This report is not a review of ammonia and amino acid metabolism in skeletal muscle. Numerous papers have addressed these issues(16,19,20,24,32,36,38). The original goal was to compare four common experimental paradigms for skeletal muscle metabolic physiology (the canine and rat hindlimb muscle models and leg metabolism of humans during either “whole body” exercise or knee extensor exercise). We planned to compare the dynamics of ammonia and amino acid metabolism in these models in response to both increasing exercise intensity and duration. We quickly realized that there were little or no data for most of these models for many of the amino acids. Even when we narrowed the focus to a few critical amino acids, large gaps in the data sets were apparent. Thus we narrowed our focus to examine ammonia and the metabolism of alanine, glutamine, and glutamate. Even with these restrictions complete comparisons are seldom possible.
Ammonia, alanine, glutamine, and glutamate are compounds found in both the circulation and in free metabolic pools in the muscle. They can be metabolized rapidly in the active muscle and can be exchanged not only between the muscle and plasma but also between the plasma and other tissues such as the liver. Thus to quantify and understand the metabolism, one has to measure and/or control all of these processes. Rarely has this been done. While this metabolic situation could be evaluated by various labeling techniques, these methods have been predominantly applied to muscle to study protein turnover and leucine oxidation. Thus, these studies will not be included in this review.
The four compounds are involved in many metabolic pathways; a major purpose of these pathways appears to be the “processing” of alpha amino nitrogen. Ammonia can be formed from AMP deamination to IMP via the enzyme AMP deaminase. Free ammonia can also be formed from glutamate through the action of glutamate dehydrogenase. In addition, some of the free ammonia can combine with glutamate to form glutamine (note that this requires energy (ATP) and a source of glutamate). Both glutamine and ammonia can be released from the muscle during exercise. In addition, alanine can be formed from pyruvate and glutamate and can be released. Once again glutamate is required and whether this represents an “alpha amino group” leaving the muscle depends on whether the glutamate was synthesized from free ammonia and alpha ketoglutarate in the muscle at that time. Similarly, glutamine will represent either one or two “alpha amino groups” depending on the glutamate source. Finally, aspartate and IMP can be metabolized to fumarate and in doing so generates an AMP which can be deaminated to ammonia and IMP. This process is the purine nucleotide cycle (PNC) and it requires energy (GTP). It also requires aspartate, and it is likely that during exercise this amino acid is predominantly formed from glutamate, placing another demand on the intramuscular glutamate pool. These demands are met at least in part by both branched-chain amino acid metabolism and the uptake of glutamate from the plasma. Studies that explore these processes in detail must determine the exchange of these compounds with the plasma and also measure the changes in the intramuscular concentrations of all of these metabolites.
This review will evaluate the information available from investigations of the four experimental models that are commonly employed in this area of research. All of these paradigms are valid and offer a number of advantages. Nevertheless, problems have developed in the literature, not from these studies, but from other scientists applying the conclusions of these studies to other experimental circumstances with little or no recognition of the limitations of the original work. We will begin with a brief consideration of the strengths and limitations of each model as it applies to the focus of our report. Then we will review and compare what has been established for each of the processes with each model. In preparing these comparisons wherever possible we have put the data for all models into the same unit. This was problematic because in some cases as we could not establish the mass of active tissue. In these circumstances we did not index the data for mass. This limitation restricts our ability to quantify exchange of any one of the metabolites and to compare it with the changes in its intramuscular concentration.
Regardless of the species or model, all of the metabolites being considered are exchanged between the muscle and perfusate, blood, or plasma. Any study that does not measure this exchange is potentially missing a critical portion of the net metabolism. Similarly, any study in which intramuscular concentrations of the amino acids are not determined is risking a large error because of a factor that we term “washout.” The larger the intramuscular concentration, the greater the potential washout error; thus it would be very small for ammonia but large for glutamine. For example, assume that the muscle concentration of glutamine is 15 mmol·kg-1 wet wt and the subject is exercising for 60 min. If the active muscle mass is 8 kg(a value often assumed for a single leg), the intramuscular free pool of glutamine is 120,000 μmol. If during the hour of exercise, 5% of the pool was lost to the circulation (washed out), i.e., the intramuscular concentration decreased to 14.25 mmol·kg-1, this would be seen as a release (and apparent production) of 100 μmol·min-1 from the leg. Such a decline in intramuscular concentration would generally be considered very modest and such a flux large. Thus if both flux and changes in intramuscular concentration are not measured, erroneous conclusions can be reached. In a recent study (18) we made both measures. In a group of untrained subjects the net glutamine release over 3 h of exercise was 7,078 μmol (a mean flux of 39 μmol·min-1), but when the change in intramuscular concentration was allowed for, the net production was 4,352 μmol; a difference and potential error of 64%.
THE MODELS: STRENGTHS AND LIMITATIONS
The “Whole Body” Human
This is the “model” that most scientists who study muscle metabolism want to understand. However, its complexity often forces us to turn to other experimental approaches. When we refer to the “whole body” human, we are referring to the protocol in which the subject performs two-legged exercise and has arterial and venous femoral catheters for the determination of leg blood flow and for direct Fick measures of metabolites. The major “strength” of this model is that it is the actual situation that we are usually striving to understand. The blood flow, the nutrient supply, the hormonal environment, the neural control, etc. are“natural.”
There are several methodological concerns. In 1971 Jorfeldt and Wahren(22) demonstrated that the leg blood flow and(a-v)O2 difference accounted for 24 ± 6% of the pulmonary˙VO2 at rest and 72 ± 8% of the increase in pulmonary˙VO2 during steady-state exercise. In our experience this relationship is quite reproducible within a subject for a given catheter placement; however, it is likely that the placement is such that the venous drainage of the more proximal muscles is missed. Hence only 72% of the increase in ˙VO2 is accounted for in this method. In addition, differences in placement and anatomy probably account for the ± 8% variation among subjects. Despite these factors investigators have often measured the (a-v) O2 difference for the active leg and pulmonary˙VO2, and using the mean values of 24 and 72% of pulmonary˙VO2 they have estimated the leg blood flow. Presumably this must introduce an error of at least ± 8% which has never been acknowledged in these studies.
