Carbonic Anhydrase Injection Provides Evidence for the Role of Blood Acid–Base Status in Stimulating Ventilation after Exhaustive Exercise in Rainbow Trout

It is now well established that severe, exhaustive exercise causes an increase in arterial in fish and that this is a major contributor to blood acidosis early in the recovery period. The cause remains unclear and controversial, with various workers advocating adrenergic inhibition of HCO₃⁻ flux through the red blood cells (RBCs), the titration of plasma HCO₃⁻ to CO₂ by protons extruded from RBCs and/or muscle tissue, or disequilibria associated with gill transit time limitations as major causes (see Wood and Perry, 1985; Perry, 1986; Wood, 1991; Randall and Perry, 1992; Thomas and Perry, 1992; Nikinmaa and Vihersaari, 1993). The present paper addresses the possible functional significance of this response, rather than its origin, because, at first glance, a phenomenon that exacerbates extracellular acidosis at a time of intense excess pot-exercise O₂ consumption (EPOC, classically termed ‘O₂ debt’; Scarabello et al. 1992) appears to be maladaptive.

In particular, we tested the hypothesis first advanced by Wood and Perry (1985) that an important purpose of this post-exercise respiratory acidosis is to help to maintain hyperventilation during recovery from exhaustive exercise, thereby ensuring correction of the O₂ debt. The classical view has been that ventilation in fish is linked solely to an O₂ drive (Randall, 1982; Shelton et al. 1986). However, in the last few years there has been increasing evidence for an important secondary drive linked to pH and/or in both elasmobranchs (Heisler et al. 1988; Heisler, 1989; Graham et al. 1990; Wood et al. 1990) and teleosts (Perry and Wood, 1989; Kinkead and Perry, 1991). Such a pH/ -based drive could play an important role in stimulating ventilation during the post-exercise period, when arterial O₂ levels are relatively normal and input from proprioceptors sensing muscle activity is presumably minimal.

In order to test this hypothesis, our goal was to alleviate the post-exercise respiratory acidosis by injection of bovine carbonic anhydrase into the circulation prior to exercise. Blood plasma normally lacks this enzyme; added carbonic anhydrase should increase the HCO₃⁻ dehydration capacity of the blood, thereby increasing CO₂ excretion and reducing build-up and acidosis after exercise. Our original rationale was that addition of exogenous carbonic anhydrase to the extracellular compartment should compensate for inhibited access of plasma HCO₃⁻ to RBC carbonic anhydrase caused by exercise-induced catecholamine mobilization (Wood and Perry, 1991; Perry et al. 1991). Shortly after the initiation of this project, two reports appeared showing that exogenous carbonic anhydrase added to trout blood in vitro would also attenuate the decrease in extracellular pH caused by adrenergic activation of RBC Na⁺/H⁺ exchange (Motais et al. 1989; Nikinmaa et al. 1990). The mechanism appears to be that carbonic anhydrase catalysis of HCO₃⁻ dehydration relieves a strong acid disequilibrium in the plasma caused by the protons extruded via Na⁺/H⁺ exchange. Thus, added carbonic anhydrase might alleviate both classical respiratory acidosis (due to increase) and ‘disequilibrium acidosis’.

In order to validate the experimental approach, it was first necessary to determine the extent to which exogenous carbonic anhydrase persisted in trout blood in vivo, its effects on resting blood acid–base and O₂ status, and its influence on both aerobic and anaerobic exercise performance. The next step was to evaluate its effectiveness in alleviating elevation and the associated acidosis, and to determine whether arterial blood O₂ variables were altered. The final step was to determine the effects of carbonic anhydrase on post-exercise ventilation and O₂ consumption and to investigate possible causative mechanisms, such as alterations in blood acid–base status itself, plasma catecholamine mobilization (Randall and Taylor, 1991) and brain acid–base status (Wood et al. 1990). The results provide functional significance in terms of ventilatory control for the phenomenon of elevation and associated respiratory acidosis after exercise.

Experiments were performed on 102 adult rainbow trout [Oncorhynchus mykiss (Walbaum); 150–450 g] and 30 juvenile rainbow trout (1.5–4.5 g), obtained from Spring Valley Trout Farm, Petersburg, Ontario, and held at seasonal temperatures in flowing dechlorinated Hamilton tapwater (moderately hard water; composition as in Scarabello et al. 1992). The fish were acclimated to the experimental temperature (15±1°C) for at least 7 days prior to experiments. All of the adult fish (series 1, 2, 4, 5, 6 and 7) were fitted with an indwelling dorsal aortic catheter for blood sampling and treatment injection (Soivio et al. 1972) and allowed to recover for 24–48 h in darkened individual fish boxes prior to testing. A standard protocol to induce exhaustion was used in all exercise experiments with adult fish. Each trout was transferred, without air exposure, to a 500 l tank and then vigorously chased for 6 min, following which the fish was returned to its chamber. Juvenile fish (series 3) were exercised aerobically as described below using a 150 l acrylic swimming respirometer (Graham and Wood, 1981).

