Evidence for the Presence of an Electrogenic Proton Pump on the Trout Gill Epithelium

Carbon dioxide and proton excretion acidifies water as it passes over the gills, whereas NH₃ excretion raises pH. The effect of carbon dioxide and proton excretion dominates at water pH above 5.3. Below pH5.3, the effect of ammonia excretion dominates and the pH of water increases as it passes over the gills (Lin and Randall, 1990).

Rainbow trout gills constitute a tight epithelium, permeable to NH₃ and CO₂. Carbon dioxide is excreted from the blood as CO₂, the gill epithelium being impermeable to HCO₃⁻ (Perry et al. 1982). It has been suggested that a portion of the carbon dioxide entering the gill epithelium is hydrated, forming bicarbonate, which is then exchanged for chloride across the apical membrane. The addition of 4-acetamido-4’-isothiocyanatostilbene-2-2’-disulphonic acid (SITS), known to block C1⁻/HCO₃⁻exchange in red blood cells, to the water results in a rise in blood pH in trout (Perry et al. 1981) and a reduction in Cl⁻ uptake (Perry and Randall, 1981).

The trout gill NH₃ permeability coefficient of 6×10⁻³cms⁻¹ (Avella and Bornancin, 1989) is intermediate between values reported for the toad bladder and mammalian kidney tubule. It has been suggested that ammonia excretion, although dominated by NH₃ diffusion (Hillaby and Randall, 1979; Cameron and Heisler, 1983), is also mediated by Na⁺/NH4⁺ exchange on the apical surface (Payan, 1978; Wright and Wood, 1985). The stimulatory effect of NFL»* on sodium flux, however, could be explained in terms of a pH effect of the ammonia addition (Cameron and Kormanik, 1982), and Avella and Bornancin (1989) concluded that the balance of evidence was against the presence of Na⁺/NH₄⁺ exchange across the apical surface of trout gills. They considered the trout gill to be similar to other tight epithelia, such as frog skin and toad bladder, in that passive sodium uptake from water is indirectly coupled to an active electrogenic proton transport system.

The object of this study was to determine the relative contributions of these various components to the change in pH of water as it passes over the gills.

Rainbow trout Oncorhynchus mykiss (Walbaum) weighing 202–592 g were maintained in outdoor fibreglass tanks supplied with flowing dechlorinated Vancouver tapwater. Fish were fed daily with commercial trout pellets and feeding was suspended for at least 48h prior to experimentation.

Surgery was performed on each fish under general anaesthesia (1:10000 MS222 solution, pH adjusted to 7.5 with NaHCO₃) to fix an opercular cannula for sampling expired water. Fish were then confined, but not physically restrained, in a black chamber to recover for at least 24 h. This black chamber was supplied with aerated dechlorinated tapwater during the recovery period. Three hours prior to the experiment, the water supply was switched to the aerated test solution (40 mmol l⁻¹ NaCl and 0.5 mmol l⁻¹ CaCl₂ in dechlorinated tapwater, from Wright et al. 1986) with a buffering capacity (β) of 81μequivl⁻¹pHunit⁻¹. The test solution had the same ionic strength as the buffer solution used to calibrate the pH electrodes. By using this test solution, we reduced the response time of the pH electrode, increased its stability and thus obtained more precise water pH measurements. Temperature was regulated to that of tapwater with a cooling coil.

Three sets of experiments were carried out using a recirculating system connected to the black chamber. One was a control experiment with the test solution only. The other two were amiloride and SITS treatments, with either amiloride or SITS added to the test solution to give a final concentration of 0.1 mmol l⁻¹. In each set of experiments the fish was exposed for 30 min to four environmental pH levels - pH7, pH6, pH5 and pH4. The volume of the recirculating system was 61 and it was aerated and controlled at temperatures between 6.5 and 8.5°C (winter experiments). A magnetic stirring bar was used in the reservoir to ensure complete mixing. The pH of the test solution was adjusted to 7 by adding 0.1 mol l⁻¹ NaOH or 0.1 mol l⁻¹ HC1 and then quickly lowered to 6, 5 and 4 by adding 0.1 mol l⁻¹ HC1 at the beginning of each different environmental pH exposure. The amount of acid added was measured precisely and pH was recorded to construct buffer curves. The water pH of the recirculating system changed slightly (<0.3 unit) over the 30-min experimental period, and no attempt was made to stabilize pH during this period. Inspired and expired water samples (approximately 5 ml each) were withdrawn from the outlets of the glass electrode chambers at the end of each 30-min exposure. Between each exposure there was a 15-min exchange time during which the water was mixed and pH changed to a new level. The results of pH 7 and pH 6 treatments were pooled for data analysis because there was no significant difference between them.