To relate the flux of metabolites measured with this technique to the active muscle, one must know the mass of this tissue. Several studies(4,23) have assumed a mass of 8 kg, and we know of no study within the present research focus that has actually estimated the mass for the specific subjects; the 8-kg value comes from formulae for the“standard 70-kg man” and these investigators apply the 8-kg assumption to all subjects regardless of size or body build. In our studies(14,15,18,26,27) with the knee extensor, the mean estimated mass of the quadriceps varies from 1.9 to 3.6 kg among studies and within a study the range can also be large. Thus, it is risky to assume a standard active muscle mass for all subjects in all studies. Furthermore, if the above criticism of the blood flow estimate is correct and a significant and variable portion of the active muscle mass is“missed” because of the catheter placement, then two large sources of error have been introduced before the key metabolites have even been measured! Considering the example presented above for the“washout” factor, if one has errors in blood flow and/or in the assumed active muscle mass, then the overall error could be enormous.
An example of how these assumptions can be carried to the extreme is found in a review by Wagenmakers et al. (38). They present(a-v) ammonia, glutamine, alanine, and glutamate data for a single McArdle's patient and then assume the blood flows without even measuring (a-v) O2 and an active muscle mass (they used 4 rather than the usual 8 kg). They did not consider washout and reported the fluxes per kilogram of active muscle!
The Human Knee Extensor Model
As reported by Dr. Saltin in this symposium, this model avoids some of the limitations described. However, one must consider whether it really represents the “whole human.” The blood flow per unit muscle is very high(much higher than could be provided to active muscle during whole body exercise), and the disturbance in the endocrine system is minimal. This does present the opportunity to increase a specific hormone by infusion, but is the control situation “normal”? Can we assume that the quadriceps is representative of all human muscle? As with any study of the human during voluntary exercise, the data from a muscle biopsy is a mixture of muscle fiber types and probably a mixture of fibers that were continuously active, some that were never active, and some that were active for only a portion of the exercise.
The Rat Hindlimb Model
This model is probably one of the most widely used to study different aspects of muscle metabolism. The two main approaches are to either exercise or perturb the “whole” rat and then quickly excise the muscles of the hindlimb, or to isolate the vasculature and the muscles of the hindlimb and artificially perfuse them. One of the main reasons for the popularity of the perfused rat hindlimb model is the ability of the investigator to have complete control over the quality and quantity of delivered substrates, ions, hormones, etc. Similarly, the duration and frequency of muscle stimulation can be controlled as well as the rate of muscle perfusion. Anatomically the rat hindlimb has the advantage of having a well-defined muscle mass and distinct and easily separated fiber types. Lastly, genetically uniform strains are available and the “life style” of the animals are easily controlled. All these advantage make this model well adapted to studying many different aspects of muscle metabolism.
On the other hand, the model has a number of limitations, particularly when using it to study amino acid and ammonia metabolism during long-term exercise. While the quality and quantity of perfusate is controlled, rarely has attention been given to the amino acid and/or ammonia concentration in the perfusate. Thus the exchange of these substances may not be the same as they are in vivo. Furthermore, the hindlimb is a heterogenous collection of muscles and fiber types. The metabolic characteristics among these fiber types are much greater than those in the human Vastus lateralis muscle. The metabolic and enzymatic characteristics (particularly for protein metabolism) of the muscles of the rat hindlimb are also vastly different from the human(16). In addition, in examinations of amino acid and ammonia metabolism during prolonged exercise, rarely have anteriovenous (a-v) differences for these parameters been made. When they have, it is unclear which muscles of the hindlimb are contributing to the a-v difference at a given point in time. Some of the muscles of the rat hindlimb fatigue relatively quickly. As a result, after only a brief exercise duration it is not clear which muscle groups are active and which are “metabolically dead.” Similarly, the mode of stimulation of the muscle recruits the fibers in reverse order to the “size principle.” This model is well suited for investigating a wide range of metabolic events; however, because of the artificial nature of the preparation and the type, size, and metabolic characteristics of the muscle mass used, it may not be suitable for examining amino acid and ammonia metabolism during prolonged exercise.
The Canine Hindlimb
With this model the muscles of the canine hindlimb can be used to examine different aspects of muscle metabolism. In some cases catheters are placed into the femoral artery and vein, and metabolism across the whole hindlimb is investigated. In contrast, a more “isolated” in situ preparation using only the gastrocnemius muscle has been used. Under these circumstances all blood vessels not directly supplying or draining the muscle are surgically occluded. Cannulae are then inserted into the popliteal vein and gracilis artery to facilitate blood flow and direct Fick measurements exclusively for the muscle. Several advantages of this model are that only a single, large muscle is contributing to the arteriovenous difference and the muscle can be serially biopsied. Furthermore, the vascular bed is auto-regulating and the intact circulation allows normal influences of the liver and other tissues on metabolites and hormones to be maintained. Finally, the muscle is highly oxidative with a fiber distribution similar to the human vastus lateralis muscle; once the muscle has been isolated, the sciatic nerve can be electrically stimulated, which allows the muscle to be used to examine long-term work.
Although this model is useful and it allows quantitative measures to be made, there are several limitations. For example, the electrical stimulation of the sciatic nerve is usually confined to twitch protocols. Also the recruitment of the muscle fibers is in reverse order to that during voluntary muscle contraction. There are also possible species differences in the distribution and isoforms of enzymes, and it is very difficult to control the“lifestyle” and gene pool of the animals used. Furthermore, while the vascular bed is auto-regulating and “normal” recirculation occurs, the animal is anesthetized and has had major surgery. These factors could alter the normal composition of the blood.
It is clear that the differences among both the models and the species complicates comparisons (16). Later in this review we will highlight specific species differences in the metabolism of ammonia and key amino acids. Differences even exist both within and among species in the fundamental property of amino acid concentrations in resting muscle(Table 1). For alanine there is general agreement regarding the resting concentration although the range reflects some variation among studies. In sharp contrast, the variation in glutamine between species(3.6-15.4 mmol·kg-1) and within a species (3.6-11.1 mmol·kg-1) is very large. The between species variation might be explained by methods, diet, training, etc.; however, the differences among different muscles of the rat are present even within a single study, suggesting that fiber types are critical. Similar findings are present for glutamate.
METABOLIC RESPONSES OF EACH MODEL
The “Whole-Body” Human Model
Broberg and Sahlin (4), Broberg et al.(5), Eriksson et al. (8), and Katz et al. (23) have published most of the work with this model in this area. They have found that ammonia efflux increases in an exponential relationship with exercise intensity. Two aspects are particularly noteworthy. First, the efflux becomes significant at low power outputs (30-40%˙VO2max) (see Fig. 1, panel A) when there is no change in intramuscular IMP and when few scientists would speculate that hypoxia is occurring. This has serious implications for the etiology of the ammonia. Secondly, the flux rate in high intensity exercise is extremely variable among their studies (see Fig. 1, panel A; efflux ranges from approximately 30 to 220 μmol·min-1 from one leg).