In those experiments involving carbonic anhydrase injection, a standard dose of 10 mg kg⁻¹ of bovine erythrocytic carbonic anhydrase (Sigma) dissolved in 1 ml kg⁻¹ of Cortland saline was selected, on the basis of preliminary experiments with a range of doses. This represents approximately 25000 Wilbur-Anderson units kg⁻¹ (1 Wilbur-Anderson unit causes the pH of 0.012 mol l⁻¹ veronal to drop from 8.3 to 6.3 in 1 min at 0°C). The carbonic anhydrase was washed in with a further 1 ml kg⁻¹ of Cortland saline. Saline-injected control fish received 2 ml kg⁻¹ of Cortland saline only.

This series was designed to assess the stability of the chosen dose of carbonic anhydrase in the blood in vivo over the intended time course of subsequent experiments. Carbonic anhydrase activity was quantified by measurement of the HCO₃⁻ dehydration rate of the blood using the [¹⁴C]bicarbonate technique of Wood and Perry (1991). This in vitro assay measures the rate at which plasma HCO₃⁻ is converted to evolved CO₂ when the blood is exposed to an external gas phase where the is essentially zero.

Fish (N=6) were injected with 10 mg kg⁻¹ carbonic anhydrase via the dorsal aortic catheter at a time corresponding to just after control 1 (C1) in series 2, 4, 5 and 6. One hour later, a time corresponding to control 2 (C2), a 1.0 ml blood sample was withdrawn for measurement of HCO₃⁻ dehydration rate. A second 1.0 ml sample was withdrawn at 3.5 h, a time corresponding to 2 h post-exercise in these series. A pre-injection reference sample (1.0 ml) corresponding to C1 was taken from a separate group of trout (N=8). This approach was used so as to avoid depleting the haematocrit of the injected fish to the point (approximately 15%) where the endogenous HCO₃⁻ dehydration rate due to the presence of RBCs becomes dependent upon haematocrit (Wood and Perry, 1991; Perry and Gilmour, 1993). Blood samples were heparinized (175 i.u. ml⁻¹), tonometered for 15 min with a humidified precision gas mixture (⁠⁠, balance N₂) at 15±1°C, and then processed through the [¹⁴C]bicarbonate assay. The assay was performed exactly as described by Wood and Perry (1991), except that the radioisotope (2 μCi=74 kBq) was added in a different injection volume (50 μl, rather than 20 μl) of 5 mmol l⁻¹ HCO₃⁻ Cortland saline. Scintillation counting with quench correction on an LKB Rackbeta 1217 counter was employed to measure 14C radioactivity in plasma (50 μl in 10 ml of Amersham ACS II) and filter papers (in 10 ml of customized cocktail containing 2.0 g of PPO + 0.1 g of POPOP in 0.8 l of toluene + 0.2 l of 95% ethanol).

This series was designed to assess the influence of carbonic anhydrase injection on arterial blood acid–base, CO₂ and O₂ status in resting fish (N=8) over the intended time course for sampling in the exercise experiments. Fish were not exercised in this experiment, and carbonic anhydrase (10 mg kg⁻¹) was injected immediately after the C1 sample. Blood samples (600 μl) were withdrawn at C1 (1.5 h prior to intended exercise), at C2 (0.5 h prior to intended exercise) and 0, 0.25, 0.5, 1 and 2 h after the period of intended exercise. Samples were analysed for pHa, plasma ⁠, whole-blood ⁠, haemoglobin, haematocrit and plasma lactate. Approximately 300 μl of blood from each sample was returned to the fish, together with sufficient non-heparinized Cortland saline to compensate for the volume deficit.

The exhaustive exercise protocol used in series 4, 5, 6 and 7 induces both aerobic and anaerobic exercise. To assess whether the aerobic exercise performance was affected by carbonic anhydrase injection, critical swimming speeds (U_crit) were determined in saline and carbonic-anhydrase-injected trout using the methods of Brett (1964). Juvenile, rather than adult, trout were used for this experiment as the maximum speed of the swimming respirometer (Graham and Wood, 1981) was not sufficient to exhaust the adult fish. Fish were freeze-branded (Mighell, 1969) approximately 2 weeks prior to testing so that each fish could be identified visually when tested as part of a group in the respirometer. Approximately 1 h prior to testing, the fish were weighed and injected via the caudal vein with carbonic anhydrase (10 mg kg⁻¹) or saline alone, using the same total volumes of saline per unit mass as for adult fish. A modified 50μl Hamilton syringe with a shortened needle was employed. Data were taken only from fish in which the injection was made cleanly into the vessel with a minimum of struggling and which exhibited no post-injection blood loss or necrosis.

