Fuels and pathways as designed systems for support of muscle work

Nothing is new, it is often said, except arrangement. And so it may be with the way animals adjust muscle energy metabolism according to specific locomotory needs. All animals are able to harness at least four sources of utilizable energy for the support of muscle work: ATP hydrolysis per se, phosphagen hydrolysis, anaerobic fermentations and oxidative metabolism. But all animals do not arrange and use all these sources in the same way. By close inspection of innovative ways that the same source of energy can be used, this paper aims to clarify: (i) mechanisms which ensure that the right energy source is activated at the appropriate time at the correct rate, and (ii) the properties of fuels which place limits or constraints on different energy-yielding processes.

The immediate source of energy tapped by most endergonic processes in the cell is, of course, that released by hydrolysis of ATP. In most tissues, the rates of such processes are low because most metabolic pathways operate rather sluggishly. However, muscle is exceptional: during transition from rest to work, its rate of ATP cycling or turnover (in flux units defined as μmolg^-1 min^-1) can increase by orders of magnitude, rather than by small percentage changes. In metabolic terms the first step in this transition involves activation of actomyosin (ATPase)-mediated hydrolysis of ATP, with maximum flux rates in man estimated at nearly 400units (Table 1). To maintain such high rates of work using exclusively endogenous ATP would require large storage depots in muscle, a requirement nature chooses not to meet, presumably because it would generally disrupt both the rates and regulation of metabolism (Atkinson, 1977). Fast-twitch or white type muscle (WM) usually contains only about 5-8 μmolg^-1 ATP while slow-twitch oxidative or red muscle (RM) usually contains even less (Guppy, Hulbert & Hochachka, 1979; Howald, Von Glutz & Billeter, 1978). If only these endogenous ATP supplies were available, they obviously would be very quickly exhausted during high work rates. Instead, as is well known, ATP levels are remarkably refractory to increased flux in a variety of species (see Gadian et al. 1981 ; Sutton, Jones & Toews, 1981; Sahlin, Palmskog & Hultman, 1978; Driedzic & Hochachka, 1976). The only way high ATPase rates can be sustained with minimal change in ATP level is through balanced ATP replenishment. At highest power output, the required replenishment of ATP is provided by phosphagen hydrolysis.

Skeletal muscle working at near maximal rates can be viewed as a temporary and fairly isolated system (Petrofsky et al. 1981). To match the need for ATP at such times, a minimum set of criteria must be met for a phosphagen to be a useful endogenous fuel.

All currently known phosphagens are substituted phosphoguanidinium compounds. The two best studied systems are phosphocreatine (PCr) and phosphoarginine (PArg), [formed from creatine (Cr) and arginine (Arg) by creatine phosphokinase (CPK) and arginine phosphokinase (APK) respectively]. Creatine itself is formed from methionine, glycine and arginine; in man, this occurs in liver and pancreas, while in some species the first step occurs in the kidney and the second in the liver (Walker, 1979). Creatine is accumulated in skeletal muscles, myocardium and brain, but to my knowledge mechanisms controlling storage amounts are unknown, although feedback repression of synthesis in the liver may limit total availability at these sites (Walker, 1979). Nevertheless, by usual standards, the total pool sizes of PCr + Cr are high (usually about 30μmolg^-1 in WM and somewhat lower in RM, heart and brain). On short- and long-term basis, the amount of PCr available for burst work seems to be controlled by hypertrophy, not by concentration adjustments (Holloszy & Booth, 1976). This is not the case for PArg. It too can be stored at high concentrations (up to 50 μmol g^-1 in some species) but in this case there seems to be a reasonable correlation between burst work capacities and the amount of PArg stored (De Zwaan, 1983).

