Integrating Metabolic Pathways in Post-Exercise Recovery of White Muscle

There have been many studies addressing the metabolic changes occurring in fish white muscle during recovery from high-intensity exercise (Milligan and Wood, 1986a,b;Milligan and McDonald, 1988; Dobson and Hochachka, 1987; Mommsen and Hochachka, 1988; Pearson étal. 1990) but no single study has determined all of the variables required to understand the control of post-exercise energy metabolism. While measurement of total (i.e. free+bound) tissue levels of adenylates and phosphocreatine (PCr) can give some insight into the energy status of the tissue, the appropriate metabolite ratios and calculated free concentrations are much more informative.

This study addresses changes in energy metabolism following high-intensity exercise in rainbow trout using analysis of pH (intracellular, arterial) and the substrates of carbon (lactate, pyruvate, glycogen) and energy (Cr, PCr, ATP, ADP, AMP, IMP) metabolism. Our goal was to obtain a more comprehensive appraisal of the regulation of recovery metabolism than is presently available, through the use of appropriate metabolite ratios and calculation of free metabolite concentrations.

Rainbow trout [Oncorhynchus mykiss (Walbaum)] of both sexes (mean mass 560±89g; S.E.M.; N=75), obtained from West Creek Trout Ponds, Aldergrove, BC, were held outdoors in a circular tank supplied with flow-through dechlori-nated tap water at 8–12°C. Fish were fed daily by hand to satiation (Ocor Pacific Salmon Feed; Moore-Clarke).

Trout (starved for one day) were anaesthetized in a buffered (NaHCO₃, 2 g l⁻¹) MS-222 solution at a concentration of 1:6000. Cannulations were performed during forced ventilation with a buffered, chilled (15°C), aerated (100% oxygen) MS-222 solution (1:16000). Fish were cannulated in the dorsal aorta (Soivio et al. 1972) and allowed to recover for 48h in an opaque black acrylic box (5cm × 40cm × 20cm) at 8–12°C.

Fish were transferred by net to a Brett-type swim tunnel. After 15–20min at 0.25 body lengths s⁻¹, flow was increased to the maximum the fish could attain. The speed was held constant until the fish could no longer maintain this velocity (about 3min), at which time the speed was decreased by about 30%. This speed was held for about 5 min and then gradually increased and the cycle begun again. The flow rate was continually oscillated in this way, as the maximum attainable speed gradually decreased, until the fish could no longer maintain its position in the swim tunnel even at the slowest speeds. At this point, the fish did not respond to being grasped by the investigator. The entire exercise procedure took approximately 25–30 min to complete.

Fish were either killed immediately following exercise (at exhaustion) or transferred back to the holding box. Fish were then randomly assigned to a treatment group and killed after 2, 4, 8 or 24 h of recovery, as required. Resting (unexercised) fish were kept in the holding box for at least 48 h prior to sampling.

At the specified sampling time, 2 ml of blood was withdrawn via a dorsal aortic cannula and placed on ice. Fish were then killed by a 2 ml injection of Somnitol (65 mg ml⁻¹ sodium pentobarbitol) to minimize struggling. Once ventilation ceased, a 1 cm thick cross-sectional slice was taken immediately posterior to the dorsal fin and freeze-clamped in aluminium tongs pre-cooled in liquid nitrogen.

A section of epaxial white muscle was ground to a fine powder in a mortar cooled in liquid nitrogen. Care was taken to remove any pieces of skin and bone prior to grinding. The sample was kept submerged in liquid nitrogen throughout. The sample (approximately 1 g) was then transferred to a pre-weighed chilled test tube containing 1 ml of ice-cold 7 % perchloric acid (PCA) and re-weighed immediately any remaining liquid nitrogen had boiled off. 3 ml of cold PCA was added and the mixture was homogenized using an Ultra turrax tissue grinder. Throughout homogenization the tube was held in a slurry of salt water and ice at –5 to –10°C. Two samples (200 μl) of the homogenate were removed and stored at –80°C for later determination of glycogen. The remainder was centrifuged at 12 000 g for 9 min (4°C). A sample of the supernatant was removed and neutralized with Tris and KOH and centrifuged at 12000g for 3 min at 4°C. The supernatant pH was checked and adjusted with 10 mol l⁻¹ KOH or HO as necessary. The neutralized extract was stored at –80°C until needed. The extraction efficiency of this method for the labile high-energy phosphates was assessed, and both ATP and PCr were shown to undergo less than 5 % hydrolysis.

