Characterization of Red Blood Cell Metabolism in Rainbow Trout

In recent years, fish red blood cells (RBCs) have emerged as an exciting experimental system for studying intracellular pH (pHi) regulation and oxygen transport during physiological stress such as strenuous exercise (e.g. Nikinmaa et al. 1984; Wood and Perry, 1985; Primmett etal. 1986; Milligan and Wood, 1987), hypercapnia (e.g. Perry et al. 1987; Vermette and Perry, 1988; Perry et al. 1989) and hypoxia (e.g. Fievet et al. 1987; Boutilier et al. 1988). In particular, elevated plasma catecholamine levels have been shown to play a central role in RBC pHi homeostasis in the face of extracellular acidosis, by stimulation of proton extrusion and cell swelling via Na⁺/H⁺ exchange. This exchange does not have a direct energy requirement, but is a secondarily active mechanism dependent upon the electrochemical gradient for Na⁺ entry established by Na⁺/K⁺-ATPase at metabolic expense (see reviews by Nikinmaa, 1986; Nikinmaa and Tufts, 1989; Motais and Garcia-Romeu, 1989). Catecholamines may thereby elevate RBC Na⁺ concentrations, stimulate Na⁺/K⁺ transport activity and oxygen consumption (Bourne and Cossins, 1982; Ferguson and Boutilier, 1988; Ferguson et al. 1989) and lower intracellular nucleoside triphosphate (NTP) levels (Nikinmaa, 1983, 1986; Milligan and Wood, 1987; Ferguson and Boutilier, 1989). The reduction in fixed negative charge associated with reduced NTP levels and cell swelling contributes additionally to pHi elevation via the Donnan effect, and the lower concentrations of H⁺ and NTP, both negative modifiers of O₂ binding to haemoglobin, favour O₂ loading in the face of extracellular acidosis.

Little is known about the metabolic substrates for these events, or even about the basic metabolic pathways of teleost RBCs. Early studies demonstrated that fish RBCs consume O₂ at a significant rate, in contrast to mammalian erythrocytes (Hunter and Hunter, 1957; Eddy, 1977), and very recent studies have concluded that aerobic metabolism accounts for more than 90 % of resting NTP production in salmonid RBCs (Ferguson and Boutilier, 1988; Ferguson et al. 1989). In vitro, RBC NTP levels are sensitive to the level of oxygenation, and RBCs incubated under anaerobic conditions produce lactate (Greaney and Powers, 1978; Tetens and Lykkeboe, 1981; Milligan and Wood, 1987; Ferguson and Boutilier, 1989; Ferguson et al. 1989). Measurement of RBC enzyme activities in the yellow perch (Perca flavescens) suggested that the tissue is largely glycolytic, but with a significant capacity for catabolism of glucose and NADPH production by the pentose-phosphate shunt (Bachand and Leray, 1975). One need for NADPH production is believed to be the continual requirement for reduced glutathions (GSH) production to protect against oxidation (e.g. Marshall et al. 1990).

With this background in mind, we undertook the present study utilizing rainbow trout Oncorhynchus mykiss to characterize RBC metabolism in terms of: (1) ability to oxidize different substrates; (2) substrate preferences, kinetic and competitive interactions and relationship to total oxidation rates; and (3) the importance of other pathways such as the pentose-phosphate shunt. This study laid the groundwork for a companion investigation (Wood et al. 1990) of the effects of strenuous exercise and post-exercise recovery on RBC metabolism, and a mechanistic analysis of the relative roles of acid-base variables and catecholamine and substrate levels in the observed changes in metabolism induced by exercise.

Rainbow trout (Oncorhynchus mykiss=Salmo gairdneri; 150–400 g) were obtained from Thistle Springs Trout Farm, Ashton, Ontario, and held at The University of Ottawa in April-May or from Spring Valley Trout Farm, Petersburg, Ontario, and held at McMaster University in June-August. At both locations, the fish were housed in large fibreglass tanks at 9–12°C in running dechlorinated tapwater; no apparent effects of source or holding conditions were apparent in our measurements. Fish were fed ad libitum on a commercial diet, but were fasted for 24–48 h prior to use. Dorsal aortic cannulae (PE50 polyethylene tubing; Soivio et al. 1972) were implanted under MS-222 anaesthesia (1:10000 w/v; Sigma), and the fish allowed to recover for 24-48h in darkened Perspex chambers served with aerated, flowing tapwater at the experimental temperature (10±l°C).

