Modes of Ammonia Transport Across the Gill Epithelium of the Marine Teleost Fish Opsanus Beta

Ammonia, which is the dominant product of amino acid catabolism in fishes, is excreted predominantly across the gill epithelium. Evidence has been presented for the presence of various potential transport pathways in this epithelium, including basolateral and apical Na⁺/ NH₄⁺exchange, basolateral Na⁺/ NH₄⁺/ 2C1⁻ cotransport, paracellular and/or transcellular diffusion of NH₃, and paracellular diffusion of NH4⁺ (see Evans & Cameron, 1986, for a recent review). We have recently utilized the isolated, perfused head of the dogfish pup (Squalus acanthias) to quantify the role of these pathways in ammonia transport across the branchial epithelium of an elasmobranch (Evans & More, 1988). The current study used the isolated, perfused head of the Gulf toadfish (Opsanus beta) to examine the role of each of these pathways in a teleost because the gills of marine teleosts have been found to have a much higher cationic permeability than those of marine elasmobranchs (Evans, 1979) and, therefore, might be expected to excrete a larger percentage of ammonia as NH₄⁺ (Evans & Cameron, 1986). In fact, preliminary studies of this species have suggested that ionic diffusion of NH₄⁺ may be quite substantial (Goldstein et al. 1982) and that basolateral Na⁺/NH₄⁺ exchange may also play a role (Claiborne et al. 1982).

Sexually mature specimens of Opsanus beta were collected by a commercial shrimp fisherman in the Gulf of Mexico, near Cedar Key, Florida throughout the year. Individuals utilized in the present study were significantly smaller (about 20–40 g) than those used in previous studies (about 150–250 g: Oduleye et al. 1982). Animals were maintained at room temperature (21–24°C) in glass aquaria with charcoal/cotton filtration, and were not fed for at least 24 h before experimentation. Atlantic sea water (32–34 %o) was collected from the running seawater system at Marineland, Florida.

Perfusion of the isolated head followed essentially the procedure of Payan & Matty (1975) as modified by Claiborne & Evans (1980). Fish were injected with 1000 units of sodium heparin 20–30 min before anesthesia with 0·01 g l⁻¹ MS-222, and weighed to the nearest 0·1g. The pericardial cavity was exposed, the heart ligated between the atrium and ventricle, and the bulbus arteriosus/ventral aorta cannulated with about 3 mm of PE 50 tubing attached to a perfusion line composed of PE90 tubing joined to Gilson Tygon tubing (i.d. =0·76mm). The body was immediately removed from the head behind the pectoral fin, and perfusion initiated, so that the branchial vasculature was ischemic for only 2–5 min. The swimbladder and gut were removed distal to ligatures, but the fiver was usually not ligated before its removal since preliminary experiments indicated that ligation of hepatic vessels sometimes produced some postbranchial swelling. The mouth was then sewn shut around an irrigation input of PE 160 tubing attached to Tygon tubing (i.d. =4·25 mm). Finally, the dorsal edge of the opercular opening was sewn shut to ensure filling of the buccal cavity before the irrigate left the gill chamber. The head was fitted with a plastic collar using stainless-steel surgical staples and held inverted in a translucent plastic cylinder by a rubber dam (proximal end of a small balloon). Cannulation of the dorsal aorta is not possible in this preparation so efferent outflow of perfusate was collected as the total effluent from the cut end of the head. Perfusion was driven by a Gilson Minipuls 2 pump at constant flow (about 600μlmin⁻¹) which produced afferent pressures of 2·67–4·66 kPa, in the range of that measured in larger animals in vivo (J. B. Claiborne & D. H Evans, unpublished results). Perfusion pressures were monitored, via a T-connection to the perfusion line, by a Gould Statham pressure transducer and recorded on a Gilson Duograph to the nearest 0·13 kPa. The perfusion flows, as well as the pressure transducer, were calibrated before each experiment. All pressures were corrected for internal perfusion-line resistance, which was measured before each experiment. Perfusion pulsations produced by the Minipuls 2 pump were damped by a small cylinder in the perfusion line which was partially filled with perfusate (Windkessel). Irrigation flow was set at about 10 ml min⁻¹ (about 40 ml 100 g⁻¹ min⁻¹), with postbranchial irrigate draining directly out of the opercular opening, through a hole in the bottom of the perfusion chamber, and into a beaker through a small funnel. The irrigate (100 ml of sea water) was bubbled with air and recycled via a Cole-Parmer Masterflex pump. Perfusate was bubbled with 1% CO₂ in air provided from commercial, premixed gas cylinders. The composition of the Ringer’s solution is presented in Table 1. Unless otherwise stated, perfusates contained 10⁻⁷ moll⁻¹ epinephrine, which produced spontaneous and continuous ventilation in most preparations. Ventilatory rates were not quantified, but approximated the relatively low rates observed in vivo for this bottom-dwelling species.