After approximately 1 h in the respirometer, the flow was started at 6 cm s⁻¹ (approximately 1 body length s⁻¹; BL s⁻¹) and increased in steps of 6 cm s⁻¹ at 1 h intervals thereafter. Respiratory frequency (fR) was recorded visually during each period. The exact time of exhaustion was recorded; fish were considered to be exhausted once they became impinged on the rear screen and were unresponsive to gentle prodding with a blunt rod. Exhausted fish were removed through a siphon tube (30 mm i.d.) while the test with the other fish remaining in the group continued until the last fish had stopped swimming. Fork length was measured (typically 6–7 cm) and critical swimming speed (U_crit, in BL s⁻¹) for each fish was calculated according to Brett (1964). Three runs involving five fish from each treatment (10 fish per run) were performed and the results were combined (N=15 per treatment).

This series with adult fish was designed to assess whether anaerobic exercise performance was affected by injection of carbonic anhydrase, by measuring the accumulation of lactate in white muscle. A second goal was to provide a preliminary indication whether carbonic anhydrase injection was successful in alleviating the normal post-exercise increase in ⁠. Blood samples (600 μl, with replacements of 300 μl blood plus 300 μl saline) were drawn at C1, followed immediately by injection of carbonic anhydrase (N=6) or saline (N=6). A second pre-exercise sample was taken at C2. Exhaustive exercise was commenced 24 min later. At the end of 6 min of exercise (0 h post-exercise), a third blood sample was taken. The fish was then immediately killed by a blow on the head; a sample of epaxial white muscle was immediately excised from below the dorsal fin and freeze-clamped in liquid N₂. Blood samples were analysed for pHa, plasma ⁠, whole-blood ⁠, haemoglobin, haematocrit and plasma lactate. Muscle samples were analyzed for lactate and water content.

This series concentrated on the blood acid–base, CO₂ and O₂ changes in the arterial blood of trout during the first 2 h of post-exercise recovery in the presence or absence of carbonic anhydrase. The blood sampling (600 μl, with replacements of 300 μl of blood plus 300 μl of saline) and measurement protocols (pHa, plasma ⁠, whole-blood ⁠, haemoglobin, haematocrit, plasma lactate) were identical to those of series 2. Fish were injected with either carbonic anhydrase (10 mg kg⁻¹; N=11) or saline (N=8) immmediately after the C1 sample. Exhaustive exercise (6 min) was commenced 24 min after the C2 sample. During recovery, blood samples were drawn at 0, 0.25, 0.5, 1 and 2 h post-exercise.

The primary objective of this series was to determine whether carbonic anhydrase injection altered the magnitude or pattern of post-exercise hyperventilation and O₂ consumption. A secondary goal was to determine whether plasma catecholamine mobilization was altered. For this experiment, it was not possible to use trout fitted with ventilation masks for direct measurement of ventilatory water flow rate in a Van Dam chamber (e.g. Playle et al. 1990) because the mask interfered with normal swimming. We therefore we adopted an indirect approach based on the Fick principle (Dejours, 1973).

Trout of a uniform mass (about 350 g) were used for these experiments. In addition to the dorsal aortic catheter (PE 50, with a PE 160 grommet), each fish was fitted with a buccal catheter (PE 160, with a PE 240 grommet) for measurement of inspired O₂ tension and one opercular catheter on each side (PE 60, with a PE 200 grommet) for measurement of expired O₂ tension (Fig. 1). The catheters were positioned as described by Holeton and Randall (1967; see their Fig. 1), but a 90° bend was heat-moulded into the PE60 immediately outside the grommet to allow the opercular catheters to trail backwards during exercise or forward during respirometry (Fig. 1). After surgery, the fish was allowed to recover for 48 h in a custom-built respirometry chamber fitted with sampling ports for measurement of O₂ tension in inflowing and outflowing water ⁠. The respirometer was designed for minimum deadspace by tapering to fit the fish’s body; total volume was 870 ml (approximately 520 ml with the fish in place). Water flow rate was 1 ml g⁻¹ min⁻¹ for resting measurements and 2 ml g⁻¹ min⁻¹ for post-exercise measurements. Water samples for O₂ measurements from the five sampling sites (⁠⁠, right ⁠) were siphoned continually (at a head of -1 cm) into glass sampling syringes.

The primary goal of this final series was to test whether the ventilatory differences attributable to carbonic anhydrase injection in series 6 could be correlated with differences in the intracellular pH (pHi) of the brain. An additional objective was to determine whether post-exercise intracellular acidoses in red and white muscle were affected by carbonic anhydrase. Single, terminal blood and tissue samples for pHi measurements were taken from three treatment groups: saline-injected at rest (N=8); saline-injected, sampled 0.25 h after exhaustive exercise (N=7); and carbonic-anhydrase-injected (10 mg kg⁻¹), sampled 0.25 h after exhaustive exercise (N=7). In each case, the injection was administered at a time corresponding to C1.

