry method (38) using BSA as a standard. An equal volume of sample buffer (20% glycerol, 125 mM Tris·HCl, 4% SDS, 100 mM dithiothreitol, 0.02% bromophenol blue, pH 6.8) was added to the homogenate, and the mixture was boiled and subjected to SDS-PAGE on 8% acrylamide gels. Gels were electroblotted onto nitrocellulose membranes, and primary antibodies (Abs) were used to detect the corresponding protein levels: rabbit anti-μ-calpain (1:750), rabbit anti-m-calpain (1:750), rabbit anti-Ub-protein conjugates (1:1,000), rabbit anti-caspase-3 (1:750), rabbit anti-hypoxia-inducible factor-1α (HIF-1α) (1:1,000) and mouse anti-β-actin (1:2,000). Abs for µ-calpain and m-calpain were from Cell Signaling Technology; for Caspase-3, Ub-protein conjugates, and β-actin from Santa Cruz Biotechnology; and for HIF-1α from Novus Biologicals. Primary rabbit and mouse Abs were detected by peroxidase-conjugated secondary anti-rabbit (1:5,000) and anti-mouse (1:5,000) Abs, respectively, and visualized with an ECL Western blot detection kit (Amersham Biosciences, Piscataway, NJ). Band intensities were quantified with the ImageJ program (version 1.41o, National Institutes of Health, Bethesda, MD). The Ub-protein conjugates were assessed as the sum of densitometric values of all bands (i.e., the entire lane). The densitometry of bands was analyzed three times, and the average value was reported. Densitometric values were normalized to β-actin for each sample and divided by the average of the normoxia group (Control), which was considered to be 1. Plasma glucose concentration was determined by a glucose oxidase method (Glucox 500; Doles, Goiânia, Brazil). Plasma and muscle catecholamines were measured as previously described (24) by HPLC (LC-7A, Shimadzu Instruments) with a 5-μm Spherisorb ODS-2 reverse-phase column (Sigma-Aldrich). Insulin was measured by radioimmunoassay with a commercial kit (DPC-MEDLAB, São Paulo, Brazil), and corticosterone was evaluated according to Elias et al. (17). Insulin sensitivity was assessed from fasting insulin and glucose levels and by the previously validated homeostasis model assessment (HOMA) index in rats (8) as follows: HOMA-IR = fasting glucose (mmol/l) × fasting insulin (AU/ml)/22.5. For this, blood samples were collected immediately after the last cycle of hypoxia, after 12 h of food deprivation. Plasma glucose and insulin were measured according to the methods described above. Regular insulin (0.75 U/kg body wt) was intravenously injected (jugular vein). Blood glucose levels were measured on samples obtained from the tail vein with a glucometer (Accu-Chek) at 0, 4, 8, 12, and 16 min after insulin injection. The corresponding 4–16 min values were used to calculate the glucose disappearance constant (kITT; obtained in insulin tolerance test) according to Bonora et al. (7). The distribution and variance homogeneity were tested, and the appropriate statistical test was employed (as indicated in figures and table). Multiple comparisons were made with two-way ANOVA and the Bonferroni test. Unpaired Student’s t-test was used for comparing two groups. The program Prism 5.0 (GraphPad, San Diego, CA) was used to conduct the statistical analyses. Data are expressed as means ± SE. Differences were considered significant at P < 0.05. RESULTSAIH increased plasma levels of corticosterone, glucose, and insulin but drastically reduced the rate of glucose oxidation in EDL muscles, suggesting insulin resistance (Table 1). HOMA-IR was significantly increased, whereas kITT obtained in the insulin tolerance test was decreased (Table 1), confirming the decreased insulin sensitivity of the hypoxic rats. Neither plasma levels of norepinephrine (3.6 ± 0.4 ng/ml in control rats vs. 4.3 ± 0.8 ng/ml in hypoxic rats; n = 8) and epinephrine (9.4 ± 1.2 ng/ml in control rats vs. 8.6 ± 1.9 ng/ml in hypoxic rats; n = 8) nor the muscle content of norepinephrine (100 ± 7 ng/g in control rats vs. 92 ± 6 ng/g in hypoxic rats; n = 8) was affected by hypoxia. Neither body weight nor muscle weight was affected by AIH (data not shown). Collectively, these data demonstrate that AIH activates the secretion of corticoids and induces insulin resistance.
Soleus and EDL muscles from fed rats exposed to AIH were isolated to evaluate the in vitro rates of overall protein synthesis and protein degradation and the activity of various proteolytic systems. In both soleus and EDL muscles, rates of overall proteolysis increased after exposure to hypoxia (Fig. 1A), an effect that was no longer observed 15 h after the hypoxic stress (soleus: 0.297 ± 0.006 nmol Tyr·mg−1·2 h−1 in control rats vs. 0.286 ± 0.018 nmol Tyr·mg−1·2 h−1 in hypoxic rats; EDL: 0.159 ± 0.016 nmol Tyr·mg−1·2 h−1 in control rats vs. 0.190 ± 0,009 nmol Tyr·mg−1·2 h−1 in hypoxic rats; n = 6). The rise in proteolysis was also observed in animals maintained inside the chambers without food (Fig. 1B) and in fed animals exposed to hypoxia only for 2 h (soleus: 0.352 ± 0.022 nmol Tyr·mg−1·2 h−1 in control rats vs. 0.427 ± 0.017 nmol Tyr·mg−1·2 h−1 in hypoxic rats; EDL: 0.217 ± 0.018 nmol Tyr·mg−1·2 h−1 in control rats vs. 0.307 ± 0.015 nmol Tyr·mg−1·2 h−1 in hypoxic rats; n = 4–6; P < 0.05). In contrast to proteolysis, the baseline rate of protein synthesis in soleus and EDL muscles did not differ between AIH and control rats. Consistently, the addition of insulin in vitro stimulated the rate of muscle protein synthesis equally in both groups (Fig. 1, C and D). Fig. 1.Effect of AIH for 8 h (6% O2 for 40 s at 9-min intervals) on overall proteolysis in fed (A) and fasted (B) rats, and on rates of protein synthesis in muscles from fed rats incubated in the presence (+Ins) or absence of insulin (C and D). Animals were maintained inside chambers with (fed) or without (fast) food for 8 h. Values are means ± SE of 5–8 muscles. Student’s t-test was used for statistical analyses.*P < 0.05 for AIH vs. control. †P < 0.05 for presence of insulin vs. absence. As shown in Fig. 2, the overall proteolysis rise observed in soleus and EDL muscles from AIH rats was accompanied by a 17% and 40% increased activity of UPS, respectively. Furthermore, AIH induced a 20% increase in the levels of Ub-protein conjugates (Fig. 3) and a twofold increase in gene expression of atrogin-1 and MuRF-1, two key Ub-protein ligases involved in muscle atrophy (Fig. 4). Fig. 2.Proteolytic activities of ubiquitin (Ub)-proteasome system (UPS) and lysosomal and Ca2+-dependent pathways in soleus (A) and EDL (B) muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Values are means ± SE of 6–8 muscles. Student’s t-test was used for statistical analyses.*P < 0.05 for AIH vs. control. Fig. 3.Total level of ubiquitin (Ub)-protein conjugates in myofibrillar fractions (A) and protein levels of μ-calpain [full-length (80-kDa form), autolyzed (76-kDa form), and total content; B], m-calpain (C), and caspase-3 (D) in EDL muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Values are means ± SE of 4–8 muscles. Student’s t-test was used for statistical analyses.*P < 0.05 for AIH vs. control. Fig. 4.Effect of adrenalectomy (ADX) on mRNA expression of Ub-ligases (atrogin-1 and MuRF1; A and C) and autophagic markers (LC3b and GABARAPl1; B and D) in soleus and EDL muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Values are means ± SE of 6–10 muscles. A.U., arbitrary unit. A 2-way ANOVA and the Bonferroni test were used for statistical analyses.*P < 0.05 for AIH vs. control. AIH increased the activity of lysosomal system estimated in isolated soleus (80%) and EDL (~1.5-fold) muscles (Fig. 2) by two different protocols. Consistently, AIH increased the mRNA expression of the autophagy-related genes LC3b (2-fold) and GABARAPl1 (2-fold) in both soleus and EDL muscles (Fig. 4). In both soleus and EDL, the proteolytic effect of AIH was paralleled by activation (50%) of Ca2+-dependent proteolysis (Fig. 2). As previously described, removal of the NH2-terminal region of the 80-kDa catalytic subunit of the calpains by autolysis can be used as a marker of calpain activation (70). We thus assessed µ-calpain autolysis in the soluble fraction of soleus muscles extracts. As shown in Fig. 3, AIH increased μ-calpain autolysis (76-kDa form), while the full-length form (80-kDa form) was significantly reduced. The total content of μ-calpain, m-calpain, and caspase-3 was not altered by hypoxia (Fig. 3). To investigate the role of glucocorticoids in the AIH-induced carbohydrate and protein metabolism changes, a group of animals was submitted to ADX and exposed to hypoxia. As expected, ADX reduced plasma levels of corticosterone in rats submitted to normoxia (3.0 ± 0.6 vs. 8.0 ± 1.9 µg/dl in sham operated; n = 5, P < 0.05) and abolished the peak of this hormone induced by hypoxia (3.9 ± 0.6 vs. 17.9 ± 3.4 µg/dl in sham operated; n = 5). The higher plasma levels of glucose (125 ± 3 mg/dl in AIH vs. 83 ± 8 mg/dl in AIH + ADX rats; n = 5, P < 0.05) as well as the lower kITT index (2.4 ± 0.4%/min in AIH vs. 5.5 ± 0.5%/min in AIH + ADX rats; n = 5, P < 0.05) induced by hypoxia were prevented by ADX. ADX did not affect the proteolytic activity in mice exposed to normoxia but completely blocked the stimulatory effect of AIH on mRNA levels of Ub-ligases atrogin-1 and MuRF-1 in both soleus and EDL muscles (Fig. 4). The transcriptional activation of LC3b and GABARAPl1 induced by AIH was also abolished in adrenalectomized rats (Fig. 4). As shown in Fig. 5, there was a tendency toward increased expression of mitochondrial fusion genes (MFN1 and MFN2) in EDL muscles from rats exposed to hypoxia, but this did not reach statistical significance. ADX significantly suppressed expression of MFN2 mRNA by >50% in hypoxic rats, while MFN1 expression showed trends toward lower expression that were not statistically significant (Fig. 5). The expression of mitochondrial fission genes (Fis-1 and DNM1L) were not altered by either AIH or ADX (Fig. 5). The rise in overall proteolysis, UPS, and lysosomal and Ca2+-dependent proteolysis observed in rats exposed to hypoxia was also blocked by ADX in EDL muscles (Fig. 6). Similar findings were observed in soleus (data not shown). Since the removal of adrenal gland leads to the depletion of hormones produced by the adrenal cortex and medulla, we investigated whether or not the surgical removal of adrenal medulla (ADMX) would affect the AIH-induced proteolysis increase in EDL muscles. As previously shown (28), ADMX depleted plasma levels of epinephrine but did not affect plasma levels of corticosterone (data not shown). In contrast to ADX, ADMX did not block the stimulatory effect of AIH on UPS activity (0.068 ± 0.005 nmol Tyr·mg−1·2 h−1 in Control + ADMX vs. 0.087 ± 0.005 nmol Tyr·mg−1·2 h−1 in AIH + ADMX rats; n = 7; P < 0.05). To investigate whether or not the AIH-inducible metabolic alterations would be associated with HIF-1α changes, protein levels of this transcriptional factor were measured. The protein content of HIF-1α was significantly higher (~5-fold) in soleus compared with EDL under normoxia and hypoxia conditions. Neither AIH nor ADX altered the HIF-1α protein levels in both soleus and EDL muscles (Fig. 7). Fig. 5.Effect of adrenalectomy (ADX) on mRNA expression of mitochondrial fission (Fis-1 and DNM1L; A) and fusion (MFN1 and MFN2; B) markers in EDL muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Values are means ± SE of 6–10 muscles. A 2-way ANOVA and the Bonferroni test were used for statistical analyses. *P < 0.05 for AIH vs. control. Fig. 6.Effect of adrenalectomy (ADX) on overall proteolysis (A), ubiquitin (Ub)-proteasome system (UPS) activity (B), capacity of lysosomal degradation (C), and Ca2+-dependent proteolytic activity (D) in EDL muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Data for sham-operated animals (control and AIH) are the same as presented in Fig. 2. Values are means ± SE of 5–8 muscles. A 2-way ANOVA and the Bonferroni test were used for statistical analyses.*P < 0.05 for AIH vs. control. Fig. 7.Effect of adrenalectomy (ADX) on protein content of HIF-1α in soleus (A) and EDL (B) muscles from rats exposed to 8 h of AIH (6% O2 for 40 s at 9-min intervals). Values are means ± SE of 6–8 muscles. A 2-way ANOVA and the Bonferroni test were used for statistical analyses. DISCUSSIONThe present work shows that intermittent exposure to normobaric hypoxia for a short period of time (8 h) in fed rats elevates plasma corticosterone and induces insulin resistance, an effect that is accompanied by an increase in the rate of overall proteolysis in both soleus and EDL muscles. These effects are transitory, since they are no longer observed in animals euthanized 15 h after the end of the hypoxia protocol, but cannot be explained by reduced energy uptake because similar findings were observed in animals exposed to a very short period of hypoxia as early as 2 h and in another group maintained in the chamber for 8 h without food. That the rise in protein degradation after 8 h of hypoxia exposure was probably a direct consequence of the increase in plasma corticosterone is clearly indicated by the demonstration that these catabolic effects are not observed in adrenalectomized animals. Indeed, it has been extensively shown that treatment with the synthetic glucocorticoid dexamethasone induces a rapid increase in the rate of protein degradation in skeletal muscles (12, 69) and in muscle cells (2). The fact that in the normoxic control animals overall proteolysis in soleus and EDL muscles from adrenalectomized animals is not lower than in those from sham-operated control animals is in agreement with the previous demonstration (69) that basal protein breakdown is not dependent on glucocorticoid, a hormone that acts predominantly under stress conditions (e.g., fasting). As muscle protein degradation was elevated in animals under AIH, we further investigated the activity of different proteolytic pathways. The data clearly show that hypoxia amplified the lysosomal activity as well as activities of UPS and Ca2+-dependent proteolysis, which were accompanied by upregulation of Ub-conjugates and µ-calpain autolysis, respectively, and eventually led to increased degradation of skeletal muscle proteins. Interestingly, a few hours of hypoxia increased expression of genes involved in proteasomal (atrogin-1 and MuRF1) and lysosomal/autophagic (LC3b and GABARAPl1) protein degradation in both soleus and EDL muscles. In agreement with these results, it has been shown that 24-h exposure of L6 myotubes to 1% O2 increased both chymotrypsin-like and caspase-like proteasomal activities (9). Moreover, expression of Bnip3, an autophagic gene, and atrogin-1 mRNAs was increased in mice subjected to sustained hypoxic conditions for 4 days (25). On the other hand, 8-h exposure to intermittent 6% hypoxia for 4 days in mice induced preferential atrophy of the diaphragm, which correlated with greater activation of the autophagy but not calpain or muscle-specific E3 ubiquitin ligase pathways (25). The conflicting results could be attributed to differences in the type of muscle fibers, species, and age of animals studied as well as the protocol of hypoxia exposure used. Caspase-3 is another cysteine protease that may alter muscle protein metabolism under stress conditions (57); however, the present data show that its level as well as that of m-calpain were not changed by hypoxia. It has been reported recently that fast-twitch muscles such as the EDL are more sensitive to sustained 8% hypoxia-induced muscle atrophy than slow-twitch muscles (e.g., soleus) (60). This conclusion appears consistent with the present finding that IH-induced UPS activation was more prominent in the EDL than the soleus muscle. Because the increase in UPS induced by glucocorticoid treatment is greater in EDL than in the soleus (12), and chronic hypoxia increases expression of the glucocorticoid receptor and its target genes mainly in the EDL (60), it is reasonable to speculate that AIH induced an increased sensitivity of the fast-twitch muscles to glucocorticoids. Accordingly, the present study provides evidence that the observed changes in proteolytic pathways induced by AIH are driven by endocrine factors. Of particular interest was the finding that ADX abolished the hypoxia-induced increase in UPS activity and atrophy-related genes, suggesting that glucocorticoids are essential for activation of this proteolytic process in both soleus and EDL muscles. These findings are consistent with the previous observations that ADX abrogates the activation of UPS in different catabolic conditions, including fasting (69) and diabetes (42). Although previous studies have demonstrated that glucocorticoids specifically augment the degradation of myofibrillar proteins (62), which is a nonlysosomal process (20, 63), and do not affect cathepsin D activity in muscle (47), the findings of the present work clearly show that ADX abolished the AIH-induced upregulation of lysosomal activity and LC3b and GABARAPl1 mRNA. Therefore, these data suggest that macroautophagy might be modulated by glucocorticoids during short-term IH and are in agreement with recent studies demonstrating that in vivo (25) and in vitro (64) treatment with synthetic glucocorticoids, like dexamethasone, can increase expression of several autophagy genes, including Atg5, LC3b, Becn1, and Sqstm1. Since long-term exposure to hypoxia negatively affects the rates and efficiency of ATP synthesis and is associated with mitochondrial dysfunction (29), we investigated whether the processes of mitochondrial fission and fusion might be affected by AIH. Although the mRNA levels of MFN1/2 found in sham-operated animals exposed to hypoxia were not significantly different from the normoxic control group, it is noteworthy that the mRNA levels of MFN2 in adrenalectomized animals exposed to hypoxia were significantly lower than in the AIH group. Interestingly, it has been shown that skeletal muscle shows a high abundance of Mfn2 (3, 52), a protein whose gene expression is upregulated under several conditions, including exposure to cold (58) and exercise (10). Because these conditions are characterized by high levels of glucocorticoids and catecholamines, the present findings suggest a functional linking between MFN2 and the adrenal gland. Given the fact that Mfn2 knockdown causes alterations in mitochondrial metabolism, characterized by reduced mitochondrial membrane potential and cellular oxygen consumption, as well as by depressed substrate oxidation (3, 44), it is likely that proteins that participate in mitochondrial fusion play a regulatory role in the mitochondrial metabolism under conditions of intermittent low oxygen availability. Further studies need to be carried out to confirm this hypothesis. The present data also show that the hypoxia-induced increase in the activity of the Ca2+-dependent proteolytic system is blocked by ADX. Evidence for a close association between glucocorticoids and Ca2+-dependent proteolysis has also been obtained in previous studies showing that dexamethasone stimulates Ca2+-dependent proteolysis in cultured L6 myotubes (66) and that sepsis, a catabolic state related to hypoxia, increases calpain gene expression in a glucocorticoid-dependent manner (5, 19). Results from the present study clearly show that hypoxia led to insulin resistance, an effect that is dependent on glucocorticoids. It has been extensively demonstrated that glucocorticoids induce insulin resistance in skeletal muscles (49, 68) and that this metabolic disturbance causes muscle wasting by mechanisms that involve suppression of PI3K/Akt signaling, leading to activation of the UPS causing muscle protein degradation (59). Although these data suggest that the dysfunction of protein metabolism observed here in muscles from animals exposed to AIH might be related to insulin resistance, the direct proteolytic effect of glucocorticoids on skeletal muscle cannot be ruled out. In addition to the deleterious effect of glucocorticoids, insulin resistance has also been associated with tissue-specific regulation of HIF-1α during mild chronic IH (50). Although HIF-1 protein, the master regulator of the hypoxic response, has been extensively shown to regulate glycolytic and oxidative (56) metabolism, it is still unclear whether or not this factor can modulate muscle protein metabolism. In agreement with previous studies (23) the present work shows (Fig. 7) that the protein content of HIF-1α was significantly higher in soleus (a typical oxidative muscle) than in EDL (glycolytic muscle). However, we did not detect any change in HIF-1α protein levels induced by AIH in either soleus or EDL muscles (Fig. 7), suggesting that this transcriptional factor is not involved in the AIH-inducible proteolytic metabolic changes. In contrast to proteolysis, we did not find any effect of AIH on rates of protein synthesis estimated by the incorporation of 14C-Tyr into total muscle proteins. This finding is in line with a recent study showing that hypoxia (8% O2 up to 4 days) does not inhibit protein synthesis through mTORC1 signaling in vivo, as phosphorylation of its downstream targets 4E-BP1 and S6K1 in hypoxic mice was similar to control mice (62). In another study in humans, Akt and S6K1 displayed a higher phosphorylation state after 4-h exposure to hypoxia following a meal (14). Although these data show that protein synthesis is maintained or even increased in response to AIH for 8 h, we cannot rule out the possibility that IH might affect the biosynthetic machinery during longer periods of exposure. Indeed, Morrison et al. (43) have reported that the predominant mechanism of muscle wasting in patients with emphysema is a fall in muscle fractional rate of protein synthesis, which is accompanied by an overall fall in whole body protein turnover. Therefore, caution should be taken before asserting that protein degradation is the main cause of skeletal muscle loss induced by prolonged environmental hypoxia in animals and humans. Furthermore, the present data do not allow us to conclude whether or not the AIH-induced changes can persist after the end of the stress stimulus. In summary, the present data suggest that glucocorticoids are essential for induction of insulin resistance in a HIF-1α-independent manner, which is associated with activation of UPS, the autophagic-lysosomal system, and Ca2+-dependent proteolysis, as well as with alterations of the process of mitochondrial fusion in rat skeletal muscles during short-term IH. These observations are important from a clinical standpoint because they might reflect the atrophy that occurs when the period of hypoxia is prolonged and imply that the catabolic response in skeletal muscle during acute hypoxia might be prevented by treatment with a glucocorticoid antagonist. GRANTSThis work was supported by grants from the Fundação de Amparo à Pesquisa do Estado de São Paulo (Fapesp 08/06694-6, 12/24524-6, 12/18861-0, 15/21112-7) and from the Conselho Nacional de Pesquisa (CNPq 305149/12-1, 306624/15-0). F. Przygodda received a fellowship from the Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior-Pró-Reitoria de Extensão (CAPES-PROEX). DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSF.P., L.H.M., J.M., N.M.Z., and L.G.B. performed experiments; F.P., L.H.M., J.M., D.A.G., and B.H.M. analyzed data; F.P., L.H.M., J.M., D.A.G., I.C.K., and L.C.N. interpreted results of experiments; F.P. prepared figures; F.P., L.H.M., B.H.M., I.C.K., and L.C.N. drafted manuscript; F.P., B.H.M., I.C.K., and L.C.N. edited and revised manuscript; J.M., D.A.G., N.M.Z., L.G.B., B.H.M., I.C.K., and L.C.N. approved final version of manuscript. We thank Maria Antonieta R. Garófalo, Elza Aparecida Filippin, Lilian do Carmo Heck, and Victor Diaz Galban for their technical assistance. REFERENCES
Page 11multipotent cells have received great attention because of their potential capacity to repair and remodel peripheral tissues. In murine skeletal muscle, both blood vessel growth and regeneration of muscle fibers have been suggested to receive contributions from bone marrow (BM)-derived cells (10). The extent to which this occurs in humans is poorly studied, but recently we demonstrated that BM-derived cells had contributed to skeletal muscle fibers and endothelial cells in women transplanted with male BM (51) to a degree comparable with earlier findings in mice (28). In humans, exercise increases the circulating level of different BM-derived cell subsets with proposed effects on skeletal muscle remodeling, including both monocytes (MOs) and endothelial progenitor cells (EPCs) (1, 29, 42). In mice, incorporation of BM-derived cells into skeletal muscle fibers is increased with exercise (28, 37). Taken together, these data indicate that BM-derived cell recruitment to skeletal muscle is activated by exercise. However, a difficulty with interpreting existing data concerning the circulating level of these cells is that different sets of markers and analytical procedures have been used for quantification. In this study, the levels of MO subpopulations were measured according to the new nomenclature that groups them into a classical, an intermediate, or a nonclassical subset based on their expression of CD14 and CD16 (22, 58), where MOs expressing CD16 have been termed proinflammatory based on their pattern of cytokine production and antigen-presenting capacity (57). Existing data provide evidence that only the proinflammatory subset of MOs are recruited to skeletal muscle upon injury, phagocytose cellular debris, and induce myoblast proliferation (3). Thus we aimed to elucidate whether the distinctive subsets of MOs show different dynamics in their mobilization to, and disappearance from, the circulation with exercise. Furthermore, with a flow cytometric protocol developed for increased specificity and sensitivity in the rare-event analysis of EPCs, the CD45dimCD34+VEGFR2+ marker profile was used to quantitate EPCs (22). This specific cell population was recently proposed to contain the “true” circulating EPCs (48). Earlier studies in humans have mainly focused on recruitment of cells with progenitor cell characteristics to the circulation, although it is of key importance to understand how the cells identify the site of regeneration. For example, skeletal muscle remodeling with exercise occurs almost exclusively in the muscle tissue involved in exercise. MOs and EPCs have been reported to utilize similar mechanisms to enter the peripheral tissue, where selectins initiate a rolling interaction with the endothelial monolayer and the adhesion molecules ICAM-1 and VCAM-1 mediate firm adhesion (8, 12, 18, 23, 31, 35). In mice, endurance exercise has been shown to increase the expression of E-selectin, P-selectin, and platelet endothelial cell adhesion molecule-1 in skeletal muscle together with an increased adhesion and transmigration of leukocytes across the endothelium (34). These data provide an idea about possible recruitment mechanisms, but no studies to our knowledge have examined their expression in human skeletal muscle after exercise. In the present study we aimed to increase the understanding regarding plausible mechanisms of exercise-induced recruitment of circulating cells to skeletal muscle. Our hypothesis was that an exercise bout increases the number of circulating cells shown to be involved in skeletal muscle remodeling, concurrent with increased expression of endothelial cell adhesion molecules in the exercised skeletal muscle. Such findings would support these processes as involved in skeletal muscle remodeling. METHODSTwenty healthy, nonsmoking, moderately active male subjects were included in the study, with 10 subjects in group 1 and 10 subjects in group 2. In group 1, the mean (range) age, height, and weight were 25 (20–31) yr, 182 (174–187) cm, and 78 (64–87) kg, respectively. Their mean (range) maximal oxygen consumption (V̇o2max) was 54 (43–59) ml·kg−1·min−1. In group 2, the mean (range) age, height, and weight were 27 (20–34) yr, 179 (173–187) cm, and 75 (69–83) kg, respectively. Their mean (range) V̇o2max was 50 (41–59) ml·kg−1·min−1. No significant differences were found between the groups with respect to age, height, weight, or V̇o2max. Electrocardiography was performed at rest before the experiment to exclude any pathology. The subjects received no medical treatment and had no medical history. All subjects performed 60 min of cycling exercise on an electrodynamically loaded cycle ergometer. During the first 20 min, subjects cycled at 60 rpm at a work rate corresponding to 50% of V̇o2max, after which the work rate was increased to a work load corresponding to 65% of V̇o2max for a further 40 min. The perceived exertion was measured every 10th minute with the Borg RPE scale (rating perceived exertion on a scale of 6–20). The following protocol was used to change load: At 30 min: effort rated <13, increase by 10 W. At 40 min: effort rated <15, increase by 10 W; 17, decrease by 10 W; ≥18, decrease by 20 W. At 50 min: effort rated <16, increase by 10 W; ≥18, decrease by 10 W. The subjects were allowed to drink water ad libitum during the experiment. In group 1, blood samples for flow cytometric analysis and complete blood count were drawn from an antecubital vein into EDTA tubes at rest (Pre), directly after exercise (Post), and 30 min (Post 30) and 2 h (Post 2 h) after exercise. In group 2, muscle biopsy samples were obtained from the vastus lateralis muscle after local anesthesia without epinephrine at rest and 2 h after exercise with the percutaneous needle biopsy technique (6). Blood samples for plasma collection were drawn from an antecubital vein into EDTA and heparin tubes at rest (Pre), directly after exercise (Post), and 30 min (Post 30) and 2 h (Post 2 h) after exercise. All biopsy samples were frozen within 15 s in liquid nitrogen and stored at −80°C until analysis. Before the study, the experimental protocol was explained to all the subjects and their informed consent was obtained. The study was approved by the Ethics Committee at Karolinska Institutet, Stockholm, Sweden. All antibodies, Trucount tubes, and the lysis buffer were from BD Biosciences (San Jose, CA). For assessment of monocytes, 20 μl of CD45-PerCP (clone 2D1, no. 345809), 20 μl of CD14-FITC (no. 345784), and 20 μl of CD16-PE (clone 3G8, no. 555407) MAb was used per Trucount tube (no. 340334). The corresponding fluorescence-minus-one (FMO) control was performed with 20 µl of CD45-PerCP and 20 μl of CD14-FITC added to a polystyrene tube. For analysis of EPCs, 20 μl of CD45-FITC (clone 2D1, no. 345808), 5 μl of CD34-APC (clone 8 G12, no. 345804), 20 μl of CD309/VEGFR2-PE (no. 560494), and 5 μl of 7-aminoactinomycin D (7-AAD; no. 559925) were used. The corresponding FMO control was performed with 20 μl of CD45-FITC and 5 μl of CD34-APC. Fifty microliters of blood was added to each tube by reverse pipetting. The tubes were incubated for 20 min at room temperature in the dark before 450 μl of 1× lysis solution (no. 555899) was added to each tube. The tubes were incubated for 10 min at room temperature in the dark and then protected from light and analyzed within 1 h. Sample acquisition was performed on a FACSAria cell sorter with FACSDiva 6.1 software (BD Biosciences) with a flow rate of <10,000 events/s. Fluorescence compensation was performed with anti-mouse Igκ CompBeads (BD Biosciences, no. 552843) and the respective antibodies. The analysis of monocyte subpopulations and EPCs was performed as in the paper by Hristov et al. (22). For analysis of monocytes, a gate (P1) was first set on a forward scatter (FSC)/side scatter (SSC) plot to remove debris (Fig. 1A). Then, P1 was displayed on a SSC/CD45 plot, and all CD45+ events were included in P2 (Fig. 1B). P2 was thereafter displayed on a SSC/FSC plot, and gate P3 was set around the monocyte population (Fig. 1C). The P3 population was thereafter shown on a CD14/CD16 dot plot, and the different monocyte populations were defined with the quadrant marker (Fig. 1D). The bead number was obtained from an ungated CD16/CD45 plot (gate P4 in Fig. 1E). For analysis of EPCs, gate P1 was set on a SSC/FSC scatterplot to remove debris (Fig. 2A). P1 was displayed on a SSC/7-AAD plot to only include 7-AAD− events (P2, Fig. 2B). Gate P2 was shown on a SSC/CD45 plot to include all CD45+ events (P3, Fig. 2C). Gate P3 was thereafter displayed on a SSC/CD34 plot, and a gate (P4) was set around the CD34+ events (Fig. 2D). The events in P4 were thereafter shown on a SSC/FSC plot, and gate P5 was set to include the lymph-blast scatter region (Fig. 2E). The events in P5 were subsequently shown on a SSC/CD45 plot, and the CD45dim events were included in P6 (Fig. 2F). Finally, the CD45dimCD34+ cells in P6 were displayed in a CD34/VEGFR2 plot, and the VEGFR2+ cells were included in gate P7 (Fig. 2G). The threshold for P7 was set with the FMO control. The bead number was obtained from an ungated CD34/CD45 plot (P8, Fig. 2H). Fig. 1.Flow cytometric analysis of absolute counts of monocyte subsets in whole blood. Gate P1 was set on a SSC/FSC dot plot to remove debris (A). Thereafter, the events included in P1 were displayed on a SSC/CD45 dot plot, and gate P2 was set to include CD45+ events (B). The events in P2 were shown on a SSC/FSC dot plot, and gate P3 was set around the monocyte population (C). Subsequently, the events in P3 were displayed on a CD14/CD16 dot plot and separated into the 3 different monocyte subsets with the quadrant marker (Q) (D). The bead number, gate P4, was obtained from an ungated CD16/CD45 plot (E). F: the population hierarchy. Fig. 2.Flow cytometric analysis of absolute counts of endothelial progenitor cells (EPCs) in whole blood. Gate P1 was set on a SSC/FSC dot plot to remove debris (A). P1 was displayed on a SSC/7-AAD dot plot, and all 7-AAD− events were included in gate P2 (B). Gate P2 was shown on a SSC/CD45 plot to include all CD45+ events in gate P3 (C). Gate P3 was thereafter displayed on a SSC/CD34 plot, and gate P4 was set around the CD34+ events (D). The events in P4 were thereafter shown on a SSC/FSC plot, and gate P5 was set to include the lymph-blast scatter region (E). The events in P5 were subsequently shown on a SSC/CD45 plot, and the CD45dim events were included in P6 (F). Finally, the CD45dim/CD34+ cells in P6 were displayed in a CD34/VEGFR2 plot, and the VEGFR2+ cells were included in gate P7 (G). The threshold for P7 was set with the FMO control. The bead number, gate P8, was obtained from an ungated CD34/CD45 plot (H). I: the population hierarchy. The complete blood count on subjects from group 1 was performed by the Karolinska University Laboratory. The values for hemoglobin were used to calculate the exercise-induced hemoconcentration according to the Dill and Costill equation (11). The cell counts were adjusted by the change in blood volume. Blood samples from group 2 collected in EDTA tubes or heparinized tubes were centrifuged at 1,000 g for 10 min. An additional centrifugation step of the plasma at 10,000 g for 10 min at 4°C was included to remove platelets. The samples were immediately frozen in liquid nitrogen and stored at −70°C until further analysis. The plasma concentrations of stromal cell-derived factor-1α (SDF-1α), VEGF-A, and granulocyte colony-stimulating factor (G-CSF) were determined with a sandwich enzyme-linked immunoassay (R&D Systems, Minneapolis, MN) according to the manufacturer’s directions. The minimum detectable levels of these kits were 47 pg/ml for SDF-1, 9 pg/ml for VEGF, and 4.6 pg/ml for G-CSF. Plasma albumin was analyzed at the Karolinska University Laboratory with the conventional Modular P2 technique and used as control for plasma volume changes with exercise. The blood samples for catecholamine analysis were collected from six subjects in ice-cold heparinized tubes and centrifuged at 2,000 g for 15 min. The plasma samples were then immediately frozen in liquid nitrogen and stored at −70°C until further analysis. The catecholamine concentrations were analyzed at the Karolinska University Laboratory by HPLC with electrochemical detection (19). Total RNA was prepared from the muscle biopsies and human umbilical vein endothelial cells (HUVECs) with the TRIzol method (Invitrogen, Carlsbad, CA) and quantified by measuring the absorbance spectrophotometrically at 260 nm. The integrity of total RNA was confirmed by a 1% agarose gel electrophoresis. RNA (1 μg) was reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Applied Biosystems, Carlsbad, CA) and random hexamer primers (Roche Diagnostics, Mannheim, Germany) in a total volume of 20 µl. For human samples, ICAM-1, VCAM-1, E-selectin, SDF-1, IL-6, and VEGF-A were ordered as gene assays on demand (Hs00164932_m1, Hs00365486_m1, Hs00950401_m1, Hs00171022_m1, Hs00174131_m1, and Hs00900055_m1; Applied Biosystems). GAPDH was selected as the endogenous control to correct for potential variations in RNA loading (Hs99999905_m1; Applied Biosystems). All reactions were performed in 96-well MicroAmp Optical plates with the ABI-PRISM 7700 Sequence Detector (Applied Biosystems). The amplification mixes contained 5 µl of cDNA sample, 2× TaqMan Universal PCR Master mix and 20× Assay Mix in a final volume of 25 µl. The thermal cycling conditions were 2 min at 50°C, 10 min at 95°C, and then 45 cycles of 15 s at 95°C and 1 min at 65°C. The frozen muscle biopsy specimens were cut into 5-µm sections and placed on SuperFrost Plus microscope slides (Thermo Fisher Scientific, Waltham, MA). The sections were fixed for 10 min in acetone at −20°C, followed by three 3-min washes in PBS. The slides were blocked in 1% BSA-PBS for 30 min in a humid chamber and were then incubated with the primary antibody overnight at 4°C in a humid chamber. The antibodies used were goat anti-human ICAM-1 (diluted 1:200; BBA17, R&D Systems), goat anti-human VCAM-1 (diluted 1:100; BBA19, R&D Systems), and goat anti-human E-selectin (diluted 1:100; BBA18, R&D Systems). After the slides were washed, they were incubated for 60 min at room temperature with secondary Texas red-conjugated donkey anti-goat (diluted 1:500; (ab6883, Abcam, Cambridge, UK). The slides were evaluated by microscopy before subsequent staining with primary mouse anti-human CD31 antibody (diluted 1:400; M0823, Dako, Glostrup, Denmark) followed by the Alexa Fluor 488-conjugated secondary rabbit anti-mouse antibody (diluted 1:500; A11059, Molecular Probes, Eugene, OR), both incubated for 60 min at room temperature. After the slides were washed in PBS, they were mounted in Vectashield DAPI (Vector Laboratories, Burlingame, CA). For the negative control, the primary antibodies were excluded from the protocol. HUVECs (no. C-003-5C, Life Technologies, Carlsbad, CA) were cultivated in M200 medium (Life Technologies) with 20% fetal bovine serum (FBS) in tissue flasks coated with 0.1% gelatin. At passages 2–3, the HUVECs were seeded in 24-well culture plates and stimulated for 2 h with either serum-free M200 medium (control) or recombinant protein diluted in serum-free M200 medium. The proteins used were VEGF-A and IL-6 (R&D Systems) at concentrations of 100 ng/ml. The stimulations were performed eight times in duplicate. Data for monocyte and EPC numbers and plasma protein and catecholamine levels were analyzed with one-way repeated-measures (RM) ANOVA. The Tukey test for all pairwise comparison was used as post hoc analysis. A power analysis was performed for the EPC data using G*Power version 3.1.9.2 to determine the subject number required to obtain a significant effect of exercise. The partial η2 utilized for calculating the effect size f was obtained by dividing the sum of squares (SS) between time points with the total SS. The data for plasma VEGF-A were not normally distributed, and therefore the RM ANOVA on ranks was used. The statistical analyses of mRNA data from skeletal muscle and HUVECs were conducted on ratios of target to endogenous control; a ΔCT (where CT is threshold cycle) value was obtained by subtracting the GAPDH CT value from the corresponding target CT value. The expression of each target was then determined by 2−ΔCT, which provides target gene expression related to the housekeeping gene expression in each sample. To ensure normal distribution of the related target gene to housekeeping gene expression, logarithmic transformation to the base 10 was performed on mRNA data from skeletal muscle and HUVECs. The mRNA data were then analyzed with Student’s t-test. All analyses except for the EPC power analysis were performed with SigmaPlot 13.0 (Systat Software, San Jose, CA). Differences were considered significant at P < 0.05. Data are presented as means ± SD unless otherwise stated. RESULTSTo assess the effect of exercise on levels of circulating cells with proposed effects on skeletal muscle remodeling, we measured the absolute number of MOs and EPCs by flow cytometry. The data for cell numbers and hemoglobin are presented in Table 1. The number of classical CD14++CD16− monocytes was changed in a biphasic pattern: significantly increased directly after exercise and also at 2 h after exercise (P < 0.001). The intermediate CD14++CD16+ monocytes (Post vs. Pre, P < 0.05) and nonclassical CD14+CD16++ monocytes (Post vs. Pre, P < 0.001) were significantly increased directly after exercise. For CD45dimCD34+VEGFR2+, there was a trend toward a significant increase with exercise (P = 0.08). The power analysis performed utilizing the data obtained for both CD45dimCD34+ and CD45dimCD34+VEGFR2+ cells displayed that with 14 subjects there would have been a significant effect on EPC number with exercise with a power > 0.80 and α-level 0.05.
The data for complete blood counts are presented in Table 2. No statistical comparisons were performed for these values.
To characterize possible mechanisms behind changes in circulating cells, we analyzed factors demonstrated in other conditions to mobilize cells to the circulation. The data for G-CSF, SDF-1, VEGF-A, catecholamines, and albumin are presented in Table 3. The concentration of G-CSF protein in the plasma was increased significantly at Post 2 h without adjustment to albumin. When adjusted to the plasma albumin concentration, the G-CSF level was not elevated after exercise (P = 0.09). Similar to G-CSF, after adjustment to the plasma albumin concentration the concentration of SDF-1 did not change in response to exercise (P = 0.44). Another factor suggested to mobilize cells to the circulation is VEGF-A. However, the concentration of VEGF-A protein in the plasma did not change in response to exercise (P = 0.3). Recruitment of cells from the marginal pool to the circulation has been suggested to occur through activation of the adrenergic system (5). Therefore, we wanted to establish whether changes in epinephrine and norepinephrine occurred within the present exercise model. The plasma levels of both epinephrine (P < 0.01) and norepinephrine (P = 0.001) were significantly increased directly after exercise and restored to resting levels at Post 30.
To assess whether exercise induces factors known to recruit MOs and EPCs from the circulation, we analyzed the expression of cell adhesion molecules and chemoattractants in skeletal muscle. The expression of the housekeeping gene GAPDH did not change with exercise (data not shown). The mRNA levels of ICAM-1 (P = 0.03; Fig. 3A), VCAM-1 (P = 0.04; Fig. 3B), and E-selectin (P = 0.04; Fig. 3C) were significantly upregulated at Post 2 h as compared with Pre. However, the mRNA level of E-selectin was extremely low, with a mean CT value in the preexercise biopsies of 36.7 at cDNA dilution 1:5 corresponding to 50 ng RNA/well. IL-6 (P = 0.04; Fig. 3D) and VEGF-A (P = 0.002; Fig. 3E) mRNA levels were significantly increased 2 h after exercise. The SDF-1 mRNA level did not change in response to exercise (P = 0.2; Fig. 3F). Double staining for ICAM-1, VCAM-1, or E-selectin and the endothelial marker CD31 showed that ICAM-1, VCAM-1 and E-selectin were localized to the endothelium (Fig. 4). The staining did not differ between Pre and Post 2 h (data not shown). E-selectin also showed a weak staining pattern on the sarcolemma of the muscle fibers (Fig. 4H). The slides treated with secondary antibodies only showed no staining (data not shown). Fig. 3.ICAM-1 (A), VCAM-1 (B), E-selectin (C), IL-6 (D), VEGF-A (E), and SDF-1 (F) mRNA levels in skeletal muscle at Pre and Post 2 h. Values (in arbitrary units) are presented as means + SD; n = 10. *P < 0.05, **P < 0.01. Fig. 4.Representative images (×20 magnification) of skeletal muscle sections with immunofluorescent staining for ICAM-1 (B), VCAM-1 (E), and E-selectin (H). Images show that the adhesion molecules are expressed on endothelial cells stained for CD31 (A, D, and G). C, F, and I display overlays. To evaluate whether endothelial cell adhesion molecules are induced by factors known to increase in skeletal muscle with exercise, we stimulated HUVECs with VEGF-A and IL-6. With VEGF-A stimulation, ICAM-1 mRNA levels did not change (P = 0.4; Fig. 5A). VEGF-A stimulation significantly increased the mRNA levels of VCAM-1 (P = 0.02) and E-selectin (P = 0.02) compared with control (Fig. 5, B and C). Stimulation with IL-6 for 2 h did not significantly increase the level of ICAM-1 (P = 0.7; Fig. 5A), VCAM-1 (P = 0.3; Fig. 5B), or E-selectin (P = 0.2; Fig. 5C). Fig. 5.ICAM-1 (A), VCAM-1 (B), and E-selectin (C) mRNA levels in HUVECs after 2-h stimulation with VEGF-A or IL-6 compared with unstimulated control (CTRL). Values (in arbitrary units) are presented as means + SD; n = 8. *P < 0.05. DISCUSSIONIn the present study, we report that in young healthy men a single bout of exercise had the following effects: 1) increased the circulating level of MO subsets with different patterns; 2) increased plasma catecholamines known to influence cell mobilization; 3) upregulated the expression of the recruiting factors ICAM-1, VCAM-1, and E-selectin in the skeletal muscle; and 4) increased the expression of VEGF in skeletal muscle. Furthermore, stimulation of HUVECs with VEGF-A increased their expression of VCAM-1 and E-selectin. Exercise induces remodeling processes such as angiogenesis, extracellular matrix remodeling, and fiber growth in skeletal muscle. Importantly, several studies have demonstrated that, in humans, exercise increases the circulating level of different BM-derived cell subsets with proposed effects on skeletal muscle remodeling, including both MOs and EPCs (1, 29, 42). MOs play a key role in murine skeletal muscle regeneration and have proangiogenic effects (3, 13, 30, 41, 47, 53), and EPCs have been shown to participate in vessel growth in mice (4, 38). The importance of these processes is highlighted by the finding that skeletal muscle regeneration in mice was demonstrated to be severely diminished when MOs and macrophages were depleted (3, 30, 53). Here we add to the existing literature regarding exercise-induced cell mobilization by utilizing a novel flow cytometric protocol designed to enumerate MOs and EPCs, which employs a lyse/no-wash procedure, dead cell exclusion, and determination of absolute cell numbers (22). Furthermore, the analysis was performed on fresh whole blood with red blood cell lysis, which has been shown to yield higher numbers of viable EPCs than density gradient centrifugation and prior preservation of cells (9). By using cell-counting beads, we determined the absolute number of MOs and EPCs per microliter of blood. In light of the leukocytosis known to occur with exercise, we find that the present protocol is advantageous because it quantifies absolute cell counts instead of the percentage of cells in a certain gate, which instead measures relative proportions. In the present study, the number of CD16-positive MOs was significantly increased directly after exercise and was back at preexercise level at 30 min after exercise. This transient increase of CD16-positive MOs with acute exercise was shown in a previous study and could be substantially reduced with a β-adrenergic blocker (50). All subpopulations of MOs increase with epinephrine infusion in healthy subjects (26), which suggests a catecholamine-dependent mechanism behind the increase seen with exercise. By contrast with the CD16-positive MOs, the number of CD14++CD16− MOs remained increased 2 h after exercise. One possible explanation for this finding could be that the quick increase in MOs seen with exercise is because of catecholamine-induced release from the marginal pool, while the sustained increase of classical MOs after exercise could represent another mobilizing stimulus. Another interpretation, proposed by Gabriel et al. (14), is that the faster disappearance of CD16-positive MOs from the circulation after exercise is because of their migration into peripheral tissues such as skeletal muscle. Interestingly, we recently demonstrated that the chemokine fractalkine increased in human skeletal muscle endothelial cells after an exercise bout similar to that in the present study (52). Because human CD16-positive MOs have been demonstrated to express the fractalkine receptor CX3CR1, and undergo efficient binding and transendothelial migration in response to fractalkine (2), we hypothesize that an exercise bout similar to that in this study may promote recruitment of CD16-positive MOs to exercised skeletal muscle. Even though an established surface marker combination defining EPCs is lacking, it was recently proposed that the CD45dim fraction of CD34+VEGFR2+ cells contains the “true” circulating EPCs (48). With the enumeration protocol by Hristov et al. (22) utilized in the present study, the absolute number of CD34+VEGFR2+ EPCs exceeded what has been reported before (55), which may imply that the number of circulating cells with angiogenic capacity recruited with exercise is higher than previously suggested. We report a similar fold change in EPC numbers as observed earlier after a single bout of exercise in healthy humans, even though we just showed a trend toward a significant increase (P = 0.08). There was, however, a larger interindividual variation in the number of EPCs compared with MOs, and the post hoc power analysis displayed that 14 subjects would have been required to detect a significant effect of exercise on the number of EPCs. Our interpretation of the present data is that exercise likely induces an increase in EPCs in the circulation. Like MOs, EPCs can be mobilized to the circulation by sympathetic activation (24). In addition, certain cytokines and chemokines, e.g., G-CSF, SDF-1, and VEGF-A (20, 27, 32, 39, 54), have been shown to mobilize EPCs. When adjusted for hemoconcentration, no increase of SDF-1 or G-CSF in plasma was observed in the present study, similar to earlier findings after a short supramaximal exercise bout (33). The level of VEGF-A in plasma was unaltered with exercise. This indicates that no release occurred for either of these chemokines and that the increase seen in SDF-1 and G-CSF directly after exercise just represented the transient exercise-induced decrease in plasma volume. Thus, even though we cannot rule out the importance of other chemokines, the present findings support earlier reports in that increased plasma concentrations of catecholamines influence EPC mobilization (24, 40). Nevertheless, even though cells are increased in the circulation, they need to traverse the endothelium to reach the site of tissue remodeling. In animal models, recruitment of MOs and EPCs to peripheral tissues occurs through increased expression of the endothelial adhesion molecules ICAM-1, VCAM-1, and selectins (8, 12, 18, 23, 31, 35), which have all been shown to be regulated at the transcriptional level (7, 36, 49). In the present study, ICAM-1, VCAM-1, and E-selectin were all increased at the mRNA level 2 h after exercise, although the gene expression of E-selectin was very low. Furthermore, their expression on endothelial cells in skeletal muscle tissue was confirmed with immunofluorescent staining. Interestingly, a mechanism proposed for the faster disappearance of CD16-positive MOs than classical MOs from the circulation after exercise is their higher expression of lymphocyte function-associated antigen 1 (the receptor for ICAM-1), which enables firm attachment to ICAM-1-expressing endothelial cells in peripheral tissues (14). Here we add to this hypothesis by demonstrating an increase of ICAM-1 expression in skeletal muscle with exercise. Our present study and those of others indicate that exercise increases both VEGF-A and IL-6 (25, 56), which also are known to induce the transcription of ICAM-1, VCAM-1, and E-selectin (15, 16, 21). With IL-6 stimulation, there was no change in expression in either of the adhesion molecules. However, this could be the result of the need for soluble IL-6 receptor for endothelial cells to respond to IL-6, as previously reported (43). With VEGF-A stimulation, the expression of E-selectin and VCAM-1 in HUVECs was increased, while the ICAM-1 level did not change. This finding suggests a cross talk between skeletal muscle and endothelial cells, with VEGF-A as one possible mediator behind upregulation of endothelial adhesion molecules with exercise. An alternative route in activation of these factors in endothelial cells is increased shear stress induced by the higher blood flow during exercise. Interestingly, while increased shear stress has been demonstrated to increase ICAM-1 expression, it rather downregulates the levels of VCAM-1 and E-selectin (46). Thus it is likely that different mechanisms govern the increased expression of adhesion molecules seen in the present study. In light of our findings and the current literature, it is tempting to speculate that shear stress induces a quick upregulation of ICAM-1, while a subsequent increase in E-selectin and VCAM-1 is due to stimulation by growth factors. The further invasion of circulating cells into the tissue depends on proteases cleaving the extracellular matrix, and we and others have shown that matrix metalloproteinases increase in levels and activity in the skeletal muscle tissue with exercise (17, 44, 45), adding to the idea of exercise as a remodeling stimulus contributing to recruitment of circulating cells. The present data support the idea that in young healthy men an aerobic exercise bout results in the recruitment of MOs and EPCs, cells with proposed effects on skeletal muscle remodeling, into the circulation. This occurs concurrently with an increased expression in the skeletal muscle of adhesion molecules with a known capacity to guide cells into the peripheral tissue. Whether these processes are induced in other study populations and with other exercise regimes should be investigated. GRANTSThis study was supported by grants from the Swedish National Centre for Research in Sports, the Swedish Medical Research Council, and the Marcus and Marianne Wallenberg Foundation. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSA.S., E.R., E.J., and T.G. conceived and designed research; A.S., E.R., and T.G. performed experiments; A.S. analyzed data; A.S., E.J., and T.G. interpreted results of experiments; A.S. prepared figures; A.S., E.R., and T.G. drafted manuscript; A.S., E.J., and T.G. edited and revised manuscript; A.S., E.R., E.J., and T.G. approved final version of manuscript. REFERENCES
Page 12altered visual acuity in astronauts exposed to microgravity is a newly described disorder commonly manifesting as a hyperopic shift in vision (farsightedness) (54). Clinically, there are retinal, optic nerve, globe, and optic nerve sheath abnormalities reported in retrospective studies (46, 54) and prospectively aboard the International Space Station (ISS) (55). Of primary concern is an increased incidence and severity of imaging findings with mission duration (46) and the development of optic disc edema and choroidal folds, which can result in permanent injury to the optic nerve papilla and retinal fiber layers. The range of pathological findings indicates elevated intracranial pressure (ICP) as a potential component, hence NASA’s internal term, visual impairment intracranial pressure (VIIP) syndrome (64); however, a definitive etiology has yet to be established. Improved understanding of the VIIP syndrome therefore remains necessary to mitigate complications and safely achieve new human space exploration milestones. Although there is no terrestrial-based analog environment that perfectly mimics microgravity, head-down tilt (HDT) is commonly used to redirect the head-to-foot hydrostatic gradient vector (Gz) and promote cephalad-directed fluid shifting (36). Cephalad fluid shifting is an invariable property of microgravity attributable to removal of hydrostatic gradients that are normally present at 1G on earth (66). This redistribution of bodily fluid is hypothesized to be a major pathophysiological component of spaceflight-induced visual alterations (64). Although severe visual changes are not common in HDT bed rest studies (81), diminished visual acuity (36), optic nerve sheath distention (86), and peripapillary retinal nerve fiber layer thickening (81) have been reported with various HDT protocols, which are features associated with the VIIP syndrome (3, 54, 55). These observations support HDT as a potential analog of the VIIP syndrome. A prospective terrestrial-based microgravity HDT analog study was therefore implemented to study the physiological components that may lead to the VIIP syndrome. Our study design also included exposure to elevated ambient CO2 levels, which is an environmental factor present on board the ISS (47). We hypothesized that cephalad fluid shifting with elevated ambient CO2 could possibly interact in a manner that alters physiological pathways that may be associated with the VIIP syndrome. As a secondary aim, our study investigates the effects of brief exposure to high levels of CO2 (3%) to simulate an acute elevation of CO2, which may occur around astronauts while in confined spaces with reduced airflow, in proximity to large groups of astronauts or during sleep attributable to the lack of movement and gravitational dissipation (47, 61). The former situations have been anecdotally associated with headaches (47), which is a symptom sometimes related to elevated ICP. As both HDT and elevated ambient CO2 levels can increase ICP (67, 70), it is useful to review the concept of ICP homeostasis. Related to the nearly inelastic properties of the cranium, ICP homeostasis is dependent on regulation of constant intracranial volume (63). ICP has both pulsatile and static components (17). Pulsatile ICP is created by oscillatory increases in intracranial volume attributable to cerebral blood flow (CBF) and characterized by a pressure-amplitude and rise time (19). Pulsatile ICP is modulated by passive downstream venous and cerebrospinal fluid (CSF) volume outflow, which occurs with a slight phase delay relative to the arterial input (45). In contrast, static ICP represents a low-pass-filtered pressure, which is actively modulated by CSF formation rate (13, 35). Collectively, passive and active modulation of ICP are components of the compensatory reserve system (16). With progressive reduction of compensatory reserve capacity, craniospinal compliance, defined as the ratio of volume change to pressure change (ΔV/ΔP), can become insufficient, resulting in exponential increases in ICP with additional volume (16). ICP stability is therefore dependent on the compensatory reserve system to maintain maximal craniospinal compliance under varying physiological or pathological conditions. In the case of normal pressure hydrocephalus (NPH), shown by magnetic resonance elastography to exhibit reduced craniospinal compliance (21), there is predictably increased ICP pulsatility (20). Given that craniospinal compliance decreases when rotating from upright to the supine position (2), we hypothesize that the additional cephalad fluid shifting associated with HDT (57) would further reduce craniospinal compliance and also result in increased ICP pulsatility. Although previous analog studies of the VIIP syndrome have capitalized on quantitative MRI (qMRI) to evaluate structural (76) or physiological changes (57) related to HDT, we elected to combined qMRI measurements of volumetry, bulk flow, and diffusion to simultaneously evaluate changes in intracranial structure and physiology. We hypothesize that this approach could better delineate how cephalad fluid shifting and elevated ambient CO2 exposure interact. By design, this was a short-duration preliminary study with the ultimate goal to develop the rationale, feasibility, and methodology for larger and longer-duration analog studies. METHODSThe local institutional review boards of Germany (Ethics Committee of the Medical Council of North Rhine) and the USA (Baylor College of Medicine) approved this HIPAA-compliant prospective analysis of six healthy adult subjects. Subjects were medically examined before enrollment and excluded from the study with evidence of inability to sleep supine; drug, medication, or alcohol abuse; smoking; psychiatric or psychological disorders; gastro-esophageal abnormalities; elevated risk of thrombosis; spinal abnormalities; and ophthalmological conditions. Written and informed consent was obtained before inclusion in the study. The study was previously described (57a). Briefly, all study procedures, including baseline data collection and study intervention periods, were performed in a facility (http://www.dlr.de/envihab) specifically designed for terrestrial-based analog studies of spaceflight with equipment for postural manipulations and precise adjustment of ambient CO2 concentrations [German Aerospace Center (DLR), Cologne, Germany]. Environmental conditions (temperature, lighting, humidity, pressure, and atmospheric gases) were standardized and remotely monitored. The investigators and subjects were blinded to the particular mixture of atmospheric gases used in each condition to avoid any bias in operator measurements or subject response. Body mass index (BMI), height, and age were recorded at the start of the study. Mean arterial pressure (MAP) was measured just before each MRI session. Height and MAP were used to estimate mean cerebral arterial pressure (CAP) during supine and −12° HDT conditions as previously described (58). All subjects were placed on a caffeine-free diet with standardized nutritional and fluid intake. Although the majority of previous HDT studies were performed at −6°, as this angle is considered a suitable model for spaceflight effects on orthostatic tolerance (36), it is somewhat arbitrary and typically not chosen to evaluate cerebrovascular physiology. −12° HDT was selected in our study for pragmatic reasons related to briefer exposure and also to accentuate parameters associated with cephalad fluid shifting, such as facial edema and jugular venous distension (3, 36, 57). Six healthy male subjects (mean age: 41 ± 5 yr, mean BMI: 26.2 ± 2.0 kg/m2), enrolled in two groups of three subjects each, were tested in a randomized, double-blinded crossover design study with two HDT conditions: 26.5-h exposure to −12° HDT with ambient air (0.04% CO2) and −12° HDT with 0.5% CO2 (Table 1) through room air conditioning. The level of atmospheric CO2 enrichment was set to 0.5% because it is the average CO2 concentration on board the ISS (47) and the limit for occupational exposure to CO2 in Germany. At the end of each HDT condition, subjects also underwent acute, 2.5-h exposure to 3% CO2 from a gas cylinder via a facial mask to simulate exposure to higher levels of CO2. Three MR scans were taken for each condition (total of 6 scans per subject): 1) baseline scan in the supine position with ambient air (0.04%), 2) after 26.5-h HDT, and 3) during exposure to 3% CO2 at 28-h HDT. A readjustment period of 1 wk was held between conditions. A ventilation mask was worn for all interventions (ambient, 0.5% CO2, and 3% CO2), which required a slight rightward head rotation to accommodate the ventilation mask tubing in the head coil. The −12° HDT was maintained during MRI using a cross-linked polyethylene foam wedge with high-friction coating that kept the torso and legs elevated upon transfer to MR table and during scanning.
MRI studies were performed on a dedicated 3-T Siemens Biograph scanner, with a 16-channel head coil, maximum slew rate of 200 mT·m−1·s−1 and maximum gradient amplitude of 45 mT/m (Siemens Healthcare, Erlangen, Germany). All flow studies used a retrospectively synchronized phase-contrast cine sequence with peripheral plethysmographic triggering and parameters shown in Table 2. Heart rate (HR) in beats per minute (bpm) was recorded during the phase-contrast acquisition. Phase-contrast data were analyzed with freely available software Segment version 1.9 R3656 (http://segment.heiberg.se) (30).