Trout were initially infused with 0.5 ml of heparinized (50 units ml⁻¹) Cortland saline (Wolf, 1963); blood was then withdrawn slowly until the fish showed the first signs of struggling or disorientation, at which point sampling ceased. This procedure usually yielded approximately 2ml of blood per fish, but occasionally large trout yielded up to 5 ml. In all experiments, except those for enzyme measurements, blood from several fish was pooled, heparinized and kept on ice for up to 30 min until use. This pooled blood was then placed in 20 ml glass vials in volumes of either 700 μl (for CO₂ production determinations) or μl (for O₂ consumption determinations) for gassing and incubations. The use of pooled blood allowed a paired or matrix experimental design; each pooled preparation was considered as one sample for statistical purposes.

Whole blood (approximately 2ml) was withdrawn from individual fish, centrifuged at 13 000g for lmin in pre-weighed microcentrifuge tubes, plasma was decanted, and both fractions immediately frozen at −80°C for 1 month prior to assay. Thawed plasma (50 μl) was used directly in enzyme assays. Packed RBCs were homogenized in a Brinkmann Polytron with 3vols of ice-cold 50 mmol l⁻¹ imidazole-HCl, pH 7.4 (at room temperature). This homogenate was centrifuged at 13 000g for 1 min, and the supernatant (10μl) was used directly in enzyme assays. Assays were buffered with 50 mmol l⁻¹ imidazole, pH 7.4 (unless noted below). Total volume was 1ml, and enzyme activities were monitored at 24.0±0.2°C by following the appearance/disappearance of NAD(P)H at 340 nm spectrophotometrically in an LKB 4050 Ultrospec II connected to a chart recorder (except for GNase) using procedures described by Mommsen et al. (1980) with slight modifications.

Malate dehydrogenase (E. C.1.1.1.37) (MDH). 0.15mmoll⁻¹ NADH, 0.5 mmol l⁻¹ oxaloacetate.

Citrate synthetase (E.C.4.1.3.7) (CS). Absorbance read at 412nm (E=13.6), buffer was 50mmoll⁻¹ sodium Hepes, pH8.0, O.lmmoll⁻¹ 5,5′-dithiobis-2-nitrobenzoic acid (DTNB), 0.3 mmol l⁻¹ acetylcoenzyme A, 0.5 mmol l⁻¹ oxaloacetate.

Isocitrate dehydrogenase (E.C. 1.1.1.42) (IDH). 0.4mmoll⁻¹ NADP⁺, 4 mmol l⁻¹ MgCl₂, 0.6 mmol l⁻¹ threo-D(+)-isocitrate.

Glucose-6-phosphate dehydrogenase (E.C.1.1.1.49) (G6PDH). 0.4mmoll⁻¹ NADP⁺, 7 mmol l⁻¹ MgCl₂, 1 mmol l⁻¹ glucose-6-phosphate.

6-Phosphogluconate dehydrogenase (E.C.1.1.1.44) (6PGDH). 0.8 mmol l⁻¹ NADP⁺, 20mmoll⁻¹ MgCl₂, 3.0mmoll⁻¹ 6-phosphogluconate.

Phosphofructokinase (E.C.2.7.1.11) (PFK). 0.12 mmol l⁻¹ NADH, 2 mmol l⁻¹ ATP, 50 mmol l⁻¹ KC1, 10 mmol l⁻¹ MgCl₂, li.u. alpha-glycerophosphate dehydrogenase, 5i.u. aldolase, 5i.u. triosephosphate isomerase, 5 mmol l⁻¹ fruc-tose-6-phosphate.

Pyruvate kinase (E.C.2.7.1.40) (PK). 0.12mmoll⁻¹ NADH, 2.5mmoll⁻¹ ADP, 10/zmoll⁻¹ fructose-l,6-bisphosphate, 30mmoll⁻¹ KC1, lOmmoll⁻¹ MgCl₂, 20i.u. LDH, 2.5 mmol l⁻¹ phosphoenolpyruvate.

Lactate dehydrogenase (E.C.l.1.1.27) (LDH). Forward direction, 0.12 mmol l⁻¹ NADH, 2 mmol l⁻¹ pyruvate. Reverse direction, 2.0 mmol l⁻¹ NAD⁺, 250 mmol l⁻¹ L-lactate.