Broberg and Sahlin (4) Katz et al.(23) have reported very close agreement with the IMP accumulation and muscle ammonia increases during intense exercise, leading to the conclusion that AMP deamination is the source of the ammonia. This is in agreement with the work with the rodent model (see below), but it should be noted that these studies did not allow for ammonia that was released from the muscle and did not consider glutamine changes. Thus it is feasible that other sources of ammonia production exist even in this circumstance.
Broberg and Salhin (4) have published the only data regarding ammonia efflux during prolonged exercise. They found that moderate exercise caused a release of ammonia early in the exercise and that this increased progressively as the duration was extended (efflux was approximately 20 μmol·min-1 at 20 min and had increases to 90μmol·min-1 for 1 leg by 65 min). This does not match the responses in intramuscular IMP, and the authors speculated that the ammonia was formed from aspartate metabolism in the PNC but could offer no supporting evidence for this theory. It is noteworthy that measurements such as glutamine efflux were not made and the assumptions outlined above for leg blood flow and the mass of active muscle in the limb were employed.
How critical glutamine is as an alpha amino carrier in this model has not been examined in detail. Katz et al. (23) reported that glutamine release increased significantly in maximal exercise but presented no data. They do not appear to have considered that this release could mean that their determinations of ammonia efflux would underestimate the alpha amino nitrogen flux unless it was caused by washout. Eriksson et al.(8) reported a progressive and apparently exponential increase in glutamine release with increasing exercise intensity with the flux at 80% ˙VO2max being six times greater than rest (i.e., approximately 90 μmol·min-1 for 1 leg); however, the data were very variable and the efflux was not statistically significantly greater than rest. (Our experience and our impression from values in the literature is that glutamine release is highly variable physiologically both between and within subjects.) The potential problems of estimation of blood flow, active muscle mass, and washout were not considered. The only report of glutamine release during prolonged exercise that we are aware of is by Martin et al.(28). They found that the resting efflux of glutamine from one leg rose from 17 μmol·min-1 for one leg at rest to 43 μmol·min-1 after 40 min at 60% ˙VO2max; however, in this study the data were variable and nonsignificant.
The responses of alanine are far better established in this model; Felig and Wahren (9,39,40) reported that alanine was released from the exercising leg of humans and that the efflux appears to increase with both exercise intensity and duration. Both Eriksson et al.(8) and Katz et al. (23) confirmed that the release increases with exercise intensity and that the release is greater than rest at power outputs as low as 50-60% (8). Resting values are typically 20-30 μmol·min-1 for one leg and rise to 100-170 μmol·min-1 with demanding exercise, and during 30-50% ˙VO2max the efflux can rise to approximately 90μmol·min-1 for one leg over 4 h.
As with the other parameters, our knowledge of the responses of glutamate is limited to few studies. Nevertheless, they are consistent. Glutamate is taken up by human muscle at a rate one to three times greater than the resting rate (approximately 10 μmol·min-1 for 1 leg) during exercise. This is observed at power intensities as low as 40% ˙VO2max, and it does not appear to be altered by increased intensities. There are no measures of glutamate uptake relative to exercise duration. Both with ammonia and these key amino acids, the limited number of studies demonstrate that the responses are not dependent on large energy demands. Rather, at very moderate, steady-state conditions there is an increased exchange between the circulation and the active muscle, implying that there is increased metabolism under conditions when energy supply can easily match the demand.
There is tremendous variability in the effect of exercise on the intramuscular concentrations of glutamine and alanine. It is impossible to generalize about the expected change other than to say that in most cases the changes are quite small. In contrast, there is remarkable consistency in the reports for glutamate. It is very clear that the concentration decreases rapidly with the onset of exercise and then remains at a constant level(resting values are 3-3.5 mmol·kg-1 wet wt and they decline to 1-2 mmol·kg-1 within a few minutes or less) for up to 3.5 h of exercise (2,17,19,23). There is no evidence that exercise intensity is a major factor in this change. Glutamate is the only amino acid to show either a large or dramatic change in concentration. This is particularly surprising considering it is also the only amino acid to have a large uptake by the active muscle. This decline presumably reflects the central role that glutamate plays in so many transamination reactions. The regulation of these reactions and in the management of the intramuscular glutamate pool is one of many areas that require a great deal of study.
While this is the “model” that many scientists want to understand, little has been established quantitatively regarding its metabolic characteristics. In addition to the problems outlined above concerning reliable estimates of the active muscle mass and accuracy of blood flow measurement, there has not been a single study with this model in which flux and intramuscular concentrations of ammonia, glutamine, glutamate, and alanine were all determined. In fact, very rarely have any of these measures been reported. Most investigations have only reported the responses in the plasma concentration of ammonia. The only interpretation that one can place on this measure is that it is a reliable, qualitative measure of ammonia release in situations when the plasma concentration rises. That is, in every study that we are aware of using healthy subjects in which ammonia flux was measured when the ammonia concentration increased, the latter was associated with an increase in release. However, a lack of change in concentration does not necessarily indicate that there was no ammonia release, and changes in the plasma concentrations of amino acids are even less reliable as a reflection of flux. If one were to apply the data from the single knee extensor model (see below) to the whole body responses, the findings would strongly suggest that these limitations in the data for whole body exercise are extremely serious. In fact, we suggest that under most experimental circumstances it is impossible to conclude anything based exclusively on a change in the plasma concentration of an amino acid.
The Knee Extensor Model
We have conducted a series of studies(14,15,18,26,27) using this model to examine aspects of ammonia and amino acid metabolism; the exercise intensities and durations have spanned a wide range. Thus, by integrating these data one can begin to get a more complete impression of the dynamics of amino acid and ammonia metabolism. We emphasize that our work is not definitive and that we view it as providing a basis for more detailed investigations. There are limitations; for example, some studies did not include intramuscular determination of ammonia (18) or amino acids (14,18). Nevertheless the findings demonstrate that the metabolism of the human quadriceps in this paradigm is similar to that in whole body exercise and that this approach offers a number of advantages in quantifying the responses.