CSF hydrodynamics were evaluated using a cine phase-contrast sequence synchronized to the arterial pulse wave. To optimize flow velocities, the sequence was acquired at a plane perpendicular to the midcerebral aqueduct (Fig. 1), which generates a bidirectional CSF waveform (Fig. 2A). The margins of the aqueduct were manually delineated on the magnitude images by selecting only pixels displaying a characteristic CSF waveform as previously described (45). The resulting pixels were used to measure aqueductal cross-sectional area and CSF flow velocity. Aqueductal CSF velocity amplitude as a measure of CSF pulsatility was calculated from the difference between maximum flow velocity in the positive and negative directions (Fig. 2A). Fig. 1.Origin of cerebral spinal fluid (CSF) pulsatility. This is a sagittal midline illustration of the brain showing the relationship of the choroid plexus (black arrows) to the lateral and third ventricle. CSF flow velocity was measured at the mid cerebral aqueduct (*). The yellow arrows represent the CSF pressure wave caused by transient expansion of the choroid plexus during arterial systole. A rise in intraventricular pressure within the lateral (Lv) and the third (3rd) ventricles causes CSF to flow between the 3rd and fourth (4th) ventricles (dashed yellow arrows). The large transparent white arrows represent CSF flow caused by arterial expansion of the brain parenchyma. The resulting pressure wave is directed toward the central portion of the brain compressing the ventricles, resulting in net CSF outflow through the aqueduct (dashed yellow arrows). Venous outflow from the choroid plexus drains to the torcula herophili (TH), combining with the venous outflow from the superior sagittal sinus and then ultimately to the internal jugular veins (not shown). Fig. 2.Phase-contrast-derived waveforms. A: example of a CSF waveform used to derive the CSF velocity amplitude measurement. Note that CSF flow is bidirectional and that there is a range of velocities within the lumen of the aqueduct in each cardiac phase. The average velocity of this range (MEAN) is shown. Positive velocity is toward the lateral ventricle and negative velocity toward the 4th ventricle. CSF flow velocity amplitude (transparent gray bar) = phase of maximum positive flow velocity (MAX) (solid arrow) – phase of maximum negative flow velocity (MIN) (dashed arrow). B: example of an internal carotid artery (ICA) velocity waveform. Resistive index was calculated as follows: [peak systolic velocity (PSV) – end diastolic velocity (EDV)] divided by PSV. Velocity amplitude of the ICA is shown (transparent gray bar). Note that the CSF phase of maximum negative flow velocity occurs at a delay compared with the PSV. Cine phase-contrast MR is the gold standard for noninvasive CBF determination (65). Volumetric blood flow of the internal carotid artery (ICA) represents ~82% of the total CBF (65) with the remainder supplied by the vertebral arteries. Our analysis focused on CBF derived from the internal carotid artery component, as it provides the large majority of volumetric flow and is less subject to hypoplasia or aplasia as seen with the vertebral artery (80). Anatomic variation of vessel diameter could potentially affect resistive index values; thus the vertebral artery was excluded from the study design (43). CBF was acquired at a plane perpendicular to the ICA just above the carotid bulb centered between the second and third cervical vertebrae. ICA margins were manually delineated bilaterally on a single phase of maximum flow velocity and then automatically propagated for the remaining 50 phases. The phase of maximum flow velocity was also used to determine cross-sectional area. Each phase image was then checked manually and adjusted as needed to maintain accurate delineation of the vascular margins. This technique was preferred over semiautomatic edge detection, which required more extensive manual correction. Peak-systolic velocity (PSV), end-diastolic velocity (EDV), and stroke volume were tabulated for each ICA. Resistive index was calculated from PSV and EDV values (Fig. 2B) and used as a measure of vascular impedance (10). Diffusion-weighted imaging (DWI) was used to measure mean diffusivity of CSF of the lateral ventricle, which is sensitive to changes in temperature, bulk flow, and pulsatile flow (44) as an additional measure of CSF physiology. DWI and 3D T1-weighted acquisitions were obtained using parameters listed in Table 3. DWI data were processed (i.e., masked, coregistered, decoded, and diagonalized) using methodology previously described (28). The average lateral ventricular mean CSF diffusivity was estimated after fusing the T1WI with the DWI data applying methods previously described (26, 27). Lateral ventricular CSF (lvCSF) volume was computed using Statistical Parametric Mapping (SPM) in Matlab as described elsewhere (27, 38).
Neck flexion and head rotation are factors that can alter ICP (88). MRIs were quantified to determine whether head or neck orientation potentially influenced qMRI results. To assess orientation between conditions, consistent anatomic reference points were utilized. Flexion angle of the neck using the straight portion of the vertebral arteries as a horizontal reference of the cervical spine position was measured relative to the vertical plane. Head rotation angle using the septum pellucidum as a vertical reference of head position was measured relative to the horizontal plane (Fig. 3). Fig. 3.Example of the degree of neck flexion and head rotation in head-down tilt (HDT) compared with supine in the same subject. A–C: large field of view sagittal localizers with visualization of the large vessels. Note the vertical dashed line oriented parallel to the vertebral artery from inferior end plate of the second vertebral body to approximately the sixth vertebral body, indicating the longitudinal axis of the straight portion of the vertebral artery (arrows). The angle between vertebral artery and the vertically oriented white line represents the angle of neck flexion of the cervical spine. D–F: axial reconstructed T2-weighted images at the level of the septum pellucidum (arrowheads). The angle between the white dashed line parallel to the septum pellucidum and the horizontally oriented white line represents the angle of head rotation. Actual lines and values used to generate Table 5 are also shown. All statistical analyses were conducted using MATLAB R12.1 Statistical Toolbox v 3.0 (Mathworks, Natick, MA). Statistical significance was set at P < 0.05 without adjustment for multiple comparisons. Paired t-tests compared correlated or repeated measurements on the same subjects. RESULTSCombined results aligned relative to condition are shown in Table 4 with the exception of CSF flow measurements from the first subject. CSF velocity in this subject exceeded the programmed velocity encoding range (VENC = 10 cm/s) in the first session resulting in aliased flow velocities. The VENC was increased to 15 cm/s in subsequent sessions. ICA CBF, CSF velocity amplitude, lvCSF mean diffusivity, and lvCSF volume responses are shown for each subject in Fig. 4. The average height of the subjects was 176.9 ± 3.8 cm, which was used to estimate CAP before and during −12° HDT (Table 4).
Fig. 4.Individual subject responses relative to condition (x-axis). Subjects A–C = ●. Subjects D–F = ○. A: ICA cerebral blood flow rate (left + right). B: CSF velocity amplitude. C: lateral ventricle CSF (LvCSF) mean diffusivity. D: LvCSF volume. There was no significant difference between repeat baseline measurements on the same subjects with the exception of HR, which showed a 4% variance (P = 0.009) indicating overall stability of qMRI measurements and evidence of an adequate readjustment period. There was a 13–29% significant increase in HR increased from baseline to HDT + ambient air and HDT + 3% CO2. This was in contrast to an 11% significant decrease in baseline MAP following HDT + ambient air, which was partially reversed following brief exposure to 3% CO2 (P = 0.33) (Table 4). There was no significant difference in CAP between baseline and HDT in ambient air or 0.5% CO2 (Table 4). There was a 12% significant increase in CAP with exposure to 3% CO2 following HDT with 0.5% CO2 (Table 4). There was a 6–8% significant increase in ICA resistive index from baseline to HDT + ambient air and HDT + 3% CO2 (Table 4). ICA resistive index was stable between HDT + 3% CO2 and each HDT condition (0.5% CO2; ambient air) (P = 0.76; P = 0.94). CBF baseline values are consistent with summated individual ICA values using phase-contrast MR previously reported (right ICA = 253 ± 98 ml/min; left ICA = 240 ± 54 ml/min) (14). There was a 17–20% significant decrease in CBF from baseline to HDT with either concentration of CO2, which was partially reversed by brief 3% CO2 exposure (P = 0.13) (Table 4). There was a statistically significant 21% increase in CSF velocity amplitude from baseline to HDT + 3% CO2 following HDT + ambient air (Table 4). There was no statistically significant difference in aqueductal or ICA cross-sectional areas between baseline and any condition. This indicates that change in aqueductal flow velocity was irrespective of cross-sectional area. LvCSF volume and mean diffusivity at baseline are similar to that previously reported in healthy controls (2,790 ± 100 mm2/s) (26). There was a 2–3% significant increase in lvCSF mean diffusivity with HDT in all conditions (Table 4). A 2–3% increase in lvCSF volume was statistically significant with HDT + ambient air and HDT + 3% CO2 following HDT + 0.5% CO2 (Table 4). There were no significant differences in head rotation or neck flexion angle between conditions (Table 5). Head rotation and neck flexion angle in the same subject in supine and HDT positioning are shown in Fig. 3. Findings suggest that any changes in qMRI measurements between conditions were not likely due to changes in head or neck orientation.
DISCUSSIONOverall, we found that HDT causes a decrease in CBF and a simultaneous increase in lvCSF volume. These alterations are relevant to the VIIP syndrome model because enlargement of the lateral ventricle (51, 84) and many of the potential causes of decreased CBF are associated with reduced craniospinal compliance. We hypothesize that the combination of decreased craniospinal compliance and augmentation of CBF following 3% CO2 exposure acts synergistically to enhance CSF pulsatility. Potential transmission of increased intracranial CSF pulsatility to the orbital structures could represent a new paradigm in VIIP syndrome pathogenesis. Decreased CBF at 26.5 h of HDT is consistent with another short-duration study showing a similar trend at 4.5 h with −6°, −12°, and −18° HDT (57) and indicates that this phenomenon is reproducible and sustained with increased exposure time. Because cardiac output has been shown to increase with HDT (73), the observed decrease in CBF is unlikely related. Alternatively, because CBF is proportional to the arterial-venous (a-v) pressure gradient divided by vascular resistance (40), decreased CBF is likely derived from increased arterial and/or venous resistance and/or a reduction in the a-v pressure gradient. Arterial vasoconstriction has been shown to decrease CBF in hindlimb-suspended rats (87). Similarly, in our study, the elevated resistive index with HDT in ambient air at 26.5 h could indicate increased arterial vasoconstriction. Increase in cerebral artery resistive index, which peaked at 4–5 days, was also confirmed in a −6° HDT study using Doppler ultrasound (3). Vasoconstriction with HDT is hypothesized because of a myogenic cerebrovascular autoregulation response (29). The concept of cephalad fluid shifting in microgravity was derived from the observation of facial congestion, jugular venous distention (4), and increased IJV cross-sectional area of up to 47% during spaceflight (3). Because CBF was shown to be inversely related to internal jugular vein cross-sectional area during HDT (57), we hypothesize that the increased inertia of venous blood mass pooling between the cranium and heart would decelerate blood flow according to Newton’s second law of motion (force = mass × acceleration) assuming a constant hydrodynamic force (8). In support of this hypothesis, internal jugular vein velocity was noted to decrease significantly from supine to −12° HDT (57) although increased vascular resistance arising from arterial vasoconstriction could also decelerate flow. The net effect would be dampened venous outflow, reduced craniospinal compliance, and increased ICP pulsatility. The hydrostatic gradient generated by HDT is dependent on fluid density, fluid column height, and gravitational acceleration (49). With HDT, pressure in the jugular veins and cerebral venous structures including the cortical bridging veins would be expected to increase compared with supine, as predicted by the hydrostatic indifference point, located just below the diaphragm (69). Similarly, hydrostatic forces generated by HDT would augment CAP (58). However, estimated mean CAP at 26.5 h of HDT was not significantly changed from baseline levels. A possible explanation is the diuretic effect induced by HDT (37). This would result in decreased plasma and blood volume (76) and a concomitant reduction in blood pressure (75). A decrease in MAP at 26.5-h HDT compared with increased MAP reported at 4.5-h HDT (57) is consistent with this hypothesis. In a −20° HDT experiment with cats, both MAP and sagittal sinus pressure increased initially with a greater increase in MAP (42), which would increase the a-v pressure gradient. Sagittal sinus pressure remained stable over 2 h; however, there was a gradual decline in MAP resulting in only a mildly decreased a-v gradient compared with baseline (42). Although unconfirmed in our study because of the lack of direct vascular pressure measurements, a mild decrease in the a-v pressure gradient could contribute to a reduction in CBF. Additionally, HDT-induced elevated intracranial venous pressure can stiffen cortical bridging veins (34), thereby decreasing craniospinal compliance (5) and enhancing ICP pulsatility. HDT is thought to reverse positional ischemia in patients of acute stroke (11); however, the results of our study would caution otherwise. This concern is also supported by a study in healthy adults that showed decreased regional blood flow in the prefrontal cortex during −15° HDT, which is a region of the brain involved in attention, reasoning, and executive function (68). The Monro-Kellie doctrine developed over two centuries ago to characterize the pressure-volume relationship of ICP assumed rigidity of the cranium and incompressibility of the intracranial contents (63). However, recent studies have shown that added intracranial volume can be absorbed by slight expansion of the cranium (31), indicating some inherent cranial elasticity. Furthermore the dural venous sinuses have shown compressibility with elevated ICP (77). These factors in combination with the viscoelastic properties of brain parenchyma (62) would help account for lvCSF volume expansion with −12° HDT and recovery of baseline volume following the 7-day readjustment phase. Enlargement of the lateral ventricles could result from increased venous pressure with accumulation of CSF from decreased reabsorption at the arachnoid granulations (6). A predicted increased CSF hydrostatic gradient (56) could also result in a direct cranial shift of CSF volume from the spinal canal. Dysfunctional lymphatic contractility secondary to HDT (23) can impede CSF reabsorption within the perivascular spaces of the olfactory nerve arachnoid sheath (78) and also expand CSF volume. CSF volume expansion with any or all of these mechanisms could cause reduction of cranial elasticity, stiffer brain tissue, and a complementary volume decrease in the more compliant venous compartment (12, 51), either of which would decrease craniospinal compliance. The net effect of ventriculomegaly is therefore augmented ICP pulsatility in response to pressure waves from arterial blood flow (51). A previous study with up to 90 days of bed rest at −6° HDT noted that ventricular enlargement correlated only with posterior brain rotation possibly attributable to compression of the dural venous sinuses at the vertex with resultant venous outflow obstruction altering CSF flow dynamics (76). In this study, there was no significant change in ventricular volume after HDT when analyzing the entire cohort (76). We hypothesize that −12° HDT pursued in our study creates a more significant alteration in CSF flow dynamics and may help to explain a significant difference in ventricular volume after HDT in our study at 26.5 h. The CSF waveform is intrinsically related to CBF by two proposed mechanisms. Bering in 1955 suggested that arterial pulsations in the choroid plexus act as a CSF pump (7) causing oscillatory expansion and contraction of the choroid plexus within the lateral and third ventricles with resultant bidirectional displacement of CSF through the cerebral aqueduct (25, 50) (Fig. 1). More recently, because of the apparent absence of pressure gradients surrounding the choroid plexus calculated by 4D phase-contrast flow analysis (60), an alternative theory was proposed suggesting that arterial perfusion of the cortex and subcortical tissues creates a pressure wave that is directed toward the ventricular system, resulting in ventricular compression (72) (Fig. 1). Regardless of mechanism, the end result is elevated intraventricular CSF pressure, which drives aqueductal CSF outflow. Therefore, any reduction of craniospinal compliance would be expected to augment intraventricular pressure during arterial pulsation and increase CSF flow velocity through the cerebral aqueduct. Hypercapnia-induced arterial vasodilation is a known cause of intracranial hypertension attributable to increased CBF with progressive volume expansion of the intracranial vascular bed (33, 67). However, the stability of the resistive index after exposure to the potent vasodilator effects of 3% CO2 (74) suggests that hypercapnia-induced vasodilatation may be attenuated in HDT. Alternatively, increased CBF following 3% CO2 exposure could reflect increased cardiac output, MAP, and HR that can occur with hypercapnia (39). Additionally, a decrease in intrapleural pressure associated with increased depth of respiration attributable to hypercapnia (24) would decrease right atrial pressure and facilitate venous return to the heart (41). Improved venous flow to the heart from the jugular veins could result in lower venous outflow resistance and increase CBF. Although increased CBF would appear beneficial, there is an important downside. Increased pulsatile CBF in conjunction with reduced craniospinal compliance could increase ICP pulsatility during HDT (51). We hypothesize that this is manifested by augmented CSF pulsatility following 3% CO2 exposure. According to Poiseuille’s law for laminar flow, the larger amplitude in CSF flow velocity would correspond to a larger variation in pressure gradient at the cerebral aqueduct over the cardiac cycle (71). Augmented ICP pulsatility with HDT is supported by the discovery of increased intracranial diameter pulsatility, a surrogate marker of ICP pulsatility, after a 1-min exposure to HDT shown by pulse-phase lock-loop ultrasonography (53). Active downregulation of CSF production inferred in astronaut and animal microgravity studies (22, 32, 45, 59) may help explain gradual normalization of static ICP with HDT (15, 48, 82). However, with fluctuating CO2 levels as occurs aboard the ISS (47), ICP homeostasis may be more difficult to achieve. In support of this hypothesis, ICP pulsatility was most severe during exposure to 3% CO2 that followed ambient atmosphere, suggesting evidence of adaptation to chronic 0.5% CO2. ICP susceptibility to rising CO2 levels is also suspected by an ISS study showing that the risk of headache, a possible symptom of elevated ICP, doubles with each 1-mmHg increase in CO2 partial pressure (47). Furthermore, pulsatile ICP was elevated in 100% of patients with idiopathic intracranial hypertension who benefited (reduced visual disturbance and headaches) from CSF diversion regardless of the fact that 50% of these patients had normal mean ICP preoperatively (18). This latter observation suggests that ICP pulsatility can contribute to pathophysiology even in the absence of elevated static ICP. The relatively small difference in static ICP during parabolic flight-induced zero gravity and baseline supine posture (48) indicates that static ICP with normal atmospheric levels of CO2 may not be a VIIP syndrome risk factor. However, the long-term effects of acute elevations in ambient CO2 in conjunction with reduced craniospinal compliance causing episodes of increased ICP pulsatility warrants further investigation. Because CBF is an important component of intracranial thermal dissipation (85), we hypothesize that increased lateral ventricular mean diffusivity induced by HDT is a manifestation of elevated brain core temperature (26). Although reduced mean diffusivity of the lateral ventricle would be expected with exposure to 3% CO2 because of the cooling effects of increased CBF (85), the paradoxically increased lateral ventricular mean diffusivity may alternatively reflect increased CSF pulsatility (44), as indicated by the identification of augmented CSF velocity amplitude. Increased optic nerve sheath diameter (OSND) and flattening of the posterior globe (1) are abnormalities found in astronauts exposed to spaceflight (46, 55). The latter finding is associated with acquired hyperopia (79), establishing an anatomic correlate for altered visual focus. Because the pressure in the perineural subarachnoid space surrounding the optic nerve has a linear relationship with ICP (52), perineural transmission of increased ICP pulsatility could play a role in pathogenesis of orbital abnormalities. The hydrodynamic forces developed during HDT that can pulsate and expand the rigid cranium (53, 83) and enlarge the lateral ventricle could also be potentially transmitted to the optic nerve sheath, causing gradual increase in ONSD and compression of the posterior globe via a water hammer mechanism. This hypothesis is also supported by a study that showed increased OSND with −30° HDT following robotic prostate surgery (86). There are important limitations of our study. 1) The number of subjects that participated in this preliminary study was small and included all males. 2) Invasive measurements of ICP were not obtained in this healthy adult population, and thus changes in ICP were inferred by CSF flow characteristics. 3) The study represents findings from a ground-based analog of spaceflight with simulated conditions, and therefore extrapolation of results to real microgravity warrants caution. A case in point was an unanticipated drop in central venous pressure upon entering microgravity, opposite to that shown by −6° HDT (9). 4) The effect of 3% CO2 in the absence of HDT was not studied as a control condition, and thus the synergistic effect of the combination of HDT and 3% CO2 on CSF pulsatility will require additional validation. Our study illustrates the importance of the combination of ambient CO2 and HDT-induced cephalad fluid shifting as a ground-based model to study the VIIP syndrome. HDT results in changes in anatomy and physiology that can be associated with reduced craniospinal compliance and a diminished capacity to maintain ICP homeostasis. In this altered state, brief exposure to high levels of CO2 augments CSF pulsatility, a potential marker of increased ICP pulsatility. We propose that the pulsatile hydrodynamic forces generated can remodel orbital structures via water hammer effect transmitted to the optic nerve sheath. Finally, because microgravity is an inevitable factor in space travel for the foreseeable future, reduction of CO2 levels aboard the ISS and future long-range spacecraft should be given consideration as an environmental countermeasure of the VIIP syndromes. GRANTSThis study was funded by a grant from the National Space Biomedical Research Institute via NASA NCC9-58 and Center for Space Medicine, Baylor College of Medicine. The DLR contribution to this study was funded by DLR-internal cost object 2475 115. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSL.A.K., K.M.H., S.H., and B.M. analyzed data; L.A.K. and K.M.H. interpreted results of experiments; L.A.K. and K.M.H. prepared figures; L.A.K. and K.M.H. drafted manuscript; L.A.K., K.M.H., A.E.S., K.M.-G., J.R., D.D., D.A.G., and E.M.B. edited and revised manuscript; L.A.K., K.M.H., and E.M.B. approved final version of manuscript; D.A.G. and E.M.B. performed experiments. We are very thankful to our test subjects. Without their selfless contribution, this research would not have been possible. SPACECOT Investigators Group was composed of Jose I. Suarez, Chethan P. Venkatasubba Rao, Jonathan Clark, Eusebia Calvillo, Edwin Mulder, Gary Strangman, Sushmita Datta, Haleh Sangi-Haghpeykar, Ulrich Limper, Brian Stevens, Mathias Basner, Jad Nasrini, Martin Wittkowski, Matthias Putzke, Henning Stetefeld, Christian Dohmen, Tobias Weber, Petra Frings-Meuthen, Peter Gauger, Worfram Sies, Kläus Muller, Claudia Stern, and Annette von Waechter. REFERENCES
Page 13a viscous fluid lining layer containing pulmonary surfactant, which is produced by type II alveolar cells, protects pulmonary airways and alveoli. Surfactant reduces the surface tension dynamically, reduces the energy of breathing, and plays a role in host defense. Furthermore, surfactant biophysical properties protect airways and alveoli from surface-tension-induced liquid instabilities that can obstruct airways and impede gas exchange (8, 11, 16). In disease states such as acute respiratory distress syndrome (ARDS), extreme pulmonary edema occurs, leading to flooding of the air spaces. Surfactant properties can become deranged because of infiltration of plasma proteins that competitively adsorb to the air-liquid interface (48). ARDS has a mortality of 30–40% (~75,000 US deaths annually) and arises from insults such as bacterial infection (sepsis), liquid aspiration, or noxious gas inhalation. A hallmark of ARDS is the presence of fluid-filled lungs with high-surface-tension interfaces. ARDS afflicts individuals in the prime of their life, and survivors may suffer from long-term neurological deficits, depression, and decreased quality of life (20). Furthermore, ICU length of stay correlates with decreased quality of life, so improved treatment can have a long-term impact on patients and their families (19). While surfactant inactivation during ARDS and lung injury is significant, surfactant replacement therapies have generally not been clinically successful (42). Recent model studies suggest that while surfactant replacement therapy is effective in the neonatal lung, the same delivery technique in the adult lung leads to a poorly mixed, highly nonlinear distribution that could cause inadequate delivery (8). Therefore, to date there are no medical treatments for acute lung injury/ARDS other than supportive care revolving around mechanical ventilation. Unfortunately, mechanical ventilation can cause ventilation-induced lung injury (VILI) through mechanisms associated with volutrauma and atelectrauma (30, 33). Therefore, mechanical ventilation is a “double-edged sword” because it is necessary for recovery and yet can exacerbate lung damage. Our goal is to develop an understanding of the surfactant distribution behavior and biomechanical stimuli-response relationships associated with ventilation to provide guidance toward improved ventilation approaches that can reduce VILI. A number of studies have investigated the biomechanical interactions leading to atelectrauma at the airway scale (12, 31) and at the cellular level (5, 7, 23, 25). Those studies focused on the pressures necessary to recruit individual airways, the mechanical stimuli created by recruitment, and the biological responses to those stimuli. Those studies indicate that the pressure gradient near the tip of the finger of air that removes the liquid obstruction provides the strongest correlation to the degree of epithelial damage and barrier function (14, 22, 24, 25). However, those studies investigated only individual airways and did not address the complex geometries associated in the lung. The behavior of a bifurcating network is much more complex than that of individual airways and has important physiological implications. For example, nonuniform recruitment could result in large regional variations of ventilation. In turn, this can result in a “baby lung” scenario that results in volutrauma to those portions of the lung that are patent, while other regions remain atelectic. A full understanding of this type of interaction would require a multiscale model that links airway and alveolar fluid-structure behaviors that incorporate physicochemical interactions (2, 9, 17, 32, 36, 39). Successful full-lung models will require an understanding of the interactions that occur in the region of a bifurcation and the relative responses of daughter airways. In a previous study (45), we initiated investigations of the recruitment of bifurcating airways, following prior studies by others (3). That study investigated the microfluidic behavior of reopening a simple symmetric Y-shape channel—this demonstrated pulmonary surfactant’s unique ability to maintain nearly symmetric recruitment of daughter airways, which could improve the likelihood of uniform recruitment of a multigenerational bifurcated network and thus maximize the potential for global opening. Mathematical modeling provided insight that suggests that the dynamically altered pressure drop across the air-liquid interface of the propagating finger of air stabilized the uniform reopening of the nearly symmetric system. That analysis illustrated why simply lowering the surface tension was not sufficient for uniform recruitment; rather, a velocity-dependent surface tension provides a variable pressure drop that is instrumental in stabilizing the system. The present study follows up on the investigation of nearly symmetric bifurcations with the goal of identifying how the carina geometry and daughter airway asymmetry interact with surfactant properties to influence the uniformity of recruitment. The goal of this study was to investigate how microscale features of the bifurcation in concert with surfactant properties and distribution can affect larger-scale recruitment phenomena. This, in turn, may lead to an understanding of how ventilation processes could be modified to improve ventilation homogeneity in the entire lung through a redistribution of surfactant near airway bifurcations that would encourage the simultaneous reopening of asymmetric obstructed daughter airways. METHODSExperiments were designed to discriminate between different mechanisms that can lead to, or ameliorate, nonuniform recruitment of bifurcating airways. These mechanisms are described in detail in Ref. 45 and synopsized here. We follow Refs. 13 and 34 to partition the bubble pressure drop along the fluid-occluded section of a daughter channel as consisting of three components (Fig. 1): PBubble=PCap+PHyd+PEnd(1) Fig. 1.Schematics of 3 patterns of asymmetric bifurcations. A: asymmetric width designs (W1>W2), with consistent inlet width (Inlet-Consistent = IC, W1/W2 = I1/I2), and with inverted inlet width (Inlet-Inverted = II, W1/W2 = I2/I1). B: Length-Asymmetric design. All designs share the same parent channel width, W0 = 208 μm, sum of daughter section width, W1+W2 = 300 μm, and Ch.1 section length, L1 = 4,800 μm. PCap is the pressure jump that exists over the air-liquid interface at the progressing bubble tip, which is related to the nonequilibrium normal stresses (law of Laplace) at low capillary number (Ca), PCap∼2γRInterface(2) where RInterface is the interfacial radius of curvature. PCap can therefore be modulated by a change in RInterface or by physicochemical modifications of the surface tension, γ.PHyd is the hydraulic pressure loss due to viscous flow downstream of the bifurcation, which is approximated by channel Poiseuille flow, PHyd∼8μURHyd2L(t)(3) where μ is viscosity, U is the velocity, RHyd is the hydraulic radius, and L(t) is the length of the fluid-filled segment downstream of the air-liquid interface.PEnd is the pressure at the end reservoir. In our model system (see Fig. 1A), the daughter channels share a common end reservoir, the pressure at which can be defined as the reference pressure, PEnd=0(4) When the bubble splits to open the daughter channels (Ch.1 and Ch.2 in Fig. 1B), the pressure drop from the bubble to the end reservoir must balance. Therefore, the sum of the capillary and hydraulic pressure drops must be equivalent through Ch.1 and Ch.2: (PCap)1+(PHyd)1=(PCap)2+(PHyd)2(5) So, (2γ1(RInterface)1)+(8μU1(t)L1(t)(RHyd2)1)=(2γ2(RInterface)2)+(8μU2(t)L2(t)(RHyd2)2)(6) In our experimental models we perturb the geometry to investigate the effect of modification of the daughter channel widths (W1 and W2), which simultaneously influence the hydraulic radius given as RHyd=WD / (W+D) in each of these branches. In addition, we perturb the inlet channel widths (I1 and I2), which modulates RInterface only at the entrance (a near-field effect). By methodically investigating the reopening behavior with these physical models using fluids of different physicochemical properties, we can satisfy our goal of identifying geometric and transport properties that influence the uniformity of airway reopening in branching networks. All microfluid experimental devices were constructed with polydimethylsiloxane and a standard photolithography fabrication method (44). The interior of the flow chamber was treated with vinyl-terminated polydimethylsiloxane (DMS-V05, Gelelst) after exposure to oxygen plasma; this process sustains the hydrophilic surface properties throughout the duration of our experiments (38). Each asymmetric model was created as a variation of the symmetric model shown in Fig. 1A, which is a two-dimensional version of idealized geometric properties of pulmonary bifurcations (29). We define the parent channel as channel 0 (Ch.0) and the daughter channels as channels 1 and 2 (Ch.1, Ch.2). The following properties exist in all models: 1) the depth of all channels is uniform, D = 150 μm; 2) the parent channel width is W0 = 208 μm; 3) the sum of the daughter channel widths is conserved, W1+W2 = 300 μm; 4) the daughter channels bifurcate at angle of 35° with initial branching curvature Rbi = 525 μm; 5) the carina tip radius of curvature is Rc = 75 μm; 6) each daughter links to outflow tubes ~700 μm from the bifurcation; and 7) daughter branches are connected to a unified outlet downstream (2,400 μm from bifurcation) so as to apply a uniform downstream pressure. The goal of this study was to identify the importance of geometric and physicochemical interactions and their effect on recruitment uniformity in asymmetric bifurcations. As such, we developed three distinct bifurcation variants to discriminate between capillary and hydraulic interactions that contribute to flow behavior. These variants are shown in Fig. 1, with specific dimensions listed in Table 1, and described below.
In these models the daughter channels have asymmetric widths (W1>W2) but symmetric lengths (L1 = L2). This design sets each entrance width to be equal to the related daughter channel width (I1 = W1, I2 = W2), hereafter referred to “Inlet-Consistent” and denoted by the subscript “IC.” As described in Experimental Design Principles, this introduces an asymmetry of PCap both at the entrance and within the daughter branches. Furthermore, it introduces an asymmetry of PHyd. Without physicochemical interactions that cause dynamic surface tension, this asymmetry should lead to an initial preferential reopening of Ch.1 because it would have a lower PCap. These devices are designed to have symmetric channel widths (W1 = W2) but asymmetric channel lengths (L1/L2 = 0.974) as shown in Fig. 1B. With this design, asymmetric flows are initiated solely by a differential in PHyd. Since PHyd ~ L / RHyd2 this system has a hydraulic pressure ratio at the bifurcation entrance equal to(PHyd)1 /(PHyd)2=L1 / L2=0.974. This hydraulic pressure ratio is equivalent to experiments with (152/148)IC apparatus. Since no differential in PCap exists, it is predicted that a significantly smaller degree of reopening asymmetry will exist in comparison to (W1/W2)IC experiments. Therefore, gradual increment of preferential opening of Ch.1 will be expected if no surfactant-induced physicochemical dynamic interactions on the moving bubble tip interface exist. As with (W1/W2)IC, the daughter channels have asymmetric widths (W1>W2) and symmetric lengths (L1 = L2), but this design is used to evaluate the sensitivity of carina position by moving the carina tip (Fig. 1A, dotted line) so that the inlet width (I) is equal to the opposite daughter channel width (I1 = W2, I2 = W1), hereafter referred to “Inlet-Inverted” and denoted by the subscript “II.” This design was chosen to highlight the inlet flow behavior, since the capillary pressure ratio is inverted from the capillary pressure ratio in the daughters, [(PCap)1 /(PCap)2]Inlet=[(PCap)2 /(PCap)1]Channel and [(PCap)1>(PCap)2]Inlet while maintaining [(PHyd)1<(PHyd)2]Channel before the bubble bifurcation (see Table 1). So, under constant-surface tension conditions the inlet characteristics will cause an increased propensity for Ch.2 to begin opening preferentially, even though the daughter channel asymmetry is likely to result in long-range preferential flow in Ch. 1. Therefore comparison of the inlet behavior with various surfactant solutions will provide insight into the effect of nonuniform surfactant distribution on the air-liquid surface of the progressing bubble tip at the inlet region (45). Various test solutions were used to investigate the influence of physicochemical behavior on reopening uniformity. We chose these solutions on the basis of static and dynamic surface tension properties, as acquired by measurement with a barrier close/open five-cycle experiment using a Langmuir trough (KSV NIMA Biolin Scientific) with barrier speed of 1.0 mm/s at 37°C. The constant barrier speed is set to approximate average bubble speed in the bifurcation microfluid apparatus (see Experimental Procedure). The static surface tension, γ0, was obtained from the measurement before the start of the isotherm barrier close/open experiment. Δγ is reported as the difference of maximum and minimum surface tension in the last three cycles of oscillation. Dulbecco’s phosphate-buffered saline (DPBS) is a buffer solution and is used as the solvent phase for all surfactants in the experiment. This solution provides a constant surface tension γ = 65 dyn/cm under static and dynamic tests. Sodium dodecyl sulfate (SDS) is an anionic surfactant with a high molecular diffusion rate near the air-liquid interface. The concentration (C = 1.73 mg/ml) is slightly below its critical micelle concentration. This solution reduces the surface tension to approximately γ0 = 30 dyn/cm and has only a very small dynamic surface tension variation, Δγ = 4 dyn/cm. Infasurf (calfactant) (ONY) is an exogenous pulmonary surfactant analog used in surfactant replacement therapy. Concentrations are selected to investigate the importance of dynamic surface tension. With these concentrations, we found γ0 = 53, 31, 24 dyn/cm and Δγ = 27, 55, and 30 dyn/cm for C = 0.01, 0.1, and 1.0 mg/ml, respectively. This suggests that different concentrations of Infasurf will display various degrees of sensitivity to the carina tip and channel asymmetries throughout the experiments. Table 2 summarizes the static and dynamic surface properties of these test solutions obtained from a Langmuir trough/Wilhelmy plate (LTWP) surfactometer for DPBS and SDS and a pulsating bubble surfactometer for Infasurf, as described in discussion.
Particle image velocimetry (PIV) is well established and one of the most versatile tools to acquire whole field instantaneous flow fields under complex conditions (1). In past years, our laboratory has developed a microscale PIV (μ-PIV) system specialized for analysis of microscale interfacial flows during airway reopening (37, 46, 47). The modified μ-PIV system is depicted in Fig. 2B. Optics principles and system operation are unchanged from the system described in Yamaguchi et al. (45, 46). Volumetric illumination is provided by a dual-pulse Nd:YAG laser (λ = 532 nm, power = 15 mJ/pulse, duration = 4 ns; New Wave Laser Pulse Solo Mini, New Wave Research-ESI, Fremont, CA). The test solution is seeded by 1.0-μm fluorescent particles that have excitation/emission peaks at 535/575 nm to provide a return signal (λ = 550 nm) with minimal background noise. The PIV camera (PowerView Plus, TSI, Minneapolis, MN) CCD provides 2k × 2k effective pixel images with pixel resolution of 0.658 μm/pixel. In combination with a selected objective lens (×10/0.30 Plan Fluor, Nikon), the focal plane was set to the center of channel depth at ~15 μm. The camera and laser timing is controlled by a multichannel synchronizer and software (model 610035, TSI). The process control, image display, and postprocessing were coordinated by Insight 4G (TSI). Fig. 2.Schematic of fluid apparatus basic design (A) and experimental setup (B). Linear motor with microsyringe (i) and the translating stage (37) are used to set initial bubble position in frame i, so that the observation window is moved to frame ii. Bubble progression and PIV image acquisition are controlled by PC-linked syringe pump (ii), CCD camera, and the pulsed laser system. *Detailed optical configuration of the PIV is available in Refs. 45 and 47. All test solutions contained 0.04 vol% of 1.0-μm-diameter fluorescently dyed particles (Nile red, λ = 535/575 nm; Molecular Probes) in order to obtain fluorescent images with sufficient vector resolution for computing average flow rate of each channel instantaneously. The image acquisition observation window is adjusted to have the bifurcation carina at the left edge of each image (Fig. 2A, frame ii). With this orientation, we capture the image of the interface as it migrates through the system and observe the downstream liquid phase for interrogation of instantaneous flow velocity. The combination of computer-controlled linear actuator/syringe pumps (electromagnetic direct linear motor P01-23x80/30x90 and E200-AT, Linmot) and the motorized sliding stage allows us to set the initial bubble tip position ~10 mm upstream of the bifurcation point (Fig. 2A, frame i), before each run. The unified inlet (left reservoir in Fig. 2A) is open to the air, and the occluded fluid is drained from the unified outlet (right reservoir in Fig. 2A) at Q = 1.17 μl/min to achieve an average flow velocity in the parent channel of approximately U = 0.5 mm/s. During each trial, image acquisition consists of 80 pairs of images at a rate of 7.25 pairs/s. The bubble tip appears after fully developed flow has commenced as shown in Fig. 3, after 4 s from the initiation of the flow. This provides a sufficient duration to capture the entire bubble bifurcation process and progression in daughter sections in the frame. Each experiment is completed n = 5 times for statistical analysis. Fig. 3.Example of fluorescent particle image in the vicinity of a bifurcation. Bright region demonstrates the emission of high concentrations of fluorescent particles in the occluding liquid phase. Instantaneous bubble velocities and corresponding positions in daughter sections can be closely estimated from the PIV analysis in the dashed rectangular areas. The interrogation of fluorescent images employs a recursive Nyquist grid with an FFT correlation engine and a Gaussian peak algorithm with a first 64 × 64-pixel interrogation window and a second smaller 32 × 32-pixel (with 50% overlap) interrogation window. An expected range of the progressing bubble speed in the daughter section is ~0–0.9 mm/s; therefore Δt = 4.0 ms is set as the time delay between each pair of fluorescent particle images in order to have a displacement of fewer than 8 pixels (25% of interrogation window size). The present image acquisition system provides a pair of 2k × 2k pixel images with pixel resolution of 0.658 μm/pixel at 7.25 sets/s. Therefore each set of images can generate a 128 × 128 vector field with spatial resolution of 10.53 μm/vector at a temporal resolution of 7.25 frames/s. Since the daughter channel widths are ~150 μm, the observation frame provides ~14 vectors, which resolves sufficient detail to estimate velocity profile and maximum velocity at center line. Since the focal plane of the μ-PIV measurement is adjusted to the middle of the channel depth, the maximum velocity of the channel can be closely approximated and then converted to an average flow rate by using rectangular duct flow profiles, following Ref. 45. We define t = 0 as the moment when the progressing interface reaches the carina. To assess reopening behavior, we monitor the instantaneous positions of the bubble tip positions downstream of the carina for each branch [B1(t), B2(t)] in Fig. 1. To do so, we integrate the instantaneous flow velocity beginning from the time that the interface reaches the carina [t = 0 s; B1(0) = B2(0) = 0], and we use conservation of mass to predict the position by assuming that the residual film thickness is insignificant. Since the Ca for the system is very low (∼10−6), this estimate of bubble position is accurate (6, 10, 18, 21, 43). Henceforth, we use the bubble position ratio, B1(t)/B2(t), as the key parameter to quantify the reopening uniformity. RESULTSAs described in Description of the Asymmetric Bifurcation Microfluidic Device, three distinct types of asymmetric models were fabricated: (W1/W2)IC, (L1/L2), and (W1/W2)II. To validate the accuracy of these designs, we measured the flow rate ratio Q1/Q2 for single-phase viscous flow through each branch. For example, the slightly asymmetric group, (152/148)IC, (152/148)II, and L1/L2 = 0.974, were designed to obtain the same hydrodynamic resistance and should therefore have an identical flow rate ratio, Q1/Q2 = 1.027. These flow rate ratios were evaluated experimentally with the μ-PIV system. The experimental data in Table 1 confirmed that we have fabricated asymmetry groups that provide three different distinctive degrees of asymmetry and that the variation in asymmetry between each subgroup is <3%. Representative data of experiments with constant surface tension (DPBS) and dynamic surface tension (0.1 mg/ml Infasurf) are presented in Fig. 4, with t = 0 defined as the instant when the bubble tip impinges the carina. These data represent the positions of n = 5 runs, and each data point has a 5–10% standard deviation (data not shown). The presented data sets are selected from the smallest degree of asymmetry [(152/148)IC, L1/L2 = 0.974, and (152/148)II], since the more highly asymmetric systems show less sensitive behavior to different test solutions. Furthermore, DPBS and C = 0.1 mg/ml Infasurf were selected to demonstrate the impact of physicochemical interactions on reopening asymmetry, since these responses bracket the behavior of other solutions. Fig. 4.Representative data set of airway reopening experiments under Inlet-Consistent (a); Length-Asymmetric (b), and Inlet-Inverted (c) conditions. Each data point represents the average of 5 trials. As shown previously (45), the stabilizing effect of pulmonary surfactant is clearly visible in all asymmetric patterns. Overall, asymmetric designs (W1/W2)IC (a) and L1/L2 (b) demonstrate a consistent flow asymmetry that diverges flow to Ch.1, which has the lower resistance. As hypothesized above, modification of (W1/W2)IC causes a greater divergence to Ch.1 because modification of W simultaneously affects PHyd and PCap, while variation of L only affects PHyd. To further decouple the effects of the capillary pressure drop from the hydraulic resistance, we include studies in which the inlet width ratio I1/I2 is inverted from the downstream width ratio W1/W2, as described above and denoted (W1/W2)II. These experiments [(c), Fig. 4] demonstrate that the inlet effect on PCap induces Ch. 2 to reopen first, even though the overall hydraulic resistance in Ch. 1 is lower than that of Ch. 2 (as shown by the flow rate ratio in Table 1). Once Ch. 2 begins to reopen, however, the lower PHyd in Ch. 1 eventually leads the interface of Ch.1 to increase its relative velocity. The time at which the interfacial tip positions are equivalent (B1 = B2) is termed the “reconvergence time” (RT), and it is evident that physicochemical interactions induced by surfactant (C = 0.1 mg Infasurf) substantially increases RT. Below, we use RT as evidence of the nonuniform surfactant distribution along the interface near the progressing bubble tip, which affects reopening uniformity. DISCUSSIONFigure 5 and Figure 6 present log-log plots of the reopening asymmetry behavior for (W1/W2)IC and L1/L2 experiments, respectively; the x-axis presents the interface position ratio B1/B2, and the y-axis presents the velocity ratio U1/U2 at the time when B1 = 450 μm. Figures 5A and 6A represent the specific data, and Figs. 5B and 6B schematically illustrate the general trends. We chose the location B1 = 450 μm because that represents the position that would correlate to the end of a generation following Wiebel and Gomez (40), which describes the average length-to-diameter ratio of an airway in the respiratory zone as L/D ~ 3. Fig. 5.Width-Asymmetric/Inlet-Consistent (IC) experiments. A: raw data. B: trend representation. *Data from Yamaguchi et al. (45). Fig. 6.Length-Asymmetric experiments. A: raw data. B: trend representation. Scale of velocity ratio is almost 10 times smaller than IC data (Fig. 5), even though initial flow rate ratios are nearly identical. *Data from Yamaguchi et al. (45). In Figs. 5 and 6, perfect reopening symmetry is represented by the point B1/B2 = U1/U2 = 1. If asymmetric reopening were consistent (i.e., a constant nonunity velocity ratio), then at any given time the data would exist at B1/B2 = U1/U2. Figure 5 demonstrates that symmetric reopening under (W1/W2)IC conditions does not occur and that in most cases B1/B2≪U1/U2, indicating that the reopening asymmetry increases rapidly. These trends are also evident in Fig. 6 (L1/L2 experiments), but the degree of reopening asymmetry is significantly smaller. We can understand the asymmetric reopening behavior of Fig. 5 through (W1/W2)IC data (a) in Fig. 4. It is clear from the time-dependent behavior in Fig. 4 that two distinct regions exist. These consist of an initial region with simultaneous nearly symmetric motion, followed by an abrupt transition to asymmetry. The initial region is driven by a transitional increase in PCap, since RCap is reduced as the interface splits to enter into the daughter segments. This pressure increase requires a small time for the yield pressure to be reached, and the bubble progresses slowly to a position that is approximately one channel width downstream (B∼W), with both daughter vessels reopening as the interfaces slowly creep at nearly the same speed. Subsequently, the reopening rates diverge substantially; Ch.1 accelerates quickly, while Ch.2 decelerates (especially with high constant surface tension, as with DPBS). In contrast, for length-asymmetric channels [L1/L2 (b)] Fig. 4 and Fig. 6A show that the relative velocities are much smaller. In this case, Fig. 6A shows that the dynamic surface tension induced by Infasurf can cause nearly symmetric reopening to occur up to B1 = 450 μm despite the fact that the apparatus has a clear channel length asymmetry. The overall trends from of the (W1/W2)IC and (L1/L2) experiments are shown by schematic representations in Fig. 5B and Fig. 6B, respectively. Figure 5B shows that an increase in the width ratio substantially increases the asymmetry. For example, only a 15% differential in width, (160/140)IC, results in a 100-fold change in relative velocity when a constant high-surface-tension fluid, DPBS, is used. However, the introduction of reduced static surface tension (SDS) or high dynamic surface tension (Δγ/γ0) (0.01Infa, 0.1Infa, and 1.0Infa) substantially reduces the magnitude of asymmetric reopening. Figure 6B does not demonstrate as strong an influence of L1/L2 on reopening asymmetry, likely because it only explicitly modifies PHyd. However, decreasing the surface tension reduces the asymmetry, and this behavior is made more prominent with large dynamic surface tension, Δγ/γ0. The relative effects of PCap and PHyd can be understood from dimensional analysis for constant-surface tension fluids using Eqs. 2 and 3. Assuming a moderate surface tension γ = 50 mN/m (dyn/cm) and an obstruction fluid of viscosity equivalent to H2O, the magnitude ratio of the two contributors to the pressure drop, PCap/PHyd, based on idealized geometric properties of respiratory airways (29), is PCap/PHyd ∼ O(102). The significantly larger impact of PCap in the present apparatus elucidates why the influence of width asymmetry is so much greater than that observed in length-asymmetric models. The magnitude of this effect is demonstrated by the gray-shaded region of Fig. 5B, which represents the entire domain of Fig. 6. The relative importance of PCap helps to explain the extraordinary sensitivity of the system to dynamic surface tension. The variable surface tension modifies the relative surface tension ratio between Ch.1 and Ch.2 (γ1/γ2). This serves to equilibrate the hydraulic pressure drop in each segment by increasing the PCap in the branch that opens more quickly. To quantify this relationship, we hypothesized that a combination of sufficiently strong surface flow and slow sorption rates of the pulmonary surfactant generates an interfacial surfactant concentration difference between the daughter segments that increases the relative surface tension along the fast reopening branch in comparison to the slow branch (45). Therefore, by modulating the surface tension a differential in PCap is created between the branches. Since the total pressure drop is fixed (Eq. 5), this modifies the pressure available for flow, PHyd, which helps to equilibrate the bubble progression. For example, consider the simple model of dynamic surface tension from (45), γeff=f(β,U)∼γ0+βU(7) where γ0 is the equilibrium surface tension and β is inversely related to the rate of surfactant sorption at the surface—this is a measure of the deficiency of surfactant sorption to maintain a constant surface tension when an interface is not static. In that analysis, it was hypothesized that a large β could cause a differential surface tension between asymmetrically reopening daughter channels that helped to equilibrate reopening velocities.To experimentally estimate this influence in our system, we used a LTWP system to evaluate the dynamic surface tension for DPBS and SDS. For Infasurf we measured the dynamic surface tension with a pulsating bubble surfactometer, since Infasurf produces low surface tension under high compression that can cause errors with a LTWP (28). Specifically, we measured Δγ = γmax − γmin as the maximum differential in the dynamic surface tension in hysteresis loop. For LTWP, we used a fixed barrier compression velocity, UCompress = 1 mm/s, which is an order of magnitude estimate of the average surface expansion rate during the bubble penetration. For the pulsating bubble surfactometer our data are from an oscillation frequency f = 10 cpm, which provides an interface expansion rate of UCompress ~ 1 mm/s. βest=ΔγUCompress(8) which is tabulated in Table 2.Figure 5 and Figure 6 reveal the strong correlation between the increase of the dynamic surface tension parameter β and a reduction in the reopening asymmetry. For example, SDS has very low dynamic effects (β = 20 dyn·s/cm2 and Δγ/γ0 = 0.12) because it consists of small and highly mobile molecules that can adsorb rapidly to the developing interface. In contrast, Infasurf at C = 0.1 mg/ml (0.1Infa) has a much larger dynamic effect (β = 115 dyn·s/cm2) with the greatest fractional change in surface tension (Δγ/γ0 = 0.74). This occurs because of the low molecular mobility of the surfactant components that results in reduced diffusion and a slow adsorption rate that cannot adsorb rapidly enough to maintain a high surface concentration. In general, it is clear that a large dynamic surface tension is important for stabilizing the reopening velocities in daughter branches—this response may be of physiological importance for reducing regional ventilation heterogeneity. The quantity and distribution of surfactant molecules that exist along the bubble tip interface are important factors to understand the physicochemical effect. Since dynamic surface tension results from the low mobility of pulmonary surfactant molecules and its relationship to velocity, a similar surfactant distribution will exist before the bifurcation point as it does afterward. Thus, the existence of the surfactant distribution should impact the initial stage of the bubble progress in the daughter airways, because it will set up the initial balance of γ1 and γ2 at inlets through setting the initial distribution of surfactant entering Ch.1 and Ch.2. Figure 7 provides a schematic of this concept. Here, the streamlines and location of a converging stagnation point illustrate the focusing of surfactant on the interface, which could impact the surfactant transport in the neighborhood of the bifurcation. Figure 7A provides experimentally obtained liquid phase streamlines from progressing infinite-long air bubble tip, with the converging stagnation point (black dot) accumulating surfactant at the center of the interface, as estimated from fluorescent microparticle imaging (47). Fig. 7.A: schematic of experimentally obtained liquid-phase flow pattern near progressing air bubble tip in straight cylindrical channel (47). B and C: likely surface-phase surfactant transport during the interface migration near the bifurcation in an Inlet-Consistent bifurcation (B) and an Inlet-Inverted bifurcation (C). When a progressing bubble interface with a large-β surfactant solution is cleaved by the bifurcation carina, the surfactant accumulation at the converging stagnation point will be preferentially directed to the wider entrance, (see Fig. 7B for the Inlet-Consistent case). This causes the ratio ΔPCap1/ΔPCap2 to be reduced at the entrance in comparison to that of smaller-β surfactants because γ1 < γ2 and R1 > R2. Therefore Ch.1 reopens preferentially at the entrance, but the dynamic surfactant behavior modulates the reopening rate once the flow is developed. In contrast, with the Inlet-Inverted case (see Fig. 7C), the converging stagnation point enters Ch.2 because its entrance region is wider, and this funnels the surfactant preferentially to the narrow channel. As the interface migrates into Ch.2, the narrowing entrance concentrates surfactant on the interface, which further reduces the surface tension in the narrower channel. The converse is also true—a reduced quantity of surfactant is convected to Ch.1, and the diverging width causes a relative increase in surface tension in the entrance region. To test this hypothesis, we developed Inlet-Inverted models (see Fig. 1A), in which the narrower daughter channel has an expanded entrance width equal to that of the wider channel’s width, and vice versa. So, for example, as shown in Table 1, the channel (155/145)II has a width ratio of W1/W2 = 155/145 but an inlet ratio of I1/I2 = 145/155. Experiments using this Inlet-Inverted geometry were designed to obtain indirect evidence of the existence of the surfactant distribution and its effect on reopening symmetry. The local effect of the entrance behavior on the Inlet-Inverted model (c) is shown in Fig. 4. With DPBS (β = 0 dyn·s/cm2 and Δγ/γ0 = 0.12) purely geometric issues are at play. Ch.2 initially reopens preferentially because the wider entrance results in ΔPCap1 > ΔPCap2; however, after only 400 μm the Ch.1 bubble position takes a lead as W1>W2 dominates the relationship. In Fig. 4, the impact that the inverted inlet poses on the bubble progress is denoted by the reconvergence time (RT). With dynamic surface tension [Fig. 4; c, 0.1 mg/ml Infasurf], ΔPCap is reduced at the Ch.2 entrance by both geometric and physicochemical effects. Therefore, Ch.2 initially reopens preferentially and continues to reopen faster than Ch.1 far into the daughter channel despite the fact that the channel width ratio becomes W1>W2. The increase of the RT indicates that the physicochemical interaction magnifies the difference between ΔPCap1 and ΔPCap2 at the entrance. Since all geometric features and flow rate balance before the bubble bifurcation are identical for two experiments in Fig. 4 (c), we conclude that the increase of the RT is caused by increased γ1-to-γ2 ratio at the entrance followed by the interfacial compression on Ch.2 (concentrating surfactant at the interface in a manner similar to that of a Langmuir trough) and expansion on Ch.1, which leads to surfactant enrichment in Ch. 2 and depletion in Ch. 1 until surfactant adsorption/desorption compensates. Figure 8 presents the velocity ratio for Inlet-Inverted cases; in comparison to the data in Fig. 5 for Inlet-Consistent cases, it can be seen that far less reopening asymmetry exists (and in some cases it is reversed). Figure 9 elucidates the entrance effects by presenting RT vs. Δγ/γ0 for (152/148)II experiments. In these experiments the difference in hydraulic diameter at the entrance is <3%; however, the RT increases by nearly threefold with an increase in dynamic surface as shown with 0.1Infa (C = 0.1 mg/ml Infasurf). The correlation of the increase of RT with Δγ/γ0 indicates that the surface tension differential between Ch.1 and Ch.2 within the entrance region increases with the dynamic surface properties, which, as described above, is associated with an increased amount of accumulated surfactant preferential directed into Ch.2. This provides experimental evidence of the existence of a nonuniform surfactant distribution along the interface front and the importance of dynamic surface properties of pulmonary surfactant, which was predicted in our previous work (45). It is also important to note that microscale entrance effects, such as small asymmetric placement bifurcation carina, can significantly affect global behavior of multichannel reopening because of the physicochemical interaction of pulmonary surfactant. Fig. 8.Data from Width-Asymmetric/Inlet-Inverted (II) experiments. *Data from Yamaguchi et al. (45). Fig. 9.Relationship between the dynamic surface tension property (Δγ/γ0) and the reconvergence time in the Inlet-Inverted (152/148)II apparatus. From this analysis, it is important to note that surfactant with a large β exists because of slow sorption properties that allow the creation of high-concentration regions due to convective focusing near converging stagnation points. However, these low sorption rates could lead to a high global surface tension, and this has been shown to lead to epithelial cell damage. Therefore, it is important to stabilize reopening uniformity while also reducing damage to the delicate tissue. For example, although the value of β is large for C = 0.01 mg/ml Infasurf, this concentration is not protective during steady-state reopening. In contrast, C = 1.0 mg/ml Infasurf has a higher sorption rate due to an increased concentration and has nearly the same value of β (see Table 2) but is highly protective (4, 25) because of a lower mean surface tension. Pulsatility may affect this optimization by inducing changes in the local concentration field. A paired t-test was conducted for statistical comparison across groups for the velocity and position ratios shown in Figs. 5 and 6. This analysis indicates that the results comparing surfactant (SDS or Infasurf) with surfactant-free (DPBS) solutions are statistically significant with P ≪ 0.001 for all channels except (160/140)IC. In that single case C = 0.1 mg/ml Infasurf provided results significantly different from DPBS (P ≪ 0.001), while other concentrations of Infasurf showed less significance (P < 0.1) and SDS showed an insignificant difference. The lack of significance for SDS occurred because the yield pressure was not achieved, and the reduction of significance was a result of the lower values of β, which led to only slow motion in the narrow channel. Inlet-Inverted and Length-Asymmetric studies always demonstrated highly significant differences between surfactant and DPBS. This analysis shows that, in general, the dynamic surface tension properties afforded by Infasurf are important for inducing uniform airway recruitment in asymmetric bifurcations. This experimental in vitro study is intended to elucidate interfacial interactions that may influence the uniform recruitment of obstructed airways in the lung. However, this study has limitations that may reduce its direct applicability. Idealized microfluidic bifurcating channels were developed to investigate the importance of reopening geometry on the uniformity of airway recruitment in bifurcating systems, addressing fundamental principles associated with multiphase flows and the relationship between the capillary and viscous pressure drops. The use of bifurcation models with a rectangular cross section fulfilled this goal. While these structures permitted the careful examination of physical and physicochemical interactions, they did not investigate characteristics of more complex three-dimensional bifurcating geometries found in the lung. For example, we did not investigate the variation of the bifurcation angle that occurs during respiration or its orientation with respect to the gravitational vector. In our studies we investigated the motion of long fluid-filled sections—it is likely that shorter plugs of liquid will obstruct the lung. This would introduce segments of trapped gas and multiple interfaces that could alter the system behavior. The structural compliance was also not investigated; clearly this would modify the geometry and interfacial instabilities based upon the local pressure (41). In reality, the terminal bronchioles and acini impose elastic resistance (PEnd) that may act to balance the resistances and lead to symmetric reopening depending on the magnitudes of PHyd and PCap. In the present study, we purposely omitted the elastic resistance from the system to evaluate the influence of the branch-dependent ΔPCap accurately. These idealizations allow us to tease apart the physicochemical contributions that impact reopening at the microscale. Also, the formalizations reduce the relevance to a true lung, where asymmetries may be time dependent and less subtle than those investigated in this study. Nevertheless, we explored the physical principles that can be exploited to develop a clinical approach. Further studies could investigate the sensitivity of our findings by introducing “noise” to the same structure or by introducing nonuniform forcing. Additionally, the dynamic surface tension parameters β and Δγ/γ0 were estimated from Langmuir trough and pulsating bubble experiments. While this analysis provides insight into the dynamic surface tension properties, the relationship between the interfacial and liquid phases in the surface tension experiments do not precisely model the transport processes in our bifurcation studies, and therefore the physicochemical models cannot entirely mimic the dynamics of the complex system. Despite this limitation, the parameters β and Δγ/γ0 provide insight into the importance of dynamic surface tension in the uniformity of airway recruitment. We have investigated the importance of dynamic surface tension in the uniformity of recruitment of idealized asymmetric pulmonary bifurcations. To complete this study, we isolated the effects of capillary and viscous contributions to the reopening of airways branching from the bifurcation. This demonstrated that the surface tension difference between branches can substantially enhance recruitment uniformity in an asymmetric bifurcation, and that these surface-tension effects dominate viscous interactions. This reopening normalization occurs from two processes: 1) a local change in the surfactant distribution at the entrance of the bifurcation that disproportionally cleaves surfactant and 2) a velocity-dependent dynamic surface tension that occurs because of sorption rate limitations. We identified these processes by examining the sensitivity of recruitment uniformity to the placement of the carina and by variation of the downstream airway geometry. Through these studies we demonstrated that microscale effects in the region of the bifurcation have important implications that can sustain uniform reopening far into the daughter airways. Obviously, one cannot modify the precise configuration of an airway bifurcation geometry in real life. However, this study demonstrates that the surfactant concentration distribution can contribute substantially to uniform reopening. Our prior theoretical and experimental studies (15, 35) demonstrate that pulsatile flow may improve the distribution of surfactant through the use of a short retraction phase. The present study suggests that the local redistribution of surfactant may improve reopening uniformity while maintaining sufficient surfactant to decrease ΔPCap effectively, since flow reversal would cause the accumulating stagnation point to sweep from the tip toward the thin film (and back) during an oscillation phase. Currently we cannot set the phase a priori because of the geometric complexity of the lung; therefore variability, such as high-frequency and variable-frequency ventilation, may play an important role in providing conditions that could lead to homogeneous airway reopening. Further studies will investigate whether this flow reversal could prevent the preferential distribution of surfactant molecules and reduce the degree of sensitivity of the carina placement. GRANTSNational Science Foundation Grant CBET-1033619 supported this study. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSE.Y. and L.P.N. performed experiments; E.Y. and L.P.N. analyzed data; E.Y. interpreted results of experiments; E.Y. prepared figures; E.Y. and D.P.G. drafted manuscript; E.Y. and D.P.G. edited and revised manuscript; E.Y. and D.P.G. approved final version of manuscript. Page 14significant exertional dyspnea is a major determinant of quality of life and mortality in clinical populations (3, 6, 11). Patients with restrictive lung diseases such as interstitial lung diseases and chest wall disorders commonly experience debilitating exertional dyspnea (31). Previous research has applied chest wall strapping in healthy individuals to restrict lung inflation. This simulated the breathing pattern and exertional dyspnea observed with restrictive lung disease (15, 18, 26, 30, 32). Although this is an imperfect model of restrictive lung disease, chest wall strapping allows testing that is not feasible in clinical populations, with the potential to improve understanding of the neurophysiological mechanisms of dyspnea associated with restrictive lung disease. Dyspnea is a complex phenomenon. However, there is consensus that dyspnea depends on the relationship between ventilatory drive and the mechanical output of the respiratory system (34, 39). Multiple studies in healthy individuals have compared dyspnea when chemoreflex ventilatory stimuli are substituted for part of the exercise ventilatory response (1, 16, 20, 21, 32). These studies compared dyspnea at the same ventilation (isoventilation) and observed similar levels of dyspnea at equivalent levels of ventilatory drive and mechanical output of the respiratory system (16, 32). The similarity of dyspnea at isoventilation supported the notion that exertional dyspnea depends on the degree of reflex ventilatory stimulation and is independent of whether this reflex ventilatory stimulation is due to chemoreflex stimulation or exercise (1, 32). There is less evidence to support this notion in restrictive lung disease (31) or with chest wall strapping (32). O’Donnell et al. (32) applied chest wall strapping during incremental exercise with and without additional dead space as a reflex ventilatory stimulus. These authors concluded that dyspnea at isoventilation did not significantly change, although dyspnea was compared at different time points within the incremental exercise protocol. It is known that dyspnea may increase over time when ventilation is stable during constant exercise (9, 33). Therefore, exercise duration may have confounded the comparison of dyspnea at isoventilation in this study. We reasoned that the use of constant, instead of incremental, exercise would allow comparison at isoventilation without the confounding effect of exercise duration. This study therefore aimed to determine whether added dead space modifies dyspnea at isoventilation, during constant exercise with and without chest wall strapping. We hypothesized that in either condition the intensity of dyspnea would not be altered when dead space provided a component of the exercise ventilatory stimulus. MATERIALS AND METHODSA convenience sample of 11 healthy men aged 28 ± 2 yr (range 18–44 yr) with height 178.6 ± 2.2 cm, weight 76.4 ± 3.3 kg, and body mass index 23.9 ± 0.7 were recruited from university staff and students. As in other studies of chest wall strapping, we excluded women due to potential for breast discomfort. Exclusion criteria included any condition that might influence safety, exercise capacity, or dyspnea, including smoking or obesity. This study was approved by the Griffith University Research Ethics Committee and conformed to the Declaration of Helsinki. All participants provided written informed consent. Participants were required to attend the laboratory, at the same time of day, on four occasions. The first occasion determined maximal aerobic capacity and gas exchange threshold. The second visit determined the exercise workloads that, with the application of 0.6-l dead space, elicited comparable minute ventilation (isoventilation) to that achieved while exercising at 90% of gas exchange threshold without the application of dead space. The remaining two visits were designed to compare exercise at 90% of gas exchange threshold, with isoventilation produced by breathing through added dead space while exercising at a lower constant level of work, for both unrestricted breathing and chest wall strapping (Fig. 1). Fig. 1.Experimental design for visits 3 and 4. Visits 3 and 4 were in random order, and the 2 exercise conditions within each visit were also randomized. Ventilation was only matched within visit. Spirometry was undertaken according to American Thoracic Society/European Respiratory Society guidelines, and participants were familiarized with inspiratory capacity maneuvers (28). An incremental cycle ergometer (Lode Excalibur Sport V2.0; Groningen, The Netherlands) cardiopulmonary exercise test was undertaken with 30-s work rate increments of 12.5 or 15 W, aiming for an exercise duration of 10–12 min (2). Gas exchange parameters for this and subsequent visits were measured breath by breath, using fast-response O2 and CO2 analyzers (MedGraphics CPX/D; MGC Diagnostics, St. Paul, MN) and a bidirectional pressure differential pneumotachograph (preVent; MGC Diagnostics). Gas exchange threshold was independently determined by two observers using the V-slope [i.e., point of change in the slope of the CO2 consumption (V̇co2) vs. oxygen consumption (V̇o2) relationship] and ventilatory equivalents methods (2). With disagreement, a third observer assessed gas exchange threshold, and consensus was reached. Four cycle exercise tests were undertaken to determine the work rate that would produce isoventilation while breathing through dead space. Two exercise tests were conducted with unrestricted breathing and two with chest wall strapping, with 20-min recovery between each. One test involved cycling for 10 min, at the work rate equal to 90% of gas exchange threshold. The other test involved two or three contiguous 5-min exercise bouts at different constant work rates below 90% of gas exchange threshold while breathing through dead space. Linear interpolation was applied to the steady-state ventilation (average ventilation after 4 min at constant work) plotted against the two work rates with dead space that most closely approximated ventilation at 90% of gas exchange threshold. This predicted a constant work rate that would produce isoventilation. This process was repeated for the two exercise tests with chest wall strapping. The order of these visits was randomized. One visit required exercise with unrestricted breathing, and the other involved chest wall strapping. At each visit, two cycle tests were undertaken in random order using a constant work rate protocol (2): 1) at 90% of gas exchange threshold while breathing room air; and 2) at a lower intensity while breathing through 0.6-l added dead space. The lower exercise intensity was designed to achieve steady-state isoventilation within that visit. Four exercise conditions were therefore examined: 1) control (CTRL): unrestricted breathing, at 90% of gas exchange threshold; 2) CTRL+dead space (DS): unrestricted breathing with added dead space, at a work rate that elicited isoventilation to CTRL; 3) CWS: chest wall strapping that restricted lung inflation, at 90% of gas exchange threshold; and 4) CWS+DS: chest wall strapping that restricted lung inflation, while breathing with added dead space, at a work rate that elicited isoventilation to CWS. If chest wall strapping was applied (see below), this was gradually and progressively adjusted until the target forced vital capacity (FVC) was achieved without undue discomfort. Spirometry was performed before each test. Each individual then rested for 3 min before commencing the predetermined work rate for 10 min. Three inspiratory capacity maneuvers (28) were performed during the 3-min rest period, and one inspiratory capacity was repeated every 2 min during exercise. Continuous pulse oximetry was recorded from a Masimo SET Rad-9 oximeter (Irvine, CA) every 30 s. Heart rate was continually monitored with a CM5 electrode configuration (Lohmeier M607; Munich, Germany) and recorded every 30 s. Blood pressure was measured with a manual sphygmomanometer after 7 min of cycling. Individuals completed postexercise spirometry within 2 min of completing exercise. At the completion of exercise, participants indicated the intensity of leg fatigue, described as the “degree of effort of the leg muscles during pedalling” using a numeric rating scale (0–10) with 0 indicating no leg fatigue and 10 indicating the most severe leg fatigue they had ever experienced. Individuals rested for a minimum of 20 min before undertaking the second test, with metabolic recovery confirmed by heart rate. Participants were blinded to the dead space intervention, and block randomization ensured a balanced order of allocation to the chest wall strapping and dead space interventions. Dyspnea was described to participants as “an uncomfortable breathing sensation.” During visits 1 and 2, participants were asked to consider previous experiences of dyspnea intensity and to relate this to the numeric rating score scale, such that their use of this scale would be consistent in further visits. There was reinforcement of the need to attend separately to the sensations of dyspnea, leg fatigue, and the overall perceived effort of exercise. Dyspnea intensity was continuously rated by participants during all exercise tests (8, 23) using a potentiometer to indicate numeric rating score (0–10, with 0.5 increments; Ref. 29), with 0 indicating no dyspnea and 10 indicating the most severe dyspnea previously experienced. This was recorded every 30 s. Dyspnea unpleasantness and the dominant quality of dyspnea were reported for the last 3 min of the 10-min constant work test using the Multidimensional Dyspnea Profile version 2 (4). This included recall of an overall rating of intensity of dyspnea for the phrase “I feel short of breath.” A custom-made chest wall strapping device was used, with three inelastic circumferential counterlevered Velcro straps, similar to previous studies (27, 32). Pilot testing showed that excessive discomfort caused by chest wall strapping limited the reduction in FVC to ≤20%. As a result, an abdominal support device (modified lumbosacral corset with circumferential straps) similar to that used by Miller et al. (27) was added to prevent ventral abdominal displacement, which allowed participants to tolerate greater reduction in FVC comfortably. Baseline spirometry was measured before chest wall strapping was applied. Sequential incremental tightening of the straps (while exhaling to residual volume) continued until FVC had fallen by a target of 30% (accepted range 20–40%) below the baseline. No participant reported undue discomfort, likely related to regular feedback during tightening of the inelastic straps. The chest wall strapping device was applied with the upper strap at the level of the axillae and the lower strap adjacent to the costal margin. When chest wall strapping was applied in visit 3 or 4, the straps were adjusted until the FVC was within 0.2 l of the FVC achieved in visit 2. For exercise without dead space, participants breathed through a nonrebreathing valve (Hans Rudolph, Kansas City, MO) attached distal to the pneumotachograph and mouthpiece. The nonrebreathing valve was attached to two limbs of 0.6-l tubing that were open to ambient air, with identical low resistance (41, 43). Each limb was made of 35-mm internal diameter tubing. Blinded rebreathing through dead space was achieved by removing the valve completely from one limb of the nonrebreathing valve and occluding the other valve (41, 43). Participants could not see the function of the nonrebreathing valve or the modification of the valve system between tests. If the mean ventilation for the period of 4–10 min during the test at 90% of gas exchange threshold was not within 10% of the mean ventilation with dead space, then data from that visit were excluded. This mismatch of ventilation occurred in 8/22 possible occasions. At the end of that visit, a 2nd test with dead space was performed at a different constant work rate for 10 min, which allowed a new prediction of the work rate required to produce isoventilation. An additional visit with the same condition (CTRL or CWS) was then performed using the new predicted work rate with dead space and the same random order of tests. This additional visit produced isoventilation in every case and was analyzed. All values are expressed as means ± SE except where indicated. Analysis was performed with IBM SPSS version 19 (SPSS, Chicago, IL) using repeated-measures analysis of variance for visits 3 and 4. For exercise, conditions (CTRL, CTRL+DS, CWS, and CWS+DS) were compared for the period of 4- to 10-min (30-s epochs) constant exercise. For spirometry, conditions were compared pre- and postexercise. Three post hoc pairwise comparisons (CTRL vs. CTRL+DS, CWS vs. CWS+DS, and CWS vs. CTRL) were performed if there was a significant overall main effect. Fisher exact test was used to compare the quality of dyspnea between conditions. Comparisons were considered significant if P < 0.05. RESULTSEleven men with peak V̇o2 43.7 ± 2.4 ml·kg−1.min−1 consented to participate and completed the study protocol. CWS reduced FVC by 30.4 ± 2.2% of baseline (P < 0.001; Table 1). Forced expiratory volume in 1 s (FEV1; P < 0.001) and FVC (P = 0.03) increased after exercise, and there was a greater increase with CWS compared with CTRL (FEV1: P = 0.019; FVC: P = 0.034). There was an increase in FVC with addition of DS in both CTRL (P < 0.001) and CWS (P = 0.025) conditions (Table 1). FEV1 did not change when DS was added to either CTRL or CWS.
