The Distribution of Branchial Carbonic Anhydrase and the Effects of Gill and Erythrocyte Carbonic Anhydrase Inhibition in the Channel Catfish Ictalurus Punctatus

The enzyme carbonic anhydrase (CA) was originally discovered in mammalian erythrocytes (Meldrum & Roughton, 1932) ; since then it has been found in red blood cells of every vertebrate species examined (reviewed by Maren, 1967a,b; Sanyal, 1984). The enzyme, which catalyses the reversible hydration of carbon dioxide, is known to be important in the transport and excretion of respiratory CO₂ both in animals that breath air and in those that breath water (Cameron, 1979; Swenson, 1984; Randall & Daxboeck, 1984; Perry, 1986). While erythrocyte CA appears to be most critical to CO₂ exchange, the major organs of respiration, the lung and the gill, also contain significant levels of CA activity (Effros, Shapiro & Silverman, 1980; Henry, Dodgson, Forster & Storey, 1986; Houston & McCarty, 1978; Houston & Mearow, 1982). Thus, respiratory tissue CA could also play a role in CO₂ excretion.

The respiratory role of pulmonary CA has been thoroughly investigated. Carbonic anhydrase is associated with the pulmonary endothelial membrane (Whitney & Briggle, 1982; Henry et al. 1986). As such, it is available to catalyse the dehydration of plasma HCO₃⁻, facilitating CO₂ excretion and maintaining pulmonary capillary equilibrium (Effros, Chang & Silverman, 1978; Bidani & Crandall, 1978; Bidani, Crandall & Forester, 1978; Klocke, 1978, 1980). However, the potential respiratory role of branchial CA is still uncertain.

Originally, CA activity in erythrocytes of fish was viewed as being sufficient for normal CO₂ transport and excretion (Maren, 1967a), and it was suggested that branchial CA functioned in hydrating respiratory CO₂ within the epithelial cells to H⁺ and HCO₃⁻ for ion transport (Maetz, 1971; Kirschner, 1979). Conflicting evidence was presented by Haswell & Randall (1976, 1978) discounting the role of erythrocyte CA and assigning a respiratory function to branchial CA by default (Haswell, Randall & Perry, 1980). These ideas were shown to be incorrect (Cameron, 1978; Obaid, Critz & Crandall, 1979), and further experiments confirmed the respiratory function of erythrocyte CA (Wood, McDonald & McMahon, 1972; Perry, Davie, Daxboeck & Randall, 1982). The potential contribution of branchial CA to CO₂ excretion has never been completely disproved, and evidence is still being accumulated indicating a role for the enzyme in the elimination of respiratory CO₂ in some species (Swenson, Hildesley & Maren, 1982; Swenson, Maren & Hildesley, 1984; Swenson & Evans, 1984).

CA in fish gills has been histochemically localized to both the ion-transporting chloride cells and the respiratory lamellae (Dimberg, Hoglund, Knutsson & Ridderstrale, 1981), but the quantitative distribution of CA within the gill is unknown. Also, no attempt has been made to correlate branchial CA distribution, its availability to plasma HCO₃⁻, and the respiratory and acid-base parameters of blood in the intact organism. The presence of CA on the basal membrane of the gill and its availability to plasma would allow for a respiratory function of the gill enzyme, but this possibility has never been investigated. The objectives of this study were to investigate the quantitative subcellular distribution of branchial CA, and to investigate the comparative contributions of erythrocyte and branchial CA to respiratory CO₂ excretion.

Channel catfish (Ictalurus punctatus) weighing between 500 and 1200 g were obtained from commercial ponds in Angleton, Texas. They were held in 2000-1 fibreglass tanks equipped with recirculating biofilters at 23 ± 1°C. Since animals were held for only a few days before use, they were not fed.