As in whole body exercise, ammonia efflux increases in an curvilinear fashion with increasing power output (see Fig. 2, panel B), and it also increases progressively with exercise duration. In addition, we have demonstrated that within seconds of the onset of intense exercise(14) the active muscle releases ammonia. This ammonia is not “trapped” in the muscle because of conversion to ammonium during the metabolic acidification associated with exercise(14). The concentration gradient from muscle to blood does not increase. This means that the assumption that the increase in intramuscular ammonia during intense exercise represents net ammonia production is incorrect if the muscle is being perfused. In as little as 3 min of exercise, approximately 25% of the ammonia leaves the muscle(14). Thus determination of the change in intramuscular ammonia underestimates the net formation without even considering the possible incorporation of ammonia into glutamine.
This in turn suggests that we must reconsider the generally accepted conclusion that in intense exercise all ammonia comes from AMP deamination since this idea is based on the stoichiometric matching of intramuscular ammonia and IMP changes. There is no question that AMP deamination is a major source of ammonia, but until these other metabolic “routes” for ammonia are evaluated, one cannot be certain that AMP deamination is the only active process producing ammonia. However, as we have discussed, to“match” the efflux of a metabolite to changes in intramuscular concentration, one must have an accurate assessment of both blood flow and the mass of the active muscle.
When the relationship between net ammonia production and IMP increase are considered (Table 2), there is good agreement between the two metabolites during intense exercise; however, during prolonged exercise it is obvious that net IMP production can account for a very minor portion of the ammonia formation. If the net glutamine formation is also considered, the“IMP contribution” becomes even less. Glutamine release is elevated four- to six-fold above the resting efflux of approximately 5μmol·kg-1·min-1 during exercise, although unlike ammonia it does not appear to increase with exercise intensity. This could be because glutamine production requires energy, and this limits the production rate during more intense exercise. In addition, the capacity of fast-twitch fibers to produce glutamine may be less than that for ammonia.
There are several interpretations for the lack of an ammonia-glutamine-IMP relationship. It may be that specific motor units are recruited and then recover, allowing for a transient increase in IMP that is then resynthesized while the ammonia is released more slowly. The IMP concentration in the biopsy would be the average of resting, inactive, and active fibers. This seems unlikely to be a major problem since no detectable increase in IMP is observed in biopsies until late in exercise as fatigue approaches, and this is also shown with single fiber analysis (34). In contrast, the muscle ammonia is moderately increased throughout the exercise, and the ammonia and glutamine release progressively increases. Thus, it appears that metabolic processes are continuously and progressively liberating alpha amino groups. This could be occurring by the deamination of the branched-chain amino acids and/or aspartate. The rate of these processes is greatest in intense exercise, but the total quantity liberated is greatest in prolonged exercise. Our knowledge regarding these events is very limited since most of the investigations using the rodent model or the “whole body” human have focused on brief, intense exercise.
Alanine release is increased four- to five-fold over the resting rate of approximately 5 μmol·kg-1·min-1 during prolonged knee extensor exercise, and this continues throughout the activity. As with glutamine, there is no indication that this process is particularly sensitive to exercise intensity. This is surprising since pyruvate production will obviously increase dramatically with exercise intensity. Alanine aminotransferase is supposed to be a near equilibrium enzyme, which suggests that availability of the other substrate, glutamate, may limit this step.
The responses of glutamate are in contrast to the other amino acids in that it is taken up by active muscle in large quantities. Typical resting values are 2-3 μmol·kg-1·min-1, and during prolonged exercise this increases to 5-15μmol·kg-1·min-1. In addition, despite the uptake there is a large and rapid decline in the intramuscular concentration which is remarkably similar to that seen in whole body exercise. The muscle continuously takes up glutamate at a rate that is three- to six-fold greater than that at rest, and yet the intramuscular concentration declines within the first few minutes of exercise. The decrease appears to be independent of exercise intensity and is very similar to that seen in whole body exercise. The concentration remains at this level for up to 3 h of exercise, and the only factor that we have observed that will modify the response is endurance training. In this case, the general response is very similar, but the actual intramuscular concentrations at rest and in exercise are higher(18).
Glutamate is not only the only amino acid taken up in considerable quantities during exercise, but it is also the only one showing a large intramuscular decline. The decline might be thought to compromise the many processes that use this amino acid, but certainly the large, continuous productions of ammonia, glutamine, and alanine do not support such a conclusion. It should be noted that presumably glutamate will be produced by transamination of the branched-chain amino acids; several other processes(e.g., ammonia, aspartate, and alanine formation) would conserve the glutamate carbons as alpha ketoglutarate. Glutamate is only completely“lost” from the muscle when glutamine is released. In the other processes the alpha amino group may exit, but the carbons are retained in another amino acid or as alpha ketoglutarate. Their ultimate fate depends on the activity of other pathways, an area about which we know almost nothing. Nevertheless, it is obvious that the combination of the decline in the intramuscular glutamate pool and the uptake of glutamate cannot account for the total demand for glutamate during prolonged exercise. The active muscle must have additional sources of this amino acid.
The data reviewed above are very similar to that reported for the human during whole body exercise. The knee extensor model allows for a far greater quantification of both exchange of substances between the circulation and the muscle as well as determination of the accumulation or washout of the substance from a known mass of active muscle. Thus the data allow us to investigate the plasma clearance of these substances. In our studies of ammonia and amino acid metabolism, we can demonstrate that the plasma“load” of ammonia, alanine, and glutamine are large (in the range of 10-17,000 μmol over 3 h). However, clearance mechanisms are so sensitive and dynamic that the arterial plasma concentration changes only a few micromoles. In one situation, we estimated that the quadriceps released over 16,000 μmol of glutamine in 3 h and the plasma concentration decreased 3μM during this time. Similarly, the active muscle takes up large quantities of glutamate; this can reach 11,000 μmol in 3 h (18) and yet the muscle concentration declines to a constant level and also the plasma concentration changes very little (increasing from 13-45 uM). This suggests that the splanchnic processes are extremely responsive to both the supply and the removal of amino acids. We know little about the regulation of these processes, but this model presents an ideal situation for studying them while large and relatively steady demands are being placed on them.
The Rat Hindlimb Model
As outlined previously, because of limitations in the metabolic characteristics of the muscles of the rat hindlimb, usually only short term, high intensity stimulation protocols have been used. Thus, there are few data and no complete data (i.e., both intramuscular and flux) for ammonia, glutamine, alanine, or glutamate over a continuous, prolonged exercise period. Furthermore, it has been suggested that electrical stimulation versus voluntary contraction results in some differences in the metabolism of some of these parameters. Despite these limitations and the lack of “real time” data, when the available data are examined as a whole, many observations about the relationship of these parameters regarding fiber type, exercise intensity, and duration can be made.