Creatine phosphokinase (E.C.2.7.3.2) (CPK). 0.2mmoll⁻¹ NADP⁺, l.Ommoll⁻¹ ADP, lOmmoll⁻¹ AMP, 4 mmol l⁻¹ glucose, 5mmoll⁻¹ MgCl₂, 2i.u. G6PDH, 5i.u. hexokinase, 50 mmol l⁻¹ creatine phosphate.

Phosphoenolpyruvate carboxykinase (E.C.4.1.1.32) (PEPCK). 0.12 mmol l⁻¹ NADH, 0.5 mmol l⁻¹ phosphoenolpyruvate, 20 mmol l⁻¹ NaHCOj, 1 mmol l⁻¹ MnCl₂, 8i.u. MDH, 0.2 mmol l⁻¹ deoxyguanosine diphosphate.

Fructose-l,6-bisphosphatase (E.C.3.1.3.11) (FBPase). 0.2mmoll⁻¹ NADP⁺, 15 mmol l⁻¹ MgCl₂, lOi.u. phosphoglucose isomerase, 2i.u. G6PDH, 0.1 mmol l⁻¹ fructose-l,6-bisphosphate.

Glutamate dehydrogenase (E.C.1.4.1.3) (GDH). 0.12mmoll⁻¹ NADH, lmmoll⁻¹ ADP, 250mmoll⁻¹ ammonium chloride, O.lmmoll⁻¹ EDTA 14 mmol l⁻¹ alpha-ketoglutarate.

Glutaminase (GNase). 150 mmol l⁻¹ potassium phosphate, 50 mmol l⁻¹ Tris (pH 8.6), 0.2 mmol l⁻¹ EDTA, 20 mmol l⁻¹ glutamine, incubated for 30 min, terminated with perchloric acid. Glutamate was subsequently determined in neutralized extracts by the glutamate dehydrogenase assay.

After addition of extract and monitoring of control absorbance changes, reactions were initiated by adding a small volume (25–100 μl) of the last item listed (usually one of the substrates). In all cases control activity was less than 5 % and was subtracted from the activity with substrate.

Carbon dioxide production of pooled whole blood was measured from [U-¹⁴C]L-lactate (90-105mCimmol⁻¹; ICN), [U-¹⁴C]D-glucose (310mCimmol⁻¹; ICN), [U-¹⁴C]L-alanine (135mCimmol⁻¹; ICN), [l-¹⁴C]o-glucose (60mCimmol⁻¹; Amersham) and [l-¹⁴C]oleate (57 mCi mmol⁻¹; NEN) according to the methods of French et al. (1981) as adapted by Walsh et al. (1988). The protocol required a modified CO₂ trapping system. Briefly, 20ml glass vials were set up in a shaking water bath; each vial contained 700μl of blood plus 150 μ1 of saline with 125 i.u. of heparin, unlabelled substrate, and any other effectors. The vials were sealed with a rubber septum, fitted with inflow and outflow ports for gassing, and a suspended well containing a fluted glass filter paper (the primary trap). The outflow port was connected via a short length of PE50 tubing to the bottom of a similar vial filled with a mixture of 1 ml of Carbotrap II (Baker) and 1 ml of ethanol. This secondary trap served to collect any ¹⁴CO₂ liberated by gassing during the incubation.

To control acid-base status, the vials were continually gassed as open systems with a humidified 99.75% air/0.25% CO₂ mixture (representative of typical in vivo resting via a Wösthoff 301a-F gas-mixing pump. After an initial 15 min equilibration period, the run (usually 2 h) was started by the addition of labelled substrate (approximately 0.5 μCi per vial) to the blood by injection through the septum. At the end of the run, gassing was stopped, 100 μl of 1 mol l⁻¹ hyamine hydroxide was injected onto the filter in the centre well, and the blood was acidified with 0.1ml of 70% perchloric acid to stop metabolism and liberate ¹⁴CO₂. The sealed vials were then shaken vigorously at room temperature for a further 1.5 h to ensure complete collection of ¹⁴CO₂ in the primary trap. Combined radioactive ¹⁴CO₂ from both traps was used to calculate total CO₂ production rate from the single labelled substrate, based on the specific activity, length of incubation, wet mass of RBCs and a correction for ¹⁴CO₂ release from simultaneously run chemical control vials. Specific activity was measured by counting the stock label for ¹⁴C radioactivity and measuring the total substrate concentration in dummy vials. Chemical control vials contained heparin, labelled and unlabelled substrates and saline or plasma; no difference was observed between these two treatments. Dummy vials (no radioactivity) were set up in parallel for measurement of metabolites, catecholamines, haemoglobin and acid-base parameters at the beginning and end of the 2 h run. All vials were shaken gently during the incubation and temperature was maintained at 10±l°C. In one experiment, specific activity of [¹⁴C]lactate, produced by RBCs in vitro from [U-¹⁴C]glucose, was measured in the residual perchloric acid extract following CO₂ collection by the methods of Roca et al. (1985) with no further modification.