For a phosphagen to be a useful ATP source, its ∼P group must be transferable to ADP at high rates and at appropriate times. In the case of PCr, this capacity is based upon (i) very high activities of the cytosolic isozyme of CPK and (ii) appropriate kinetic properties (Hochachka, Dobson & Mommsen, 1983a). Within any species, CPK activities are highest in WM, lower in RM, myocardium and brain; comparisons between species indicate that the most CPK occurs in animals capable of high burst speeds, sluggish species having lower activities. A similar relationship appears to hold for PArg kinases (De Zwaan, 1983). These adjustments in enzyme amount per gram muscle would, in themselves, favour high rates of ATP formation from phosphagen when and where needed, a process further facilitated by the kinetic properties of these enzymes.

Another characteristic of a useful phosphagen is that, except at extremes, its utilization should proceed with minimal effect on ATP levels. In this regard, compounds such as PCr can be viewed as being fitted for the job because the equilibrium constant (K_eq) for ATP formation by the CPK reaction is large; at pH 7·0, the K_eq for reaction (1) above is about 2 X 10⁹ (Dawson, Gadian & Wilkie, 1978; Gadian et al. 1981). As a result PCr can be almost completely converted with no change in ATP concentrations (Table 2). It would not do to use another compound (such as nucleotide triphosphate, NTP), with a Keq for hydrolysis similar to that of ATP because the concentration ratio of ATP/ADP would change along with that of NTP/NDP. With this kind of phosphagen, muscle metabolic transitions would sustain much larger fluctuations in ATP levels; using up three-quarters of such a hypothetical ∼P donor, for example, would cause a 75 % decline in ATP concentration. That is why compounds such as PCr, rather than NTPs, have evolved as muscles phosphagens and why the seemingly extreme equilibria of their ∼P group transfer reactions are, in this role, advantageous.

When fully depleting the phosphagen supply the CPK reaction might be expected to lead to ionic and charge perturbations. For example, in the resting state, 30 mmol 1^-1 PCr^2- would require similar amounts of a divalent counterion. Cr is of course uncharged and its accumulation in burst work would appear to leave the cell with an anion gap. This problem is alleviated by CPK coupling to ATPase, since in terms of charge, P ²” is equivalent to PCr^2- ; moreover, the former is a better buffer at near-neutral pH values than all known phosphagens. So the stoichiometry not only avoids serious osmotic or charge imbalances, but simultaneously minimizes perturbing effects of any [H⁺] changes.

The value of phosphagens as fuels would be greatly amplified if they were also usable in aerobic metabolism. For PCr, at least three such aerobic functions have been suggested; namely, in the shuttling of P_i between sites of ATP formation and utilization (Bessman & Geiger, 1981), in the preferential channelling of ATP to SR-bound ATPase (Levitskii et al. 1977), and in the facilitated diffusion of ATP (Meyer, Sweeney & Kushmerick, 1984).

In summary, then, two potential advantages can arise from using phosphagens in support of burst muscle work: high power output and effective buffering of ATP levels. We can assert with confidence that the first is more important than the second, because numerous invertebrates (England & Baldwin, 1983) and at least one teleost (G. P. Dobson, P. W. Hochachka & T. P. Mommsen, in preparation) are known to sustain very large drops in [ATP] during intense muscle work. Thus maintenance of [ATP] duringintense muscle work is neither universal nor absolutely necessary. On the other hand, in all systems we know about the power output of muscles working on phosphagen as fuel is higher than that obtainable from any other pathways.

Phosphagens, however, show some critical limitations, the most obvious being the 30-50 mmol I^-1 maximum storage level (placing a modest ceiling on the total amount of muscle work supportable by this fuel). Again, we can make this assertion with confidence, since muscles in many animals store less, but none are known which store more, than this amount. While my analysis cannot specify the exact basis for limits to phosphagen storage, the most likely reasons seem to be either: (i) that phosphagens are strongly anionic (and if levels were too high they would begin to create undesirable effects), or (ii) that the amount of phosphagen-derived P₁ could lead to significant metabolic perturbations. Whatever the underlying reasons, the modest limit put on storage amount means that phosphagen-based work is only possible for very short times (5-10 s in mammals, for example) ; any further muscle work therefore requires a back-up pathway for ATP replenishment.