This study employed the recirculating system described above. Three concentrations of amiloride were utilized and their effects on CO₂, ammonia and net proton excretion determined. Each experiment started with a 1-h control period, with the fish resting in the recirculating system containing the test solution alone. Inspired and expired water samples were taken at 30 min and 60 min. The system was flushed with fresh test solution and amiloride was added to the system to give a final concentration of 0.1, 0.5 or 1mmoll⁻¹. Recirculation was restored and the amiloride treatment lasted for another hour. Water samples were taken at 30 min and 60 min. The buffer capacities of inspired and expired water were measured later by titrating the stored water samples using 0.1 moll⁻¹ NaOH and 0.1 moll⁻¹ HC1. The results of the 30-min and 60-min sampling were pooled for data analysis since there was no significant difference between them. Experiments were performed during the summer at temperatures between 13 and 17°C.

Using the same recirculating system, 0.1 mmol l⁻¹ vanadate or 0.1 mmol l⁻¹ acetazolamide was added to the external water to examine their effect on CO₂, ammonia and proton excretions. Each experiment started with a 1-h control period followed by a 1-h treatment period. Since the ammonia accumulation in the system was very low (less than 100μmol l⁻¹ after 2h), flushing the system with fresh test solution at the beginning of the treatment period was considered unnecessary. For the vanadate treatment, the pH of the test solution was adjusted to between 7 and 8 and, 30min later, to between 8 and 9 by titrating with 0.1 mol l⁻¹ NaOH, in order to investigate proton excretion in the pH range 7-9. Freshly made 3 mmol l⁻¹ sodium orthovanadate (Na₃VO₄) solution was boiled and the cooled solution was neutralized with 0.1 moll⁻¹ HC1. After the control period, £00 ml of the 3 mmol l⁻¹ vanadate solution was added to the 6-1 recirculating system to obtain a final concentration of 0.1 mmol l⁻¹. External water pH was adjusted to the control value during the two 30-min treatment periods. For the acetazolamide treatment, water pH was adjusted to around 7 and acetazolamide was added to the system to give a final concentration of 0.1 mmol l⁻¹ after the control period. The same set of measurements as described in ‘amiloride treatment’ was performed every 30 min in both control and treatment periods. Temperature in these experiments was 7–9°C (spring experiments).

Inspired and expired water pH were monitored during the whole experimental period with combination glass pH electrodes housed in two water-jacketed glass chambers (Wright et al. 1986). Inspired pH (pH_in) and expired pH (pHe_ex) values were recorded at each sampling.

Total carbon dioxide contents of inspired water [CO₂]_in and expired water [CO₂]_ex were measured immediately with a Carle gas chromatograph (model III) containing a CO₂ discriminating column (porapak Q) (Boutilier et al. 1985; Lenfant and Aucutt, 1966).

Total ammonia contents of inspired water [Amm]_in and expired water [Amm]_exwere measured by a micro-modification of the salicylate-hypochlorite assay with frozen water samples (Verdouw et al. 1978). To ensure that there was no ammonia loss from water in the recirculating system, two experiments were carried out in which known amounts of NH4CI were added to the system without fish. No loss from the system occurred.

Bicarbonate concentrations in inspired water [HCO₃⁻]_in and expired water [HCO₃⁻]_ex were calculated from [CO₂]_in, pH_in and [CO₂]_ex, pHe_ex, respectively, by the Henderson-Hasselbalch equation, using the $pKCO2$ and $αCO2$ values from Boutilier et al. (1985). The difference in [HCO₃⁻] between the expired and inspired water, [HCO₃⁻]_ex–[HCO₃⁻]_in, was equivalent to the calculated proton addition due to CO₂ excretion by the fish (assuming no HCO₃⁻ excretion). Carbonate formation is negligible over the pH range.

Ammonium ion concentrations in inspired water [NH4⁺]_in and expired water [NH4⁺]_ex were calculated from [Amm]_in, pH_in and [Amm]_ex, pH_ex, respectively, by the Henderson-Hasselbalch equation, using the pX_Amm value from Cameron and Heisler (1983). The difference in [NH₄⁺] between the expired and inspired water, [NH4⁺]_ex—[NH4⁺]_in, was equivalent to the calculated proton consumption due to NH₃ excretion by the fish (assuming no NH₄⁺ excretion).