Heads were perfused for about 30 min after preparation to allow for complete removal of blood from the branchial and cephalic vasculature and to achieve stable afferent perfusion pressures. Thus, time zero in most experiments with this preparation was at least 45 min after initial perfusion of the head.

To assess the integrity of this preparation, and therefore its usefulness for an investigation of the modes of ammonia transport across the branchial epithelium, we measured the surface micromorphology of gill tissue, the actual structural leak of perfusate at the end of approximately 2h of perfusion, and the transepithelial electrical potentials across the head.

To examine gill surface morphology, gill arches were removed from intact or perfused heads (after at least 2 h of perfusion) and fixed for at least 24 h in 2 % glutaraldehyde in 0·1 moll⁻¹ cacodylate-HCl buffer (pH 7·4) adjusted to the proper osmolarity with sucrose. Fixed tissues were rinsed in distilled water, dehydrated through a graded ethanol series, critical-point dried in liquid CO₂, mounted on a specimen stub, and coated with a gold-palladium alloy. Mounted tissues were then viewed and photographed at a specified magnification on a Hitachi model S-415A scanning electron microscope set at an acceleration voltage of 15 kV.

The measurement of actual structural leaks was monitored as previously described (Evans & More, 1988). At the end of each experimental measurement of ammonia effluxes, the heads were perfused for a final 20 min with Ringer’s solution containing 0·018% (2·3×10⁻⁴ moll⁻¹) erioglaucine (Acid Blue 9, molecular weight 783 Da; Sigma Chemical Co.). The amount of dye appearing in the irrigate during the efflux period was computed by comparing the absorbance (at 632 nm; Beckman model 34 spectrophotometer) of a sample of irrigate with a standard curve prepared by adding various amounts of dye to 100 ml of sea water.

The method described by Claiborne & Evans (1984) was used for measuring the TEP across the perfused head of O. beta. PE 100, 3 % agar-Ringer-filled bridges connected the irrigate and the perfusate in the open peritoneal cavity to calomel electrodes immersed in saturated KC1 solutions. The electrical PD between the electrodes (perfusate relative to irrigate) was measured on a Keithley model 616 digital electrometer. To provide good electrical contact on both sides of the branchial epithelium, the irrigate outflow from the perfusion chamber was blocked for a short period to allow the irrigation solution to fill the chamber to the levels of the gills. Both perfusion and irrigation pumps were then turned off for a few seconds, since stable TEPs could not be recorded during fluid flow. Asymmetries between the two bridge tips (sum of calomel electrode differentials and tip potentials in two different salt media) were monitored by placing the bridge tips in beakers filled with samples of irrigate or perfusate, connected via another agar-Ringer bridge. The asymmetries were usually less than 1 mV, stable over the course of the experiment, and always subtracted from the TEPs measured across the gills. TEPs were measured when the head was irrigated with sea water and after the irrigation solution had been changed to either Na⁺- or CE⁻ -free artificial seawater solutions (Table 1). To ensure removal of external Na⁺ or Cl⁻ in these determinations, the heads were irrigated for a few minutes with the appropriate ion-free solution before the perfusion chamber was partially filled with the postbranchial irrigate for the TEP measurement. Standard protocol was to measure TEPs sequentially (for approximately 1 min): in SW, Na⁺-free SW, SW, CU-free SW and finally SW. In parallel experiments, the same recording system was used to measure the TEPs across intact O. beta, with a PE 10, agar-Ringer bridge inserted through a 19 gauge needle into the peritoneal cavity and the other bridge in the external, seawater or ion-substituted bath. Animals were anesthetized (0·01 % MS-222 initially, 0·005 % during TEP measurement) for these experiments and transfers to Na⁺- or Cl⁻-free solutions were made with a small handnet. TEPs across the perfused heads were also monitored in a separate series of experiments. These mimicked the protocols outlined below for the ammonia efflux experiments in order to determine any changes in the TEP produced by these protocols which might have had an effect on the diffusion of NH₄⁺ across the perfused head.