Before any experiment, individuals were anaesthetized in MS-222 (tricaine methanesulphonate, Sigma) and were fitted with two catheters while being ventilated artificially. A blood-sampling catheter (PE 60) was implanted in the dorsal aorta (DA) according to the method of Soivio, Nyholm & Westman (1975); a second catheter (PE 160), for pressure recording, was placed through the operculum and fastened with a sleeve (PE 200) against the outer surface. Catheters were filled with heparinized Cortland saline (Wolf, 1963).

Following surgery, catfish were transferred to darkened individual Plexiglas chambers with running dechlorinated tap water and allowed to recover for approximately 24 h.

Three control blood samples (approx. 1 ml) were taken from resting catfish via the DA catheter, before a treatment was administered. Blood pH was measured with a thermostatted glass microcapillary electrode (Radiometer G297/K497; Radiometer pHM71 blood gas analyser); was determined with a thermostatted cell and O₂ electrode (Radiometer E5046/D616). The remaining blood was then immediately centrifuged for about 15 s in an Eppendorf Microfuge. Total CO₂ (CT) was measured on 20-μl samples of plasma with a Capni-Con total CO₂ analyser (Cameron Instruments Co.). The values of pH and CT were used to calculate with values for solubility calculated from Boutilier, Heming & Iwama (1984) and the pK′calculated using the equation given by Heisler (1984).

Catfish were given an injection of one of the following carbonic anhydrase inhibitors: quaternary ammonium sulphanilamide (QAS-chloride or QAS-bromide; Ferrosan), acetazolamide (Sigma) or benzolamide (the gift of Dr Thomas H. Maren, the University of Florida), through the DA catheter. The inhibitors were dissolved in Cortland saline; the pH values of the stock solutions were between 7·40 and 8·00, depending on the inhibitor and the concentration (dilute NaOH was added to dissolve acetazolamide and benzolamide fully). Initial measurements on QAS indicated that the inhibitor was rapidly distributed throughout the total extracellular fluid (ECF) pool. Thus, the inhibitor dosages were calculated to yield an initial ECF concentration of Immoll⁻¹ for QAS and 10 μmoll⁻¹ for acetazolamide and benzolamide after an approximate 1:100 dilution from stock solutions. ECF volume was estimated at approximately 20% of total body weight (Cameron, 1980); volume of injectate was between 1 and 3 ml, depending on the weight of the fish. Blood samples were taken at regular intervals between 10 min and 24 h after injection, and blood O₂ and acid-base variables were determined as described above. Plasma and packed red cells were saved for subsequent measurements (see below).

The opercular catheter was connected to a pressure transducer (Micron MP15D) and strain gauge preamplifier (Coulbourn Instruments, Type A) to record gill ventilation frequency (fG) and opercular pressure amplitude (POP); these data were displayed on a thermal recorder (Watanabe). Heart rate (fH) and blood pressure in the DA (PDA) were also analysed via a pressure transducer (Cobe) connected to the DA catheter. These variables were analysed immediately before injection of a CA inhibitor and at intervals between 1 and 30 min post-injection. Opercular pressure was analysed as the mean of 10–15 individual pressure excursions, while mean systolic pressure was taken from the average of 10 pressure measurements taken over a 1-min period. Changes in these values with time were statistically analysed using a two-way analysis of variance (ANOVA), no replication, with the fiducial limit of significance at P<0 ·05. The ventilatory and cardiovascular responses to benzolamide were measured on only three animals and were not statistically analysed.

Carbonic anhydrase activity was measured in both plasma and haemolysates of packed red cells (lysed by the addition of distilled water). The catalysed rate of CO₂ hydration was determined according to the electrometric method of Maren & Couto (1979) as modified by Henry & Kormanik (1985). Protein concentrations were determined using the Coomassie brilliant blue dye-binding technique (Bio Rad Laboratories), and CA activity was reported as μmol CO₂ mg protein ⁻¹ min ⁻¹.