For total oxygen consumption measurements ⁠, pooled whole blood samples (1400μl) were additionally heparinized (125 i.u. in 50 μl of saline) and then gassed with 99.75% air/0.25% CO₂ for 15 min as outlined above. Samples were then transferred by gas-tight Hamilton syringe to an measurement chamber thermostatted to 10±l°C. This consisted of a 1.35 ml glass chamber, similar to that described by Tucker (1967), which was fitted with a Radiometer E5046 electrode and stirring flea. The output was displayed on a Brown Boveri flatbed chart recorder. was calculated from the slope of the record over time as fell from the starting value (approximately 20.7 kPa) to no lower than 18.0 kPa, a region over which trout haemoglobin is fully saturated (Eddy, 1971). The system was calibrated empirically by measuring simultaneously in 16 pairs of identical samples (selected from a wide range of experimental treatments) by the present method, and by the method described by Ferguson et al., (1989). In the latter, the decline in total O₂ content over a 2h period in a sealed syringe was measured by the method of Tucker (1967). There was a linear proportional relationship between the two methods (r=0.78, N=16, P < 0.001) with the method yielding values that were on average 34.5 % of the total O₂ method. The advantage of the present technique is that determinations are complete in about 15 min, whereas the O₂ content method takes 2h at 10°C to detect reliable changes, or necessitates a rise in temperature to speed up the determination (see Ferguson et al. 1989). A minor disadvantage is that the system becomes less reliable at lower haematocrits; samples with haematocrits less than 14% were not used.

Blood samples for acid-base and metabolite measurements were drawn from the incubation vials through the septum into Hamilton gas-tight syringes. Wholeblood pH (pHe) was measured using a Radiometer PHM-71 meter and G97-G2 micro-capillary electrode thermostatted to the experimental temperature. Samples were removed for NTP and haemoglobin (Hb) analysis; the remainder of the sample was then transferred to a pre-weighed micro-centrifuge tube. Plasma was separated and an RBC pellet obtained by centrifugation at 13 000g for 2 min. Plasma was decanted for determination of catecholamine, glucose, lactate and alanine levels; packed cell mass and haematocrit were determined gravimetrically. RBC pHi was measured by the freeze-thaw method of Zeidler and Kim (1977), using the same electrode. Total carbon dioxide content was measured on 100 μl of true plasma using a Corning 965 CO₂ analyzer. Plasma and [HCO₃⁻] were calculated using the Henderson-Hasselbalch equation and appropriate dissociation and solubility constants listed in Boutilier et al. (1984). Whole-blood NTP, haemoglobin (Hb) and plasma lactate concentrations were measured spectrophotometrically using commercial kits (Sigma), and standard enzymatic methods were used to measure plasma alanine (GraBl and Supp, 1983) and plasma glucose and RBC glycogen concentrations (Keppler and Decker, 1974). A separate plasma sample was mixed with 25 mmol l⁻¹ each of reduced glutathione and EGTA, stored at -80°C for no more than 3 days, and analyzed for catecholamines (epinephrine and norepinephrine) by HPLC (Woodward, 1982, as applied by Vermette and Perry, 1988). Biochemicals were purchased from Sigma or Boehringer-Mannheim. All other chemicals were reagent grade.

All values are presented as means±ls.E.M. (TV). Statistically significant differences (P < 0.05) were assessed using Student’s two-tailed t-test, paired or unpaired format as appropriate to the design of the experiments.