Most studies examining ammonia metabolism in the exercising rat hindlimb have been conducted by Dudley and Terjung (7), Goodman and Lowenstein (12), and Terjung et al.(29-31). They have concentrated specifically on the relationship between fiber types, IMP, and ammonia production. As would be expected with an increase in stimulation intensity (frequency), there was a corresponding increase in intramuscular ammonia concentration in all fiber types. The increase was also much more pronounced in the fast-twitch compared with that in the slow-twitch fibers (see Fig. 2). Judging from the changes in the intramuscular ammonia levels, the greatest potential for ammonia production occurs in the fast-twitch white, followed by fast-twitch red, and finally slow-twitch fibers. Most of these experiments used a brief, high intensity stimulation protocol; it is conceivable that the fast-twitch white fibers may not be the major ammonia source during prolonged, submaximal exercise (nor do the authors make such claims).
Interestingly, there does not seem to be the same relationship between IMP and ammonia production in all fiber types (see Fig. 2). The ammonia concentration closely matches the intramuscular changes in IMP in the fast-twitch white fibers following both electrical stimulation and voluntary running (7,29,31). The fast-twitch red fibers show only very small increases in IMP, and these are approximately 25% less than the ammonia levels at the same time points for the same conditions (7,29,31). This discrepancy is even more dramatic in the slow-twitch fibers, where any change in IMP is difficult to detect, although the ammonia levels increase several- fold(29). Terjung et al. (29) suggest that this difference is not because of ammonia production in the fibers where little or no IMP is produced, but rather results from an uptake of ammonia from the circulation by these muscles groups. However, the data could indicate that significant ammonia production occurs from sources other than AMP deamination (i.e., amino acid metabolism). For example, if all the IMP (which cannot diffuse from the cell) and ammonia contents in the different muscles are summed, there is still more than 1 mmol·kg-1 of ammonia produced above that which could be accounted for from IMP accumulation alone. Furthermore, this calculation does not allow for the efflux of ammonia (or glutamine) from the muscles, and thus the mismatch between IMP and ammonia production could be even bigger. These data add support to the belief that AMP deaminase is not the only source of ammonia production during exercise and also demonstrate that differences exist in the processes by which the various muscle fibers produce ammonia and IMP.
Few studies using this model report ammonia flux. However, Goodman and Lowenstein (11) have reported that at rest the muscles of the rat hindlimb release ammonia. When the muscles were stimulated to contract at an increasing intensity, the release of ammonia also increased(Fig. 1, panel C). Similarly, blood ammonia sampled after incremental treadmill running showed increasing ammonia concentrations(29). Turcotte et al. (37) demonstrated that the greatest ammonia release from stimulated rat muscle occurred early (within the first 5 min) and then steadily declined as exercise progressed in contrast to that reported for human muscle (see above). One possible reason for the finding that slow-twitch fibers have a greater glutamine concentration at rest (see Table 1) is that they have a greater glutamine synthetase activity. Similarly, the synthesis of glutamine requires energy and the higher levels in the slow-twitch fibers may reflect their greater capacity to maintain energy supply via oxidative means.
Meyer and Terjung (30) did not find any significant change in the intramuscular glutamine level during tetanic contractions lasting up to 5 min. Goodman and Lowenstein (12) reported drops of up to 23% with twitch protocols (1, 4, 5, and 6 Hz) ranging from 15 to 30 min. Furthermore, the drop in intramuscular glutamine tended to be greater as the twitch frequency increased. In a study by Dohm et al.(6) in which rats were run to exhaustion or swam for 2 h, intramuscular glutamine decreased by 19 and 15%, respectively. In all studies examined for this review, the muscle of the rat hindlimb released glutamine at rest (range = 6.7-16.0 μmol·min-1·kg-1)(12,21,35). However, it has been reported that upon stimulation the rate of glutamine release decreases(Fig. 1), and this was more pronounced as the stimulation frequency increases (12). This is in contrast to the data for humans (see above) and may be explained because the large proportion of the limb is fast-twitch fibers and these quickly fatigue. Thus the data gathered after the first few seconds may represent a much smaller active mass that is composed of more oxidative fibers. However, there are few data in this area.
Studies are consistent with regard to the response of intramuscular alanine during exercise; there is an early increase in the intramuscular alanine level followed by a gradual decrease (30,41). If the exercise is continued for long enough, the alanine concentration declines below resting values. At rest the rat hindlimb releases alanine, and during exercise this appears to continue (Fig. 1). However, as with glutamine there are not enough data to determine the precise response over an entire exercise bout.
The higher glutamate content in resting slow-twitch fibers(Table 1) may reflect a greater need for this precursor for glutamine synthesis. During exercise each fiber type shows a dramatic drop in glutamate content of 30-50%. While one study reported a very marginal uptake of glutamate (21), others have shown a release at rest (35). Unfortunately, no data are available on glutamate flux during exercise.
The major limitation with the rat model is the lack of data during prolonged exercise. Most studies have only pre and postexercise values and these are generally only for muscle concentrations, and the complementary flux data are lacking. The most probable reason for the latter limitation is that the muscle group has a very small absolute flow of blood; thus it would be difficult to obtain rapid samples of suitable volume for analysis. However, the model is best suited for the study of brief, intense exercise, and even if more studies were conducted with a prolonged exercise protocol they would be limited because of factors such as the inability to have serial biopsies from the same muscle and also the rapid fatigue of much of the hindlimb tissue.
The Canine Model
The canine hindlimb represents a potentially suitable model for examining amino acid and ammonia metabolism during prolonged exercise because it is not susceptible to many of the limitations listed above for the rat model. A number of excellent studies used this model to examine a single aspect of ammonia or amino acid metabolism, but this has usually been at rest or during acidosis, hypoxia, etc.
To our knowledge the most complete study using this model has been conducted in our laboratory (25). It involved the stimulation of the canine gastrocnemius muscle for 60 min at an intensity equivalent to 80-100% of twitch ˙VO2max. As with the rodent model, it appears that the higher the stimulation frequency, the greater the increase in intramuscular ammonia. The greatest increase in intramuscular ammonia was observed after 5 min of stimulation, and as stimulation continued there was no further increase. Similarly we observed that upon stimulation there was a dramatic increase in the release of ammonia by the muscle which peaked after 5 min (Fig. 1). This increase was also greater for the more intense (15-20 fold above rest) as compared with the less intense (4-8 fold) stimulation protocol. Also in agreement with the data for the rodent model was a progressive decline in ammonia release during the more intense protocol despite only a modest amount of fatigue. These data suggest that canine muscle has the capacity to produce substantial quantities of ammonia and that the majority of the production occurs very early during the exercise bout. The latter finding may be caused by the electrical stimulation.