High performance liquid chromatography (HPLC) was used to measure ATP, ADP, AMP and IMP, each clearly separated from other purine and pyrimidine nucleotide phosphates. The procedure was carried out using an LKB2152 HPLC controller and 2150 titanium pump coupled to a 2220 recording integrator. The separation was performed on an Aquapore AX-300 7 μm weak anion exchanger (Brownlee laboratories) eluting at 2 ml min⁻¹ at 55°C (Parkhouse et al. 1987, with modifications). Elution was isocratic for the first5min, using 60 mmol l⁻¹ KH₂PO₄ (pH 3.2) followed by a linear gradient from 60 mmol l⁻¹ KH₂PO₄ (pH 3.2) to 750 mmol l⁻¹ KH₂PO₄ (pH 3.5) over 10 min. This concentration and pH were then maintained for 12 min. The column was re-equilibrated for 6 min with starting buffer before the next run. Adenylates were detected at 254 nm using a Bio-Rad flow-through ultraviolet monitor. Standard curves were constructed for all metabolites over the relevant concentrations by preparing a mixed solution in 60 mmol l⁻¹ KH₂PO₄. The standard curves were linear throughout the range required (r²=0.99). The coefficient of variation between duplicates was always less than 5 %.

Analytical reagent grade KH₂PO₄ was pre-purified (Reiss et al. 1984, with modifications) by passing a Imoll⁻¹ stock solution through a Bio-Rad Econo column, 5 cm diameter × 30 cm) packed with an anion exchange resin (AG1X8, chloride form), a cation exchanger (chelex 100, sodium form) and activated charcoal (14–60 mesh). The solution was kept at 4°C and constantly recirculated through the column by a peristaltic pump. Immediately prior to use, the stock buffer was diluted, brought to the appropriate pH and vacuum filtered (0.22 μm). With this technique, baseline disturbances caused by the impurities present in the high-concentration buffer were reduced by 74 %.

All determinations were performed on a Perkin-Elmer Lambda 2 ultraviolet/ visible spectrophotometer. Lactate, pyruvate, PCr and glucose were measured using routine NAD⁺/NADH-linked assays, according to Bergmeyer (1974). If the coefficient of variation between duplicate determinations was greater than 5 %, a third run was performed and the outlier discarded. All assays were validated with appropriate standards.

Mean intracellular pH (pHi) was determined using 5,5 dimethyl oxazolidine 2,4 dione (DMO) distribution (Milligan and Wood, 1986b) and a tissue homogenate technique (Pörtner et al. 1991).

Approximately 12 h prior to sampling, trout were injected through the dorsal aortic cannula with 10 μCi of [³H]mannitol and 2.5 μCi of [¹⁴C]DM0 (New England Nuclear) in Cortland’s fish physiological saline (total volume 500 ul). The pH of the arterial blood sample collected just prior to Somnotol injection was measured within 4 min using a Radiometer microelectrode (type E5021) main-tained at 10°C using a recirculating water bath and linked to a Radiometer 26 pH meter.

White muscle levels of ³H and ¹⁴C radioactivity were measured in 1 ml of the PCA extract used for metabolite assays using Amersham ACSII aqueous fluor. All samples were counted on an LKB 1214 Rackbeta liquid scintillation counter using dual-label quench correction. Total tissue water was determined by drying a 1–2 g sample to constant mass in an oven at 75 °C. The pK for DMO was taken from Malan et al. (1976).

Approximately 200 mg of muscle was ground under liquid nitrogen and added to a 1.5 ml Eppendorf tube. The tube was then quickly filled with a solution of 160 mmol l⁻¹ KF and 1 mmol l⁻¹ disodium nitrilotriacetate, stirred briefly with a needle and capped. The insoluble fraction was spun down (3–5 s in a microcentri-fuge at room temperature). Samples of the supernatant were taken for repeated measurement of pH as with the blood samples. Variation in pH between replicates from the supernatant was usually less than ±0.005 pH units and from the same muscle powder less than ±0.02 pH units.

Free cytoslic concentrations of ADP and AMP were calculated while taking into account the dependence of the equilibrium conditions on pH and [Mg²⁺] (Connett, 1988, 1990). We assessed three variables that reflect the energy state of the cell: Cr charge=[PCr]/([Cr] + [PCr]), R_ATp=[ATP]/([ATP]±[ADP]_f± [AMP]_f); and EC=([ATP]±0.5[ADP])/([ATP]±[ADP]±[AMP]). The values of equilibrium constants (K_eq) selected (Table 1) were corrected to conditions of ionic strength equivalent to that in vivo (μ = 0.17–0.2) and wherever possible to a temperature of 10°C using the Van’t Hoff equation. Implicit in these calculations is the assumption that all species involved are homogeneously distributed throughout the cell.