Under 02-limiting conditions, anaerobic glycogenolysis or glycolysis appears to be the main ‘back-up’ mechanism for ATP replenishment after phosphagen supplies are depleted. In invertebrates, the potentials also exist for energetically more efficient fermentations pathways, such as glycogen → succinate or glycogen→ propionate, as well as others, with or without coupling to aspartate → succinate (see Hochachka, 1980).

Since the high power output supported by anaerobic glycolysis (second only to phosphagen hydrolysis) also exceeds a critical threshold (Petrofsky et al. 1981), muscle in this state is still a relatively isolated system. As with phosphagens, this sets conditions and limits which must be met in order for a compound to be a useful fermentable fuel for muscle work.

This criterion is well met by muscle glycogen, which in vertebrate WM can be stored at about 100 /xmol glucosyl unit g^-1 ; a level which can be adjusted according to need. In invertebrates with higher relative dependence on this pathway, muscle glycogen levels may constitute a significant fraction of the tissue weight. To increase efficiency of storage, some muscles resort to specific glycogen-membrane associations or to larger intracellular granules (Hochachka, 1980). Some gastropods store galactogens along with glycogen but the former do not appear to be used for muscle work (De Zwaan, 1983). In many marine invertebrates, the free amino acid pool is large and several of these (aspartate and the branched chain amino acids in particular) can be used at low rates to supplement glycolysis (Collicutt & Hochachka, 1977).

This property is particularly well met by glycogen which yields 3 mol ATP/ glucosyl unit. Thus, the complete fermentation of 100μmolg^-1 generates at least 300 μmol ATP, and the total work possible on this fuel is about 10 times that supportable by PCr (because the latter is stored at only 30 μmol g^-1 and only generates maximally the same amount of ATP). The amplification in the aspartate→ propionate path is only two-fold, as it is in the glucose → lactate path ; that is one reason why glycogen is a better fuel for anaerobic muscle work than either aspartate or glucose.

Current evidence indicates that the main reason why sequences such as aspartate → propionate or branched chain amino acids → volatile fatty acids are not much utilized to support anaerobic muscle work is because they do not supply sufficient power (see De Zwaan, 1983). In contrast, the high power output of anaerobic glycolysis is probably the chief reason why this pathway is almost universally used in support of anaerobic work (Table 1 ). The ‘flare up’ characteristics of muscle glycolysis are based upon high catalytic activities of glycolytic enzymes and upon the occurrence, at virtually every step in the glycogen→ lactate conversion, of a muscle-specific isozyme form or forms. The control properties of glycolysis are largely determined by the catalytic and regulatory properties of the muscle isozymes of phosphorylase, phosphofructokinase (PFK) and pyruvate kinase (PK), although other isozymes may also play a role (Hochachka et al. 1983a). Stimulation of muscle glycolysis is triggered by a variety of signals (hormones, ions and metabolites), although the mechanisms which ensure that the pathway is appropriately phased in with other ATP-generating pathways is frequently under-emphasized. Nevertheless, during anaerobic work it is often observed that phosphagen hydrolysis usually precedes full glycolytic activation; PCr utilization can, for example, approach completion before glycolysis is phased in. This is because glycolysis is competing with CPK as a source of ATP (APK in invertebrate systems). Three enzymes are involved : CPK, phosphoglycerate kinase (PGK) and pyruvate kinase (PK). These enzymes are present in the 10^-4moll^-1 range (Lebherz, Petell, Shackelford & Sardo, 1982), while free [ADP] in resting muscle occurs in the 10^-6moll^-1 range (McGilvery, 1975). During early phases of work, both the high activity of CPK and its affinity for ADP ensure that this pathway dominates (see above). On the other hand, the kinetic properties of PGK and PK are suitably tailored for back-up function. In the case of PK, for example, the K_d for phosphoenolpyruvate (PEP) is drastically lowered from a non-physiological range to values between 0·01 and 1 mmol 1^-1 on addition of ADP. PEP has the same effect on binding of ADP (i.e. the binding of one substrate leads to about an order-of-magnitude increase in enzyme affinity for the co-substrate, Dann & Britton, 1978). This means that as anaerobic work continues and [ADP] rises, PK becomes ever more competitive and serves to complete the coupling of myosin ATPase and glycolysis (see Hochachka et al. 1983a). Although the rate of work supportable by this system is related in some complex way to the catalytic potentials of the enzyme pathway per se (Emmett & Hochachka, 1981), it is clear that the enzyme content of a pathway cannot be elevated without limit because of potentially debilitating effects of the end products. It is the need to minimize such problems that defines another criterion of useful fermentable fuels.