There does not appear to be any consistent change in the intramuscular content of glutamine during exercise. On the other hand, intramuscular alanine behaves similarly to that in rodent muscle where there is an early increase in its concentration and then as exercise progresses its content decreases, sometimes below resting values(1,3,13,25,33,42). As in both the human and the rat, glutamate drops quite dramatically very quickly after the initiation of stimulation and then remains relatively unchanged throughout the remainder of the exercise bout.
The studies reviewed for this paper all consistently show a release of glutamine (9.4-14.3 μmol·min-1·kg-1) and alanine (6.7-23.7 μmol·min-1·kg-1) from both the whole hindlimb and the isolated gastrocnemius muscle(3,10,25,33). In contrast, an uptake of glutamate (3.6 μmol·min-1·kg-1) has been reported for the whole hindlimb (33), while a release(0.2-1.3 μmol·min-1·kg-1) was found for the isolated gastrocnemius muscle (25). These differences may be caused by the surgery or the number of tissues contributing to the a-v difference between preparations. During exercise both glutamine and alanine were released to a greater extent than at rest (25). The greatest release of these amino acids also came after 5 min of stimulation and then slowly declined over the next 55 min of stimulation. Furthermore, the release of glutamine was lower for the more intense than the less intense stimulation protocol. An interesting observation in the canine gastrocnemius model was the consistent release of glutamate from the muscle, even during exercise. In fact, the release of glutamate actually increased upon stimulation (25) and remained elevated for the first 30 min of exercise. This is a unique finding as most muscle preparations show an uptake of glutamate during exercise. These data are not easily explained and may be a species-specific difference in glutamate metabolism.
In general, the metabolism of ammonia and amino acids in canine muscle behave in a fashion similar to that of human muscle during prolonged exercise. The efflux of ammonia and glutamine and alanine are qualitatively and quantitatively very similar to that for the leg extensor model, with the exception that the ammonia efflux declines with exercise duration. The major aspect not in agreement is that noted above for glutamate. However, this conclusion is based on very few data. Furthermore, these data are limited to studies using electrical stimulation for muscle contraction, and no data are available for ammonia and amino acid metabolism during voluntary exercise in the dog.
Various paradigms using both human and nonhuman models have been employed to investigate ammonia metabolism during exercise. It is obvious from the comparisons conducted here that there are a number of similarities in the findings for the various models and also some differences. The latter result from such factors as basic species differences, the mode of inducing the muscle activity, and various limitations in how the data are collected. These differences have led to some confusion in the literature, but it should also demonstrate that each model has strengths that result in it being optimal for use in addressing specific hypotheses.
These models for studying muscle metabolism each have limitations; generally the researchers employing the model recognize these limitations, but frequently other scientists apply these results to a different situation in a different species. For example, numerous papers report ammonia data for humans(frequently that from a mixed venous sample) in which the data are interpreted on the basis of studies of rodent muscle. Often it is not recognized that the species differences result in large differences in concentrations of amino acids in the muscle, in differences among fiber types, and in different enzyme complements and/or different isozymes.
Understanding the physiological responses of the human during whole body exercise is often the ultimate goal in our research. However, when this“model” is used, its limitations frequently restrict the depth of understanding. It is difficult to establish the mass of active muscle represented in blood samples. Researchers have rarely, if ever, studied the full compliment of amino acids together with ammonia using direct Fick measurements in association with muscle biopsies. Thus the available data are incomplete.
The leg extensor model allows a more direct quantification of human muscle metabolism, and several studies combine both direct Fick measures and intramuscular data. It is reassuring that these data are generally in close agreement with the more qualitative data from investigations of whole body exercise in humans. These data have allowed us to begin to quantify the responses; it is obvious that ammonia production is very closely related to AMP deaminase activity in intense exercise. However, it can occur independent of IMP accumulation within a few minutes of mild exercise; this is even more apparent during prolonged, moderate exercise. The evaluation of the net production of “free” ammonia requires the quantification of both the intramuscular concentration and the quantity released. Furthermore, under most circumstances if this evaluation is not combined with similar measurements for glutamine, one is grossly underestimating the total alpha amino liberation.
The rat hindlimb model has been used extensively for examining intense exercise. The results consistently demonstrate a close relationship between IMP formation and intramuscular ammonia in fast-twitch white fibers. Thus, this is a useful model to examine this aspect of ammonia metabolism in detail. However, ammonia efflux has rarely been measured, and after a brief time a major portion of the muscles fatigue and are not active. This can present problems in resolving what functional portion of the muscle one is studying. This is analogous to the problem with studying the human and not knowing what portion of the muscle group(s) are actually active.
In prolonged exercise the rat and canine models show a pattern of ammonia release that is different from that seen in the whole body and knee extensor models. In these models the greatest release occurs early (within 5 min) and then either remains constant or decreases (depending on the stimulation intensity). The most logical explanation for this difference is that the electrically stimulated muscle has all fibers recruited. As a result, there is probably a greater ammonia production from the fast-twitch fibers early in the exercise; then these fibers fatigue. This pattern may possibly be even more exaggerated in the rat model given the metabolic characteristics of the different muscles of the hindlimb. Nevertheless, the magnitude of ammonia production per unit of muscle appears to be comparable across models.
It is interesting that the mechanisms responsible for ammonia production are probably the same among the experimental models, but the capacity of the various mechanisms may differ greatly between models as well as between fiber types within a given model. This difference is illustrated in the rat model where fast-twitch fibers have much greater capacity to produce ammonia than do the slow-twitch fibers. This has been speculated to be the same for the human, but this observation has been based solely on the end-exercise accumulation of IMP in different fiber populations. There is little doubt that in human subjects as exercise intensity increases to maximum, there is greater AMP deaminase activity and ammonia production from this source. It has also been shown that AMP deaminase activity is greater in fast than slow-twitch fibers in humans. However, during mild exercise and/or prolonged exercise with normal fiber recruitment, i.e., in voluntary exercise, it is most likely that a substantial part of the ammonia production occurs from other sources (i.e., branched-chain amino acid metabolism), particularly in the slow-twitch fibers. This idea is supported by the finding that human muscle has little or no increase in IMP at exhaustion in prolonged exercise, except for a modest increase in the fast-twitch fibers. While fiber type differences have not been studied, canine muscle certainly is also characterized by a lack of IMP accumulation despite continuous ammonia production during prolonged activity. This can also be seen to some extent in the slow-twitch fibers of the rat hindlimb muscles; even with high intensity stimulation, there was very little IMP accumulation despite substantial intramuscular ammonia increases. Therefore, it is apparent that at the same relative work intensity rat muscle probably has a greater ammonia production from AMP deaminase activity, whereas in either canine or human muscle the ammonia production arises more from other sources.