The differences in total CO₂ content between inspired water and expired water with different treatments are presented in Table 1. [CO₂]_ex–[CO₂]_in represents the CO₂ excretion rate if we assume that ventilation rate is constant. There was no significant difference in [CO₂]_ex–[CO₂]¡_n between control and drug-treated animals except in the case of acetazolamide. CO₂ excretion increased when fish were exposed to 0.1mmoll⁻¹ acetazolamide in water.

When water pH was approximately neutral, [HCO₃⁻]_in was always greater than [HCO₃⁻]_ex (Fig. 1), indicating that the water was acidified as it passed over the gills by something other than CO₂ addition. When proton excretion was inhibited by 0.1 mmol l⁻¹ vanadate, 0.1 mmol l⁻¹ acetazolamide or 0.5 or 1 mmol l⁻¹ amiloride (see Figs 5,6), the extent of bicarbonate dehydration decreased and [HCO₃⁻]_ex increased. Therefore, [HCO₃⁻]_in–[HCO₃⁻]_ex was significantly lower than the control value. In addition, the control [HCO3⁻]_in–[HCO₃⁻]_ex value at pH7.6 was greater than that at pH6.6, indicating that the excretion of acid equivalents was higher at pH 7.6.

As external water pH decreased, the ammonia excretion rate was not statistically different from that at neutral pH within control, 0.1 mmol l⁻¹ amiloride and 0.lmmoll⁻¹ SITS treatments (Fig. 2). The difference between control and SITS treatments, tested by ANOVA, was not significant, but there was a significant difference between control and amiloride treatments (notice that control, amiloride and SITS treatments were performed on different group of animals). Ammonia excretion of the same animals was not significantly inhibited by O.lmmoll⁻¹ amiloride (Fig. 3). However, higher concentrations of amiloride induced a reduction in ammonia excretion, by 58% with 0.5 mmol l⁻¹ amiloride and by 87% with 1 mmol l⁻¹ amiloride (Fig. 3). Vanadate and acetazolamide had no significant effect on ammonia excretion.

In external water below pH 6, net proton excretion was approximately zero (Fig. 4), and the addition of either 0.1mmoll⁻¹ amiloride or 0.1mmoll⁻¹ SITS had no effect on net proton excretion (Fig. 4). At neutral pH, however, there was a net proton excretion across the gill epithelium. This excretion was also unaffected by the addition of either 0.1 mmoll⁻¹ SITS or 0.1 mmol l⁻¹ amiloride. Increasing amiloride concentration in the external medium induced a reduction in proton excretion (Fig. 5), but more than 50% of the net proton excretion was still sustained even in the presence of 1 mmoll⁻¹ amiloride. 0.1mmoll⁻¹ vanadate treatment resulted in reductions of net proton excretion by 58% and 67% when fish were exposed to neutral and moderately alkaline water, respectively (Fig. 6). Acetazolamide treatment also caused a 48% reduction in net proton excretion (Fig.6).

The relationship between net proton excretion and the pH difference across the gill epithelium is illustrated in Fig. 7. When water pH is below blood pH by more than 2.5 units, there is no net proton excretion; that is, any change in water pH can be accounted for by CO₂ hydration/HCO₃⁻ dehydration and/or NH₃ protonation. Net proton excretion was completely inhibited at low external pH. When the pH difference was more than –2.5 units, net proton excretion increased with inspired water pH. Acid excretion was maximal when inspired water pH was equal to or higher than blood pH.

Fig. 8 shows the relationship between net proton excretion and expired water carbon dioxide levels in a neutral environment. There is a general tendency for increased $PCO2$ in expired water to lead to an increase in net proton excretion.

The observation that SITS treatment did not result in a reduction of carbon dioxide excretion across the gills indicates that C1⁻ /HCO₃⁻ exchange is playing only a minor role in carbon dioxide excretion. If 10% of the total CO₂ excretion is in the form of HCO₃⁻ in exchange for Cl⁻, then blockage of this exchange pathway may result in an elevation of $PCO2$ in the fish gill, which, in turn, will enhance the molecular CO₂ diffusion and re-establish normal CO₂ excretion rates. The extent of conversion of CO₂ to HCO₃⁻ in expired water is unimportant for CO₂ excretion, because water pH has no effect on carbon dioxide excretion (Lin and Randall, 1990), and is consistent with the view that the rate-limiting step in carbon dioxide excretion is C1⁻ /HCO₃⁻ exchange across the red blood cell membrane (Perry et al. 1982). SITS had no measurable effect on acid excretion across the trout gill in this study, but Perry et al. (1981) reported that SITS reduced chloride influx and caused an alkalosis in trout after 6 h of exposure. This indicates that the presence of a C1⁻ /HCO₃⁻ exchange mechanism on the fish gill has only a minor effect on acid-base balance of the fish, below that detectable in this study.