In the first place, it is usually overlooked just how much H⁺ is cycled through normoxic metabolism, with no net accumulation or depletion. A simple calculation can show that in a 70-kg man (with a resting metabolic rate of about 700 mmol O₂ h^-1) over 100 mol of protons are produced per day. The main sink for this huge quantity of H⁺ is oxidative metabolism, and a close balance between rates of H⁺ production and removal ensures stable pH. Although it is well known that the balance of this system breaks down during anaerobiosis, it is usually also overlooked that pathways such as glycogen → lactate represent net H⁺-consuming reactions at metabolically relevant pH values. At pH 7’4, for example, this pathway consumes 0-4H⁺/glucosyl unit, while the pathway glycogen → propionate at this pH consumes an order of magnitude more H⁺/glucosyl unit (Hochachka & Mommsen, 1983), indicating that both H⁺ stoichiometry and ATP yield are pathway specific. In both cases, it is the coupling with ATPase that leads to net H⁺ production. This important insight emphasizes that the metabolically relevant relationship is between H⁺ and ATP, not between H⁺ and glucosyl units. For the (glucose → lactate)-ATPase couple, 1 μmol of ATP is cycled/μmol H⁺ produced, whereas for the (glycogen → lactate)-ATPase couple, 1·5 μmol of ATP are cycled through/μmol H⁺ generated. In succinate fermentation, 2μmol ATP are cycled/μmol H⁺, whereas in glycogen→ propionate, up to 3·5 μmol ATP are cycled/μmolH⁺ formed (Hochachka, 1983). The strategy of maximizing the amount of ATP turned over per mol of H⁺ produced in muscles working anaerobically at its limit would mean a balance between H⁺formed and H⁺consumed during ATP cycling, as in aerobic metabolism (Vaghy, 1979).

This could be achieved by alcohol fermentations, but such pathways do not occur in muscle for they generate their own set of problems. Nevertheless, it is clear that Nature can, and has, invented anaerobic ATP-yielding pathways which, as in oxidative metabolism, fully consume the H⁺ released during ATP hydrolysis. Why, then, do only a few animals employ such mechanisms? One possibility is that protons perse, may play important, but thus far largely unrecognized, roles during O₂-limiting periods.

From current data, we have considered four such roles for H⁺ production: (i) the creation of conditions which are more favourable for unloading available O₂ (i.e. the Bohr shift) ; (ii) the facilitation of phosophagen hydrolysis; (iii) the establishment of a pH optimum for glycolysis, or at least for allowing its catalytic or regulatory potentials to be expressed; and (iv) the facilitation of lactate efflux (Hochachka & Mommsen, 1983). Even if H⁺ favours some or all of these related functions, its production cannot proceed indefinitely. That may explain why glycogen is stored in muscle far below possible levels; in some species (Hochachka, 1980), it may be stored at over 1 mol 1^-1 in liver (where the main end product of glycogen metabolism is glucose, not lactate and H⁺). That is also an important reason why a number of properties - such as (i) stored [glycogen], (ii) catalytic capacities of glycolytic enzymes, (iii) myosin [ATPase] and (iv) muscle buffering power - all co-adapt. Upward adjustments in any one of the parameters usually means adjustments in them all. Yet even the best adapted glycolytic systems tend to be critically time-limited, which is why sustained muscle work at any reasonable rates must be supported by aerobic, ATP-replenishing pathways.