Significant activities of enzymes of glycolysis (pyruvate kinase, lactate dehydrogenase), the tricarboxylic acid cycle (malate dehydrogenase), the pentosephosphate shunt and NADPH metabolism (glucose-6-phosphate dehydrogenase, 6-phosphogluconate dehydrogenase, isocitrate dehydrogenase) and phosphagen metabolism (creatine phosphokinase) were detected in rainbow trout RBCs (Table 1), indicating a potential for activity of all these pathways. These values are substantially lower than those observed in salmonid liver (Mommsen, 1986), consistent with the lower metabolic rates (see below). We attempted to measure six other enzymes (glutamate dehydrogenase, citrate synthase, phosphoenolpyruvate carboxykinase, fructose 1,6-bisphosphatase, phosphofructokinase and glutaminase), but could detect no activity. However, with the exception of glutaminase (which is assayed non-spectrophotometrically), this lack of detection could be due to the relatively low homogenate volumes we were forced to use to avoid spectral interference by Hb; alternatively, some enzymes may be freeze-labile. Significant activities of malate dehydrogenase, lactate dehydrogenase, creatine phosphokinase and pyruvate kinase were also detected in blood plasma (Table 1). Since the Hb concentration ratio between plasma and cell homogenates was considerably lower than the comparable ratio of enzyme activities, it is unlikely that these plasma enzymes were derived from post-sampling lysis of RBCs. Enzyme activities have been reported previously in fish plasma, and they appear to increase with stress (Wells et al. 1986). It would be interesting to determine the isozyme types and tissue origins of the enzymes found in rainbow trout plasma before and after stress.

Our treatments of rainbow trout blood resulted in preparations with high viability, shown by stable NTP values and NTP/Hb ratios (Table 2) comparable to values from in vivo studies (e.g. Milligan and Wood, 1987). The flow-through gassing system maintained blood and acid-base parameters (Table 2) very close to measured resting in vivo values in this same batch of fish (Wood et al.1990). Equilibration was complete within 15 min; an important precaution was to ensure that the inflowing gas was fully humidified at the experimental temperature to prevent changes in the hydration state of the blood. Plasma catecholamine, glucose and lactate levels (Table 2) were slightly higher than prior studies from the same laboratory (Perry et al. 1989; Wright et al. 1989) but comparable to in vivo values measured on this same batch of trout (Wood et al. 1990). Plasma alanine levels were in the range of salmonid values (e.g. Mommsen et al. 1980). RBC intracellular glycogen levels were very low in comparison to values for rainbow trout liver (Perry et al. 1988; Wright et al. 1989), but detectable, whereas intracellular alanine, lactate and glucose were not detectable (Table 2). All measured variables appeared to be stable throughout the 120 min incubation period, except plasma lactate which increased significantly. Note, in particular, the stability of epinephrine and norepinephrine. In confirmation, we found that when much higher levels of epinephrine (350 nmol l⁻¹) were added to blood in vitro, the loss rate was only about 10% h⁻¹. This conflicts with the recent report of Tetens et al. (1988) that catecholamines are extremely labile in trout blood incubated in vitro; the reason for this difference is unknown.

Overall metabolic rates of RBCs (Table 2) were low in comparison to other highly active preparations such as trout hepatocytes (Mommsen et al. 1988) and fish gill cells (Perry and Walsh, 1989). O₂ consumption rates were comparable to previous measurements on salmonid erythrocytes (Eddy, 1977; Ferguson and Boutilier, 1988; Ferguson et al. 1989). Assuming typical values for whole-trout metabolic rate, haematocrit and blood volume (e.g. Milligan and Wood, 1982), the RBCs would account for 2–3% of resting whole-animal ⁠.

Rainbow trout RBCs were presented (separately) with several different labelled substrates, at typical resting plasma levels, to determine their ability to oxidize these fuels to CO₂. Under resting conditions, glucose was clearly the dominant of the four substrates, followed by lactate; both were oxidized at substantially higher rates than alanine and oleate (Table 2). Interestingly, the combined resting CO₂ production rates from glucose, lactate, alanine and oleate were substantially lower than measured O₂ consumption rates (Table 2). These data suggest that other substrates are also important in supporting RBC metabolism in vivo. The extremely low level of intracellular glycogen would appear to exclude it as a significant contributor. However, since there are many amino acids in fish plasma, and several types of fatty acids, the total quantitative contribution of these substrate types may be important to total metabolic rate. In mammalian red blood cells, adenosine may be metabolized to several purine catabolites and pentose sugars; the pentoses can then enter glycolysis (and ultimately the citric acid cycle) via the pentose-phosphate shunt (for a review, see Joshi and Palsson, 1989). The presence or absence of this pathway should be examined in fish red blood cells. It is likely that other metabolic substrates may also be important in fish red blood cells. However, one methodological caution is appropriate: measurement of ¹⁴CO₂ production has the potential to underestimate metabolic rates artificially, compared to direct measurement of oxygen consumption, if recycling of meta-bolically generated CO₂ into synthetic pathways (via HCO₃⁻) occurs. Carbon skeleton synthesis has not been evaluated in fish red blood cells, so we cannot assess the extent of recycling. One appropriate additional experiment to evaluate adenosine metabolism would be to test if added adenosine stimulates oxygen consumption rate.

