Nacl Uptake by the Branchial Epithelium in Freshwater Teleost Fish: An Immunological Approach to Ion-Transport Protein Localization

In freshwater fish, the active uptake of Na⁺and Cl⁻is necessary for ionic homeostasis. The consequence of living in a hypo-osmotic medium and having a large surface area (the gill) subject to large outward ion gradients is the continuous loss of salts by passive diffusion and gain of water by osmosis. The fish manages the water gain by producing copious amounts of urine; however, this also adds to ion loss problems, because the kidney is not capable of reabsorbing all the salts. Active branchial ion uptake is crucial for ion homeostasis. The uptake of Na⁺and Cl⁻is achieved by electroneutral Na⁺/H⁺and Cl⁻/HCO₃⁻exchange mechanisms (Krogh, 1939).

Na⁺uptake is mediated by an amiloride-sensitive Na⁺/H⁺exchange that is either directly or indirectly coupled and shows saturation kinetics (Wright, 1991; Potts, 1994). It is difficult to explain Na⁺uptake using the directly coupled carrier-mediated Na⁺/H⁺exchange (NHE) mechanism because of the absence of physiologically relevant gradients to drive the exchange process under many natural conditions (for a review, see Lin and Randall, 1995). The Na⁺levels in the epithelium (6.4–16.5 mmol l⁻¹in isolated pavement cells and MR cells; Li et al., 1997) are typically greater than in the freshwater environment (<1 mmol l⁻¹), where the Na⁺chemical gradient is insufficient to drive Na⁺/H⁺exchange (Avella and Bornancin, 1989). Wright (1991) calculated that a pH gradient of 0.3 units would be required to drive Na⁺uptake via a Na⁺/H⁺exchanger, but acid excretion and Na⁺uptake have been measured at water pH values well below that of the epithelium (pHi 7.4; Wood, 1991). However, a Na⁺/H⁺exchanger could explain Na⁺uptake by a fish in water of 0.5 mmol l⁻¹[Na⁺], pH 8.0 (Wright, 1991). The amiloride concentration typically used (10⁻⁴mol l⁻¹) does not distinguish between the Na⁺/H⁺exchanger and epithelial Na⁺channel (ENaC) routes of Na⁺uptake (see, for example, Wright and Wood, 1985; Clarke and Potts, 1998b). The Na⁺/H⁺-exchange-specific inhibitor, amiloride analogue 5-(N,N-dimethyl)-amiloride (DMA) had no effect on Na⁺uptake rates in isolated gill filament preparations (Clarke and Potts, 1998a).

The alternative model employs an apical electrogenic vacuolar-type H⁺-ATPase (V-ATPase) and an amiloride-sensitive Na⁺channel. This indirectly coupled exchange has been proposed for frog skin (Harvey and Ehrenfeld, 1986) and turtle bladder (Stetson and Steinmetz, 1985). The frog skin/turtle bladder model for Na⁺uptake has also been proposed by Avella and Bornancin (1989) to operate in freshwater fish. This model has been further substantiated by the work of Lin and Randall (1995). The basis for the mechanism is the apical V-ATPase, which pumps protons across the apical membrane, creating a membrane potential capable of driving passive Na⁺influx via an amiloride-sensitive channel against its concentration gradient. Recently, Fenwick et al. (1999) have been able to demonstrate that inhibition of the V-ATPase by bafilomycin A1 reduces the rate of Na⁺uptake. The protons are provided from CO₂ hydration catalyzed by intracellular carbonic anhydrase (Lin and Randall, 1991). The accumulated HCO₃⁻is thought to exit the cell via a basolateral Cl⁻/HCO₃⁻exchanger with an associated Cl⁻channel; however, experimental evidence is lacking (Wright, 1991). The movement of Na⁺across the basolateral membrane is facilitated by the Na⁺/K⁺-ATPase (Richards and Fromm, 1970; Payan et al., 1975).

The immunolocalization of V-ATPase to the apical membrane of pavement cells (Lin et al., 1994; Sullivan et al., 1995) and correlative data from X-ray microanalysis (Morgan et al., 1994; Morgan and Potts, 1995) and morphometric ion-flux studies (for a review, see Goss et al., 1995) suggest that Na⁺uptake is performed by pavement cells. However, Lin et al. (1994) also reported apical V-ATPase labelling in MR cells. In addition, high levels of Na⁺/K⁺-ATPase activity were associated with the MR cell, and the amounts in pavement cells were below the level of detection (Witters et al., 1996). Apart from the information that an amiloride-sensitive Na⁺channel exists on the apical surface, the localization of Na⁺channels to the pavement cell or MR cell apical membrane has yet to be demonstrated.

Evidence for an epithelial Cl⁻/HCO₃⁻exchanger in fish gills comes from kinetic, pharmacological and correlative morphological studies (for a review, see Goss et al., 1995). Cl⁻influx has been shown to be stimulated by infusion of HCO₃⁻(Kerstetter and Kirschner, 1972), and there is a good correlation between the rates of Cl⁻uptake and base secretion (de Renzis and Maetz, 1973). More recently, analysis using two-substrate enzyme kinetics has provided further evidence for a 1:1 Cl⁻/HCO₃⁻exchange mechanism (Wood and Goss, 1990). The use of the anion transport inhibitor 4-acetamido-4′-isothiocynato-stilbene-2,2′-disulphonic acid (SITS) in the external bathing medium resulted in a reduction in the rate of Cl⁻influx after both short-term (66 %; Perry et al., 1981) and longer-term (71 %; Perry and Randall, 1981) treatments. Interestingly, no effect of short-term exposure to SITS was seen on net proton efflux (Lin and Randall, 1991). No studies on a basolateral SITS-sensitive anion exchanger have been conducted. However, apically applied thiocyanate (SCN⁻; a non-specific anion transport inhibitor) has been shown to inhibit Cl⁻uptake, while SCN⁻injection resulted in no change in the rate of Cl⁻influx (de Renzis, 1975).