Intracellular pHi was measured by the DMO (5,5-dimethyl-2,4-oxazolidinedione) technique; the method has been described in detail and critically evaluated previously (Milligan and Wood, 1985, 1986a,b; Wright et al. 1988). In brief, 12–13 h prior to the planned time of killing, trout were injected, via the dorsal aortic catheter, with a 1 ml kg⁻¹ dose of 7 μCi ml⁻¹ (=259 kBq ml⁻¹) [¹⁴C]DMO (New England Nuclear; specific activity 1850 mBq mmol⁻¹) plus 28 μCi ml⁻¹ (=1036 kBq ml⁻¹) [³H]mannitol (New England Nuclear; specific activity 101.4 GBq mmol⁻¹) in 140 mmol l⁻¹ NaCl, flushed in with an equal volume of Cortland saline. Mannitol was used as the extracellular space marker as it penetrates the blood–brain barrier better than do other labels (Munger et al. 1991). At sampling, a 2 ml blood sample was withdrawn from the dorsal aortic catheter, and then the fish was immediately killed by a blow on the head. The whole brain, a sample of epaxial white muscle just below the dorsal fin and a sample of adductor mandibularis red muscle from the opercular region were excised and processed for pHi measurement. Arterial pH, plasma water content and plasma [³H] and [¹⁴C] radioactivities were measured on the blood samples. Water content and [³H] and [¹⁴C] radioactivities were measured on digested tissue samples. Digestion, scintillation counting and water content measurements on plasma and tissue samples for pHi determination were carried out exactly as described by Milligan and Wood (1986b) and Wright et al. (1988). Intracellular pH in each tissue sample was calculated using the relevant equations from Wright et al. (1988), employing pHa as representative of extracellular pH (pHe).

Accurate measurements are difficult at the low tensions and temperatures present in fish blood. To obtain maximum accuracy, the recommendations of Boutilier et al. (1978) were followed to avoid ‘memory’ effects in the electrode. The electrode was pre-equilibrated to a close to the expected value and the sample was replaced at 2 and 4 min, with a final equilibration time of 6 min. For comparative purposes, was also calculated from the Henderson–Hasselbalch equation using appropriate values for αCO₂ and pK′ for rainbow trout plasma from Boutilier et al. (1984) and for mammalian plasma extrapolated to fish temperatures from Severinghaus et al. (1956).

All data are expressed as means ±1 S.E.M. (N). Statistically significant differences (P⩽0.05) were assessed using paired or unpaired Student’s two-tailed t-tests, as appropriate to the design of the experiments. Whenever multiple comparisons were involved, the t-value was adjusted by the Bonferroni procedure.

Injection of 10 mg kg⁻¹ of bovine carbonic anhydrase into resting trout elevated the HCO₃⁻ dehydration rate of the blood by about 40% (Table 1). There was essentially no attenuation of this increase over the 3.5 h time course of the intended experiments, indicating that carbonic anhydrase is quite stable when infused into the circulatory system.

The effects of carbonic anhydrase injection on arterial acid–base and O₂ status were evaluated in resting fish over the time course of the intended experiments. Some small but significant changes were seen. These included an increase in pHa, a decrease in plasma [HCO₃⁻] and a decrease in ⁠, effects that persisted more or less throughout the whole experimental period (Table 2). increased at the first sample after injection (C2), but decreased gradually thereafter. These effects reflected initial increases in haematocrit and [haemoglobin] and subsequent declines due to blood sampling (not shown); ⁠, [O₂]/[Hb], MCHC and plasma [lactate] did not change (Table 2).

Saline injection controls were not performed in this experiment; however, it was evident that the changes associated with carbonic anhydrase injection were all exhibited by the time of the C2 sample (Table 2). Therefore, the resting data of series 4, 5 and 6 (i.e. the C1 and C2 samples), where saline controls were performed, were employed to evaluate whether the observed changes in resting fish were specific to carbonic anhydrase. The results (Table 3) indicated that the effects could only be partially attributed to the actions of carbonic anhydrase. Saline injection alone caused significant decreases in and [HCO₃⁻], though these effects were only half the size of the changes induced by carbonic anhydrase. These small changes might have resulted from dilution or a small increase in ventilation induced by disturbance. increased slightly but not significantly with both treatments, probably reflecting the same phenomenon as the significant increase in in series 2 (Table 2), and indicating that the effect was not specific to carbonic anhydrase. However, a small increase in pHa was seen only with carbonic anhydrase. We conclude that, in resting fish, carbonic anhydrase injection induces only very minor decreases in and [HCO₃⁻] and increases in pHa, changes that are close to the limit of reliable detection.

Carbonic anhydrase injection had no effect on aerobic swimming performance, as assessed by the U_crit of juvenile trout (Table 4). There were also no significant effects on respiratory frequency (fR) at any stage in the aerobic swim test; the maximum values of fR, and the frequencies in the period during which exhaustion finally occurred, are illustrated as representative values in Table 4.