Essentially the same protocol as described previously (Evans & More, 1988) for the perfused head of Squalus acanthias was used in the present study. Unless indicated, the perfusate in all experiments contained 300–400 μmoll⁻¹ NH₄C1. Initial time-control experiments involved the determination of the consistency of ammonia efflux over four 20-min efflux periods. All other efflux experiments involved an initial 20-min control efflux period, followed by another one or two 20-min efflux periods. 10-min wash periods, during which the perfusate and irrigate were changed, separated all efflux periods. The importance of branchial cell deamination in total ammonia efflux was determined by perfusing the head with ammonia-free, polyvinylpyrrolidone (PVP)-free Ringer’s solution, followed by a second efflux period with ammonia added to the perfusate. The putative role of basolateral Na⁺/NH₄⁺ exchange in ammonia efflux was measured by following a control efflux period with a second period in which the head was perfused with Ringer containing 10⁻⁴ moll⁻¹ ouabain. The potential involvement of basolateral Na⁺/NH₄⁺/2C1⁻ cotransport was monitored in another series of experiments by adding 5×10⁻⁵ moll⁻¹ bumetanide to the perfusate during a second efflux period. The potential role of an apical Na⁺/NH₄⁺ exchange was examined by adding 10⁻³ moll⁻¹ amiloride (see Evans & More, 1988) to the irrigate during a second efflux period. To investigate possible amiloride involvement with basolateral Na⁺/NH₄⁺ exchange, ouabain was added to the perfusate during a third efflux period in these experiments to inhibit basolateral Na⁺/K⁺-activated ATPase. To investigate this interaction further, in another series of experiments, amiloride was added to the irrigate during a third efflux period, with ouabain in the perfusate during the second and third efflux periods. Finally, to quantify the relative permeabilities to NH3 and NH₄⁺, control efflux periods were followed by about 10-fold increases in either the perfusate NH₃ or NH4⁺ concentrations, according to the method described previously (Goldstein et al. 1982; Evans & More, 1988). (NH₃ diffusion obviously will be down partial pressure, rather than concentration, gradients; however, for comparative purposes we have equated concentration with partial pressure, assuming constant solubility of the gas in the experimental solutions.) In all protocols the ammonia concentration in perfusates or irrigates from each experimental period was determined by the phenyl-hypo-chlorite method of Solorzano (1969). Similarly, the pH of perfusate and irrigate in each experiment was determined on a Radiometer PHM84 pH meter, with the electrode calibrated with buffer in saline solution to approximate more closely the ionic concentration of the experimental solutions. The actual NH₃ or NH4⁺ concentrations of experimental solutions were calculated by the Henderson-Hasselbalch equation, using a pK of 9·35 according to the equations of Cameron & Heisler (1983).

Adrenalin was manufactured by ESI Pharmaceuticals, amiloride and ouabain were obtained from Sigma Chemical Co., and bumetanide was supplied by Leo Pharmaceutical Products.

Experimental results are given as means ±S.E. (N) and statistical differences between experimental means were determined by Student’s t-test (using paired data when appropriate). Calculations of slopes for NH₃-stimulated and NH₄⁺-stimulated ammonia fluxes, and corrections for changes in the NH₃ gradient in the latter experiments were performed using Multiplan (Microsoft) on an Apple lie computer.

