A New Analysis of Ammonia and Sodium Transport Through the Gills of the Freshwater Rainbow Trout (Salmo Gairdneri)

Freshwater teleosts compensate for the constant diffusional loss of ions by active branchial absorption of sodium and chloride from the hypo-osmotic environment. It has been shown that the mechanisms of Na⁺ and Cl⁻ uptake are independent, and Krogh (1938) proposed that Na⁺ is exchanged for NH₄⁺, which at the same time facilitates excretion of nitrogenous waste. According to Smith (1929), Cameron & Wood (1978) and Evans (1982), loss of ammonia occurs mainly through the gills. Further investigations have provided more precise information for determination of the physicochemical parameters such as the pK and solubility coefficient for ammonia in water and plasma at different temperatures (Kormanik & Cameron, 1981; Cameron & Heisler, 1983).

However, some controversy exists over the endogenous ion exchange in Na⁺ uptake (Maetz et al. 1976; Evans, 1985). Thus, in rainbow trout, Salmo gairdneri, Na⁺ uptake has been coupled with NH4⁺ excretion (Kerstetter & Keeler, 1976; Payan, 1978a,b), H⁺ excretion (Kerstetter et al. 1970; Cameron & Heisler, 1983; Heisler, 1984) or with both H⁺ and NH4⁺ excretion (Evans, 1985; Wright & Wood, 1985). Moreover, recent studies have led to a new appraisal of the relative NH3 and NH₄⁺ gradients and Na⁺ transport across fish gills (Wright & Wood, 1985; Evans & Cameron, 1986).

In the light of this recent work, the present investigation employed the isolated-perfused head of rainbow trout, a technique allowing constant control of both the internal and external components, to study Na⁺ absorption coupled with H⁺ or NH4⁺ excretion.

Rainbow trout (Salmo gairdneri) of either sex and weighing about 200 g were obtained from a local source (St Jeannet, Alpes-Maritimes, France). The fish were housed in large circular glass fibre tanks supplied with aerated city tap water (NaCl ≈ 160 μmoll⁻¹; Ca²⁺ ≈ l·6mmol I⁻¹; K⁺ ≈ 80μmoll⁻¹; pHμ8·0) at constant temperature (15°C); they were kept unfed for 1 week before the experiments.

The isolated-perfused head preparation was prepared as described by Payan & Matty (1975) and Girard & Payan (1977). This technique involved the setting up of a circulation in the isolated head allowing measurements of branchial water and ionic permeability. Before experimentation, a blood sample was taken from the caudal vein for chemical analysis.

The head of a previously heparinized trout was sectioned 1cm behind the pectoral fins, and the gills were then irrigated by a current of fresh water and temporarily perfused under constant pressure. The perfusion liquid had physicochemical parameters similar to those of the trout blood (Bornancin et al. 1985). The perfusion was performed at a constant flow rate (150 ml h⁻¹100 g⁻¹) chosen to avoid any excess pressure in the lamellae. The input pressure was constantly controlled by a recorder.

These conditions have been proved to prevent any oedema formation: this was regularly controlled by morphological examination of the gills at the end of the preparation. During experimentation, the head was perfused for a 10-min period before sample collection. Arterial, venous and external samples were collected at regular intervals. The experiments never exceeded lh.

Primary lamellar sodium influx values were calculated but not stated because no significant changes appeared during our experiments, and most Na⁺ absorption in fresh water (80%) has been shown to occur through the secondary lamellae (Gardaire et al. 1985; Avella et al. 1987).

The word ‘ammonia’ is used to refer to the total ammonia (T_Amm) in solution; particular chemical forms are referred to as NH₃ or NF₄⁺.

According to the experimental design, one of the following techniques was chosen.

1. In a closed external circuit, ammonia excretion (⁠ expressed in μmolh⁻¹100 g⁻¹) was measured by the external appearance of ammonia as follows:

where Δ [T_Amm]_e represents the difference in external ammonia concentrations (μmol I⁻¹) during the time (Δ t, expressed in min) between two sample collections; V_e is the volume of the external medium (≈0·1701); and W is the fish mass (g).

External ammonia concentration was determined by the manual method of Koroleff (1969).