In the gill of freshwater fish, Cl⁻uptake is thought to occur through the MR cells. Evidence for this location is based on correlative studies of MR cell fractional surface area and Cl⁻fluxes (for a review, see Goss et al., 1995). In addition, various disturbances resulting in changes in intracellular ion concentrations, measured using X-ray microanalysis, tend to support this idea of a freshwater MR chloride cell (Morgan et al., 1994; Morgan and Potts, 1995).

In this paper, we demonstrate the presence and branchial distribution of the amiloride-sensitive Na⁺channel, the Na⁺/H⁺exchanger and the SITS-sensitive anion exchanger immunologically through the use of non-homologous antibodies. If the V-ATPase/epithelial Na⁺channel model is to be valid, then co-localization must be demonstrated. Information on the cell types involved is provided by the specific labelling of MR cells by a Na⁺/K⁺-ATPase antibody (Witters et al., 1996). The specificities of the antibody probes used are verified by western analysis. Data are presented on two widely studied freshwater species of fish, the rainbow trout (Oncorhynchus mykiss) and the tilapia (Oreochromis mossambicus).

Four adult tilapia Oreochromis mossambicus Peters were obtained from the SkekKipMei fish market, Kowloon, Hong Kong. The fish were kept in glass aquaria within a recirculation system and fed commercial fish food. The fish weighed approximately 500 g and were kept for less than a week. The water temperature was 26 °C, and lighting conditions were not controlled. Water composition was (in mmol l⁻¹): Na⁺0.435, Cl⁻0.457, Ca²⁺0.395, pH 8.1.

Rainbow trout Oncorhynchus mykiss (Walbaum) were obtained from local suppliers in the Vancouver and Ottawa areas (Canada) and maintained under local conditions. Ottawa water contained (in mmol l⁻¹) Na⁺0.15, Cl⁻0.15, Ca²⁺0.45, pH 7.5–7.7; Vancouver water contained (in mmol l⁻¹) Na⁺0.02, Cl⁻0.01, Ca²⁺0.03, pH 5.8–6.4. Fish were fed commercial trout chow and kept under a natural photoperiod.

Fish were quickly netted and killed by a blow to the head or an overdose of anaesthetic (MS-222, Syndel). The second gill arch from both the left and right sides was excised to be fixed (see below) or freeze-clamped (liquid nitrogen) and stored frozen (−70 °C) for later western analysis and immunohistochemistry (unfixed/frozen preparations). Some fish had their gills perfused for 5 min with ice-cold heparinized Cortland’s saline via the cannulated bulbous arteriosis, to remove blood, prior to freeze-clamping.

For investigations by light microscopy, excised gill arches were fixed in 3 % paraformaldehyde (PFA)/phosphate-buffered saline (PBS), pH 7.4, or Bouin’s solution (24 % formalin, 5 % acetic acid, 71 % saturated aqueous picric acid, pH 2.2) for 2 h at room temperature (22–26 °C) or overnight at 4 °C. Following fixation, tissues were rinsed in PBS and 10 % sucrose/PBS and either frozen in liquid nitrogen or processed for paraffin embedding (fixed/paraffin-embedded preparations) (Paraplast, Fisher Scientific).

For immunoelectron microscopy, smaller pieces of tissue (pairs of filaments) were fixed by a two-step 3 % PFA/20 mmol l⁻¹ethylacetimidate/PBS and 3 % PFA, 0.1 % glutaraldehyde, PBS fixation procedure (Tokuyasu and Singer, 1976). Following fixation for 1–2 h at room temperature, the tissue was rinsed in PBS and free aldehyde groups were quenched in a 50 mmol l⁻¹NH₄Cl solution for 20 min. The tissue was then dehydrated through a decreasing temperature (room temperature to −20 °C) ethanol series (1 h in 30 %, 50 %, 70 %, 95 % ethanol and three times for 1 h in 100 % ethanol), gradually infiltrated with resin (ethanol:Unicryl; 2:1, 1:1, 1:2 for 30 min at −20 °C and 100 % Unicryl twice for 1 h and overnight at −20 °C) and transferred to Beem caps (BioCell Intl., UK) for embedding (Scala et al., 1992). The resin was polymerized under ultraviolet light at −10 °C over 3 days.

Cryosections (5–10 μm) were cut on a cryostat at −20 °C, collected onto either 0.01 % poly-L-lysine-coated (Sigma) slides or electrostatically charged slides (SuperFrost Plus, Fisher Scientific) and fixed in acetone at −20 °C for 5 min. Sections were then air-dried.

Paraffin sections (5 μm) were collected onto charged slides, air-dried at 37 °C overnight, and dewaxed though a series of xylene baths and rehydrated through an ethanol series finishing in PBS. Sections were circled with a hydrophobic barrier (DakoPen, Dako DK), blocked with 5 % normal goat serum (NGS)/0.1 % bovine serum albumin (BSA)/0.05 % Tween-20 in PBS (TPBS), pH 7.4, for 20 min and incubated with primary antibody diluted 1:50 to 1:200 in 1 % NGS/0.1 % BSA/TPBS, pH 7.3, for 1–2 h at 37 °C. Sections were then rinsed in 0.1 % BSA/TPBS or PBS followed by incubation with secondary antibody conjugated to fluorescein isothiocyanate (1:50; FITC; Chemicon International Inc.), Texas Red (1:100, Molecular Probes or Jackson ImmunoLab) or Cy3 (1:200, Sigma) for 1 h at 37 °C. Following rinsing with 0.1 % BSA/TPBS, sections were mounted with glycerol/PBS or VectaShield and viewed on a Zeiss AxioPhot photomicroscope with the appropriate filter sets.