Carbonic anhydrase injection also had no effect on anaerobic swimming performance, as assessed by the intracellular concentrations of lactate in white muscle (Table 5) immediately after exhaustive exercise (time 0 h). This experiment also provided a preliminary indication that carbonic anhydrase injection reduced the extent of post-exercise build-up and pHa depression, both of which were significantly attenuated relative to those of saline-injected controls (Table 5). Plasma lactate and concentrations were not affected.

This series extended the results of series 4, demonstrating that carbonic anhydrase injection reduced the post-exercise increase in by about 50% relative to saline-injected controls. This difference persisted through 1 h of recovery (Fig. 2A). In concert, the post-exercise fall in pHa was also attenuated over the same time course (Fig. 2B). Decreases in plasma (not shown) and [HCO₃⁻] (Fig. 2C) were slightly greater in the carbonic anhydrase treament, but the differences were significant only at one time point (0.25 h; Fig. 2C). These data suggest that carbonic anhydrase attenuated mainly the respiratory component (i.e. elevated ⁠) of post-exercise acidosis, with minimal effect on the metabolic component (i.e. from fixed acidic equivalents). This was confirmed by calculation of the plasma which exhibited no significant differences between the two treatments at any time after exercise (Table 6). Plasma [lactate] changes were also similar in the two groups early in recovery. Curiously, plasma [lactate] elevations were lower in carbonic-anhydrase-treated fish from 0.5 h onwards, a difference which reached 40% by 2 h.

Despite these marked effects on blood-acid base variables, carbonic anhydrase treatment had virtually no effect on blood O₂ variables. Post-exercise changes in (Fig. 3A), (not shown) and [O₂]/[Hb] (Fig. 3B) were very similar, apart from a significantly lower in the carbonic anhydrase group at 2 h (Fig. 3A). Haematocrit (not shown), [haemoglobin] (not shown) and MCHC (Fig. 3C) also underwent identical fluctuations in the two treatments, with a marked haemoconcentration (elevated haematocrit and [haemoglobin]) and red cell swelling (lowered MCHC) over the first 0.5 h after exercise. Indeed, in none of the experimental series was there any significant difference in [haemoglobin], haematocrit or MCHC between carbonic-anhydrase-injected fish and saline-injected controls.

The primary focus of this series was on ventilatory effects (Fig. 4). However, blood acid–base responses were recorded for reference, and responses to carbonic anhydrase were very similar to those of series 5. These included attenuation of both the post-exercise acidosis (Fig. 5A) and the post-exercise elevation in (Fig. 6A), effects that were actually slightly greater and longer-lasting than in series 5 (Fig. 2). The carbonic-anhydrase-injected fish were significantly different from the saline-injected controls through 2 h post-exercise, at which time blood acid–base status had been restored close to resting levels.

Carbonic anhydrase injection had no effect on ventilatory variables at rest, but greatly attenuated the normal post-exercise hyperventilation (Fig. 4). In saline-injected control fish, increased to about six times resting levels at 0.25 h post-exercise and remained elevated at 4.75 times resting levels through 2 h post-exercise (Fig. 4A). In contrast, only increased about threefold at 0.25 h post-exercise and had returned to resting levels by 2 h in trout treated with carbonic anhydrase. This difference was due to a large attenuation of the post-exercise increase in VS,R (Fig. 4B) and a smaller reduction in the post-exercise increase in fR (Fig. 4C). Differences between the two groups were significant throughout the post-exercise period.

Carbonic anhydrase injection had no effect on resting ⁠, but altered the post-exercise increase in (Fig. 5B). Despite the large attenuation of the hyperventilatory response, was only slightly (and not significantly) lower in the carbonic anhydrase treatment at 0.25 h and 0.5 h post-exercise than in saline-injected control fish (Fig. 5B). However, at 1 h and 2 h, the difference became significant. Indeed, by 2h, was no longer elevated above resting levels, in marked contrast to the controls. In accordance with these differences in and ⁠, carbonic anhydrase greatly attenuated the post-exercise decrease in an effect that was significant throughout the recovery period (Fig. 5C). (not shown) followed a pattern similar to that seen in series 5 (Fig. 3A), with virtually identical values in the two groups until 0.25 h post-exercise, followed by a small difference thereafter. was significantly lower in the carbonic anhydrase group than in the saline group at 0.5 h (75.6±3.6 versus 95.7±6.4 mmHg) and 1 h (86.3±3.0 versus 104.4±5.4 mmHg), but not at 2 h.

The mobilization of plasma catecholamines was markedly affected by carbonic anhydrase injection (Fig. 6B,C). Peak levels of adrenaline (approximately 120 nmol l⁻¹) and noradrenaline (approximately 80 nmol l⁻¹) at 0.25 h post-exercise in control fish were reduced to approximately one-third of these levels by the carbonic anhydrase treatment. Although catecholamine levels declined thereafter in both groups, a significant difference for adrenaline persisted through 1 h of recovery.