Internal ammonia concentration was determined by automatic analyser (Technicon) with a dialysis membrane according to Logsdon (1960). Concentrated NH4CI solution was the source for all experiments involving ammonia addition, and actual ammonia measurements were undertaken immediately after experimentation.

Net hydrogen flux (⁠ in μequivh⁻¹100 g⁻¹) was determined from the variations of H⁺ concentration in the external medium as a function of time. Assuming that NH₃ was equilibrated with NH₄⁺ at the expense of H⁺ in the external medium, the sum of the titratable acidity measured plus ammonia excretion was termed the net apparent H⁺ efflux (Kirschner et al. 1973; Cameron, 1986; Vermette & Perry, 1987).

The H⁺ concentration of the external solution (buffered with 2 mmol I⁻¹ imidazole) was determined by titration with 2 mmol l⁻¹ NaOH and calculated from the quantity of base added to reach the starting pH. The samples were first aerated overnight, to eliminate respiratory CO₂ produced by the head, and measurements were then made on 2-ml samples with a radiometer autotitrator in conjunction with an autoburette unit and a titrigraph recorder.

Data shown in figures are means ± S.E. Student’s unpaired t-tests were used to compare most of the sample means. However, when appropriate, Student’s paired t-tests were used and are mentioned in the text.

A low external Na⁺ concentration of 200μmol l⁻¹ [approximate K_m of sodium absorption (Avella et al. 1987) and see Fig. 12] was employed so that the rate of Na⁺ influx could be readily enhanced by increasing the internal ammonia concentration, if a Na⁺/NH4⁺ exchange existed. When the internal ammonia concentration was gradually increased from 0 to 1 mmol I⁻¹ at constant pH (Fig. 1), no significant change in sodium absorption was observed. However, ammonia excretion increased as a function of internal ammonia concentration, showing a complete uncoupling between sodium and ammonia transport.

Basal ammonia excretion in the external medium was measured in the absence of ammonia in the perfusion fluid in an external closed circuit and was equal to 5·7 ± 0·7 μmolh⁻¹100 g⁻¹ (N = 4). Ammonia was produced by the branchial epithelium as a result of cellular metabolism of irrigated tissues. Glutamine, which is rich in NH₂ residues, would stimulate endogenous ammonia production, if added to the perfusate. Branchial cell uptake of glutamine is followed by an enzymatic deamination producing the nitrogenous waste, ammonia (Goldstein & Forster, 1961). After glutamine addition, basal ammonia excretion to the external! medium was stimulated by 80% within 6 min (Fig. 2). Because of the closed external circuit, external ammonia accumulated and the flux progressively decreased (see below). Excretion of ammonia in the efferent arterial fluid (basal excretion) was also observed.

Experiments with a closed external circuit (see Wright & Wood, 1985) showed that the decrease in ammonia excretion was proportional to the internal-external ammonia gradient (Fig. 3).

When the external medium was open (to prevent any ammonia accumulation which could artificially reduce the efflux), ammonia branchial clearance was measured under conditions of increasing internal [T_Amm], Ammonia excretion was directly proportional to the internal ammonia concentration within the range studied (100-900μmol l¹, Fig. 4). The limits chosen were physiological values observed in vivo (231·2 ± 33·6 μmol I⁻¹; N = 5). Addition of sodium (1 mmol I⁻¹) to the external medium stimulated ammonia excretion.

The buffer (Tris-H₂SO₄, pH 8·0; 50 mmol l⁻¹) was used to alter the environmental ability to donate protons. Tris buffer, with a pK of about 8·0, is a good proton donor to the NH₃/NH4⁺ system, which has a pK of around 9·3. After a control period of 20min, buffer was added to the external medium, causing an immediate response which resulted in a 50% increase in ammonia excretion (Fig. 5).

Acetazolamide, the carbonic anhydrase inhibitor which reduces intracellular proton production and causes an alkaline shift in the intracellular pH (Hersey & High, 1971), was introduced into the perfusion fluid, causing an inhibition of sodium influx (50%) within 15 min in the presence of 160μmoll⁻¹ internal ammonia (Fig. 6). The same phenomenon was apparent when ammonia was absent from the internal medium. Because sodium entry decreased both with or without ammonia in the perfusion fluid, the action of acetazolamide cannot be interpreted simply as an effect on Na⁺/NH4⁺ exchange.