Normal rabbit and normal mouse serum (Sigma) were routinely used as negative controls to account for non-specific background labelling. Non-specific labelling of the secondary antibody conjugates and tissue autofluorescence were accounted for by substituting buffer for the primary and buffer for both the primary and secondary antibodies, respectively, in the above protocol.

An antigen retrieval technique was employed to enhance the immunoreactivity of tissue sections with the various antisera used (see below) (Brown et al., 1996). This technique exposes epitopes masked during tissue fixation or causes the refolding of proteins in such a way that they possess epitopes not normally present. The SDS pre-treatment was conducted on sections mounted on charged slides only. Following dewaxing of paraffin sections or air-drying of cryosections, the sections were circled with a hydrophobic marker and flooded with 1 % SDS in PBS, pH 7.3, for 5 min at room temperature. The sections were then rinsed in a stream of PBS and put through a series of PBS baths (Coplin jars; 5, 10, 15 min) with gentle agitation. The sections tended to be very hydrophobic, so care had to be taken when applying the blocking buffer to make sure that the sections did not dry. The standard immunolabelling protocol was then followed (above).

Ultrathin sections were prepared on a Reichert ultramicrotome and collected onto either Formvar or Formvar/carbon-coated nickel grids. Following air-drying, sections were rehydrated by floating grids on drops of PBS. Grids were then transferred to drops of diluted primary antiserum (1:100) and incubated at room temperature for 1 h. They were then rinsed and transferred to drops of diluted secondary antiserum (1:100) conjugated to colloidal gold (10 or 20 nm diameter, Sigma, Chemicon) and incubated for 1 h at room temperature. They were rinsed with PBS and fixed for 10 min in 1 % glutaraldehyde/PBS and rinsed again in double-distilled water before counter-staining with lead citrate and saturated uranyl acetate. Sections were viewed on a Philips 300 transmission electron microscope and photographed with Kodak EM plate film 4489.

Tissue for Western analysis was usually prepared using the crude membrane method of Zaugg (1982). Briefly, gill tissue was scraped with a microscope slide into ice-cold SEI buffer (0.3 mol l⁻¹sucrose, 0.02 mol l⁻¹EDTA, 0.1 mol l⁻¹imidazole, pH 7.3). Tissue was then homogenized using a Potter–Elvehjem tissue grinder. The homogenate was centrifuged at 2000 g for 10 min, the supernatant was discarded and the pellet resuspended in 2.4 mmol l⁻¹deoxycholate acid in SEI buffer using the homogenizer. The homogenate was again centrifuged at 2000 g for 10 min, and the supernatant was saved. Total protein was measured using the method of Bradford (1976) and a BSA standard.

A more refined method was also used for preparing membrane proteins for western analysis. Saline-perfused gill tissue was prepared by differential centrifugation using a protocol adapted from those of Flik and Verbost (1994) and Dubinsky and Monti (1986). The gills were perfused with cold heparinized modified Cortland’s saline for 5 min, and the gill arches were excised. The tissue was scraped with a glass microscope slide onto an ice-cold piece of glass (approximately 2.5 g of material from a 150 g fish). The scrapings were then suspended in erythrocyte lysis buffer (nine parts 0.17 mol l⁻¹NH₄Cl, one part 0.17 mol l⁻¹Tris-HCl, pH 7.4, 10 μg ml⁻¹aprotinin; 40 ml for approximately 2.5 g of scraped tissue) for 20 min at room temperature. Intact branchial cells and pieces of tissue were collected by low-speed centrifugation (200 g for 10 min at 4 °C), and the supernatant was discarded. The pellet was resuspended in 20–40 ml of hypotonic buffer (25 mmol l⁻¹NaCl, 1 mmol l⁻¹Hepes-Tris, pH 8.0, 1.0 mmol l⁻¹dithiothreitol, 10 μg ml⁻¹aprotinin) and homogenized using a loose-fitting Dounce homogenizer (20–40 strokes). A low-speed centrifugation followed to remove nuclei and cellular debris (550 g for 10 min at 4 °C). The supernatant was decanted and centrifuged at 13 000 g for 10 min at 4 °C. The mitochondrial pellet was discarded. The supernatant was decanted and centrifuged at 33 000 g for 45 min at 4 °C, and the membrane pellet was saved. The resulting pellet was resuspended with a 23 gauge needle in 0.5–1.0 ml of suspension buffer (100 mmol l⁻¹mannitol, 5 mmol l⁻¹Hepes, pH 7.6, 10 μg ml⁻¹aprotinin) and either frozen or layered on a sucrose step gradient (5 % to 25 % sucrose, 1 mol l⁻¹KBr, 10 mmol l⁻¹Hepes, pH 7.4).

An attempt was made to isolate apical membranes from basolateral membrane fractions using a sucrose gradient. The membrane suspension was layered on top of the sucrose step gradient (25, 20, 15, 10 and 5 % sucrose) and centrifuged at 100 000 g for 120 min at 4 °C. The layers were removed by puncturing the side of the tube with a 23 gauge needle and 10 ml syringe. Membrane fractions were resuspended in 5 ml of suspension buffer and centrifuged at 150 000 g for 15 min. Pellets were resuspended in 200 μl of suspension buffer, and 50 μl samples were taken for total protein measurement; the remainder was frozen (−70 °C). Total protein was measured using a modified Bradford (1976) method and a BSA standard (Simpson and Sonne, 1982).

Gill homogenates were diluted to 1 μg μl⁻¹in Laemmli’s buffer (Laemmli, 1970). Proteins were separated by polyacrylamide gel electrophoresis (PAGE) under denaturing conditions, as described by Laemmli (1970), using a vertical mini-slab apparatus (Bio-Rad, Richmond, CA, USA). Proteins were transferred to either Immobilon-P (Millipore) or nitrocellulose membranes using a semi-dry transfer apparatus (Bio-Rad). Blots were then blocked in 3 % skim milk/TTBS (0.05 % Tween 20 in Tris-buffered saline: 20 mmol l⁻¹Tris-HCl, 500 mmol l⁻¹NaCl, 5 mmol l⁻¹KCl, pH 7.5).