Minute ventilation was not altered by CWS but was increased when DS was added to either CTRL or CWS (P < 0.001 for each; Table 2). CWS reduced inspiratory capacity by 0.47 ± 0.14 l compared with CTRL (P = 0.007). Inspiratory capacity did not change when DS was added to either CTRL or CWS.
There was no significant change in V̇o2 or V̇co2 over time for the period of 4- to 10-min constant exercise. Across all conditions, ventilation increased by 1.78 ± 0.76 l/min between 4- and 10-min exercise (P = 0.041; Fig. 2 and Table 3). Fig. 2.Ventilation and dyspnea plotted against duration of exercise for unrestricted breathing (A) and chest wall strapping (B). *Ventilation was greater with CTRL+DS compared with CTRL (P < 0.05). **Dyspnea intensity was less with CWS+DS compared with CWS (P < 0.01).
Each individual achieved isoventilation for both CTRL and CWS conditions. This required a mean of 4.7 visits (range 4–6). Visit 3 or 4 was repeated when required to achieve isoventilation, with analysis restricted to the isoventilation visit. To elicit isoventilation, work rate was reduced by 33.5 ± 3.5% with CTRL+DS and 57.2 ± 5.9% with CWS+DS, with a greater reduction in work rate with CWS+DS compared with CTRL+DS (P < 0.01). Steady-state ventilation was higher with CTRL+DS compared with CTRL (Δ1.3 ± 0.55 l/min; P = 0.038) and unchanged with CWS+DS compared with CWS (Δ−0.7 ± 0.8 l/min; P = 0.45). Ventilation was 15% higher with CWS compared with CTRL (Δ7.2 ± 1.5 l/min; P = 0.001). Dyspnea was lower with CWS+DS compared with CWS (3.40 ± 0.52 vs. 4.51 ± 0.53; Δ−1.11 ± 0.28; P = 0.003; Fig. 2). Dyspnea was unchanged with CTRL+DS compared with CTRL (1.93 ± 0.49 vs. 2.17 ± 0.43; Δ−0.24 ± 0.20; P = 0.244). Dyspnea intensity increased with CWS compared with CTRL (Δ2.34 ± 0.30; P < 0.001). Across all four conditions, dyspnea increased by 1.00 ± 0.30 between 4- and 10-min constant exercise (P = 0.007). Dyspnea unpleasantness was greater with CWS than CTRL (Δ2.50 ± 0.27; P < 0.001; Fig. 3). There was a trend for unpleasantness to reduce with CWS+DS compared with CWS (Δ−0.68 ± 0.35; P = 0.082). Unpleasantness was unchanged with CTRL+DS compared with CTRL (Δ−0.05 ± 0.20; P = 0.82). The ratio between unpleasantness of dyspnea and intensity of dyspnea (using the intensity reported for the phrase “I feel short of breath” in the Multidimensional Dyspnea Profile) was 0.67 ± 0.24 for CTRL (n = 7 as intensity was 0 in 4 participants), 0.82 ± 0.28 for CTRL+DS (n = 7 as intensity was 0 in 4 participants), 0.93 ± 0.14 (n = 11) with CWS, and 0.92 ± 0.12 (n = 11) with CWS+DS (P = 0.56 across all conditions; n = 7). Fig. 3.Dyspnea unpleasantness during exercise. ‡Dyspnea unpleasantness was greater with CWS compared with CTRL (P ≤ 0.001). There was a trend to reduced unpleasantness with CWS+DS compared with CWS (P = 0.08). The phrase selected best to describe dyspnea showed a trend to altered overall distribution when comparing CWS with CTRL (P = 0.087; Table 4), with no effect of adding DS to CTRL or CWS (P = 0.68). There was also no significant change in the frequency that each individual descriptor phrase was selected when comparing CWS with CTRL, CTRL+DS with CTRL, or CWS+DS with CWS. However, there was a trend (P = 0.090) to increased frequency of selecting “My chest and lungs feel tight or constricted” when comparing CWS with CTRL. This descriptor phrase was also selected more frequently with chest wall strapping (CWS and CWS+DS) compared with unrestricted breathing (CTRL and CTRL+DS; P = 0.001).
For unrestricted breathing, the descriptor phrases corresponding to effort/work of breathing [“I am breathing a lot (breathing rapidly, deeply or heavily)” or “My breathing requires muscle work or effort”] were selected in 14/22 tests, “My breathing requires mental effort or concentration” in 6/22 tests, and no phrase could be selected in 2/22 tests. For chest wall strapping, participants selected “My chest and lungs feel tight or constricted” as the dominant quality of dyspnea in 10/22 tests. Phrases corresponding to effort/work of breathing (“I am breathing a lot” or “My breathing requires muscle work or effort”) were the other most frequent dominant quality, selected in 9/22 tests. “I am not getting enough air, I feel hunger for air, or, I am smothering” was selected in 2/22 tests, and “My breathing requires mental effort or concentration” in 1/22 tests. The intensity of dyspnea was greater with CWS compared with CTRL for each of six descriptor phrases for dyspnea (P < 0.01 for each; Table 5). The intensity of dyspnea also tended to be less with CWS+DS compared with CWS for each descriptor phrase, although only the global phrase for dyspnea intensity (“I feel short of breath”) was significantly reduced (P = 0.022). The intensity of dyspnea was similar between CTRL and CTRL+DS for each quality of dyspnea. However, with unrestricted breathing, the intensity was frequently 0 for the phrase “My chest and lungs feel tight or constricted,” which prevented valid comparison.
CWS was accompanied by an increase in perceived leg fatigue compared with CTRL at the same exercise intensity (Δ1.14 ± 0.29; P = 0.003; Fig. 4). Perceived leg fatigue was reduced with CTRL+DS compared with CTRL (Δ−0.5 ± 0.22; P = 0.049) and with CWS+DS compared with CWS (Δ−2.14 ± 0.61; P = 0.006). Fig. 4.Leg fatigue during exercise. *Leg fatigue was reduced with CTRL+DS compared with CTRL (P < 0.05). **Leg fatigue was reduced with CWS+DS compared with CWS (P < 0.01). ‡Leg fatigue was increased with CWS compared with CTRL at the same constant work rate (P < 0.01). CWS was associated with increased respiratory rate and reduced tidal volume (P < 0.001 for each; Table 3). Tidal volume increased (Δ143 ± 42 ml; P = 0.006) and respiratory rate decreased (Δ−0.5 ± 0.2 breaths/min; P = 0.017) with CTRL+DS compared with CTRL. Tidal volume and respiratory rate were unchanged with CWS+DS compared with CWS (Δ32 ± 29 ml, P = 0.30; −1.2 ± 0.8 breaths/min, P = 0.15; respectively). Inspiratory capacity was reduced with CWS compared with CTRL (P < 0.001). Inspiratory capacity did not significantly change when DS was added to either CTRL or CWS. Expiratory reserve volume was reduced (P < 0.001) with CWS compared with CTRL. Expiratory reserve volume was not altered when DS was added to either CTRL or CWS. There was no significant change in inspiratory reserve volume across any of the four conditions. Pulse oximetry O2 saturation (SpO2) did not significantly change across the four conditions (Table 3). End-tidal Pco2 was higher with CWS+DS compared with CWS (Δ3.6 ± 0.7 mmHg; P = 0.001) and higher with CTRL+DS compared with CTRL (Δ1.2 ± 0.3 mmHg; P = 0.002). End-tidal Pco2 was lower with CWS compared with CTRL (Δ−2.0 ± 0.6 mmHg; P = 0.009). Compared with CTRL, CWS did not significantly change V̇o2 but was associated with increased V̇co2, respiratory exchange ratio, systolic/diastolic blood pressure, and heart rate (Table 3). V̇o2 and V̇co2 were unable to be measured with DS, as the breath-by-breath system had algorithms that assumed room air was inspired at the mouthpiece. DISCUSSIONThe primary finding of this study was that when chest wall strapping restricted lung inflation, exercise alone was a more potent stimulus to dyspnea than the combination of additional dead space and lower intensity exercise. In comparison, when breathing was unrestricted, dead space did not modify dyspnea intensity at isoventilation. A secondary finding was that chest wall strapping was accompanied by increased perceived leg fatigue, compared with unrestricted breathing at the same exercise work rate, an observation that could be related to the increased unpleasantness caused by restricted breathing during exercise. Chest wall strapping was also associated with indirect, albeit weak, evidence that sympathetic activity (increased heart rate and blood pressure) and anaerobic metabolism (increased respiratory exchange ratio) may have been increased. During constant exercise with unrestricted breathing, dyspnea intensity and unpleasantness at isoventilation were not altered by dead space. There was (on average) a 33% reduction in constant work rate with dead space to achieve isoventilation. A study that employed an incremental exercise protocol also observed no significant change in dyspnea intensity with dead space (32), consistent with our observations. With unrestricted breathing, dead space caused a small change to a slower and deeper breathing pattern at isoventilation, which has previously been demonstrated (19, 25, 32, 41). In studies of exercise with dead space, tidal breathing was on the steep part of the respiratory compliance curve at a level of ventilation that is comparable with our study (16, 32). It can be assumed that the small increase in tidal volume with dead space at isoventilation, without a significant change in inspiratory capacity or inspiratory reserve volume, would not have disturbed the normal relationship between neural drive and the mechanical output of the respiratory system (16). We infer that in healthy individuals, dyspnea intensity does not depend on whether reflex ventilatory stimulation is from exercise alone or a combination of exercise and dead space. This is consistent with other studies in healthy individuals that suggest that dyspnea is related to reflex stimulation of the brain stem and its motor output to the respiratory muscles (1, 20, 21). With chest wall strapping, our participants reported average dyspnea of 4.5 numeric rating score units, indicating moderate dyspnea. However, when isoventilation was achieved with dead space and lower exercise intensity, there was a large relative decrease in dyspnea. With chest wall strapping, ventilation, breathing pattern, inspiratory capacity, and SpO2 were similar with dead space, and there was only a small increase in end-tidal CO2. The reduction in dyspnea was therefore unexpected. The exercise study by O’Donnell et al. (32) also examined chest wall strapping with or without dead space, using an incremental protocol where the work rate was increased every 3 min. These authors reported that dead space did not significantly change dyspnea at isoventilation (32). The degree of restriction of lung inflation with chest strapping was comparable (FVC reduced by 35% of baseline) with the current study. However, comparisons were made at different durations of exercise. A type 2 error might have explained the lack of any significant change in dyspnea. Chest wall strapping was reapplied in a separate visit with dead space, and the authors did not report whether variability in FVC between visits could have confounded the effect of dead space on dyspnea (32). In our study, there was little difference in dyspnea between CWS and CWS+DS after 3 min of constant exercise, with a greater difference observed after a longer duration of constant exercise. The 3-min work rate increment used in the study by O’Donnell et al. (32) may have explained their observation of comparable dyspnea with CWS+DS. A mean 57% reduction in exercise intensity was required to produce isoventilation with chest wall strapping. There was a greater reduction in exercise intensity with CWS+DS compared with CTRL+DS, as a fixed volume of dead space had a greater ventilatory stimulant effect when chest wall strapping caused a decrease in tidal volume. The reduced leg muscle work appears to have resulted in a reduction in dyspnea with dead space. This experiment cannot definitively determine how reduced leg muscle work caused a reduction in dyspnea, but there are several potential explanations that are each at least indirectly associated with reduced leg muscle work. First, the reduction in work rate that produced less leg muscle work may also have caused less respiratory muscle fatigue. A study of seven healthy men with chest wall strapping demonstrated diaphragm fatigue after 10-min constant exercise, at an exercise intensity that was similar to the CWS condition in our study (42). Diaphragm fatigue was potentially due to competition for O2 delivery between “overworked” respiratory (27) and leg muscles. It is known that diaphragm fatigue activates the neural metaboreflex, which leads to marked sympathetic activation, increased heart rate and blood pressure, and reduced limb blood flow (37, 38). Increased perception of leg fatigue occurred in healthy subjects when the work of breathing was experimentally increased, leading to competition for blood flow between limb and respiratory muscles (13). Competition between limb and respiratory muscles has also been observed during exercise in patients with respiratory disease (5, 40). In this context, we found evidence of increased sympathetic activity (increased blood pressure and heart rate), increased anaerobic metabolism (increased V̇co2 and respiratory exchange ratio), and increased perception of leg fatigue. This provided circumstantial evidence that significant competition for O2 may have occurred with the CWS condition, which may have caused respiratory muscle fatigue (7). As the CWS+DS condition was associated with reduced leg muscle work, there could have been less competition for O2 delivery. A consequent reduction in respiratory muscle fatigue could have improved the relationship between neural drive and the mechanical output of the respiratory system, reducing dyspnea. It must be emphasized that our study design cannot address the relative contributions of leg muscle and respiratory muscle work to dyspnea. However, reduced respiratory muscle fatigue, associated with less competition for O2 delivery, is one plausible explanation for the reduction in dyspnea with reduced leg muscle work. Another possible explanation is that reduced leg muscle fatigue may have directly reduced dyspnea. Peripheral afferent information from fatigued leg muscles may increase dyspnea (12, 36). Limb metaboreceptors stimulated by fatigue are believed to increase dyspnea more than ventilation (36, 39). There was a large reduction in perceived leg muscle fatigue in the CWS+DS condition, which may indicate a reduction in peripheral afferent information from fatigued leg muscles. This may explain the reduction in dyspnea at isoventilation, but it is also possible that this is explained by a global reduction in symptoms related to a reduced general sense of discomfort (36). Alternatively, impeded venous return and cardiac output in the CWS condition (27) may have contributed to dyspnea via increased vascular congestion, which is known to increase ventilatory drive via large vessel afferent firing (14). Thus the lower dyspnea with CWS+DS may also have resulted from the lowered cardiac output secondary to reduced workload, which reduced vascular congestion. A further potential explanation is that different central neural pathways may be involved in the stimulation of ventilation and dyspnea with dead space compared with exercise. The neural pathways mediating dyspnea with dead space are unknown (35). Previous studies in normal individuals (1) and patients with COPD (10) have shown that voluntary hyperventilation is associated with less intense dyspnea than exercise-induced ventilation, suggesting that a different neural pathway mediates dyspnea during voluntary hyperventilation. However, this seems a less plausible explanation for our findings, since dyspnea was not reduced with dead space when breathing was unrestricted. As previously shown, chest wall strapping significantly increased exertional dyspnea (32). Chest wall strapping was associated with typical pathophysiological findings of restrictive lung disease with exercise, including increased ventilation, reduced inspiratory capacity, a more rapid, shallow breathing pattern (24, 31), and reduced resting expiratory reserve volume (32). During exercise, tidal volume increased above resting levels by a reduction in inspiratory reserve volume, without any change in expiratory reserve volume from the reduced resting levels. It is known that reduced activation of slowly adapting pulmonary stretch receptors may increase dyspnea (22), and reduced activation of pulmonary stretch receptors associated with the reduced end-expiratory lung volume and tidal volume may have contributed to the increase in dyspnea with chest wall strapping. Chest wall strapping did not change inspiratory reserve volume during exercise, indicating that chest wall strapping caused a similar absolute reduction in tidal volume and inspiratory capacity. A previous study also found that chest wall strapping increased exertional dyspnea without changing the inspiratory reserve volume (18). During incremental maximal exercise with chest wall strapping, reduced inspiratory reserve volume predicted increased dyspnea (32). These authors suggested that reduced inspiratory reserve volume may indicate ventilatory constraint in the setting of increased ventilatory drive during maximal exercise, which contributes to exertional dyspnea (32). These contrasting findings might be explained by the different methods employed, with our study examining constant work exercise at a ventilation that did not approach ventilatory limitation (17). With chest wall strapping, chest tightness was the phrase selected best to describe exertional dyspnea in 46% of tests. Chest tightness has previously been observed to be frequent with chest strapping (32), although those authors did not ask participants to choose the best descriptor and reported that other phrases were also frequently selected. In contrast, patients with interstitial lung disease selected chest tightness less frequently than most other descriptors (31). It can therefore be suspected that chest wall strapping poorly simulates the qualitative experience of dyspnea in interstitial lung disease. It has been speculated that chest wall strapping may better simulate the quality of dyspnea in restrictive disorders characterized by reduced chest wall compliance (32). We observed that each quality of dyspnea increased with CWS compared with CTRL. This suggests that increases in all qualities are responsible for the increase in overall dyspnea intensity and unpleasantness observed with chest wall strapping. A limitation of the current study is that we did not measure lung mechanics or functional residual capacity and hence are uncertain about the extent to which our experimental conditions modeled restrictive lung disease. Furthermore, the acute simulation of restrictive lung disease via chest wall strapping is not able to mimic respiratory muscle and peripheral muscle changes expected with a chronic restrictive lung disease. Our experimental model may have limited applicability to some clinical populations. However, our observations suggest that chest wall strapping may better simulate the qualitative aspects of dyspnea experienced by patients with chest wall disorders. It may therefore be a more useful model to understand physiological determinants of dyspnea in chest wall disorders that are not chronic, such as in patients with pleural effusion. The experimental design required rigorous matching of ventilation. However, this introduced potential for bias as some participants required a greater number of visits to achieve this matching. In addition, there was variability of the degree of reduction in exercise load to produce isoventilation. This was inherent to the study design, which employed a standard dead space volume identical to that used by previous investigators (32). We did not measure respiratory muscle fatigue or leg blood flow to determine whether there was reduced respiratory muscle fatigue and/or less competition for O2 delivery in the CWS+DS compared with the CWS condition. This study aimed to compare dyspnea at isoventilation with a similar level of restricted lung inflation. FVC was mildly increased with dead space, which raises the possibility that the degree of restriction was not identical. We cannot identify a plausible explanation why dead space would increase FVC but FEV1 and inspiratory capacity were not altered by dead space, suggesting a similar degree of restricted lung inflation at isoventilation. In conclusion, in healthy individuals, dyspnea was unchanged when dead space substituted for part of a constant exercise stimulus to ventilation. This confirms that dyspnea does not depend on the mode of reflex ventilatory stimulation in health. However, when chest wall strapping was imposed to simulate restrictive lung disease, dead space combined with lower intensity exercise presented a less potent stimulus to dyspnea. Therefore, dyspnea associated with chest wall strapping appears to reflect the degree to which ventilation is stimulated by leg muscle work. When dead space is substituted for part of the exercise ventilatory stimulus, a constant exercise protocol may lead to different conclusions about the effect on dyspnea. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSL.A.G. and L.A. conception and design of research; L.A.G., R.L., N.R.M., T.J.C., and L.A. performed experiments; L.A.G. analyzed data; L.A.G., I.B.S., N.R.M., and L.A. interpreted results of experiments; L.A.G. and N.R.M. prepared figures; L.A.G. drafted manuscript; L.A.G., R.L., I.B.S., N.R.M., T.J.C., and L.A. edited and revised manuscript; L.A.G., R.L., I.B.S., N.R.M., T.J.C., and L.A. approved final version of manuscript. Page 15the intestine is a large organ and a major determinant of whole body energy homeostasis through its control over nutrient absorption and release of gut hormones during digestion (6). Evidence demonstrating the potential role of the intestine in the pathogenesis of obesity and insulin resistance is rapidly increasing. In type 2 diabetes, there is a continuous deterioration of intestinal endocrine function (16), and alterations in the intestinal microbiota content have been shown to be associated with the development of insulin resistance in humans and animals (8, 9, 26). Splanchnic glucose uptake (GU) accounts for up to 60% of total glucose metabolism after an oral glucose load. In insulin resistance, splanchnic GU is impaired and plays a role in the pathogenesis of hyperglycemia in type 2 diabetes (10, 27). Our laboratory has previously shown that tissue-specific intestinal GU from circulation into enterocytes is impaired in the insulin-stimulated state, i.e., intestinal insulin resistance exists, in obese and type 2 diabetic subjects (29). The role of intestinal insulin resistance in the pathology of type 2 diabetes is unclear; however, it has been suggested that intestinal insulin resistance leads to abnormalities in the signaling mechanism responsible for the GLUT2-mediated GU in the small intestine, particularly in the jejunum, leading to increased transepithelial or lumen to blood glucose exchange, causing hyperglycemia (3). Regular exercise training enhances skeletal muscle insulin sensitivity (11, 20, 23, 35) in working muscles. Exercise training also enhances the regulation and utilization of lipids in the skeletal muscle (13, 19, 22, 42). The training-induced adaptations in muscle substrate metabolism and oxidative capacity lead to improvements in the whole body metabolism and insulin sensitivity. Although muscle is widely studied, previous data about the effects of exercise on abdominal organs concern mainly the liver and pancreas, and data are limited about the effects of exercise on intestine (28, 33, 36). Thus it is not known whether exercise training could enhance intestinal substrate metabolism, and whether any possible changes would be reflected in the insulin sensitivity of the whole body. Our laboratory has previously shown that 2 wk of low-volume, high-intensity interval training (HIIT) and moderate-intensity continuous training (MICT) increase both aerobic capacity and whole body and main working skeletal muscle insulin-stimulated GU in sedentary, middle-aged men (7). In the present study, using the intestine data from this same clinical trial (NCT01344928), our aim was to quantify the effects of exercise on tissue-specific, insulin-stimulated glucose and fasting free fatty acid uptake (FFAU) from circulation into the intestine (duodenum, jejunum, and colon) using positron emission tomography (PET) and radiotracers 2-[18F]fluoro-2-deoxy-d-glucose (FDG) and 14(R,S)-[18F]fluoro-6-thia-heptadecanoic acid (FTHA) before and after HIIT and MICT. We hypothesized that the higher training volume instead of the intensity would strain the intestinal metabolism more and thus lead to the increased intestinal insulin-stimulated GU and decreased FFAU after MICT compared with HIIT. Additionally, to explore possible mechanisms behind the changes in intestinal GU and FFAU, we also studied healthy Wistar rats, which underwent corresponding HIIT and MICT interventions, and analyzed the intestinal protein expression of GLUT2 and CD36. We hypothesized that training would increase the expression of GLUT2 and CD36 in enterocytes more after MICT than HIIT. MATERIALS AND METHODSTwenty-eight middle-aged, sedentary individuals were recruited and randomized into two groups; one with 2 wk of HIIT and the other with 2 wk of MICT. The subjects were nonobese [aged 40–55 yr, peak O2 uptake (V̇o2peak) < 40 ml·kg−1·min−1] and had no previous experience of active exercise training. The inclusion and exclusion criteria of the recruitment process have been described in detail previously (24). Two of the subjects withdrew during the intervention, one from the HIIT group due to exercise-induced hip pain and one from MICT group due to personal reasons. This left 13 subjects in each group. The purpose, nature, and potential risks involved in participating in the study were explained in detail, and informed consent was obtained before any measurements were performed. The study was approved (NCT01344928) by the local ethical committee of the Hospital District of South-Western Finland (decision 95/180/2010 §228) and carried out in compliance with the Declaration of Helsinki. Initial screening included a physical examination, an oral glucose tolerance test (OGTT), and a V̇o2peak test to assess the participant’s health, glycemic status, and aerobic capacity. The participants then underwent two PET imaging sessions on 2 different days. On the first day, [18F]FTHA and PET were used to measure, under a fasting state, the free fatty acid uptake in different intestinal regions (duodenum, jejunum, and colon) and the quadriceps femoris (QF) and deltoid muscles [the muscle results were taken from our laboratory's previous publication (7)]. On the second day, [18F]FDG and PET were used to measure the insulin-stimulated GU in the intestine and the muscles during hyperinsulinemia. Once again the muscle results used were from our laboratory's previous publication (7). An overnight fast of at least 10 h was required before the OGTT and PET measurements. Participants were also asked to abstain from any caffeinated and alcoholic beverages, and to avoid strenuous exercise 48 h before these studies. After the 2-wk exercise training intervention, all measurements were repeated, starting with [18F]FTHA PET 48 h after the last exercise session and continuing with a [18F]FDG PET after 72 h, and finally an OGTT and V̇o2peak test were done after 96 h (Fig. 1). Fig. 1.Study design. Subjects were studied on 3 separate days before and after the exercise intervention. OGTT, oral glucose tolerance test; PET, positron emission tomography; FTHA, 14(R,S)-[18F]fluoro-6-thia-heptadecanoic acid ([18F]FTHA); PET-FDG, [18F]fluoro-2-deoxy-d-glucose ([18F]FDG). Participants were randomized into HIIT and MICT exercise groups, and both training groups had six supervised training sessions within 2 wk. Each HIIT session consisted of 4–6 × 30 s exercise bouts of all-out cycling efforts (Wingate protocol, load 7.5% of the whole body weight, Monark Ergomedic 828E, Monark, Vansbro, Sweden) with 4 min of recovery in between the exercise bouts (5). All of the participants were familiarized with the HIIT training protocol (2 × 30 s bouts) before they were randomized into training groups. MICT training consisted of 40–60 min of cycling at a moderate intensity (60% of V̇o2peak intensity). In both groups, the training was progressive, and in the HIIT group the number of cycling bouts increased from four to five and finally to six, and in the MICT group the training time increased from 40 to 50 min and then to 60 min in every second training session. Participants underwent four PET sessions: one [18F]FTHA PET and one [18F]FDG PET before and after the training intervention. Antecubital veins of both arms were cannulated for the PET studies. One catheter was used to inject the radiotracers [18F]FTHA and [18F]FDG, whereas the other one was for blood sampling. To arterialize the venous blood samples, the arm was heated using an electronically powered cushion. On the first PET scan session, intestinal free fatty acid uptake was measured using [18F]FTHA PET in a fasting state. [18F]FTHA radiotracer [155 (SE 0.4) MBq] was injected, and dynamic imaging of the abdominal region (frames 3 × 300 s) was acquired, starting, on average, at 46 min after the tracer injection. This was followed by a femoral region scanning (QF) (frames 3 × 300 s), starting ~65 min after the tracer injection. Finally, the shoulder region (deltoid) (frames 3 × 300 s) was scanned, starting ~90 min after the tracer injection. On the second day, intestinal GU was measured using [18F]FDG under euglycemic hyperinsulinemic clamp. On average 87 (SE 1) minutes after the start of the clamp, [18F]FDG [156 (SE 0.5) MBq] was injected, and similar time frames were acquired, as described earlier for [18F]FTHA scans, starting at 49, 70, and 90 min after the tracer injection. Arterialized blood samples were obtained at regular intervals during both the [18F]FTHA and [18F]FDG scans to measure the plasma radioactivity to calculate the tracer input function. An automatic gamma-counter (Wizard 1480, Wallac, Turku, Finland) was used to measure the plasma radioactivity. A GE Discovery TM ST system (General Electric Medical Systems, Milwaukee, WI) was used to acquire the PET/computerized tomography (CT) images. CT images were acquired for anatomical references. The imaging data obtained from the PET scanner were corrected for dead time, decay, and photon attenuation, and the images were reconstructed using the 3D-OSEM method. Carimas 2.7 (http://turkupetcentre.fi/) was used to manually draw the regional tubular three-dimensional regions of interest (ROIs) on sections of the descending duodenum, the jejunum, and the transverse colon, using CT images as anatomical reference. The tubular ROIs were carefully drawn to outline the intestinal wall, while avoiding the intestinal contents and external metabolically active tissues (17). From these regional (duodenum, jejunum, and colon) the ROI time activity curves were extracted. The rate constant (Ki) for the uptake of the radiotracer ([18F]FTHA, [18F]FDG) into the cells was calculated using tissue time activity curves obtained from the duodenum, jejunum, and colon, and a tracer input function using a fractional uptake rate method, as previously described (17). Regional glucose and free fatty acid uptakes were calculated by multiplying region-specific Ki by the corresponding plasma glucose or free fatty acid concentration, respectively. For GU, the products were further divided by a lumped constant of 1.15 (17), and a recovery coefficient of 2.5 (17) was applied for the colonic GU to take into account the partial volume effect (4, 25). For the duodenal and jejunal GU, no recovery coefficient was needed. The ROIs for the deltoid and QF muscles were drawn as explained previously (7). As previously described (24), the V̇o2peak was determined by performing an incremental bicycle ergometer test (Ergoline 800s, VIASYS Healthcare, USA) with direct respiratory measurements using a ventilation and gas exchange (Jaeger Oxycon Pro, VIASYS Healthcare, Germany) at the Paavo Nurmi Centre (Turku, Finland). Initial exercise intensity was 50 W, and after every 2 min the exercise intensity was increased by 30 W until volitional exhaustion. V̇o2peak was expressed as the highest 1-min mean oxygen consumption. The workload at the last 2 min of the test was averaged and used as a measure for maximal performance. The peak respiratory exchange ratio was ≥1.15 and peak blood lactate concentration, measured from capillary samples obtained immediately and 1 min after exhaustion (YSI 2300 Stat Plus, YSI Life Sciences, USA), was ≥8.0 mmol/l for all the tests. A peak heart rate (RS800CX, Polar Electro, Kempele, Finland) within 10 beats of the age-appropriate reference value (220 – age) was true in all except one participant in both groups and in both pre- and posttraining tests. Therefore, the highest value of oxygen consumption was expressed as V̇o2peak and not maximal O2 uptake (V̇o2max). The euglycemic hyperinsulinemic clamp technique was used as previously described (7). Insulin was infused at a rate of 1 mU·kg−1·min−1 (Actrapid; Novo Nordisk, Copenhagen, Denmark), and blood samples were taken every 5–10 min to adjust the exogenous glucose infusion and to maintain the plasma glucose concentration as closely as possible to the level of 5 mmol/l. Euglycemic hyperinsulinemic clamp was performed after the subjects had fasted at least for 10 h. Insulin (Actrapid, 100 U/ml, Novo Nordisk, Bagsvaerd, Denmark) infusion was started with the rate of 40 mU·min−1·m−2 during the first 4 min. After 4 min and up to 7 min, infusion rate was reduced to 20 mU·min−1·m−2, and, after 7 min to the end of the clamp, it was kept constant at 10 mU·min−1·m−2. Glucose infusion was started 4 min after the start of the insulin infusion with a rate of subject’s weight (kg)·0.1−1·g−1·h−1. At 10 min, glucose infusion was doubled, and after that it was adjusted further according to plasma glucose levels to maintain the steady-state level of 5 mmol/l. Arterialized venous blood samples were collected before the clamp and every 5–10 min to measure the plasma glucose concentration for adjusting the glucose infusion rate. Arterialized plasma glucose was determined in duplicate by the glucose oxidase method (Analox GM9 Analyzer; Analox Instruments, London, UK). Whole body insulin-stimulated GU rate (M-value) was calculated from the measured glucose values collected when the subjects had reached the steady state during the PET scan that was started 87 min (SE 1) after the start of the clamp. FDG-PET study was performed when the subject had reached the stable glucose concentrations at the level of 5 mmol/l (within 5% range for at least 15 min) after positioning into the PET scanner. Adipose tissue depot masses were measured with MRI. MRI scans were performed using Philips Gyroscan Intera 1.5 T CV Nova Dual scanner (Philips Medical Systems). Abdominal area axial T1 weighted dual fast field echo images (echo time 2.3 and 4.7 ms, repetition time 120 ms, slice thickness 10 mm without gap) were obtained. To measure different adipose tissue masses, the images were analyzed using SliceOmatic software version 4.3 (http://www.tomovision.com/products/sliceomatic.html). To obtain the mass, the pixel surface area was multiplied with the slice thickness and the density of adipose tissue, 0.9196 kg/l (1). A 2-h, 75-g OGTT was conducted after the subjects had fasted for 12 h. Blood samples were collected at 0, 15, 30, 60, 90, and 120 min after the glucose ingestion to determine the glucose and insulin levels. Measurements of oxidized low-density lipoprotein (LDL) and oxidized high-density lipoprotein (HDL) were based on spectrophotometric analyses of oxidized lipids in lipoproteins isolated by precipitation methods (2). Whole body fat percentage was measured at the Paavo Nurmi Centre using a bioimpedance monitor (InBody 720, Mega Electronics, Kuopio, Finland). Twenty-four male Wistar rats were randomly divided into three groups: HIIT (n = 8), MICT (n = 8), and control (CON) (n = 8). At the central animal laboratory of the University of Turku, the animals (aged between 8 and 12 wk) were housed under standard conditions (temperature 21°C, humidity 55 ± 5%, lights on from 6:00 AM to 6:00 PM) with free access to food and tap water. Before the exercise intervention, rodents’ body weight, body fat mass, and lean tissue mass were measured using EchoMRI-700 (Echo Medical Systems LLC, Houston, TX), OGTT and V̇o2max test were performed, and free-living energy consumption measured. Animals in the HIIT and MICT groups had 10 exercise sessions within 2 wk. Each HIIT exercise session was composed of 8–10 × approximately 30 s swimming bouts with 1-min resting period after each bout. Animals in the HIIT group had extra weights of 30–50 g tied to the waist to force them to make all-out efforts. Animals in the MICT group started with 40-min swimming exercise, and thereafter the exercise duration was increased by 10 min every second session until 80 min was reached in the last two sessions. In the MICT group, the rats did not bear any additional weights. One day after the last training session, OGTT was performed, which followed V̇o2max tests on the second and third day after the last exercise session. Thereafter, the animals were kept in the metabolic cages for 2 days. Animals were killed 5 days after from the last exercise session, and intestinal samples from duodenum were collected for protein expression analyses. All animal procedures were approved by the National Animal Experimental Board (ESAVI/5053/04.10.03/2011) and were performed in accordance with the guidelines of the European Community Council Directives 86/609/EEC. The frozen duodenal tissue pieces were homogenized on ice in a lysis buffer (150 mM NaCl, 1% NP-40, 0,5% sodium-deoxycholate, 0,1% SDS, 50 mM Tris·HCl, pH 8.0), supplemented with a protease inhibitor cocktail with an Ultra-Turrax T25 (Ika-Werke). The protein concentration was then quantified with the Thermo Scientific Pierce BCA protein assay kit (Thermo Fisher Scientific) before the sample denaturation with SDS loading buffer containing β-mercaptoethanol (Sigma-Aldrich) in +95°C for 5 min. Samples were run on a 10% SDS–polyacrylamide gel and, after electrophoresis, transferred onto a nitrocellulose membrane (Santa Cruz Biotechnology). An incubation with 5% (wt/vol) milk diluted in Tris-buffered saline-Tween 20 (0,02 M Tris-buffered saline, 0.1% Tween 20) was used to block the unspecific binding sites before the overnight incubation in +4°C with the following primary antibodies: GLUT2 (no. 07–1402, Millipore), CD36 (no. sc-9154, Santa Cruz Biotechnology), vascular endothelial growth factor 2 (VEGFR2) (no. NB-100-530, Novus Biologicals), and β-actin (no. sc-8432, Santa Cruz Biotechnology). The fluorescent signal from the secondary antibodies IRDye 800CW Donkey anti-Rabbit lgG (H+L) and IRDye 800CW Donkey anti-Mouse lgG (H+L) (LI-COR Biosciences) was detected by using the LI-COR Odyssey CLx Imager (LI-COR). The intensities were normalized to a reference band in each membrane, and the relative values were used for fold-change calculations. Body composition was measured using EchoMRI-700 (Echo Medical Systems, Houston, TX). Each animal was scanned before and after the exercise intervention, and body fat mass and lean tissue mass were measured. The aerobic capacity was studied by measuring the V̇o2max with rat single-lane treadmill (Panlab- Harvard Apparatus). Animals were familiarized to the rat single-lane treadmill (Panlab-Harvard Apparatus) for 3 days before the V̇o2max test. The test started after a warm-up period. During the test, the angle of the treadmill was 25°, and the speed was increased by 3 cm/s after every other minute until exhaustion. OGTT was performed after 6-h fast. Glucose (20% wt/vol, 1 ml /100 g) was administered orally, and tail vein glucose was measured at 0, 30, 60, 90, and 120 min with a Precision Xceed Glucose Monitoring Device (Abbott Diabetes Care, Abbot Park, IL). Whole body energy expenditure was measured with a metabolic cage (Oxylet system, Panlab, Harvard Apparatus, Spain) over 48 h. The energy expenditure was calculated according to the measured carbon dioxide (CO2) production and oxygen (O2) consumption and averaged over 24 h. Descriptive statistics shown in Tables 1 and 2 and Figs. 2–4 are based on model-based means [95% confidence intervals (CI)]. Association between the anthropometrics, glucose profile, and the lipid profile and the training groups, time points, and time × training interaction were performed with hierarchical linear mixed model, using the compound symmetry covariance structure for time. Transformations (logarithmic or square root) were done to (insulinfasting; HDL; colonic, QF, and deltoid GU; duodenal, jejunal, colonic, and QF free fatty acid uptake) to achieve the normal distribution assumption. All tests were performed as two-sided, with a significance level set at 0.05. Correlations were calculated using Pearson r. In the animal study, one-way analysis of variance was used. All of the analyses were performed using SAS System, version 9.3 for Windows (SAS Institute, Cary, NC). RESULTSThe effects of exercise on whole body fat percentage, aerobic capacity (V̇o2peak), and whole body insulin sensitivity (M-value) have been published in our laboratory's previous study (7). Total, LDL, and HDL cholesterol levels decreased significantly after training (Table 1). In the cholesterols, the only difference between the training modes was the greater decrease in LDL cholesterol in the HIIT group compared with the MICT group (P = 0.03, time × training).
Colonic insulin-stimulated GU improved in the MICT group (+37%), while no response was observed in the HIIT group (±0%) (P = 0.02 time × training) (Fig. 2). Jejunal GU tended to respond differently between the training modes, with only MICT increasing the uptake (HIIT − 4%, MICT + 13%, P = 0.08 time × training) (Fig. 2). Both exercise modes decreased the free fatty acid uptake in the duodenum (P = 0.001, time, Fig. 2), and MICT tended to also decrease the uptake in the colon (HIIT 0%, MICT −38%, P = 0.08 time × training, Fig. 2). The jejunal GU associated positively with aerobic capacity (V̇o2peak) [pretraining (Pre): r = 0.46, P = 0.03; posttraining (Post): r = 0.45, P = 0.03] and negatively with visceral fat mass (Pre: r = −0.42, P = 0.05; Post: r = −0.45, P = 0.03). GU both in the jejunum (Pre: r = −0.31, P = 0.15; Post: r = −0.50, P = 0.02) and duodenum (Pre: r = −0.12, P = 0.59; Post: r = −0.53, P = 0.02) associated negatively with HcA1c levels. In the MICT group, the GU in the colon associated positively (Pre: r = 0.17, P = 0.63; Post: 0.68, P = 0.03) (Fig. 3), and the duodenal free fatty acid uptake negatively (Pre: r = −0.38, P = 0.31; Post: r = −0.94, P = 0.01), with the whole body GU after the training. QF and deltoid muscle results in these subjects have been published elsewhere (7). For comparison purposes, those results have been added to Fig. 2. Fig. 2.Insulin-stimulated glucose uptake (top) and fasting free fatty acid uptake (bottom) in different tissues before and after 2 wk of either high-intensity interval training (HIIT; solid triangles) or moderate-intensity continuous training (MICT; shaded squares). The muscle [quadriceps femoris (QF) + deltoid] results have been adapted from Eskelinen et al. (7). All values are expressed as model-based means, and bars are confidence intervals (95% CI). P value for time interaction, the groups behaved similarly for the change in parameter with no differences between the training modes. P value for time × training interaction, the groups behaved differently for the change in parameter with significant difference between them. Fig. 3.Correlation between insulin-stimulated jejunal glucose uptake and aerobic capacity (V̇o2peak; top) and visceral fat mass (middle) in pooled analysis of moderate-intensity continuous training (MICT; shaded squares) and high-intensity interval training (HIIT; solid triangles) subjects. Bottom: correlation between insulin-stimulated colonic glucose uptake and whole body glucose uptake (M-value) in MICT (shaded squares) subjects. There was a significant increase in the body weight and fat free mass of all the animal groups, indicating age-related growth during the study intervention (Table 2). While the fat percentage increased in the CON group, it significantly decreased in both HIIT and MICT groups after the training. There were no differences in glucose values at time points 0 min and 120 min or in the glucose AUC in any of the group × s. The aerobic capacity (V̇o2max) tended to improve significantly in both HIIT and MICT groups compared with the CON group [Pre: HIIT: 70.07 (66.2, 74.0); MICT: 71.2 (67.3, 75.1); CON: 69.0 (65.1, 72.9) ml·min−1·kg−0.75; Post: HIIT: 72.9 (69.0, 76.8); MICT: 72.8 (68.9, 76.7); CON: 68.9 (65.0, 72.8) ml·min−1·kg−0.75 (95% CI), P = 0.05]. GLUT2 protein expression in the rat intestine was significantly higher in the HIIT and MICT groups compared with CON group [HIIT: 19,090 (12,930, 28,190); MICT: 11,606 (7,651, 17,604); CON: 4,141 (2,730, 2,141) arbitrary units (95% CI), P < 0.01]. Also, CD36 expression was higher in the HIIT and MICT groups compared with CON group [HIIT: 635 (366, 1,100); MICT: 696 (387, 558); CON: 79 (44, 63) arbitrary units (95% CI), P < 0.05]. Whereas VEGFR2 was only higher in the HIIT group compared with MICT and CON groups [HIIT: 704 (477, 976); MICT: 345 (193, 541); CON: 294 (147, 491) arbitrary units (95% CI), P < 0.05]. No significant differences were observed in GLUT2, CD36, or VEGFR2 expression between the HIIT and the MICT groups.