Intracellular pH (pHi) was measured in saline-injected fish at rest, and in both saline-injected and carbonic-anhydrase-injected animals at 0.25 h after exercise (Fig. 7). At rest, extracellular pH (i.e. pHa) was about 7.8, and intracellular pH was higher in brain (approximately 7.6) than in red or white muscle (both approximately 7.25). After exercise, the pH values of all compartments fell significantly. As before, the extracellular acidosis (i.e. fall in pHa) after exercise was reduced by carbonic anhydrase. Brain pHi fell by about 0.2 units after exercise, but this response was unaffected by carbonic anhydrase. In red and white muscle, pHi fell by about 0.25 and 0.45 units, respectively, but again this response was not significantly altered by the carbonic anhydrase treatment.

Carbonic anhydrase injection was employed to test whether elevation and the associated respiratory acidosis were involved in driving post-exercise hyperventilation. Our goal was to reduce the post-exercise increase in and the resulting fall in pHa without directly interfering with either exercise performance or other responses to exercise. Overall, the approach was highly successful. The chosen dose (10 mg kg⁻¹) was selected to compensate for an observed 35% relative inhibition of the HCO₃⁻ dehydration rate (i.e. the rate of conversion of plasma HCO₃⁻ to CO₂) observed in blood samples taken from trout after exhaustive exercise (Wood, 1994). The exogenous carbonic anhydrase actually raised the blood HCO₃⁻ dehydration rate by about 40% and persisted without reduction in the bloodstream in vivo throughout the 3.5 h experimental period (Table 1). The effects on resting blood acid–base and O₂ status were very small (Tables 2, 3). Exercise performance, assessed by both aerobic (U_crit; Table 4) and anaerobic (white muscle [lactate]; Table 5) criteria, was not impaired. Post-exercise changes in arterial O₂ variables (Fig. 3A,B), MCHC (Fig. 3C; indicative of RBC swelling), haematocrit and [haemoglobin] were largely unaffected, indicating that alterations in the primary O₂-based drive to ventilation did not occur to complicate the results. Most importantly, the treatment achieved about a 50% reduction in the extent of post-exercise elevation (Figs 2A, 6A; Table 5) and pHa depression (Figs 2B, 5A, 7; Table 5). However, there was one significant, and unexpected, complicating feature associated with the carbonic anhydrase treatment: the mobilization of plasma catecholamines associated with exhaustive exercise was reduced by about 65% (Fig. 6B,C). The potential implications of this effect are discussed below.

Throughout this study, was measured, rather than calculated, to counter the argument that the elevation in post-exercise is an artefact of using the Henderson–Hasselbalch equation at a time of ‘disequilibrium acidosis’. In every series, increased significantly after exercise, whether the values were measured directly or calculated indirectly using the Henderson–Hasselbalch equation from measurements of plasma and pHa. Furthermore, this was true whether or not the fish were treated with carbonic anhydrase (which should relieve ‘disequilibrium acidosis’; Motais et al. 1989; Nikinmaa et al. 1990), and irrespective of whether the pK ′ values of Severinghaus et al. (1956) or Boutilier et al. (1984) were used in the calculations. Fig. 8 shows a representative data set, from series 5, and illustrates several interesting features. First, the post-exercise elevation of was in fact larger and longer-lasting when measured directly than when calculated. Second, this was true whether or not the fish had been treated with carbonic anhydrase. Third, even when calculated in the presence of exogenous carbonic anhydrase, the post-exercise increase in was still significant. Finally, under all conditions, estimates calculated using the pK′values of Severinghaus et al. (1956) were closer to measured values than were estimates calculated using the pK′values of Boutilier et al. (1984). The latter was surprising, as the Severinghaus pK′ values are for mammalian plasma extrapolated to fish temperatures, while the Boutilier et al. (1984) pK′ values are based on rainbow trout plasma assayed at the correct temperature. Nevertheless, we conclude that the post-exercise increase in is a real phenomenon and that the use of calculated values will give a conservative estimate of the response.

The entry of plasma HCO₃⁻ into the RBCs through the band 3 HCO₃⁻/Cl⁻ exchange protein for subsequent dehydration by erythrocytic carbonic anhydrase is generally considered to be the rate-limiting step in CO₂ excretion (Perry, 1986; Perry and Laurent, 1990). Recently, Perry and Gilmour (1993) have confirmed this in vitro using the radioisotopic assay of Wood and Perry (1991). In particular, they showed that lysis of trout RBCs, which makes erythrocytic carbonic anhydrase directly available to plasma HCO₃⁻, will increase the rate of CO₂ excretion over a wide range of blood haematocrits and plasma HCO₃⁻ levels. However, in the present study, the addition of exogenous carbonic anhydrase to the blood plasma of resting trout in vivo had only very minor effects on pHa and plasma HCO₃⁻ (Tables 2, 3). This result suggests that access of plasma HCO₃⁻ to erythrocytic carbonic anhydrase via Cl⁻/HCO₃⁻ exchange is not an important limiting factor at rest in vivo. If there is an ‘apparent diffusive limitation’ on branchial CO₂ transfer as suggested by some models (e.g. Malte and Weber, 1985; Piiper, 1989), it most probably occurs at some other point in the system.