The data from the leg extensor model has illustrated that glutamine is an important and critical amino acid when considering total alpha amino exchange during prolonged exercise. When this amino acid is ignored, the total ammonia flux could be underestimated by as much as 50%. The production of glutamine, catalyzed by glutamine synthetase, is energy requiring. It was observed in the canine model that the greater the exercise intensity, the lower the glutamine production. It was suggested that this was because of the greater energy need for contraction; thus less energy was available for glutamine synthesis. In an examination of the human data, the same pattern appears to emerge. When the exercise intensity was increased from 70 to 80% of ˙VO2max, approximately 1100 μmol less glutamine was released by the muscle over the same exercise period (15,27). Little data are available for the rat; however, it has been shown that glutamine synthetase activity is higher in slow than in fast-twitch fibers. It was also observed that these fiber types have the highest and lowest resting glutamine levels, respectively. The limited available data for the rat also indicate that glutamine production is decreased from rest following electrical stimulation. Since the stimulation protocols used in the rat were of high intensity and short duration, they would require a substantial amount of energy.
All the models appear to demonstrate the same metabolic response for alanine metabolism. There is an early production and increase in muscle alanine levels which slowly decline as exercise progresses and glycogen decreases. While alanine is continuously released in large quantities, this probably represents carbon skeletons from carbohydrates (but only approximately 1% of the net glycolytic flux) and nitrogen from branched-chain amino acids. There appear to be no differences in the ability of one fiber type to produce alanine; however, the data for making this speculation are extremely limited.
Glutamate is a central amino acid in muscle metabolism, being involved as a substrate or a product in many reactions. This can be seen in rat skeletal muscle in which the highest resting intramuscular concentration of glutamate occurs in slow-twitch fibers. This matches the potential need for glutamate as substrate for glutamine production. Human muscle takes up glutamate during exercise, while canine muscle releases it. There is no immediate explanation for this finding, which may be caused by either the trauma of the surgery required for the canine model or a species-specific difference. On the other hand, all models show a rapid drop in the intramuscular glutamate pool within a few minutes of exercise, and then it remains at a low and constant value, even during hours of exercise. Thus, the intramuscular free pool is not a direct source of glutamate after the initial few minutes. This suggests that the intramuscular glutamate concentration is very tightly regulated at a low level during exercise, and this is can be maintained for hours despite large demands on the amino acid by several pathways. In addition, even when human muscle takes up large quantities of glutamate continuously during exercise, this quantity falls well short of the net demand. These responses present numerous questions about the regulation of glutamate.
In this review we have taken one small area of metabolism in which our understanding is limited, that of ammonia and amino acid metabolism. We feel that the various experimental models are all useful and provide valuable insight into the physiology. Each of these models has demonstrated in various ways that ammonia and amino acid metabolism are extremely dynamic and that their regulation is complex. It is also apparent that in many metabolic circumstances glutamate, alanine, and glutamine are critical amino acids with vital roles during exercise. However, each of these models for studying muscle metabolism has limitations which should be remembered when one is trying to interpret and extrapolate data among models.
Several common experimental models have been employed; there is no“best” model. The rat hindlimb is a very good model to examine aspects of AMP deaminase regulation during intense exercise particularly in different fiber types, whereas the canine muscle preparation has distinct advantages for following the temporal responses of ammonia and amino acid metabolism during prolonged exercise. The leg extensor model has distinct advantages for studying ammonia metabolism in either intense and prolonged exercise in humans, but ultimately most of us want to understand the metabolic events in humans during whole body exercise. Each has its strengths and weaknesses, and the optimal approach depends on the scientific question. Regardless of the procedure the scientist must resolve several basic questions including: what muscle groups and/or what fiber types are active? Are both the intramuscular concentrations and the flux of the various metabolites accounted for in the design? What is the mass of the active tissue? These are critical in the quantitative assessment of the metabolic events.
The work of T. E. Graham has been supported by NSERC of Canada. The studies using the leg extensor model were supported by grants from the Danish Medical Research Council, the Danish Natural Sciences Research Council, and Team Danmark held by our colleagues at the Copenhagen Muscle Research Center. DAM currently holds a Canadian MRC scholarship and is working at the Copenhagen Muscle Research Center.
Address for correspondence: T. Graham, Human Biology and Nutritional Sciences, University of Guelph, Guelph, Ontario, Canada N1G 2W1. E-mail:email@example.com.
1. Begum, S. J., M. M. Reddy, O. Ramakrishna, K. Indira, and K. S. Swami. Skeletal muscle protein metabolism under denervation atrophy in dog canis domesticus. Ind. J. Physiol. Pharmacol.
2. Bergstrom, J., P. Furst, and E. Hultman. Free amino acids in muscle tissue and plasma during exercise in man. Clin. Physiol.
3. Biolo, G., D. Chinkes, X. Zhang, and R. R. Wolfe. A new model to determine in vivo
the relationship between amino acid transmembrane transport and protein kinetics in muscle. J. Par. Ent. Nutr.
4. Broberg, S. and K. Sahlin. Adenine nucleotide degradation in human skeletal muscle during prolonged exercise. J. Appl. Physiol.
5. Broberg, S., J. Wahren, and L. S. Eriksson. Ammonia metabolism during exercise in patients with liver cirrhosis. In:Advances in Ammonia Metabolism and Hepatic Encephalopathy
, P. B. Soeters, J. H. P. Wilson, A. J. Meijer, and E. Holm (Ed.). Stockholm: Elsevier Science, 1988, pp. 130-140.
6. Dohm, G. L., G. R. Beecher, R. Q. Warren, and R. T. Williams. Influence of exercise on free amino acid concentrations in rat tissues. J. Appl. Physiol.
7. Dudley, G. A., and R. L. Terjung. Influence of aerobic metabolism on IMP accumulation in fast-twitch muscle. Am. J. Physiol.