Total magnesium was calculated for values of [Mg²⁺]_f between 0 and 20 mmol l⁻¹ using a binding site approach (Connett, 1985). Known values for total magnesium concentration of rainbow trout white muscle (approximately 19 mmol l⁻¹ both at rest and at exhaustion; Parkhouse et al. 1987) were then used to select the most likely value for the actual intracellular [Mg²⁺]_f. Since this approach yields only an estimate of [Mg²⁺]_f, the effects of reasonable variations in [Mg²⁺]_f on the model calculations were assessed.

All data are reported as means±S.E.M. Metabolite differences between rest and/or the various recovery times were assessed using a one-way analysis of variance (ANOVA) followed by Tukey’s HSD on those variables that showed a significant F ratio. The data were log transformed where necessary.

In the case of pH, all statistical analyses were performed using calculated [H⁺]. The two techniques for estimating intracellular pH were compared using non-parametric tests (sign test, signed-rank test and Wilcoxon signed-rank test) because the DMO distribution technique produces pH estimates with greater variance than the homogenate technique.

The homogenate technique gave estimates of pHi which were not statistically distinguishable from the estimates obtained using the DMO distribution technique (Table 2). Since the homogenate technique appeared to yield greater precision, all subsequent calculations were based on estimates of pHi derived from the homogenate technique.

There was a stoichiometric relationship between [ATP] and [IMP] at all times (Fig. 1). Since neither [AMP] nor [ADP] changed significantly during exercise or recovery, total purine nucleotides (TPN) also did not change (Table 3). Similarly, [PCr] + [Cr] remained unchanged, with the decrease in [PCr] due to exercise being accounted for by the increase in [Cr], In contrast to [ATP] (Fig. 1), [PCr] was fully recovered by 2 h post-exercise (Table 3) and did not change significantly there-after. Changes in creatine charge paralleled the changes in [PCr] (Table 3), declining to approximately 16% of the resting value at exhaustion. By 2h post-exercise, creatine charge had fully recovered.

[Lactate] increased more than 10-fold at exhaustion and this increase could be accounted for by the decrease in glycogen content (Table 4). Recovery of [lactate] to resting levels was relatively slow (approximately 2 μmolg⁻¹ h⁻¹) but by 24h [lactate] had returned to resting levels. Glycogen repletion was similarly slow but by 24h [glycogen] was significantly different from the value at exhaustion. [Pyruvate] increased at exhaustion to nearly three times the resting value and remained elevated over the first 2h of recovery, after which time it decreased rapidly, recovering entirely by 8h post-exercise. The lactate/pyruvate ratio increased at exhaustion and over 8 h of recovery such that by 4 h post-exercise it was significantly different from the resting value (Table 4) but by 24 h it was fully recovered.

Plasma [lactate] increased nearly 20-fold at exhaustion and remained elevated through 8 h of recovery, but by 24 h was not significantly different from the resting value.

Intracellular [Mg²⁺]_f was estimated to be close to 10 mmol l⁻¹ at rest, increasing to as high as 15 mmol l⁻¹ at exhaustion. In general, the effects on the model calculations of changing [Mg²⁺]_f were small; in most cases less than 5%. Calculated [ADP]_f showed the greatest variation. Resting [ADP]_f ranged from 18.2±2.9 to 22.3±3.6 nmol g⁻¹ tissue, with [Mg²⁺]_f between 1 and 20 mmol l⁻¹.

[AMP]_f increased significantly at exhaustion although [ADP]_f did not. By 2 h of recovery both had declined by at least an order of magnitude (Table 5). [ADP]_f and [AMP]_f gradually increased over the remainder of the recovery period so that by 24 h post-exercise neither was significantly different from the resting values.

Energy charge (EC) and R_ATP estimated using free metabolite concentrations decreased slightly at exhaustion but had recovered by 2h post-exercise (Table 5). In contrast, EC calculated without taking into account that large fractions of the ADP and AMP pool are bound in vivo (EC_total) decreased significantly at exhaustion and remained depressed until 24 h post-exercise (Table 5).