According to Boron & De Weer (1976), addition of ammonia to the external medium is followed by a rapid cellular diffusion of NH₃, which binds H⁺, causing intracellular alkalinization. Ammonia removal, however, is immediately followed by intracellular acidification. This treatment was applied to the isolated-perfused head preparation (Fig. 7): 20 mmol I⁻¹ ammonia was added to the external medium for a 20-min period (phase b) and then quickly removed (phase c).

During ammonia exposure, sodium entry was inhibited by 40% (phase b). In this condition, a rapid (within 30s) alkalinization of the efferent arterial fluid was observed for each preparation exposed to external ammonia. The efferent arterial pH increased significantly from 7·50 ±0·04 to 7·80 ±0μ05 (N = 6) (Student’s paired t-test: Δ pH = 0·300 ± 0·016, P = 0·0001). As the perfusion fluid contained a coloured indicator, acidification was noticed immediately, and the pH measurements of the arterial effluents were performed with an electrode without delay on efferent samples. This result provides evidence for NH₃ diffusion through the branchial epithelium (cellular and/or paracellular pathways) and could argue for an intracellular alkalinization responsible for sodium influx inhibition.

Conversely, the removal of both external and internal ammonia, which could result in an intracellular acidosis, led to a stimulation of Na⁺ absorption (phase c).

The decrease of internal pH by lunit stimulated sodium uptake by 40% (Fig. 8). Moreover, in the same conditions, ammonia excretion was significantly inhibited by 30% and fell from 51 ± 3/μmol h⁻¹100 g⁻¹ at pH8·0 to 35 ± 5 μmol h⁻¹ 100g⁻¹ at pH7·0 (N = 3) (Student’s paired t-test: difference in rates of ammonia excretion = 15·6 ± 5·0 μmolh⁻¹100 g⁻¹, P = 0·004).

During external acidification, sodium and ammonia excretion were measured simultaneously (Fig. 9). To maintain the imposed pH, the external milieu was buffered, and to prevent ammonia accumulation the latter was open and continuously renewed. When external pH was decreased from 8-0 to 6-0, a marked stimulation (100%) of ammonia excretion was observed and sodium absorption was inhibited by 80%.

Sodium and ammonia transport had become completely ‘uncoupled’.

The diuretic amiloride is known to inhibit sodium entry and/or Na⁺/H⁺ exchange, according to the experimental concentration employed (Benos, 1982). It was added to the external medium to exert effects on the mucosal side of the epithelium. Amiloride (Merk Sharp & Dohme, Chibret) was used initially at 10⁻⁴moll⁻¹ to modify sodium and ammonia fluxes. After a control period, when sodium influx was stable, an immediate fall in sodium absorption was observed (Fig. 10). Sodium entry quickly stabilized and the percentage inhibition was calculated. In these conditions, ammonia excretion was stable: 20·0 ± 1·9 μmol h⁻¹100g⁻¹ for controls and 16·9 ± 2·6μmolh⁻¹100g⁻¹ (N = 4) during amiloride exposure (no significant difference: Student’s paired t-test: difference between rates of ammonia excretion = 3·0 ± 4·3 μmol h⁻¹100 g⁻¹, P = 0·26).

Different concentrations of the diuretic (range 10⁻⁶-10⁻³moll⁻¹) were also tested, producing an inhibition of sodium influx which was calculated as a percentage of the control flux (Fig. 11). The concentration required for 50% inhibition (IC₅₀) was 10⁻⁵moll⁻¹.

Further, sodium influx was measured as a function of the external sodium concentration (Fig. 12), in the presence (curve C) or absence (curve A) of external amiloride (10⁻⁴moll⁻¹). In contrast to the previous experiments, amiloride was present from the beginning of the experiment. Under these conditions, the remaining sodium influx increased linearly as a function of the external sodium concentration. This regression line (curve C) could be taken to represent the diffusional fraction of the total sodium absorption. When diffusional fluxes (curve Q were subtracted from total fluxes (curve A) a saturation curve was obtained (curve A’) and the kinetic parameters could be calculated: V_max = 42·8 ± 1·9 μmol h⁻¹100g⁻¹; = 305 ± 45 μmoll⁻¹ (N= 6).