The concentration of injected CA inhibitor in plasma was measured by the method of Maren (1960). Briefly, a standard amount of CA (purified bovine erythrocyte CA, Sigma) was titrated with increasing amounts of a known concentration of inhibitor, thus producing a standard curve for CA inhibition vs inhibitor concentration. The CA standard was then titrated with plasma, and the inhibitor concentration was determined from a comparison of the amount of the resultant CA inhibition with that from the standard curve. Similar concentrations and volumes of inhibitor and plasma were used to avoid any dilution artefacts.

In one group of fish, total gill CA activity was also measured after treatment with acetazolamide. In these fish, the inhibitor was given via the caudal artery while the animal was under anaesthesia. The gills were ventilated with oxygenated water and the animals were kept under anaesthesia for the duration of the experiment. At 30 min post-injection, the heart was exposed and a catheter was inserted into the bulbus arteriosus. The gills were perfused with calcium-free, heparinized saline made hypoxic by bubbling with nitrogen to promote vasodilation. The heart was cut to allow for drainage, and the gills were perfused for 5–10 min until they appeared clear. The gill arches were cut out and the filaments were removed, placed into four volumes of cold buffer and homogenized (see below). Only filaments that were completely free of red cells or clots were used. The homogenate was centrifuged for 20 min at 10 000 g at 5°C (Sorvall RC-2B), and the supernatant was assayed for CA activity as described above.

Catfish were anaesthetized, and the gills were perfused free of erythrocytes as described above. Filaments were dissected free of the gill arch and placed into approximately four volumes of cold buffer (225 mmol l⁻¹ mannitol, 75 mmol I⁻¹ sucrose, 10 mmol l⁻¹ Trizma base, adjusted to pH 7·40 with phosphoric acid) (Henry & Kormanik, 1985; Henry et al. 1986). Branchial tissue was homogenized with a motor-driven Teflon/glass homogenizer. The crude homogenate was then subjected to differential centrifugation at 4°C to produce three pellet fractions and a final supernatant: (1) cellular debris (whole cells, nuclei and large cell fragments) (270g for 15 min), (2) a mitochondrial pellet (7500g for 20min), and (3) a microsomal pellet (110 000g for 90min). The remaining supernatant represented the cytoplasm (Henry et al. 1986).

Each fraction was then assayed for CA activity as described above. To differentiate between loosely bound CA and more integrally associated CA, each fraction was also resuspended in buffer and vigorously agitated by vortexing. The pellet was then recentrifuged, the supernatant was drawn off, and the pellet was resuspended in fresh buffer. CA activities in the resuspended pellet and the supernatant were determined. This allowed for a rough estimate of integral vs peripheral particulate CA to be made, and it also gave a measure of total recovery after the pellet was washed.

Plasma concentrations of sodium and potassium were measured by flame photometry (Radiometer FLM3). Chloride concentrations were determined by coulometric titration (Buchler-Cotlove).

Carbonic anhydrase activity was found in all subcellular fractions of gill tissue (Fig. 1). The final supernatant (cytoplasm) contained by far the greatest level of CA activity, making up approximately 56% of the total gill enzyme activity by weight and 67% of the specific activity (Table 1). Spectrophotometric determination of haemoglobin (at 570 nm) in the supernatant fraction indicated that potential contamination from erythrocyte CA was negligible. Of the three pellet fractions, the cell debris fraction contained the highest levels of CA activity: over 10-fold greater than either mitochondrial or microsomal CA activity (Fig. 1). Upon resuspension and vigorous washing of the cell debris pellet, approximately 60% of the initial particulate-associated CA activity was transferred to the supernatant (Fig. 2). Thus, it appears that most of the CA activity in the cell debris pellet is not true particulate-associated CA, but rather it is soluble CA of the gill fragments and whole cells that is liberated after disruption of the pellet. The same pattern is seen for mitochondrial CA (Fig. 2). The microsomal CA, however, is more characteristic of true membrane-associated CA, as fully 95 % of the particulate CA remains in the pellet fraction after agitation and washing (Fig. 2). This membrane-associated CA made up about 9% of the total branchial CA activity by weight and nearly 11 % of the specific activity (Table 1). Throughout the fractionation and washing procedures, percentage recovery was at least 80% or better (Figs 1, 2).