Blots were incubated with primary antiserum diluted (1:250 to 1:1000) in TTBS for 1 h at room temperature or overnight at 4 °C with agitation. Following a series of washes with TTBS, blots were incubated with either goat anti-rabbit or anti-mouse antibody conjugated to horseradish peroxidase or alkaline phosphatase (Sigma). Bands were visualized by enhanced chemiluminescence with horseradish peroxidase (Amersham) or 5-bromo-4-chloro-3-indolyl phosphate/Nitroblue Tetrazolium reaction with alkaline phosphatase. Normal rabbit serum and normal mouse serum were substituted for primary antibodies (see above) to assess non-specific immunoreactivity.

The V-ATPase was immunolocalized using a rabbit polyclonal antibody raised against a synthetic peptide corresponding to a sequence from the catalytic 70 kDa A-subunit of the bovine V-type H⁺-ATPase complex (CSHITGGDIYGIVNEN; Südhof et al., 1989) (Protein Service Laboratory, University of British Columbia, Canada) conjugated to keyhole limpet haemocyanin using a maleimide linker (Pierce). A pre-immunization serum was collected for use as a control. Sera were tested by peptide enzyme-linked immunosorbent assay (ELISA) and western blot analysis of gill homogenates.

A polyclonal antibody raised against the same peptide (Südhof et al., 1989) has been used to identify the distribution of the V-ATPase A-subunit in mammalian kidney (Madsen et al., 1991; Kim et al., 1992) and rainbow trout gill (Lin et al., 1994). Using rat kidney as a positive control tissue, we found identical results to those published above.

A rabbit polyclonal antibody raised a peptide sequence from the E-subunit of the V-ATPase was kindly donated by S. F. Perry and J. N. Fryer (University of Ottawa). This is the same antibody used by Sullivan et al. (1995). We were, however, unable to reproduce the results of Sullivan et al. (1995) or Perry and Fryer (1997) on rainbow trout with the sample of antiserum provided.

The amiloride-sensitive Na⁺channel was immunolocalized using a rabbit polyclonal antibody raised against a synthetic peptide corresponding to amino acid residues 411–420 of the β-subunit of the human epithelial Na⁺channel clone (βhENaC; CGEKYCNNRDF; D. J. Benos, unpublished results). The IgG used was purified on a protein A column (0.8 mg ml⁻¹in glycine/Tris buffer, pH 7.5). The α-subunit bovine ENaC (αbENaC) rabbit polyclonal antibody was raised against a full-length fusion protein (Ismailov et al., 1996) generated from the cDNA αbENaC clone (Fuller et al., 1995). The IgG was purified from whole serum on a protein A column (1.4 mg ml⁻¹glycine/Tris buffer, pH 7.8). A rabbit polyclonal antibody was raised against biochemically purified bovine renal papilla amiloride-sensitive epithelial Na⁺channel complex (αENaC; Sorscher et al., 1988). The IgG fraction was purified from whole serum on a protein A column (1.48 mg ml⁻¹glycine/Tris buffer, pH 7.4). This antibody has been used in a number of different studies to localize Na⁺ channels sensitive to both low and high concentrations of amiloride in a number of different tissues (e.g. intestine, Smith et al., 1993; kidney, Brown et al., 1989).

Gill Na⁺/K⁺-ATPase was immunolocalized using a monoclonal antibody specific for the α-subunit of chicken Na⁺/K⁺-ATPase (Takeyasu et al., 1988). The antibody (α5) developed by D. M. Fambrough (Johns Hopkins University, MD, USA) was obtained from the Developmental Studies Hybridoma Bank maintained by the University of Iowa Department of Biological Sciences, Iowa City, IA 52242, USA, under contract NO1-HD-7-3263 from the National Institute of Child Health and Development (NICHD). The antibody was purchased as culture supernatant (0.9 mg ml⁻¹). This antibody is now in routine use for identifying gill MR cells (Witters et al., 1996; T. H. Lee et al., 1998) and, in cultured cells, Witters et al. (1996) have been able to demonstrate that Na⁺/K⁺-ATPase-positive cells (using the same α5 antibody) corresponded to sites of cells labelled with the mitochondrial marker dimethylaminostyrylmethylpyridinium iodine (DASPMI).

Polyclonal antibodies were generated against rainbow trout erythroid band 3 protein (AE1) purified by SDS–polyacrylamide electrophoresis (under reducing conditions) (Cameron et al., 1996). Using western analysis, this antibody has been shown to cross-react with trout and lamprey erythrocyte preparations (Cameron et al., 1996).

We also attempted to use a rabbit polyclonal antibody generated against the native 43 kDa fragment cytoplasmic domain of human erythroid band 3 (Verlander et al., 1988; Philip Low Department of Chemistry, Purdue University, West Lafayette, IN, USA). This antibody has been used to immunolocalize the Cl⁻/HCO₃⁻anion exchanger in the basolateral membrane of collecting duct α-type intercalated cells (Verlander et al., 1988; Drenckhahn et al., 1987; Madsen et al., 1991; Kim et al., 1992). We also tested a monoclonal antibody generated against a synthetic peptide (NRSLAGQSGQGKPR) corresponding to amino acid residues 871–884 in the deduced primary structure for human kidney AE2 (Martínez-Ansó et al., 1994; Eduardo Martínez-Ansó, Department of Medicine and Liver Unit, University Clinic and Medical School, University of Navara, Pamplona, Spain). This amino acid sequence corresponds to the Z-loop, characteristic of the non-erythroid Cl⁻/HCO₃⁻anion exchanger. This antibody has been shown to cross-react with the luminal membrane of shark rectal gland tissue (George et al., 1998). We were, however, unable to obtain cross-reactivity with teleost fish gill tissue either by western analysis or by immunohistochemistry with either the human AE1 or AE2 antibodies. Rat kidney and human erythrocytes were used as positive controls.