Photomicrographs of gill filaments, lamellae and interlamellar filamental surface cells selected from non-perfused and perfused heads of O. beta are shown in Fig. 1. It is clear that perfusion of the head for over 2h with teleost Ringer containing 3% PVP and 10⁻⁷ moll⁻¹ epinephrine did not result in any noticeable structural damage.

The erioglaucine leak measured in 35 experiments was only 0·32 ± 0·07 % in a 20-min period after over 3h of perfusion of the head preparation, indicating substantial structural integrity, even after this prolonged period of perfusion.

The TEP measured across the toadfish head in sea water was significantly more negative (about 5 mV) than that monitored in the intact fish in the present study (Fig. 2). Replacement of the irrigate Na⁺ or Cl⁻ produced hyperpolarization (about 5 mV) or depolarization (about 10mV), respectively, of the same magnitude as that seen in intact animals, suggesting similar relative Na⁺ and Cl⁻ permeabilities in vitro and in vivo. When the irrigate was returned to sea water after ionic substitutions in the perfused head, the TEP returned to initial levels, indicating that the substitutions themselves did not result in any significant alterations in the relative ionic permeabilities. In the intact animal, the TEP did not return to seawater levels after Cl⁻ substitution.

The perfused toadfish head maintained relatively consistent total ammonia efflux for periods up to 80 min (about 2h total perfusion) (Table 2). During the third, 20-min period the efflux appeared to decline, but it returned to control levels during the fourth period, so we assume that the decline was not physiologically important. The afferent perfusion pressure also remained consistent during the 80 min, indicating that the hemodynamics of the branchial vasculature had not deteriorated during that period.

When the toadfish head was perfused with ammonia-free Ringer’s solution, the total ammonia efflux was essentially zero. Addition of 300–400μmoll⁻¹ NH₄CI to the perfusate stimulated the efflux substantially (Fig. 3), indicating that, under the conditions of these experiments, branchial cell ammoniogenesis played a vanishingly small role in the excretion of ammonia.

When 10⁻⁴ moll⁻¹ ouabain was added to the perfusate, after a control 20-min efflux period, the total efflux of ammonia was reduced by 22% (17·8 ± 1·9μmol 100g⁻¹ h⁻¹vs 22·8 ± 2·1μmol 100g⁻¹ h⁻¹; P<0·01; N = 17). In these experiments, the afferent perfusion pressure increased slightly, but significantly (3·33 ± 0·13 vs 4·12 ± 0·13 kPa; P< 0·001) when the perfusate contained ouabain.

The addition of 5×10⁻⁵ moll⁻¹ bumetanide to the perfusate, after a control 20-min efflux period, did not significantly affect the total efflux of ammonia (38·1 ± 4·2 vs 35·9 ± 3·2μmol 100g⁻¹ h⁻¹; N = 6). The afferent perfusion pressure also was not altered in these experiments (3·19 ± 0·67 kPa in both cases).

When 10⁻³ moll⁻¹ amiloride was added to the irrigate, after a control 20-min efflux period, the ammonia efflux declined from 28·8 ±6·2 to 17·4 ± 4·4 μmol 100g⁻¹ h⁻¹ (P = 0·01; N = 5), suggesting that apical Na⁺/NH₄⁺ exchange plays a measurable role in ammonia transport across the toadfish gill. However, when 10⁻⁴ moll⁻¹ ouabain was added to the perfusate during a third period (with amiloride still present in the irrigate) the ammonia efflux did not change (16·1 ±2·6μ mol100g⁻¹h⁻¹, P>0·2). In these experiments, addition of neither amiloride nor ouabain had a significant effect on the afferent perfusion pressure (data not shown). When another series of heads was perfused with ouabain in the perfusate during an initial flux period, and amiloride was then added to the irrigate during the second period (ouabain still present), there was no significant change in the ammonia efflux (13·8 ±2·2 vs 9·2 ± 2·3μmol 100 g⁻¹ h⁻¹; P>0·l; N = 9). Addition of amiloride to the irrigate did not alter afferent perfusion pressures (data not shown), but addition of ouabain increased the pressures slightly (3·72 ± 0·27 vs 4·52 ± 0·27kPa, P < 0·01).