On transition from rest to maximum work in mammals, the metabolic rates of the muscles involved increase by an order of magnitude or more. The necessary ATP is derived from the complete oxidation of carbohydrates, fats and proteins or amino acids. Interestingly, most of the enzymes of /3-oxidation, the Krebs cycle, and the electron transfer system do not appear to occur in isozymic form, so the dependence upon muscle-specific activating signals is even greater than in anaerobic metabolism. Another distinguishing feature of oxidative metabolism is that the complete combustion of most substrates takes place in at least two separate compartments, the cytosol and the mitochondria. This introduces problems of transport of (i) carbon substrateperse, (ii) reducing equivalents and (iii) adenylates per se — all of which may influence flux rates (Holloszy & Coyle, 1984; Tager et al. 1983). Because no single fuel feeds oxidative metabolism and because aerobically working muscles remain perfused, relatively open systems, the criteria for useful substrates depend upon whether the fuel is stored in muscle (endogenous) or in other depots (exogenous).

The main endogenous fuels in mammalian muscles are glycogen and triglycerides. Both pools are expandable during short-term training or in long-term adaptation (Holloszy & Booth, 1976). In salmon and squid, for example, expendable proteins are dominant energy sources during migration, although there is uncertainty as to pathways of mobilization of the resultant amino acids (Mommsen, French & Hochachka, 1980). The amino acid pool is also utilizable for aerobic metabolism in many invertebrates (Storey & Storey, 1983). Interestingly, levels of endogenous fuels in oxidative fibre need not be as high as in glycolytic fibre because of higher molar yield of ATP.

The most striking difference observed in comparing phosphagens and glycogen in anaerobic ATP generation with fuels for aerobic metabolism is the immense increase in yield of ATP/mol of the starting substrate. For triglyceride (calculated for glycerol +3 palmitate), free fatty acids (calculated for palmitate), glycogen, glucose, proline and lactate, the respective molar yields of ATP are 403, 129, 38, 36, 21 and 18, assuming complete combustion; these values are 18-400 times higher than the molar yield of ATP from phosphagen hydrolysis, and 6-130 times the yield from anaerobic glycogenolysis. The advantages for sustained muscle work are self-evident.

What then are the functions of AMP deaminase in species which possess it? The most likely answer is that by supplying IMP for adenylsuccinate synthetase, AMP deaminase sets the stage for fumarate formation from asparate and, thus, for augmenting Krebs cycle intermediates when they are needed. This would explain the occurrence of this enzyme in vertebrate muscles, as well as its absence in invertebrates, where proline and glutamate are abundant and serve anapleurotic roles (De Zwaan, 1983; Storey & Storey, 1983). It also explains why IMP, in this view, a potentially adaptive end product of aerobic metabolism, may have arisen.

This obvious need for glucose by vertebrates is met by liver glycogen depots ; for FFA, by adipose fat. When lactate and amino acids are used as fuels they are mobilized mainly from glycogen and protein stores in non-working muscles (see below).

Because aerobically-working muscles are perfused, a critical feature of any exogenous substrate is that it must be capable of sustaining adequate flux rates. In units defined as μmol g^-1 min^-1, the flux rates required depend upon the molar yield of ATP. Thus an ATP turnover rate in human muscle of 20μ molg^-1 min^-1 could be supported by a palmitate flux of 0·15 units, a glucose flux of about 0·6 units, a lactate flux of about 1 ·2 units and an O₂ flux of about 3 · 3 units. The highest flux is required by O₂, which of course is an important reason why its transfer from lungs or gills to muscle mitochondria is facilitated by haemoglobins (or their analogues) and by muscle myoglobins. On the other hand, the more energetically efficient a fuel is (in terms of ATP/substrate burned) the less rapidly it need be fluxed from depot site to working muscle. That is why enzymes (e.g. acylCoA synthetases) mobilizing efficient fuels can occur at lower activities per gram muscle than those mobilizing relatively inefficient fuels or those operating in anaerobic metabolism.