In contrast, after exhaustive exercise, the present results indicate that access of plasma HCO₃⁻ to erythrocytic carbonic anhydrase becomes an important factor in the apparent diffusive limitation on CO₂ excretion. The attenuation of elevation by approximately 50% by exogenous carbonic anhydrase occurs at a time of elevated CO₂ production by the tissues (Steffensen et al. 1987; Milligan and McDonald, 1988; Scarabello et al. 1992), elevated cardiac output (Neumann et al. 1983) and catecholamine mobilization (Fig. 6). In vitro, catecholamines cause a β₁-adrenergic inhibition of the HCO₃⁻ dehydration rate of trout RBCs (i.e. the rate of conversion of plasma HCO₃⁻ to CO₂) as measured by the [¹⁴C]bicarbonate assay (Wood and Perry, 1991; Perry et al. 1991; Wood and Simmons, 1994). Blood sampled from exhaustively exercised trout also exhibits a reduction in HCO₃⁻ dehydration rate, compared with that of resting fish, when assayed in vitro under the same conditions (Wood, 1994). This contrasts with the observations of Tufts et al. (1988), who used a very different assay procedure; possible reasons for this difference have been discussed by Wood and Perry (1991) and Perry et al. (1991). Therefore, we speculate that, in vivo, catecholamine mobilization during and after exhaustive exercise similarly inhibits the rate of HCO₃⁻ dehydration through the RBCs and that this limitation, at a time of increased CO₂ loading to the system from metabolism and reduced gill transit time, results in an increase in ⁠.

It might be questioned why carbonic anhydrase injection only attenuated and did not abolish the increase in ⁠. The probable explanation is that insufficient carbonic anhydrase was used to block the phenomenon completely. Assuming a distribution volume about equal to the blood volume (approximately 50 ml kg⁻¹; Milligan and Wood, 1982), the concentration of carbonic anhydrase achieved in vivo was about 0.2 mg ml⁻¹, and the resulting activity measured (Table 1) agreed closely with an in vitro dose–response curve presented by Perry and Gilmour (1993). According to these authors, the level used here represents only about 30% of a saturating concentration, or about a 40% increase above the resting in vivo level for whole blood, in agreement with the present data (Table 1). At this time of increased CO₂ load to the system, it may be necessary to add more carbonic anhydrase than is needed to compensate for the 35% inhibition relative to resting levels (Wood, 1994).

Attenuation of the post-exercise depression of pHa by carbonic anhydrase could result not only from the reduced build-up (i.e. reduction of classical respiratory acidosis), but also from alleviation of ‘disequilibrium acidosis’ (see Introduction; Motais et al. 1989; Nikinmaa et al. 1990). For example, Motais et al. (1989) found that injection of a saturating dose of carbonic anhydrase (50-fold higher than that used here) into either a tonometer of trout blood or the extracorporeal blood loop of an intact rainbow trout greatly attenuated the decrease in blood pH caused by adrenergic activation of RBC Na⁺/H⁺ exchange. The calculation (equation 5) is a useful diagnostic tool because it uses the assumption of equilibrium conditions to subtract the respiratory component from the total acidosis. The remaining was the same in the saline and carbonic anhydrase treatments (Tables 5, 6). This suggests that, in the present experiments, the influence of carbonic anhydrase was explained mainly by its action in reducing respiratory acidosis.

It was necessary to use an indirect method based on the Fick principle (Dejours, 1973) in order to measure changes in in trout which would freely exercise to exhaustion. This approach has been used in numerous studies and has been critically evaluated in a few. The principal limitation is the requirement for six different measurements (Fig. 1), the most problematic of which are the values. The approach works best in those species where the anatomy ensures that a well-mixed expired sample can be collected (e.g. skate; Graham et al. 1990). In trout, which have large, non-discrete opercular openings, the method has been criticized because of variability in over time and with respect to exact opercular catheter location (Davis and Watters, 1970). In our experience, the former is not a serious problem and can be minimized by continually siphoning through the catheters at a low rate, rather than aspirating samples from time to time. We sought to minimize the latter source of variability by standardizing the catheter location and by measuring separately for each opercular chamber. The resulting determinations for trout at rest (approximately 700 ml kg⁻¹ min⁻¹; Fig. 4A) were about twice those routinely measured in our laboratory at this temperature by use of ventilation masks on fish in Van Dam chambers (e.g. Playle et al. 1990). While it is possible that absolute values of and VS,R were somewhat overestimated by the Fick approach used there, there is no reason to believe that the relative changes seen after exercise were in error.