When the perfusate concentration of NH₃ was specifically increased by increasing perfusate pH from about 6·8 to 7·8 during two successive 20-min efflux periods, the total ammonia efflux increased significantly (Fig. 4), with a slope of 1·55 ± 0·291100g⁻¹h⁻¹ [(μmol100g⁻¹h⁻¹)/(μ moll⁻¹); N=11]. The afferent pressures were not altered by the pH changes in the perfusate in these experiments (data not shown).

Specifically increasing the perfusate NH₄⁺ concentration, by raising the NH₄CI content while lowering the pH, also stimulated the total ammonia efflux (Fig. 5), in this case with a slope of only 0·016 ± 0·0041100 g⁻¹h⁻¹ (N=13). Afferent pressures were not altered in these experiments (data not shown). To ensure that the apparent stimulation of ammonia efflux by increased basolateral NH₄⁺ concentration was not biased by stimulation of basolateral Na⁺/NH₄⁺ exchange, we repeated these experiments with 10⁻⁴ moll⁻¹ ouabain in the perfusate. In this case, the slope of the NH₄⁺-stimulated ammonia efflux was 0·012 ± 0·0041100 g⁻¹h⁻¹ (N = 26), not significantly different (P>0·4) from the previous experiment, indicating that basolateral Na⁺/NH₄⁺ exchange was not stimulated by raising perfusate NH₄⁺ concentrations. In these experiments, the afferent pressures were increased (P < 0·01) from 3·33 ±0·13 to 3·86±0·27kPa by the addition of ouabain to the perfusate.

Measurement of the TEP in a separate series of experiments, which combined the perfusate and irrigate manipulations outlined above, demonstrated that the alterations in the electrical potentials across the gills were minor (<2 mV) in all cases (data not shown).

Perfusion of the toadfish head for periods exceeding 2 h did not produce any obvious gill surface abnormalities (Fig. 1), in contrast to the obvious edema produced in the eel (Anguilla australis) holobranch (Ellis & Smith, 1983) or the rainbow trout (Salmo gairdneri) head (Part et al. 1982) perfused for much shorter Periods with adrenalin-free Ringer’s solution.

The extremely small leak (0·3%) in this preparation is even less than that described for the perfused dogfish shark pup head preparation (0·7 % ; Evans & More, 1988), and demonstrates directly that this preparation is structurally sound. If we assume that the actual gill permeability to erioglaucine is zero (taking the average perfusate ammonia concentration as approximately 400μ moll⁻¹, the perfusion flow as 650μ1 min⁻¹, and an animal mass of 30 g), an erioglaucine leak of 0·3% for 20 min would produce an apparent ammonia efflux of only 0·15 μmol 100 g⁻¹ h⁻¹, approximately 1 % of that measured in the ammonia experiments described below.

Afferent perfusion pressures, assumed to be proportional to branchial and cephalic resistance, were stable in this preparation for at least 80min (Table 2), indicating that, like the perfused dogfish pup head (Evans & More, 1988), the toadfish head is hemodynamically patent for the entire length of the experiments described here. The afferent pressures monitored in the present experiments were very similar to those measured, under control conditions, (2·79 and 4·12 kPa) in an earlier series of experiments on the toadfish perfused head (Claiborne et al. 1982).