In general, uptake at the working muscle varies with delivery rate (perfusion) and with plasma [substrate], so this feature requires provision for plasma [substrate] varying with working needs. Plasma glucose often does not satisfy this provision at low work rates, but may do so at high ones (Jones et al. 1980). Plasma FFAs on the other hand increase in availability during modest exercise making them a quantitatively important fuel (Maughan, Williams, Campbell & Hepburn, 1978) ; in man, however, their contribution decreases during heavy exercise (Jones et al. 1980).

The use of lactate as a carbohydrate fuel appears to confer at least three advantages. First, it conserves endogenous glycogen in aerobic fibres. Secondly, the high activities in FG fibres of isozymes which are kinetically suited for lactate generation and, in more oxidative type fibres, of lactate oxidase isozymes ensures high substrate flux capacities. Thirdly, lactate fluxes can be geared to energy needs: by the F₂ flux (uptake by muscle) which is dependent upon plasma [lactate] and can vary over ranges at least as large as the scope for aerobic metabolism, and by the Fi flux (generation of lactate) which varies with the energy needs of aerobically-working muscles (i.e. varying with ⁠). These relationships suggest that the greater the running speed, the higher the steady state plasma [lactate], the fluxes of lactate to aerobically working muscles, and the In fact, during heavy exercise in man, lactate may be favoured as a substrate while the use of plasma FFAs seems actually to slow down (Jones et al. 1980), an organization presumably selected because of the higher power output achievable by lactate metabolism.

It is widely observed in vivo that as exercise intensity increases, rises to a maximum level at which time further muscle work depends upon anaerobic glycolysis. The final question thus arises of how this integration of aerobic and glycolytic pathways is achieved. Although many regulatory signals may be involved (see Hochachka & Somero, 1984), the main mechanisms seem to revolve around acceptor control of muscle mitochondrial function. According to current evidence (e.g. Chance et al. 1981), mitochondria in non-working muscle are in state 4: flux rates through the electron transfer system (ETS) are low because of limiting [ADP] or [Pi]. During muscle work, mitochondrial respiration and phosphorylation rates increase (the mitochondria are said to enter state 3) as [ADP] and/or [P₁] increase. While the overall response in vitro is hyperbolic, at K_m substrate levels respiration increases directly with [ADP] or [P₁] (Jacobus, Moreadith & Vandegaer, 1982), which is presumably also why work rates vary in a linear fashion with [Pi]/[PCr], quantified in vivo (Chance et al. 1981). Aside from being consistent with recent studies of isolated mitochondria (Jacobus et al. 1982), these in vivo studies also satisfactorily explain the advantage, with endurance training, of increasing the capacities of the electron transfer system (ETS) and oxidative phosphorylation (Davies, Packer & Brooks, 1981): for any given submaximal work rate in trained muscle, flux through the ETS per gram will be lower, because the amount of ETS per gram is up to two-fold higher. Muscle mitochondria from endurance-trained individuals, therefore, almost necessarily operate at lower [ADP], Secondly, because in vivo operation occurs at lower [ADP] ranges and because the maximum capacity for state 3 flux is elevated, the overall metabolic scope for activity is elegantly expanded. Finally, when the state 3 capacity is surpassed in either trained or untrained muscle, [ADP] and [Pi] rise even further, creating conditions favouring glycolytic competition for both. This ensures in all species, irrespective of speed capacity or size (Seeherman, Taylor, Maloiy & Armstrong, 1981), a phasing in of anaerobic glycolysis at the appropriate time : when the ATP replenishing capacity of oxidative phosphorylation is being surpassed.