It has long been hypothesized that Na⁺/H⁺(NH4⁺) electroneutral exchange is the principal mechanism of acid-base regulation in the gill epithelium of fish (Wright and Wood, 1985). This antiport exchange process is blocked by 0.1 mmol l^{− 1} amiloride, a very potent and relatively specific inhibitor of sodium transport in a wide variety of cellular and epithelial transport systems (Benos, 1982). 84% and 94% reductions of the Na⁺ uptake by the gills of intact freshwater rainbow trout exposed to 0.1mmoll^{− 1} amiloride in the external medium were reported by Perry and Randall (1981) and Wright and Wood (1985), respectively. In our studies, this concentration of amiloride had no effect on either net proton or ammonia excretion when compared with control values from the same animals (Figs 3 and 5). Our experimental conditions are similar to those of Perry and Wright. This indicates that sodium influx and either proton or ammonium ion efflux are not directly coupled (see also Avella and Bornancin, 1989).

It is well documented that proton transport in mammalian kidney (Steinmetz, 1985), amphibian urinary bladder (Al-Awqati, 1978; Steinmetz, 1986) and frog skin (Ehrenfeld et al. 1985) is mediated by an electrogenic proton pump. The gill epithelium in freshwater fish is considered to be ‘tight’ (Sardet, 1980) and resembles frog skin and turtle bladder epithelia functionally and morphologically. Our studies demonstrated that proton excretion in trout was unaffected by low concentrations of amiloride but was inhibited by vanadate, acetazolamide and low water pH, which is associated with a reversal of gill transepithelial potential (McWilliams and Potts, 1978; Ye et al. 1991). In addition, elevated water $PCO2$ stimulated proton excretion in fish gills. All these phenomena have been reported for frog skin (Ehrenfeld et al. 1985; Ehrenfeld and Garcia-Romeu, 1977) and turtle bladder (Al-Awqati, 1978; Steinmetz, 1986). Considering all of these lines of evidence together, we agree with Avella and Bornancin (1989) and conclude that the fish gill has an electrogenic proton pump in the mucosal membrane, similar to that reported for frog skin and toad bladder, rather than a Na⁺/H⁺ exchange mechanism.

The electrogenic proton pump, or H⁺-translocating ATPase, on the apical membrane removes protons from the cell and generates a negative potential on the inner side of the apical membrane (Fig. 9). Sodium influx, driven by the negative potential, occurs via a sodium channel that is highly sensitive to amiloride. A Na⁺/K⁺(NH4⁺)-ATPase in the basolateral membrane pumps sodium out of the cell into the blood. Thus, proton excretion and sodium uptake are intimately, but indirectly, linked. Since Avella et al. (1987) showed that branchial sodium uptake was proportional to the number of chloride cells in the gills and since proton pumps consume energy and chloride cells are rich in mitochondria and can supply the energy demand, we conclude that the electrogenic proton pump is located in the chloride cell (Fig. 9).

Ammonium ions can replace potassium on the Na⁺/K⁺-ATPase and thus enter the cell and form NH₃ and protons (Evans et al. 1989). The deprotonation of NH₄⁺ could supply the proton pump and NH₃ could diffuse passively across the apical membrane into the water. Although much less sensitive than the Na⁺ channel, the Na⁺/K⁺-ATPase in the basolateral membrane can be inhibited by amiloride that has entered the cell when applied in high concentrations to the mucosal side (Knauf et al. 1976; Kleyman and Cragoe, 1988). Thus, the reduced proton and ammonia excretion in 0.5 and O.lmmoll⁻¹ amiloride treatments could be accounted for by the inhibitory effect of amiloride on Na⁺/K⁺(NH4⁺)-ATPase in the basolateral membrane. In support of this contention, Evans et al. (1989) showed that amiloride did not affect ammonia excretion if the perfused head of the toadfish was pretreated with ouabain, which blocks Na⁺/K⁺(NH4⁺)-ATPase. Proton excretion in frog skin was inhibited by 0.5 mmol l^{− 1} amiloride by 35% but was not affected by 0.05 mmol l^{− 1} amiloride, whereas sodium uptake was completely abolished. If we assume that the ventilation rate of the fish was 100mlmin^{− 1} (Lin and Randall, 1990), we can compare the ammonia excretion rate with the net proton excretion rate under amiloride treatments (Table 2). The reduction in ammonia excretion was equivalent to that in proton excretion, indicating the possibility that NH₃ and protons were both originating from NH4⁺ transported into the epithelium via the Na⁺/K⁺(NH4⁺)-ATPase in the basolateral membrane (Fig. 9).