Carbonic anhydrase activity in crude lysates of the erythrocytes was much higher than in any gill fraction. Specific activity of the lysate was over four-fold greater than branchial cytoplasmic CA activity and over 20 times greater than CA activity in the gill microsomes (Table 1).

The two CA inhibitors, acetazolamide and benzolamide, have a residence time in the catfish plasma of not more than 24 h. Both compounds appear to wash out of the plasma slowly and at approximately the same rate (Fig. 3A). Erythrocyte CA activity was virtually 100% inhibited 10min after injection of acetazolamide; this level of inhibition persisted over the initial 4h and then dropped to about 80% by 24 h (Fig. 3B). For benzolamide, CA inhibition at 10 min was only 80 % and did not reach 100% until 30min after injection. Benzolamide-induced CA inhibition was also present (80%) at 24 h.

For branchial tissue perfused with acetazolamide, CA inhibition was complete by 30min after injection; the time course of branchial enzyme inhibition was not followed past that point.

Quaternary ammonium sulphanilamide (QAS) was difficult to use in the catfish. The compound was cleared from the plasma rapidly; an initial injection of 1 mmol l⁻¹ produced a circulating concentration of less than 10 μmol I⁻¹ after 10min. At a concentration high enough to produce a circulating level of 400 μmoll⁻¹ QAS, the compound was fatal.

At rest, fish exhibited breathing patterns characterized by periods of apnoea in which gill ventilation ceased (e.g. Fig. 4). During the apnoeic periods, heart rate and blood pressure also fell. Injection of either acetazolamide or benzolamide caused a transient apnoea and bradycardia immediately following the injection (Fig. 5). The time courses of the responses to both inhibitors were similar; gill ventilatory frequency was significantly depressed immediately after injection but returned to control values within 5 min (Fig. 5). The depression in fG was strongly correlated with a transient reduction in dorsal arterial (Fig. 6). Opercular pressure was not affected in the acetazolamide-treated fish but was stimulated by benzolamide. Heart rate was depressed significantly by acetazolamide, but it recovered within 5 min and actually increased over the subsequent 25 min (Fig. 5). Dorsal aortic pressure increased transiently, but significantly, following an injection of acetazolamide, but it decreased after administration of benzolamide (Fig. 5). The transient changes in ventilatory and cardiovascular parameters were often accompanied by a loss of equilibrium in the animal.

Injection of acetazolamide or benzolamide resulted in the development of an acidosis (Figs 7 and 8, respectively). The predominant component of the acidosis was respiratory: increased approximately three-fold, from 3 to 9mmHg (1 mmHg = 133-3 Pa), and pH fell by 0·3 units following treatment with both inhibitors. Virtually all of the increase in took place within the initial 2-h period following inhibitor injection. The respiratory acidosis was still maximal at 24 h after treatment. Compensation of the acidosis was slow, with no appreciable [HCO₃⁻] increase until 4 h post-treatment, and only partial compensation was achieved at 24 h. A small, non-respiratory component accompanied the respiratory acidosis for about 2h following treatment with either inhibitor. This is probably a result of the disruption of normal ventilatory patterns and the concomitant drop in arterial (<10 μmoll⁻¹) had no effect on acid-base status.

The concentrations of the major ions in the plasma, Na⁺, Cl^- and K⁺, were not significantly altered (P >0 ·05, F-test) over the time course of the treatment by either acetazolamide or benzolamide. The strong ion difference (SID), which in this case was estimated as ([Na⁺] + [K⁺]−[Cl⁻]), did not exhibit a clear pattern over the initial period of the experiment for either inhibitor (Fig. 9). There was virtually no net change in the SID over the first 4h of treatment, the time when erythrocyte and branchial CA inhibition was maximal and during which the respiratory acidosis was uncompensated. Between 4 and 24 h, however, there was a slight increase in the SID which corresponded to the small degree of compensation. Throughout the experiment, however, the changes in the SID and the changes in plasma HCO₃⁻ were well matched.