Although experimental conditions were complex (there is an acidic mucus layer around the head and respiratory CO₂ is produced in the external medium), proton excretion was measured (Table 1). With ammonia (1 mmol I⁻¹) in the perfusion fluid, titratable acidity was very low, but in the absence of internal ammonia it increased. Thus, as defined in Materials and methods, total proton excretion can be considered as the sum of titratable acidity and ammonia excretion. These results (Table 1) confirm the presence of a substantial proton excretion, which could be correlated quantitatively with Na⁺ influx.

Total ammonia excretion would stop under our experimental conditions (of temperature and pH) if the difference in concentration between the internal and external compartment decreased to 50/zmoll⁻¹ (see possible extrapolation on Fig. 3). Based on the solubility of ammonia in various media (fresh water, plasma, etc.) and on the characteristics (i.e. pK) of the NH₃ – NH₄⁺ equilibria (Cameron & Heisler, 1983), the corresponding pNH₃, [NH₃] and [NHZ] values were calculated in another experimental situation in conditions where ammonia diffusion would stop (Table 2). These values indicate that excretion as a diffusion process should be totally inhibited when there is no difference in pNH₃ between internal and external media (internal pNH₃ = external pNH₃ = 46nmHg). In the present case, the calculated difference between the internal and external total ammona concentrations decreased to 60μmoll⁻¹ (0·200mmoll⁻¹–0·140 mmoll⁻¹) by the end of the experiment, when the external pH fell to 7·63. The predicted experimental value (observed in Fig. 3: 50μmoll⁻¹) and the calculated value (60 μmol I⁻¹) are remarkably close, which would argue for a diffusion of NH₃ as a function of the ΔpNH₃ between the internal and external media.

Although some authors have shown that gills are permeable to NH₄⁺ (Maetz, 1973; Goldstein et al. 1982; Evans, 1985), the uncharged form NH₃ is generally considered to be highly lipid-soluble and to diffuse rapidly through biological membranes. Non-ionic diffusion of free base NH₃ was first postulated as the mechanism of renal excretion of ammonia by Pitts (1948). Manning (1964) proposed that NH₃ penetrates cell membranes much more rapidly than does the ionic form. Many authors since then have studied NH₃ diffusion through the gill epithelium (Kirschner et al. 1973; Cameron & Heisler, 1983; Wright & Wood, 1985). Maetz (1973) reported that NH₃ diffusion occurred through the gill epithelium of goldfish. More recently, Evans (1984) showed the epithelium of elasmobranchs to be much more permeable to NH₃ than to NH₄⁺, the former diffusing 300 times faster than the latter.

It is clear, therefore, that ammonia excretion could take place via the diffusion of the free base NH₃ into the external medium, where it would bind hydrogen ions to form relatively nondiffusible ammonium ions. NH₃ diffusion could be stimulated by further binding and NH₃ trapping, which would occur in the external medium: (1) if H⁺ is excreted (addition of external Na⁺ possibly enhancing a Na⁺/H⁺ exchange as in Fig. 4) or (2) with increasing proton reserves in the external medium (addition of a buffer as in Fig. 5; external acidification in presence of a buffer as in Fig. 9). In the internal medium or in the cells, NH₃ is regenerated through continuous dissociation of NH4⁺ to NH₃ and H⁺. Excretion of total ammonia will continue as long as internal pNH₃ exceeds external pNH₃.

Cameron & Heisler (1983) calculated a value of 1·3 × 10⁻⁵cm⁻¹ for D, which they called the ‘gill permeability coefficient’, and compared this value to the coefficients of diffusion for O₂ and CO₂. However, the values should only be compared with great caution, because NH₃ diffusion rate is determined not only by the membrane permeability to the free base but also by the pH on both sides of the membrane, and by the flow rate of the perfusion fluid, which could modify the instantaneous amount of NH₃ available to be excreted.