Recovery of fish white muscle from exercise is very slow. The exact time when a fish is deemed ‘recovered’ is highly dependent on which metabolite is measured. One of the most interesting observations of this study is the rapid recovery of [PCr] when [ATP] is approximately half of the resting levels. Conclusions about energy status at this time would be highly dependent on which variable was favoured. Once a metabolite is chosen, the conclusion may be erroneous if the concentration term is inappropriate (i.e. using absolute instead of relative concentrations or total instead of free concentrations). When two common indices of adenylate status (R_ATP and EC) are calculated using free metabolite concentrations (Table 5), it is clear that adenylate status recovers much faster than the changes in [ATP] would suggest. Comparison of energy charge calculated with free vs total metabolites illustrates the importance of considering the free pools only. Arguments based on [ATP] alone are less meaningful. This is particularly true in trout white muscle, which has an active purine nucleotide cycle (Mommsen and Hochachka, 1988). Indeed, calculation of free adenylate concentrations offers insight into the regulation of the purine nucleotide cycle and its temporal separation into deaminating and reaminating arms during exercise and recovery.

During exercise, a large fraction (80%) of the ATP in white muscle is stoichiometrically converted to IMP (Fig. 1) via AMP deaminase. The purpose of this enzyme, the deaminating arm of the purine nucleotide cycle, is to maintain adenylate status (i.e. R_ATP) at high levels to support mechanical work. AMP deaminase is expected to be exquisitely sensitive to the changes in [AMP]_f observed in this study (Table 5). At rest [AMP]_f is less than 2 μmol l⁻¹, well below the apparent Michaelis constant (K_m,_app) of rabbit muscle AMP deaminase (400 μmol l⁻¹, Smiley and Suelter, 1967). The activity of AMP deaminase in rainbow trout white muscle is approximately 40 μmol min⁻¹ g⁻¹ (Fijisawa and Yoshino, 1987). If, as a first approximation, we assume that the rate is linearly related to [AMP]f up to the K_m,_app, then, at the [AMP]_f observed at exhaustion, there would be enough AMP deaminase activity to produce about 9 μmol IMP g⁻¹ muscle over the 30min protocol (assuming 6 μmoll⁻¹ [AMP]_f), close to the observed increase in [IMP] of about 5 μmolg⁻¹. At rest and during recovery, [AMP]_f is much lower and IMP production would be expected to decrease. At the [AMP]_f calculated for the 2 h post-exercise group (Table 5) at most 0.042 μmol g⁻ IMP could be formed over the same 30 min period, less than 0.5% of the rate expected during exercise. These changes in [AMP]_f may be one of the causes of the separation of the purine nucleotide cycle into two temporally distinct arms in fish white muscle (Mommsen and Hochachka, 1988). Furthermore, the import-ance of changes in [AMP] is obscured if the total [AMP], rather than [AMP]_f, is considered.

During recovery, adenylate status is an important signal to intermediary metabolism and, again, it is the free concentration of metabolites such as ADP that influences rates of glycolysis and mitochondrial respiration. In the present study, resting [ADP]_f represents less than 3 % of the measured total ADP (see Tables 3 and 5). The cost of recovery metabolism (phosphagen resynthesis, glycogen repletion, ion homeostasis) is largely met by mitochondrial metabolism. The availability of free ADP relative to ATP ([ATP]/[ADP]_f) is a critical determinant of mitochondrial respiration in skeletal muscle, but the exact nature of the relationship is complex (see Balaban, 1990). Although total [ADP] changes little in rest, exercise and recovery (Table 3), [ADP]_f undergoes pronounced changes (Table 5), which have an important influence on white muscle metab-olism. There is a strong dependence of mitochondrial oxygen uptake on ADP availability at these low concentrations (Moyes et al. 1992). At [ATP]/[ADP]_f ratios greater than 200, as is the case in trout at rest (Fig. 1 and Table 5), white muscle respiration is less than 10 % of the maximal capacity (Moyes et al. 1992). In fact, if [ATP]/[ADP]_f is an important determinant of respiration in vivo, it is perplexing that, relative to rest, it becomes even more inhibitory during recovery when metabolic demands are expected to be greater. In a parallel study, we suggest that the added cost of recovery, as indicated by the excess post-exercise oxygen consumption, is met by mitochondria, which are stimulated primarily by changes in [phosphate] and influenced by pHi (Moyes et al. 1992). Instead of a signal for mitochondrial respiration, the elevated post-exercise [ATP]/[ADP]_f apparent in this study may be involved in regulation of carbohydrate metabolism.