Rabbit polyclonal antibodies generated against glutathione S-transferase fusion proteins incorporating the C-terminal 87 amino acid residues of NHE2 (antibody 597; Tse et al., 1994) and the C-terminal 85 amino acid residues of NHE3 (antibodies 1380 and 1381; Hoogerwerf et al., 1996) were used. These antibody have been used in a number of different studies for immunolocalization and western analysis of NHE2 and NHE3 (He et al., 1997; M. G. Lee et al., 1998; Levine et al., 1993; Sun et al., 1997) including NHE3 using antibody 1381 in two species of teleost fish (O. mykiss and Pseudolabrus tetrious; Edwards et al., 1999).

Three monoclonal antibodies generated against a maltose-binding protein fusion protein that contained the C-terminal 131 amino acid residues of NHE3 were also tested (Biemesderfer et al., 1997). We were unable to demonstrate cross-reactivity of gill tissue by either western analysis or immunohistochemistry with these commercial monoclonal antibodies (Chemicon International, CA, USA).

The rabbit polyclonal anti-peptide antibody to the A-subunit of the V-ATPase cross-reacts with a population of squamous epithelial cells covering the lamellae and filament on the upstream or efferent (leading) side of the gill (Fig. 1A). There is an absence of immunoreactive epithelial cells towards the downstream or afferent (trailing) side of the filament (Fig. 1B). Pillar cells, erythrocytes and mucocytes were not labelled. In western blots, bands in the 70–80 kDa range are recognized (Fig. 1E).

The distribution of Na⁺/K⁺-ATPase, determined using the mouse monoclonal antibody to the α-subunit of the Na⁺/K⁺-ATPase, is mainly restricted to a population of cells concentrated on the afferent side of the filament (Figs 1D, 2C; see Fig. 6A). These cells are frequently found on the trailing edge and in the interlamellar spaces of the filament epithelium. They also are found on the lamellae, although generally towards the base. There are clearly fewer immunoreactive cells towards the efferent side of the filament (Fig. 1C). Immunopositive cells are strongly labelled, ovoid or cuboidal in appearance and sometimes associated with an apical crypt. Nuclei can also be made out as a negative image against the cytoplasmic labelling. Mucocytes that have a similar shape to MR cells and that also appear on the trailing edge of the filament epithelium show no labelling (see Fig. 6A). Control labelling with normal mouse serum and buffer also produces negligible levels of fluorescence (Fig. 2D). In western blots, a 116 kDa band and a weaker 97 kDa band are recognized (Fig. 1F).

The polyclonal antibody generated against trout erythrocyte AE1 (AE1t) cross-reacts strongly with tilapia erythrocytes regardless of the fixation conditions used (Fig. 2A). However, to achieve cross-reactivity with epitopes in the epithelium, a pre-treatment of sections with SDS is required. In SDS-treated sections, the antibody cross-reacts with the apical region of cells that are also strongly immunoreactive for Na⁺/K⁺-ATPase (Fig. 2C). There are, however, Na⁺/K⁺-ATPase-immunoreactive cells not associated with apical AE1t. The number of double-labelled cells is greatest in the afferent region of the filament epithelium. Control incubations of sections with normal rabbit serum result in negligible levels of fluorescence (Fig. 2B). In western blots, the antibody cross-reacts with a 110 kDa band from saline-perfused gill tissue as well as erythocytes (Fig. 3). An additional band at approximately 70 kDa is the result of antibody cross-reactivity with a contaminant in the electrophoresis set-up and is probably of microbial origin (Marshall and Williams, 1984).

In the tilapia, the rabbit polyclonal antibody raised against the β-subunit of the human epithelial Na⁺channel specifically labels a population of squamous cells in the epithelium on the efferent side of the filament (Fig. 4A). Both lamellar and filament epithelial cells stain in this region. The labelled cells are in the same location as those that were positive for the V-ATPase but do not extend as far towards the afferent side of the filament (Fig. 4A). Pillar cells and erythrocytes show no immunoreactivity. Normal rabbit serum IgG (Fig. 4B,C) and buffer control sections were also negative. Western blots of tissue homogenates recognize a band at approximately 98 kDa (Fig. 4D).

We were unable successfully to use either the αbENaC or biochemically purified amiloride-sensitive Na⁺channel antibodies for immunohistochemical localization of the epithelial Na⁺channel. This was despite trying a number of antigen retrieval techniques and fixation protocols. However, in western blots, cross-reactivity can be demonstrated (Fig. 5A,B). The αbENaC and ENaC antibodies both recognize a band at approximately 74 kDa.

In the freshwater tilapia gill, the rabbit polyclonal antibody against NHE2 cross-reacts with cells in both the lamellar and filament epithelia (Fig. 6B,D). In the afferent region, immunoreactive cells are found predominantly in the interlamellar space of the filament epithelium. These cells are round in appearance and frequently associated with Na⁺/K⁺-ATPase-immunoreactive cells (Fig. 6A). In contrast, immunoreactive cells in the efferent region are squamous and not associated with Na⁺/K⁺-ATPase-immunoreactive cells (Fig. 6D). Pillar cells, mucocytes and erythrocytes show negligible levels of immunoreactivity.

In western blots, the NHE2 antibody cross-reacts with a doublet at approximately 87 kDa, as reported for this protein (Fig. 7). However, there are also immunoreactive bands at approximately 56, 60 and 100 kDa. A comparison of separated purified membrane homogenates on a 5 % to 25 % sucrose gradient probed for NHE2 and Na⁺/K⁺-ATPase (as a basolateral membrane marker) indicates that the 87 kDa bands are not in the basolateral fraction of the sucrose gradient fraction.