The most important finding of the present study was the marked attenuation of post-exercise hyperventilation and EPOC caused by the administration of carbonic anhydrase prior to exercise (Figs 4, 5). This finding provides adaptive significance for the phenomenon of ‘CO₂ retention’ and strong support for the original hypothesis of Wood and Perry (1985) that ‘the rise in and/or the associated fall in plasma pH may help maintain hyperventilation during the post-exercise period, thereby ensuring correction of the O₂ debt’. The present experiments were not designed to separate the relative importance of elevations in and decreases in pHa in driving hyperventilation and, in any event, the two are not independent variables.

Nevertheless, it was of interest to examine the relationships between ventilation and changes in and between ventilation and changes in pHa. For this analysis, the for each fish of series 6 at each post-exercise time was expressed as a simple ratio to its at C2, and absolute changes in and pHa were calculated relative to their values at C2. Data were grouped over intervals of 0.5 mmHg and 0.05 pH units respectively. In each case, the relationship between relative and or ΔpHa could be expressed as a simple exponential and was highly significant (Fig. 9). However, the correlation with ΔpHa (r=-0.94; Fig. 9A) was stronger than that with (r=0.73; Fig. 9B). In this regard, it is interesting that two separate studies on elasmobranchs (dogfish, Heisler et al. 1988; Heisler, 1989; skate, Graham et al. 1990; Wood et al. 1990) found correlations between changes in arterial acid–base status and relative under conditions where the primary O₂ drive to ventilation was more or less constant. In both, the correlation with ΔpHa was stronger than that with ⁠, as in the present investigation. Indeed, in the study of Heisler (1989), the slope of the exponential relating ΔpHa and relative (−3.35) was very similar to that of the present study −4.07; Fig. 9A).

Heisler (1989) suggested that a secondary respiratory drive might be provided by ‘pH sensitive structures in, or closely related to, the arterial bloodstream’. The acid–base status of the brain is important in this regard in higher vertebrates (Shelton et al. 1986), and Wood et al. (1990) found that increases in ventilatory flow rates in the skate correlated with depressions in the pHi of the brain, as well as with depressions in pHa. However, in the present study, there was no difference in the mean pHi of the brain between carbonic-anhydrase- and saline-infused fish (Fig. 7) at a time when pHa was much lower and much higher (Fig. 4A) in the former. Therefore, it appears more likely that the detector system in the trout responds directly to the pHa of the arterial blood.

Our original idea was that carbonic anhydrase acted to alleviate the consequences of catecholamine mobilization – i.e. the elevation in and plasma acidosis associated with β₁-adrenergic activation of RBC Na⁺/H⁺ exchange (Perry et al. 1991). Therefore, the reduction in would be related directly to the alleviation of these effects. However, this interpretation is complicated by the unexpected finding of much lower plasma levels of catecholamines after exercise in carbonic-anhydrase-treated fish (Fig. 6B,C). Regressions between relative and plasma adrenaline concentration or noradrenaline concentration or the sum of these two catecholamines (similar to those of Fig. 9) yielded much weaker but significant correlations (r=0.27–0.41, N=52–57). This raises several alternative scenarios. (i) Because these fish mobilized lower levels of catecholamines, there was less adrenergic stimulation of the RBCs and therefore less build-up and acidosis, and therefore less stimulation of ⁠. (ii) Because these fish became less acidotic, they mobilized lower levels of catecholamines, and there was less effect on the RBCs, etc. (iii) Because these fish mobilized lower levels of catecholamines, there was less direct stimulation of by catecholamines themselves. From the present results, we cannot eliminate any of these possibilities.

The role of circulating catecholamines in direct ventilatory stimulation remains highly controversial, at least in trout (Aota et al. 1990; Playle et al. 1990; Kinkead and Perry, 1990, 1991; Kinkead et al. 1991, 1993; Randall and Taylor, 1991; Perry et al. 1992; Randall and Perry, 1992). The nature of the stimuli for catecholamine release is another area of controversy (see Randall and Perry, 1992; Thomas and Perry, 1992, for surveys of current ideas). Nevertheless, there certainly exists evidence that the extent of acidosis and the extent of catecholamine mobilization are related (Boutilier et al. 1986; Tang and Boutilier, 1988). We therefore favour some combination of the above scenarios such that the lower degree of acidosis and the reduced catecholamine mobilization were closely matched to one another by feedback loops in the carbonic-anhydrase-injected fish, resulting in a lower post-exercise ventilation. Whatever the exact details here, our overall conclusion is that elevation and the associated respiratory acidosis play an important role in driving hyperventilation after exhaustive exercise.

We thank Rod Rhem and Marnie Goldstein for excellent technical assistance. Supported by NSERC research grants to C.M.W.