When Na⁺ is added to the external medium (Fig. 4: [Na⁺] = 1 mmoll⁻¹; [T_Amm] = 900/xmoll⁻¹; ⁠), rises to 3·6 × 10⁻⁶ cm²s⁻¹ which could indicate that NH₃ diffusion responds to H⁺ external production and would be consistent with a sodium-proton correlation.

The permeability coefficient for NH₃, calculated as ⁠, is equal to 6 × 10⁻³ cm s⁻¹ and is intermediate between the values reported for turtle urinary bladder (3·0 × 10⁻⁴cms⁻¹; Schwartz & Tripolone, 1983) and renal tubules (in rat: l-0 × 10⁻²cms⁻¹, Oelert et al. 1968; in rabbit: l·6 × 10⁻²cms⁻¹, Garvin et al. 1987).

The possibility of NH₄⁺ excretion through simple diffusion cannot be ruled out (Evans & Cameron, 1986). However, the diffusion of ammonium ions does not need to occur to account for total ammonia excretion in the gill. Moreover, in particular conditions, active excretion of ammonia (possibly through a Na⁺/NH₄⁺ exchange) may take place either in a restricted environment rich in ammonia, to get rid of ammoniacal metabolic byproducts (Cameron & Heisler, 1983) or, after exhaustive exercise leading to extracellular acidosis, to contribute with Na⁺/H⁺ exchange to the regulation of acid-base equilibrium (Wood, 1988).

In general, the results described above lead us to conclude that Na⁺ influx and total ammonia excretion are not linked and the notion of Na⁺/NH4⁺ exchange has been rejected.

This was clear when the pH of the perfusion fluid was decreased by 1 unit (Fig. 8 and text): since the ratio [NH₃]/[NH4⁺] is equal to 10^pH-pK, the ratio must have decreased 10-fold. Total ammonia concentration being constant and equal to 1 mmol I⁻¹ in the internal medium, [NH₃] decreased from 28 to 3μmoll⁻¹ and [NH4⁺] increased slightly (2-5%) from 972 to 997/zmoll⁻¹. This acidification phenomenon will similarly affect the branchial cell; although the existence of intracellular buffers will give some attenuation, it should be followed by a trend towards intracellular acidification affecting the [NH₃]/[NH4⁺] ratio. From the Na⁺/NH4⁺ exchange hypothesis, even a small decrease in pH should slightly stimulate the exchange. However, this is not consistent with our observations that ammonia excretion decreased (30%: see text) when sodium influx increased markedly (Fig. 8) during internal acidification.

Earlier studies concluded that a Na⁺/NH4⁺ exchange existed in trout (Kerstetter & Keeler, 1976; Payan, 1978a,b). Payan (1978a,b), however, hypothesized that part of the Na⁺ influx was independent of internal [NEU^4-]. Even for a very low ammonia excretion rate (3·3 μmolh⁻¹100 g⁻¹) he still observed a significant sodium influx (20·8 μmolh⁻¹100 g⁻¹) and although amiloride inhibited Na⁺ influx, he observed no such effect on NH₄⁺ excretion. Hence, the stimulatory effect that an experimental increase in internal ammonia concentration had on sodium influx, as observed in trout (Kerstetter & Keeler, 1976; Payan, 1978a,b) and goldfish (Maetz & Garcia-Romeu, 1964), could instead be explained by the decrease in internal pH resulting from the internal addition of ammonia (Cameron & Kormanik, 1982; Cameron & Heisler, 1983).

An alternative hypothesis is required to explain our results. Indirect modulation of epithelial cell pH (Figs 6, 7, 8) brought adjustments in sodium influx that are consistent with the proposed Na⁺/H⁺exchange.

In Fig. 9, the inhibition of sodium absorption during external acidification could be explained as a competition of external H⁺ with external Na⁺ on the sodium site of the Na⁺/H⁺ exchanger (Paris & Pouyssegur, 1983). Then amiloride, which could totally inhibit sodium influx, can be assumed either to inhibit directly the apical Na⁺/H⁺ exchange or to decrease sodium conductance in the apical region of the cells.