The transepithelial electrical potential (–5 mV) measured across the toadfish head irrigated with sea water is significantly more negative than that measured across intact, anesthetized fish in a parallel series of experiments (Fig. 2), suggesting relatively higher Cl⁻ conductances in vitro. However, it is quite close to the seawater TEPs measured in the same species in other in vivo studies (–8 mV: Evans & Cooper, 1975; —6 mV: Kormanik & Evans, 1979; –9 mV: Kormanik & Evans, 1982). Irrigation with Na⁺-free artificial seawater solution hyperpolarized the TEP by some 5 mV in both the perfused head and intact fish. Slightly higher hyperpolarizations (about 7 mV) were observed in previous, in vivo studies (Evans & Cooper, 1975; Kormanik & Evans, 1982). Replacement of the irrigate Cl⁻ with benzenesulfonate depolarized the TEP across the perfused head and intact fish by some 10 mV, significantly less than the degree of depolarization (20 mV) caused by Cl⁻ substitution in an earlier in vivo study (Kormanik & Evans, 1979). It is important to note that we observed that the TEPs were more stable and more rapidly attained after ionic substitutions using the perfused head than in the intact fish in the present experiments. For example, after Cl⁻ substitution and transfer back into sea water, the TEP across the perfused head approached that measured previously in sea water, whereas that across the intact fish was extremely variable, slow to stabilize (data not shown) and 7 mV depolarized from the initial seawater TEP (Fig. 2). One might suggest, therefore, that placement of the internal bridge in the postbranchial perfusate of the perfused head provides a more accurate measurement of the true TEP than implantation of a bridge in the muscle mass or peritoneal cavity of an intact, anesthetized fish. Most importantly, the present data demonstrate that reproducible TEPs, of the same order as in vivo TEPs, can be measured across the perfused head of O. beta. Therefore, changes in passive ionic fluxes produced by alteration of electrochemical gradients can be monitored and factored out in any subsequent studies of ion transport by this preparation.

Taken together, these data show clearly that the perfused head of O. beta maintains structural, hemodynamic and electrical integrity, even after prolonged (1–2 h) perfusion, when compared with the intact fish.

The control efflux of ammonia from the perfused head in these experiments (Table 2) was somewhat below the 26μmol 100 g⁻¹ h⁻¹ described for the intact, normocapnie toadfish in one earlier experiment (Evans, 1982), but equal to that found in a similar study (15 μmol 100 g⁻¹ h⁻¹; Evans, 1977) in hypercapnic toadfish (11 μmol 100 g⁻¹ h⁻¹; Evans, 1982) and in an earlier study of ammonia transport across the perfused toadfish head (16μ mol 100 g⁻¹ h⁻¹; Claiborne et al. 1982). It is clear, therefore, that the preparation maintains near in vivo rates of ammonia transport. In addition, the relative consistency over the course of the current experiments (Table 2) supports, again, the patency of this preparation.

We found that ammoniogenesis by the branchial epithelium plays no measureable role in the ammonia excretion by the perfused toadfish head (Fig. 3). The quantitative importance of branchial cell ammoniogenesis in the total gill extrusion of ammonia by fishes has rarely been investigated. Pequin (1962) found that extraction of blood ammonia could account for all the ammonia excretion across the intact carp (Cyprinus carpio) and Goldstein et al. (1964) found similar results with the marine sculpin Myoxocephalus scorpius. However, Payan (1978) demonstrated that some 5 % of the ammonia excretion by the perfused trout head could be accounted for by branchial cell production, and Cameron & Heisler (1983) calculated similar levels using arteriovenous differences in intact trout.

The observation that addition of ouabain to the perfusate reduced the rate of ammonia efflux by 22 % corroborated our earlier study (Claiborne et al. 1982) of O. beta, which demonstrated that ammonia efflux from the perfused head was sensitive to perfusate ouabain and K⁺ concentrations, and suggests again that basolateral Na⁺/NH₄⁺ exchange plays a measureable role in transporting ammonia across the toadfish branchial epithelium. An alternative explanation of these results is that inhibition of basolateral Na⁺/K⁺-activated ATPase by ouabain indirectly inhibits a basolateral Na⁺/K⁺/2C1⁻ cotransporter. This bumetanide-sensitive carrier has been shown to be sensitive to NH₄⁺ (Kinne et al. 1986; O’Grady et al. 1987) and to play a role in NH₄⁺ reabsorption in the thick ascending limb of the rabbit nephron (Garvin et al. 1988; Good, 1988). Our earlier study of ammonia transport across the dogfish pup gill epithelium demonstrated that a small, but measureable, ammonia efflux took place via a bumetanide-sensitive system (Evans & More, 1988). However, in the present experiments, bumetanide addition to the perfusate did not inhibit ammonia efflux (see Results), consistent with the conclusion that the ouabain effect was directly on basolateral Na⁺/NH₄⁺ exchange.