This first metabolic screening of RBCs was performed on blood from resting fish, with relatively low substrate levels in the plasma (Table 2). These concentrations can change markedly in vivo during stresses such as hypoxia (Dunn and Hochachka, 1987; Wright et al. 1989; Boutilier et al. 1988), hypercapnia (Perry et al. 1988) and strenuous exercise (Wood and Perry, 1985; Milligan and Wood, 1987; Wood et al. 1990). Therefore, we wished to determine the kinetic and competitive relationships for glucose, lactate and alanine in order to clarify possible substrate preferences in vivo. As oxidation rates for oleate were so low, this substrate was not examined further.

Under all substrate conditions tested, the slowest oxidation rates were always obtained for alanine (Fig. 1). Over most of the range of comparable substrate levels (except extremely high values), lactate oxidation rates exceeded glucose oxidation rates (Fig. 1). However, it must be remembered that resting levels of lactate in plasma were much lower than resting levels of glucose, which explains why glucose oxidation was greater than lactate oxidation under resting conditions. There are stress conditions (e.g. exercise, hypoxia) where plasma lactate may equal or exceed plasma glucose concentration. From the kinetic relationships in Fig. 1, lactate oxidation would clearly exceed glucose oxidation under such circumstances. For this reason, we decided to compare lactate and glucose oxidation rates in direct competition experiments using the ranges seen after exhaustive exercise (Wood et al. 1990). Variation in glucose concentration from 0 to 10 mmol l⁻¹ did not affect the rate of lactate oxidation at either high or low lactate concentration. However, lactate had a small inhibitory effect on glucose oxidation at high concentrations of lactate and glucose (Table 3). Thus, there are circumstances when oxidative substrate preference, at least between glucose and lactate, may change.

To check whether alterations in lactate concentration affected total RBC ⁠, plasma lactate was varied from control levels to 20 mmol l⁻¹ in one experiment on a single pool of blood (not shown). There was no change in ⁠, reflecting the fact that lactate oxidation accounted for only a small fraction (<10 %) of over this entire range.

The dependence of lactate and alanine oxidation rates on substrate concentrations yielded unequivocal hyperbolic relationships, which were well described by standard Michaelis-Menten analysis (Fig. 1). In both cases, K_m values were well within the physiological range for the respective substrate concentrations in the blood, and the upper end of the physiological range produced near V_max velocities in vitro. However, glucose oxidation did not follow such a clear hyperbolic curve, but continued to increase even at supraphysiological levels (Fig. 1). The K_m (again in the physiological range) and V_max estimates are, therefore, at best approximate. This lack of saturation, plus our demonstration of the presence of significant activities of glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase in RBCs (Table 1), led us to suspect that CO₂ was being produced from glucose via the pentose-phosphate shunt as well as by the tricarboxylic acid cycle.

To examine this possibility, we measured glucose oxidation with both uniformly labelled and (l-C)-labelled glucose. The ratio of CO₂ production from (1-C)-labelled glucose (0.128±0.047 μmolg⁻¹h⁻¹) to CO₂ production from uniformly labelled glucose (0.390±0.159 μmolg⁻¹ h⁻¹) was 0.403±0.011 (N=6), indicating an appreciable pentose-phosphate shunt activity (Wood and Katz, 1958). Prior studies (Ferguson et al. 1989) and our initial data (Table 2) also indicated that a measurable quantity of lactate was produced in rainbow trout blood during in vitro incubation, despite highly oxygenated conditions. Therefore, in the same experiment, we also measured the production of total lactate and [^l4C]lactate from uniformly labelled glucose. Lactate production was 0.051 ±0.031μmol lactate g⁻¹h⁻¹ (N=6) or approximately 28% of total glucose metabolism to CO₂ and lactate. This result directly confirms initial observations that significant lactate production from glucose can occur in rainbow trout RBCs despite the presence of high O₂ concentrations and partial pressures. Cancerous cells also exhibit this ‘excess aerobic glycolysis’ (Lehninger, 1982), and interesting parallels may become evident upon further study of the regulatory aspects of glycolysis in fish red blood cells.

This research was supported by NSF and NIH (PJW), NSERC of Canada (CMW and SFP) and CNRS (ST). Claude Lemieux provided valuable intellectual stimulation. We thank Dr Tom Mommsen for commenting on an earlier draft of this paper.