The antibodies against the NHE3 isoform (antibodies 1380 and 1381) cross-react with unidentified material in the basal portion of the epithelium that is also recognized by the normal rabbit serum control. This unidentified material has the appearance of a grape-like cluster.

In the rainbow trout, the distribution of the V-ATPase (Lin et al., 1994) and Na⁺/K⁺-ATPase (Witters et al., 1996) have been previously described using the same antibodies. Western analysis data have also been presented by Lin et al. (1994) for the V-ATPase A-subunit.

The trout reared in Vancouver tapwater (ion-poor) have (A-subunit) V-ATPase widely distributed throughout the branchial epithelium (Fig. 8A,B). This pattern of labelling is similar to that reported by Lin et al. (1994). However, immunoreactivity is also observed within branchial mucocyte mucin granules. There are many Na⁺/K⁺-ATPase-immunoreactive cells throughout both the lamellar and filament epithelia. The epithelial Na⁺channel β-subunit has an identical staining pattern to that of the V-ATPase in trout (Fig. 8C). Western analysis reveals immunoreactivity of the α5 Na⁺/K⁺-ATPase antibody with a band at approximately 116 kDa, the V-ATPase A-subunit anti-peptide antibody with a band at approximately 70 kDa and the β- and α-subunits of the epithelial Na⁺channel with bands at approximately 98 and 74 kDa, respectively (Fig. 9).

The gill tissue of three of the four Ottawa trout examined had a discontinuous apical distribution of V-ATPase, a pattern of labelling similar to that reported by Sullivan et al. (1995).

The number of Na⁺/K⁺-ATPase-immunoreactive cells is also not as great as observed in the Vancouver trout. The fourth Ottawa trout had a V-ATPase labelling pattern similar to that observed in Vancouver trout.

Immunogold labelling of the V-ATPase is associated with the apical plasma membrane of both MR and pavement cells (Fig. 10). There is also clustered subapical labelling of electron-dense areas suggestive of vesicle labelling. The density of labelling decreases towards the basal portion of the epithelial cells.

We were unable to make the V-ATPase E-subunit anti-peptide antibody used by Sullivan et al. (1995) cross-react with tissue sections or western blots. Also, the NHE2 and NHE3 antibodies 597 and 1380/1381, respectively, did not cross-react. The antibody against trout erythroid band 3 protein cross-reacts only with erythrocytes and never with the branchial epithelium. A number of antigen retrieval techniques were used (1 % SDS/PBS, trypsin and heat), but all yielded negative results.

This is the first study to identify the distributions of the apical Cl⁻/HCO₃⁻and Na⁺/H⁺exchangers and epithelial Na⁺channels in the branchial epithelium of a fish. In the two freshwater species examined, the patterns of immunolabelling do show some similarities, but also have some major differences. Fig. 11A,B summarizes the immunolocalization data collected from the tilapia and trout.

The apical localization of the anion exchanger in the tilapia MR cell provides convincing evidence that these presumed freshwater chloride cells are aptly named. In seawater fishes, the evidence that branchial MR cells are involved in the active efflux of Cl⁻has, until now, been far more convincing than the evidence that Cl⁻uptake occurs via the freshwater MR cell (Perry, 1997). The tilapia MR cell apical anion exchanger is immunogenically related to the erythrocyte band 3 of trout (Fig. 2A). This is unlike the pattern seen in higher vertebrates, in which non-erythroid band-3-like proteins are restricted to the basolateral membrane domain of acid-excreting cells (A- or α-type cells of bladder and cortical collecting duct; Drenckhahn et al., 1987; Alper et al., 1989). In mammals, the apical Cl⁻/HCO₃⁻exchanger is an AE2 isoform (Alper et al., 1997) that is not immunogenically related to band 3 (AE1) and shows less sensitivity to disulphonic stilbenes (e.g. SITS; Cohen et al., 1978). In the trout, apically applied SITS has been shown to inhibit Cl⁻uptake and to cause alkalosis, results consistent with the presence of an apical Cl⁻/HCO₃⁻exchanger (Perry and Randall, 1981). Sullivan et al. (1996) were also able to localize band 3 mRNA by in situ hybridization using a 28-mer oligonucleotide probe to the filament interlamellar epithelium, which is typically populated by MR cells. Thus, unlike higher vertebrates, it appears that fish make use of an apical band-3-like anion exchanger for apical Cl⁻/HCO₃⁻exchange.

We would like to note that immunolabelling of gill tissue with the trout AE1t antibody was possible only after pre-treating the sections with detergent (SDS). In retrospect, such a treatment makes sense because the polyclonal antibody was generated against band 3 protein purified under denaturing conditions (SDS–PAGE). Thus, the conformational changes that occur in the native protein with denaturation and subsequent refolding (upon removal of SDS) result in different epitopes. The polyclonal antibody, which is composed of a battery of different antibodies that recognize different epitopes on the band 3 protein, is capable of recognizing the erythrocyte band 3 protein in a number of different species. These epitopes may not be present in the native protein of either the gill epithelium or erythrocyte, accounting for the negative results observed without pre-treatment. Erythrocytes of both species showed strong cross-reactivity regardless of pre-treatment. It also is possible that SDS pre-treatment exposes epitopes masked by tissue fixation (protein crosslinking); however, heat denaturation did not enhance reactivity.