Two main questions dominate current interest in carbohydrate metabolism during recovery from high-intensity exercise in fish white muscle. What is the fate of lactate? What is the pathway of glycogen resynthesis? Glycogen could be formed from lactate in situ or via export to the liver, where hepatic gluconeogene-sis produces glucose, which serves as a precursor for white muscle glycogen (Cori cycle). There is abundant evidence to suggest that Cori cycling is not necessary, at least for some skeletal muscles. In salmonids, there is little evidence for Cori cycling as an important route of lactate metabolism and glycogen resynthesis. After injection of [¹⁴C]lactate into the blood of coho salmon following exhaustive exercise, more than 80% of the total blood radioactivity was recovered as lactate (Milligan and McDonald, 1988). Less than 10% of the glycogen resynthesized following exhaustive exercise in the trout can be accounted for by glucose uptake into white muscle, based on the uptake of 2-deoxyglucose (T. G. West, P. M. Schulte and P. W. Hochachka, unpublished results). Since the sum of [glycogen] + 2[lactate] is equal in rest and throughout recovery, it is unlikely that oxidation is an important fate of lactate in recovery in this study. It appears that, in fish, the majority of the lactate formed during exhaustive exercise is converted to glycogen within the white muscle. The route for this pathway has not been established but the results of this study are consistent with a pathway involving reversal of pyruvate kinase.

In liver gluconeogenesis, phosphoenolpyruvate carboxykinase (PEPCK) and pyruvate carboxylase are required to reverse the step catalyzed by pyruvate kinase in glycolysis. Pyruvate kinase is normally thought to be an irreversible enzyme. Skeletal muscle of vertebrates lacks pyruvate carboxylase and two schemes have been suggested to bypass this step in skeletal muscle: (1) PEPCK+malic enzyme and (2) reversal of pyruvate kinase. Connett (1979) has argued that the metabolite concentrations found in recovering frog skeletal muscle suggest that pyruvate kinase is too far from equilibrium to be reversed. He suggests the PEPCK-malic enzyme route is more feasible in frog. Fish white muscle, however, also lacks PEPCK (e.g. Moon and Johnston, 1980). Consequently, glyconeogenesis in trout white muscle may require the reversal of pyruvate kinase. Reversal requires high [ATP] and [pyruvate] and low [ADP] and [PEP] (Dyson et al. 1975). The [ATP]/[ADP_f] ratios observed in this study (500–2000) are higher than those observed in liver (20, Morikofer-Zwez and Walter, 1989) and those employed by Connett for frog skeletal muscle (30–80). They are similar to ratios reported in mammalian studies (e.g. Shoubridge and Radda, 1984). Unfortunately, PEP was not measured in our study so we cannot assess how close to equilibrium lie the reactants of pyruvate kinase. Newsholme and Crabtree (1986) suggest that if the ratio of the mass action ratio to the equilibrium constant is between 0.2 and 5, the reaction can be assumed to be near equilibrium and reversible in vivo. Using this as a guideline, PEP concentrations of 10–60 μmol l⁻¹ would bring this reaction near to equilibrium in recovery. [pH-independent K_eq = [pyruvate] × [ATP]/ ([PEP]x[ADP]_f × [H⁺])=2 × 10¹¹ (moll⁻¹)⁻¹, as used by Connett (1979).] These concentrations are physiologically realistic.

The high [ATP]/[ADP]_f ratios seen during recovery could be required to reverse flux through pyruvate kinase. The fact that high [ATP]/[ADP]_f ratios may also limit oxidative metabolism well below its capacity (see Moyes et al. 1992) may be of lesser significance. Thus, there may be a trade-off involved in which the rate of glycogen synthesis is optimized relative to the rate of oxidative phosphoryl-ation. ATP replenishment via the purine nucleotide cycle need not occur prior to these events.

Fish white muscle exhibits, albeit to an exaggerated degree, many of the typical characteristics of fast twitch glycolytic muscles from all animals. Energy metab-olism and carbohydrate metabolism in these types of muscle must be regarded as an integrated system responding to a number of controlling factors including [ADP] _f and pHi. Energy metabolism and purine nucleotide cycling, however, are less closely linked, resulting in a dissociation of [ATP] and energy status. By utilizing the calculated free concentrations of ATP, ADP and AMP, it is possible to understand more fully these relationships and the control of muscle metabolism during exercise and recovery.

We would like to thank Dr Peter Arthur for helpful discussions. This study was supported by a Natural Sciences and Engineering Research Council of Canada operating grant to P.W.H. and a postgraduate scholarship to P.M.S.