DISCUSSIONIn the present study, the effects of 2 wk of exercise training, HIIT and MICT, on intestinal substrate uptake from circulation were studied in healthy, untrained, middle-aged men. The data show that MICT increases insulin-stimulated GU, while both training modes decrease FFAU in the intestine, and that intestinal insulin-stimulated GU correlates positively with aerobic capacity and negatively with visceral fat and HbA1c. In addition both training modes increased GLUT2 and CD36 protein expressions in rat enterocytes. To our knowledge, this is the first study that provides evidence about the beneficial effects of exercise training on the intestinal substrate metabolism and an additional mechanism by which exercise improves whole body metabolism. The intestinal GU values during hyperinsulinemia in the present study agree with our laboratory's recent data in healthy lean controls and obese subjects (17, 29). Studies by Honka et al. (17) and Mäkinen et al. (29) show that insulin increases the intestinal GU compared with fasting state in healthy lean controls, but the increase is blunted in obese subjects. This means that the intestine is an insulin-sensitive organ and intestinal insulin resistance exists in obesity. Furthermore, it was shown that, in obese subjects, intestinal insulin resistance is ameliorated after rapid weight loss (17, 29). In enterocytes, glucose is transported from blood to lumen by GLUT2 transporter proteins (40). In obesity and intestinal insulin resistance, there is an impairment in the insulin-stimulated GLUT2 internalization in the enterocyte; which has been suggested to restrain the normal GU in the intestine (41). In the present study, the insulin-stimulated intestinal GU before the training intervention was at the same level as the healthy controls in our laboratory's previous study (29). Insulin-stimulated GU improved in the colon (+37%) and tended to improve in the jejunum in the MICT group after the training, while it remained essentially unchanged in the HIIT group. To study the mechanisms behind the exercise-induced improvements in intestinal GU in our human data, we performed corresponding short HIIT and MICT training interventions in healthy rats. As GLUT2 is responsible for the uptake of glucose from basolateral membrane in the intestine (21), we hypothesized that exercise would increase the expression of GLUT2 in enterocytes to enhance the intestinal GU, and that the increase would be higher in MICT compared with HIIT due to higher training volume. We found that both HIIT and MICT increased intestinal GLUT2 expression in rats with no differences between the groups. The reason why the increased GU was seen only after MICT in humans, whereas GLUT2 expression increased in both training groups in rats, is unclear. However, it might be that, although 2 wk of low-volume HIIT was enough to induce changes in protein level in rats, longer time is need to be able to detect a change in tissue level noninvasively in humans. The discrepancy in GU in different parts of the intestine agrees with the findings of Mäkinen and coworkers (29) and may be due to the differences in the location of GLUT2 receptor in the enterocytes (41). In humans, GLUT2 has been observed in the apical membrane of an enterocyte in the jejunum, but not in the duodenum (3). The discrepancy in substrate uptake in different parts of the intestine is possibly also related to the different digestive tasks between the small and large intestines and how exercise training strains these mechanisms. The results in this study demonstrate a decreased free fatty acid uptake in the duodenum after the training intervention in both training groups. The digestion and delivery of dietary fats throughout the body is mediated by the small intestine. In the small intestine, inside the enterocytes, the dietary fats are resynthesized into triacylglycerols (TAG) and secreted into the circulation or stored in cytoplasmic lipid droplets. Postprandially, the increased secretion of TAG from the small intestine leads to an increment in the circulating TAG levels; however, during a fast, the levels decrease as a result of clearance by peripheral tissues (30). Recently, Hung and coworkers (18) showed that, in rodents, endurance training leads to enhanced lipid turnover and more efficient fatty acid oxidation for energy utilization within the enterocytes. Our data regarding the higher CD36 expression, in both HIIT- and MICT-trained rats, is in agreement with the results of Hung et al. (18). Despite the higher CD36 expression, the reduced intestinal FFAU after training in the present study could be due to the more efficient fatty acid oxidation. This is because enhanced fatty acid oxidation means that less fatty acids are needed to produce the same amount of energy. Another possible mechanism for the decreased intestinal FFAU could be the reduced free fatty acid flux in the intestine. In fact, we found in the present study an almost significant (P = 0.052, Table 1) drop in the levels of circulating plasma free fatty acids after the training during the FTHA-PET study (fasting). The lower free fatty acid levels can be explained by decreased visceral fat mass and increased whole body insulin sensitivity posttraining, as both reduce the adipose tissue lipolysis and thereby circulating FFAs (Table 1) (31, 34, 38). At the moment, little is known about the different mechanisms of how exercise training could strain the intestinal metabolism, yet some data exist about exercise and the splanchnic bed. Splanchnic blood flow reduces during dynamic training and as a function of exercise intensity. However, it has been shown that the reduction in splanchnic blood flow during exercise attenuates as a response to long-term training (32, 33). The smaller reduction in splanchnic blood flow during exercise after regular training seems to be related to the enhanced vasodilation and reduced vasoconstriction of splanchnic and renal vasculature, which further could indicate improved nutrient supply and utilization during exercise in a trained state (33). In the present study, we did not measure intestinal blood flow in humans. In rodents, we found higher VEGFR2 (a marker of angiogenesis) expression level in enterocytes in HIIT compared with MICT and CON groups (Fig. 4). Thus angiogenesis could also be one factor explaining the attenuated reduction in the intestinal blood flow shown after exercise training (33). The difference in VEGFR2 levels between the groups in the present study might be due to higher transient reduction of flow into the splanchnic area during HIIT compared with MICT. HIIT is an extremely intense exercise mode, and, during the intervals, the body concentrates to supply blood mainly to the working muscles, which may induce the hypoxic condition in the splanchnic area and further stimulate intestinal angiogenesis. Other possible factors regulating intestinal metabolism could be peristaltic movements and colon transit time (37, 43). Fig. 4.Top: relative expression of CD36, GLUT2, and VEGFR2 in duodenum; n = 6–8. All values are expressed as model-based means, with error bars representing the confidence intervals (95% CI). *P value < 0.05. Bottom: Western blots of CD36 (75 kDa), GLUT2 (55 kDa), and VEGFR2 (105 kDa). Animals without a detectable band were excluded from the analysis. HIIT, high-intensity interval training; MICT, moderate-intensity continuous training; CON, control group. We used two different training modes in this study. These both included six training sessions within an intervention period of 2 wk. Both the time spent during the training (time HIIT 15 min vs. MICT 300 min) and the average calculated energy consumption during the training [403 and 2680 kcal, respectively (7)] were much less in HIIT than MICT. Despite this difference, both training modes improved whole body insulin sensitivity (M-value, HIIT 12% and MICT 7%) and aerobic capacity (V̇o2peak, HIIT 6% and MICT 3%) without significantly different responses between the training modes. In contrast to this, intestinal metabolism seems to be more sensitive to MICT than HIIT. As intestine mediates the delivery of nutrients throughout the body, it may be that the aerobic training mode and longer exercise time per session in MICT compared with HIIT challenges the intestinal metabolism more and thus may be a more rapid and effective way to improve intestinal metabolism. It is also possible that the difference in the daily habitual physical activity levels or in dietary intake affects the observed findings. In the present study, subjects were instructed not to perform any additional physical activity, except daily normal living, and they reported having done so. However, no pedometer or any other device was used to follow the activity. Thus we cannot completely rule out the possible effect of habitual physical activity on our results. Subjects were also instructed to maintain their normal dietary habits, and they kept dietary logs for 3 days before and during the exercise intervention. According to the dietary logs, there were no changes in the total caloric intake or in the caloric content before and after the intervention in either study group (data not shown). Most of the beneficial effects of exercise on the whole body are attributed to skeletal muscles, and thus it is interesting to compare these intestinal findings to our laboratory's previous findings concerning skeletal muscles in these same subjects (7). In skeletal muscles, both training modes increased insulin-stimulated GU in the main working muscle, the QF, whereas no changes were observed in deltoid and other upper body muscles (Fig. 2). In addition, no significant changes were observed in the FFAU in any of the studied muscles (7). Adding the findings from the present study to the overall picture, it is interesting to note that intestinal metabolism seems to respond more readily to MICT than the metabolism in the nonworking upper body muscles (Fig. 2). Previously, intestinal insulin-stimulated GU has been shown to be associated with whole body GU (M-value), in both healthy and obese subjects (17). Our data are in line with these previous findings, showing that whole body GU associates positively with insulin-stimulated GU in the colon and inversely with the duodenal free fatty acid uptake. Furthermore, the jejunal GU correlated positively with the V̇o2peak and negatively with visceral fat mass and HbA1c, which are both known risk markers for metabolic diseases. Thus, although exercise training induces major health benefits through the body’s muscular system, its effects on the intestine, with an average weight of 3–4 kg and surface of 200–300 m2, also warrants further research. There are some limitations in this study. The first is the location of the intestine. This is because, even though the duodenum has a relatively fixed location in the abdomen, the distal segments of the intestine move within the abdomen. This issue was addressed by confirming the drawn ROIs with a CT scan. Second, the results might have been affected by spillover and partial volume effects due to the transaxial resolution of the PET scanner and the thinness of the intestinal mucosal wall. However, this effect was demonstrated to be minimal in our laboratory's previous validation study (17). Third, in this study, we measured the substrate uptake from the circulation into the enterocytes without knowing the release from the enterocytes into the circulation (i.e., from lumen to circulation). Fourth, due to the radiation dose limits, we could not perform the [18F]FDG and [18F]FTHA PET scans both at fast and during euglycemic hyperinsulinemic clamp. Thus we studied the FFAU at fasting state and GU during euglycemic hyperinsulinemic clamp, in situations when the FFAU and GU, respectively, are at their highest. Finally, the exercise duration in this study was only 2 wk. Although this kind of intervention has been shown to be effective (7, 12, 14, 15, 44), it must be emphasized that the findings show only the early training response and, therefore, the long-term effects of these training modes on intestinal metabolism should be studied further in future experiments. In conclusion, this study shows that intestinal insulin sensitivity associates positively with aerobic capacity and inversely with the metabolic risk markers visceral adiposity and HbA1C. Two weeks of regular training (HIIT and MICT) were shown to already improve aerobic capacity and whole body insulin sensitivity and, specifically, MICT to induce positive changes in intestinal substrate metabolism in middle-aged, healthy men. The changes in intestinal substrate uptake seem to be related to improvements in GLUT2 and CD36 protein levels. It is likely that regular long-term training has pronounced effects on intestine and whole body metabolism, and thus the role of exercise training on intestinal substrate uptake in patient populations warrant further studies. GRANTSThis study was conducted within the Centre of Excellence in Cardiovascular and Metabolic Diseases and supported by the Academy of Finland, the University of Turku, Turku University Hospital, and Åbo Akademi University. The study was financially supported by European Foundation for the Study of Diabetes, the Emil Aaltonen Foundation, the Hospital District of Southwest Finland, the Orion Research Foundation, the Finnish Diabetes Foundation, the Ministry of Education of the State of Finland, the Academy of Finland (Grants 251399, 251572, 256470, 281440, and 283319), the Paavo Nurmi Foundation, the Novo Nordisk Foundation, and the Centre of Excellence funding. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSK.K.M., J.-J.E., J.T., T.I., M.Y.-K., K.A.V., R.P., E.L., and K.K.K. analyzed data; K.K.M., E.L., K.K.K., and J.C.H. interpreted results of experiments; K.K.M. prepared figures; K.K.M. drafted manuscript; K.K.M., P.N., K.K.K., and J.C.H. edited and revised manuscript; A.M.S., J.-J.E., J.T., T.I., M.Y.-K., K.A.V., J. Kapanen, T.J.G., M.H.-S., O.S., N.S., M.A., K.K.K., and J.C.H. performed experiments; J. Knuuti conceived and designed research; P.N., K.K.K., and J.C.H. approved final version of manuscript. REFERENCES
Page 16the classic physiological responses to hypoxia are well documented with increases in heart rate and respiratory rate among the primary acute responses that serve to maintain systemic oxygen delivery despite low arterial oxygen saturation (26). While important, some of these physiological adjustments may influence the thermoregulatory responses to heat stress and exercise. For example, hypoxia may reduce plasma volume (45), which could impair sweating and lead to a redistribution of blood flow (8). Studies performing exercise at simulated high altitude have reported lower steady-state local sweat rates and thermosensitivities in hypoxic conditions using altitude-specific %V̇o2peak (32, 33), but exercise at fixed absolute workloads (and presumably fixed rates of heat production) have resulted in no differences between normoxia and hypoxia in steady-state sweating or its relationship to esophageal temperature (45). On the other hand, cutaneous vasodilation has been reported under normothermic hypoxic conditions (56), and an increased vasodilatory response to local heating after ~9 h of hypoxia has been observed at rest (36). Yet, a lower cutaneous vascular conductance has been reported on the forearm, but not the chest, during whole body passive heating (38) and exercise at a fixed external workload at 30°C (45). Whether any potential differences in cutaneous vasodilatory control in hypoxia are sufficient to alter the changes in core temperature during exercise in the heat at a fixed heat production is, however, unknown. Additionally, the choice of exercise intensity for studies comparing thermoregulatory responses between participants is extremely important as confounding factors associated with differences in metabolic heat load and/or body size between individuals have the potential to either cause or mask differences in core temperature and sweating (11, 14, 27). It has been widely believed for many years that exercise intensity relative to an individual’s rate of maximum oxygen uptake (i.e., % V̇o2max) is a critical determinant of thermoregulatory responses (3, 20, 54). However, a recent series of studies by our research group has suggested that the importance of %V̇o2max may be greatly exaggerated (27). Nevertheless, our evidence to date has been exclusively derived from an independent group approach. While participant groups in these studies were carefully selected to isolate the independent influence of fitness and thus relative intensity [i.e., matched for body mass, body surface area (BSA), and sex but with different levels of V̇o2peak], individual responses in core temperature are notoriously variable (24). A clear limitation is therefore that variation in other factors (e.g., partial acclimation status, adiposity) that inevitably arises when comparing different groups of participants could have obscured a potential influence of relative intensity. Indeed, there remain many recent examples of studies using %V̇o2peak to compare thermoregulatory responses between groups including obese individuals (1), children (37), and heart failure patients (4). Resolving whether %V̇o2peak exerts an influence independently of metabolic heat production and evaporative heat balance requirement is therefore an urgent priority. One way to truly isolate the influence of relative intensity within an individual is to employ a hypoxic environment. By reducing the inspired fraction of oxygen (FIO2) to create a normobaric hypoxic environment, the V̇o2peak for a given individual will be lower (35, 41). Thus, %V̇o2peak can be manipulated for a given work rate (and therefore heat production) by altering FIO2. This approach also permits the comparison of exercise in normoxia vs. hypoxia at a matched %V̇o2peak, but with different rates of heat production. Given the traditional logic that core temperature (3, 54) and sweating (20) during exercise are altered, or even determined, by %V̇o2peak, a higher core temperature would be expected in hypoxia compared with normoxia at the same rate of heat production. However, the evidence for such a phenomenon is weak (2, 21, 45), and observations of thermoregulatory responses to exercise in hypoxia have been inconsistent. Given these inconsistencies in the literature and the various protocols used to study thermoregulatory control in hypoxia (i.e., fixed %V̇o2peak vs. fixed workload), the independent influence of hypoxia on thermoregulatory responses warrants determination. The aims of the present study were therefore to assess 1) the independent influence of hypoxia on core temperature, sweating, and skin blood flow during exercise at a fixed metabolic heat load; and 2) the within-subject influence of relative exercise intensity on the control of thermoregulatory responses independently of heat production and the evaporative requirement for heat balance (Ereq) using hypoxia to differentiate relative vs. absolute workloads. It was hypothesized that thermoregulatory responses would not differ between normoxia and hypoxia during exercise at fixed rates of heat production and Ereq, irrespective of large differences in %V̇o2peak. However, during exercise at a fixed %V̇o2peak, systematic differences in thermoregulatory responses were expected between normoxia and hypoxia secondary to different rates of heat production and Ereq to maintain a given %V̇o2peak. METHODSThe experimental protocol was approved by the Health Sciences and Science Research Ethics Board at the University of Ottawa (H09-14-12) and conformed to the Declaration of Helsinki. All participants in the study voluntarily provided written informed consent and completed a Physical Activity Readiness Questionnaire and American Heart Association/American College of Sports Medicine Health/Fitness Facility Pre-participation Screening Questionnaire before participation. Based on the mean effect sizes for changes in core temperature from previous studies (6, 27) the minimum required sample size to determine significant differences (α = 0.05, β = 0.8) in rectal temperature was seven [G*power v3.1.5 (16)]. Therefore, eight (1 woman) nonheat-acclimated, healthy, and active participants [1.75 (0.06) m, 70.2 (6.8) kg, 25 (4) yr, 54 (8) ml·kg−1·min−1] were recruited to participate in the study; at the time of data collection the female participant was using constant-release hormonal contraception (IUD). All participants reported no history of cardiovascular, respiratory, metabolic, or neurological disorders and were asked to refrain from consuming any alcohol or caffeine, as well as performing any strenuous activity, 24 h before testing. Participants were also asked to maintain similar habits, such as sleep and diet, the night before and the day of experimental sessions, which were all separated by at least 48 h. The study consisted of five laboratory visits: two preliminary trials and three experimental trials. During the first preliminary session, height and body mass were measured. Peak oxygen consumption (V̇o2peak) was measured in normoxia (FIO2 = 0.21) and again in normobaric (~743 mmHg) hypoxia (FIO2= 0.13) during the preliminary sessions in a counterbalanced order. The maximal tests to determine V̇o2peak were performed on a semirecumbent cycle ergometer (Lode; Corival, Groningen, Netherlands) and began at 100 W with increases of 20 W every minute thereafter until physical exhaustion, based on the recommendations from the Canadian Society for Exercise Physiology (13). The two V̇o2peak tests were performed on separate days and in a counterbalanced order between participants. All sessions were performed in an environmental chamber at the University of Ottawa. During hypoxia sessions, O2 extractors (CAT12; Altitude Control Technologies, Lafayette, CO) connected to the climate-controlled chamber (volume of ~64 m3) allowed for a stabilized FIO2 level. Prior to each trial, urine specific gravity was measured to ensure that each participant was below the cut-off value for euhydration (<1.025) (29). The first two trials were performed in a counterbalanced order and consisted of 45 min of cycling at a fixed metabolic heat production (Hprod) of ~7 W/kg in either normoxia (FIO2 = 0.21; NORM) or normobaric hypoxia (FIO2 = 0.13; HYP1). Exercise intensity was set to elicit a fixed Hprod between trials as described previously by Cramer and Jay (11). The trials were completed in a compensable environment, at 34.4 (0.2)°C, 46 (3)% RH, to compare steady-state thermoregulatory responses. During a third trial (FIO2 = 0.13; HYP2), 45 min of cycling was completed to compare steady-state thermoregulatory responses at an intensity corresponding to the same %V̇o2peak as in NORM but at a lower rate of metabolic heat production. Subjects were clothed in standardized athletic shorts (and sports bra for the woman) and sandals. Once all equipment (see Instrumentation below) was in place and functioning, the participant rested in a seated position for a 30 min baseline period inside the chamber. Upon completion of exercise, there was a ~45 min postexercise rest period during which the participant was deinstrumented and measurements of maximum skin blood flow were recorded under normoxic conditions. Oxygen consumption (V̇o2) and carbon dioxide production (V̇co2) were measured by indirect calorimetry (Vmax Encore; Carefusion, San Diego, CA). Metabolic energy expenditure (M) was subsequently estimated by Eq. 1. Hprod was calculated by subtracting the rate of mechanical work (W) from M (Eq. 2), as follows: M=V˙O2(RER−0.70.3)ec+(1.0−RER0.3)ef60(BSA)(1,000)[W/m2](1) Hprod=M–W(W/m2)(2) where RER is the respiratory exchange ratio (V̇co2:V̇o2), and ec and ef are the caloric equivalents per liter of oxygen for the oxidation of carbohydrates (21.13 kJ) and fats (19.62 kJ), respectively. Clothing insulation and evaporative resistance were considered negligible for all calculations given the minimal coverage of the ensemble. Heat losses via convection were determined by: C=hc(Tsk−Ta)[W/m2](3) hc=8.3v0.6[W·m−2·K−1](4) where hc refers to the convective heat transfer coefficient and can be used for a seated subject facing an air velocity of 0.2–4.0 m/s (47), Tsk is the mean temperature of the skin (°C), Ta is the temperature of the ambient air (°C), and v is the air velocity, which was estimated to be 1.2 m/s in the chamber.Radiative heat losses were calculated as follows: R=hr(Tsk−Ta)[W/m2](5) hr=4εσ·BSArBSA·[Tsk+Tr2+273.15]3[W·m−2·K−1](6) where hr refers to the radiative heat transfer coefficient (W·m−2·K−1) and Tr refers to the mean radiant temperature of the environment (°C), which is assumed to be equivalent to Ta, ε is the weighted area emissivity of the skin, set to 0.95 (ND), σ is the Stefan-Boltzmann constant (5.67 × 10−8 W·m−2·K−1), and BSAr/BSA is the effective radiative area of the body (ND), which can be estimated as 0.70 for seated subjects (31). Total respiratory heat losses can be calculated by adding convective (Cres; Eq. 7) and evaporative (Eres; Eq. 8) respiratory heat loss (7, 23): Cres=VEρCP(Te−Ti)60(BSA)[W/m2](7) Eres=VEρ(He−Hi)60(BSA)hv[W·m−2](8) where V̇e refers to the rate of ventilation (l/min), ρ refers to the density of the air (kg/m3), CP refers to the specific heat capacity of dry air (kJ/kg·K−1), Te refers to the temperature of the expired air (assumed to be 37°C), Ti refers to the temperature of the inspired air, which is equivalent to ambient air (°C), He refers to the humidity ratio of expired air (g/kg), Hi refers to the humidity ratio of inspired air (g/kg), and hv refers to the latent heat of vaporization of water (J/kg·K−1). The required amount of evaporative heat loss to maintain heat balance (Ereq) can be calculated as: Ereq=Hprod−(C+R+Cres+Eres)[W/m2](9) All heat balance parameters were calculated in W/m2 but are displayed in either W or W/kg throughout this article. Rectal temperature (Tre) was monitored with a pediatric thermistor (TM400; Covidien, Mansfield, MA) inserted ~20 cm past the anal sphincter. Esophageal temperature (Tes) was also measured using a pediatric thermistor inserted through the nasal passage and into the esophagus with the bottom of the probe resting at approximately the level of the right atrium (42). Skin temperature (Tsk) was measured at four sites (48) via thermistors integrated into heat flow sensors (2,252 Ohms; Concept Engineering, Old Saybrook, CT). The heat flow sensors were placed onto the skin with double-sided adhesive disks and surgical tape (Transpore; 3M, London, ON, Canada). All thermometry data were recorded on a National Instruments data acquisition unit (model NI cDAQ-9172) at a sampling rate 0.2 Hz. Data were simultaneously displayed and recorded in spreadsheet format on a personal computer (Dell Inspiron 545) with LabVIEW software (National Instruments). Local sweat rates (LSR) were measured from ventilated capsules placed on the skin of the upper back (lateral portion), forearm, and forehead. Influent anhydrous air flowed through the capsule at a rate of 1.00 l/min. Flow rates were measured with an Omega FMA-A2307 flow rate monitor (Omega Engineering, Stamford, CT). The vapor content of the effluent air was measured by capacitance hygrometers (series HMT333; Vaisala, Helsinki, Finland). LSR values were calculated from the recorded flow rate and the difference in vapor content of the influent and effluent air, normalized to the area of skin under the capsule (expressed in mg·cm−2·min−1). In addition to LSR, whole body sweat losses (WBSL) were estimated from the difference of pre- and postexercise body mass measurements (Combics 2, Sartorius, Mississauga, ON, Canada), corrected for metabolic and vapor mass losses from respiration (44). Heat-activated sweat gland density (HASGD) was determined using the iodine paper method (18) adjacent to the forearm sweat capsule. Sweat expulsions (i.e., the number of activated sweat glands) produced dark purple spots on the paper which were counted using Image J software (49). Forearm LSR was divided by HASGD to determine sweat gland output (SGO) expressed in µg·gland−1·min−1. Blood pressure was monitored with an automated unit (Tango M2; SunTech, Raleigh, NC) and a 3-lead ECG setup (Q-Stress v3.3; Quinton, Bothell, WA). Mean arterial pressure (MAP) was calculated from the addition of 1/3 systolic blood pressure and 2/3 diastolic blood pressure at rest, while exercising values were calculated by a heart rate corrected formula as described by Razminia et al. (50). As an index of skin blood flow (SkBF), red blood cell flux (LDF) was measured on the forearm and upper back using laser-Doppler flowmetry (Periflux System 5000; Perimed, Järfälla, Sweden). Cutaneous vascular conductance (CVC) was calculated as LDF (AU)/MAP (mmHg) and expressed as both arbitrary units (AU/mmHg) and as a percentage of maximum values (%CVCmax) determined during 45 min of normoxic postexercise local heating of the measurement area (~1 cm2) to 44°C. Oxygen saturation (SpO2) and heart rate (HR) were recorded every 5 s with a Rainbow SET pulse oximeter (Radical-7, Masimo, Irvine, CA). One blood sample of ~6 ml was taken at 0 and 45 min of exercise and was analyzed for hematocrit (Hct) and hemoglobin (Hb) using a photometer (HemoPoint H2 Meter; StanBio Laboratory, Boerne, TX) and microcuvettes (Alere, Orlando, FL). Each draw was replaced with an equivalent volume of 0.9% NaCl saline solution. Changes in plasma volume (ΔPV) were determined using the method described by Dill and Costill (15). Ratings of perceived exertion (RPE) were taken at rest and every 15 min during exercise using the Borg scale [6–20]. Two-way repeated-measures ANOVAs with the independent variables of condition (2 levels: normoxia or hypoxia) and time (four levels: 0, 15, 30, 45 min) were used to analyze the dependent variables of ΔTes, ΔTre, ΔTsk, LSRarm, LSRback, LSRhead, CVC, %CVCmax, HR, MAP, RPE, V̇o2, RER, V̇e, and SpO2. Separate ANOVAs were performed to compare fixed Hprod trials (i.e., NORM vs. HYP1) and trials matched for %V̇o2peak (i.e., NORM vs. HYP2). When a significant interaction between time and condition was detected, individual time points were compared by t-tests with a Holm-Bonferroni correction. Partial eta squared (ηp2) was reported, where values of 0.01, 0.09, and 0.25 correspond to small, medium, and large effect sizes, respectively (9). Maximum CVC values were compared by a one-way ANOVA. Mean values throughout the trials for % V̇o2max, Hprod, workload, mechanical efficiency, HASGD, SGO, respiratory heat losses, and changes in PV were compared by a paired-sample Student’s t-test. Cohen’s d values were reported for the t-tests where d values of 0.2, 0.5, 0.8 are indicative of small, medium, and large effect sizes, respectively (9). All data are reported as means (standard deviation). Alpha was set at the P = 0.05 level. All statistical analyses were performed with Prism GraphPad v6.0 for Windows (La Jolla, CA). RESULTSMean values for Hprod, Ereq, external workload, and %V̇o2peak are presented in Table 1. By design, there was no difference in heat production (P = 0.20, d = 0.49) or Ereq (P = 0.17, d = 0.73) in the fixed Hprod trials between NORM and HYP1 (Table 1), while workload was slightly lower in HYP1 than NORM (P = 0.03, d = 0.61) due to decrements in mechanical efficiency (NORM: 16.8 (1)%; HYP1: 15.7 (1)%; P < 0.01); however, %V̇o2peak was significantly greater in HYP1 compared with NORM (P < 0.001, d = 2.46) due to a 27% reduction in V̇o2peak from 3.78 (0.67) l/min in normoxia to 2.74 (0.36) l/min in hypoxia (P < 0.001, d = 1.92). Conversely, when the trials were matched for %V̇o2peak, heat production (P < 0.001), Ereq (P < 0.001), and external workload (P < 0.001) were all significantly lower in HYP2 compared with NORM (Table 1).
During the fixed Hprod trials, while significant differences in %V̇o2peak were observed, changes in Tes over time (Fig. 1A) were not different between conditions (P = 0.69, ηp2 = 0.07) with a ΔTes after 45 min of 0.64 (0.22)°C in NORM and 0.63 (0.29)°C in HYP1. There was also no interaction between time and condition on changes in Tre (P = 0.77, ηp2 = 0.05) with a ΔTre after 45 min of 0.76 (0.19)°C in NORM and 0.75 (0.24)°C in HYP1 (Fig. 1B). In contrast, when exercising at a matched %V̇o2peak (but different Hprod), changes in Tes over time (Fig. 1A) were significantly smaller in HYP2 compared with NORM (P < 0.01, ηp2 = 0.52), with a 45 min ΔTes of 0.42 (0.21)°C in HYP2, while changes in Tre over time (Fig. 1B) were also smaller in HYP2 compared with NORM (P < 0.01, ηp2 = 0.85), with a 45 min ΔTre of 0.56 (0.22)°C in HYP2. The change in Tsk over time was not different between fixed Hprod trials (P = 0.82, ηp2 = 0.04) with 45 min ΔTsk values of 0.63 (0.26) °C in NORM and 0.58 (0.18)°C in HYP1, while ΔTsk was significantly lower in HYP2 compared with NORM (P = 0.04, ηp2 = 0.32) with 45 min ΔTsk values of 0.50 (0.19)°C in HYP2. Fig. 1.Changes in esophageal (Tes, A) and rectal (Tre, B) temperatures as a function of time while cycling in normoxia (NORM, 21% FIO2), hypoxia at the same heat production as in NORM (HYP1, 13% FIO2), and hypoxia at the same %V̇o2peak as in NORM (HYP2, 13% FIO2). *Significant difference between NORM and HYP2 (P < 0.05). During the fixed Hprod trials (NORM and HYP1), again despite differences in %V̇o2peak, LSR on the forearm (Fig. 2, left) did not respond differently over time between conditions (P = 0.30, ηp2 = 0.18) with values after 45 min of 1.21 (0.18) mg·cm−2·min−1 in NORM and 1.28 (0.21) mg·cm−2·min−1 in HYP1. There was also no significant interaction between time and condition on LSR of the upper back (P = 0.93, ηp2 = 0.02) with values after 45 min of 1.06 (0.17) mg·cm−2·min−1 in NORM and 1.05 (0.31) mg·cm−2·min−1 in HYP1 (Fig. 2, middle), nor did an interaction exist between time and condition on LSR of the forehead (P = 0.91, ηp2 = 0.03) with LSR values after 45 min of 1.01 (0.36) mg·cm−2·min−1 in NORM and 1.03 (0.32 mg·cm−2·min−1 in HYP1 (Fig. 2, right). However, in the %V̇o2peak-matched trials LSR was attenuated in HYP2 compared with NORM on the forearm (P < 0.01, ηp2 = 0.74) with LSR after 45 min of 0.77 (0.20) mg·cm−2·min−1 (Fig. 2, left). LSR on the upper back (P < 0.01, ηp2 = 0.77) and the forehead LSR (P < 0.01, ηp2 = 0.65) were also lower in HYP2 relative to NORM with values after 45 min of 0.76 (0.18) mg·cm−2·min−1 (Fig. 2, middle) and 0.62 (0.29) mg·cm−2·min−1 (Fig. 2, right), respectively. Fig. 2.Local sweat rates (LSR) from the forearm (left), upper back (middle), and forehead (right) as a function of time while cycling in normoxia (NORM, 21% FIO2), hypoxia at the same heat production as in NORM (HYP1, 13% FIO2), and hypoxia at the same %V̇o2peak as in NORM (HYP2, 13% FIO2). *Significant difference between NORM and HYP2 (P < 0.05). Heat-activated sweat gland density and sweat gland output are displayed in Fig. 3. During fixed Hprod trials, HASGD was not different between conditions (P = 0.51, d = 0.21), with 93 (29) glands/cm2 in NORM and 101 (45) glands/cm2 in HYP1. When matched for %V̇o2peak HASGD was lower in HYP2 than NORM with 70 (26) glands/cm2 (P = 0.01, d = 0.85). Conversely, SGO was not different between conditions in fixed Hprod trials (P = 0.64, d = 0.12), with values of 13.2 (5.5) µg·gland−1·min−1 in NORM and 12.6 (5.3) µg·gland−1·min−1 in HYP1. Nor was SGO different when matched for %V̇o2peak (P = 0.48, d = 0.62) with a value of 12.0 (1.8) µg·gland−1·min−1. Fig. 3.Mean heat-activated sweat gland density (HASGD) and sweat gland output (SGO) while cycling in normoxia (NORM, 21% FIO2), hypoxia at the same heat production as in NORM (HYP1, 13% FIO2), and hypoxia at the same %V̇o2peak as in NORM (HYP2, 13% FIO2). *Significant difference between NORM and HYP2 (P < 0.05). CVC data are displayed in Fig. 4. In the fixed Hprod trials, there tended to be an interaction between time and condition on CVC in AU at the forearm (P = 0.06, ηp2 = 0.33), but not the back (P = 0.35, ηp2 = 0.14), between NORM and HYP1. There was also a trend toward a significant interaction between time and condition on %CVCmax at the forearm (P = 0.09, ηp2 = 0.30), suggesting %CVCmax increased in a greater fashion over time in HYP1 than in NORM. While there was no evidence of an interaction at the upper back, there tended to be a main effect of hypoxia with steady-state %CVCmax ~30% higher in HYP1 compared with NORM (P = 0.09, ηp2 = 0.60). At a fixed %V̇o2peak, there was no interaction between time and condition on CVC in AU at the arm (P = 0.64, ηp2 = 0.09) or back (P = 0.99, ηp2 < 0.01), nor was there an interaction on %CVCmax at either the forearm (P = 0.31, ηp2 = 0.18) or upper back (P = 0.45, ηp2 = 0.12) between NORM and HYP2. Mean maximum CVC values in AU (Fig. 4) from the arm and back following local heating were not different between all trials [P = 0.17, ηp2 = 0.18; NORM: 3.45 (1.04), HYP1: 2.64 (0.89), HYP2: 3.65 (0.73)]. Fig. 4.Cutaneous vascular conductance (CVC) from the arm and upper back expressed as arbitrary units (AU) and as a percentage of maximum values (%CVCmax) while cycling in NORM (normoxia; 21% O2), HYP1 (fixed Hprod hypoxia; 13% O2), and HYP2 (matched %V̇o2peak hypoxia; 13% O2). Steady-state (30–45 min) mean arterial pressure (MAP) and maximum CVC values obtained during 45 min of local heating (CVCmax) are also displayed on right. Mean MAP, Hb, Hct, and ΔPV values for both studies are all displayed in Table 2. In the fixed Hprod trials, the HR response was greater over time in HYP1 compared with NORM (P < 0.01, ηp2 = 0.71) with mean values of 113 (18) beats/min in NORM and 134 (14) beats/min in HYP1 (Fig. 5A), MAP was not different between conditions (P = 0.29, ηp2 = 0.16), SpO2 decreased over time in HYP1 compared with NORM (P < 0.01, ηp2 = 0.54) with mean values of 96 (1)% in NORM and 76 (4)% in HYP1 (Fig. 5B), and reductions in PV after 45 min of exercise tended to be greater in HYP1 compared with NORM (P = 0.06, d = 0.81). In the %V̇o2peak-matched trials, hypoxia had no effect on HR (P = 0.17, ηp2 = 0.21) with mean values of 113 (18) beats/min in NORM and 119 (16) beats/min in HYP2, and no effect on MAP (P = 0.91, ηp2 = 0.02), or ΔPV (P = 0.17, d = 0.41), but SpO2 decreased over time in HYP2 with mean values of 76 (3)% compared with NORM (P < 0.01, ηp2 = 0.56).
Fig. 5.Heart rate (A) and oxygen-hemoglobin saturation (SpO2, B) responses as a function of time while cycling in normoxia (NORM, 21% FIO2), hypoxia at the same heat production as in NORM (HYP1, 13% FIO2), and hypoxia at the same %V̇o2peak as in NORM (HYP2, 13% FIO2). *Significant difference between NORM and HYP2 (P < 0.05). †Significant difference between NORM and HYP1 (P < 0.05). Ventilatory and respiratory responses are displayed in Fig. 6. During the fixed Hprod trials, V̇o2 (P = 0.12, ηp2 = 0.38) and RER (P = 0.35, ηp2 = 0.29) were not different over time between NORM and HYP1, while minute ventilation (P < 0.01, ηp2 = 0.69) was greater in HYP1 compared with NORM. Conversely, when matched for %V̇o2peak, V̇o2 (P < 0.01, ηp2 = 0.95) was lower in HYP2 compared with NORM, while RER was not different in HYP2 (P = 0.56, ηp2 = 0.19), and the V̇e (P = 0.01, ηp2 = 0.59) was significantly lower in HYP2 than in NORM. Mean respiratory heat losses during exercise [NORM: 35 (2) W; HYP1: 41 (3) W] were greater (P < 0.01, d = 2.74) in HYP1 at a fixed Hprod and were lower [NORM: 35 (2) W; HYP2: 32 (3) W] in HYP2 when matched for %V̇o2peak (P = 0.02, d = 1.07). Fig. 6.The rate of O2 consumption (V̇o2, A), respiratory exchange ratio (RER, B), and rate of ventilation (V̇e, C) over time while cycling in normoxia (NORM, 21% FIO2), hypoxia at the same heat production as in NORM (HYP1, 13% FIO2), and hypoxia at the same %V̇o2peak as in NORM (HYP2, 13% FIO2). *Significant difference between NORM and HYP2 (P < 0.05). †Significant difference between NORM and HYP1 (P < 0.05). In the fixed Hprod trials, RPE values (Borg) were higher over time in hypoxia (P < 0.01, ηp2 = 0.44) with mean values of 11 (2) in NORM and 13 (2) in HYP1. Conversely, when exercising at a matched %V̇o2peak RPE values were lower over time in HYP2 (P < 0.01, ηp2 = 0.43) with a mean value of 10 (1). DISCUSSIONThe present results conclusively demonstrate, using a within-subject experimental design, that thermoregulatory responses of core temperature and sweating during exercise at a fixed rate of Hprod, and therefore fixed evaporative heat balance requirement (Ereq), are unaffected by acute hypoxia. At the same time, we provide evidence illustrating that exercise at a fixed %V̇o2peak but different Hprod (and Ereq), does not determine time-dependent changes in core temperature or local sweat rates. The current study isolated the influence of hypoxia itself on thermoregulatory responses to exercise by fixing the rates of heat production between normoxia and hypoxia (NORM vs. HYP1). It follows that no independent effect of hypoxia on core temperature or sweat rates was observed. Previous assessments of thermoregulation at simulated high altitude (i.e., hypoxia) have compared thermo-effector responses at altitude-specific %V̇o2peak (32, 33) and found that exercise at 60% of altitude-specific V̇o2peak attenuated steady-state forearm sweat rates in the high-altitude trials (4,575 m). However, these observations (32, 33) can be attributed to differences in heat production and, therefore Ereq, stemming from a reduced V̇o2peak at high altitude (i.e., hypoxia) and not to hypoxia per se. Indeed, our current results demonstrate that LSR are the same when Ereq is fixed (NORM vs. HYP1) and lower when Ereq was reduced to match %V̇o2peak in HYP2 (Fig. 2). Dipasquale et al. (16) examined LSR in normobaric hypoxia using pharmacological induction of sweating and reported that a peripheral effect of hypoxia suppressed sweating on the forearm by 16% compared with normoxia. However, in the current study, the lower steady-state LSR (at all sites) observed in HYP2 compared with NORM were mediated by fewer activated sweat glands and not differences in mean SGO (Fig. 3), which indicates that hypoxia did not have a peripheral influence on the sweat gland. Other studies have previously reported no effect of hypoxia on core temperature during exercise at fixed absolute external workloads (2, 28, 34) and also no difference in chest LSR during fixed workload exercise between normoxia and hypoxia (45). However, two of these studies reported a higher LSR on the forehead during normothermic (~22°C) exercise in hypoxia compared with normoxia (28, 34). It is noteworthy, however, that while the same fixed workloads were used in normoxia and hypoxia in all of these studies, heat production was not measured. In the current study, different external workloads were required in NORM and HYP1 to generate a fixed heat production (Table 1) due to differences in mechanical efficiency. It was suggested that higher sweat rates in glabrous skin (i.e., forehead) may have been potentiated by greater perceived strain relating to nonthermal factors such as increased HR and dyspnea due to hypoxia (28). In support, greater forehead LSR have also been observed in normoxic conditions between groups of low and high aerobic fitness, independently of Ereq, which was also associated with greater perceptual strain (10). The reasons for the absence of an influence of %V̇o2peak on forehead LSR (Fig. 2, right) in the present study is unclear. It is possible that perceptual strain was insufficiently high to induce greater sweating from regions sensitive to such a stimulus as a result of low workloads in hypoxia. Indeed, cycling resistance was <100 W in the current study, and despite that workload tended to be lower in HYP1 compared with NORM at a fixed heat production, RPE was greater in the hypoxic condition given the greater %V̇o2peak (Table 1). Yet RPE was also lower in HYP2 compared with NORM when matched for %V̇o2peak, which could potentially indicate a differential response between RPE and relative exercise intensity based on V̇o2peak. However, workloads were ~60 W in HYP2 compared with ~90 W in NORM, which suggests RPE during cycling could also be influenced by peripheral sensation of force production relating to different cycling workloads (6). Additionally, such large differences in V̇o2peak within a given subject between normoxia and hypoxia allows for a unique situation in which the influence of %V̇o2peak on changes in core temperature and sweating can be assessed without potential confounding factors that arise when using a between-group experimental design (e.g., mass, BSA, body fat %, partial acclimation status) (7, 11). Recently, Jay et al. (27) demonstrated that differences in aerobic capacity between mass-matched groups of high and low fitness do not influence thermoregulatory responses during exercise at the same Hprod. The current study lends further support to this previous finding (27). It was found that despite differences in %V̇o2peak of 15–20%, core temperature (Fig. 1) and sweating (Fig. 2) were unaffected by hypoxia during exercise at a fixed Hprod. Conversely, to match %V̇o2peak in hypoxia to normoxia, Hprod (and thus Ereq) was lowered, resulting in smaller changes in rectal and esophageal temperatures (Fig. 1), as well as attenuations of LSR in hypoxia at all sites (Fig. 2). This conclusively demonstrates that setting exercise intensity at a fixed percentage of V̇o2peak to compare core temperature and sweating responses, as traditionally advised (3, 20, 54), should be avoided if Hprod is different between conditions, individuals, or groups. Rather, fixing Hprod [per unit mass, i.e., W/kg (8, 9)] has been demonstrated to be a more suitable method of setting exercise intensity when the change in core temperature is the primary outcome measure (11). While it was not part of the specific design of the present study, we observed a very narrow range of oxygen-hemoglobin saturation values (SpO2; Fig. 5B) in our participants with 95% confidence intervals of 74–79%. Therefore, we can confidently state that we successfully elicited a fixed level of hypoxemia in the present study. Accordingly, HR was higher during hypoxic exercise by ~20 beats/min at a fixed rate of heat production (and V̇o2) compared with normoxia (Fig. 5A). It is likely that the demand for oxygen delivery by skeletal muscles was greater given the reductions in oxygen saturation, such that increases in cardiac output via HR were necessitated. There is also evidence of a resetting of baroreflex control in hypoxia leading to an increased HR (22), which is consistent with higher resting HRs in both HYP1 and HYP2 by ~8 beats/min compared with NORM in the current study. Conversely, HR did not differ between NORM and HYP2 while exercising at a similar %V̇o2peak, but much lower absolute V̇o2 in HYP2. While hypoxia exerts an autonomic reflex response, as evidenced by increased sympathetic vasoconstrictor nerve activity (52), many vascular beds also exhibit a vasodilator response including the coronary (58), cerebral (57), skeletal muscle (8), and cutaneous circulations (56). Yet reduced vascular resistance in beds such as the coronary and cerebral circulations must be accompanied by physiological compensations to maintain MAP, either by increases in HR, as currently observed, or perhaps due to sympathetic vasoconstrictor tone in other vascular beds (40). Although hypoxia-mediated reductions in arterial partial pressure of oxygen stimulate higher V̇e (39, 59), which might be thought to contribute to greater rates of heat loss (21), the difference in ventilation (Fig. 6) observed in the current study only resulted in an extra 6 W of respiratory heat losses in HYP1 compared with NORM. A difference of this magnitude would be equivalent to <7 g of required sweat evaporation over 45 min, whereas mean WBSL after 45 min in HYP2 was 458 (107) g; thus any increases in respiratory heat loss related to hypoxia would be obscured by the variability of the measurement. Greenleaf et al. (21) also proffered that increases in respiratory heat loss must be compensated by reductions in heat loss via other physiological mechanisms to maintain a constant core temperature. They suggested peripheral blood flow was reduced because estimates of tissue conductance were correspondingly lower at 2,000 and 4,000 m of altitude in a hypobaric hypoxia chamber compared with increases in respiratory heat losses. However, their calculations of tissue conductance were most likely confounded by large reductions in skin temperature (−2.5°C) in the high altitude conditions as a result of the greater evaporative efficiency from high air velocities and actual peripheral blood flow was not measured. In support of Greenleaf et al. (21), Miyagawa et al. (45) reported lower forearm blood flow (FBF) during hypoxic exercise in the heat (30°C, 50% RH) compared with normoxia; however, other studies have also reported no effect (51) or greater (5, 53, 56) cutaneous blood flow during hypoxic exposures. Rowell et al. (51) measured FBF during normothermic exercise and found no effect of hypoxia (FIO2 = 0.12), while Black and Roddie (5) and Sagawa et al. (53) both observed a greater FBF during passive hypoxic exposures. More recently, Simmons et al. (56) used laser-Doppler flowmetry (LDF) to directly measure SkBF on the forearm and found that CVC was higher during passive exposure to isocapnic hypoxia in normothermia. Lawley et al. (36) confirmed these findings in local heating (44°C) of the forearm during prolonged hypoxia exposure (9 h; FIO2 = 0.12). Alternatively, Low et al. (38) recently demonstrated the potential for regional differences in the measurement of SkBF during passive exposure to heat and hypoxia (FIO2= 0.13). Using LDF, they reported no effect of hypoxia on CVC at the chest and lower CVC during hypoxia on the forearm. Ultimately, LSR in their study were unaffected by hypoxia, and core temperature was not compromised by possible reductions in CVC on the forearm during whole body heating. The current study, which was apparently the most strenuous combination of hypoxia (FIO2 = 0.13) and exercise-heat stress (34.5°C, 46% RH) performed to date, also observed a potentially higher steady-state %CVCmax on the forearm and upper back in hypoxia compared with normoxia at a fixed Hprod (Fig. 4). It has previously been suggested that cutaneous blood flow is determined by the rate of heat production (17); therefore, differences in CVC between fixed Hprod trials (NORM vs. HYP1) might indicate an independent role of hypoxia in the regulation of SkBF. This lends partial support to the results of Simmons et al. (56) and Lawley et al. (36) and suggests a potential vasodilatory effect of hypoxia during exercise in the heat. However, it must be noted that CVC values in AU were not different between trials (Fig. 4). Mathematically, %CVCmax in the current study was probably higher primarily as a result of lower average maximum CVC values in HYP1 compared with NORM, but maximum CVC values in HYP2 were similar to NORM (Fig. 4), which indicates that the possible difference in maximum CVC values between NORM and HYP1 was not consistently driven by hypoxia. Several studies have suggested potential implications for heat dissipation from the skin when skin blood flow is altered (25, 30, 45, 55, 56); however, our current results demonstrate that this was not translated to differences in sweating or skin or core temperatures (Fig. 1). Although supraphysiological reductions in skin blood flow impair local sweat rates during passive heating (60), physiologically relevant differences in skin blood flow (such as in the present study) may not be as important for thermoregulatory capacity given the small temperature gradient between the skin and environment unless sweating is altered, which was not the case in the current study (Fig. 2). Based on work from our research group over the last five years including the present study, it can be stated conclusively that relative exercise intensity (%V̇o2peak) does not determine changes in core temperature (11, 12, 27). Our previous investigations used between-group designs, but the current study, using hypoxia, truly isolated the independent influence of %V̇o2peak within the same individual. Our previous studies have demonstrated that normalizing heat production for body mass eliminates systematic differences in core temperature that seem to arise from differences in body size (11). It has also been suggested that heat production relative to lean body mass might be a strong predictor of core temperature given the large “heat sink” effect of fat free mass (1, 19); however, we have recently provided evidence demonstrating that such an approach also leads to systematically different thermoregulatory responses (14). While we demonstrated that thermoregulatory responses are not influenced by relative exercise intensity per se, that is not to say that there is no effect of physical training, which itself has the potential to induce partial acclimation. Any heat-related adaptations that may accompany training or repeated elevations in core temperature or sustained sweating may augment heat dissipation and therefore thermoregulatory capacity, particularly in hot and uncompensable conditions. The current study was performed in compensable conditions, so if an influence of relative exercise intensity exists on the upper limits of heat loss it would not have been observed. This study isolated the independent influence of hypoxia on thermoregulatory responses to exercise in hot conditions and determined that acute hypoxia does not alter thermoregulatory control. It was also conclusively demonstrated that there is no within-subject influence of relative exercise intensity on changes in core temperature or sweating responses independently of heat production. Exercise set at a fixed %V̇o2peak led to systematic differences in core temperature and local sweat rates. On the other hand, exercise set at a fixed Hprod and Ereq eliminated systematic differences in thermoregulatory responses, irrespective of large differences in %V̇o2peak. GRANTSThis project was supported in part by a Natural Sciences and Engineering Research Council of Canada (NSERC) equipment grant (O. Jay, P. Imbeault), NSERC Discovery Grant (P. Imbeault), and a student research grant from the Gatorade Sports Science Institute (G. Coombs). G. B. Coombs was supported by a University of Ottawa Master’s Scholarship, M. N. Cramer was supported by an Ontario Graduate Scholarship, and N. Ravanelli was supported by an NSERC Postgraduate Scholarship (PGS-D). DISCLOSURESThe authors declare no conflicts of interest, financial or otherwise. AUTHOR CONTRIBUTIONSG.B.C., P.I., and O.J. conceived and designed experiments; G.B.C., M.N.C., and N.R. performed experiments; G.B.C. analyzed data; G.B.C., M.N.C., N.R., P.I., and O.J. interpreted data; G.B.C drafted manuscript; G.B.C., N.R., M.N.C., P.I., and O.J. edited, revised, and approved final draft of the manuscript. The authors thank Jean-François Mauger for technical assistance, Yannick Plante for help during data collection, Dr. Tony Carlsen for input on statistical analysis, and finally the participants for the devotion of their time. REFERENCES
Page 17AbstractAlthough all-out exercise protocols are commonly used, the physiological mechanisms underlying all-out exercise performance are still unclear, and an in-depth assessment of skeletal muscle bioenergetics is lacking. Therefore, phosphorus magnetic resonance spectroscopy (31P-MRS) was utilized to assess skeletal muscle bioenergetics during a 5-min all-out intermittent isometric knee-extensor protocol in eight healthy men. Metabolic perturbation, adenosine triphosphate (ATP) synthesis rates, ATP cost of contraction, and mitochondrial capacity were determined from intramuscular concentrations of phosphocreatine (PCr), inorganic phosphate (Pi), diprotonated phosphate (H2PO4−), and pH. Peripheral fatigue was determined by exercise-induced alterations in potentiated quadriceps twitch force (Qtw) evoked by supramaximal electrical femoral nerve stimulation. The oxidative ATP synthesis rate (ATPOX) attained and then maintained peak values throughout the protocol, despite an ~63% decrease in quadriceps maximal force production. ThusATPOX normalized to force production (ATPOX gain) significantly increased throughout the exercise (1st min: 0.02 ± 0.01, 5th min: 0.04 ± 0.01 mM·min−1·N−1), as did the ATP cost of contraction (1st min: 0.048 ± 0.019, 5th min: 0.052 ± 0.015 mM·min−1·N−1). Additionally, the pre- to postexercise change in Qtw (−52 ± 26%) was significantly correlated with the exercise-induced change in intramuscular pH (r = 0.75) and H2PO4− concentration (r = 0.77). In conclusion, the all-out exercise protocol utilized in the present study elicited a “slow component-like” increase in intramuscular ATPOX gain as well as a progressive increase in the phosphate cost of contraction. Furthermore, the development of peripheral fatigue was closely related to the perturbation of specific fatigue-inducing intramuscular factors (i.e., pH and H2PO4− concentration). NEW & NOTEWORTHY The physiological mechanisms and skeletal muscle bioenergetics underlying all-out exercise performance are unclear. This study revealed an increase in oxidative ATP synthesis rate gain and the ATP cost of contraction during all-out exercise. Furthermore, peripheral fatigue was related to the perturbation in pH and deprotonated phosphate ion. These findings support the concept that the oxygen uptake slow component arises from within active skeletal muscle and that skeletal muscle force generating capacity is linked to the intramuscular metabolic milieu. all-out exercise protocols are increasingly being utilized as an integral component of experimental designs (7, 22, 56, 58). The appeal of these protocols stems from the potential to determine multiple important physiological parameters in a single testing session. Indeed, maximal oxygen uptake (V̇o2max), a gold standard in human physiology (19, 48), is elicited during all-out cycling exercise lasting longer than 90 s (7, 58). Furthermore, the 3-min all-out cycling protocol allows for the determination of critical power (CP) (for review see Refs. 7, 53), a parameter with great utility in both health and disease (25, 44, 55). This protocol has been adapted to provide equivalent parameters for other exercise modalities (4, 6, 9, 42, 57), such as critical force (CF), during a 5-min all-out knee-extensor protocol (6). However, despite growing use of these protocols, the physiological mechanisms underlying performance during all-out exercise are still under investigation, and an in-depth assessment of skeletal muscle bioenergetics (i.e., sources and rates of ATP synthesis) is currently lacking. Interestingly, all-out exercise exhibits an oxygen uptake (V̇o2) slow component (7, 18, 52, 56, 59). The V̇o2 slow component, originally described for constant-work-rate exercise above the lactate threshold, is manifest as additional V̇o2, predominantly arising from a loss of efficiency within the exercising skeletal muscle (43, 45, 47), and is closely related to exercise tolerance (40). During all-out exercise, the V̇o2 slow component likely arises as a result of all motor units being recruited at the onset of exercise and motor units dropping out as fatigue ensues (i.e., decreased force production), while V̇o2max is maintained (7, 18, 52, 56, 59). Importantly, these findings demonstrate that progressive motor unit recruitment is not requisite for the V̇o2 slow component and therefore provides an interesting approach to study the mechanisms by which the V̇o2 slow component originates. In particular, it remains to be determined whether the slow component during all-out exercise is still evident during small-muscle-mass exercise and actually arises from the exercising skeletal muscle. Remarkably, the magnitudes of intramuscular metabolic perturbation during the 3-min all-out cycling and 5-min all-out knee-extensor protocols are very similar (8, 52). Specifically, intramuscular phosphocreatine concentration ([PCr]) decreased to ~20–25% of resting concentrations and muscle pH decreased to ~6.7, while muscle inorganic phosphate concentration ([Pi]) and blood lactate concentration increased to ~500% and ~1,200% of resting concentrations, respectively (8, 52). For high-intensity cycling exercise, the magnitude of intramuscular perturbation ([Pi] and pH) is closely related to peripheral fatigue (2). However, the robustness of this relationship across exercise modality and exercising muscle mass remains unknown. Of note, the 5-min all-out knee-extensor protocol both induces a high degree of peripheral fatigue [~50% reduction in potentiated quadriceps twitch force (Qtw)] (6) and allows for phosphorus magnetic resonance spectroscopy (31P-MRS) measurements of intramuscular metabolites (8), making this an ideal exercise paradigm to further assess the relationship between the magnitude of intramuscular perturbation and the development of peripheral fatigue. Therefore, the purpose of this study was to quantitatively characterize skeletal muscle bioenergetics and peripheral fatigue during all-out exercise. Specifically, we utilized 31P-MRS to determine the rate and source of ATP production and electrical femoral nerve stimulation to determine peripheral fatigue for all-out intermittent isometric single-leg knee-extensor exercise. We hypothesized that during the all-out exercise 1) the oxidative ATP synthesis rate (ATPOX) would reach and then remain at peak values, while force production would progressively decrease, indicating an intramuscular V̇o2 slow component; 2) the ATP cost of contraction would progressively increase; and 3) in terms of fatigue, the reduction in quadriceps Qtw would be related to the change in both intramuscular [Pi] and pH. METHODSEight healthy men (age 28 ± 5 yr, stature 178 ± 4 cm, and body mass 77 ± 8 kg) volunteered and provided written informed consent to participate in this investigation. All experimental procedures were conducted in accordance with the Declaration of Helsinki and were approved by the Institutional Review Boards of the University of Utah and the Salt Lake City Department of Veterans Affairs Medical Center. Subjects were instructed to abstain from vigorous activity during the 24 h preceding each visit to the laboratory and to arrive at the laboratory having abstained from food and caffeine during the preceding 3 h. Subjects were tested in the laboratory a minimum of twice with at least 72 h between visits. Single-leg intermittent isometric knee-extensor exercise (3-s contraction and 2-s relaxation) was performed for all exercise protocols. This was conducted in a semirecumbent position (~15° elevation of the trunk), with the knee of the leg to be exercised situated over a custom-built knee support (~45° knee joint angle), the ankle fixed to an immovable strain gauge (SSM-AJ-250, Interface), and nonelastic straps positioned over the hips and thigh. The choice of exercising leg was balanced for dominance across subjects. During the initial visit, subjects performed 60 maximal voluntary quadriceps contractions (MVCs) over 5 min with neuromuscular function assessments before and immediately after exercise (i.e., at the 60th MVC). This 60-MVC protocol with 5 min of recovery was then completed inside a whole body magnetic resonance imaging (MRI) system during the second visit. An audio recording cued the start and stop of each 3-s contraction, but no information regarding the duration of exercise was provided to the subjects. The integrated force, mean force, and peak force were determined for each of the 60 MVCs, and CF was defined as the mean force of the final 6 MVCs (6). During the initial visit, neuromuscular function assessments were conducted before and immediately after exercise. The femoral nerve was stimulated with a constant-current stimulator (model DS7AH, Digitimer), with the anode placed in the femoral triangle and the cathode placed between the greater trochanter and the iliac crest. Low-intensity single-pulse stimuli (200-μs pulse width, 100–150 mA) were used to locate the optimal position of the stimulating electrode, defined as the location evoking the greatest force production. The electrodes were then fixed in position until all measurements for the visit were conducted. The stimulation intensity was increased in 20-mA increments until maximal force was obtained. The stimulator intensity was then set to 120% of this, to ensure supramaximality. For the evaluation of quadriceps function, Qtw was performed 2 s after 3-s MVCs both before exercise and then immediately after exercise (i.e., at the 60th MVC). For each Qtw, the contraction time (CTQtw) and half-relaxation time (0.5RTQtw) were determined. During each MVC, a superimposed twitch was delivered and voluntary activation of the quadriceps was calculated as VA = [1 − (superimposed twitch/Qtw)] × 100. A clinical 2.9-T MRI system (Tim-Trio, Siemens Medical Systems, Munich, Germany) operating at 49.9 MHz (31P resonance) and a dual-tuned 31P-1H surface coil (110-mm 1H coil loop surrounded by 31P single-loop coil with a diameter of 125 mm) with linear polarization were utilized to acquire 31P-MRS data (RAPID Biomedical, Rimpar, Germany). The surface coil was secured over the midthigh with elastic straps, and advanced localized volume shimming was performed after a three-plane scout proton image was acquired to ensure that all major quadriceps muscles were sampled. Two fully relaxed spectra were acquired (3 averages per spectrum and a repetition time of 30 s) before the 60-MVC protocol commenced. Throughout exercise and recovery, MRS data acquisition was conducted with a free-induction decay pulse sequence with a 2.56-ms adiabatic-half-passage excitation RF pulse, a repetition time of 2.5 s, a receiver bandwidth of 5 kHz, 1,024 data points, and 2 averages per spectrum. Comparisons between fully relaxed and partially relaxed spectra were utilized to quantify saturation factors. A time-domain fitting routine using the AMARES algorithm (51) incorporated into CSAIPO software (33–35) was utilized to determine absolute and relative concentrations of intramuscular PCr, Pi, H2PO4−, and ATP. Intracellular pH was calculated from the chemical shift difference between the Pi and PCr signals. The free cytosolic [ADP] was calculated from [PCr] and pH with the creatine kinase (CK) equilibrium constant (KCK = 1.66 × 109 M−1) and the assumption that PCr represents 85% of the total creatine content (23). Resting concentrations were calculated from the average peak areas of the two fully relaxed spectra and assuming a resting [ATP] of 8.2 mM (16). To account for Pi splitting, the pH corresponding to each Pi pool was calculated separately as pH1 and pH2 on the basis of the chemical shift of each peak relative to PCr, such that the overall pH was then calculated as pH=pH1×(areaPi1/total Pi area)+pH2×(areaPi2/total Pi area) The concentration of H2PO4− was calculated as (32)H2PO4−=[Pi]/(1+10pH-6.75) Free cytosolic adenosine monophosphate (AMP) was calculated based on the equilibrium of the adenylate kinase reaction corrected for the effects of pH and assuming a free magnesium concentration of 1 mM (15). Relative amplitudes were corrected for partial saturation due to the repetition time relative to T1 with the fully relaxed spectra acquired at rest. The rate of ATP production from the breakdown of PCr through the CK reaction (ATPCK, mM/min) was calculated from the change in [PCr] for each time point of the exercise period (27): ATPCK=ΔPCr/Δt Based on the sigmoid relationship between ATPOX (mM/min) and free cytosolic [ADP], the rate of mitochondrial ATP production was calculated as ATPOX=Vmax/(1+(Km/[ADP])2.2) in which Km (the [ADP] at half-maximal oxidation rate) is ~30 μM in skeletal muscle (27), 2.2 is the Hill coefficient for a sigmoid function (24), and Vmax is the peak rate of in vivo oxidative ATP synthesis (see PCr recovery kinetics).During exercise, changes in intramuscular pH result from glycogen breakdown to pyruvate and lactate, proton efflux, buffering capacity, protons produced by oxidative phosphorylation, and the consumption of protons by the CK reaction (27). Assuming that the glycogenolytic production of 1 mol of H+, when coupled to ATP hydrolysis, yields 1.5 mol of ATP, the ATP production from anaerobic glycolysis (ATPGLY) can be deduced from the total number of protons (P) produced throughout exercise (20, 27, 28): ATPGLY=1.5×P whereP=HCK++Hβ+−HOX++Hefflux+ HCK+ (in mM/min) was calculated from the time-dependent changes in [PCr] and from the stoichiometric coefficient (γ): HCK+=−γ×ATPCK where γ is the proton stoichiometric coefficient of the coupled Lohmann reaction as previously described (31). Hβ+ (in mM/min) was calculated from the apparent buffering capacity βtotal (in slykes, millimoles acid added/unit change in pH) and from the rate of pH changes:Hβ+=−βtotal×ΔpH/Δt whereβtotal=βnonbicarbonate-nonPi+βPi+βbicarbonate whereβnonbicarbonate-non-Pi=βa−βPi in which βa was determined from the initial change in PCr (ΔPCri) and alkalinization of pH (ΔpH) (10):βa=γ×(PCri×ΔpH) βPi was determined based on the dissociation constant of the buffer (K) according to the standard formula (11): βPi=(2.303×H+×K×[Pi])/(K+H+)2 where K = 1.77 × 10−7. In agreement with previous studies and assuming that muscle is a closed system during exercise (11, 28), βbicarbonate was set to zero. HOX+ (in mM/min) was calculated from the factor m = 0.16/[1 + 10(6.1 – pH)], which accounts for the amount of protons produced through oxidative ATP production (28, 29):HOX+=m×ATPOX Hefflux+ (in mM/min) was calculated for each time point of exercise using the proportionality constant λ relating proton efflux rate to ΔpH (28, 29):Hefflux+=−λΔpH This proportionality constant λ (in mM·min−1·pH unit−1) was calculated during the recovery period: λ=−Vefflux/ΔpH During the recovery period, PCr is regenerated throughout the CK reaction as the consequence of oxidative ATP production in mitochondria. Thus Hefflux+ can be calculated from the rates of proton production from the CK reaction (HCK+, in mM/min) and mitochondrial ATP production (HOX+, in mM/min) on one side and the rate of pH changes on the other side. At this time, ATP production is exclusively aerobic and lactate production is considered negligible: Vefflux=βtotal×ΔpH/Δt+γ×ViPCr+m×ATPOX To improve precision, a modified version of this calculation was used (28, 29), in which the total proton disappearance (i.e., ∫Edt) is estimated cumulatively from the start of recovery and then fitted to an exponential function to obtain the initial recovery rate (Vefflux). The total ATPase rate (ATPTOTAL, in mM/min) was calculated for each time point as ATPTOTAL=ATPOX+ATPCK+ATPGLY and the anaerobic ATPase rate (ATPANA, mM/min) was calculated for each time point asATPANA=ATPCK+ATPGLY The ATP cost of contraction (in mM/N) was calculated as the ratio between ATPTOTAL and the force integral. The ATPOX and ATPANA gains were determined as the respective ATP synthesis rate normalized to the force integral. All values were calculated for each minute of exercise. The [PCr] kinetics during the recovery period were described by a monoexponential curve: [PCr](t)=[PCr]end+Δ[PCr]×(1−e−(t/τp)) where [PCr](t) is the [PCr] at a given time t, [PCr]end is the [PCr] at end exercise, Δ[PCr] is the amount of PCr resynthesized during the recovery period, and τ represents the time constant of the PCr offset kinetics. The initial rate of PCr resynthesis (ViPCr) was calculated from the derivative of the monoexponential equation at the onset of recovery:ViPCr=k×Δ[PCr] where Δ[PCr] is the amount of PCr resynthesized during the recovery period and the rate constant k = 1/ τ (27). The peak rate of in vivo oxidative ATP synthesis (Vmax) was calculated with ViPCr and the [ADP] at end exercise (50):Vmax=ViPCr×[1+(Km/[ADP]end2.2)] where Km (the [ADP] at half the highest oxidative rate) is ~30 μM in skeletal muscle (27).Model variables were determined with an iterative process by minimizing the sum of squared residuals (RSS) between the fitted function and the observed values. Goodness of fit was assessed by visual inspection of the residual plot and the frequency plot distribution of the residuals, with the χ2 values and the coefficient of determination (r2) calculated as follows (39): r2=1−(SSreg/SStot) Force, 31P-MRS, ATP synthesis, ATP cost of contraction, ATPOX gain, and ATPANA gain were analyzed by one-way ANOVAs with repeated measures. A two-way ANOVA with repeated measures was used to analyze the percent contribution of ATPOX and ATPANA to ATPTOTAL. Tukey’s post hoc analyses were conducted when significant main effects were detected. Preexerecise-to-postexercise comparisons for MVC, Qtw, and VA were made with Student’s paired t-tests. ATPOX was compared with the 95% confidence interval (CI) around Vmax to determine whether maximal oxidative ATP production was attained. The relationships between neuromuscular function measurements and intramuscular metabolic perturbations were assessed with Pearson product moment correlation coefficients. Significance for the statistical analysis was accepted at P < 0.05. Results are presented as means ± SD, except in figures, where SE is used for clarity. RESULTSThe peak MVC force attained during the protocol was 568 ± 126 N, which significantly decreased to 234 ± 50 N by the final MVC (Fig. 1). Throughout the majority of the exercise protocol both the peak force and the integrated force per MVC significantly decreased over time (Fig. 1). However, as illustrated, force did not significantly change over the final 30 s of the test and the resulting CF was 209 ± 41 N, which corresponded to 38 ± 10% of the peak force during the protocol. The intramuscular metabolic responses to the 60-MVC protocol and postexercise recovery are presented in Fig. 2. All variables changed significantly as a function of time during both the exercise and recovery periods. Intramuscular pH remained significantly below, and [H+] significantly above, baseline values for the entire duration of the 5-min recovery period, while [PCr], [Pi], [H2PO4−], and free energy of ATP hydrolysis [ΔGATP] were not significantly different from baseline values by the end of this period. The force generating capacity of the quadriceps was significantly reduced immediately after exercise, as measured by the pre- to postexercise reduction in MVC. This decrease was accompanied by significant central (ΔVA: −12 ± 11%) and peripheral (ΔQtw: −52 ± 26%) fatigue (Table 1). The fall in Qtw was significantly correlated with the decrease in intramuscular pH and increase in intramuscular [H2PO4−] (Fig. 3), while the correlations with the increase in intramuscular [H+] and [Pi] tended to be significantly related, with P values of 0.06. Fig. 1.Force development during the 5-min all-out intermittent isometric single-leg knee-extensor protocol. Subjects performed a series of 60 intermittent maximal voluntary contractions (3-s contraction, 2-s relaxation) over 5 min. Peak force and mean force were determined per maximal voluntary contraction. Fig. 2.Intramuscular metabolic perturbation during the 5-min all-out intermittent isometric single-leg knee-extensor protocol. Intramuscular metabolite concentrations were determined with phosphorus magnetic resonance spectroscopy. †Significantly different from baseline.