The co-localization of the V-ATPase and epithelial Na⁺channel in both tilapia and trout greatly increases the viability of this Na⁺uptake model in freshwater fishes. This is also the first study to identify the distribution of the epithelial Na⁺channel in fish (Harvey, 1992). However, it is interesting that the co-localized V-ATPase and epithelial Na⁺channel are found exclusively in pavement cells in a particular area of the branchial epithelium in one species (tilapia) but in a mixed population of pavement cells and MR cells throughout the branchial epithelium in the other species (trout). It seems that the exclusive pavement cell distribution in the tilapia follows the predicted distribution (Goss et al., 1995; Perry, 1997), while that in the trout does not. It is not clear in the trout whether Na⁺uptake differs in pavement cells and MR cells. It may be that the higher Na⁺/K⁺-ATPase activities associated with the MR cell are required to aid Na⁺uptake. There is nothing to preclude the possibility that both Na⁺and Cl⁻uptake mechanisms can be found in a subpopulation of branchial MR cells. In the trout, Na⁺influx has been shown to be sensitive to SITS and Cl⁻uptake to amiloride (Perry and Randall, 1981). The amphibian skin MR cells have both an apical epithelial Na⁺channel and anion-exchange proteins together with the V-ATPase (γ-cell; Larsen, 1991).

We were able to identify the distribution of the epithelial Na⁺channel only using the polyclonal antibody against the β-subunit; results with the α-subunit antibody were negative. Perhaps it is not surprising that the β-subunit of the epithelial Na⁺channel was recognized in fish while the α-subunit was not since there is greater amino acid identity within the β-than the α-subunit isoform groups within higher vertebrates (human compared with frog; β-subunit 79 %, α-subunit 59 %; Garty and Palmer, 1997).

The abundance of Na⁺/K⁺-ATPase associated with the unique tubular system of MR cells allows the clear identification of this cell type from pavement cells and mucocytes (Pisam and Rambourg, 1991). However, it should be noted that, although labelling is absent from the pavement cell, it is considered to be present but in amounts below the level of detection by immunohistochemistry (and autoradiography) since Na⁺/K⁺-ATPase is generally considered to be a ubiquitous basolateral membrane protein (Hootman and Phillpot, 1979; Witters et al., 1996). There is also the possibity that the α5 antibody used is unable to detect the α-subunit isoform(s) of Na⁺/K⁺-ATPase present in pavement cells. In the case of the pavement cells with an apical Na⁺uptake mechanism, Na⁺/K⁺-ATPase is probably present to facilitate the basolateral movement of Na⁺(Richards and Fromm, 1970; Payan et al., 1975).

The finding of NHE2 cross-reactivity in the branchial epithelium of the freshwater tilapia was not expected. However, Wright (1991) calculated that it is possible to operate a Na⁺/H⁺exchange for Na⁺uptake in the water conditions in which these fish were raised (0.5 mmol l⁻¹[Na⁺], pH 8). Also, since the tilapia is euryhaline, the Na⁺/H⁺exchanger may be expressed in fresh water in the expectation of movement into a more saline environment where it could potentially operate for acid excretion or Na⁺uptake. Claiborne et al. (1999) have identified an NHE2-like nucleotide sequence in the marine sculpin Myoxocephalus octodecimspinosus, although they were unable to confirm the presence of an NHE3 homologue (Blackston et al., 1997). We were also unable to detect any specific NHE3 cross-reactivity in either the tilapia or trout using a battery of five antibodies. However, Edwards et al. (1999) have recently been able to demonstrate NHE3 cross-reactivity using antibody 1381 in rainbow trout. The pattern of immunolabelling appears similar to that of NHE2 in the filament epithelium of tilapia (Fig. 6B). NHE3 cross-reactivity in the tilapia was found to be non-specific (not different from the normal rat serum negative control).

The localization of NHE2 to subapical cells adjacent to MR cells is interesting and again may be due to the tilapia being a euryhaline fish. On the basis of the location of these cells, it appears that they may be a freshwater-type accessory cell. These cells have been found in a number of freshwater euryhaline fishes (Pisam et al., 1988, 1989). The function of the accessory cell is poorly understood, and the function of the NHE2 in such an arrangement is not known.

Within the literature, there are two conflicting reports regarding the distribution of the V-ATPase in the freshwater rainbow trout (O. mykiss) (Lin et al., 1994; Sullivan et al., 1995). Lin et al. (1994), using an anti-peptide antibody directed against the A-subunit (Südhof et al., 1989), found extensive apical staining of the branchial epithelium and concluded that both epithelial pavement cells and MR cells expressed the V-ATPase. Sullivan et al. (1995), using an anti-peptide antibody directed against the E-subunit, found a small population of lamellar cells with apical immunoreactivity. On the basis of immunogold studies, they concluded that labelling was restricted to pavement cells.

Recently Evans et al. (1999) have suggested that the differences may be due to the rearing conditions of the fish. Vancouver tapwater has a much lower pH and ionic strength (pH 5.8–6.4; [Na⁺] 0.02 mmol l⁻¹) than Ottawa tap water (pH 7.5–7.7; [Na⁺] 0.15 mmol l⁻¹). Laurent et al. (1985) have shown that such ion-poor conditions result in the proliferation of branchial MR cells, implying a greater need for ion uptake. With a central role in driving Na⁺uptake, increased V-ATPase activity would be predicted under such conditions such as those in Vancouver trout. This does explain the more extensive distribution seen in Vancouver trout, but does not explain why the V-ATPase should be found in both MR cells and pavement cells in Vancouver trout.