The overwhelming majority of CA activity in the gills of the catfish is cytoplasmic, rendering it unavailable to plasma HCO₃⁻. Bicarbonate has been shown not to permeate artificial membranes (Gutknecht, Bisson & Tosteson, 1977), the endothelial membrane of the lung (Effros, Mason & Silverman, 1981) or the basal branchial membrane of fish (Perry et al. 1982). It is highly unlikely that HCO₃⁻ enters the gill, and thus the major fraction of branchial CA can be excluded from having a role in respiratory CO₂ excretion. Cytoplasmic CA has been localized primarily to the chloride cells of the teleost gill (Dimberg et al. 1981), where it undoubtedly functions in branchial ion transport processes (Kirchner, 1979). The level of CA activity in the cytoplasm of catfish gills is remarkably similar to that of the gills of other euryhaline organisms (mostly arthropods) that also probably use similar mechanisms of branchial ion transport and regulation (Henry, 1984).

Of the three particulate fractions, the cell debris pellet contained the highest initial level of CA activity. The removal of most of the CA activity after resuspension and a single washing, however, suggests that the cell debris pellet itself does not represent a significant independent fraction of gill CA activity. Rather, most of the CA activity is probably soluble CA, associated with intact cells and cell fragments, and liberated by the washing procedure. The mitochondrial pellet also possesses extremely low levels of CA activity, the bulk of which does not appear to be integrally associated with that fraction. Microsomal CA activity, although relatively low, appears to be integral as virtually none was lost from the fraction after washing. Although this fraction of the total gill CA activity is potentially available to mobilize plasma HCO₃⁻ for CO₂ excretion, it is extremely small. Gill microsomal CA specific activity is only about 4% of that in the erythrocyte; this difference must be very conservative, since erythrocyte CA specific activity is minimized by the high haemoglobin concentration and microsomal CA specific activity is magnified by the low protein concentration in that fraction. Thus it is doubtful that gill CA could contribute significantly to CO₂ excretion.

Acetazolamide, benzolamide and QAS were originally used in the hope that they would selectively inhibit intracellular and extracellular CA, and thus help to elucidate the role of membrane-associated (branchial) vs intracellular (erythrocytic) CA. The use of these inhibitors in conscious fish was, however, unexpectedly complicated. Injection of all three inhibitors elicited potent systemic responses. The loss of equilibrium and, in particular, the altered ventilatory patterns suggest that the initially high inhibitor concentration in the plasma had toxic side effects on the central nervous system. The depressed gill ventilation and subsequent hypoxaemia probably stimulated the bradycardia and pressor response, which are typical teleost reflex responses to hypoxia (see Shelton, Jones & Milsom, 1986, for a recent review). These effects were minimized, but not entirely eliminated, by giving the injections over a 2-min period. It is likely that the impaired branchial gas transport contributed to the initial acidosis by raising arterial levels. The prolonged hypoxaemia may have additionally resulted in increased plasma lactate levels, thus contributing to, or accounting for, the observed transient metabolic acidosis.

QAS in the ECF at concentrations below 10 μmoll⁻¹ had no effect on blood acid-base status. This compound was probably restricted to the ECF (Henry, 1987), and the lack of an effect suggests that there is no significant fraction of CA (i.e. branchial membrane-associated CA) available to ECF HCO₃⁻. The concentrations that were sub-lethal to the animal were also probably close to or below the K_i (inhibitor constant) of the inhibitor (Henry, 1987) ; thus, it is also possible that some putative membrane-associated CA remained uninhibited. The other two inhibitors used, acetazolamide and benzolamide, are much more potent, and the concentrations used were between 100 and 1000 times their K₁ values, respectively (Sanyal, 1984). Both compounds permeated the erythrocytes, although benzolamide moved more slowly. This difference in permeability allowed for an initial separation of the putative effects of extracellular vs intracellular CA inhibition (i.e. 20 min after injection), but that difference was much less than previously reported for other systems (Maren, 1967a).