Fig. 3.Relationship between quadriceps fatigue and intramuscular metabolites. Data are expressed as % difference from baseline to end exercise for intramuscular pH and hydrogen ion ([H+]), inorganic phosphate ([Pi]), and diprotonated phosphate ([H2PO4−]) concentrations vs. preexercise-to-postexercise % difference for the potentiated quadriceps twitch force (Qtw). ATP synthesis rates and ATP cost of contraction data are presented in Fig. 4. ATPCK, ATPGLY, ATPANA, and ATPTOTAL significantly decreased throughout the exercise test. ATPOX remained within the Vmax 95% CI (30.8 ± 8.7 mM·min−1·N−1, 95% CI: 23.5–38.1 mM·min−1·N−1) throughout the entire protocol. There was a significant, but transient, increase in ATPOX during the protocol, with a small significant difference between the second and fifth minutes. The ATP cost of contraction significantly increased as a function of time during the exercise. The change in ATP cost of contraction was not significantly correlated with changes in any of the neuromuscular function measures. Data for the percent contribution of ATPOX and ATPANA to ATPTOTAL are presented in Fig. 5. The percentage of ATPTOTAL arising from ATPANA significantly decreased over the first 2 min, while the percentage arising from ATPOX significantly increased over the first 2 min, with no significant changes in either variable thereafter. The percent contribution from ATPOX was significantly greater than ATPANA for minutes 2–5 (Fig. 5). During exercise at CF (i.e., the final 30 s), the percentage of ATPTOTAL arising from ATPOX (83 ± 13%) was significantly greater than ATPANA (17 ± 13%). The ATPOX gain significantly increased throughout the exercise, while the ATPANA gain significantly decreased from the first minute to the second minute and did not significantly change thereafter (Fig. 6). Fig. 4.Adenosine triphosphate (ATP) synthesis rates and ATP cost of contraction during 5-min all-out intermittent isometric single-leg knee-extensor protocol. Rate of ATP synthesis through creatine kinase reaction (ATPCK), anaerobic glycolysis (ATPGLY), cumulative anaerobic metabolism (ATPANA), and oxidative phosphorylation (ATPOX), total ATPase rate (ATPTOTAL), and ATP cost of contraction were determined for each minute of exercise. Significantly different: †from 1st minute, ‡from 2nd minute, *from 3rd minute (P < 0.05). Fig. 5.Contribution of oxidative and anaerobic adenosine triphosphate (ATP) production during 5-min all-out intermittent isometric single-leg knee-extension protocol. Rates of ATP synthesis through oxidative phosphorylation (ATPOX) and anaerobic metabolism (ATPANA) are expressed relative to total ATPase rate (ATPTOTAL). Significantly different: *between conditions, †from 1st minute, ‡from 2nd minute (P < 0.05). Fig. 6.Oxidative and anaerobic adenosine triphosphate (ATP) gain during 5-min all-out intermittent isometric single-leg knee-extension protocol. Gains were determined as the rate of ATP synthesis through oxidative phosphorylation (ATPOX) and anaerobic metabolism (ATPANA) normalized to the integrated force. Significantly different: †from 1st minute, ‡from 2nd minute, *from 3rd minute (P < 0.05). DISCUSSIONThis study utilized 31P-MRS to quantitatively characterize skeletal muscle bioenergetics during all-out exercise. Consistent with our first hypothesis, ATPOX attained and remained at peak values, while force production progressively decreased throughout the all-out exercise. As a result, the ATPOX gain progressively increased in a “slow component-like” manner, similar to the V̇o2 slow component documented during all-out cycling exercise. Furthermore, the ATP cost of muscle contraction progressively increased throughout the all-out exercise, which is consistent with our second hypothesis. This provides evidence of an increased phosphate cost of force generation across time with such all-out exercise. In agreement with our third hypothesis, the exercise-induced reduction in Qtw was related to the changes in intramuscular [H2PO4−] and pH. This finding reinforces the link between the magnitude of intramuscular perturbation and peripheral fatigue that has previously been documented for large-muscle-mass cycling exercise. In combination, the findings of this study offer a novel, in-depth assessment of skeletal muscle bioenergetics during all-out exercise, while also providing mechanistic insight in the determinants of the V̇o2 slow component and neuromuscular fatigue during exercise. The force profile and intramuscular metabolic perturbation (both magnitude and time course) in this study were similar to previous reports for both the 5-min all-out knee-extensor and 3-min all-out cycling protocols (6, 8, 52, 53). However, of importance, the present work builds upon and extends the findings of these previous studies by assessing the sources and rates of ATP synthesis throughout the all-out exercise (Fig. 4). Consistent with the role of PCr as an energy buffer, ATPCK peaked during the first minute of exercise and then rapidly decreased to no longer measurably contribute to ATPTOTAL over the third through fifth minutes of exercise. ATPGLY peaked during the second minute and then decreased, yet still contributed ~20% of the ATPTOTAL during the final 3 min of the exercise. Concomitantly, ATPOX attained maximal values (i.e., not different from Vmax) during the first minute of exercise, which were maintained throughout the remainder of the protocol. In concert, the time course changes in ATPANA and ATPOX resulted in a peak in ATPTOTAL during the first minute of exercise, which progressively decreased thereafter. This is consistent with the suggestion that, during fatiguing exercise, muscle force production declines so that intramuscular [ATP] is maintained within a certain range (21). Additionally, the proportion of ATPTOTAL arising from ATPANA and ATPOX progressively shifted from ~50/50% during the first minute to ~20/80% over the final 3 min of exercise (Fig. 5). These findings provide a novel, comprehensive characterization of skeletal muscle bioenergetics during all-out exercise. Importantly, normalizing ATPTOTAL to force production unveiled a progressive increase in the ATP cost of contraction from the second minute of exercise onward (Fig. 4). For this exercise paradigm, the increased ATP cost of contraction is likely not related to a progressive recruitment of higher-order muscle fibers, as all motor units appear to be recruited at the onset of such all-out exercise (6, 36, 49, 56). Moreover, motor unit recruitment, assessed by electromyography, has been documented to progressively decrease during all-out cycling and knee-extension exercise (6, 56). The findings of the present study also demonstrate that the increased ATP cost of contraction is likely not related to changes in the contractile properties of the muscle fibers, as the change in ATP cost was not related to changes in neuromuscular function (i.e., Qtw, CTQtw, and 0.5RTQtw). Rather, the dramatic decline in MVC force (~63%) in the face of a moderate decrease in ATP synthesis (~40%) lends support to the hypothesis of persistent metabolism in muscle fibers that are contributing less to the overall force production as a result of fatigue. In this regard, it is interesting to note that the ATPANA gain attained a plateau after 3 min of exercise, while the ATPOX gain increased continuously throughout the exercise (Fig. 6). Although it is not possible to definitively determine the metabolic contribution of motor units comprising different muscle fiber types with the present experimental design, these results are consistent with findings in isolated single myocytes that fatigue-resistant muscle fibers exhibit a greater oxidative ATP cost of contraction during fatiguing exercise, whereas more fatigable muscle fibers reduce metabolic demand proportionally to the fall in tension (17). The all-out exercise-induced central and peripheral fatigue and intramuscular metabolic perturbation were closely related in the present study (Fig. 2). Specifically, the reduction in Qtw was correlated to changes in [H2PO4−], [H+], [Pi], and pH. These findings are consistent with a previous report from our group that documented similar relationships during high-intensity cycling exercise (2). Collectively, these findings demonstrate that the link between exercise-induced peripheral fatigue and intramuscular metabolic perturbation is robust across exercise modalities involving quite different amounts of muscle. Mechanistically, the robustness of these relationships provides further, in vivo, evidence of an important role of these metabolites in the development of peripheral fatigue (1). Interestingly, while a low cellular pH or elevated [Pi] has been demonstrated to diminish skeletal muscle function (13, 30), it has recently been demonstrated that these metabolites can also interact synergistically in the development of peripheral fatigue (41). Moreover, the apparent differences in the time course changes between the intramuscular metabolites and force production in the present study further suggest that fatigue development is likely a complex process with synergistic mechanisms. The present data suggest that an intramuscular “V̇o2max” was attained throughout the all-out exercise, as ATPOX values were similar to the peak rate of in vivo oxidative ATP synthesis (i.e., Vmax). This finding is consistent with the attainment of pulmonary V̇o2max during whole body all-out exercise protocols (7, 9, 22, 52, 53). However, such findings, at the muscle level, bolster the use of the all-out exercise protocol as a simple and practical test to determine CF and muscle aerobic capacity in a single test. Interestingly, ATPOX maintained maximal values despite a ~63% reduction in force production in the present study, resulting in the progressive increase in ATPOX gain throughout the exercise (Fig. 6). Importantly, this reveals an intramuscular slow component in the rate of oxidative phosphorylation elicited during the all-out exercise that arose from a progressive loss of muscle efficiency within the exercising skeletal muscle. These findings further support that the V̇o2 slow component arises from the exercising skeletal muscle (45, 47) and that progressive recruitment of higher-order muscle fibers is not requisite to evoke the V̇o2 slow component (18, 52, 56, 59). These findings are also consistent with the expression of a pulmonary V̇o2 slow component during a 3-min all-out cycling test (52, 56). From the outset of discussing the implications for CF, although it is recognized that CP and CF are different, as the implications of assessing both CP and CF are very similar, for clarity, only the term CF will be used here. The growing body of evidence supports that CF is predominantly determined by oxidative energy production (3, 5, 12, 14, 26, 37, 38, 46, 52, 54). This interpretation comes from assessing the alteration in CF as a result of manipulations in oxygen delivery. However, the intramuscular contribution of ATPOX and ATPANA to ATPTOTAL has not been quantified during exercise at CF. In this study, it was demonstrated that ATPOX and ATPANA contributed ~80% and ~20% to ATPTOTAL during exercise at CF (i.e., final 30 s), respectively (Fig. 5). These findings support the concept that CF is predominantly determined by oxidative energy production and explains why manipulating oxygen delivery has such a significant impact on CF. Importantly, however, the findings of this study demonstrate that CF is not fully independent from ATPANA (specifically ATPGLY), which contributed ~20% to ATPTOTAL. In light of the observation that intramuscular metabolites do not progressively change during exercise at or slightly below CF (26, 46, 52), this ~20% ATPANA contribution is likely such that energy utilization and replenishment are in equilibrium and, therefore, no net change occurs. In support of this interpretation, pH was constant across the latter portion of the exercise protocol in the present study, despite the continued ~20% contribution from ATPANA. The quantitative characterization of skeletal muscle bioenergetics during all-out exercise revealed that within this paradigm a small muscle mass exhibits an intramuscular V̇o2 slow component, as well as a progressive increase in the phosphate cost of force generation. Additionally, the intramuscular metabolic perturbation was closely related to the development of peripheral fatigue during the all-out exercise. Collectively, these findings provide direct evidence supporting the concept that the V̇o2 slow component arises from within the active skeletal muscle and that the force generating capacity of skeletal muscle is closely linked to the intramuscular metabolic milieu. GRANTSThis study was supported by National Heart, Lung, and Blood Institute Grants (HL-103786, HL-116579, HL-091830, and K99 HL-125756); Department of Veterans Affairs Rehabilitation Research and Development Merit Awards (E6910-R and E1697-R), SPiRE Grants (E1572-P and E1433-P), and Senior Research Career Scientist Award (E9275-L); and the Flight Attendant Medical Research Institute (YFEL141011). DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSR.M.B., G.L., T.J.H., M.A., and R.S.R. conceived and designed research; R.M.B., G.L., T.J.H., M.A., and R.S.R. performed experiments; R.M.B., G.L., T.J.H., M.A., and R.S.R. analyzed data; R.M.B., G.L., T.J.H., M.A., and R.S.R. interpreted results of experiments; R.M.B., G.L., T.J.H., M.A., and R.S.R. prepared figures; R.M.B., G.L., T.J.H., M.A., and R.S.R. drafted manuscript; R.M.B., G.L., T.J.H., M.A., and R.S.R. edited and revised manuscript; R.M.B., G.L., T.J.H., M.A., and R.S.R. approved final version of manuscript. This work was conducted at the Utah Vascular Research Laboratory housed in the Salt Lake City Department of Veterans Affairs Medical Center. REFERENCES
Page 18AbstractSkeletal muscle contraction results from molecular interactions of myosin “crossbridges” with adjacent actin filament binding sites. The binding of myosin to actin can be “weak” or “strong,” and only strong binding states contribute to force production. During active shortening, the number of strongly bound crossbridges declines with increasing shortening velocity. Forcibly stretching a muscle that is actively shortening at high velocity results in no apparent negative consequences, whereas stretch of an isometrically (fixed-length) contracting muscle causes ultrastructural damage and a decline in force-generating capability. Our working hypothesis is that stretch-induced damage is uniquely attributable to the population of crossbridges that are strongly bound. We tested the hypothesis that stretch-induced force deficits decline as the prevailing shortening velocity is increased. Experiments were performed on permeabilized segments of individual skeletal muscle fibers obtained from human subjects. Fibers were maximally activated and allowed either to generate maximum isometric force (Fo), or to shorten at velocities that resulted in force maintenance of ≈50% Fo or ≈2% Fo. For each test condition, a rapid stretch equivalent to 0.1 × optimal fiber length was applied. Relative to prestretch Fo, force deficits resulting from stretches applied during force maintenance of 100, ≈50, and ≈2% Fo were 23.2 ± 8.6, 7.8 ± 4.2, and 0.3 ± 3.3%, respectively (means ± SD, n = 20). We conclude that stretch-induced damage declines with increasing shortening velocity, consistent with the working hypothesis that the fraction of strongly bound crossbridges is a causative factor in the susceptibility of skeletal muscle to stretch-induced damage. NEW & NOTEWORTHY Force deficits caused by stretch of contracting muscle are most severe when the stretch is applied during an isometric contraction, but prevented if the muscle is shortening at high velocity when the stretch occurs. This study indicates that velocity-controlled modulation of the number of strongly bound crossbridges is the basis for the observed relationship between stretch-induced muscle damage and prevailing shortening velocity. forced lengthening of fully activated, isometrically contracting skeletal muscle disrupts myofibrillar ultrastructure (4, 24, 25, 28, 29, 36) and results in a reduction in force-generating capacity (4, 23–25, 29). The functional deficit increases with the magnitude of the applied strain (4, 23, 25, 29). Such “lengthening contractions” are of interest because they are encountered in common daily activities and can result in functional deficits and delayed-onset pain (1). In contrast, forced lengthening of a fully activated skeletal muscle fiber that is shortening at maximum velocity results in no apparent adverse consequences (2, 3). Instead, activated fibers benefit from periodic rapid shortening-rapid stretch cycles as evidenced by improved striation uniformity and enhanced stability of isometric force production (2). Force generation and fiber shortening result from cyclic interactions between myosin crossbridges extending from myofibrillar thick filaments and actin binding sites on adjacent thin filaments. For a given level of thick and thin filament overlap, force generation is greatest when no shortening is allowed (isometric contraction) and declines according to a fiber type-specific force-velocity relationship as external constraints are removed and shortening occurs (34). The velocity-dependent decline in force that occurs with shortening is the result of a reduction in the number of strongly bound force-producing crossbridges (30), which reaches a minimum when shortening velocity is maximum (35). The contrast between susceptibility to stretch-induced damage at the two extremes of shortening velocity (zero and maximum), coupled with the inverse relationship between shortening velocity and the number of strongly bound crossbridges, provided the rationale for the working hypothesis that stretch-induced damage is a function of the number of strongly bound crossbridges present at the onset of the stretch. To address this working hypothesis, we tested the hypothesis that stretch-induced force deficits decline as prevailing shortening velocity is increased. Single stretches were applied to permeabilized, fully activated human skeletal muscle fibers during maximum isometric force production (force = Fo), during shortening at near-maximum velocity (force near 0), and during shortening at an intermediate velocity that resulted in force maintenance of ≈50% of Fo. Fiber Fo was evaluated before and after the stretch to allow calculation of the force deficit resulting from the stretch. Additional experiments were performed to assess fiber stiffness as a function of shortening velocity and to determine the time course of the fiber transition from a damage-susceptible isometric state to a protected state associated with high-velocity shortening following a sudden release from an isometric contraction. METHODSHealthy, young males (21.0 ± 3.4 yr, mean ± SD, n = 4) underwent medical history and physical examination screening before biopsy. Participants had no evidence of musculoskeletal, neurologic, or cognitive impairment. Written informed consent was obtained from each participant. All procedures were approved by the University of Michigan IRBMED institutional review board and were conducted in accordance with the Declaration of Helsinki. A core needle biopsy of the vastus lateralis (VLT) muscle was taken from the right leg, midthigh, laterally overlying the VLT muscle belly. Following shaving and preparation of the skin with ChloraPrep solution (BD, San Diego, CA), 3 ml of 1% lidocaine with 1:200,000 epinephrine was infiltrated subcutaneously for local anesthesia. Care was taken to avoid injection of the VLT with the local anesthetic. A 5 mm incision was made with a scalpel in the skin and fascia overlying the biopsy site. Approximately 50 mg of VLT were obtained atraumatically with a University College Hospital (London, UK) skeletal muscle core biopsy needle inserted through the incision and into the muscle. The skin was then closed with DermaBond (Johnson & Johnson, New Brunswick, NJ). The muscle sample was removed quickly from the biopsy needle and placed immediately into cold skinning solution. Bundles of fiber segments ~4–6 mm in length and 0.5 mm in diameter were dissected from the samples. Following dissection, bundles were immersed for 30 min in skinning solution to which the nonionic detergent Brij 58 had been added (0.5% wt/vol). The bundles of permeabilized fiber segments were then placed in storage solution and maintained for 12–16 h at 4°C followed by storage at −80°C for up to 3 mo before use. On the day of an experiment, a bundle was removed from storage solution and placed in relaxing solution on ice. Individual permeabilized fiber segments (hereafter referred to as “fibers”) were pulled manually from the bundle with fine forceps and transferred to an experimental chamber containing relaxing solution maintained at 15°C. All measurements were performed exclusively on type 1 (“slow”) fibers, sourced uniformly from among the four subjects. The type 1 fibers were distinguished from type 2 (“fast”) fibers by observing visual cues during fiber isolation (reduced transparency, absence of sarcolemma festooning) and performing an abbreviated force-velocity test (7). Single fiber contractility experiments were performed with techniques that have been described in detail previously (7, 8, 32). Briefly, one end of the fiber was secured to a force transducer (model 403A; Aurora Scientific, Aurora, ON, Canada) and the other to the lever arm of a servomotor (model 322C, Aurora Scientific) while immersed in relaxing solution. The solution-changing system (model 802D, Aurora Scientific) consisted of eight separate glass-bottom chambers machined into a moveable, temperature-controlled aluminum plate coated with an inert, thermally conductive material (Tufram). Movement of the plate with respect to the fiber was achieved by remote-control of two stepper motors, one to lower and raise the chamber array and the other to translate the plate to a new chamber position. For all experiments reported here, the length of the relaxed fiber was adjusted to obtain an average sarcomere length of 2.7 µm, determined by projecting a laser diffraction pattern produced by the fiber onto a calibrated target screen. After setting sarcomere length, fiber length (Lf) and cross-sectional area (CSA) were measured. An activation-relaxation sequence was accomplished by transferring the fiber from the chamber containing relaxing solution to a chamber containing preactivating solution for 3 min and then immersing it in a chamber containing activating solution (pCa ≈4.5) to elicit Fo. The fiber was then relaxed by returning it to the chamber containing relaxing solution. Specific force was calculated for each fiber as Fo/CSA. Fiber stiffness was assessed in a separate series of experiments in which very small (0.002 Lf) stretches were applied at a constant velocity of 0.5 Lf/s and the force responses recorded (11, 18). Stiffness was calculated as ΔF/ΔL, where ΔF was the difference between the force levels at the onset and end of the stretch, and ΔL was 0.002 Lf. Stiffness-probing stretches were applied: 1) immediately after application of a 0.1 Lf shortening movement at high velocity (0.7 Lf/s), 2) immediately after application of a 0.1 Lf shortening movement at intermediate velocity (0.02 Lf/s), and 3) during a plateau in isometric force. Thus the stretches were applied from prevailing force levels of ≈0, ≈50, and 100% Fo, respectively, all from a fiber length of 1.0 Lf. Because the amplitude of the stretch employed to assess stiffness was well below that required to produce force deficits (23, 25, 29), it was possible to obtain all three stiffness measurements from each fiber in the series. A representative sequence of stiffness assessments is illustrated in Fig. 1. Fig. 1.Stiffness measurement protocol. A: representative force and length records are shown at low resolution for an entire activation-relaxation sequence during which 3 stiffness-probing stretches of ΔL = 0.002 Lf were applied. The length perturbations visible in the sequence are: a large (0.2 Lf) slack-inducing step release in fiber length (inserted to indicate the location of 0 force in the experimental records), maintained for 30 ms before rapidly (within 1 ms) returning the fiber to its original length (a); shortening by 0.1 Lf at high velocity (0.7 Lf/s) followed by a slack-inducing (0.1 Lf) step release maintained for 30 ms before returning the fiber to its original length (b); and shortening by 0.1 Lf at moderate velocity (0.02 Lf/s) followed by a slack-inducing (0.1 Lf) step release maintained for 30 ms before returning the fiber to 1.0 Lf (c). The purpose of the 30 ms periods of unloaded shortening was to ensure that the fiber was not damaged during the subsequent returns to 1.1 Lf (b) and 1.0 Lf (c). The final event in the sequence (d) was a 0.002 Lf stretch applied to the isometrically contracting fiber. Note that, with the exception of the force response in d, the 0.002 Lf stretches and corresponding force responses are not visible at these time and amplitude resolutions. B: details of the stiffness-probing stretches and force responses on expanded time and amplitude scales. In all cases the stretch amplitude was 0.002 Lf, completed within 4 ms (0.5 Lf/s) and held for 4 ms. For the stiffness measurements during shortening at 0.7 Lf/s and 0.02 Lf/s, the stretch was applied at the end of the period of controlled shortening and the slack-inducing step release was applied after the 4 ms hold period that followed the stretch. The force calibration bar shown near the center force record applies to all 3 force responses. The time calibration bar shown near the center fiber-length record applies to all 6 records. Representative measurements of ΔF and ΔL are shown for the isometric case (right-most force and length records). Note that all 3 stiffness-probing stretches were applied in the absence of fiber slack from a fiber length of 1.0 Lf. Lf, fiber length; ΔF, the difference between the force levels at the onset and end of the stretch; ΔL, 0.002 Lf. The relationship between prevailing shortening velocity and stretch-induced force deficits was determined for a series of muscle fibers using three separate consecutive maximum activation-relaxation cycles. The first and third “bracketing” activations were identical in all respects and established pre- and postintervention Fo from which force deficits (% of preintervention Fo) were calculated (4). Both were initiated from a resting sarcomere length of approximately 2.7 µm and sustained for 20 s before returning to relaxing solution. For both bracketing activations, the fiber was maintained at a constant length corresponding to Lf (“isometric”) with the exception of a brief period midcontraction during which a rapid shortening movement was applied and then removed 30 ms later. The shortening movement was sufficient in size to slacken the fiber, thereby indicating on the experimental record the force transducer output level corresponding to zero force (7, 8, 32). For the second activation, the fiber was treated exactly as in the bracketing activations except that, instead of relaxing the fiber after 20 s of activation, one of three different stretch protocols was applied. The protocols were: 1) application of a 0.1 Lf shortening movement at high velocity (0.7 Lf/s) just before a single stretch (Fig. 2A), 2) application of a 0.1 Lf shortening movement at intermediate velocity (0.02 Lf/s) just before the stretch (Fig. 2B), or 3) performing the 0.1 Lf stretch on an isometrically contracting fiber (Fig. 2C). Thus, the stretch was applied from prevailing force levels of ≈0, ≈50, and 100% Fo, respectively. In all cases, the stretch movement of 0.1 Lf was complete within 1 ms and was held for 4 ms before the fiber was returned to its prestretch length. For the protocols in which stretches were applied after the fiber had shortened by 0.1 Lf, activation was initiated from a fiber length of 1.1 Lf, thereby ensuring that, for all protocols, stretches were applied from a consistent fiber length of 1.0 Lf. In addition, for the protocol that required application of stretch after high-velocity shortening (0.7 Lf/s), care was taken to confirm that the prevailing fiber force was >0 at the time the stretch was applied (i.e., the fiber was not slack). Separate fibers were used for each stretch-induced force deficit determination in this series of experiments. Fig. 2.Protocol for determining the relationship between prevailing shortening velocity and stretch-induced force deficits. Representative experimental records are shown for stretch applied during high-velocity shortening (0.7 Lf/s) (A), during shortening at moderate velocity (0.02 Lf/s) (B), and to an isometrically contracting fiber (C). The paired force and length records on the left show the entire activation-relaxation sequence. Arrows indicate the times at which the stretches were applied. The paired force and length records on the right show the corresponding details of the stretch event on an expanded time scale. In all cases the stretch was 0.1 Lf in amplitude, completed within 1 ms and held for 4 ms before the return to prestretch length. The force calibration bar shown in A applies to all force records. The 2 s time calibration bar shown in A applies to all long time-base records on the left, and the 10 ms time calibration bar shown in A applies to all expanded time-base records on the right. The sudden decline in force evident in the long time-base records 10 s after the onset of force was caused by a large (0.2 Lf) slack-inducing step release in fiber length, inserted to indicate the location of zero force in the experimental records. The slack was maintained for 30 ms before returning the fiber to its original length with a step stretch. We next probed the time course of the transition of fibers from the isometric state, which is susceptible to stretch-induced force deficits, to the high-velocity shortening state, which is protected from stretch-induced force deficits. This was done in a series of experiments in which a stretch sufficient in size to produce force deficits in an isometrically contracting fiber was applied to fibers at various times after a step-release (complete within 1 ms) from the isometric state to the unloaded shortening (high-velocity) state. The stretch extended fiber length by 0.1 Lf relative to the prerelease length, was complete within 1 ms, and was held for 4 ms before the fiber was returned to its original length. The intervals between release and stretch were 5, 10, and 20 ms. As with the experiments that examined the relationship between shortening velocity and stretch-induced force deficits, bracketing isometric contractions served to monitor any decline in force attributable to the stretch. Note that, because some shortening took place during the 5, 10, or 20 ms of high-velocity shortening that preceded the stretch, the effective stretches were slightly greater than the nominal 0.1 Lf and increased as the duration of shortening increased. Specifically, based upon typical unloaded shortening velocities for human type 1 VLT fibers at 15°C (7), we calculate that effective stretches were approximately 0.105, 0.11, and 0.12 Lf following 5, 10, and 20 ms of high-velocity shortening, respectively. Representative records illustrating the transition time experiments are shown in Fig. 3. Also included for comparison is a response to a stretch applied to an isometrically contracting fiber (Fig. 3A). Fig. 3.Protocol for determining the time course of the transition from isometric state to high-velocity shortening state. Representative experimental records are shown for stretch applied to an isometrically contracting fiber (A) and to a slack fiber after 5 ms (B), 10 ms (C), and 20 ms (D) of unloaded shortening. The paired force and length records on the left show the entire activation-relaxation sequence. Arrows indicate the times at which the stretches were applied. The paired force and length records on the right show the corresponding details of the stretch event on an expanded time scale. In all cases the stretch was nominally 0.1 Lf in amplitude, completed within 1 ms and held for 4 ms before returning the fiber to its original length. The effective stretch amplitudes were slightly greater than 0.1 Lf, and increased with increasing slack time (see text for explanation). The force calibration bar shown in A applies to all force records. The 2 s time calibration bar shown in A applies to all long time-base records on the left, and the 10 ms time calibration bar shown in A applies to all expanded time-base records on the right. The sudden decline in force evident in the long time-base records 10 s after the onset of force was caused by a large (0.2 Lf) slack-inducing step release in fiber length, inserted to indicate the location of zero force in the experimental records. The slack was maintained for 30 ms before returning the fiber to its original length with a step stretch. Note that the records shown in A are the same as those shown in Fig. 2C. The skinning solution was composed of the following (in mM): 125 potassium propionate, 20 imidazole, 5 ethyleneglycol-bis (B-aminoethyl ether) tetraacetic acid, 2 MgCl2, 2 ATP; pH 7.0. The composition of the storage solution was identical to skinning solution with the exception that glycerol was substituted for 50% of the water volume. The relaxing, preactivating, and activating solutions were modified from Moisescu and Thieleczek (26) and have been described in detail previously (7, 8, 32). Potassium propionate was obtained from TCI America, and all other compounds were obtained from Sigma Chemical. Results are presented as means ± standard deviation (SD). Statistical analyses were performed using JMP software (SAS Institute, Cary, NC) and included one-way analysis of variance (ANOVA) and Student’s t-test, as indicated in the figure legends. For tests in which the ANOVA indicated significance, individual differences were determined by a Tukey’s honest significant difference post hoc test. The level of significance was set a priori at P < 0.05. RESULTSA total of 152 fibers were studied. Mean (± SD) Lf, CSA, and specific force were 1.46 ± 0.20 mm, 6,900 ± 2,730 µm2, and 152 ± 30 kPa. Fiber shortening velocity was controlled to modulate the number of strongly bound, force-producing crossbridges indicated, in turn, by the force maintained at the three velocities employed (30). During the fiber stiffness experiments (illustrated in Fig. 1) and force deficit experiments (illustrated in Fig. 2), fibers were allowed to shorten from 1.1 to 1.0 Lf at controlled velocities of 0.7 and 0.02 Lf/s before a stretch was applied. Forces recorded just before stretch were F/Fo = 0.019 ± 0.005 when shortening velocity was 0.7 Lf/s and F/Fo = 0.525 ± 0.033 when shortening velocity was 0.02 Lf/s (n = 32). During the same two series of experiments, isometric force (shortening velocity = 0 Lf/s) recorded just before stretch was F/Fo = 1.01 ± 0.01 (n = 32). The three force-velocity pairs resulting from these experiments are plotted in Fig. 4 to show the effect of shortening velocity on the ability of the fibers to maintain force. The continuous line in Fig. 4 is a rectangular hyperbola (15, 34) fitted (least-squares regression) to the means of the force values at velocities of 0.7 Lf/s, 0.02 Lf/s, and 0 Lf/s (isometric). The parameters of fit were a/Fo = 0.012, b = 0.022 Lf/s, Fo′/Fo = 1.01, and Vmax = 1.85 Lf/s, where a and b are the force and velocity asymptotes, respectively, Fo′ is the intercept with the force axis, and Vmax is the intercept with the velocity axis (7). The fitted curve is in good agreement with previously reported force-velocity results in which each curve was defined by 8–12 force-velocity data pairs obtained from a much larger number of human VLT type 1 fibers (7). Fig. 4.Relationship between shortening velocity and force maintained during shortening. Force was measured just before stretch during experiments of the types illustrated in Figs. 1 and 2. All forces are expressed relative to maximum isometric force (Fo) as determined during the preceding bracketing activation (means ± SD, n = 32, error bars omitted where they fall within the mean symbols). For the velocity > 0 measurements, ramp shortening was initiated from a fiber length of 1.1 Lf and force was measured at the time the fiber length reached 1.0 Lf. The continuous line is a rectangular hyperbola fitted to the means of the measured force values. The stiffness of a calcium-activated muscle fiber increases with the number of strongly bound crossbridges (11, 18, 19). At a shortening velocity of 0.02 Lf/s, fiber force and stiffness were 53.4 ± 3.3 and 59.1 ± 2.4% of their respective maximum isometric levels (n = 12). Further increasing shortening velocity to 0.7 Lf/s caused fiber force and stiffness to decline to 1.8 ± 0.4 and 11.3 ± 0.1% of their respective maximum isometric levels. The decline in fiber stiffness with increasing shortening velocity was linearly related to the decline in force maintained during shortening as shortening velocity was increased (Fig. 5A). Fig. 5.Fiber stiffness and force deficits as a function of prevailing force. Measurements were made during shortening at 0.7 Lf/s (near maximum velocity of shortening and near zero force maintenance), at 0.02 Lf/s (≈50% force maintenance), and during isometric contractions (0 Lf/s, 100% force maintenance). x-Axes are force at time of stiffness or deficit measurement relative to maximum isometric force (Fo) as determined during the preceding bracketing activation. Velocity of shortening at the time of each measurement is labeled. A: stiffness. Stiffness [(ΔF/Fo)/(ΔL/Lf)] was calculated at all 3 velocities (0.7, 0.02, 0 Lf/s) for each of 12 fibers, and mean isometric (0 Lf/s) stiffness was used to normalize all 36 stiffness measurements. Symbols are means ± SD of the normalized values, n = 12 (error bars omitted when they fall within the mean symbols). Line is least-squares fit to all data (R2 = 0.99, slope = 0.90, intercept = 0.12). All 3 stiffness means are different from each other (1-way ANOVA, Tukey’s honest significant difference (HSD) post hoc, P < 0.0001). The stiffness experiments are illustrated in Fig. 1. B: force deficits caused by 0.1 Lf step stretch. Symbols are means ± SD, n = 20 (error bars omitted when they fall within the mean symbols). Line is least-squares fit to all data (R2 = 0.72, slope = 23%, intercept = −1.5%). *Different from isometric; ‡different from results obtained at shortening velocity of 0.02 Lf/s (1-way ANOVA, Tukey’s HSD post hoc, P < 0.001). The possibility that stretch applied during high-velocity shortening caused no deficit could not be rejected (Student’s t-test, P = 0.66). The force deficit experiments are illustrated in Fig. 2. The deficit in maximum force generating capacity was greatest (23.2 ± 8.6%, n = 20) when the 0.1 × Lf stretch was applied during the plateau of a maximum isometric contraction. When the stretch was applied to fibers shortening at 0.02 Lf/s and maintaining a force level of 51.9 ± 3.3% of Fo, the resulting force deficit fell to 7.8 ± 4.2%. Further increasing shortening velocity to 0.7 Lf/s, which is near maximum for type 1 fibers from human VLT at 15°C, resulted in very little force maintenance (2.0 ± 0.6% of Fo) and no force deficit (0.3 ± 3.3%, P = 0.66). Peak forces attained during stretch from isometric, ≈50% isometric, and ≈2% isometric were F/Fo = 1.59 ± 0.18, 0.99 ± 0.06, and 0.39 ± 0.11, respectively. The relationship between force deficit and prevailing force at the time of stretch is shown in Fig. 5B. Stretches applied to isometrically contracting fibers sustained significant force deficits, but fibers that had shortened by 0.1 Lf at a velocity of 0.7 Lf/s were unaffected by stretch (Figs. 2A and 5B). The elapsed time between the onset of shortening and the stretch in the experiments illustrated in Fig. 2A was 143 ms. The step release–step stretch experiments illustrated in Fig. 3 were designed to probe the time course of the transition from the deficit-susceptible isometric state to the protected state associated with high-velocity shortening. The relationship between force deficit and the delay between step release and step stretch (Fig. 6) indicates that susceptibility to stretch-induced force deficits declines rapidly following the onset of high-velocity shortening. The relationship is well described by a single exponential with a time constant of 5.8 ms (Fig. 6, dashed line). Fig. 6.Force deficits as a function of duration of unloaded shortening before stretch. All stretches were to a final fiber length of 1.1 Lf. With the exception of the t = 0 ms time point (no prior shortening), the stretch was preceded by unloaded shortening initiated by applying a step release of 0.2 Lf from an initial length of 1.0 Lf during the plateau phase of a maximal isometric contraction. x-Axis is duration of the unloaded shortening that immediately preceded the stretch. Dashed line is least-squares fit of a single exponential to median force deficits at each time point (R2 = 0.94; deficit at t = 0, 26.8%; time constant, 5.8 ms). All 4 means are significantly different from each other (1-way ANOVA, Tukey’s HSD post hoc, P < 0.02, n = 20). For t = 0, 5 and 10 ms, mean force deficit was >0 (Student’s t-test, P < 0.001). For t = 20 ms, the possibility that the mean force deficit was zero could not be rejected (Student’s t-test, P = 0.23). These experiments are illustrated in Fig. 3. DISCUSSIONStretch of isometrically contracting skeletal muscle causes ultrastructural damage and functional deficits that scale with the strain imposed and the energy absorbed during the stretch (4). In contrast, high-velocity, high-strain stretches applied to muscle fibers that are shortening at maximum velocity cause no functional deficits (2). Force maintained during shortening is proportional to the number of strongly bound crossbridges (30). Accordingly, the force-velocity relationship of a muscle fiber can be viewed as the relationship between its shortening velocity and the number of strongly bound crossbridges. In the present study, fiber shortening velocity was controlled before stretch to produce one of three levels of force corresponding to a maximum, intermediate, and minimum number of strongly bound crossbridges. The observation that the force deficits resulting from stretches imposed under these three conditions scaled with the prevailing force at the onset of the stretch is consistent with our working hypothesis that the susceptibility of skeletal muscle fibers to stretch-induced damage is a function of the number of strongly bound, force-producing crossbridges present at the time that the stretch is applied. The stiffness of a calcium-activated muscle fiber increases as the number of strongly bound crossbridges increases (11, 18, 19), but multiple additional factors influence crossbridge-related fiber stiffness measurements. These include confounding contributions from sarcomeric actin and myosin filaments (5, 13, 20). In addition, a “weakly bound” crossbridge state has been identified that exists in the presence of high levels of calcium, contributes to fiber stiffness but not force, and has a detachment rate that is estimated to be 100-fold higher than that of strongly bound crossbridges (35). During shortening at maximum velocity, the number of strongly bound crossbridges falls to near zero and weakly bound crossbridges are the principal source of crossbridge stiffness (35). The weakly bound state present in activated fibers during high-velocity shortening is unique in that its kinetics are intermediate between those of the weakly bound state in relaxed fibers (10) and strongly bound force-producing crossbridges found in activated fibers. Consequently, stiffness-probing strains applied at strain rates typically employed to detect the presence of strongly bound crossbridges also produce a contribution from the population of weakly bound crossbridges in activated fibers, with the fractional contribution of the weakly bound crossbridges increasing as the strain rate is increased (35). Such stiffness measurements are thus influenced by both crossbridge binding states, with the weakly bound state contributing a strain rate-dependent fraction that constitutes the dominant source of crossbridge stiffness at the highest shortening velocities when force is near zero (35). An alternative interpretation is that the crossbridge-related stiffness measured during shortening at very low loads is sensing a population of “constitutively on” crossbridges that is postulated to be responsible for high-velocity shortening when stress in the thick filament is low (22). In addition to actin-myosin interactions, passive cytoskeletal elements also contribute to the total stiffness of a muscle fiber (31). In the absence of calcium (i.e., in passive fibers), strong crossbridge attachment is effectively absent and titin is the predominant contributor to fiber stiffness at the sarcomere lengths used in the present study (16, 31). In the presence of the high calcium levels associated with muscle activation, the stiffness of titin is thought to increase significantly (9, 14, 21) and likely accounts for some of the residual stiffness observed during high-velocity shortening when force maintenance is near zero (Fig. 5A). Thus both noncrossbridge and crossbridge sources are proposed to contribute to fiber stiffness during high-velocity shortening. Regardless of its source, the stiffness present during high-velocity shortening did not render fibers susceptible to stretch-induced force deficits. As shortening velocity was reduced from maximum to zero (isometric), force (Fig. 4) and stiffness (Fig. 5A) increased. We attribute the increases to growth of the population of isometric-type, strongly bound, force-producing crossbridges (35). Our interpretation that increases in fiber force and stiffness indicate a growing population of strongly bound force-producing crossbridges, coupled with our working hypothesis that stretch-induced damage scales with the number of strongly bound crossbridges, leads to the prediction that stretch-induced force deficits increase with decreasing shortening velocity. The findings of the present study largely support this prediction. At a shortening velocity near maximum (0.7 Lf/s), the force maintained during shortening was approximately 2% of maximum isometric force, stiffness was approximately 11% of maximum isometric stiffness, and there was no detectable force deficit in response to stretch. The same stretch applied to isometrically contracting fibers under conditions of maximum force and stiffness resulted in the greatest deficit in force (23.2%). During shortening at an intermediate velocity (0.02 Lf/s) that resulted in maintenance of approximately 50% of maximum isometric force and stiffness, the stretch resulted in a force deficit that was intermediate between those observed at the two extremes of the force-velocity relationship. The increases in force deficits with decreasing shortening velocity were also associated with a rise in the peak forces attained during stretch, in general agreement with previous studies showing that force levels during stretch and work done during stretch are the best predictors of force deficits (4). It should be noted, however, that peak forces did not exceed Fo during deficit-producing stretches initiated while shortening at the intermediate velocity of 0.02 Lf/s. Since peak forces below Fo are not normally associated with stretch-induced force deficits, this result suggests that, under some circumstances, force and work can be uncoupled as factors that contribute to muscle damage. Huxley (17) proposed that a significant number of strongly bound crossbridges are present during unloaded shortening and include a population that is positively strained and contributing to force and a population that is negatively strained and exerting drag, resulting in no net external force. Stiffness measured during unloaded shortening has subsequently been interpreted as a measure of both the positively and negatively strained, strongly bound crossbridges (12). The absence of force deficits in response to stretch during shortening at high velocity despite the presence of significant fiber stiffness (11% of maximum isometric stiffness) indicates that the source of the high-velocity stiffness does not contribute to stretch-induced deficits. This finding, coupled with an interpretation that stiffness measured during unloaded shortening indicates the presence of strongly bound crossbridges, contradicts our working hypothesis that attributes force deficits to strain of strongly bound crossbridges. The apparent contradiction is resolved by a proposed weakly bound crossbridge state that is predominant during shortening at high velocities (35) and is observable via standard stiffness measurement techniques but does not contribute to stretch-induced damage in skeletal muscle. The resistance of the weak binding state to stretch-induced force deficits could be attributable to the 100-fold increase in detachment rate compared with strongly bound crossbridges (35), allowing detachment before strain-related damage occurs. Similarly, the stiffness measured during shortening at very low loads could be sensing a population of constitutively active crossbridges (22) with kinetics that allow detection using stiffness-sensing maneuvers but protect the fiber from stretch-induced deficits. The hyperbolic shape of the force-velocity relationship of skeletal muscle (15, 34) can be roughly partitioned into a low force region within which large changes in velocity result in relatively small changes in force, a region of intermediate forces, and a high force region within which force is much more sensitive to changes in velocity. The sensitivities of the low and high force regions determine the curvature of the force-velocity relationship, a characteristic that is quantified by the parameter a/Fo, determined by fitting a rectangular hyperbola to force-velocity measurements (15, 34). Slow (type 1) fibers exhibit greater curvature (smaller a/Fo) than fast (type 2) fibers (33), and the force-velocity relationship of human type 1 fibers has a particularly severe curvature (7, 37) (see Fig. 4). The extreme sensitivity of force to velocity at high forces is illustrated by our finding that even a very small increase in shortening velocity from 0 to 0.02 Lf/s (≈2% of maximum, unloaded velocity) results in a large decline in force from 100% Fo to approximately 50% Fo. Although human type 1 fibers represent an exceptional case, the high sensitivity of force to shortening at forces near isometric is a feature of mammalian force-velocity relationships in general and human fibers in particular (37). The high sensitivity has implications for stretch-induced muscle damage in that even small, low-velocity shortening movements preceding a stretch have the potential to mitigate stretch-induced damage. Such movements likely occur in vivo as an active muscle-tendon unit transmits force. That is, even during an activity that is externally isometric, muscle undergoes some shortening at the expense of a lengthening tendon as the force generated by the muscle is transmitted to the skeleton. Any protection afforded by prior shortening against a compliant tendon would likely be diminished with age as tendon compliance is reduced (38, 39). The transition of crossbridges from the strongly bound, isometric type that are susceptible to stretch-induced damage to the protected state associated with high-velocity shortening is not an instantaneous event coincident with the onset of shortening. If stretch-induced force deficits can be taken as an indicator of the presence of strongly bound crossbridges, the results shown in Fig. 6 suggest that, for human type 1 fibers at 15°C, the shift in the crossbridge population from the isometric state to the high velocity-shortening state takes nearly 20 ms to complete. An exponential fit to the relationship between force deficits and duration of unloaded shortening (dashed line, Fig. 6) indicates a time constant of 5.8 ms for the transition. The exponential fall in deficit susceptibility is qualitatively similar to the decline in the number of strongly bound crossbridges following an abrupt transition from isometric conditions to unloaded shortening in intact frog fibers (22), in further support of a connection between force deficits and crossbridge attachment state. The finding that the deficit-susceptible isometric crossbridge state requires time to transition to a protected state during unloaded shortening is of some practical significance for the execution of the periodic rapid shortening–rapid stretch cycles that are frequently employed in experiments on permeabilized fibers (2). If the rapid shortening phase that precedes restretch is too brief (e.g., <20 ms for human type 1 fibers at 15°C), fiber damage might be incurred during the return to original length. The damage-susceptibility time course reported here could also have implications for muscle function in vivo. At physiological temperatures, the rate of transition from an injury-susceptible state to a protected state is likely faster than that observed in the present study (performed at 15°C), enhancing the impact of this protective effect. Similarly, the faster crossbridge kinetics of type 2 fibers would be expected to result in a more rapid transition to a protected state following the onset of shortening. We controlled shortening velocity in fully activated human skeletal muscle fibers to modulate the number of strongly bound crossbridges. Stiffness measurements were consistent with the existence of a population of crossbridges during high-velocity shortening (22, 35) that do not contribute to stretch-induced damage. Stiffness measurements coupled with force measurements were consistent with the recruitment of strongly bound crossbridges in numbers proportional to force as shortening velocity was reduced from maximum to intermediate to 0 (isometric). Stretches were superimposed during shortening to determine the relationship between the fraction of strongly bound crossbridges and damage susceptibility. Force deficits caused by stretch were approximately proportional to prevailing force at the time of the stretch, in support of the working hypothesis that stretch-induced damage is a function of the number of crossbridges that are strongly bound at the onset of stretch. Susceptibility to stretch-induced force deficits was also determined during the transition from isometric to unloaded shortening states and found to decay with a time constant of approximately 5.8 ms. Since force during stretch is also related to the number of strongly bound crossbridges at the time of stretch (6, 27), these findings complement previous reports that demonstrate a link between force deficits, force during stretch, and work absorbed during stretch (4). GRANTSThis work was supported by National Institutes of Health Grants R01-AG-050676, R01-AR-063649 and F32-AR-067086. A. L. Saripalli was supported by the University of Michigan Student Biomedical Research Program. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSA.L.S. and D.R.C. conceived and designed research; A.L.S. performed experiments; A.L.S., S.V.B., and D.R.C. analyzed data; A.L.S., K.B.S., C.L.M., S.V.B., and D.R.C. interpreted results of experiments; A.L.S. and D.R.C. prepared figures; A.L.S., S.V.B., and D.R.C. drafted manuscript; A.L.S., K.B.S., C.L.M., S.V.B., and D.R.C. edited and revised manuscript; A.L.S., K.B.S., C.L.M., S.V.B., and D.R.C. approved final version of manuscript. REFERENCES
Page 19AbstractHemoglobin-based oxygen carrier (HBOC)-201 is a cell-free modified hemoglobin solution potentially facilitating oxygen uptake and delivery in cardiovascular disorders and hemorrhagic shock. Clinical use has been hampered by vasoconstriction in the systemic and pulmonary beds. Therefore, we aimed to 1) determine the possibility of counteracting HBOC-201-induced pressor effects with either adenosine (ADO) or nitroglycerin (NTG); 2) assess the potential roles of nitric oxide (NO) scavenging, reactive oxygen species (ROS), and endothelin (ET) in mediating the observed vasoconstriction; and 3) compare these effects in resting and exercising swine. Chronically instrumented swine were studied at rest and during exercise after administration of HBOC-201 alone or in combination with ADO. The role of NO was assessed by supplementation with NTG or administration of the eNOS inhibitor Nω-nitro-l-arginine. Alternative vasoactive pathways were investigated via intravenous administration of the ETA/ETB receptor blocker tezosentan or a mixture of ROS scavengers. The systemic and to a lesser extent the pulmonary pressor effects of HBOC-201 could be counteracted by ADO; however, dosage titration was very important to avoid systemic hypotension. Similarly, supplementation of NO with NTG negated the pressor effects but also required titration of the dose. The pressor response to HBOC-201 was reduced after eNOS inhibition and abolished by simultaneous ETA/ETB receptor blockade, while ROS scavenging had no effect. In conclusion, the pressor response to HBOC-201 is mediated by vasoconstriction due to NO scavenging and production of ET. Further research should explore the effect of longer-acting ET receptor blockers to counteract the side effect of hemoglobin-based oxygen carriers. NEW & NOTEWORTHY Hemoglobin-based oxygen carrier (HBOC)-201 can disrupt hemodynamic homeostasis, mimicking some aspects of endothelial dysfunction, resulting in elevated systemic and pulmonary blood pressures. HBOC-201-induced vasoconstriction is mediated by scavenging nitric oxide (NO) and by upregulating endothelin (ET) production. Pressor effects can be prevented by adjuvant treatment with NO donors or direct vasodilators, such as nitroglycerin or adenosine, but dosages must be carefully monitored to avoid hypotension. However, hemodynamic normalization is more easily achieved via administration of an ET receptor blocker. hemoglobin-based oxygen carrier (HBOC)-201 is a cell- and endotoxin-free, glutaraldehyde-polymerized hemoglobin solution produced by chemical modification of hemoglobin extracted from isolated bovine red blood cells (25). HBOCs may be used in the treatment of cardiovascular disorders, and hemorrhagic shock, in particular; however, side effects include systemic and pulmonary blood pressure elevations, plasma volume expansion, lower cardiac output and reduction in heart rate (24, 39, 41, 53). Despite these potentially unfavorable effects, studies in human subjects with documented coronary disease showed that HBOC-201 had no effect on left ventricular (LV) stroke work index or any of the measured coronary function parameters (53). The most important HBOC-201 side effect is systemic and pulmonary vasoconstriction (24, 49). Consequently, this study first aimed to determine the possibility of reversing the HBOC-201 pressor effects via simultaneous administration of adenosine (ADO), a nitric oxide (NO)-independent vasodilator, or the NO-donor nitroglycerin (NTG). The pressor effect of HBOC-201 has been ascribed to scavenging of NO, an important endogenous vascular relaxing factor (5, 6, 18, 44, 45, 60). Free hemoglobin (Hb) undergoes rapid (~107 M−1·s−1) (19, 23) and irreversible reaction with NO to form metHb, where Hb kinetically behaves as a dioxygenase enzyme (1, 12). In the following slower processes, iron-NO complexes are formed that may further deplete NO concentrations (1). However, although disruption of the NO-mediated cascade may be an important contributor to transient systemic and pulmonary hypertension, it is not the only possible pathway. Oversupplying oxygen (O2) can stimulate vasoconstriction, to protect against the oxygen burst, but can also stimulate reactive oxygen species (ROS) formation that may result in further scavenging of NO (33, 61). By scavenging NO, conversion from pro-endothelin to endothelin (ET) is no longer inhibited, thereby increasing the release of this vasoconstrictor (3, 28, 52). Also, free radicals generated by the auto-oxidation of hemoglobin may contribute to the enhanced release of ET (59). Therefore, the second aim of this study was to address the potential roles of NO scavenging, ROS, and/or endothelin in the HBOC-201 systemic and pulmonary pressor effects. NO has been shown to contribute to exercise-induced vasodilation in skeletal muscle, the heart, as well as the pulmonary vasculature in many (9, 22, 51) but not all studies (13). A state of decreased NO and increased ROS and ET production resembles some aspects of endothelial dysfunction, a phenomenon that may have exaggerated effects during exercise. Hence, the third aim of this study compared the effects of HBOC-201 in resting and exercising swine. METHODSStudies were performed in accordance with the Council of Europe Convention (ETS123)/Directive (86/609/EEC) for the protection of vertebrate animals used for experimental and other scientific purposes and with approval of the Animal Care Committee at Erasmus MC, University Medical Centre Rotterdam. A total of 16 Yorkshire × Landrace swine (2–3 mo old, 22 ± 1 kg at the time of surgery) of either sex (11 female and 5 male) entered the study. After completing all experimental protocols, animals were euthanized by an intravenous overdose of pentobarbital sodium. Detailed surgical procedures have been previously described (8, 37). In brief, under deep anesthesia, a thoracotomy was performed in the fourth left intercostal space. Fluid-filled catheters were placed in the aorta, pulmonary artery, left atrium, and LV for measurement of pressure, infusion of drugs and blood sampling. In addition, a flow probe (Transonic Systems) was placed around the ascending aorta for measurement of cardiac output. All catheters were exteriorized at the back of the animal and filled with heparinized saline. The thorax was closed in layers, and the animal was allowed to recover for at least one week. Antibiotic prophylaxis (amoxicillin, 25 mg/kg iv) was provided for 5–7 days starting immediately before surgery. Immediate postoperative analgesia was provided by buprenorphine (0.015 mg/kg im), while a slow-release fentanyl patch (12 μg/h) maintained postoperative analgesia for 72 h. Studies were performed 1–3 wk after surgery, with animals resting and exercising on a motor-driven treadmill up to 85–90% of maximal heart rate. Three main protocols (as described below) were performed on different days and in random order. All chemicals were obtained from Sigma, and HBOC-201 (13 g/dl) was obtained from OPK Biotech. With swine lying on the treadmill, resting hemodynamic measurements consisting of heart rate (HR), LV pressure, first derivative of LV pressure (dP/dt), mean aortic pressure (MAP), pulmonary artery pressure (PAP), left atrial pressure, and cardiac output were obtained. Subsequenty, swine were subjected to a five-stage exercise protocol (1–5 km/h) while hemodynamic variables were continuously recorded, and blood samples were collected during the last 60 s of each 3-min exercise stage at a time when hemodynamics had reached a steady state. Blood samples were used for determination of Hb, oxygen content, and lactate using an automated blood gas analyzer (ABL800; Radiometer). After the exercise protocol was completed, animals were allowed to rest on the treadmill for 90 min, after which HBOC-201 (10 ml/kg iv) was infused over a period of 30 min. At the end of infusion, the exercise protocol was repeated. We have previously shown excellent reproducibility of the hemodynamic response in consecutive bouts of exercise (9). Also, in three pigs, we assessed the reproducibility of the hemodynamic responses to HBOC-201 infusion by administration of three separate doses of HBOC-201 (10 ml/kg iv), separated by 5 ± 1 days. After performing a control run, six animals received HBOC-201 combined with the vasodilator ADO. Administration of ADO was started 10 min after the start of HBOC-201 infusion and continued till the end of the second run. The infusion rate of ADO (25 mg/ml) was titrated to obtain a stable MAP similar to that before HBOC administration. To determine the involvement of NO in HBOC-201-induced hypertension, in six swine, the NO donor nitroglycerin (NTG) was infused starting 10 min after the start of HBOC-201 infusion. To prevent a direct interaction between the NO donor and HBOC-201, HBOC-201 and NTG were infused through separate catheters. The infusion rate of NTG (1 mg/ml) was titrated to obtain a stable MAP similar to that before HBOC administration. To further investigate the role of endogenous NO, NO production was inhibited using the NO synthase inhibitor Nω-nitro-l-arginine (l-NNA, 20 mg/kg iv) in five swine (9). After administration of l-NNA, swine underwent an l-NNA exercise trial. Ninety minutes later, HBOC-201 (10 ml/kg iv) was given to the animals, and they underwent a second exercise trial. As previously shown (57), l-NNA has a long-lasting effect so no additional l-NNA was administered before the second exercise protocol. Loss of NO reduces ROS scavenging and may increase the production of the potent vasoconstrictor ET. To determine the involvement of ET and ROS in HBOC-201-induced hypertension, HBOC-201 was infused after prior administration of an ET-receptor antagonist or a cocktail of ROS scavengers. After completing a control exercise protocol, animals were allowed to rest on the treadmill for 90 min. Then, the mixed ETA and ETB receptor (ETA/ETB) antagonist tezosentan was intravenously administered over 10 min in a dose of 3 mg/kg iv (slow bolus), followed by a continuous infusion of 6 mg·kg−1·h−1 iv in four swine (38). HBOC-201 (10 ml/kg iv) was started upon completion of the tezosentan slow bolus. When HBOC-201 infusion was completed, the exercise protocol was repeated. We used a mixture of different substances to scavenge all ROS during HBOC administration. The mixture consisted of N-acetylcysteine (NAC; 150 mg/kg iv), 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (Tempol; 30 mg/kg iv), and mercaptopropionyl glycine (MPG; 1 mg/kg iv) (26). NAC is an aminothiol and synthetic precursor of intracellular cysteine and glutathione and is thus considered an important antioxidant (6). It is generally assumed that the antioxidant and free radical scavenging activities of NAC are attributable to increasing intracellular glutathione levels; however, NAC also possesses a reducing property through its thiol-disulfide exchange activity (2, 26). Tempol is a stable piperidine nitroxide and scavenges superoxide anions in vitro and may act as a SOD mimetic (26). Tempol also reduces the formation of hydroxyl radicals either by scavenging superoxide anions or hydroxyl radicals (via the Fenton or Haber-Weiss reactions) (26). N-2-mercaptopropionylglycine (MPG) is a synthetic thiol compound that is not highly radical specific and scavenges different types of ROS, including O2·−, ONOO−, and ·OH (2, 26, 57). After completing a control exercise protocol, the animals were allowed to rest on the treadmill for 90 min. Then, the scavenger mixture was administered in five swine, starting 10 min before the HBOC-201 infusion. The administration of NAC and Tempol was completed before administration of HBOC-201, while MPG infusion continued throughout HBOC-201 administration and the subsequent exercise protocol. Digital recording and offline analysis of hemodynamic variables have been described in detail elsewhere (10, 11, 38). Systemic vascular resistance (SVR) was computed as mean aortic blood pressure divided by cardiac output. Pulmonary vascular resistance (PVR) was computed as mean pulmonary arterial pressure minus mean left atrial pressure divided by cardiac output. Body lactate production/consumption was calculated as the product of cardiac output and arterio-mixed venous lactate difference. Statistical analysis of hemodynamic data was performed with SPSS 22 [IBM (Armonk, NY) released 2013. IBM SPSS Statistics for Windows, Version 22.0]. Since no differences between male and female swine were found in the response to HBOC-201 administration alone, data from both sexes were pooled. The effects of drug treatment and exercise were compared using a two-way ANOVA for repeated measures. When significant effects were detected, post hoc testing was performed using paired or unpaired t-test, with Bonferroni correction. Statistical significance was accepted when P ≤ 0.05. Data are presented as means ± SE. RESULTSAdministration of HBOC resulted in significant pressor effects in the systemic and pulmonary circulations with an increase in MAP (27 ± 3 mmHg) and PAP (14 ± 1 mmHg). These pressor responses were accompanied by a probable baroreflex-mediated decrease in HR, which together with a slight decrease in stroke volume, resulted in a decrease in cardiac output (Table 1). These pressor effects were the result of significant systemic and pulmonary vasoconstriction, as evidenced by significant increases in SVR and PVR. There was no sign of anaerobic metabolism, as arterial and mixed venous lactate levels (Table 2) and body lactate consumption (not shown) were maintained. The hemodynamic responses to HBOC-201 administration occurred during the first 10 min of HBOC-201 administration after which they stabilized. Moreover, a second and third HBOC administration with 5–7 days washout in between yielded hemodynamic responses similar to the first administration (Fig. 1).