We have been using the same anti-A-subunit antibody as Lin et al. (1994) and can confirm their observations but, unfortunately, attempts to detect a reaction with the anti-E-subunit antibody used by Sullivan et al. (1995) were unsuccessful. However, we have been able to examine gill tissue collected from Ottawa trout and can report a similar labelling pattern to that observed by Sullivan et al. (1995) using the A-subunit antibody. In addition, we have also performed double-labelling experiments and can clearly show that some of the apical V-ATPase labelling is associated with Na⁺/K⁺-ATPase-immunoreactive cells (MR cells). This observation would contradict the immunogold studies of Sullivan et al. (1995), but in their paper they admit that their observations were not exhaustive and did not entirely exclude the possibility that a subpopulation of MR cells may also apically express the V-ATPase. It is also possible that different isoforms of the E-subunit are expressed in pavement cells and MR cells. Studies of the mammalian kidney have shown that there are heterogeneous forms of the E-subunit that have different tissue and membrane distributions (Hemken et al., 1992). The antibody that Sullivan et al. (1995) employed was generated against the same synthetic peptide used by Hemken et al. (1992) in the development of a battery of monoclonal antibodies. The immunologically determined distribution of the E-subunit was not the same for all the antibodies, suggesting that different variants of the E-subunit may exist and possibly impart functional variation to the V-ATPase. Immunolocalization of the 56 kDa (B) and 70 kDa (A) subunits was found not to differ in the rat kidney (Brown et al., 1987). A comparison of the results obtained on trout from Ottawa but using the A-subunit probe suggests that the immunoreactivity observed by Sullivan et al. (1995) is localized to gill MR cells and not only to pavement cell as had been suggested.

An interesting hypothesis put forward by Feng and Forgac (1994) suggests that reducing conditions favour V-ATPase activity or, conversely, that oxidizing conditions inhibit activity. When the A-subunit Cys254 is oxidized, it forms a disulphide bond with Cys532, inactivating the ATPase. It has been argued that this ‘redox modulation’ allows the V-ATPase to be active in vesicle acidification under the reducing condition of the cytoplasm, but when the V-ATPase cycles through the plasma membrane it is inactive. It may also explain why the plasma membrane V-ATPase is frequently found in MR cells where reducing conditions can be maintained by mitochondrial cytochrome c oxidase (Harvey and Wieczorek, 1997). How does the V-ATPase manage to operate in the gill which is, of course, the central gas-exchange organ in fishes and subjected to high The mitochondria may act as reducing agents within close proximity, but it is unclear what may be occurring in the pavement cells. In the brown bullhead (Ictalurus nebulosus), Goss et al. (1992, 1994) have found pavement cells that contain many mitochondria and that show an increase in microvillar density with hypercapnia. Hypercapnia has been shown to increase V-ATPase activity (Lin and Randall, 1993) and the level of V-ATPase protein expression (Sullivan et al., 1995). These cells were distinguished from ‘MR cells’ by the absence of a tubular system in the brown bullhead (Goss et al., 1992, 1994). In tilapia, which have only upstream (leading edge) pavement cells expressing the V-ATPase, mitochondria may be numerous; however, immunogold analysis of pavement cells in the trout by Sullivan et al. (1995) demonstrated that these cells were characterized by their absence of mitochondria. We have also used mitochondrial density as a characteristic for distinguishing MR cells from pavement cells and have observed apical immunogold labelling in cells with few mitochondria. Immunocytochemistry places limitations on tissue fixation or morphological preservation in favour of preservation of antigenicity, making the identification of the tubular system within the cell uncertain. Alternatively, the trout V-ATPase might possess an A-subunit isoform that has a reduced sensitivity to oxidizing conditions. Nitrate (NO₃⁻) acts as an oxidizing agent inhibiting the V-ATPase by promoting the formation of disulphide bonds (Dschida and Bowman, 1995). The degree of inhibition of ATPase activity by NO₃⁻in purified membranes is similar to that by bafilomycin A1 (J. M. Wilson, unpublished data). The question of redox modulation of the teleost gill V-ATPase will probably be resolved once the A-subunit has been cloned.

In the tilapia, there is an interesting upstream distribution of the H⁺-excreting mechanisms (V-ATPase and NHE2). One possible explanation of this arrangement may be to aid ammonia unloading by boundary-layer acidification (Randall et al., 1991). This seems a more plausible explanation than the requirement for some specific condition necessary for Na⁺uptake. Acidification of the boundary layer by both H⁺excretion and hydration of respiratory CO₂ (CO₂+H₂O→H⁺+HCO₃⁻) aids ammonia excretion by maintaining the transbranchial ammonia partial pressure (P_NH) gradient by removing NH₃ by protonation to form ⁠. The transbranchial P_NH gradient can account for the majority of the total ammonia efflux (Cameron and Heisler, 1983), and the disruption of the boundary-layer pH effect by addition of buffer to the water inhibits ammonia excretion (Wright et al., 1989; Wilson et al., 1994). The contribution of CO₂ hydration to boundary-layer acidification would be greatest downstream, decreasing upstream as the levels of transiting blood decrease. Presumably, upstream acidification of the boundary layer by the V-ATPase and Na⁺/H⁺exchanger would aid ammonia elimination whilst downstream alkalization by Cl⁻/HCO₃⁻exchange would not interfere. The absence of such a relationship in trout may be related to survival in almost ion-free waters and a greater need for Na⁺uptake, H⁺efflux and/or boundary-layer acidification imparted by dietary and/or environmental factors.

In the tilapia, we have developed an almost complete picture of the organization of NaCl uptake and acid–base regulatory mechanisms, whereas in the trout the picture is far from complete. The absence of cross-reactivity should not be taken to mean that that a particular transporter is not present in the gill. There are numerous technical reasons for the absence of cross-reactivity. However, the partial picture of the trout gill is important in highlighting the points that the organization of the gill for ionoregulation and acid–base regulation need not be uniform among fishes and that generalizations need to be taken with some caution. There are at least 25 000 different species of teleost fishes, and our understanding of the few we have studied is still far from complete.

Financial support for this work was provided by an NSERC operating grant to D.J.R. and an EC Marie Curie TMR grant and UBC fellowship to J.M.W. We would also like to acknowledge the support of members of the City University of Hong Kong, the Institut de Biologie Moléculaire et Cellulaire – CNRS (Jules Hoffman) and the CEPE-CNRS, where some of the work was conducted.