The addition of ouabain, but not bumetanide, to the perfusate did produce vasoconstriction of the gill vasculature, increasing the afferent perfusion pressure by some 0·8kPa above the baseline of 3·33kPa (see Results). However, since a plot of the ammonia efflux against the afferent pressure of all the experiments not utilizing ouabain (perfusates with about 400 μmoll⁻¹ ammonia, pH about 7-8) did not demonstrate any significant correlation (Fig. 6), we assume that the slight increase in afferent pressure produced by ouabain in this and other experiments (see above) did not, itself, produce any change in the apparent ammonia efflux.

The addition of amiloride to the irrigate did significantly inhibit ammonia efflux, but the observation that subsequent addition of ouabain to the perfusate did not inhibit the ammonia efflux further (see Results) suggests that 10⁻³ moll⁻¹ amiloride on the apical surface may have inhibited basolateral Na⁺/K⁺-activated ATPase (Soltoff & Mandel, 1983), which provides the energy for basolateral Na⁺/NH₄⁺ exchange, rather than apical Na⁺/NH₄⁺ exchange. This proposition is supported by the results of our next protocol which demonstrated that amiloride did not inhibit ammonia efflux when ouabain was present in the perfusate (see Results). These data are consistent with the proposition that apical Na⁺/NH₄⁺ exchange does not play a role in the transport of ammonia across the toadfish gill. Our previous study of the perfused dogfish pup head also demonstrated that apical Na⁺/NH₄⁺ exchange was unimportant in the transport of ammonia across the gill epithelium in this species. Previous studies of intact S. acanthias (Evans, 1982) and O. beta (Evans, 1977) suggested the presence of apical Na⁺/NH₄⁺ exchange because, in both species, removal of external Na⁺ reduced ammonia efflux. However, this could have been due to cessation of apical Na⁺/H⁺ exchange (in both species the H⁺ efflux was reduced to at least zero), which would have affected the NH₃ gradient, and thereby the efflux of ammonia (Evans & Cameron, 1986). In addition, removal of external Na⁺ could have affected paracellular NH4⁺ conductance (Zadunaisky, 1984; Evans & Cameron, 1986).

The data in Figs 4 and 5 demonstrate that the ammonia efflux across the gill epithelium of the toadfish is stimulated by an increase in either the NH₃ or the NH₄⁺ concentration of the perfusate. The stimulation by both NH₄⁺ and NH₃ is somewhat at odds with earlier data (Goldstein et al. 1982) which appeared to demonstrate that only changes in perfusate NH4⁺ stimulated ammonia efflux across O. beta gills, suggesting a zero permeability to NH₃ However, in the earlier study, specific NH₃ gradients were not monitored, and it is unlikely that any epithelium is totally impermeable to NH₃. Our finding that ammonia can traverse the toadfish gill epithelium by ionic diffusion of NH₄⁺ confirms the earlier study (Goldstein et al. 1982), as well as our study of the dogfish pup gill (Evans & More, 1988), a recent study of the intact, marine teleost, Myoxocephalus octadecimspinosus (Claiborne & Evans, 1988) and studies of the turtle bladder (Schwartz & Tripolone, 1983) and the rabbit renal proximal straight tubule (Garvin et al. 1987).