Fig. 1.Systemic and pulmonary hemodynamics at rest and during exercise following administration of hemoglobin-based oxygen carrier (HBOC)-201 alone (n = 3 pigs, left), demonstrating reproducibility of 3 separate administrations of HBOC-201, and in combination with infusion of adenosine (ADO) (n = 6 pigs in full crossover study design, right). MAP, mean arterial pressure; SVR, systemic vascular resistance; PAP, pulmonary artery pressure; PVR, pulmonary vascular resistance. *P ≤ 0.05, **P ≤ 0.10, compared with control; †P ≤ 0.05, compared with HBOC-201 alone; ‡P ≤ 0.05, effect of HBOC+ Ado different from HBOC-201 alone. The pressor response to HBOC-201 was maintained during exercise (Fig. 1). In the systemic circulation, both MAP and SVR remained elevated throughout the exercise protocol as compared with control, although the elevation of SVR tended to wane with incremental levels of exercise. Similarly, in the pulmonary circulation both PAP and PVR remained elevated at all exercise intensities. Coinfusion of ADO was carefully titrated to maintain MAP at a level similar to MAP before HBOC-201 infusion (Fig. 1). Dosages required to stabilize MAP fluctuated throughout the experiment, but they were on average 0.17 ± 0.01 mg·kg−1·min−1 (range between 0.08 and 0.38 mg·kg−1·min−1). Although the HBOC-201-induced changes in SVR and PVR were abolished by ADO (Fig. 1), PAP tended to remain slightly higher (P = 0.08) due to a slight increase in left atrial pressure (not shown). MAP increased by ~15 mmHg upon cessation of ADO infusion (not shown). The exogenous administration of NO, by coinfusion of the NO-donor NTG, was also titrated to counteract systemic pressor responses to HBOC-201. The dose of NTG required to stabilize MAP increased from 0.11 ± 0.01 mg·kg−1·min−1 at 20 min of HBOC-201 infusion to 0.22 ± 0.06 mg·kg−1·min−1 upon completion of HBOC-201 infusion (P = 0.05) and remained essentially unchanged during the exercise protocol, being 0.16 mg·kg−1·min−1 at maximal exercise (range from 0.06 to 0.49 mg·kg−1·min−1). This dose of NTG negated the HBOC-201-induced increase in SVR as well as PVR and, thereby, the elevated pressures in these vascular beds (Fig. 2). Similar to ADO, MAP increased upon cessation of the NTG-infusion (not shown). Fig. 2.Systemic and pulmonary hemodynamics at rest and during exercise following administration of the nitric oxide (NO)-donor nitroglycerin (NTG; left) or the eNOS inhibitor Nω-nitro-l-arginine (l-NNA; right) in combination with HBOC-201. *P ≤ 0.05, compared with control; †P ≤ 0.05, compared with HBOC-201 alone; ‡P ≤ 0.05, effect of HBOC+ NTG different from HBOC-201 alone; §P ≤ 0.05, §§ P ≤ 0.1 compared with l-NNA alone; n = 6 pigs (NTG) or 5 pigs (l-NNA) in a full crossover study design. For the sake of clarity, statistics comparing HBOC-201 with control are not shown, but they are identical to Fig. 1. Endothelial NOS (eNOS) blockade with l-NNA resulted in peripheral vasoconstriction, as evidenced by a significant increase in SVR and an increase in MAP. The increase in MAP was accompanied by increases in LV systolic pressure, as well as left atrial pressure, and probably, by a baroreflex-mediated decrease in HR and cardiac output, as stroke volume was not altered (Fig. 2). However, subsequent infusion of HBOC-201 did not result in a further increase in MAP or SVR. In contrast to the findings in the systemic circulation, HBOC-201 induced an increase in mean PAP and PVR even in the presence of l-NNA (Fig. 2). Thus HBOC-201 induced further pulmonary vasoconstriction following the vasoconstriction produced by l-NNA, both at rest and during exercise, suggesting that, in addition to scavenging of NO, HBOC-201 exerts its vasoconstrictor effect through another pathway in the pulmonary circulation. Of note, when HBOC-201 and l-NNA were coinfused, systemic pressor responses appeared to be increased, as compared with the effect of HBOC-201 alone (Fig. 2), indicating that not all NO is scavenged by HBOC-201. Administration of the mixed ETA and ETB receptor antagonist tezosentan reduced MAP by 10 ± 4 mmHg (P < 0.05), and negated the systemic hypertension caused by subsequent HBOC-201 by preventing the increase in SVR (Fig. 3). Also, in the pulmonary circulation, HBOC-201 had no effect on either pulmonary pressure or PVR, in the presence of tezosentan (Fig. 3). These data suggest that activation of the endothelin system is an important contributor to the vasoconstrictor response to HBOC-201. Fig. 3.Systemic and pulmonary hemodynamics at rest and during exercise following administration of the endothelin ETA/ETB-receptor blocker Tezosentan (n = 4 pigs in full crossover study design; left) or a reactive oxygen species (ROS) scavenger cocktail comprised of N-acetylcysteine (NAC), Tempol, and mercaptopropionyl glycine (MPG; n = 5 pigs in full crossover study design; right) in combination with HBOC-201. *P ≤ 0.05, compared with control; †P ≤ 0.05, compared with HBOC-201 alone; ‡P ≤ 0.05, effect of HBOC+ Tezo or ROS different from HBOC-201 alone. For the sake of clarity, statistics comparing HBOC-201 with control are not shown, but are identical to Figure 1. ROS scavenging in itself had no significant effect on MAP (ΔMAP 12 ± 8 mmHg, P = 0.21). Coinfusion of ROS scavengers with HBOC-201 slightly reduced the effect of HBOC-201 on mean arterial pressure but did not significantly affect SVR (Fig. 3). Similarly, in the pulmonary vascular bed, no reduction of pressor effects could be detected at rest, and the effects of HBOC-201 tended to be exacerbated during exercise following administration of the ROS scavenger cocktail (Fig. 3). DISCUSSIONIn the present study we report, in accordance with previous publications (40, 46, 60, 63–65), that intravenous administration of HBOC-201 resulted in systemic and pulmonary hypertension as a result of vasoconstriction, which was maintained during exercise. Pressor responses could be prevented by coinfusion of NTG or ADO both at rest and during exercise; however, this required careful titration of the dosage of these vasodilators. eNOS inhibition prevented HBOC-201-induced increase in systemic vasoconstriction, and it reduced but did not abolish HBOC-201-induced pulmonary vasoconstriction. ETA/ETB blockade with tezosentan prevented the HBOC-201-induced pressor responses in the systemic and pulmonary vasculature, while ROS-scavenging tended to blunt the pressor response in the systemic but not pulmonary vasculature. The main hemodynamic effects of HBOC-201 occurred during the first 10 min of its administration and were maintained throughout the entire infusion and after infusion. Repeated administration of HBOC-201, following complete washout, induced virtually identical effects, corroborating results from ECMO priming with HBOC-201 in piglets (21, 62) and indicating that no immune reaction occurred in response to the protein and that repeated administration is safe. Also, plasma clearance of HBOC-201, which has been shown to follow first-order pharmacokinetics with an elimination half time of 20 h, for either single or multiple dosage regimens (25), was comparable with previous studies (25, 34). As anticipated, HBOC-201 produced an increase in arterial Hb, met-Hb, and plasma metHb (Table 2); however, the level of metHb in this study was well below toxic levels (32), and coinfusion of NTG with HBOC did not elevate metHb levels further (not shown). The pressor effect of HBOC has been ascribed to the scavenging of NO, primarily by plasma ferrous heme, thereby lowering NO concentration (63). Our results support previous findings that NTG is capable of negating HBOC-201-induced vasoconstriction and the accompanying increases in systemic and pulmonary blood pressures (27). NTG reduces vascular resistance in small and large vessels through endothelium-independent, but NO-mediated, vasodilation (66). However, earlier studies were skeptical using NTG coadministration as a therapeutic option to negate the pressor effect of HBOC due to its short half-life, requiring continuous infusion. Moreover, profound systemic vasodilation and hypotension might occur in response to NTG, potentially jeopardizing resuscitation from hemorrhagic shock (27, 31). In the present study, we avoided NTG-induced hypotension through careful NTG titration to maintain MAP within physiological limits. Importantly, the NTG dosage required to normalize systemic pressures was also capable of normalizing pulmonary pressures. NTG was compared with ADO, a purine nucleoside and principal NO-independent vasodilator (14, 37). At ADO infusion rates that eliminated HBOC-induced systemic hypertension, the pulmonary pressor effect was not fully eliminated, despite restoration of normal PVR. The persistence of the pulmonary pressor effect was likely due to elevated left atrial pressure, secondary to adenosine-induced negative cardiac inotropy (20). Indirect oxidation of hemoglobin involves a process of cooxidation in which the methemoglobin-forming agent is cooxidized with heme iron by hemoglobin-bound oxygen (HbO2) (4). O2·− and H2O2 are produced when HbO2 accepts electrons from ferrous heme and the methemoglobin-forming agent. However, ROS scavenging only marginally influenced the pressor response to HBOC-201 in the systemic vasculature, while the pulmonary pressor response was unaffected, suggesting that ROS do not play a major role in the pressor effect of HBOC-201. Alternatively, it is possible that oxidative stress indirectly modulates vascular tone. Indeed, it has been shown that oxidative stress enhances ET production through stabilization of prepro-endothelin mRNA (15, 35, 56). In the present study, the ET-receptor blocker tezosentan was capable of negating the pressor responses of HBOC-201 in both the systemic and pulmonary vasculature. Therefore, it is plausible that pressor effects of HBOC-201 may result from disinhibition of endothelin synthesis and release (17, 47). ET is a potent and long-lasting vasoconstrictor and ET-receptor blockade could, therefore, potentially provide another strategy to oppose HBOC-201-induced vasoconstriction. Although tezosentan is a relatively short-acting receptor blocker with a half-life of 3 h (7), longer acting ET-receptor blockers are available. ET-receptor blockade would require less patient monitoring and have a lower risk of inducing hypotension. To our knowledge, this is the first study to analyze the effect of a cell-free oxygen carrier on exercise hyperemia. In the normal healthy vasculature, exercise-induced vasodilation is regulated via an intricate interplay of vasoactive molecules, including NO, ROS, and endothelin (9, 22, 30). As outlined above, HBOC-201 could potentially scavenge NO and enhance production of ROS and ET and could, therefore, interfere with exercise-induced vasodilation. However, although the pressor responses of HBOC-201 were essentially unaffected by exercise, SVR did decrease during exercise following administration of HBOC-201, indicating that exercise-induced vasodilation is essentially intact. To assess the role of NO in HBOC-201 pressor effects at rest and during exercise, HBOC administration was repeated following eNOS inhibition. If indeed, scavenging of NO is the main contributor to the pressor effect of HBOC-201, eNOS inhibition would be expected to block a further pressor effect by HBOC-201. Indeed, in accordance with previous studies (50, 64), following inhibition of eNOS and the consequent vasoconstriction, no additional vasoconstriction was induced by HBOC-201 infusion in the systemic vasculature. In contrast, HBOC-201 did result in a further increase in PVR. It is not clear why the pulmonary and systemic vasculature responded differently to the combination of l-NNA and HBOC-201. However, it is possible that l-NNA did not completely inhibit eNOS in the pulmonary circulation, although this is unlikely given the high dose (20 mg/kg iv) of l-NNA administered. An alternative explanation for the divergent effects in the systemic and pulmonary vasculature is that it has been shown that the nature of the chemical interaction between NO and Hb is dependent on the amount of oxygen present. Formation of FeIINOHb occurs principally when Hb is deoxygenated (T-state) in peripheral tissue. NO bound to the heme-group can be transferred to a specific cysteine residue (β93Cys) upon reoxygenation of Hb in the lung, resulting in formation of SNO-Hb (54). This SNO-Hb formation can also occur directly but only when Hb is oxygenated (R-state). From SNO-Hb, NO can be either released or transferred to another thiol group, thereby preserving part of NO signaling (36, 55). Thus S-nitrosylation of Hb is governed, in part, by the state of the Hb molecule undergoing an allosteric shift from R to T shift during passage in the circulatory system (29). These varying degrees of Hb S-nitrosylation at different molecular states (R and T) may explain, at least in part, the different hemodynamic responses to HBOC-201 in the systemic and pulmonary vasculature following NOS inhibition by l-NNA. In peripheral microvessels, because of low oxygen tension, SNO-Hb levels are low and NO released from SNO-Hb may be consumed by biological targets and catabolic reactions, such as those mediated by GSNO reductase (55). Consequently, the availability of bioactive NO may be closely coupled to de novo synthesis by NOS that is inhibitable by l-NNA, leaving little residual NO for scavenging by HBOC-201. By contrast, bioactive NO in the form of SNO-Hb is abundant in lungs and well protected inside erythrocytes but, upon release, is susceptible to scavenging by free Hb, manifesting as a further increase in PVR following eNOS inhibition. A third explanation may be that the vasoconstrictor effect of HBOC-201 is not solely mediated through scavenging of NO. A ROS scavenging cocktail failed to appreciably alter hemodynamic responses to HBOC-201 and, unlike either NTG or ADO, failed to restore SVR or PVR to control levels. As glutaraldehyde-polymerized HBOC in itself was shown to exhibit catalase-like properties, it is possible that the increase in free radicals induced by administration of this HBOC was negated by HBOC itself (58). Indeed, lipid peroxidation as measured by thiobarbituric acid reactive substances was not significantly elevated by glutaraldehyde-polymerized HBOC. Similar to our study, these observations suggest that HBOC-201 fails to significantly stimulate ROS formation or that any HBOC-induced increase in ROS is adequately scavenged by endogenous antioxidants. However, it is possible that in certain disease states generally characterized by elevated oxidative stress (57), such as reperfusion following myocardial infarction or in the presence of severe endothelial dysfunction, HBOC may exacerbate oxidative stress, either directly or through scavenging NO. In vitro studies have suggested that HBOCs may amplify ROS formation that could, in turn, react with NO to generate nitroxide radicals (33) and/or uncouple NO synthase secondary to insufficient cofactors tetrahydrobiopterin (BH4) and NADPH required to convert l-arginine to l-citrulline and NO (43). The pattern of SVR and PVR during exercise with HBOC-201 very much resembles the pattern found with ET antagonism, as we previously showed that the vasodilator effect of ET-receptor blockade waned with increasing exercise intensity in the systemic circulation while it increased in the pulmonary vasculature (38). Moreover, reduced bioavailability of NO and/or oxidative stress could contribute to overexpression of ET (3). Although we did not measure plasma ET-levels in the present study, an increase in ET-mediated vasoconstriction as a cause of the pressor effects of HBOC-201 is consistent with the ability of tezosentan to negate these pressor effects both at rest and during exercise. Importantly, dosing of the ET-receptor blocker tezosentan did not need to be altered during exercise, and endothelin-antagonists by themselves have only very modest effects on hemodynamics, making endothelin-antagonists clinically attractive antagonists of the pressor effect of HBOC-201. Finally, several clinical trials have shown that other HBOC products with other compositions than HBOC-201 may increase the risk of myocardial infarction and death (42). However, clinical studies conducted with HBOC-201 in patients with documented cardiovascular disease showed both intravenous and intracoronary infusion of HBOC-201 to be safe and well tolerated (53). Although plasma ET-1 measurements could possibly strengthen the conclusion that HBOC-201-induced vasoconstriction is ET mediated, a number of both physiological and methodological issues complicate interpretation of such measurements. First, ET is released for more than 80% into the abluminal side, while less than 20% is secreted into the lumen side. Hence circulating ET-1 does not reflect the local concentration of ET-1 in the vessel wall. Second, an ET-mediated pressor effect may be caused by changes in ET-receptor sensitivity through altered nitrosylation of the ET-receptors. Finally, we have previously found that l-NNA and indomethacin yield a false-positive result in plasma ET-1 measurements (22). Similarly, an interaction with HBOC-201 cannot be excluded, making it difficult to interpret the results. Therefore, blocking ET receptors with tezosentan is the best way to assess the interaction of HBOC-201 with the ET system. HBOC-201 can disrupt hemodynamic homeostasis, mimicking some aspects of endothelial dysfunction, resulting in elevated systemic and pulmonary blood pressures. HBOC-201 induced vasoconstriction is mediated by scavenging NO and likely by upregulating ET production. Pressor effects can be restored by NO donors or direct vasodilators, such as nitroglycerin or ADO, but dosages must be carefully monitored to avoid hypotension. However, hemodynamic normalization was more easily achieved via administration of an ET receptor blocker. Future studies should focus on coadministration of long-acting ETA receptor antagonists (e.g., ambrisentan or sitaxentan) and, although oxygen-derived free radicals do not appear to play a significant role in HBOC-201-induced pressor responses of healthy subjects, the possible role of ROS in HBOC-induced vasoconstriction in subjects with documented preexisting endothelial dysfunction would be of interest. GRANTSThis study was supported by Navy Bureau of Medicine (BUMED) congressional funding World Universities Network (WUN) 604771N.9737.001.A0315. DISCLOSURESG. P. Dubé was previously employed by Biopure Corporation and OPK Biotech, LLC. P. F. Moon-Massat was previously employed by Biopure Corporation. We have no further conflicts of interest to report. AUTHOR CONTRIBUTIONSY.J.T., D.P.d.W.-M., M.t.L.H., and D.M. performed experiments; Y.J.T., D.P.d.W.-M., M.t.L.H., and D.M. analyzed data; Y.J.T., D.J.D., and D.M. interpreted results of experiments; Y.J.T. prepared figures; Y.J.T. and D.M. drafted manuscript; Y.J.T., P.F.M.-M., G.P.D., D.J.D., and D.M. edited and revised manuscript; Y.J.T., P.F.M.-M., G.P.D., D.J.D., and D.M. approved final version of manuscript; P.F.M.-M., D.J.D., and D.M. conceived and designed research. Experiments were performed at the Division of Experimental Cardiology, Department of Cardiology, Thoraxcenter, Erasmus University Medical Center Rotterdam, The Netherlands. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, nor the U.S. Government. P. F. Moon-Massat is a military service member (or an employee or contractor of the U.S. Government). This work was prepared as part of his official duties. Title 17 U.S.C. §105 provides that “Copyright protection under this title is not available for any work of the United States Government.” Title 17 U.S.C. § defines a U.S. Government work as work prepared by a military service member or employee of the U.S. Government as part of that person’s official duties. REFERENCES
Page 20AbstractImpaired endothelial function is observed with aging and in those with low cardiorespiratory fitness (V̇o2peak). Improvements in endothelial function with exercise training are somewhat dependent on the intensity of exercise. While the acute stimulus for this improvement is not completely understood, it may, in part, be due to the flow-mediated dilation (FMD) response to acute exercise. We examined the hypothesis that exercise intensity alters the brachial (systemic) FMD response in elderly men and is modulated by V̇o2peak. Forty-seven elderly men were stratified into lower (V̇o2peak = 24.3 ± 2.9 ml·kg−1·min−1; n = 27) and higher fit groups (V̇o2peak = 35.4 ± 5.5 ml·kg−1·min−1; n = 20) after a test of cycling peak power output (PPO). In randomized order, participants undertook moderate-intensity continuous exercise (MICE; 40% PPO) or high-intensity interval cycling exercise (HIIE; 70% PPO) or no-exercise control. Brachial FMD was assessed at rest and 10 and 60 min after exercise. FMD increased after MICE in both groups {increase of 0.86% [95% confidence interval (CI), 0.17–1.56], P = 0.01} and normalized after 60 min. In the lower fit group, FMD was reduced after HIIE [reduction of 0.85% (95% CI, 0.12–1.58), P = 0.02] and remained decreased at 60 min. In the higher fit group, FMD was unchanged immediately after HIIE and increased after 60 min [increase of 1.52% (95% CI, 0.41–2.62), P < 0.01, which was correlated with V̇o2peak, r = 0.41; P < 0.01]. In the no-exercise control, FMD was reduced in both groups after 60 min (P = 0.05). Exercise intensity alters the acute FMD response in elderly men and V̇o2peak modulates the FMD response following HIIE but not MICE. The sustained decrease in FMD in the lower fit group following HIIE may represent a signal for vascular adaptation or endothelial fatigue. NEW & NOTEWORTHY This study is the first to show that moderate-intensity continuous cycling exercise increased flow-mediated dilation (FMD) transiently before normalization of FMD after 1 h, irrespective of cardiorespiratory fitness level in elderly men. Interestingly, we show increased FMD after high-intensity cycling exercise in higher fit men, with a sustained reduction in FMD in lower fit men. The prolonged reduction in FMD after high-intensity cycling exercise may be associated with future vascular adaptation but may also reflect a period of increased cardiovascular risk in lower fit elderly men. aging is associated with chronic low-grade inflammation, oxidative stress, and impaired nitric oxide (NO) bioavailability that contribute to endothelial dysfunction and large artery stiffness (58, 59). Endothelial dysfunction is considered an important prognostic factor and precursor to the development of atherosclerosis (23, 49) and is strongly associated with the risk of cardiovascular events (23, 61). In addition, endothelial dysfunction is suggested to contribute to other age-associated disorders including cognitive impairment and insulin resistance (64, 66, 76). As such, interventions that prevent or slow the detrimental changes in endothelial function are important in reducing the cardiovascular risk and mortality associated with increasing age (60, 61). Importantly, age-associated endothelial dysfunction, measured using flow-mediated dilation (FMD) of the brachial artery (63), can be attenuated with both regular physical activity (75) and exercise training (16, 24). Results of cross-sectional studies indicate that exercise-trained older adults have preserved endothelial function (17, 42, 48, 53) and reduced cardiovascular disease risk (63), compared with those who are not habitually active. This adaptive response is commonly attributed to the repeated episodes of elevated blood flow, and consequently shear stress, observed during acute exercise that induces vascular adaptation (22). While the positive impact of chronic aerobic exercise on endothelial function is well described, the significance of the transient changes observed in endothelial function with acute exercise is less clear (15). To elucidate which forms of exercise are most likely to benefit cardiovascular health and function, recent studies have focused on the acute FMD response and how it is modulated by factors such as exercise intensity. Some evidence suggests that the FMD response to acute exercise may be biphasic, involving an immediate decrease, followed by a transient increase in FMD before returning to baseline levels (15). This may represent the acute initiation of an adaptive response and be linked to the long-term benefit provided by exercise training on endothelial function at rest (24). This response is suggested to be exaggerated following acute higher intensity exercise, e.g., a larger immediate reduction followed by transient improvement in FMD (3, 11, 15, 33), and may contribute to recent observations of larger improvements in FMD following high-intensity interval exercise (HIIE) compared with moderate-intensity continuous exercise (MICE) training (50, 56). We hypothesize that the biphasic FMD response would be further exaggerated in individuals with low cardiorespiratory fitness. To date, there have been no comparisons of the FMD response to acute exercise between individuals of a higher and lower cardiorespiratory fitness. There is a strong association between a higher cardiorespiratory fitness and maintenance of FMD with aging (42). HIIE training improves cardiorespiratory fitness in healthy elderly adults to a greater extent than MICE training (29), suggesting that it may also modulate the acute FMD response to exercise. Despite this, no study has investigated the influence of a lower and higher cardiorespiratory fitness on the FMD response following acute exercise in the elderly. We therefore aimed to determine whether the effect of acute exercise on FMD differed between MICE and HIIE cycling in elderly males, when controlling for both exercise work and duration. In addition, we assessed the influence of cardiorespiratory fitness on the acute effect of exercise intensity on the FMD response between participants with higher and lower cardiorespiratory fitness. In line with previous findings in the young (3, 11), we hypothesized that acute HIIE would stimulate greater immediate reductions in endothelial function compared with MICE, with subsequent elevation in FMD after 60 min. We also hypothesized that this overall response would be attenuated in those with a higher cardiorespiratory fitness. METHODSParticipants underwent four laboratory visits, each following an overnight fast, refraining from alcohol and exercise for 24 h and caffeine for 12 h, before each visit. Participants consumed a standardized snack (4 oat breakfast biscuits, 20 g carbohydrate, and 8 g fat) 3 h before attending the laboratory, and the macronutrient content of this snack was unlikely to influence endothelial function (25, 74). Visit 1 consisted of baseline measurements of height, body mass, and estimated body composition using bioimpendence scales (BC 545N; Tanita). After 10 min of supine rest, blood pressure was measured using a manual sphygmomanometer, which was followed by a maximal cycling test to determine cardiorespiratory fitness (V̇o2peak) and peak power output (PPO). Experimental visits (visits 2, 3, and 4) were randomized, counterbalanced, and consisted of two separate acute cycling exercise conditions (moderate-intensity continuous vs. high-intensity interval) or a no-exercise control condition. Blood pressure and brachial FMD were assessed at baseline following 20 min of supine rest and then repeated at 10 and 60 min following exercise/control. Laboratory conditions were standardized for each visit (room temperature: 23 ± 1°C) (67). To control for diurnal variation in blood pressure and vascular function, each visit was performed at the same time of day (34) and separated by 7 days. Forty-seven healthy elderly males [means ± SD, aged 70 ± 5 y; body mass index (BMI) 25.3 ± 3.4 kg/m2] were recruited from a University of the Sunshine Coast alumni cohort and local advertisement. Participants were screened using a preexercise screening questionnaire (2, 52) and included if they were able to exercise and were nonsmokers (>12 mo no smoking history). Participants were excluded if they were aged >86 yr, had a BMI >39, or had a chronic cardiovascular or metabolic condition including uncontrolled hypertension, known heart or vascular disease, angina, and atrial fibrillation. During the study, participants were requested to continue to take all prescribed medication. Participants were informed of the methods and study design verbally and in writing before providing written informed consent. The study conformed to the Declaration of Helsinki and was approved by the University of the Sunshine Coast ethics committee. A maximal incremental cardiorespiratory fitness test was performed in an upright position on an electromagnetically braked cycle ergometer (Lode Corival, Groningen, The Netherlands). Following a 3 min warm up at 0 peak power output (W), the test began at 20 W and then increased by 10 W each min until volitional cessation. Participants were required to self-select a pedal cadence (between 60 and 90 rpm) and maintain this throughout the test. Expired respiratory gases were collected throughout the test and data were averaged every 15 s (Parvo Medics) for the determination of oxygen consumption (V̇o2; ml·kg−1·min−1). Peak VO2 was determined as the highest 15-s average over the last 60 s of maximal exercise (V̇o2peak). Heart rate was measured continuously using 12-lead ECG (Mortara) and recorded, along with the rate of perceived exertion (RPE) using the 0–10 Borg scale, during the final 10 s of each stage. All participants reached the criteria for maximum effort based on attaining >2 of the following: a peak heart rate within 10 betas/min of predicted age-related maximum; RPE (>9); a fall in pedal cadence (>10 RPM); a plateau in V̇o2 despite an increase in workload; and a respiratory exchange ratio >1.15. Peak power output (W) was then used to establish the exercise intensity in the subsequent test visits. Following pretest measurements, participants performed 27 min of upright continuous or interval cycling exercise or no-exercise control (seated rest). Both acute exercise protocols commenced with a 3-min warm-up at 0 W, followed by either 24 min of 1) continuous moderate-intensity cycling at 40% PPO, or 2) high-intensity interval cycling involving twelve 60-s bouts at 70% PPO, with each separated by 60 s at 10% PPO. Heart rate and RPE were recorded every 2 min. This design ensured the continuous and interval cycling exercise protocols were duration and work matched. Control consisted of 27 min of seated rest with both arms relaxed and rested on a table in front. The total measurement period and timing between measurements were the same across exercise and control visits. Immediately following exercise/control (<60 s), participants were moved to the supine position and asked to remain supine for posttest FMD measurements (at 10 and 60 min). Right brachial artery blood pressure was measured in triplicate using an automated device (Sphygmocor XCEL; AtCor Medical, West Ryde, NSW, Australia) 10 min before each FMD time point to negate any effect of cuff inflation on FMD. Brachial artery FMD was used as a measure of endothelial function (67). Measurements were performed in the supine position on the right arm with the cuff placed distal to the olecranon process. High-resolution duplex ultrasound (T3000; Terason, Burlington, MA) with a 12-MHz multifrequency linear array probe was used to image the brachial artery at the distal third of the upper arm and simultaneously record the longitudinal B-mode image and Doppler blood velocity trace. The angle of Doppler insonation was 60°. Images were optimized, and settings (depth, focus position, and gain) were maintained between FMD assessments within each individual visit, and the location of the transducer was recorded and marked on the skin using an indelible marker. Following a 60-s baseline recording period, the cuff was rapidly inflated to 220 mmHg and maintained for 5 min (D. E. Hokanson, Bellevue, WA). Ultrasound recordings resumed 30 s before rapid cuff deflation (<2 s) and continued for 3 min thereafter in accordance with recommendations (12, 67). All ultrasound scans were performed by the same trained sonographer. Analysis of brachial artery diameter was performed using custom-designed edge-detection and wall-tracking software, which is largely independent of investigator bias. Recent papers describe the analysis approach in detail (12, 67). Briefly, from recordings of the synchronised artery diameter and blood velocity data, blood flow (the product of lumen cross- sectional area and Doppler velocity) was calculated at 30 Hz. Shear rate (an estimate of shear stress independent of viscosity) was calculated as four times the mean blood velocity/vessel diameter. This semiautomated software possesses an intraobserver coefficient of variation (CV) of 6.7% and reduces error, with the reproducibility of diameter measurements significantly better than manual methods (68, 77). To differentiate the cohort on the basis of cardiorespiratory fitness, each participant was stratified into lower (V̇o2peak <27 ml·kg−1·min−1) and higher (V̇o2peak >31 ml·kg−1·min−1) fitness (fit) groups based on age- and sex-specific normative data (2). These differences in cardiorespiratory fitness were closely aligned with the prior observation that cardiovascular burden and mortality are significantly reduced with a V̇o2peak >28 ml·kg−1·min−1, e.g., metabolic equivalent score of 8, in males over the age of 65 (10, 44). A three-way (fitness × protocol × time) linear mixed model (LMM) was employed to analyze changes in FMD parameters [brachial diameter, peak diameter and FMD (mm), FMD (%), time to peak, shear rate area under the curve (SRAUC), and blood flow] and blood pressure between the two fitness groups (low and high fitness) across “time” (baseline and 10 and 60 min post) during each protocol (control, moderate-, and high-intensity exercise). As variability in the baseline artery diameter and shear rate may influence the magnitude of the FMD response (69), these parameters were included in the analysis as covariates (1, 9). In line with recent recommendations (4–6), we also performed an additional three-way LMM analysis of logarithmically transformed absolute diameter change (difference between peak and baseline diameter as the outcome in millimeters), with logarithmically transformed baseline diameter and shear rate again included as covariates, specific to each FMD test. The logged absolute diameter change was then also interpreted in the conventional manner and is presented as “adjusted FMD%” for comparative purposes as suggested (8) in line with recent reports (3, 71). This allometric approach may be more accurate for scaling changes in diameter than percentage change alone, which makes implicit assumptions about the linearity of the relationship between baseline diameter and peak diameter (7). The strength of the relationships between cardiorespiratory fitness and changes in FMD after exercise and/or control was assessed using Pearson correlation coefficient. Similarly, a three-way LMM analysis was used to detect any differences in heart rate, blood pressure, and perceived exertion in response to the acute protocols between the two fitness groups (low and high fit) across time (at 2- and 6-min intervals for heart rate/RPE and blood pressure, respectively) during each protocol (control, moderate-, and high-intensity exercise). Statistically significant interactions were further investigated with multiple comparisons using the least significant difference approach (46, 55). Analyses were conducted using the Statistical Package for Social Sciences (Version 22; IBM SPSS, Chicago, IL). Statistical significance was delimited at P ≤ 0.05 and exact P values are cited (P values of “0.00” are reported as “<0.01”). Data are presented in the text as means [95% confidence interval (95% CI)] unless otherwise stated. RESULTSParticipant characteristics are presented in Table 1. Participant age was higher in the lower fit compared with the higher fit group [mean difference of 3 yr (95% CI, −1–6), P = 0.05]. Approximately one-quarter of the participants were hypertensive (30 and 26% in the lower and higher fitness groups, respectively), and all hypertensive participants were taking blood pressure–controlling medication. Resting heart rate was lower in the higher fit compared with lower fit [mean difference 6 beats/min (95% CI, 2–10), P = 0.01], but there were no differences in resting blood pressure or anthropometric variables between lower and higher fit groups.
There was a mean difference of 11 ml·kg−1·min−1 (95% CI, 8–13, P < 0.01) in V̇o2peak and 50 W (95% CI, 30–70, P < 0.01) between higher and lower fit groups. Heart rate responses were normalized for peak heart rate obtained during the cardiorespiratory fitness test. Heart rate was significantly higher during high-intensity exercise [mean 65% peak heart rate (HRpeak) (95% CI, 62–68%,)] compared with moderate-intensity exercise [mean 58% HRpeak (95% CI, 55–61%, P < 0.01)], while both were elevated compared with control [mean 37% HRpeak (95% CI, 34–40), P < 0.01]. There was no effect of fitness on the heart rate responses (P = 0.24). Similarly, mean arterial pressure was higher during high-intensity exercise [mean change of 18 mmHg (95% CI, 14–20)] compared with moderate-intensity exercise [mean change of 14 mmHg (95% CI, 11–16), P = 0.02] while both were elevated compared with control [mean change 5 mmHg (95% CI, 6–10), P < 0.01]. There was no effect of fitness on the mean arterial pressure responses (P = 0.45). RPE was higher during the HIIE [mean RPE 4 arbitrary units (95% CI, 3–5)] compared with moderate-intensity exercise [mean RPE 3 arbitrary units (95% CI, 2–4, P < 0.01)]. There was no effect of fitness on the RPE responses (P = 0.58). The CV for baseline FMD% across the three visits in this study was 11.8 ± 3.9%, which is similar to that previously reported (10.1–14.7%) (70, 77). Using test-retest data from our control condition (baseline and 10 min post), we established that the within-day CV% for FMD% was 8.06 ± 7.50%. There were no differences in resting (preexercise/control) brachial diameter, FMDmm, FMD%, or SRAUC across the three separate testing days (Table 2; P > 0.05).
There was no significant difference in resting FMD% between the lower (Table 3, Low fit) and higher fit groups (Table 3, High fit) [mean difference of 0.2% (95% CI, −0.8–0.9), P = 0.82]. SRAUC was significantly higher in the lower fit compared with the higher fit group [mean difference of 3.2 103/s (95% CI, 1.3–6.3), P = 0.04], despite no differences in baseline diameter between fitness groups [mean difference of 0.2 mm (95% CI, −0.6–0.8), P = 0.13]. Furthermore, time-to-peak diameter was significantly longer in the lower fit compared with the higher fit group [mean difference of 10 s (95% CI, 1–17), P = 0.02].
Baseline and recovery (10 and 60 min post) brachial FMD% and associated variables are detailed in Table 3, Low fit and High fit, for the lower and higher fit groups, respectively. For clarity, post hoc P values are reported only in the text. The ΔFMD% data are summarized in Fig. 1, which shows the change in FMD% from baseline during recovery (10 and 60 min post). Furthermore, individual responses in ΔFMD% are displayed in Fig. 2. Fig. 1.Delta flow-mediated dilation percentage (ΔFMD%) from baseline at 10 min post (A) and 60 min post (B) in control, moderate-intensity, and high-intensity exercise in both lower fit (open bars) and higher fit (dark bars) elderly individuals. Error bars represent SD. P ≤ 0.05, significant value. Post hoc analysis revealed the following: aP = 0.01, control 60-min ΔFMD% was significantly reduced compared with exercise; bP = 0.02, ΔFMD% significantly increased 10 min after moderate-intensity compared with high-intensity exercise; cP = 0.01, ΔFMD% significantly improved in the higher fit compared with the lower fit group 60 min after high-intensity exercise. Fig. 2.Mean (white squares) and individual (lines) ΔFMD% from baseline at 10 and 60 min after high-intensity (A and B), moderate-intensity (C and D), and control (E and F) protocols in both higher and lower fitness groups. #P ≤ 0.05, significant change from baseline FMD%. In both fitness groups, FMD decreased by 0.74% (95% CI, −1.34 to −0.03) after 60 min of recovery in control compared with baseline (P = 0.05). There was no effect of fitness on this response. There was a significant fitness × condition × time interaction for FMD% (P = 0.01). FMD% was significantly reduced compared with baseline following high-intensity exercise in the lower fit group at both 10 min [mean reduction of 0.85% (95% CI, 0.12–1.58), P = 0.02] and 60 min post [mean reduction of 0.72% (95% CI, 0.02–1.46), P = 0.05] (see Table 3, Low fit). In the higher fit group, a negligible change in FMD% was observed 10 min after high-intensity exercise [mean difference of 0.13% (95% CI, −0.73–0.98), P = 0.77]; however, there was a significant increase in FMD% compared with baseline after 60 min of 0.84% (95% CI, −0.12–1.69; P = 0.05) (see Fig. 1). The improved FMD% response following HIIE elicited a mean difference of 1.52% (95% CI, 0.41–2.62) after 60 min in the higher fit compared with the lower fit group (P = 0.01; Table 3, Low fit and High fit). In support of this difference between fitness groups, the Δ-change in FMD% after high-intensity exercise at 60 min was significantly correlated with V̇o2peak (r = 0.41; P < 0.01). Furthermore, in the higher fit group, FMD% was elevated after 60 min compared with moderate-intensity and control protocols [mean difference of 0.92% (95% CI, 0.05–1.78, P = 0.01) and 1.54% (95% CI, 0.65–2.42, P = 0.02) (Table 3, High fit). These changes in FMD% were also observed for absolute FMD (mm), with an increase 60 min following high-intensity exercise in the higher but not lower fit group (P = 0.04; Table 3, Low fit and High fit). FMD% increased significantly from baseline 10 min after moderate-intensity exercise [mean change of 0.86% (95% CI, 0.17–1.56), P = 0.02; Fig. 1] and returned to baseline levels after 60 min [mean difference to baseline of 0.30% (95% CI, −0.59–0.53)], with no effect of fitness on the response [mean between fitness group difference of 0.43% (95% CI, −0.28–1.13), P = 0.23; r = −0.13, P = 0.38]. Furthermore, the FMD% response 10 min after moderate-intensity exercise was increased compared with the high-intensity response [mean difference of 1.15% (95% CI, 0.58–1.72), P < 0.001] and control [mean difference of 1.23% (95% CI, 0.72–1.88), P < 0.001] in both fitness groups (Fig. 1). In the lower fit group, an increase in FMD% was observed 10 min after moderate-intensity exercise compared with the reduction observed after high-intensity exercise [mean difference of 1.34% (95% CI, 0.60–2.09), P < 0.001] and control [mean difference of 0.99% (95% CI, 0.23–1.75), P = 0.01] (Table 3, High fit). We also present covariate “adjusted FMD%” values (Table 3, Low fit and High fit). This analysis was consistent with our initial observations in FMD%, with a significant interaction among condition, fitness, and time (P = 0.04). Post hoc analysis revealed significant differences between the lower and higher fit groups 60 min after HIIE (P < 0.01). Resting blood flow was significantly elevated 10 min following both exercise protocols compared with control (P < 0.01) and was higher following high-intensity exercise compared with moderate-intensity [mean difference of 0.36 ml/s (95% CI, −0.03–0.66), P = 0.05]. There was no effect of fitness on the blood flow responses to exercise (P = 0.79) (Table 3, Low fit and High fit). Shear rate demonstrated a similar pattern where it was elevated 10 min after both exercise protocols compared with control (P = 0.01) and was higher immediately after high-intensity compared with moderate-intensity exercise [mean difference of 17.38 103/s (95% CI, −3.86–38.62), P = 0.01]. There was no effect of fitness on the shear rate responses after exercise (P = 0.78) (Table 3, Low fit and High fit). There was a condition × time interaction for heart rate, systolic blood pressure, and mean arterial pressure (Table 3, Low fit and High fit; P < 0.01). Heart rate was elevated by 9 beats/min (95% CI, 8–12) and by 13 betas/min (95% CI, 11–15) 10 min after moderate-intensity and high-intensity exercise, respectively, compared with rest. Mean arterial pressure was 5 mmHg (95% CI, 3–8) and 6 mmHg (95% CI, 3–9) higher 10 min after moderate- and high-intensity exercise, respectively, compared with rest. DISCUSSIONTo our knowledge, this is the first study to investigate the acute effects of exercise intensity and cardiorespiratory fitness on endothelial function in elderly men. The main findings from this study indicate that the acute effects of leg exercise on brachial FMD are dependent on both the intensity of exercise and cardiorespiratory fitness in elderly men. We observed an immediate increase in FMD following MICE that normalized after 60 min in both fitness groups. In contrast, FMD decreased immediately and 60 min following HIIE in the lower fit group, whereas FMD increased after 60 min in the higher fit group. We also observed reductions in FMD in both groups following prolonged rest during the control assessment. The FMD response to acute exercise is suggested to be biphasic (15), with an inverse relationship between exercise intensity and the recovery in brachial artery endothelium-dependent function observed in some (11, 33) but not all studies (3, 62). We attempted to capture the time-course response by measuring FMD immediately (10 min post) and 60 min after exercise in the elderly and found an exercise intensity-dependent decrease in endothelial function immediately after high-intensity exercise, which is consistent with previous findings in young (11, 33), hypertensive (39) and peripheral arterial disease patients (35). Conversely, we found an immediate increase in endothelial function after short-term moderate-intensity exercise, which has been observed in one (33) but not all (3, 11), studies in younger individuals and following 30 min of walking exercise in healthy middle-aged adults (13). The immediate improvement in FMD after MICE of 40% PPO in this study contrasts the finding of no change in FMD following cycling exercise at 50% maximal heart rate (HRmax), albeit in younger healthy individuals (11). This difference in findings may be due to the degree of baseline endothelial dysfunction in elderly compared with younger adults, with greater improvements in acute FMD observed after exercise in coronary artery disease patients with a lower baseline FMD (14). Moreover, the increase in FMD after moderate-intensity exercise normalized after 60 min, which is similar in younger adults (33). In line with the suggested effect of higher intensity exercise (>70% HRmax) on the biphasic FMD response, we observed an increase in FMD 60 min after HIIE compared with normalization of FMD after MICE in the higher fit elderly adults. This contrasts with a report by Currie et al. (13a), who found an increased FMD after both high- and moderate-intensity exercise in coronary artery disease patients. However, unlike the study by Currie and colleagues, our exercise protocols were duration and work matched, which is important as the dose of exercise affects FMD independent of intensity (33). Our study reports intensity-dependent, dose-matched differences in the biphasic FMD response in elderly adults. We provide further support that exercise intensity modulates acute endothelial function (3, 11, 19, 33) in healthy elderly adults. The rationale for assessing the acute response of endothelial function to exercise relates to the potential impact of repeated bouts of exercise on vascular adaptation (24), but whether the immediate increase or decrease in FMD after exercise in this study is important for future vascular adaptation in the elderly is unknown. Padilla et al. (45) suggested recurring periods of exercise-induced transient endothelial impairment may represent a beneficial stimulus that contributes to longer term improvements in vascular function and structure, a concept known as hormesis. That is, the initial challenge, e.g., acute reductions in FMD, leads to activation of beneficial adaptive processes (45). The acute exercise intensity-dependent reductions in FMD we observed in this study may be linked to the recent observation that HIIE training is likely more effective than MICE training in improving conduit artery endothelial function (50). Therefore, improving FMD immediately after moderate-intensity exercise (which normalized after 60 min) may not lead to beneficial long-term vascular adaptation with training. Interestingly, we observed that cardiorespiratory fitness modulates the biphasic response of FMD to high-intensity but not moderate-intensity exercise in the elderly. The sustained reductions in FMD in the lower fit individuals after high-intensity exercise may be the signal required for future vascular adaptation observed following training and increases in fitness (45, 65). Our study is the first to directly assess the effect of cardiorespiratory fitness levels on acute changes in FMD following exercise in the elderly. The positive relationship between exercise training and endothelial function is well established (41, 42), while cardiorespiratory fitness is related to training status (37) and can be modified through changes in routine physical activity (26, 43). In support of this, acute reductions in FMD have been reported in sedentary, but not active adults after both leg-press exercise (47) and maximal running (30). Whether the similarities observed in the reduced FMD response after HIIE in the present study reflect the low overall physical activity levels or the impact of low activity on reductions in cardiorespiratory fitness is not known. The acute changes of ~0.85% in FMD up to 60 min in this study are in line with previous studies that reported changes in FMD between 0.6 and 2.3% in young healthy and individuals with cardiovascular disease (11, 14, 30, 62). Our current understanding of the physiological significance in the magnitude of the acute, transient changes observed in FMD is limited, and we are guided by longitudinal evidence suggesting changes in FMD are associated with changes in cardiovascular risk, with an absolute increase in FMD of 1% associated with a ~9–17% reduction in cardiovascular risk, independent of traditional cardiovascular risk factors (23, 32). It is plausible that larger responses in acute FMD, such as the prolonged reductions in FMD observed in the lower fit after HIIE in this study, may lead to greater eventual vascular adaptations; however, this is yet to be established. The mechanisms responsible for exercise-induced, intensity-dependent changes in FMD have been proposed to include alterations in oxidative stress, inflammation, reactive oxygen species (19, 31), shear stress and shear pattern, blood pressure, baseline artery diameter, endothelin-1 expression (28), increased sympathetic nervous activity (27), or vasoconstrictors (15). As we did not assess the mechanisms of FMD changes, we can only speculate on the possible causes. We covariate controlled for exercise-induced changes in artery diameter and shear stress, so these are unlikely to be the causes of our observed differences. NO bioavailability (54) and shear stress patterns during exercise are known to directly contribute to changes in FMD (21, 70, 73). Large increases in brachial antegrade shear stress occur during cycling exercise (21) and are associated with improved FMD (73), while increases in oscillatory shear and/or retrograde flow lead to reductions in FMD (57). Increases in oscillatory flow are observed early during cycling exercise (21) but may also be augmented during interval exercise used in this study, due to the stop-start nature of the high-intensity modality. This may explain the immediate improvement in FMD after MICE compared with the reduced FMD immediately following HIIE. Reductions in FMD immediately after exercise of higher intensity, and not moderate-intensity exercise, may be related to the negative impact of induced hypertension on FMD (18, 40). We observed a larger increase in blood pressure during HIIE compared with MICE in this study, irrespective of fitness level. Interestingly, a training-associated protection against the drop in FMD exists following increases in blood pressure, albeit during resistance exercise (47), which may be linked to our observation of a prolonged reduction in FMD following HIIE in the lower but not higher fit individuals. Studies investigating the acute effect of exercise intensity on endothelial function do not commonly assess FMD across the same measurement period using a nonexercise control. This study is unique in that it offers the opportunity to assess changes in FMD during extended periods of sedentary time in the elderly. We observed a reduction in brachial artery FMD after ~120 min of “sedentary time” (baseline rest + protocol + recovery), which is not reported in younger individuals after 6 h of prolonged sitting (51). As sitting time increases all-cause and cardiovascular mortality risk in older adults (38), the vascular effects of prolonged sitting warrant investigation. In line with recent evidence (51), we showed that reductions in FMD with sedentary time can be attenuated with short-term moderate-intensity exercise. However, we also found that high-intensity exercise in lower fit individuals led to a similar decline in FMD to that of prolonged supine rest. This suggests that prescribing moderate intensity in lower fit elderly individuals might be considered before progressing to higher intensity exercise as cardiorespiratory fitness improves. A modest association exists between cardiorespiratory fitness and basal endothelial function, independent of age and health status (41). Similarly, aerobically trained middle-aged and older adults have preserved endothelial function compared with those who are sedentary (16, 17, 42, 48, 53); however, in this study investigating FMD in the elderly, there was no difference in resting brachial artery FMD between lower and higher fit groups. This may be due to normalized FMD in the higher fit following increases in artery diameter and structural remodeling observed with exercise training (36, 72) with a tendency for a larger arterial diameter in the higher fit compared with the lower fit group. It is also possible that a “ceiling” effect exists on basal FMD in the elderly, as no improvements in FMD were reported following short-term training in older, higher fit adults despite increases in V̇o2peak (20). Ischemic events typically occur in the elderly who have known cardiovascular risk factors and/or disease. It is known that regular physical activity and exercise training throughout the life span have cardioprotective and vascular effects. Recently, HIIE has become popular for its potential for additional cardiovascular benefits with shorter bouts of exercise, including improved endothelial function (50). Our findings highlight the exercise paradox, where those who are at the greatest risk of adverse responses to acute exercise have the most to gain from regular exercise (37). Elderly individuals with low fitness and endothelial function who exhibit further reductions in FMD 60 min after higher intensity exercise may be at increased, acute cardiovascular risk. The acute reduction in FMD was not observed following MICE irrespective of fitness level. Whether the acute reduction is necessary to induce vascular adaptation (see hormesis, discussed above) (45, 65) and represents a potential danger period where the vascular system may be less responsive to stress is unknown. However, higher fitness in this study did attenuate the reduction in FMD observed following HIIE, suggesting there may be an adaptive or tolerance response with improvements in cardiorespiratory fitness. However, in the elderly who are of a lower fitness and/or those who already exhibit vascular dysfunction, this type of exercise may need to be treated with caution due to the potential that vascular dysfunction is transiently exacerbated. Importantly, whether the differences in the FMD response to different acute exercise intensities reported here have longer term consequences on endothelial function and/or cardiovascular risk in healthy elderly individuals needs to be determined. In future studies, it would be interesting to have prolonged FMD measurements e.g., 2–24 h after exercise, to establish whether the biphasic pattern is delayed or persistent in the lower fit compared with higher fit individuals, particularly after high-intensity exercise. A limitation of our study is that we included controlled-hypertensive participants. Despite observing no difference in resting FMD between controlled-hypertensive and normotensive individuals, we cannot rule out the potential confounding influence of hypertension on the findings. Furthermore, we cannot rule out the potential influence of antihypertensive, statin, and β-blocker therapy on the current findings, and further work should focus on the direct impact of medication on acute postexercise FMD. As we have not reported the physical activity of participants, we cannot exclude the possibility that genetic or behavioral differences contribute to the different levels of fitness and the observed findings. We did not include measures of potential mechanisms involved in the changes in FMD we observed, and further studies are required to fully explain our findings of an interaction between exercise intensity and fitness on the acute FMD response to exercise. In conclusion, the present study illustrates the effect of exercise intensity on acute FMD responses in elderly men. Furthermore, we highlight the importance of cardiorespiratory fitness on the acute FMD response following high-intensity exercise. Increases in FMD after MICE normalized quickly. Conversely, there were prolonged increases in FMD after HIIE in those with a higher fitness, whereas lower fitness individuals exhibited sustained decreases in endothelial function. This decrease in FMD may represent the signal for an adaptive vascular response and/or endothelial fatigue in untrained elderly individuals. Further studies on the acute effects of exercise intensity on endothelial function will be important to establish if the same effect exists in elderly females and to investigate the link between changes in FMD with acute exercise and the potential for chronic adaptation with exercise training in the elderly. GRANTSThe study was funded by Australian National Health and Medical Research Council (NHMRC) Grants 1022752, 1021416, 1020955, 1003707, and 1000967 and the Queensland Government. J. Golledge holds a Practitioner Fellowship from the NHMRC (1019921) and a Senior Clinical Research Fellowship from the Queensland Government. The research of D. J. Green’s research is supported by an NHMRC Principal Research Fellowship (1090914). DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONST.G.B., M.P., M.W., F.D.R., J.G., and C.D.A. conceived and designed research; T.G.B., M.P., and M.W. performed experiments; T.G.B. analyzed data; T.G.B., D.J.G., and C.D.A. interpreted results of experiments; T.G.B. prepared figures; T.G.B. and C.D.A. drafted manuscript; T.G.B., F.D.R., J.G., D.J.G., and C.D.A. edited and revised manuscript; T.G.B., M.P., M.W., F.D.R., J.G., D.J.G., and C.D.A. approved final version of manuscript. REFERENCES
Page 21b-type natriuretic peptide (BNP) is released from ventricular cardiomyocytes and belongs to the family of cardiac natriuretic hormones. BNP secretion into the circulation is enhanced in pathophysiological conditions leading to cardiac cell stress, such as volume overload, myocardial infarction, inflammation and extreme physical exercise (4, 9, 23). The increase in BNP secretion during impaired ventricular function is used for diagnosing heart failure and for monitoring its therapy (16). BNP exerts natriuretic, diuretic, vasodilating, and renin-aldosterone-inhibitory effects (5, 19). Therefore, BNP is considered a potential beneficial counterpart to the deleterious neurohumoral activation accompanying heart failure, especially to the renin-angiotensin-aldosterone system and endothelin 1. Pharmacological administration of BNP using the synthetic analog Nesiritide was pathophysiologically a promising therapy in heart failure; nevertheless, it did not lead to improved clinical outcomes (17). In parallel, the angiotensin receptor neprilysin inhibitor LCZ696, which targets both the angiotensin type 1 receptor and the natriuretic peptide degrading peptide neprilysin, had remarkable beneficial effects (15). The difficulty to interpret these outcomes is mainly based on the incomplete understanding of the BNP pathophysiology. Besides these cardiovascular properties, BNP was recently found to exert also metabolic effects, such as the decrease in plasma glucose, reduced appetite, and increased energy expenditure (1, 8, 24). These data, taken together with the fact that BNP concentrations are reduced in patients with obesity and insulin resistance, open novel possibilities for testing the therapeutical usefulness of BNP also in patients with metabolic diseases (11, 25). Both heart failure and obesity are associated with increased activation of the pituitary-hypothalamic-adrenal axis and sympathetic nervous system (14). Data from basic research support both positive and negative BNP effects on cortisol and catecholamine secretion in vitro (3, 18). To date, little is known on the clinical impact of BNP on the pituitary-adrenal axis (18). In humans, administration of BNP in physiological dose during 25 min exerts sympathoinhibitory properties, but these effects are abolished by the increase of BNP concentrations to pharmacological levels (2). Nevertheless, this BNP administration was very short and it can be presumed that the neurohumoral adaption needs some time and might develop a new balance. In the present study, we use a randomized within-subject crossover design for testing the effects of intravenous BNP administration during 4 h on pituitary, thyroid, and adrenal hormone release in men. MATERIALS AND METHODSThe trial was approved by the local institutional review board and registered with EudraCT and clinicaltrials.gov (NCT01375153). Participants were 10 healthy men aged 21–29 yr (mean 24.3 yr), without any history of past or current heart, endocrine, liver, kidney, or malignant disease, and receiving no medications. All study participants gave signed informed consent and underwent a thorough clinical examination and basic laboratory tests before recruitment. The study was designed as a randomized, placebo-controlled, crossover, single-blinded (subject) trial and was performed in an academic medical center. Participants were examined in two study sessions separated by a minimum of 3 wk, where they received once placebo (0.9% NaCl) and once an intravenous infusion of 3 pmol·kg−1·min−1 BNP-32 during 4 h (Fig. 1A). BNP-32 containing amino acids 77–108 (Ser-Pro-Lys-Met-Val-Gln-Gly-Ser-Gly-Cys-Phe-Gly-Arg-Lys-Met-Asp-Arg-Ile-Ser-Ser-Ser-Ser-Gly-Leu-Gly-Cys-Lys-Val-Leu-Arg-Arg-His) of the proBNP prostructure was manufactured as an acetate salt from American Peptide Company (Sunnyvale, CA) and dissolved in 0.9% NaCl before administration. The study sessions started at 8:00 AM after overnight fasting, and two indwelling catheters were placed in the antecubital veins of the right and left forearm for infusions and blood sampling respectively. Blood samples were withdrawn at baseline and every 60 min afterwards. Heart rate and blood pressure were measured concomitantly. For avoiding significant BNP-induced changes in blood pressure (22), all participants remained in supine position throughout the study, until 20 min after the termination of the BNP infusion. The volume infused intravenously during the whole study session was 40 ml (10 ml/h) and the total amount of blood withdrawn was 188 ml. Fig. 1.A: schematic protocol of study sessions. B: profile of changes in plasma concentrations of NT-proBNP. Data are presented as means ± SE. ●, 3 pmol·kg−1·min−1 B-type natriuretic peptide (BNP); ○, placebo. Blood samples were obtained in native, heparin-containing tubes, in prechilled EDTA-containing tubes. and in glutathione-EGTA-containing tubes. After sampling, the EDTA-, heparin and glutathione-EGTA-containing tubes were immediately cooled on ice for ~10 min, and the native tubes were left in room temperature for ca 10 min. Afterwards, the samples were centrifuged at 3,000 rpm for 10 min and then frozen at −20°C for the later measurement of ACTH, NH2-terminal-proBNP (NT-proBNP), growth hormone (GH), prolactin, thyroid-stimulating hormone (TSH), free T4, free T3, cortisol, epinephrine, and norepinephrine. Assays were performed at the very end of the study in duplicates, with samples from both study days of each individual analyzed within one assay. All measurements were performed in the certified laboratory of the General Hospital of Vienna (http://www.kimcl.at). Prolactin, cortisol, TSH, free T4, and free T3 were measured using ECLIA, with sensitivity and coefficient of variations (CVs) being 0.05 ng/ml and 4% for prolactin, 0.04 µg/dl and 6% for cortisol, 0.01 µU/ml and 4% for TSH, 0.3 pg/ml and 5–10% for free T3, and 0.02 ng/dl and 5–7% for free T4, respectively. GH and ACTH were determined by an assay based on CLIA, with sensitivity and CVs being 0.01 ng/ml and 7.5% for GH and 5 pg/ml and 6% for ACTH. Norepinephrine and epinephrine measurements were performed by HPLC. Sensitivity was 15 ng/l for both hormones, and CV ranged from 3.7% to 5.9% for norepinephrine and from 4% to 6.5% for epinephrine. The NTproBNP measurements were performed using the Elecsys 2010 immunoassay platform (intra-assay precision <4% CV and interassay precision <5% CV; Roche Diagnostics, Rotkreuz, Switzerland). Sample size was calculated based on the assumptions α = 0.05 and β = 0.2. The primary hypothesis is that BNP changes the hormone levels by at least 20%. Assuming a variability of repeated measurements of hormones of ~15% (SD), a minimum difference of 19% can be detected with an 80% power in a sample of 10 participants. Data were analyzed using the statistical software IBM SPSS Statistics 20. All outcome parameters are expressed as means ± SE. The differences between the two study days were tested by repeated measurements-analysis of variance (RM-ANOVA). The interaction between time and treatment (time × treatment) was considered the term of interest. P < 0.05 was considered statistically significant. When appropriate (significant RM-ANOVA results), post hoc paired t-tests were performed for testing intervention-induced changes at specific time points. RESULTSThe profile of changes in plasma BNP concentrations during the study sessions has been previously published (24). Baseline BNP concentrations were 13.0 ± 1.5 pg/ml in placebo sessions and 13.3 ± 2.7 pg/ml in BNP sessions. Placebo did not influence plasma BNP levels, and BNP concentrations at the 4-h time point were 13.7 ± 1.5 pg/ml. The intravenous administration of BNP increased circulating BNP concentrations over 26-fold after the first hour (to 350 ± 27.8 pg/ml), and even further later on, reaching a 34- to 35-fold increase in BNP levels during the second, third, and fourth hour. This increase of BNP to mean levels between 350 and 466 pg/ml was not accompanied by significant changes in blood pressure, as participants remained in supine position throughout the study (Table 1) (22). Nevertheless, in accordance with previous studies using a similar protocol, BNP continuously increased the heart rate from 58.7 ± 2.9 beats/min at baseline to 68.7 ± 3.7 beats/min at the 240-min time point, with respective values in placebo sessions being 63.5 ± 2.9 beats/min at baseline and 61.2 ± 3.1 beats/min at 240 min (RM-ANOVA P = 0.019) (Table 1) (8).