The effects of amiloride externally, and acetazolamide internally, which can only occur directly on the branchial cells - as no erythrocytes are present in the perfusion fluid used for the acetazolamide effect - indicate that the exchange seems to take place on the external face of the gill epithelium.

The correlation between the magnitude of sodium influx and that of proton excretion (Table 1) adds considerable weight in favour of this exchange. When ammonia is present in the internal medium, the sum of titratable acid and NEU^4- excreted can be considered as the total efflux of protons, to counterbalance Na⁺ influx. However, in the absence of ammonia in the perfusion fluid, the gill epithelium of trout will excrete enough protons to counterbalance the influx of positive charge, entering as sodium in that case.

It has been suggested that Na⁺/H⁺ exchange occurs in trout both in perfused head preparations (Perry et al. 1985) and in vivo (Kerstetter et al. 1970; Kerstetter & Keeler, 1976; Cameron & Heisler, 1983). Results obtained in this study for H⁺ excretion are similar to those obtained by these authors and are close to the values obtained by Wood et al. (1984) for a range of internal pH conditions similar to those prevailing in our experiments.

This experimental preparation allows a precise study of ionic fluxes, but the structural complexity (Pisam et al. 1987) of, and limited access to, the gill epithelium hinder a more detailed description of the exchange mechanism.

Potentially, two mechanisms could account for our results.

1. A Na⁺/H⁺ exchange mediated by a carrier as is usually described in cells in contact with salt water, such as sea urchin eggs (Johnson et al. 1976), and for ‘leaky’ epithelia such as rat small intestine (Mürer et al. 1976), rabbit gall bladder (Cremashi et al. 1979) and renal proximal tubule (Kinsella & Aronson, 1980). Such carrier-mediated exchanges are electrically neutral, strictly coupled, and are not energy-consuming. Na⁺ entry along its electrochemical potential triggers the sodium-proton countertransport.

2. An indirect coupling of an active electrogenic proton transport and a passive sodium entry. Such a mechanism has been shown to operate in ‘tight’ epithelia in which the mucosal side is bathed by very dilute media (such as fresh water) and the serosal side by a liquid similar to plasma: urinary bladder of turtle and toad (Steinmetz et al. 1987; Ludens & Fanestil, 1972) and frog skin (Ehrenfeld & Garcia-Romeu, 1977; Ehrenfeld et al. 1985). In these epithelia, when the external medium is diluted, H⁺ excretion and Na⁺ influx are correlated, but proton excretion can also occur in situations of low Na⁺ influx (Ehrenfeld et al. 1985).

The gill epithelium of freshwater fish can be considered as being tight (Sardet, 1980; Pisam et al. 1987). In our experimental conditions, external sodium concentrations were very low (see freshwater composition in Materials and methods), always lower than or equal to 1 mmoll⁻¹. Thus, in terms of electrochemical potential, Na⁺ could not be considered as a possible trigger of a strict Na⁺/H⁺ exchange. Moreover, for the passive proton excretion to trigger the exchange, cell pH would have to decrease far below 7, which is incompatible with intracellular metabolism and is not consistent with reported values for intracellular pH. Therefore, we propose that sodium influx in trout gills is electrically coupled to an intracellular negative potential resulting from the active excretion of protons. In other words, an electrogenic proton pump participates in the genesis of a favourable electrochemical driving force for apical sodium entry.

Fig. 13 summarizes the possible roles of sodium, protons and ammonia in the branchial epithelium. The site for those exchanges through the branchial epithelium has not been determined. However, according to Avella et al. (1987), chloride cells on the secondary lamellae have been shown to play a role in branchial sodium influx and, thus, are the likely site for this transport (they were well-developed in control conditions), but NH₃ diffusion probably takes place across the whole gill surface.

This research was partially supported by the CNRS and by ‘Le Conseil Régional PACA’. We wish to thank the CEA for facilitating the purchase of radioactive products. The authors express their deepest gratitude to Drs N. Mayer-Gostan, G. De Renzis, J. Ehrenfeld and P. Payan for their continued interest and encouragement during the course of this study and for invaluable advice and critical appraisal of the manuscript. We are also grateful to Drs J. P. Bolton, J. D. Dutil, R. Balment and C. Ellory for help with the language.