The slopes in Figs 4 and 5 indicate that the toadfish gill’s permeability to NH₃ is nearly 100 times (97:1) the NH₄⁺ permeability. In our previous study we determined that the relative permeabilities were 1100:1 in the gill of S. acanthias (Evans & More, 1988). Using a gill surface area of 2·04cm²g⁻¹ (taken from data on a congeneric, O. tau, of similar mass; Hughes & Gray, 1972), we can calculate the apparent NH₃ and NH₄⁺ permeabilities of the toadfish gill epithelium and compare them with those published for other epithelia (Table 3). It is clear that the toadfish gill maintains relatively high permeabilities to both NH₃ and NH₄⁺, in the range of those described for the ‘leaky’ proximal tubule of the mammalian nephron, but much greater than those described for either turtle bladder or shark gill, both considered to be relatively ‘tight’ epithelia (Erlij & Martinez-Palomo, 1978; Evans, 1979). In particular, the relatively high permeability to NH₄⁺ of the toadfish gill correlates with the relatively high cationic permeability found in the gills of a variety of marine teleost fishes (Evans, 1979). In fact, the Na⁺ efflux from the toadfish gill is relatively low compared with that from the gills of other marine teleosts (Evans, 1979), suggesting that NH₄⁺ diffusion may be even more substantial across the gills of other marine teleosts. Since the gills of fresh- or brackish-water teleosts appear to have substantially lower ionic permeabilities (Evans, 1979), one could suggest that diffusion of NH₄⁺ might also be reduced in these species. In fact, our preliminary studies on perfused heads of O. beta acclimated to 5 % sea water have demonstrated that the slope of the NH₃-stimulated ammonia efflux is 3·67 ± 0·691100 g⁻¹ h⁻¹ (N = 6) and the slope of the NH₄⁺-stimulated ammonia efflux (with 10⁻⁴ moll⁻¹ ouabain in the perfusate to block putative basolateral exchanges) is 0·003 ± 0·0011100g⁻¹ h⁻¹ (N=6), suggesting that NH₃ permeability significantly increased, while NH₄^{+ -} permeability fell, concomitant with acclimation to reduced salinity. In fact, the ratio of permeabilities was 1200:1, close to that described for the dogfish shark pup (Evans & More, 1988), and the cationic permeability of elasmobranchs is of the same order as that of freshwater teleosts (Evans, 1979).

These data support the proposition that the gill epithelium of a marine teleost is significantly more permeable to NH₄⁺ than the gill of either a marine elasmobranch or the same teleost species acclimated to extremely low salinities. This increased permeability to the ionic form of ammonia may be correlated with the relatively higher cationic permeability of teleosts in sea water, compared with marine elasmobranchs or teleosts in reduced salinities (Evans, 1979).

Since separate measurements of the transepithelial electrical potential (during manipulations of the perfusate and/or irrigate as outlined above) indicated that changes in the TEP were minimal (see Results), we can utilize the data just described to calculate the relative importance of various putative pathways for ammonia transport across the toadfish gill epithelium. Assuming an ammonia pK of 9·35 for teleost Ringer’s solution at 24°C (Cameron & Heisler, 1983) and the pH of the perfusate (7-8), the concentration ratio of the two species of ammonia is 0·028:1·0 (NH₃:NH4⁺). Since the actual relative efflux is the product of the relative permeability and the relative concentration, the actual ratio of the effluxes of NH₃: NH₄⁺ across the gill is 2·7:1, despite the nearly 100-fold difference in their respective permeances across the epithelium. Given that 22% of the total ammonia efflux is ouabain-sensitive, 78% has to be via diffusion of NH₃ and NH4⁺. Fig. 7 summarizes these calculations and demonstrates that, despite the fact that over 50% of the total ammonia efflux is via non-ionic diffusion, ionic diffusion of NH₄⁺ and basolateral Na⁺/NH₄⁺ exchange play substantial roles in the transport of ammonia across the gill epithelium of the toadfish. In contrast, our earlier study of ammonia transport by the dogfish shark pup gill epithelium (Evans & More, 1988) demonstrated that basolateral Na⁺/NH₄⁺ exchange did not play a measureable role, basolateral Na⁺/NH₄⁺/2C1⁻ transport could account for 17% of the efflux, and diffusion of NH₃ was 12 times greater than NH4⁺ diffusion. These are the only direct, quantitative measurements of the role of various putative ammonia transport pathways across the fish gill epithelium. However, it is apparent that the importance of various pathways may vary with species, and that the relative roles of NH₃ and NH4⁺ diffusion may be correlated with the relative ionic permeabilities of the gill epithelium (Evans, 1979).

This research was supported by NSF PCM 8302621 to DHE. Figures were drawn by Daryl Harrison.