Circulating levels of NT-proBNP continuously decreased over time in both placebo and BNP sessions (effect of time in RM-ANOVA P = 0.032), but there were no differences between both study days (effect of time × treatment in RM-ANOVA P = 0.46) (Fig. 1B). We further investigated the profile of changes in circulating levels of pituitary hormones under BNP infusion. There was a tendency toward an increase in plasma ACTH (P = 0.04 for absolute differences in plasma ACTH between the 4-h time point and baseline), but this was not significant when tested using RM-ANOVA (P = 0.2) (Fig. 2A). BNP did not influence the levels of GH, prolactin and TSH (Fig. 2, B–D). There were also no differences in free T3 and free T4 concentrations (data not shown). Fig. 2.Effects of BNP on pituitary hormone concentrations. Profile of changes in plasma concentrations of ACTH (A), growth hormone (GH; B), prolactin (C), and thyroid-stimulating hormone (TSH; D). Data are presented as means ± SE ●, 3 pmol·kg−1·min−1 BNP; ○, placebo. BNP inhibited the physiological circadian decline in cortisol levels during the late morning hours, leading to significantly increased plasma concentrations of cortisol (RM-ANOVA P = 0.022, Fig. 3A). Final cortisol concentrations at the 4-h time point were 9.6 ± 0.8 µg/dl in placebo sessions and 14.4 ± 1.6 µg/dl in BNP sessions (post hoc test P = 0.043). BNP increased also circulating epinephrine and norepinephrine concentrations (RM-ANOVA P = 0.018 and P = 0.036 respectively, Fig. 3, B and C). Fig. 3.Effects of BNP on circulating cortisol and catecholamine concentrations, Profile of changes in plasma concentrations of cortisol (A), epinephrine (adrenaline; B), and norepinphrine (noradrenaline; C). Data are presented as means ± SE ●, 3 pmol·kg−1·min−1 BNP; ○, placebo. DISCUSSIONIn addition to the well-known diagnostic and prognostic utility in the clinical praxis, BNP has been investigated as a therapeutic tool in patients with heart failure. Nevertheless, the BNP effects extend beyond the cardiovascular system and include also glucose-lowering, lipolytic, and anorectic properties. Here we show that administration of BNP in a pharmacological dose increases circulating cortisol and catecholamine concentrations in the absence of changes in pituitary and thyroid hormones. Cross-sectional studies reveal a strong correlation of plasma BNP concentrations with circulating epinephrine and cortisol levels in patients with heart failure (6). Nevertheless, several experiments investigating the direct BNP effects on zona fasciculata cells have revealed not only positive, but also negative BNP effects on steroidogenesis and cortisol secretion (18). One recent publication demonstrates direct inhibitory effects of BNP on both biosynthesis and release of cortisol in human adrenocortical cells in primary culture (13). In this context, the increase in serum cortisol levels following BNP administration in men is highly likely not a direct effect but the consequence of other neurohumoral changes induced by BNP, such as natriuresis, which is a well-known BNP effect not measured in the present study (5). In contrast, the mechanisms underlying the adrenergic effects of BNP are well explained in several basic and translational studies. BNP promotes catecholamine release not only from rat pheochromocytoma PC12 cells but also from sympathetic nerve endings of guinea pigs (3). This is achieved via an inhibition of phosphodiesterase type 3, which increases intracellular cyclic AMP levels activating the protein kinase A pathway (3), In addition, systemic administration of BNP increases cardiac sympathetic activity and exerts pro-arrhythmogenic effects in mice, and these effects are reduced by β1-adrenergic blockade (21). All these data are in line with our findings on elevated circulating catecholamine concentrations and increased heart rate in men. Previous studies have shown that the BNP effects on blood pressure depend on posture, with hypotensive effects observed in subjects remaining in sitting position and no effects observed in subjects remining in supine position (22). Our experimental setup with subjects remaining in supine position throughout the study sessions avoided a BNP-induced fall in arterial blood pressure, aiming to minimize the compensatory changes in the autonomous nervous system. Nevertheless, we cannot exclude a secondary baroreceptor-mediated sympathetic activation in response to the BNP-induced reduction in right ventricle filling pressure or compensatory effects following the well-known renal effects of BNP such as in Ref. 5. Previous clinical studies on this topic have investigated only the impact of short-term BNP administration in healthy volunteers (2, 10). Brunner-La Rocca et al. (2) found out that infusion of pharmacological BNP concentrations during 25 min reduced renal and whole body norepinephrine spillover in patients with heart failure, but not in healthy subjects, where an increase in renal norepinephrine spillover was observed. Also, the administration of 2 pmol·kg−1·min−1 BNP during 2 h did not significantly change plasma catecholamines (10). The main difference between these works and our study is the length of BNP administration, as we continuously infuse BNP during 4 h and the profile of changes in circulating catecholamines displays significant changes starting from the third hour. Thus our results are in line with the data of Brunner-La Rocca et al. (2) over the initial time period of 2 h but extend the information showing that the neurohumoral balance changes in the long term. Taken together, the BNP-induced increases in circulating catecholamines, together with previous basic, translational, and clinical studies, corroborate a complex role of BNP on the sympathetic nervous system, which might depend on the amount of BNP increase, on the time of BNP administration, and on the function/dysfunction of the endogenous BNP system, with confirmed adrenergic effects during long-term BNP administration in both patients and healthy volunteers (2, 3, 7, 10, 12). The increase in stress hormones following BNP infusion could aim to compensate the well-known diuretic and natriuretic effects of BNP. These findings become clinically significant during pharmacological BNP administrations and might also play a role in the neurohumoral activation accompanying heart failure, as Lainchbury et al. (12) using a similar protocol, described a BNP-induced increase in cortisol in patients with impaired left ventricular function. Increased cortisol and epinephrine levels are independent predictors of increased mortality risk in patients with heart failure; therefore, therapeutic regimens that target the BNP system exerting positive cardiovascular effects in the absence of enhanced cortisol and catecholamine concentrations would be optimal in this population (7). We recently showed that intravenous BNP infusion induced a 33% decrease in proANP concentrations, most likely due to a decrease in the endogenous production of ANP (20). In the present study we find a mild but consistent decrease in NT-proBNP over time in all study days but no differences between placebo and BNP sessions. It is important to note that the study subjects were healthy young volunteers with very low NT-proBNP baseline levels, which got even lower during the relaxed study atmosphere in supine position, possibly reaching the lowest threshold. Based on our data, we cannot exclude significant effects of BNP administration on NT-proBNP in subjects with higher baseline NT-proBNP secretion. Taken together, we conclude that pharmacological administration of BNP increases circulating cortisol, epinephrine, and norepinephrine concentrations in healthy subjects. These properties of BNP should be further investigated in conditions associated with increased BNP secretion and might impact the outcome of therapies targeting the BNP system, as both cortisol and epinephrine levels are independent predictors of increased cardiovascular mortality risk. GRANTSThis work was supported by Austrian National Bank Research Grant 13583 (to M. Clodi). DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONSG.G., M.R., B.B.H., and G.V. performed experiments; G.G., M.R., B.B.H., A.L., M.C., and G.V. analyzed data; G.G., M.R., B.B.H., M.H., A.L., and M.C. edited and revised manuscript; G.G., M.R., B.B.H., M.H., A.L., M.C., and G.V. approved final version of manuscript; M.R. and G.V. prepared figures; M.H., A.L., M.C., and G.V. interpreted results of experiments; G.V. drafted manuscript. We thank A. Hofer (Division of Endocrinology and Metabolism, Medical University of Vienna) for excellent technical assistance. REFERENCES
Page 22skeletal muscle is a viable metabolic organ that is impacted by use and disuse (4, 6, 12). Because of immobilization, disuse, and a decline in physical activity, skeletal muscle atrophy is one of the most prominent adaptations to occur following spinal cord injury (SCI; 4, 12, 13, 27, 28). This adaptation is noted by decreased thigh cross-sectional area (CSA) of skeletal muscle up to 30–50% within a few weeks postinjury compared to able-bodied controls (4, 12). The decrease in skeletal muscle CSA is linked to several metabolic and cardiovascular diseases in persons with SCI. Therefore accurate quantification of muscle size may be prophylactically used as a simple tool to predict comorbidities after SCI. Individuals with SCI have increased fat mass and decreased lean mass compared with matched able-bodied controls (10, 28). These changes in body composition expose individuals to an increased risk of developing chronic metabolic disease, insulin resistance, type 2 diabetes mellitus, and cardiovascular disease (1, 9, 10, 23, 24). Considering the changes in body composition and the lack of access to expensive body composition equipment, body weight was previously used to derive prediction equations for whole body, trunk, and leg lean mass, with body weight being the least predictive method for leg lean mass in persons with SCI (11). Because of the large leg muscle mass and extensive muscle atrophy following SCI, accurately quantifying muscle size is becoming increasingly important in clinical and research settings. Several imaging techniques have been used to quantify whole body and regional mass, including magnetic resonance imaging (MRI), dual-energy X-ray absorptiometry, ultrasound, and computed tomography (CT). These techniques have provided accurate measurements of skeletal muscle CSA and fat mass deposition in persons with SCI (4, 7, 12, 28). Currently, MRI and CT are considered the gold standards to measure skeletal muscle CSA and adipose tissue infiltration on the basis of their high accuracy and reproducibility (2, 9, 21, 26, 29). However, the use of these techniques is expensive, time consuming, requires a medical facility, and necessitates a high level of training. Moreover, MRI requires identifying the correct anatomical regions of interest, as well as the infiltration of intramuscular fat (IMF) based on signal intensity, to accurately quantify muscle size. On the other hand, CT exposes individuals to radiation (21, 26), making MRI a more viable option for studies that involve numerous scans and capturing multiaxial slices. However, the aforementioned factors may hinder the use of MRI and CT for many clinicians and researchers, making these techniques impractical, especially in studies that require large sample sizes or recurrent measurements at varying time points. Anthropometrics, including circumferential and skinfold measurements, provide an inexpensive and viable alternative to MRI and have been effectively used to predict muscle size in other clinical populations (17, 19, 22). However, there is only one study that has validated the use of anthropometrics to measure muscle size against MRI after SCI (20). This study showed a ~70% difference in estimation of thigh muscle size between MRI and anthropometrics, with the anthropometric measurements overestimating muscle size twofold compared with MRI (20). Therefore developing an accurate field method to predict thigh muscle size specific to the SCI population could identify risk for many of the comorbidities associated with SCI. Additionally, accurate assessment of thigh muscle CSA without the use of MRI or CT would allow for the optimization of therapeutic interventions, similar to resistance training, while saving time and resources. For example, muscle hypertrophy has been noted using MRI in the knee extensors (35%) and whole thigh muscle group (28%) over the course of 12 wk of neuromuscular electrical stimulation resistance training in persons with chronic SCI (14). With increased participation in sports, field events, and exercise interventions, the SCI population could benefit from a field equation that may act as a marker not only for health but also for training status. The purpose of the current study was to validate the use of anthropometrics, including skinfolds and thigh circumference measurements, against MRI to predict muscle CSA of the thigh after accounting for IMF and femur bone CSAs. This will allow for the development of a field equation to predict the CSA of skeletal muscle in the midthigh of individuals with chronic SCI. On a translational level, developing this field equation will provide clinicians with a method to accurately quantify muscle size following SCI. METHODSTwenty-two men were recruited with ages between 18 and 50 yr old and with motor complete SCI C5–L2, American Spinal Injury Association Impairment Scale (AIS) A or B. Subject characteristics are presented in Table 1. All participants were part of a clinical trial registered at https://clinicaltrials.gov (NCT01652040). Only men were included because they were part of a study investigating the effects of testosterone replacement therapy in men with SCI. Fourteen participants were classified as having paraplegia (level of injury T3–T11), and eight were classified as having tetraplegia (level of injury C5–C7). Participants with preexisting medical conditions were excluded. These included cardiovascular disease, uncontrolled type 2 diabetes and those on insulin, pressure sores stage 2 or greater, and urinary tract infection or symptoms. The exclusion criteria were primarily established to reduce the heterogeneity commonly observed in SCI studies. Moreover, the aforementioned comorbidities in the SCI population are likely to cause an additional impact on muscle size, making it difficult to predict muscle size. After meeting inclusion criteria, participants provided written informed consent. All study procedures were approved by the McGuire Veterans Affairs Medical Center Institutional Review Board and conducted according to the Declaration of Helsinki.
All anthropometric measurements were performed on the right thigh, with the participants lying supine on a flat mat, by the same trained investigator. Thigh circumference (Thighcircum) was measured middistance between the anterior superior iliac spine and superior border of the patella in triplicate using a standard inflexible measuring tape (Executive Diameter Pocket Tape Measure; Lufkin). Subcutaneous fat thickness (SFT) was measured in triplicate at the same location using a Harpenden skinfold caliper (Baty International). Measurements were reported to the nearest 0.1 cm and repeated until the three measurements were within 0.5 cm of one another. A single vertical and anterior midline skinfold measurement was chosen to represent the midthigh in this study according to the American College of Sports Medicine guidelines for exercise testing and prescription (25). This is a feasible and clinically acceptable measurement to acquire in the SCI population. Multisite measurements are often unattainable because of extensive mobility limitations, and there are risks with placing individuals with a high level of SCI in various positions similar to prone. Knowing that circumference is equal to 2πr, where π = 3.14, the radius (r) of the whole thigh is equal to Thighcircum/2π. r=Thighcircum/2π(1) To estimate the radius of the whole muscle CSA (muscle CSAanthro), SFT was subtracted from the radius of the whole thigh. radius of muscle CSAanthro=radius of whole thigh−SFT/2(2) SFT was divided in half to represent the subcutaneous fat layer superficial to the muscle. On the basis of Eq. 1, anthropometrically predicted whole thigh CSA (thigh CSAanthro) is equal to πr2. On the basis of Eq. 2, muscle CSAanthro was calculated using the following equation: muscle CSAanthro = π [r – (SFT/2)]2. Magnetic resonance images were obtained from a General Electric Signa 1.5-T magnet (fast spin echo; repetition time, 850–1,000 ms; echo time, 6.7 ms; imaging frequency, 63.8 MHz; echo number, 1; echo train length, 3; flip angle, 90°; field of view, 20 cm; matrix size, 256 × 256). Approximately 12–15 transaxial images, 8 mm thick and 16 mm apart, were taken from the hip joint to the knee joint using a General Electric body array flex coil to measure thigh CSA. By using a localized coil, the signal-to-noise ratio is improved, resulting in high-resolution images for analysis. The acquisition time per leg was 3.5 min. Prior to each scan, the participant's legs were strapped together to mitigate involuntary muscle spasms, and participants were provided earplugs to minimize the noise of the magnet. Images were analyzed using WinVessel software (Ronald Meyer, Michigan State University). To distinguish muscle from fat, the outer perimeter of the thigh muscle group was manually traced, and the pixel signal intensity was automatically determined via the software. A bimodal histogram segmentation was plotted that contained two distinct peaks, with the first peak representing the threshold for the muscle and the second peak representing the threshold for the fat. The midpoint value of the two peaks was utilized to separate muscle from fat, as has been previously described in detail (12, 14, 16). Regions of interest were manually traced including the whole thigh CSA [thigh CSAMRI; thigh CSA = muscle CSA + subcutaneous adipose tissue (SAT CSAMRI); Fig. 1A] and whole skeletal muscle CSA (muscle CSAMRI; Fig. 1B), with these methods both including bone and IMF. The absolute skeletal muscle (muscle CSA-IMFMRI) was determined on the basis of signal intensity and excluded IMF and the femur bone. SAT CSAMRI was determined by measuring the CSA of the subcutaneous adipose tissue, defined as the area between the outside of the muscle CSAMRI and inside of the thigh CSAMRI. Bone CSA was measured by tracing outside of the cortical bone (Fig. 1C). Six sequenced transverse-axial slices were chosen to represent the midthigh and were analyzed and then averaged. To evaluate test-retest reliability, the coefficient of variability of muscle CSAMRI and bone CSA were found to be 0.5 and 0.7%, respectively. Fig. 1.Representative T1-weighted MRI of the whole thigh CSA (A), muscle CSAMRI (B), and bone CSA (C). ImageJ software (National Institutes of Health, Bethesda, MD) was used to measure subcutaneous adipose thickness (SATT). The anterior SATT was measured as the linear distance from outside of the subcutaneous fat boundary to the outside of the muscle boundary. We measured the SATT at four polar sites (four-site SATT; anterior, medial, lateral, and posterior) of the thigh and then compared the average of the four SATT measurements against the anterior SFT value taken via skinfold caliper. A single anterior SATT linear distance measurement obtained from MRI was then validated against the anterior SFT measurement. To determine differences in estimating muscle size using a single-site SFT vs. multisite SFT measurements, the regression equation derived from the polar four-site SATT as measured by MRI and SFT analysis was utilized to emulate using multiple-site measurements. This value was then used to determine the muscle CSApolar anthro, where muscle CSApolar anthro = π [r – (polar four-site SATT)]2. Polar four-site SATT was determined using the prediction equation derived from the relationship between a single-site SFT vs. the polar four-site SATT. Linear regression analyses were used to identify associations between anthropometric and MRI measurements. The standard error of the estimate (SEE) was calculated in SPSS to evaluate the accuracy of the prediction equations using the following equation: SEE=∑(Y−Y′)2N−2 where Y refers to individual data, Y′ is the mean of the data, and N is the sample size. Bland-Altman analyses (3) were used to determine the limits of agreement between anthropometric and MRI measurements. Statistical significance was accepted at P < 0.05. All values are presented as means ± SD. Statistical analyses were performed using SPSS, version 23.RESULTSSubject demographics are presented in Table 1. Participants ranged in age from 18 to 50 yr old and had a body mass index (BMI) range of 17.0–31.5 kg/m2. Age, height, weight, BMI, and time since injury were not statistically significant between Caucasians and African Americans or between paraplegic and tetraplegic participants. Means ± SD for anthropometrics and MRI are presented in Table 2. Thigh CSAanthro was correlated with thigh CSAMRI (r2 = 0.90, SEE = 17.6 cm2, P < 0.001; Fig. 2A). Muscle CSAanthro was correlated with muscle CSAMRI (r2 = 0.78, SEE = 16.6 cm2, P < 0.001; Fig. 2B) and muscle CSA-IMFMRI (r2 = 0.75, SEE = 17.6 cm2, P < 0.001; Fig. 2C), respectively. A correlation was found between SFT and the single-site SATT (r2 = 0.75, SEE = 0.38 cm, P < 0.001; Fig. 2D) and the polar four-site SATT (r2 = 0.78, SEE = 0.37 cm, P < 0.001; Fig. 2E), respectively. Muscle CSApolar anthro was correlated with muscle CSA-IMFMRI (r2 = 0.76, SEE = 16.6 cm2, P < 0.001; Fig. 2F). SFT correlated with SAT CSAMRI (r2 = 0.71, P < 0.001; Fig. 2G) with the slope revealing that for every 1-cm increase in SFT there is an ~20 cm2 increase in SAT.
Fig. 2.Linear regression (solid line) and the line of identity (dashed line) comparing MRI and anthropometric predictions for thigh CSAMRI and thigh CSAanthro (A), muscle CSAMRI and muscle CSAanthro (B), muscle CSA-IMFMRI and muscle CSAanthro (C), single-site SATT and anterior SFT (D), polar four-site SATT and anterior SFT (E), muscle CSA-IMFMRI and muscle CSApolar anthro (F), anterior SFT and SAT CSAMRI (G), and muscle CSA-IMFMRI and muscle CSApredicted including the correction for IMF and bone (H). Bland-Altman plots in Fig. 3 reveal a high level of agreement between anthropometrics and MRI, with anthropometrics leading to overestimations compared with muscle CSAMRI (mean bias = 4.0 cm2; Fig. 3A), muscle CSA-IMFMRI (mean bias = 24.8 cm2; Fig. 3B), anterior single-site SATT (mean bias = 0.8 cm; Fig. 3C), and polar four-site SATT (mean bias = 0.2 cm; Fig. 3D). Fig. 3.Bland-Altman plots between anthropometric measurements and MRI for muscle CSAanthro and muscle CSAMRI (A), muscle CSAanthro and muscle CSA-IMFMRI (B), anterior SFT and single-site SATT (C), and anterior SFT and polar four-site SATT (D). The difference between the values obtained by MRI and the values obtained from the anthropometric measurements is plotted on the y-axis, and the average of the two measurements is plotted on the x-axis. The solid line represents the mean difference between measurements, with the dashed lines representing 2 SD above and below the mean difference. Utilizing the average femur CSA and average IMF CSA derived from MRI led to the field equation developed herein. The bone CSA measured by MRI was 6.39 ± 0.98 cm2 (range 4.7–8.7 cm2), and the IMF CSA measured by MRI was found to be 14.43 ± 10.42 cm2 (range 2.4–47.2 cm2). The following field equations were derived to predict muscle CSA (muscle CSApredicted) utilizing a single midthigh circumference measurement and a single-site anterior skinfold measurement. Equation 3 estimates the whole muscle CSA without excluding IMF and bone CSAs. Equation 4 provides an estimate for the absolute muscle CSA and includes a correction factor to account for IMF CSA and bone CSAs. Muscle CSApredicted=π[(Thighcircum/2π)−(SFT/2)]2(3) Muscle CSApredicted=π[(Thighcircum/2π)−(SFT/2)]2−23.2(4) The correction factor 23.2 is derived as the sum of 14.43 cm2 (IMF CSA), 6.38 cm2 (bone CSA), and 2.4 cm2 (constant); where 2.4 cm2 is derived from the constant acquired after regressing muscle CSA-IMFMRI vs. muscle CSAanthro (Fig. 2C). The anthropometric midthigh circumference and anterior skinfold measurements were utilized in the field equation containing the correction factor and yielded a relationship between muscle CSApredicted and muscle CSA-IMFMRI (r2 = 0.76, SEE = 17.3 cm2, P < 0.001; Fig. 2H). DISCUSSIONThe current study demonstrates that thigh circumference and skinfold thickness can be used to predict thigh skeletal muscle CSA in persons with chronic motor complete SCI. We found strong relationships between MRI and anthropometric measurements and were able to derive equations to estimate midthigh muscle CSA and whole thigh CSA specifically for individuals with chronic SCI. The novelty of the current work originates from the fact that there is no clinical field equation specifically for persons with SCI that provides an accurate estimate of muscle CSA. The field equation presented provides clinicians with a simple method to estimate skeletal muscle size using easily reproducible anthropometric measurements without relying on expensive equipment. Moreover, the SCI-specific field equation accounted for bone and IMF CSAs. A slight overestimation (3.5%) of muscle CSA was found when comparing anthropometrics with MRI, although this method did not account for IMF or bone. The muscle CSA-IMFMRI was overestimated (21.5%) when using anthropometrics. Muscle CSApolar anthro, which emulated multisite skinfold measurements, led to an overestimation (14.0%) of muscle CSA-IMFMRI, although to a lesser extent than muscle CSAanthro, which was derived using only a single anterior SFT measurement. The overestimations noted suggest that anthropometrics alone may need a correction factor to quantify absolute muscle CSA. With notable muscle atrophy and increased IMF infiltration following SCI (8, 12), these findings emphasize the significance of accounting for IMF to accurately quantify muscle size and further reiterate the importance of the correction provided in the field equation. Utilization of the muscle CSApredicted equation including the correction factor for IMF and bone revealed a strong relationship and minimal (<1%; Table 2) overestimation of muscle CSA-IMFMRI when applying the anthropometric measurements into the equation. Further modeling may need to be considered based on weight and whole body fatness to predict IMF. Similar findings were highlighted when we determined the relationship between MRI findings of measuring visceral adiposity and waist circumference (15). Waist circumference was only able to predict subcutaneous adipose tissue but not visceral adiposity (15). Previous reports have also yielded overestimations of muscle when using anthropometric measurements to predict muscle size in able-bodied (18, 22, 29) as well as SCI individuals (20). Since anthropometrics are unable to account for the bone or IMF, the equation derived in this study subtracts the average femur CSA and average IMF CSA as determined by MRI. To our knowledge, this is the first time that the bone CSA and IMF CSA have been accounted for in an anthropometrically based prediction equation estimating skeletal muscle CSA after SCI. In the able-bodied population, Housh et al. analyzed individual muscle groups in the thigh via MRI to predict quadriceps CSA (r = 0.86, SEE = 5.2 cm2), hamstring CSA (r = 0.75, SEE = 3.5 cm2), and total thigh muscle CSA (r = 0.86, SEE = 9.5 cm2) and derived estimation equations for each (17). Utilizing the published Housh et al. regression equations derived from MRI (17), Mathur et al. were able to estimate the quadriceps CSA, hamstring CSA, and total thigh muscle CSA in older individuals with chronic obstructive pulmonary disease (22). The results of this study showed that anthropometric measurements, including thigh circumference and skinfold thickness, were not sensitive enough to provide an accurate representation of muscle size in older adults with chronic obstructive pulmonary disease (22). The polar four-site SATT was correlated to the anthropometric SFT (r2 = 0.78), with SFT being overestimated (11.7%) compared with the polar four-site SATT derived via MRI. When comparing SFT with the single-site SATT, SFT was systematically overestimated (41.5%) compared with SATT. These results align with a previous report from Tothill and Stewart, who found that SFT did not correspond to a single measurement of SAT as well as multisite measurement (29). SFT correlated with SAT CSAMRI (r2 = 0.71), with the slope of the linear equation derived from this comparison revealing that for every 1-cm increase in SFT there is an ~20 cm2 increase in SAT. The strong correlations and slope presented suggest that SFT can provide an estimate of SAT in the SCI population. It is worth noting that adding the polar four-site SATT to our prediction equation resulted in a similar correlation (Fig. 2F) but decreased the overestimation of absolute muscle CSA from 21.5 to 14% (Table 2). In line with our results and suggesting the feasibility of the approach presented herein, one substantial finding of the Layec et al. study was a strong correlation (r2 = 0.89) of the thigh skeletal muscle volume estimated by anthropometry vs. MRI in controls and SCI individuals (20). However, Layec et al. noted a dramatic overestimation of thigh muscle volume (~70%) when utilizing anthropometrics vs. MRI in the SCI population (20), compared with the 21.5% overestimation in the absolute muscle CSA noted in the current study. It is possible that the small sample size (n = 8 SCI, n = 8 controls) may have contributed to this discrepancy and introduced a source of error in the prediction of muscle volume. A larger homogenous sample size may be needed to develop a field method to be used with the SCI population. Furthermore, we utilized CSA to quantify muscle size, whereas Layec et al. used volumetric measurements, potentially explaining the discrepancy in results. Also, there are changes in muscle size and intramuscular and subcutaneous fat from the proximal to the distal portion of the upper leg. We chose to only analyze the midthigh to correspond to the anthropometric measurements as opposed to the entire upper leg, which could explain the variation in results. Our anthropometrically based results assume that the muscle of the thigh is a cylindrical shape and that the subcutaneous adipose tissue is distributed in a uniform manner around the circumference of the muscle. The variation in the polar four-site SATT values presented in Table 2 reveals that this assumption is unlikely, especially in individuals following SCI who have notable muscle atrophy and increased intramuscular and subcutaneous fat deposition (8, 14). Another limitation is that only men were included in the current study. Therefore the generalizability of the results should be considered with caution, and additional studies should be conducted to test the validity of this field equation in women with SCI. We chose to utilize one anterior SFT measurement instead of taking multisite measurements, as this measure is more feasible and realistic in a clinical setting. It has been observed that multisite skinfold measurements provide a more accurate estimation of subcutaneous adipose tissue (29); therefore this could limit the credence of the results provided. Taking accurate multisite skinfold measurements is challenging and often unfeasible in the SCI population because of limited mobility, spasticity, skin sensitivity, and increased risk of adverse events such as autonomic dysreflexia. To display any potential limitations of using a single anterior SFT measurement vs. multisite skinfold measurements, we utilized the polar four-site SATT regression equation to derive the muscle CSApolar anthro estimation and compared that against the muscle CSA-IMFMRI (Fig. 2F). Similar studies have been performed in able-bodied populations (2, 5, 17, 22) and generally have larger pools of participants than has been reported in the current work. The SCI population available for research studies being invariably lower than the population available for able-bodied studies and significant differences within the population including comorbidities, level of injury, and time since injury are considered limitations in generalizing our results. This study demonstrated that thigh circumference and skinfold thickness could be used to predict skeletal muscle CSA in persons with chronic motor complete SCI. The novelty of the current work originates from the fact that the prediction equation provides an estimate of absolute muscle CSA after accounting for IMF and bone CSAs. Correction for IMF and bone CSAs decreased the initial overestimation of absolute muscle CSA from 21.5 to 0.6%, suggesting that anthropometrics may be sensitive enough to predict muscle size after including a correction factor. With additional validation the proposed equation has the potential to mitigate barriers specific to the SCI population and provides an effective estimation to be used clinically. GRANTSThis work was supported by the Office of Research and Development, Medical Research Service, Department of Veterans Affairs Grant B7867-W. DISCLOSURESThere are no conflicts of interest declared by the authors. AUTHOR CONTRIBUTIONSR.C.W. and A.S.G. performed experiments; R.C.W. and A.S.G. analyzed data; R.C.W. and A.S.G. interpreted results of experiments; R.C.W. and A.S.G. prepared figures; R.C.W. and A.S.G. drafted manuscript; R.C.W. and A.S.G. edited and revised manuscript; R.C.W. and A.S.G. approved final version of manuscript; A.S.G. conceived and designed research. We thank all of the participants and Refka Khalil for participant recruitment and research coordination. We are grateful to Dr. Laura O'Brian for assistance with the editing process and Pamela Moore for help with MRI analysis. REFERENCES
Page 23AbstractElderly white, thin, nonsmoking women appear to be more susceptible to lung infections with Mycobacterium avium complex and other nontuberculous mycobacteria (NTM). It has been postulated that such disease in women is related to suppression of their cough. We hypothesized that patients with pulmonary NTM (pNTM) infections may have altered cough physiology compared with unaffected control subjects. We used capsaicin-induced cough to assess the cough reflex in pNTM subjects. Eight elderly white women with stable chronic pNTM infections and six unaffected age-matched control subjects were recruited. There was no significant difference between groups in capsaicin-elicited cough motor response, airflow pattern, or cough frequency. The urge-to-cough (UTC) score at the lowest capsaicin concentration was significantly lower in pNTM than control subjects (P < 0.05). There were no significant differences in the UTC score between pNTM and control subjects at >50 μM capsaicin. These results suggest lower UTC sensitivity to the lowest concentration of capsaicin in pNTM than control subjects. In other words, the pNTM subjects do not sense a UTC when the stimulus is relatively small. NEW & NOTEWORTHY This study investigates the cough motor response and cough sensitivity in patients with nontuberculous mycobacteria (NTM) infection. In elderly white female pulmonary NTM subjects, we demonstrated a capacity to produce coughs similar to that of age-matched control subjects but decreased cough sensitivity in response to a low dose of capsaicin compared with control subjects. These findings are important to understand the pathophysiological mechanisms resulting in NTM disease in elderly white women and/or the syndrome developing in elderly white female NTM patients. pulmonary disease due to chronic infection with Mycobacterium avium complex in elderly women with no underlying pulmonary disorders was labeled “Lady Windemere’s syndrome” by authors who hypothesized that the pathogenesis involved habitual voluntary suppression of cough due to fastidiousness (21). Although this appellation has been criticized as incorrect from the medical and the literary perspective, it continues to be used to refer to thin elderly women with pulmonary infections from any nontuberculous mycobacteria (NTM) (13, 15). The role of cough suppression has been debated, but not delineated by data (16, 20). Cough is an important airway defensive behavior to expel foreign materials and excessive secretions from the respiratory tract to decrease the risk of aspiration. A typical cough is characterized by three phases: 1) inspiratory phase, during which the lung is inflated by a large burst of activity of the diaphragm and inspiratory intercostal muscles, 2) compression phase, during which the expiratory muscles contract against a closed glottis, and 3) expiratory phase, during which the vocal folds suddenly open and expiratory muscles contract, generating a rapid acceleration of expiratory airflow followed by sustained expiratory airflow to expel materials from the airways (9). Ineffective cough is a result of impaired cough initiation and efficacy of cough motor activity (1). To achieve an efficient clearance of the airways, an accurate urge-to-cough (UTC) sensation and a functional cough motor pattern are necessary. Impaired cough initiation may be caused by reduced cough reflex sensitivity, which has been reported in smokers and patients with cystic fibrosis (6, 7, 14). In this pilot study we hypothesized that subjects with stable pulmonary NTM (pNTM) disease have decreased cough reflex sensitivity and ineffective cough airflow patterns compared with controls. We used capsaicin-induced reflex cough to assess the UTC and cough motor pattern. MATERIALS AND METHODSThe study was reviewed and approved by the Institutional Review Board of the University of Florida. Eight female subjects >65 yr of age with pNTM infections and six age-matched control subjects without respiratory symptoms or disease provided informed written consent to participate in the study. The diagnosis of pNTM infection was confirmed by a pulmonologist (K. Fennelly) on the basis of a review of symptoms, imaging, and microbiology following the 2007 guidelines from the American Thoracic Society and the Infectious Disease Society of America (10). All participants underwent standard spirometry prior to the cough physiology testing (Table 1). The apparatus for the cough physiology measurements is illustrated in Fig. 1. Subjects were seated and breathed through a facemask connected to a pneumotachograph for recording of airflow. The nebulizer (DeVilbiss Healthcare, Somerset, PA) that delivered the aerosolized solution during inspiration, with delivery duration of 2 s, was controlled by a KoKo dosimeter (nSpire Health, Longmont, CO). The KoKo dosimeter was triggered by a negative inspiratory pressure, and aerosolized solution was delivered into the facemask for 2 s after the dosimeter was triggered. Participants were asked to initially perform three voluntary coughs and then take a deep inspiration to trigger the dosimeter for delivery of the aerosolized solution. Four concentrations of capsaicin (50, 100, 200, and 500 μM) plus placebo (0 μM capsaicin) were presented, three times each, to the subject in a blinded-randomized block order. Each capsaicin presentation was separated by ∼1 min. At the end of each capsaicin presentation, subjects were asked to rate their UTC, using the modified Borg scale, from 1 (no UTC) to 10 (maximum UTC) (5, 23).
Fig. 1.Experimental setup. All signals were stored and digitized using a PowerLab data acquisition system (ADInstruments, Colorado Springs, CO). A cough was defined as an inspiration followed by a compression phase and then a high expiratory airflow. The total cough count (CTot) was determined by counting all cough airflow events for 1 min after each capsaicin presentation. Cough parameters were measured from the airflow waveforms and are illustrated in Fig. 2. Fig. 2.Cough airflow response to 200 μM capsaicin in elderly women with nontuberculous mycobacterial (NTM) infection. The following parameters were computed from the initial cough within an epoch: compression phase duration (in s), peak expiratory airflow rate (PEFR, in l/s); peak expiratory airflow acceleration time (PEAAT, in s); cough volume acceleration (PEFR/PEAAT, in l·s−1·s−1), a measure of cough effectiveness; expiratory plateau phase duration (in s); and cough expiratory flow volume (in liters). Cough parameters and UTC were analyzed using one-way ANOVA. The significance criterion for all analyses was set at P < 0.05. RESULTSThere were no significant differences in voluntary cough and capsaicin-elicited reflex cough airflow parameters between pNTM and control groups (Table 2). UTC at the lowest capsaicin concentration (50 μM) was significantly lower in pNTM than control subjects (P < 0.05). There were no significant differences in UTC between pNTM and control subjects when capsaicin concentration was >50 μM (Fig. 3). The pNTM subjects showed a trend (P = 0.1) for CTot compared with controls in the 50 μM capsaicin trials (Fig. 4). There were no significant differences in CTot between pNTM and control subjects when the capsaicin concentration was >50 μM (Fig. 4).
Fig. 3.Urge to cough in NTM (n = 8) and control (n = 6) groups. Values are means ± SE. *P < 0.05. Fig. 4.Cough number in NTM (n = 8) and control (n = 6) groups. Values are means ± SE. DISCUSSIONCough can be evoked reflexively or voluntarily, suggesting that the medullary cough central pattern generator can be affected by peripheral afferent stimuli and suprapontine descending inputs. Reflex cough is mainly elicited by stimulation of peripheral airway receptors and processed by the brain stem neural network to generate a cough motor action (2, 3). Voluntary cough involves higher brain cortical and subcortical regions to plan the initiation of motor action and consciously control the cough pattern (1). Reflex cough does not require suprapontine involvement, but these suprapontine regions can send projections to the brain stem for modulation and regulation of cough motor pattern (11, 12). This study examined both voluntary and reflex cough airflow patterns in pNTM and age-matched healthy control subjects. The results show no significant differences in voluntary or capsaicin-elicited cough airflow patterns between pNTM and control subjects. Voluntary cough is a type of cue-elicited cough that subjects were told to produce by an instructor. The capsaicin-elicited reflex cough is elicited when subjects cough in response to inhalation of capsaicin. Since voluntary and capsaicin-elicited cough are elicited by extrinsic stimuli or commands, we suggest that pNTM and control subjects have similar ability to produce an extrinsically elicited cough. These results do not address the sensitivity and pattern for intrinsic spontaneous cough. UTC is defined as one type of respiratory sensation of a need to cough. Mechanical and/or chemical irritations activate cough receptors located in the upper and lower airways that trigger the brain stem neural network for the neurogenesis of cough motor action. These peripheral airway stimuli can be gated into the suprapontine system to elicit discriminative (spatial, temporal, and intensity of stimuli) and affective (evaluative and emotional components of stimuli) aspects of the UTC sensation (4, 18, 22). Experiments based on functional MRI have shown that capsaicin-elicited cough activated sensory, cognitive, and motor neural elements in cortical and subcortical regions (17–19). Activation of somatosensory cortex, insular cortex, posterior parietal cortex, and dorsolateral prefrontal cortex is important (17–19). Recruitment of orbitofrontal cortex and cingulate cortex is likely to be involved in shaping subjects’ affective response to airway stimuli and the cognitive UTC response (8, 17, 18). Activation of the primary motor cortex represents the voluntary motor drive to cough (17–19). In this study UTC in 50 μM capsaicin trials was lower in pNTM than control subjects, but there was no significant difference between pNTM and control subjects at >50 μM capsaicin. In other words, pNTM subjects did not feel a UTC with low levels of stimulation but seemed to have normal responses to higher levels of stimulation. These data suggest that pNTM subjects have an increased sensitivity threshold to stimuli. This increased sensory threshold will decrease the stimuli to be gated into suprapontine cortical and subcortical regions; therefore, these pNTM subjects were not less motivated to cough, nor were they consciously controlling their coughs. This suggests that the pNTM subject may be at higher risk of pulmonary aspiration of small volumes. We hypothesize that one reason for recurrent pNTM infection and exacerbations is the reduced UTC for low-stimulus airway clearance coughing. In addition, pNTM subjects were diagnosed with bronchiectasis, which is associated with retained secretion in the airways. The retained secretion, which covered the surface of respiratory tract, may also blunt the sensitivity of receptors lying on the respiratory tract in response to capsaicin stimuli. The blunted sensitivity of the receptors may result in the lower UTC in pNTM than control subjects. However, high concentrations of capsaicin provide sufficient stimulus of the UTC so that pNTM and control subjects have the same degree of UTC and cough, demonstrating that the pNTM subjects can sense and produce cough in response to higher levels of stimulation. In contrast, patients with neurological diseases such as strokes or amyotrophic lateral sclerosis may not be able to protect their airway from such large-volume insults. Furthermore, there were no significant differences in cough number with capsaicin-induced cough stimulation. Coughing is a series of motor movements that can be controlled by descending motor pathways and/or spinal reflexes. We suggest that, with inhalation of suprathreshold concentrations of capsaicin, subjects in both groups would produce coughs reflexly preceded by sensory and cognition processes in the suprapontine regions. This might be the reason for the same cough motor outcomes, including cough airflow pattern and cough number, in pNTM and control subjects. A strength of this study is that the women were well matched for age and region of residence. However, because of their shorter stature and greater body weight, the body mass index of our pNTM subjects was higher than that of other cohorts, notably that of Kim et al. (16). The major limitation of this study is that we cannot determine if our findings are a result of the pNTM disease or preceded the pNTM disease. It may be that a subset of elderly women is destined to be susceptible to pNTM disease by virtue of having a lower UTC. Future studies done on larger numbers of elderly women may be able to resolve this issue. Because of the high intensity of time and labor involved with studying the cough physiology of these research participants, we designed this as a pilot study. Thus, another limitation is the small number of subjects. However, we submit that our findings provide support for pursuing this research in a larger cohort. This is the first study to investigate the cough sensitivity and motor response in NTM patients. The results suggest that elderly white women with pNTM and control subjects have similar capacity to produce coughs, but they do not sense the UTC when the intensity of “cough stimuli” is low. This study implies that these elderly white women with pNTM might have blunted airway afferent sensation and/or reduced central neural sensory processing, resulting in an increase in the capsaicin stimulation threshold for processing of the UTC and motivating the pNTM patient to cough and clear her airway. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSH.-W.T., K.F., and P.W.D. conceived and designed the research; H.-W.T., K.W.H., S.A., J.A.C., J.L.H., and P.W.D. performed the experiments; H.-W.T., K.W.H., S.A., J.A.C., J.L.H., and P.W.D. analyzed the data; H.-W.T., K.F., K.W.H., S.A., J.A.C., and P.W.D. interpreted the results of the experiments; H.-W.T. prepared the figures; H.-W.T. drafted the manuscript; H.-W.T., K.F., K.W.H., S.A., J.A.C., J.L.H., and P.W.D. edited and revised the manuscript; H.-W.T., K.F., K.W.H., S.A., J.A.C., J.L.H., and P.W.D. approved the final version of the manuscript. Present address of P. Fennelly: Division of Intramural Research, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD 20892. REFERENCES
Page 24AbstractInspiratory muscle training (IMT) has consistently been shown to reduce exertional dyspnea in health and disease; however, the physiological mechanisms remain poorly understood. A growing body of literature suggests that dyspnea intensity can be explained largely by an awareness of increased neural respiratory drive, as measured indirectly using diaphragmatic electromyography (EMGdi). Accordingly, we sought to determine whether improvements in dyspnea following IMT can be explained by decreases in inspiratory muscle electromyography (EMG) activity. Twenty-five young, healthy, recreationally active men completed a detailed familiarization visit followed by two maximal incremental cycle exercise tests separated by 5 wk of randomly assigned pressure threshold IMT or sham control (SC) training. The IMT group (n = 12) performed 30 inspiratory efforts twice daily against a 30-repetition maximum intensity. The SC group (n = 13) performed a daily bout of 60 inspiratory efforts against 10% maximal inspiratory pressure (MIP), with no weekly adjustments. Dyspnea intensity was measured throughout exercise using the modified 0–10 Borg scale. Sternocleidomastoid and scalene EMG was measured using surface electrodes, whereas EMGdi was measured using a multipair esophageal electrode catheter. IMT significantly improved MIP (pre: −138 ± 45 vs. post: −160 ± 43 cmH2O, P < 0.01), whereas the SC intervention did not. Dyspnea was significantly reduced at the highest equivalent work rate (pre: 7.6 ± 2.5 vs. post: 6.8 ± 2.9 Borg units, P < 0.05), but not in the SC group, with no between-group interaction effects. There were no significant differences in respiratory muscle EMG during exercise in either group. Improvements in dyspnea intensity ratings following IMT in healthy humans cannot be explained by changes in the electrical activity of the inspiratory muscles. NEW & NOTEWORTHY Exertional dyspnea intensity is thought to reflect an increased awareness of neural respiratory drive, which is measured indirectly using diaphragmatic electromyography (EMGdi). We examined the effects of inspiratory muscle training (IMT) on dyspnea, EMGdi, and EMG of accessory inspiratory muscles. IMT significantly reduced submaximal dyspnea intensity ratings but did not change EMG of any inspiratory muscles. Improvements in exertional dyspnea following IMT may be the result of nonphysiological factors or physiological adaptations unrelated to neural respiratory drive. inspiratory muscle training (IMT) has been studied extensively in healthy individuals and patients with chronic respiratory diseases, but the efficacy of this intervention remains controversial (20, 28). Systematic reviews have concluded that IMT improves whole body exercise performance using a range of performance-based exercise tests but does not improve peak aerobic capacity or maximal work rates during incremental exercise tests (11, 14). The purported improvements in exercise performance are thought to be related, at least in part, to reductions in exertional dyspnea ratings (39). Although IMT can reduce dyspnea during both performance-based and maximal incremental exercise tests in healthy subjects (11, 32), the physiological mechanisms for this improvement have not been adequately studied. Previous research in health and disease has demonstrated a strong relationship between diaphragmatic electromyography (EMGdi), an indirect measure of neural respiratory drive (NRD), and dyspnea intensity ratings (8, 16, 23, 36). Moreover, the ratio between NRD and the mechanical output of the respiratory system (i.e., neuromechanical coupling) is thought to be an important contributor to both the intensity and qualitative dimensions of exertional dyspnea (26). It follows that improvements in EMGdi and neuromechanical coupling of the respiratory system can reduce dyspnea. Indeed, it has been shown previously that bronchodilator-induced improvements in neuromechanical coupling in COPD are correlated with improvements in dyspnea during exercise (26). IMT may decrease the relative electrical activation of the diaphragm and improve neuromechanical coupling of the respiratory system to perform a given ventilatory task. Recent evidence also suggests that extradiaphragmatic inspiratory muscles, such as the scalene and sternocleidomastoid muscles, are heavily recruited during IMT (29). Thus, reductions in dyspnea following IMT may also be related to changes in the electrical activation of extradiaphragmatic inspiratory muscles. Accordingly, the purpose of this study was to determine whether IMT reduces exertional dyspnea intensity ratings in healthy subjects and whether improvements in dyspnea are related to improvements in inspiratory muscle electromyography (EMG) and neuromechanical coupling of the respiratory system. We hypothesized that IMT would reduce dyspnea intensity ratings at submaximal work rates, which would coincide with a reduction in EMG of the diaphragm, scalene, and sternocleidomastoid muscles with corresponding improvements in neuromechanical coupling of the respiratory system. METHODSTwenty-five young and healthy males participated in this study (NCT02243527; https://clinicaltrials.gov/). All subjects provided written, informed consent, and all procedures were approved by the Providence Health Care Research Ethics Board at the University of British Columbia and adhered to the Declaration of Helsinki. Inclusion criteria were as follows: male; self-reported physical activity levels as “moderate” or “high” according to the International Physical Activity Questionnaire (4); spirometry within normal limits; and able to read and understand English. Exclusion criteria were as follows: current or former smokers; history or current symptoms of cardiopulmonary disease; participating in a competitive endurance sport at the provincial, national, or international level; ulcer or tumor in the esophagus, a nasal septum deviation, or recent nasopharyngeal surgery; allergies to latex or local anesthetics; and contraindications to exercise testing. Subjects were randomly assigned to either an IMT (n = 12) or sham control (SC) training (n = 13) program. Subjects completed three experimental visits and 5 wk of training. Visit 1 (V1) served as a detailed familiarization and screening visit where each subject performed pulmonary function testing and a maximal incremental cycle exercise test. Visits 2 (V2) and 3 (V3) served as pre- and postintervention measurements, respectively, and consisted of the same exercise test performed on V1. Detailed ventilatory, EMG, respiratory mechanical, and sensory responses were measured throughout exercise on V2 and V3. Those in the IMT group were told that they were a part of a “respiratory muscle strength training” intervention, whereas the SC group were told they were part of a “respiratory muscle endurance training” intervention. Both IMT and SC groups performed their respective training with a POWERbreathe K3 device (HaB International, Southam, Warwickshire, UK). The K3 model is a variable flow-resistive device that employs an electronically controlled valve to apply a variable resistance over the course of inspiration. The IMT group trained 5 days/wk for 5 wk at two sessions/day (morning and evening). Each session included 30 sharp inspiratory efforts from residual lung volume. The initial intensity was set at 50% of the participant’s maximal inspiratory pressure (MIP), which was determined on V2. Participants in the IMT group were instructed to increase the training intensity freely, such that they were training at a 30-repetition maximum intensity. The POWERbreathe K3 included an inspiratory muscle warmup for the first four breaths, which were completed at an intensity less than the target training intensity. The next 26 repetitions were completed at the target intensity. Any repetition that failed to meet the target intensity did not count toward the total completed repetitions for that session. The SC group also trained for a total of 5 wk but at a fixed intensity of 10% of MIP (from V2) once/day for a total of 60 repetitions, 5 days/wk. Each training breath was described as being a slow, protracted, deliberate breath. Previously, a 6-wk intervention at 15% of MIP has been shown to be an effective sham protocol that elicits no training effect (32). Both groups performed one supervised session/wk in the laboratory to monitor MIP and to gauge the appropriateness of their training technique and intensity. The training intensity for the IMT group was increased if mouth pressures were <50% of current MIP or if subjects were, at the discretion of the research team, performing training at lower than a 30-repetition maximum. Spirometry, plethysmography, and maximal inspiratory pressure (MIP) from residual volume were collected according to previous recommendations (1, 24, 45) using a commercially available testing system (Vmax Encore 229, V62J Autobox; CareFusion, Yorba Linda, CA) and expressed in absolute terms and relative to predicted values (10, 43, 47). Handgrip strength was measured before and after training using a handheld dynamometer (model 76618; Lafayette Instrument Company, Lafayette, IN). This measurement assessed each participant’s motivation to perform a maximal voluntary contraction. The maximal handgrip strength test involves muscle groups unaffected by IMT. Therefore, an improvement in MIP without an increase in handgrip strength was indicative of a physiological change in MIP, as opposed to a greater voluntary effort in performing MIP maneuvers. Subjects performed incremental exercise tests to exhaustion on an electronically braked cycle ergometer (VIAsprint 200P; Ergoline, Bitz, Germany). Each test began with 6 min of steady-state rest, followed by a 1-min warmup of unloaded pedaling. The incremental cycle test began at 25 W and increased by 25 W every 2 min until volitional exhaustion. Participants pedaled at a freely chosen cadence (>60 rpm), and all ergometer measurements (i.e., saddle height, saddle position, handlebar angle/height) were recorded and reproduced for all subsequent exercise tests. Surface electrodes were used to assess EMG of the sternocleidomastoid (EMGscm) and scalene (EMGsca) muscles, as described previously (29). Briefly, bipolar electrodes on the scalenes were placed within the posterior triangle of the neck at the level of the cricoid cartilage (37). Electrodes were placed along the long axis of the sternocleidomastoid muscle between the mastoid process and the medial clavicle (38). The electrode placement was on the right side of the body and recorded in reference to anatomic landmarks to ensure consistency in electrode placement between visits. A wireless surface EMG system (TeleMyo DDTS; Noraxon USA, Scottsdale, AZ) was used to evaluate both EMGscm and EMGsca. A combined esophageal electrode balloon catheter was used to measure EMGdi, which was connected to a bioamplifier (bioamplifier model RA-8; Yinghui Medical Technology, Guangzhou, China) (36). The catheter was inserted through the nares after application of a topical anesthetic (Lidodan Endotracheal Spray; Odan Laboratories, Montreal, QC, Canada). Position of the catheter was determined when the EMG amplitude was lowest in the center pair and highest at the electrode pairs furthest from the center during spontaneous breathing (19). The catheter was placed at the same depth and in the same nostril on both visits. All EMG data were sampled at 2 kHz, and the signals were further digitally processed using LabChart 7.3.7 Pro software (ADInstruments, Colorado Springs, CO) with a band-pass filter between 20 and 500 Hz. All raw EMG data were converted to a root mean square using a time constant of 100 ms and a moving window. In an effort to improve the signal-to-noise ratio in the EMGsca and EMGscm signals, the average root mean square during expiration at baseline was subtracted from all subsequent EMG data. All EMG data were expressed as a percentage of maximal EMG activity achieved during any inspiratory capacity maneuver performed at rest or during exercise for a given experimental visit. Standard metabolic and ventilatory responses were measured on a breath-by-breath basis using a commercially available metabolic cart (Vmax Encore 229; CareFusion). Heart rate and arterial oxygen saturation were measured using a heart rate monitor (Polar T34; Polar Electro, Kempele, Finland) and pulse oximeter (Radical-7 Pulse CO-Oximeter, Masimo, Irvine, CA), respectively. The inspiratory capacity maneuvers used for EMG normalization purposes were also used to calculate end-expiratory and end-inspiratory lung volumes (9) and were expressed as a percentage of total lung capacity. Neuromechanical coupling of the respiratory system was determined as the ratio between EMGdi (%max) and VT (%vital capacity). Breathing discomfort, defined as “a feeling of labored or difficult breathing,” and leg discomfort, defined as “the feeling of fatigue in [their] leg muscles,” were measured using the modified 0–10 category ratio Borg Scale (2). The end points of the scale were anchored such that zero represented “no breathing/leg discomfort at all” and 10 represented “the most intense breathing/leg discomfort [they] have every experienced or could ever imagine experiencing.” Additionally, at the end of exercise, participants were asked to first state their primary reason for stopping exercise (i.e., breathing discomfort, leg discomfort, a combination of the two, or another reason) and then to choose applicable qualitative descriptors of breathlessness using a modified version of a previously published questionnaire (40) that we have used previously (3). The esophageal catheter used to measure EMGdi includes an esophageal balloon, which was connected to a calibrated differential pressure transducer (model DP15-34; Validyne Engineering, Northridge, CA) to measure esophageal pressure. The total work of breathing (Wb) was determined as the area within an averaged tidal esophageal pressure-volume loop, including a portion of a triangle that fell outside of the loop representing part of the elastic work of breathing (22). The Wb was then multiplied by breathing frequency. All physiological exercise variables were averaged in 30-s epochs. The time between 60 and 90 s of each 2-min stage was designated as our primary period of data collection. During this time, participants were reminded to look straight forward, minimizing any head or neck movement, keep a loose grip on the handlebars, and to avoid talking or swallowing to minimize contamination of our outcomes of interest. The data obtained during this period were then linked to the breathing and leg discomfort ratings and inspiratory capacity values that were collected during the final 30 s of each stage (i.e., from 90 to 120 s). Analyses of EMG and Wb were performed by a blinded assessor. Blinding was achieved through assigning a random identifier to each stored data file during analysis and removing the original file name that identified the subject group and visit. This was done to ensure neutrality when processing data that may involve a bias of selection from the assessor, such as EMGdi and esophageal pressure-volume loops for assessing the Wb. Following analysis, all files were renamed to their original identifier. An initial sample size calculation was performed on the basis of previous work (36) showing a decrease in EMGdi by 10%max correlated with a difference in dyspnea by 1 Borg unit with an α of 0.05. This calculation yielded a sample size of 11 subjects/group to detect a significant decrease in EMGdi. Statistical tests were performed using SPSS (Version 21; IBM, Armonk, NY). Baseline comparisons of pulmonary function and exercise responses between groups were made using unpaired t-tests. Pre- vs. postcomparisons of subject characteristics, pulmonary function, anthropometry, and exercise measurements were made using paired t-tests. Between-group differences in the pre- and postchanges in dyspnea, leg discomfort, ventilatory responses, and all EMG-derived variables across work rates were tested using repeated-measures analysis of variance with a Greenhouse-Geisser correction where appropriate. The between-subject factor tested was SC vs. IMT group, and the within-subject factor was work rate. First, the interaction term was evaluated. Because no significant interaction effects were observed in the current data, with the exception of MIP, only the between-subject factors were considered. These t-tests were performed on data collected at rest, standardized absolute work rates, and the highest equivalent work rate (HEWR) completed by an individual on both visits and at peak exercise, where peak exercise was defined as the highest work rate maintained for ≥30 s. Changes in qualitative descriptors of dyspnea and reasons for stopping exercise were performed with a paired McNemar’s test. Significance was set at P < 0.05, and all data are presented as means ± SD. RESULTSSubject characteristics and pulmonary function are presented in Table 1. Preintervention groups were well matched for age, mass, pulmonary function, and physical activity levels. There were no changes in self-reported physical activity when comparing pre- vs. postintervention in both groups. Peak exercise data can be found in Table 2. There were no group differences in any baseline peak exercise responses. There were no statistically significant changes in any peak exercise responses following IMT or SC training.