Ammonia elimination was not affected by vanadate or acetazolamide, indicating that proton and ammonia efflux from the gill epithelium are through different pathways. Thus, ammonium entry into the gill epithelium may affect proton excretion (Table 2), but variations in proton excretion do not appear to affect ammonia excretion. Ammonium cannot be the sole source of protons however, because proton excretion can be more than twice ammonia excretion in some instances. Elevated carbon dioxide levels appear to enhance proton excretion (Fig. 8), as observed in toad bladder (Al-Awqati, 1978), indicating that carbon dioxide is also a source of protons for the pump. Protons can be generated in the cellular compartment from CO₂ hydration catalysed by carbonic anhydrase. Acetazolamide, a traditional carbonic anhydrase inhibitor, inhibits proton excretion in fish gills (Fig. 6), as it does in frog skin and turtle bladder (Ehrenfeld and Garcia-Romeu, 1977; Steinmetz, 1986). While inhibiting proton excretion, acetazolamide also elevates carbon dioxide excretion across fish gills (Table 1). The reduction in intracellular CO₂ hydration will probably enhance the passive diffusion of CO₂. The amount of proton excretion in this study is of the same magnitude as that reported by Avella and Bornancin (1989).

Vanadate has a nonspecific inhibitory effect on ATPases and could be acting on Na⁺/K⁺-ATPase on the basolateral border of the fish gill. In our studies, more than 50% of the net proton excretion across the gill epithelium was inhibited by O.lmmoll^{− 1} vanadate applied to the mucosal membrane. De Sousa and Grosso (1979) showed that applying 1 mmol l^{− 1} vanadate to the outer surface did not affect the Na⁺/K⁺-ATPase in the basolateral membrane of frog skin. Arruda et al. (1981) showed that vanadate had no effect on the backleak of proton or bicarbonate secretion but had a direct effect on H⁺-translocating ATPase in turtle bladder. Thus, we conclude that the reduction in proton excretion observed in our studies was induced by the inhibitory effect of vanadate on the H⁺-translocating ATPase in the apical membrane. The reason that the proton excretion was not completely abolished was, presumably, because of the difficulty of vanadate reaching the action site from the mucosal side (Arruda et al. 1981).

0.1mmoll^{− 1} amiloride had no effect on the putative fish gill proton pump in open-circuit conditions. Inhibition of sodium influx should have increased membrane potential and reduced proton excretion. This did not happen; therefore, if the proton pump does exist, there must be some other counter-ion that can replace sodium. Perry and Randall (1981) found that amiloride inhibited chloride influx in the fish gill. Inhibition of both chloride and sodium influx, when fish are exposed to amiloride, would tend to ameliorate any rise in potential across the apical membrane and, therefore, permit continued functioning of the proton pump.

The relationship between net proton excretion and Δ pH across the epithelial membrane shown in Fig. 7 is very typical of proton transport mediated by an electrogenic proton pump. At constant serosal pH, net proton secretion increased linearly with luminal pH over the physiological range of urine pH (4.4-7.4). Proton secretion was maximal at higher pH (Steinmetz, 1986). A linear relationship between proton excretion and mucosal pH over a limited range was also reported in frog skin by Ehrenfeld et al. (1985). This indicates that the electrochemical gradient for protons across the membrane is a fundamental regulator of the active proton transport, as discussed by Steinmetz (1986).

We found no inhibition of ammonia excretion at low pH, whereas Wright and Wood (1985) observed a marked depression of ammonia excretion in trout exposed to acid conditions. Low pH inhibits the proton pump, which will reduce epithelial pH, lower NH₃ levels in the epithelium and reduce ammonia excretion. Acid conditions in the water, however, will also lower NH₃ levels in the water and enhance ammonia excretion. The actual rate of ammonia excretion will vary initially, depending on the balance between the two effects, changing NH₃ levels and total ammonia stores in various compartments. Ultimately, excretion will match production and there is no reason to suppose that exposure to acid conditions has any effect on ammonium production.

In conclusion, our results provide evidence that the acidification of expired water in rainbow trout in neutral water is mainly caused by a net proton excretion mediated by an active proton pump on the apical membrane of gill lamellae. This proton pump is sensitive to vanadate and acetazolamide and is modulated by ambient CO₂ and pH. It resembles the electrogenic proton pump in frog skin and turtle bladder in many features.

This research was supported by an NSERC (Canada) grant to D.J.R. The authors would like to thank Dr Yong Tang and Mark Shrimpton for their helpful discussions on these studies.