The results of these inhibitor studies also confirm the conclusions concerning the relative contributions of erythrocyte and branchial CA to CO₂ excretion. The onset and duration of the respiratory acidosis was correlated with the inhibition of erythrocyte CA. A significant acidosis did not develop until 20 min after administration of either inhibitor; at that time erythrocyte CA was between 80 and 100% inhibited. If branchial microsomal CA were involved in CO₂ excretion it would have been immediately inhibited by either acetazolamide or benzolamide circulating in the plasma, and a respiratory acidosis would have developed much sooner. This was not the case. Also, the respiratory acidosis persisted for 24 h after injection, by which time both inhibitors had been cleared from the plasma. Thus, any branchial CA involved in CO₂ excretion should have been functioning at that time, and the respiratory acidosis should have been at least partially ameliorated. Again, this was not the case. Also, at this time erythrocyte CA was still largely inhibited. These results strongly suggest that branchial CA is not quantitatively important in CO₂ excretion. Similar conclusions were obtained using an isolated, perfused fish head preparation (Perry et al. 1982).

It is not entirely clear why CA in teleost branchial tissue should lack a role in CO₂ excretion, whereas CA in the mammalian lung is known to facilitate the process. One speculation might be that branchial endothelial CA is not necessary for aquatic CO₂ excretion in fish. The high capacitance of water for CO₂, the relatively long residence time of branchial capillary blood (compared to mammalian pulmonary capillary residence time), and the presence of counter-current exchange may make the gill efficient enough at CO₂ exchange so that an additional source of CA other than that found in the erythrocyte is not required.

Aquatic organisms compensate blood acid-base disturbances by adjusting ion transport mechanisms in both gills and kidneys; this alters the blood strong ion difference (SID) and thus blood pH is readjusted (Cameron, 1976; Stewart, 1978). Branchial cytoplasmic CA becomes fully inhibited after 30 min of perfusion with acetazolamide and, although the time course of inhibition and recovery was not directly measured, presumably it is similar to that in the red blood cells. If so, gill CA would remain inhibited for 24 h, and its normal role in ion transport of providing H⁺ and HCO₃⁻ to be used as counterions for active Na⁺ and Cl⁻ uptake would be disrupted (Kirschner, 1979). Assuming there are ion transport processes involved in branchial acid-base regulation, we would expect an impairment of the ability to compensate acid-base disturbances. This prediction is supported by the data (Figs 7, 8). A comparable level of hypercapnia in normal channel catfish leads to a 5mequivl⁻¹ increase in HCO₃⁻ in 2h (from 6 to 11 mequivl⁻¹; Cameron, 1980), compared to reduction at 2h after treatment with acetazolamide or benzolamide. The relatively small compensation achieved after 24 h probably reflected gradual recovery of CA activity. At that time plasma SID and HCO₃⁻ were beginning to offset the decreased pH; erythrocyte CA was also beginning to recover from inhibition, and so presumably was branchial CA.

In summary, CA in the catfish gill does not appear to be important in respiratory CO₂ excretion. The fraction of branchial CA available to plasma HCO₃⁻ is very small compared to both the gill cytoplasmic fraction and CA in the erythrocyte. A respiratory acidosis that is induced by CA inhibition correlates with the inhibition of erythrocyte CA, and the impaired ability to compensate this acidosis correlates with the apparent disruption of branchial ion transport processes as a result of the inhibition of gill cytoplasmic CA.

The authors would like to thank Ms Anna Garcia for her technical assistance in this project. Supported by NSF DCB 87-17379 to RPH, NSF DCB 83-17914 to NJS, and NSF DCB 83-15833 to JNC. This work was also supported by funds from the Alabama Agricultural Experiment Station (AAES JI. no. 15-871273).