Adherence to interventions was good in both groups. The IMT group completed 94 ± 9% whereas the SC group completed 88 ± 13% of assigned training sessions. Successful training sessions were monitored digitally by the POWERbreathe device as well as via a diary kept by the subject. In the event of a discrepancy between the diary and POWERbreathe recordings of successful sessions, the POWERbreathe data were used. This occurred in only one subject. MIP significantly increased after 5 wk in the IMT group but not in the SC group (see also Table 1). Esophageal catheter-derived measurements were obtained in 11 IMT and 11 SC subjects. EMGdi during exercise before and after training is shown in Fig. 1, A and D. There were no significant changes in EMGdi following training in either the IMT or SC group at rest or during exercise. The scalenes were relatively inactive throughout the early stages of exercise, with EMGsca activity increasing at the HEWR and peak exercise in both groups (Fig. 1, B and E). However, there were no statistically significant changes in EMGsca in the IMT or SC groups from baseline. The EMGscm displayed a similar response with low levels of activity up to the HEWR in both groups (Fig. 1, C and F). A statistically significant decrease in EMGscm was observed at 50 W in the IMT group, but there were no significant changes at any other work rate. No changes in EMGscm were observed in the SC group. Fig. 1.Inspiratory muscle electromyography during exercise. Dashed lines within each graph and for each group connect the 150-W data point to the highest equivalent submaximal work rate (HEWR) and the HEWR to the peak exercise data point. EMGdi, diaphragm electromyography; EMGsca, scalene electromyography; EMGscm, sternocleidomastoid electromyography; IMT, inspiratory muscle training; SC, sham control. *P < 0.05. There were no significant changes in minute ventilation, tidal volume, breathing frequency, or operating lung volumes at any work rate in either group following training (Fig. 2). Similarly, there was no change in the total Wb or neuromechanical coupling for a given exercise intensity in either group (Fig. 3). Fig. 2.Ventilatory responses during exercise. Dashed lines within each graph and for each group connect the 150-W data point to the highest equivalent submaximal work rate (HEWR) and the HEWR to the peak exercise data point. Gray shaded regions represent tidal volume (i.e., the difference between end-expiratory lung volume and end-inspiratory lung volume). TLC, total lung capacity; IMT, inspiratory muscle training; SC, sham control. Fig. 3.Total work of breathing and neuromechanical coupling of the respiratory system during exercise. Dashed lines within each graph and for each group connect the 150-W data point to the highest equivalent submaximal work rate (HEWR) and the HEWR to the peak exercise data point. EMGdi, diaphragm electromyography; VT, tidal volume; VC, vital capacity; IMT, inspiratory muscle training; SC, sham control. Figure 4 shows the sensory intensity responses in both the IMT and SC groups. Subjects in the IMT group reported significantly lower dyspnea ratings at 125 W (pre: 2.2 ± 1.4 vs. post: 1.6 ± 1.5 Borg units, P < 0.05), 150 W (pre: 3.2 ± 1.5 vs. post: 2.3 ± 1.4 Borg units, P < 0.01), and the HEWR (pre: 7.6 ± 2.5 vs. post: 6.8 ± 2.9 Borg units, P < 0.05) after IMT, with no changes in the SC group. There were no changes in leg discomfort ratings in either group. There were no significant changes in the reasons for stopping exercise and the qualitative descriptors of dyspnea upon exercise cessation in either group following training (data not shown). Fig. 4.Perceived breathing and leg discomfort during exercise. Dashed lines within each graph and for each group connect the 150-W data point to the highest equivalent work rate (HEWR) and the HEWR to the peak exercise data point. IMT, inspiratory muscle training; SC, sham control. *P < 0.05. DISCUSSIONThis study is the first to comprehensively examine the neurophysiological mechanisms associated with reduced dyspnea ratings following IMT in healthy human subjects. The main findings are as follows. 1) Dyspnea intensity ratings were modestly reduced during submaximal exercise intensities in the IMT but not in the SC group, and moreover, IMT had no effect on leg discomfort ratings throughout exercise or on the qualitative dimensions of exertional dyspnea at maximal exercise; 2) IMT had no effect on inspiratory muscle EMG; and 3) IMT had no effect on ventilatory responses or neuromechanical coupling of the respiratory system during incremental cycle exercise. Collectively, these results suggest that modest improvements in dyspnea intensity ratings following IMT are not explained by improvements in key physiological outcomes known to contribute to dyspnea in health and disease. Systematic reviews in healthy populations indicate a beneficial effect of IMT on dyspnea ratings (11). Despite this finding, it must be acknowledged that not all studies demonstrate a positive effect of IMT on dyspnea (42, 46), and some studies, like ours, show only modest (i.e., <1 Borg unit) decreases in dyspnea (25). Nevertheless, several mechanisms have been proposed to explain the apparent improvement in dyspnea following IMT. The most commonly cited but as of yet untested mechanism is that motor outflow (i.e., “neural respiratory drive”) decreases for any given level of minute ventilation following IMT (12, 21). This is a reasonable hypothesis given that dyspnea intensity ratings during exercise are explained largely by an increased awareness of NRD (15, 16, 30, 36). Huang et al. (13) demonstrated that improvements in inspiratory muscle strength following IMT correlated with a reduction in inspiratory motor command measured using mouth occlusion pressure at 0.1 s. This correlation likely reflects a decrease in the percentage of motor units required for a given ventilatory task. However, the study by Huang et al. (13) did not include a control group and did not evaluate NRD during exercise or examine its association with exertional dyspnea. To our knowledge, the present study is the first to examine the effects of IMT on NRD during exercise in healthy humans. The results of our study suggest that IMT does not affect NRD, as indirectly estimated using invasive measures of crural EMGdi during incremental cycling to exhaustion. A conference abstract based on 10 COPD patients (7 IMT and 3 controls) demonstrated reductions in both EMGdi (by 12%) and dyspnea intensity ratings (by 3.3 Borg scale units) at standardized ventilations during constant load cycling following 8 wk of IMT (17). One potential explanation for this discrepancy is the fact that these COPD patients had baseline inspiratory muscle weakness and may have derived greater benefits from IMT compared with healthy subjects that were not limited during exercise by dyspnea and that had normal baseline inspiratory muscle strength. It is possible that strengthening the inspiratory muscles beyond a certain point confers no additional advantage in reducing NRD during the hyperpnea of exercise. Neuromechanical uncoupling of the respiratory system is thought to be an important contributor to the intensity and qualitative dimensions of dyspnea (26, 27). In general, the mechanical output of the respiratory system increases proportionally to the level of NRD during exercise in healthy humans. However, when VT becomes constrained or reaches a plateau/inflection, then the ratio between EMGdi (%max) and VT (%vital capacity) begins to rise. This often leads to intolerable dyspnea and gives rise to the sensation of “unsatisfied inspiration,” particularly in patients with chronic respiratory disease that have severe mechanical constraints on VT expansion (27). Given the lack of change in EMGdi and VT in the present study, it is not surprising that there was no change in our measurement of neuromechanical coupling following IMT. There is some evidence to suggest that increases in dyspnea may be associated with extradiaphragmatic inspiratory muscle activity (6). Recent work also suggests that traditional inspiratory muscle strength training protocols, as used in the present study, tend to preferentially recruit extradiaphragmatic inspiratory muscles, particularly those in the neck (29). Thus, improvements in global inspiratory muscle strength following IMT may be due to improvements in extradiaphragmatic muscle strength (e.g., intercostal muscles, scalenes, sternocleidomastoids, etc.). We attempted to address this question by examining changes in EMGscm and EMGsca during exercise. Although we observed a statistically significant reduction in EMGscm at one submaximal work rate, this was not sustained throughout exercise, and there were no changes in EMGsca at any exercise intensity. Based on these data (Fig. 1), we argue that there were no physiologically meaningful changes in EMGscm or EMGsca following IMT. We have suggested previously that diaphragmatic recruitment can be increased significantly if subjects consciously engage the diaphragm during IMT (29). We did not employ this approach in the present study to facilitate comparisons with the majority of the IMT literature in healthy subjects. Whether conscious recruitment of the diaphragm during IMT would confer greater benefits on dyspnea and perhaps reduce EMGdi requires further investigation. Collectively, the results of this study show that IMT has no effect on any physiological measurement related to the hyperpnea of exercise across the full range of ventilations (i.e., rest to maximal exercise). Based on this observation, we speculate that modest decreases in dyspnea intensity ratings following IMT may reflect some form of desensitization to repeated inspiratory muscle loading over several weeks, changes that could not be captured in our physiological measurements during exercise. However, this remains speculative and requires further investigation. Interestingly, we observed no difference in leg discomfort ratings in the current study. This is inconsistent with the findings of Romer et al. (32) during incremental cycling but is consistent with the findings of Verges et al. (44) during constant load cycling at 85% of peak work rate. If IMT provided a desensitization effect on dyspnea, it stands to reason that our IMT protocol would not impact leg discomfort ratings because IMT would not provide a similar effect to the locomotor muscles. The effect of IMT on perceived leg discomfort remains inconclusive based on the available literature. Discrepancies among studies are likely related to varying exercise testing protocols, IMT regimes, and/or differences in subject characteristics (e.g., fitness level). Attenuation of respiratory muscle fatigue is another potential mechanism whereby IMT might improve exertional dyspnea. Respiratory muscle fatigue results in a sympathetically mediated metaboreflex response that reduces limb blood flow and increases perceptions of limb and respiratory discomfort (5, 35). IMT has been shown to improve the fatigue resistance of the inspiratory muscles (33, 44) and can attenuate the respiratory muscle metaboreflex (48), which may explain, at least in part, reduced dyspnea and leg discomfort ratings following IMT in previous studies. However, we do not believe that diaphragm fatigue played a role in our dyspnea results given that we used an incremental cycling protocol, which does not normally cause diaphragm fatigue (34). We intentionally selected incremental rather than constant load cycling to track the sensory and physiological changes across the full range of ventilations and to avoid the potential confounding effects of diaphragm fatigue on dyspnea and NRD. Additional studies are needed to determine whether respiratory muscle EMG can be reduced during other exercise protocols such as time trials and constant-load exercise tests that are more likely to induce respiratory muscle fatigue. This study has some limitations that must be acknowledged. First, limitations of using multipair esophageal electrode catheters for assessing NRD are well established and have been described elsewhere (18, 19, 36). Second, we recognize that there is generally poor between-subject and between-occasion reproducibility of surface EMG measurements. Although reproducible inspiratory muscle EMG during quiet resting breathing and inspiratory threshold loading has been established (7), this has, to our knowledge, not been examined during exercise despite the widespread use of respiratory muscle surface EMG during exercise (6, 31, 37). We attempted to address this problem by carefully standardizing the skin preparation procedures, placing the electrodes in the same position during all visits and normalizing the data to maximal inspiratory contractions. Third, surface EMG can be influenced by underlying levels of subcutaneous fat. To overcome this limitation, we used lean subjects and measured skinfold thickness of the neck and found no changes pre- vs. posttraining in either group (data not shown). An additional critique is our decision to normalize all EMG data as a percentage of maximum. This was done to standardize the procedures between our catheter-derived EMG measurements and our EMG measurements using surface electrodes. Some suggest that it is better to report EMGdi in absolute values when comparing within-subject changes (23, 41). We performed this analysis (data not shown), and our conclusions regarding the lack of change in EMGdi remain the same. Finally, we acknowledge that there may have been “cross-talk” between our sternocleidomastoid and scalene EMG measurements given the close proximity of these muscles. Thus, we are not able to definitively say that we isolated these specific muscles with our surface electrodes. Nevertheless, our surface EMG data provide a good global index of extradiaphragmatic inspiratory muscle activation originating from the neck region. Five weeks of inspiratory muscle strength training resulted in modest reductions in exertional dyspnea intensity but did not change inspiratory muscle EMG, neuromechanical coupling of the respiratory system, or the ventilatory response to exercise. Thus, improvements in dyspnea in healthy individuals following IMT may be driven by nonphysiological factors or by some physiological outcomes that were not measured in the present study. Future work is needed to explore the mechanisms of dyspnea relief following both strength and endurance-based IMT using various exercise protocols across the spectrum of health and disease. GRANTSThis research was supported by a Discovery Grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada and an Infrastructure Grant from the Canada Foundation for Innovation. A. H. Ramsook was supported by the University of British Columbia 4 Year Fellowship (4YF). Y. Molgat-Seon was supported by a 4YF and a Post Graduate Scholarship from the NSERC. M. R. Schaeffer was supported by a 4YF and a fellowship from the British Columbia Lung Association. P. G. Camp was supported by a Scholar Award from the Michael Smith Foundation for Health Research (MSFHR). J. A. Guenette was supported by a Scholar Award from the MSFHR, a Canadian Institutes of Health Research Clinical Rehabilitation New Investigator Award, and a New Investigator Award from the Providence Health Care Research Institute and St. Paul’s Hospital Foundation. DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSA.H.R., P.G.C., W.D.R., L.M.R., and J.A.G. conceived and designed research; A.H.R., Y.M.-S., M.R.S., and S.S.W. performed experiments; A.H.R. and J.A.G. analyzed data; A.H.R., M.R.S., and J.A.G. interpreted results of experiments; A.H.R. prepared figures; A.H.R., L.M.R., and J.A.G. drafted manuscript; A.H.R., Y.M.-S., M.R.S., S.S.W., P.G.C., W.D.R., L.M.R., and J.A.G. edited and revised manuscript; A.H.R., Y.M.-S., M.R.S., S.S.W., P.G.C., W.D.R., L.M.R., and J.A.G. approved final version of manuscript. We thank Dr. Joseph Puyat for assistance with the statistical analysis for this study and to our subjects for their enthusiastic participation. REFERENCES
Page 25skeletal muscle is a highly plastic tissue capable of altering its morphological and metabolic phenotype according to specific physiological demands placed upon it (e.g., exercise, inactivity, and diet). Mitochondrial content, and thus oxidative capacity, can change depending on the energy demand experienced by the muscle cell. A classical feature of prolonged endurance training is an increase in mitochondrial mass and improvement in oxidative energy production (25). Conversely, skeletal muscle mitochondrial content will reduce relatively rapidly with the cessation of regular physical activity (11). In agreement with this, skeletal muscle oxidative capacity is reduced in individuals with obesity and Type 2 diabetes (26, 28), in which a reduction of daily physical activity is believed to be part of the disease pathogenesis (7). Despite their relative abundance within skeletal muscle, a reduction of mitochondrial mass is not a benign occurrence, as it has been implicated in the development of insulin resistance and Type 2 diabetes (41). Thus, maintenance of skeletal muscle oxidative capacity is required for normal metabolic function. Central to the regulation of mitochondrial mass is cellular Ca2+ (18, 38, 39). During muscle activity, release of Ca2+ from the sarcoplasmic reticulum (SR) can activate Ca2+-sensitive signaling proteins, including Ca2+/CaMKs and the Ca2+/calmodulin-dependent phosphatase calcineurin (CaN), resulting in the coordinated regulation of genes involving contractile and metabolic phenotype (13, 14). The importance of cellular Ca2+-signaling and Ca2+-handling proteins toward the oxidative phenotype of skeletal muscle is evident from several transgenic models. Mice expressing a constitutively active form of CaMK IV in skeletal muscle show a transition toward greater mitochondrial content (54), while mice null for CaN isoforms display reduced oxidative staining upon histological examination (40). Mice lacking the Ca2+-buffering protein parvalbumin (PV) within fast-twitch muscle display an increase in mitochondrial content, and consequently, these muscles become fatigue-resistant (10, 42). Conversely, ectopic expression of PV within typical slow-twitch oxidative muscle reduces mitochondrial protein expression and improves contractile kinetics (15). Together, the above studies demonstrate that genetic manipulation of Ca2+-signaling and -handling proteins is enough to alter the metabolic phenotype of skeletal muscle, even in the absence of a physiological stressor. A major regulatory protein of SR, basal cytosolic, and contractile [Ca2+] is the sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA). SERCAs are 110-kDa integral membrane proteins capable of the ATP-dependent transfer of Ca2+ ions from the cytosol to the SR lumen (50) and are responsible for maintaining a high SR [Ca2+] despite a >104 gradient favoring Ca2+ efflux (32, 49). As such, we have found SERCA to account for upward of 50% of resting skeletal muscle energy expenditure in mouse (47). Two well-established regulators of SERCA function are the homologous proteins phospholamban (PLN) and sarcolipin (SLN), which physically interact with the ATPase to slow the rate of cytosolic Ca2+ removal (4). In addition to its role in regulating muscle contractility (52, 53), we have shown that physiological levels of SLN can uncouple SERCA-mediated Ca2+ transport from ATP hydrolysis within oxidative muscle (6). As a result, more ATP is required by SERCA to pump a given amount of Ca2+ across the SR in the presence of SLN. In agreement with this, energy expenditure is reduced during submaximal exercise in SLN knockout mice (Sln−/−) despite a similar maximal aerobic capacity of control mice (5). A unique feature of SLN is its role in adaptive diet-induced thermogenesis (19). High-fat feeding has been shown to increase SLN protein expression three- to four-fold in oxidative muscle (2, 5), increasing the capacity to uncouple SERCA function. Consequently, Sln−/− mice develop an excessively obese phenotype, characterized by exaggerated glucose intolerance when given a “Western” diet (5, 20, 33). Despite a greater positive energy balance being a major causal factor of dysregulated glucose handling of Sln−/− mice, several questions remain. Given the central role of cellular Ca2+ in mitochondrial metabolism and that genetic modification of Ca2+ signaling and handling proteins can unilaterally alter muscle oxidative potential, it is unclear whether SLN ablation per se is associated with alterations in mitochondrial content. This is of particular interest given that SLN has the potential to alter both cellular [Ca2+] and [ATP] through uncoupling and is exemplified by the increased expression of mitochondrial proteins in mice overexpressing SLN (34, 48). It is not clear whether SLN can impact skeletal muscle mitochondrial adaptations to prolonged activity. Furthermore, because SLN expression is responsive to diet, it is not known whether the exaggerated diet-induced glucose intolerance of Sln−/− mice is the result of an absence of SLN-induced changes in mitochondrial signaling with high-fat feeding. Thus, the objectives of our current study are to examine traditional markers of mitochondrial content 1) in Sln−/− mice at rest, 2) in response to prolonged exercise, and 3) in response to excessive diet-induced obesity. MATERIALS AND METHODSGeneration of Sln−/− mice has been previously reported (1). Generation of experimental animals was achieved through heterozygous breeding (i.e., Sln+/− X Sln+/−) to achieve Sln−/− mice and wild-type (WT; i.e., Sln+/+) littermates on a C57BL/6J background. At ~4 wk of age, ear clippings were obtained from animals and genotyped, according to Tupling et al. (53). Prior to experimentation, all mice were group-housed in a temperature-controlled room (~22°C) on a 12:12-h reverse light-dark cycle and were given ad libitum access to rodent chow (22/5 Rodent Diet 8640; Harlan-Teklad, Madison, WI) and water. All experiments were conducted on 3- to 4-mo-old male mice and were approved by the University of Waterloo Institutional Animal Care Committee in accordance with the Canadian Council on Animal Care. Experiments involving voluntary wheel running (VWR) and corresponding sedentary (Sed) control groups are the same as those described previously (20). Briefly, animals (WT and Sln−/−) were individually housed as above, given free access to voluntary running wheels equipped with a magnetic sensor and counter balance, or a locked running wheel (Sed) for a period of 8 wk, and were allowed ad libitum access to water and rodent chow for the examination of SLN’s impact on oxidative capacity in response to voluntary activity (i.e., Chow-Sed vs. Chow-VWR). An additional ad libitum fed high-fat diet (HFD) group was included during this 8-wk period, with a locked running wheel, and run in parallel with the chow-fed Sed and VWR animals. This was done to examine the effects of SLN ablation on dietary responses of mitochondrial enzymes (i.e., Chow-Sed vs. HFD-Sed). The HFD used (42% kcal from fat; TD 88137; Harlan Teklad) has previously been shown by our group (5, 20) to result in an excessively obese phenotype in Sln−/− mice. In addition to VWR, we also examined whether SLN impacts mitochondrial adaptations to forced treadmill training. The training protocol employed was similar to previously published regimens shown to elicit increases in mitochondrial content (29, 35, 37, 44, 45), with minor modifications. Briefly, a separate group of exercise-trained mice ran 5 days/wk for eight consecutive weeks, while untrained control animals were acclimated to the motor-driven treadmill (model: Exer-6M Treadmill; Columbus Instruments, Columbus, OH) by walking at 9 m/min (0° incline) 3 times/wk for 15 min over this same period. The exercise-training program was characterized by progressive increases in running speed and duration. Trained mice began each bout with a warmup starting at 8 m/min, and the running speed was slowly increased by 1 m/min up to the target training speed for the week. The incline of the treadmill remained at 5° throughout the exercise program. A final speed of 17 m/min and a running time of 60 min were achieved at the end of 8 wk (Table 1).
Tissue was collected following 4 h of fasting and inactivity (i.e., wheel lock) for VWR and dietary experiments. Animals were euthanized by an anesthetic overdose (0.65 mg pentobarbital sodium/kg body mass; Bimeda-MTC, Cambridge, ON, Canada), soleus (SOL) muscles were dissected free of connective tissue and weighed, then immediately flash frozen in liquid nitrogen, and stored at −80°C until analyzed. SOL was chosen for our analyses as this muscle expresses SLN protein (53). All chemicals were purchased from Bio-Shop (Burlington, ON, Canada) unless otherwise indicated. Flash-frozen SOL muscles for enzyme activity measurements were homogenized 1:50 (wt/vol) using a glass mortar and pestle in ice-cold phosphate-glycerol buffer (containing in mM: 16 Na2HPO4, 4 KH2PO4, 5 β-mercaptoethanol, 0.5 EDTA (Sigma-Aldrich, St. Louis, MO), 50% glycerol (vol/vol), and 0.02% (wt/vol) BSA, at pH 7.4). Representative enzymes of the major biochemical pathways involved in mitochondrial energy metabolism were chosen for analysis and included citrate synthase (CS) and succinate dehydrogenase (SDH) for the citric acid cycle, cytochrome-c oxidase (COX) for the electron transport chain, and β-hydroxyacyl-CoA dehydrogenase (β-HAD) for β-oxidation. Maximal activities of all enzymes except COX were measured using NAD+/NADH-linked fluorometric end-point assays at room temperature (~22°C), as described previously (12, 23), and modified by Green et al. (21). All samples were measured in triplicate from the same phosphate-glycerol homogenate. SDH was measured on freshly homogenized tissue to avoid loss of enzyme activity with repeated freeze/thawing; all of the other enzymes were examined on thawed homogenate, as they are not affected by repeated thawing and storage at −80°C (12, 23). COX activity was measured using a reaction mixture consisting of 970 μl of 10 mM potassium phosphate buffer (pH 7.0) and 20 μl of reduced cytochrome c (Sigma C-2506; Sigma-Aldrich) at 37°C. The original muscle homogenate was diluted 1:10 using potassium phosphate buffer, creating a total dilution of 1:500 of the original homogenate. The reaction was initiated by adding 10 μl of dilute homogenate to the reaction mixture, and the decrease in absorbance at 550 nm was measured spectrophotometrically for 3 min. COX activity was calculated by using the measured slope and the millimolar extinction coefficient of reduced cytochrome c at 550 nm. For the treadmill training study, animals were euthanized 2 days following the final treadmill session, and the muscles were dissected and cleaned, as described above. Once cleaned, the tissue was homogenized 1:10 (wt/vol) in PMSF buffer [containing in mM: 250 sucrose, 5 HEPES, 10 NaN3 (Fisher Scientific, Fair Lawn, NJ), and 0.2 phenylmethanesulfonyl fluoride (Sigma-Aldrich), pH 7.5], frozen in liquid nitrogen and stored at −80°C until needed. All primary and secondary antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Relative protein expression of adenine nucleotide translocase (ANT; sc-9299), cytochrome-c (cyt-c; sc-13156), and COX subunit IV (COX IV; sc-69360) were determined by standard SDS-PAGE techniques as surrogate measures of mitochondrial content following exercise training. Briefly, samples were resolved on glycine gels (14% for ANT and cyt-c; 12.5% for COX IV) and transferred to PVDF membranes (0.2-μm pore size; Bio-Rad, Hercules, CA), which were then blocked in Tris-buffered saline (pH: 7.5) containing 0.1% Tween-20 (vol/vol) (TBST) and 5% (wt/vol) skim milk powder for 1 h at room temperature. Membranes were then incubated in appropriate primary antibodies diluted in TBST for 1 h at room temperature (ANT: 1:100; cyt-c: 1:2,000; COX IV: 1:5,000). Following this, membranes were washed in TBST (3 × 5 min) and incubated in appropriate horseradish peroxidase-conjugated secondary antibodies. Membranes were washed a final time (3 × 5 min), and signals were detected with an enhanced chemiluminescence kit (Amersham Pharmacia Biotech, Piscataway, NJ) and densitometry analysis using GeneSnap software (Syngene, Frederick, MD). All membranes were stained with Ponceau S (Bio-Shop) to confirm equal protein load and normalization of densitometry values. Protein content of muscle homogenates was determined by the bicinchoninic acid assay (Sigma) using BSA as a standard for both enzymatic analyses and Western blot analysis. All data were analyzed with a two-way ANOVA. When appropriate, post hoc comparisons were made using a Newman-Keuls test to compare specific mean differences. Statistical significance was considered at α ≤ 0.05. RESULTSTransgenic manipulation of Ca2+-handling proteins can alter skeletal muscle mitochondrial content and muscle metabolic phenotype (10, 14, 15, 42, 54). Because of this, we sought to determine what effect SLN ablation per se had on mitochondrial metabolism and whether the mitochondrial adaptive response to increased physical activity is altered within Sln−/− mice. Voluntary running activity for these animals is reported elsewhere (20) and is not affected by loss of SLN. The mean cumulative distance run over 8 wk was 282.7 ± 46.7 km and 281.9 ± 50.6 km for WT and Sln−/− mice, respectively (20). Within sedentary control animals, no effect of SLN ablation was observed on any of the maximal activities of the mitochondrial enzymes examined (Fig. 1). To our surprise, 8 wk of VWR did not impact the mitochondrial enzymatic response in either WT or Sln−/− mice. Because of this, we hypothesized that the metabolic demand of VWR was not sufficient to elicit an increase in mitochondrial content and reveal any potential phenotypic differences. Fig. 1.Soleus mitochondrial enzyme activities (mmol·h−1·g protein−1) of wild-type (WT) and SLN knockout (Sln−/−) mice following 8 wk of voluntary wheel running (VWR) or sedentary (Sed) behavior. A: succinate dehydrogenase (SDH) activity. B: cytochrome-c oxidase (COX) activity. C: citrate synthase (CS) activity. D: β-hydroxyacyl-CoA dehydrogenase (β-HAD) activity. Values displayed are expressed as means ± SE; n = 7–12/group. Thus, we subjected a separate cohort of experimental animals to an 8-wk treadmill-training program (described in Physical activity and dietary treatments). Relative to untrained (UTR) control animals, trained (TR) mice displayed significantly higher maximal activities (main effect of training: P < 0.05) of SDH (~12% greater than UTR animals), COX (17% greater than UTR animals), and CS (~11% greater than UTR animals) (Fig. 2), while β-HAD activity remained unchanged (P = 0.16) by exercise training. No differences in enzymatic activities were found between WT and Sln−/− mice, regardless of training status. Consistent with the training-induced increase in the rate of enzymatic activity, protein expression of cyt-c (~34% greater than UTR animals) and COX IV (~28% greater than UTR animals) were significantly greater (main effect of training: P < 0.05) following the 8-wk exercise program (Fig. 3). Additionally, ANT protein expression tended (P = 0.08) to be greater (~21% greater than UTR animals) with exercise training (Fig. 3). Again, no differences in mitochondrial protein expression were found between WT and Sln−/− mice, regardless of training status. Fig. 2.Soleus mitochondrial enzyme activities (mmol·h−1·g protein−1) of WT and SLN knockout (Sln−/−) mice following 8 wk of endurance treadmill training (TR). A: SDH activity. B: COX activity. C: CS activity. D: β-HAD activity. UTR, untrained animals. Values displayed are expressed as means ± SE; n = 9/group. *Significant main-effect of treadmill training (TR > UTR). Fig. 3.Soleus mitochondrial protein expression of UTR and TR WT and Sln−/− mice. A: representative images of adenine nucleotide translocase (ANT), cytochrome c (cyt-c), and COX subunit IV (COX IV). The Ponceau S staining (42-kDa band) was used to ensure equal protein loading and normalization of densitometry values. B: expression of mitochondrial proteins following 8 wk of treadmill training. All values are expressed relative to WT UTR. Values displayed are means ± SE; n = 9/group. *Significant main effect of treadmill training (TR > UTR). We have previously shown Sln−/− mice to develop excessive diet-induced obesity and glucose intolerance (5, 20). Given that obesity is associated with alterations in mitochondrial metabolism and Ca2+ is a major regulatory signal for mitochondrial biogenesis, it was of interest to determine whether SLN ablation affects the enzymatic response to high-fat feeding. This sedentary cohort of Sln−/− animals developed excessive diet-induced obesity and glucose intolerance relative to WT mice and is reported elsewhere (20). Eight weeks of high-fat feeding had no impact on SOL SDH, COX, CS, or β-HAD activity relative to chow-fed controls, regardless of genotype (Fig. 4). Fig. 4.Soleus mitochondrial enzyme activities (mmol·h−1·g protein−1) of WT and Sln−/− mice following 8 wk of a chow or high-fat diet (HFD). A: SDH activity. B: COX activity. C: CS activity. D: β-HAD activity. Values displayed are expressed as means ± SE; n = 7–12/group. DISCUSSIONCa2+ is an integral second messenger involved in the regulation of mitochondrial metabolism, and genetic modification of Ca2+-handling proteins can alter the metabolic phenotype of rodent skeletal muscle (10, 14, 15, 42, 54). Thus, it was of interest to us to examine the effect of SLN ablation on the activity of mitochondrial marker enzymes and protein expression under a basal state and in response to physiological challenge (i.e., exercise, high-fat feeding). While VWR was found to be insufficient to elicit an increase in mitochondrial content, forced treadmill training resulted in a stereotyped rise in the maximal activities and expression of several mitochondrial proteins, the impact of which was unaffected by the physiological levels of SLN, suggesting that SLN is not required for the regulation of skeletal muscle mitochondrial adaptations to endurance training. Additionally, high-fat feeding, which has previously been shown to elicit an approximately three- to four fold increase in SLN protein expression within oxidative muscle (2, 5), was not associated with any change in mitochondrial marker enzymes relative to chow-fed animals. Thus, our previous observation of excessive diet-induced glucose intolerance of Sln−/− mice (5, 20, 33) is not related to an alteration of mitochondrial content resulting from an inability of these mice to recruit SLN in response to high-fat feeding. Together, our findings indicate that both the basal and adaptive expression of SLN protein are not directly involved in the regulation of mitochondrial metabolism. Muscular adaptations to periods of prolonged and repeated activation are governed, in part, through the release and reuptake of SR Ca2+ stores. The oscillatory patterns of intracellular Ca2+ transients during muscle activity are unique to the excitatory input received by the muscle fiber and can drive the transcription of gene networks, resulting in specific adaptations (i.e., excitation-transcription coupling) (14). SLN has been shown to affect Ca2+ transients, specifically the decay phase, which is primarily determined by SERCA pumping activity (51). Furthermore, SLN-induced uncoupling of SERCA function adds to its potential to mediate metabolic adaptations through alteration of energy demand itself. We chose to examine whether physiological levels of SLN protein expression play any role in metabolic signaling by examining classical end-point measures of mitochondrial content and biogenesis. Surprisingly, VWR was not effective in increasing enzyme activity of mitochondrial proteins within the SOL. The volume of activity engaged in by this group of experimental animals is previously reported not to differ by genotype (20) and is similar to other studies using C57BL/6J mice (17, 30, 36). Although we were unable to examine exercise intensity and bout duration with our wheel-running system, it is possible that the intensity and duration of wheel running were not sufficient enough to necessitate an increase in mitochondrial content above basal (i.e., sedentary) levels. Thus, we employed a forced treadmill-training paradigm to further examine this question. As expected, SOL mitochondrial enzyme activity and content increased with training relative to sedentary animals and was unaffected by SLN. We have previously shown that whole body energy expenditure is reduced at submaximal running speeds in Sln−/− mice (5). However, maximal aerobic capacity is unaltered by SLN ablation (5), which is consistent with our findings herein. While not measured in the current study, it is reasonable to expect exercise performance (e.g., running time to fatigue) to be improved in Sln−/− mice due to the improved Ca2+-pumping efficiency and enhanced pumping rate of SERCA (6, 53). Thus, it appears that SLN is not required for the Ca2+-induced signaling of mitochondrial biogenesis with endurance training. One possible explanation for this may be the involvement of regulatory kinases (e.g., CaMKII, PKC, PKA), which may be activated during muscle activity. Together, these kinases can coordinately improve Ca2+ handling by acting on several SR proteins, including the Ca2+ release channel (31), SERCA (22), PLN (22, 31, 43, 46), and possibly SLN (3). Thus, it is possible that SLN-induced changes in a single Ca2+ transient during a twitch stimulus are attenuated with repeated muscle activation, resulting in similar oscillatory patterns of Ca2+ transients and subsequent metabolic adaptations between WT and Sln−/− mice. Our next objective was to examine the impact of SLN ablation on mitochondrial metabolism in response to diet-induced obesity. In obese and type II diabetic individuals and nondiabetic offspring of type II diabetics, skeletal muscle oxidative gene and protein expression is reduced (8, 26, 28, 41). However, it remains controversial as to whether a reduction in oxidative capacity of muscle leads to dysregulated glucose handling with obesity (24). We have previously shown that exaggerated glucose intolerance is a metabolic feature accompanying the excessive diet-induced obesity of the Sln−/− model (5, 20, 33). Thus, it was of interest to examine whether an association between altered skeletal muscle oxidative capacity and the development of diet-induced glucose intolerance existed with SLN ablation. Relative to chow-fed controls, no effect of caloric overload was seen on the activity of marker enzymes of the Krebs cycle, electron transport chain, or β-oxidation. Furthermore, there was no overt effect of SLN ablation on the mitochondrial enzymes examined, regardless of experimental diet. Therefore, it is unlikely that the previously observed defect in glucose handling (5, 20, 33) is the result of dysregulated mitochondrial metabolism of Sln−/− mice in response to high-fat feeding. The findings from the current study suggest that a greater positive energy balance resulting from an inability to recruit uncoupling for SERCA function is the primary cause of the severe diet-induced glucose intolerance experienced by Sln−/− animals, possibly through accretion of intramuscular lipid metabolites (9, 16, 27) or an increase in proinflammatory cytokines (33), factors known to reduce skeletal muscle glucose uptake. The findings from the current study are in contrast somewhat with several recent reports on SLN-overexpressing mice (34, 48). Constitutive overexpression of SLN driven by the human α-actin promoter has been shown to increase mitochondrial mass within fast-twitch muscle (34). Furthermore, this increase coincides with the increased protein content of several Ca2+-dependent regulators of oxidative metabolism (i.e., CaMKII, CaN), along with the master regulator PGC-1α (34). Maurya et al. (34) have concluded that SLN expression per se can control oxidative capacity through both uncoupling-induced activation of Ca2+-dependent pathways and ATP/ADP signaling pathways. Although the energy demand and simultaneous rise in intracellular Ca2+ may be causal factors driving mitochondrial biogenesis in this particular model, caution should be taken when extending results from overexpression models. The exact degree of SLN overexpression in this model relative to WT animals is unclear from these previous studies (34, 48). Thus, it is possible that a supraphysiological amount of SLN, beyond the three- to four-fold adaptive rise seen with high-fat feeding (2, 5), may necessitate an increase in mitochondrial mass to support basal energy demands. We have previously shown that basal O2 consumption of isolated soleus muscle is not lowered by SLN ablation, despite the relative proportion of ATP consumption attributed to SERCA-mediated Ca2+ pumping being reduced (5, 6), suggesting an undefined source of ATP consumption in Sln−/− muscle. In light of our previous findings (5, 6), it is not surprising that mitochondrial content is unaltered by SLN ablation, given that the absolute energy demand of SOL muscle is similar between WT and Sln−/− mice. Several limitations exist with our current study. One that must be considered when interpreting our findings is the housing temperature of our experimental animals (~22°C), which is below the thermoneutral temperature of mice (~30°C). Given that our housing paradigm represents a thermal stress that must be compensated for by increased energy expenditure, an inadvertent consequence may be a requirement for mitochondrial content to be higher at room temperature to support increased energy demand, possibly masking an overt effect of SLN ablation. However, we do not believe this to be of significant concern given that no change in mitochondrial content was observed following high fat-feeding in WT mice, which has been shown to result in a three- to four-fold adaptive increase in SLN protein expression (2, 5). In light of this, it seems unlikely that the complete absence of SLN per se at room temperature would significantly lower mitochondrial content. Furthermore, given that our mitochondrial measurements were conducted on whole muscle homogenates, we are unable to determine whether specific effects of SLN ablation on mitochondrial subpopulations exist (e.g., subsarcolemmal vs. intramyofibrillar), or by fiber type. Lastly, maximal enzyme activities and expression of individual mitochondrial proteins do not reflect mitochondrial function, which will require future examination using high-resolution respirometry in the Sln−/− model. Our findings herein demonstrate that Sln−/− mice respond similarly to WT animals with the stereotyped increase in mitochondrial content following endurance training. This suggests that Ca2+-induced signals involved in exercise-induced mitochondrial biogenesis are not altered by SLN’s effect on SERCA. Additionally, no impact of SLN ablation was found on markers of oxidative capacity in response to high-fat feeding, despite the development of an excessively obese phenotype in Sln−/− mice. Thus, the previously observed glucose intolerance of these animals is not associated with reduced mitochondrial content and is due to a greater positive energy balance as a result of an inability to recruit SLN-mediated uncoupling. In conclusion, both basal and diet-induced levels of skeletal muscle SLN expression are not required for the regulation of mitochondrial content. GRANTSThis work was supported by research grants from the Canadian Institutes of Health Research (MOP 86618 and MOP 47296 to A. R. Tupling). DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONSD.G., A.T., V.A.F., and E.B. performed experiments; D.G. and A.T. analyzed data; D.G. and A.T. interpreted results of experiments; D.G. prepared figures; D.G. drafted manuscript; D.G., A.T., V.A.F., E.B., and A.R.T. edited and revised manuscript; D.G., A.T., V.A.F., E.B., and A.R.T. approved final version of manuscript. REFERENCES
Page 26exercise-induced hypoalgesia (EIH) is a reduction in pain that occurs during or following exercise and is well demonstrated for exercise of various modalities, durations, and intensities (49). Identifying the mechanisms of pain relief from exercise involves investigations of acute (16, 42, 62) and chronic adaptations (35, 37, 59) that may be similar or distinct. For example, acute exercise might transiently reduce pain through changes in the release of analgesic substances (33, 42), whereas chronic exercise might reduce pain more through changes in the cognitive appraisal of a noxious stimulus (35). The fact that exercise acutely reduces pain suggests that changes must occur somewhere in the nociceptive pathways, but whether these changes are predominantly peripheral, central, or a combination of both, is not known. Studies in animals have identified numerous substances that contribute to EIH (e.g., opioids, cannabinoids, noradrenaline, and nitrite). These substances are synthesized and released from multiple sites within the peripheral and central nervous systems, the endocrine system, the immune system, and blood vessels (e.g., endothelium and skeletal muscle for nitric oxide, adrenal medulla and peripheral sympathetic nerves for catecholamines, the pituitary gland and immune cells for opioids, and central and peripheral neurons for endocannabinoids) (1, 4, 15, 47, 70). For exercise, whether these substances reduce pain predominantly through their actions at peripheral or central sites is not clear (12, 18, 20, 21, 23). In rats, the systemic but not central administration of ɑ2-adrenergic receptor antagonists reverses EIH (12), implying that catecholamines released during exercise reduce pain through peripheral actions. Nociceptors, through their rich expression of ligands, receptors, and neurotransmitters, have autocrine and paracrine actions and can modify input before it reaches the central nervous system (7). Many of the substances that influence nociceptor sensitivity (e.g., opioids, cannabinoids, and noradrenaline) increase in the blood during exercise (13, 19, 61, 70) and remain elevated in a manner consistent with the persistence of EIH following exercise (14), so it is reasonable that these exercise-induced changes could reduce nociceptor sensitivity. The distribution of blood during exercise (6) also corresponds, albeit not exactly, with the gradient in the magnitude of EIH across different body regions. That is, like blood flow, EIH is typically greater in exercised than unexercised limbs (50, 65). This is revealed by the relatively smaller increase in pain thresholds for the arms vs. the legs following cycling or running exercise (46). However, EIH does occur in unexercised limbs. Therefore, blocking blood flow to an unexercised limb during exercise provides a means to determine whether a blood-borne factor contributes to EIH via actions in the periphery. The important role of peripherally acting catecholamines as mediators of EIH shown in rats (12) has yet to be directly explored in humans. A recent study provided indirect evidence of a possible role of the peripheral nociceptor in EIH (36). EIH as determined by the increase in pressure pain threshold (PPT) over the biceps brachii and first dorsal interosseous muscles after exercise was induced by isometric exercise of the elbow flexors for 3 min at 40% of the participant’s maximal voluntary force. Somatosensory-evoked potentials to noxious electrical stimuli delivered to the index finger were unchanged after exercise, whereas laser-evoked potentials to noxious heat delivered to hand dorsum decreased after exercise. This was despite clear EIH in both instances. This result suggests that the peripheral nociceptor might be involved in EIH in humans (36) because laser heat stimuli activate peripheral nociceptors, whereas electrical stimuli bypass the receptors by activating the axons of the nociceptive afferents (3, 52). However, the role of central factors cannot be discounted. Animal studies have shown that centrally acting drugs can influence EIH (18, 20, 21, 23), and the occurrence of EIH in humans could also be explained by the actions of the central nervous system (26, 43, 50, 64, 65). It appears likely that both peripheral and central changes contribute to EIH, but the relative contribution of each is not known. To our knowledge, no studies have directly explored the contribution of peripheral factors to EIH in healthy adults. Therefore, the purpose of this study was to investigate whether during high-intensity lower limb cycling, EIH, as measured by an increase in PPT in a nonexercising arm, would be reduced in the other arm in which blood flow was occluded during exercise. It was hypothesized that a circulating factor with a peripheral action contributes to EIH, and therefore, the increase in PPT after exercise would be less in the occluded arm vs. the nonoccluded arm. METHODSWritten informed consent was obtained from each participant before testing. All procedures were approved by the University of New South Wales Human Research Ethics Committee (HC 14065) and conformed to the requirements of the Declaration of Helsinki (2008). Sample size calculations were performed using G*Power (version 3.1.9.2; Dusseldorf, Germany) (17). The primary outcome in this study was the effect of exercise on PPT at the occluded vs. the nonoccluded limb. Exercise has a moderate effect on increasing pain thresholds at nonexercised limbs (5, 49); however, a difference in PPT between limbs (i.e., occluded vs. nonoccluded) would be an inherently smaller effect. We estimated a physiologically meaningful difference to be in the order of 0.4 ± 0.7 kg/cm2 (means ± SD), which corresponded to approximately one-third of the typical elevation of PPT in an unexercised limb following high-intensity exercise of a large muscle group (66). Assuming an effect of this size, and using a paired sample t-test with α = 0.05 and 90% power, it was calculated that 35 participants were needed for this study. Therefore, we aimed to recruit at least 35 participants. Participants were recruited through advertisements placed on billboards around the University of New South Wales campus. Eligibility criteria included 1) apparently healthy with no history of chronic pain or chronic disease, 2) between the ages of 18 and 60 yr, and 3) absence of a current diagnosis of depression or any other major mood disorder because this can influence pain thresholds (60). Before the experiment, participants were asked to abstain from vigorous exercise for 24 h and caffeine for 4 h. Compliance to these requests was confirmed verbally at the start of the session. The experiment consisted of a single session that lasted ~1 h. The experimental procedures are outlined in Fig. 1. With the participant seated on the recumbent cycle ergometer, PPTs were determined at three sites [rectus femoris and first dorsal interosseous (FDI) muscle of both arms]. Blood flow to one arm was occluded by inflating a blood pressure cuff on the upper arm to 240 mmHg. The occluded arm was counterbalanced across participants. Once the target pressure was achieved, PPTs were again determined. The participant then either cycled at a high intensity for 5 min or remained at rest for 5 min. Immediately upon cessation of exercise, or at the end of a 5-min rest, PPTs were again determined. Occlusion to the arm was then removed and a 30-min wash-out period ensued. After this, PPTs were obtained again at rest, after which blood flow to the other arm was occluded. Once the target pressure was achieved, PPTs were reassessed. Participants then either remained at rest for 5 min, or cycled at a high intensity for 5 min, after which PPTs were again determined. The occlusion to the arm was then removed. The order of exercise and rest were randomized between participants. PPTs before occlusion were assessed to account for any influence of occlusion on PPTs and to provide a baseline to ensure that in participants who exercised first, PPTs had returned to baseline after the postexercise wash-out period. Approximately 1.5 to 2.5 min were required to complete the PPT measures. Therefore, the time of occlusion ranged from 8 to 10 min [i.e., 1.5–2.5 min for PPT assessment, 5-min intervention (rest or exercise), and 1.5–2.5 min for PPT reassessment]. Fig. 1.Experimental procedures. Pressure pain thresholds (black circles) were assessed over the rectus femoris muscle of the right leg and over the first dorsal interosseous muscles of both arms before and during upper limb occlusion using a tourniquet that was placed around the upper arm and inflated to 240 mmHg. Both arms were occluded during the experiment, but only one arm was occluded at a time and the order of the arm that was occluded first was counterbalanced across participants. During occlusion (dashed rectangles), pressure pain thresholds (PPTs) were assessed before and after 5 min of high-intensity cycle exercise, and before and after an equivalent period of quiet rest. The order of exercise or quiet rest was randomized and counterbalanced across participants. A 30-min wash-out period was included to ensure any exercise-induced alterations in pain were gone before commencing the next round of PPT measures. PPT was assessed over the right rectus femoris muscle and the FDI muscles of both arms in a random order. These measurements were interspersed so that PPT was assessed once over each site, and this was then repeated in the same order until three measurements had been obtained for each site. PPT was recorded as the average of these three measurements. Two practice trials were performed on the left rectus femoris muscle before testing to familiarize the participant with the procedure. The rubber-tipped probe of a handheld algometer (Wagner Force 10 FDX-25; Wagner Instruments, Greenwich, CT) was applied perpendicularly to the participant’s skin, and the force was increased gradually at a rate of ~1 kg/s. Participants were instructed to give a verbal command of “stop” when the sensation of pressure turned to pain. The duration of pressure application for each PPT measure was similar for each site (~5 s over the hand and ~6 s over the rectus femoris). The sleeve of a standard sphygmomanometer was placed around the participant’s upper arm and the cuff was inflated to 240 mmHg. This high pressure was chosen so that occlusion of the limb would be maintained despite elevations in systolic blood pressure during dynamic exercise (2). Each period of occlusion lasted approximately 8–10 min, which corresponded to the time it took to assess PPTs before and after exercise or rest and the 5-min intervention itself. Immediately after cuff inflation and before the intervention (pre), participants were asked to rate the intensity and unpleasantness of any pain from the pressure of the cuff on their arm using a 0–10 numerical and categorical scale (0 = no pain, 10 = worst possible pain). Ratings were also made of the intensity and unpleasantness of any painful ischemic sensations in the forearm and hand. These ratings were obtained again immediately before cuff deflation following the intervention (post). To account for any effect of altered sensation by occlusion on PPTs, tactile sensation was assessed at the same time points as the pain ratings by lightly brushing the fingertips of each participant’s occluded hand with cotton wool and asking them to describe the sensation. A familiarization trial before occlusion was provided so that participants were aware of what their “normal” sensation felt like. To reduce the influence of any residual effects from the first period of occlusion, both arms were occluded during the experiment—one for rest and the other for exercise—and the order of the arm that was occluded first was randomized. We used short-duration exercise to minimize the amount of time participants spent with their arm occluded. Short-duration, high-intensity aerobic exercise has not previously been used in studies of EIH, so we chose to use high-intensity exercise to increase the likelihood that exercise would evoke EIH (30, 39). Furthermore, recumbent cycling was used to minimize/negate any influence that forcefully gripping the handlebars during exercise might have had on PPTs at the hand. Participants were seated behind a stationary cycle ergometer (Monark 828e; Vansbro, Sweden) with their arms relaxed by their sides. Because it takes several minutes for heart rate to increase and then stabilize after the commencement of exercise, exercise intensity was based on ratings of perceived exertion (RPE; Borg’s 6–20 scale). The aerobic exercise bout therefore consisted of 5 min of high-intensity cycling in which participants were instructed to pedal at a workload corresponding to an RPE of 17 or greater (i.e., very hard). Measurements of workload and RPE were recorded every 30 s during the exercise bout and, when necessary, the workload was adjusted so that participants maintained their target intensity. Descriptive statistics were calculated using version 22 of the Statistical Package for Social Sciences (IBM Chicago, IL). Differences in pressure pain threshold at each muscle were examined with a three (time: baseline, pre, post) × two (condition: rest, exercise) repeated-measures ANOVA. Pain ratings during limb occlusion by the cuff were tested with a two (time: pre, post) × two (condition: rest, exercise) repeated-measures ANOVA. Normality of the data was assessed using the Kolmogorov-Smirnov statistic. Paired sample post hoc tests were conducted to identify the source differences in pain thresholds and pain ratings that were detected by ANOVA and to investigate differences in EIH between men and women. There have been several investigations of sex differences in EIH (44, 45, 66), but the comparison of EIH between men and women in the present study was purely a secondary analysis. The study was not designed, nor powered, to test for a sex difference in EIH. If sphericity was violated, Huynh-Feldt and Greenhouse-Geisser corrections were used when epsilon was >0.75 and <0.75, respectively. Alpha was set at 0.05 and the P value for the t-tests was multiplied by the number of comparisons for each ANOVA model. Bonferroni corrected Student’s paired sample t-tests were also used to compare change scores in PPTs to contrast the effects of rest and exercise. Cohen’s d effect sizes (ES) and 95% confidence intervals (CIs) were also calculated to aid these comparisons and were interpreted as small (0.2), medium (0.5), or large (0.8) (8). The 95% CIs of the ES were calculated using a noncentral t distribution (9). Except where stated, values are reported as the mean and 95% CI. Absolute units are used for presentation of PPT data in figures, whereas percent changes are used in text to enable easier interpretation and comparison of the results. RESULTSThirty-six volunteers participated in this study. Participant characteristics and exercise intensity data are outlined in Table 1. Most participants were undergraduate students who did not regularly participate in moderate- or high-intensity exercise and had minimal experience with cycle ergometer exercise. This was evident from the relatively high RPE during exercise despite modest work rates (Table 1). The participants’ level of perceived exertion during exercise was slightly below our target RPE of 17, but the values indicate that participants still perceived the exercise to be between hard (heavy) and very hard.
Data for PPTs are shown in Fig. 2. There was a significant effect of time [F(1.13, 39.50) = 37.75, P < 0.001] and condition [F(1, 35) = 10.37, P = 0.003], and a significant time × condition interaction [F(2, 70) = 64.32, P < 0.001] for PPT over the rectus femoris muscle. A significant effect of time [F(1.61, 56.25) = 24.78, P < 0.001], condition [F(1, 35) = 12.08, P = 0.001], and a time × condition interaction [F(1.71, 59.71) = 52.37, P < 0.001] was also observed for PPT over the nonoccluded FDI muscle. There was no effect of time [F(1.31, 46.02) = 1.86, P = 0.16] or condition [F(1, 35) = 2.42, P = 0.13] for PPT over the occluded FDI muscle; however, a significant time × condition interaction was observed [F(1.58, 55.32) = 10.37, P = 0.017]. Post hoc comparisons of pain thresholds before cuff inflation (i.e., baseline) with those measured before rest or exercise (but still during cuff inflation) showed no significant effect of occlusion on PPTs over any muscle (range of mean change, −1.4% to +3.9%, all d < 0.09 and P > 0.54). There was also no significant effect of quiet rest on PPT over any muscle (range of mean change, −1.3% to +0.9%, all d < 0.05 and P > 0.51). In contrast, exercise significantly increased PPT over the rectus femoris [+29.3 ± 13.9% (mean ± SD), d = 0.69 (0.47 to 0.90), P < 0.001], nonoccluded FDI [+22.7 ± 13.9% (mean ± SD), d = 0.56 (0.38 to 0.74), P < 0.001), and occluded FDI muscles [+8.3 ± 13.6% (mean ± SD), d = 0.19 (0.07 to 0.32), P = 0.018]. Data for the change scores of the PPTs (in kg/cm2) are shown in Fig. 3. The increase in PPT after exercise was significantly smaller in the occluded limb vs. the nonoccluded limb [−14.4 ± 13.0% (mean ± SD of the difference), d = −1.03 (−1.44 to −0.65), P < 0.001], as shown in the change scores plotted in Fig. 3. The smaller exercise-induced change in PPT for the occluded vs. nonoccluded limb was also apparent when these changes were expressed relative to the change with quiet rest [−13.2 ± 15.2% (mean ± SD of the difference)], d = −0.89 (−1.27 to −0.49, P < 0.001). That is, EIH was diminished when blood flow to that limb was occluded. Fig. 2.Pressure pain thresholds (PPTs). Individual data (gray dots) and group data (mean and 95% confidence interval, black lines) before cuff inflation (baseline) and during cuff inflation before and after rest and exercise (Ex) for PPTs at the rectus femoris muscle (A) and the nonoccluded and occluded first dorsal interosseous (FDI) muscles (B and C, respectively). *Significant increase in PPT after exercise (P = 0.018). **Significant increase in PPT after exercise (P < 0.001). Fig. 3.Change in pressure pain thresholds (PPTs). Individual data (gray dots) and group data (mean and 95% confidence interval, black lines) for the change (Δ) in PPTs with rest and exercise (Ex) at the rectus femoris muscle (left) and the nonoccluded and occluded first dorsal interosseous muscles (FDI, middle and right, respectively). Comparisons for these data were made only on the differences between the FDI sites. *Significant difference between the nonoccluded and occluded arm for the change in PPT after exercise (P < 0.001). #Significant difference between the nonoccluded and occluded arm for the change in PPT after quiet rest compared with exercise (P < 0.001). Data for ratings of pain intensity and pain unpleasantness during occlusion are shown in Fig. 4. Pain from cuff pressure and ischemic pain were rated separately. There was a significant effect of time on ratings of cuff pressure pain intensity [F(1, 35) = 60.86, P < 0.001], cuff pressure pain unpleasantness [F(1, 35) = 68.9, P < 0.001], ischemic pain intensity [F(1, 35) = 59.7, P < 0.001], and ischemic pain unpleasantness [F(1, 35) = 89.78, P < 0.001]. All pain ratings significantly increased over time during rest (range of mean increase, 2.2 to 3.1, all d > 0.91, and P < 0.001) and exercise (range of mean increase, 1.5 to 3.5, all d > 0.63. and P < 0.001), but there were no significant time × condition interactions and no differences in the increase between the rest and exercise conditions (range of mean difference, 0.07 to 0.71, all d ≥ −0.27 and ≤0.35, all P > 0.40). Fig. 4.Pain ratings during occlusion. Data to the left of the vertical dotted line in each graph show the individual data (gray dots) and group data (mean and 95% confidence interval, black lines) before and after rest and exercise (Ex) for ratings of cuff pressure pain intensity (A) and unpleasantness (B), and ratings of ischemic pain intensity (C) and unpleasantness (D) during occlusion. Data to the right of the vertical dotted lines show individual data (gray dots) and group data (mean and 95% confidence interval, black lines) for the differences in pain ratings between the prerest and postrest measures (Δ rest) and between the preexercise and postexercise measures (Δ Ex). Data to the left of the vertical dotted lines are plotted against the left-hand y-axis; data to the right of the vertical dotted line are plotted against the right-hand y-axis. *Significant increase in pain ratings from before to after (P < 0.001). Paresthesia of varying degrees was reported by all participants by the end of occlusion, but all participants were still able to feel the cotton wool on their fingertips by the end of the occlusion period. That is, light touch was diminished but still present in all participants irrespective of the rest or exercise condition. There were no significant sex differences in baseline PPTs (all P > 0.16) or PPT change scores after rest and exercise at any site (all P > 0.07). Similarly, no significant differences between men and women were observed for ratings of pain intensity or pain unpleasantness immediately after cuff inflation (all P > 0.13) or for the change in these pain ratings after rest and exercise (all P > 0.14). DISCUSSIONThe current study shows that short-duration, high-intensity aerobic exercise increases PPTs both locally and in unexercised limbs in healthy adults. This increase was significantly diminished in a limb to which blood flow was occluded. These results suggest that substances released into the blood during high-intensity dynamic exercise contribute to EIH through peripheral analgesic actions. However, other factors unrelated to the occlusion of blood flow to the limb must also be involved in EIH because PPTs were still significantly increased in the occluded limb after exercise. In the current study, short-duration exercise was used to minimize the time that participants spent with their arm occluded, but it was unclear whether this duration of exercise would be sufficient to elicit EIH. The effect of moderate-duration (more than 15–20 min) aerobic exercise on reducing pain is well demonstrated (49), particularly for higher intensities of exercise (i.e., 60–75% of maximal aerobic capacity) (41). Typically, these studies have used similar modalities of exercise (cycle ergometer) and methods of pain assessment (PPT over exercised and nonexercised sites) to those used in the current study, and have produced similar findings. That is, an increase in PPT at exercised and remote sites. For shorter-duration aerobic exercise, the effects on pain are less clear. In healthy young men, cycle ergometer interval exercise (4 × 4 min at 85% of heart rate reserve separated by a 2-min recovery at 60% of heart rate reserve) was found to reduce sensitivity to noxious thermal stimuli but have no effect on PPT, which was assessed over the muscle belly of a nonexercised muscle (extensor carpi radialis) using a handheld algometer (39). This method of PPT assessment was comparable to that used in the current study. Shorter-duration (10 min) continuous aerobic exercise has also been shown to reduce thermal (16) but not pressure pain sensitivity (30), which in this case was assessed by pain ratings every 10 s during the 2-min application of a 9.8-N stimulus to the nondominant index finger. The different measures of pressure pain may contribute to the different results. Nonetheless, our results contrast with those studies as moderate increases in PPT at both the exercised and nonexercised, nonoccluded muscles demonstrated the presence of EIH. Following exercise, we observed an increase of ~23% and 8%, respectively, in PPT at a nonoccluded and occluded limb, both of which were unexercised. However, the difference in the increase in PPT between these two sites after exercise was large and significant. That is, blocking blood flow to a limb during exercise diminished EIH in that limb by >60%. Importantly, this same effect was not observed during an equivalent period of quiet rest. These results show that blocking the peripheral delivery of agents released during exercise and carried by the blood markedly diminishes EIH. Although it was not within the scope of this study to investigate which agents contribute to this effect, data from animals and humans suggest a likely role of cannabinoids, opioids, catecholamines, and nitrite (12, 20, 22, 24, 25, 33, 42). Evidence for a significant effect of catecholamines is particularly strong given that their plasma concentrations are greatly increased during high-intensity exercise (19), and that blocking their peripheral actions reverses EIH in rats (12). Furthermore, the time course of elevated catecholamines in the blood following exercise (19) corresponds with the typical duration and decrement of EIH following the cessation of exercise (40). The analgesic action of agents released into the blood during exercise represents an interesting mechanism for EIH, which may also account for the consistent observation that EIH is greatest at body sites nearest to exercised muscle (36, 65). The gradient of blood flow to exercised vs. unexercised limbs (6, 11, 57) could provide a reasonable account for these variations in EIH if there are effects at the peripheral nociceptors. That is, the greater delivery of analgesic factors released into the blood during exercise would have greater effects on reducing pain in the exercised limb(s) and muscles that receive the majority of the blood flow, possibly through actions at the peripheral nociceptors. During moderate- to high-intensity cycling exercise, blood flow to the arms increases only slightly compared with the changes observed for the legs (6, 11, 57). Although this relative difference in blood flow to the arms and legs during cycling does not exactly match the relative increase in PPT at the arms and legs following a comparable dose of cycling (46, 65, 66), some similarities are apparent. In line with our hypothesis, a significant elevation of PPT in the occluded limb was still observed following exercise, and there are a few possible reasons for this. First, occlusion of blood flow would not be expected to entirely eliminate EIH because other peripheral changes would still occur. For example, exercise-induced increases in muscle sympathetic nerve activity would have been ongoing in the occluded arm (55, 56, 68) and could have influenced pain via the release of agents acting locally on peripheral nociceptors. Second, centrally mediated changes at spinal and/or supraspinal levels (26, 43), as well as cognitive factors, might have also contributed (34, 50, 51); however, these were not measured in the present study. It is well demonstrated that the same intensity of noxious mechanical stimulation is often perceived as less intense or unpleasant after exercise (49). Our current data do not support this, however, because ratings of cuff pressure pain and ischemic pain did not differ between the rest and exercise conditions. Although spinal, supraspinal, and/or cognitive contributors to EIH could influence ratings of cuff pressure pain and ischemic pain, our result showing no effect of exercise on these ratings is consistent with our hypothesis that a circulating factor contributes to EIH. That is, because the cuff was inflated before exercise specifically to occlude circulation, a circulating factor would not be able to influence pain ratings. Had the cuff been inflated after exercise to cause pain instead, then we would have expected an effect, although EIH is not well demonstrated for ischemic stimuli (49) and is less consistent following aerobic exercise when cuff algometry is used to assess pain (63, 64, 67). Differences in the temporal and spatial properties of cuff vs. manual algometry are likely reasons for this. For example, during cuff algometry, the widespread application of pressure over muscle, bone, and nerves is likely to activate different tissues and do so less vigorously than when a small probe is used, as evidenced by the greater time it takes to reach pain threshold during cuff algometry (~25 s for the leg and ~30 s for the arm) (28) vs. handheld algometry (approximately 6–7 s for the leg and approximately 4–5 s for the arm, present data). Handheld algometry and cuff pressure were both used in the current study, but only thresholds of pain were solicited for algometry, whereas only ratings of pain intensity were solicited for cuff pressure, and the cuff pressure stimulus was present throughout the entire exercise and assessment period compared with only transient measurements with handheld algometry. Several elements of the experimental design warrant further discussion. First, a very high cuff pressure was used around the arm. We did this to ensure that occlusion of blood flow to the limb was maintained despite large elevations in systolic blood pressure during exercise, often as high as 200 mmHg or more (2, 54). Accordingly, we chose to use a cuff pressure of 240 mmHg for all participants. This cuff pressure was immediately rated by participants as mild to moderately painful (i.e., 4/10) and was therefore of sufficient intensity to potentially induce a conditioned pain modulation or “pain inhibits pain” effect (53). Indeed, even nonnoxious pressure applied to one body site can reduce pain sensitivity at another site (31). The magnitude and reliability of conditioned pain modulation varies based on interpersonal (e.g., age, gender, attention to the conditioning stimulus and expectations about its subsequent effect on pain) (29) and experimental (e.g., the type of conditioning and test stimuli used, whether pain is assessed during or following the conditioning stimulus) factors (38), whereas psychological influences are less influential (48). However, the considerable heterogeneity and risk of bias in the existing conditioned pain modulation literature means that normative values for healthy adults have not been established. Therefore, it is unclear to what extent, if any, a “pain inhibits pain” effect should have been expected in this study. Inspection of the individual data in Fig. 2 shows that whereas occlusion did increase PPTs in some participants, many participants also experienced a reduction in their pain thresholds, and so the overall group effect of occlusion on PPT was negligible. Another consideration was whether occlusion of blood flow for 8–10 min may have directly influenced the nociceptive afferent axons (i.e., group III/IV afferents; a.k.a. A-delta and C fibers). The sensitivity of both fast- and slow-conducting afferents are influenced by ischemic pressure blocks, although it is generally accepted that the faster conducting afferents are affected first (10, 27, 58). Although there is some evidence contrary to this (69), it is likely that tactile sensation would have to disappear during occlusion before any effect of occlusion on the nociceptive afferents would be observed in the hand (32). Paresthesia of varying degrees was reported by all participants during occlusion; however, tactile sensation, albeit reduced, was still present in all participants at the end of occlusion. This shows that the approximately 10-min period of occlusion in the present study was too short to influence the axons of nociceptive afferents, which is in line with previous studies (10, 27, 32, 69) and is supported by the stability of PPTs observed during occlusion and quiet rest in the current study. The primary limitation of this study was that the experimenter who assessed PPTs was not blinded to the limb that was occluded or whether the participant had just rested or exercised. However, blinding would have been difficult for two reasons. First, obvious visible changes occur in the occluded limb during occlusion (i.e., discoloration), so the experimenter would have known which arm was occluded even if it was concealed. Second, the high-intensity nature of the exercise meant that participants were often still short of breath when PPTs were being assessed immediately after exercise, and immediate assessment was required to limit the period of occlusion. In conclusion, this study provides evidence that the reduction in pain sensation after exercise is mediated, at least partially, by factors that are released into the blood during exercise and act at the periphery. Future studies should aim to determine the agents involved in this effect (e.g., opioids, cannabinoids, catecholamines, and nitrite) and whether combining exercise with drugs that increase these substances and/or upregulate their receptors has additive effects on relieving pain. GRANTSSupport for this work was provided by Program Grant 1055084 from the National Health and Medical Research Council of Australia (NHMRC). M. D. Jones holds an Australian Postgraduate Award and J. L. Taylor holds an NHMRC Research Fellowship. DISCLOSURESNone of the authors have conflicts of interest, financial or otherwise, to be declared. AUTHOR CONTRIBUTIONSM.D.J., J.L.T. and B.K.B. conceived and designed research; M.D.J. performed experiments; M.D.J. and B.K.B. analyzed data; M.D.J., J.L.T. and B.K.B. interpreted results of experiments; M.D.J. prepared figures; M.D.J. drafted manuscript; M.D.J., J.L.T. and B.K.B. edited and revised manuscript; M.D.J., J.L.T. and B.K.B. approved final version of manuscript. We thank all participants who volunteered for this study. REFERENCES
|
When does the body experience the highest rates of glycogen storage?
Pos Terkait
Copyright © 2024 ketajaman Inc.