248 (Cell Physiol
8. Eriksson, L. S., S. Broberg, O. Bjorkman, and J. Wahren. Ammonia metabolism during exercise in man. Clin. Physiol.
9. Felig, P. and J. Wahren. Amino acid metabolism in exercising man. J. Clin. Invest.
10. Galim, E. B., K. Hruska, D. M. Bier, D. E. Matthews, and M. W. Haymond. Branched-chain amino acid nitrogen transfer to alanine in vivo in dogs. J. Clin. Invest.
11. Goodman, M. N., R. Dietrich, and P. Luu. Formation of gluconeogenic precursors in rat skeletal during fasted-refed transition.Am. J. Physiol.
259 (Endocrinol. Metab.
12. Goodman, M. N. and J. M. Lowenstein. Studies of ammonia production by skeletal muscle in situ and in perfused preparations. J. Biol. Chem.
13. Gougoux, A. and P. Vinay. Renal metabolism and ammoniagenesis during acute respiratory alkalosis in the dog. Can. J. Physiol. Pharmacol.
14. Graham, T. E., J. Bangsbo, P. D. Gollnick, C. Juel, and B. Saltin. Ammonia metabolism during intense dynamic exercise and recovery in humans. Am. J. Physiol.
259 (Endocrinol. Metab.
15. Graham, T. E., B. Kiens, M. Hargreaves, and E. A. Richter. Influence of fatty acids on ammonia and amino acid flux from active human muscle. Am. J. Physiol.
261 (Endocrinol. Metab.
16. Graham, T. E., J. W. E. Rush, and D. A. Maclean. Skeletal muscle amino acid metabolism and ammonia production during exercise. In: Exercise Metabolism
, M. Hargreaves (Ed.). Champaign, Il: Human Kinetics, 1995, p. 131-176.
17. Graham, T. E. and B. Saltin. Estimation of the mitochondrial redox state in human skeletal muscle during exercise. J. Appl. Physiol.
18. Graham, T. E., L. P. Turcotte, B. Kiens, and E. A. Richter. Training and muscle ammonia and amino acid metabolism in humans during prolonged exercise. J. Appl. Physiol.
19. Henriksson, J. Effect of exercise on amino acid concentrations in skeletal muscle and plasma. J. Exp. Biol.
20. Hood, D. A. and R. L. Terjung. Amino acid metabolism during exercise and following endurance training. Sports Med.
21. Hundal, H. S., P. Babij, P. W. Watt, M. R. Ward, and M. J. Rennie. Glutamine transport and metabolism in denervated rat skeletal muscle. Am. J. Physiol.
259 (Endocrinol. Metab.
22. Jorfeldt, L. and J. Wahren. Leg blood flow during exercise in man. Clin. Sci.
23. Katz, A., S. Broberg, K. Sahlin, and J. Wahren. Muscle ammonia and amino acid metabolism during dynamic exercise in man. Clin. Physiol.
24. Lowenstein, J. M. The purine nucleotide cycle revised.Int. J. Sports Med.
25. MacLean, D. A., J. K. Barclay, and T. E. Graham. Electrical stimulation and amino acid and ammonia metabolism in the canine gastrocnemius muscle. Am. J. Physiol.
268 (Reg. Integr. Comp. Physiol.
26. MacLean, D. A., T. E. Graham, and B. Saltin. Stimulation of muscle ammonia production during exercise following branched chain amino acid supplementation in humans. J. Physiol. (Lond.)
27. MacLean, D. A., T. E. Graham, and B. Saltin. Branched chain amino acids augment ammonia metabolism while attenuating protein breakdown during exercise. Am. J. Physiol.
267 (Endocrinol. Metab.
28. Martin, I. K., A. Katz, and J. Wahren. Splanchnic and muscle metabolism during exercise in NIDDM patients. Am. J. Physiol.
269 (Endocrinol. Metab.
29. Meyer, R. A., G. A. Dudley, and R. L. Terjung. Ammonia and Imp in different skeletal muscle fibers after exercise in rats. J. Appl. Physiol.
30. Meyer, R. A. and R. L. Terjung. Differences in ammonia and adenylate metabolism in contracting fast and slow muscle. Am. J. Physiol.
237 (Cell Physiol.
31. Meyer, R. A. and R. L. Terjung. AMP deamination and IMP reamination in working skeletal muscle. Am. J. Physiol.
32. Millward, D. J., J. L. Bowtell, P. Pacy, and M. J. Rennie. Physical activity, protein metabolism and protein requirements.Proc. Nutr. Soc.
33. Muhlbacher, F., C. R. Kapadia, M. F. Colpoys, R. J. Smith, and D. W. Wilmore. Effects of glucocorticoids on glutamine metabolism in skeletal muscle. Am. J. Physiol.
247 (Endocrinol. Metab.
34. Norman, B., A. Sollevi, and E. Jansson. Increased IMP content in glycogen-depleted muscle fibres during submaximal exercise in man.Acta Physiol. Scand.
35. Ruderman, N. B. and M. Berger. The formation of glutamine and alanine in skeletal muscle. J. Biol. Chem.
36. Tullson, P. C. and R. L. Terjung. Adenine nucleotide degradation in striated muscle. Int. J. Sports Med.
37. Turcotte, L. P., P. J. L. Hespel, T. E. Graham, and E. A. Richter. Impaired plasma FFA oxidation imposed by extreme CHO deficiency in contracting rat skeletal muscle. J. Appl. Physiol.
38. Wagenmakers, A. J. M., J. H. Coakley, and R. H. T. Edwards. Metabolism of branched-chain amino acids and ammonia during exercise: clues from McArdle's disease. Int. J. Sports Med.
39. Wahren, J., P. Felig, G. Ahlborg, and L. Jorfeldt. Glucose metabolism during leg exercise in man. J. Clin. Invest.
40. Wahren, J., P. Felig, R. Hendler, and G. Ahlborg. Glucose and amino acid metabolism during recovery after exercise. J. Appl. Physiol.
41. White, T. P. and G. A. Brooks. [U-14 C]-glucose,-alanine, and -leucine oxidation in rats at rest and two intensities of running. Am. J. Physiol.
240 (Endocrinol. Metab.
42. Wolfe, B. R., T. E. Graham, and J. K. Barclay. Hyperoxia, mitochondria redox state, and lactate metabolism of in situ canine muscle. Am. J. Physiol.
253 (Cell Physiol.
Symposium: New Insights into the Control of Human Muscle Blood Flow and Metabolism Studied in Vivo