Calcium Transport By Isolated Skin Of Rainbow Trout

The function of mitochondria-rich (MR) cells in the gills and related epithelial structures (jaw skin and opercular epithelium) of seawater teleosts is now very clearly associated with the secondary active secretion of Cl⁻ (Foskett and Scheffey, 1982; reviewed by Zadunaisky, 1984; Pequeux et al. 1988). The term ‘chloride-secreting cells’, first coined by Keys and Willmer (1932) and now often shortened to ‘chloride cells’, is an accurate description of their function in sea water. This same term is now commonly applied to ultrastructurally similar MR cells in the gills of freshwater teleosts. However, the function of these cells in freshwater fish, in particular their specific contribution to ionoregulation, remains highly controversial (reviewed by Wood, 1991). While some studies have discounted any involvement of MR cells in active Na⁺ and Cl⁻ uptake (Girard and Payan, 1980; Payan et al. 1984), others have indicated that they are the principal sites of these processes (Avella et al. 1987; Perry and Laurent, 1989; Laurent and Perry, 1990). The most recent data suggest that the MR cells in freshwater gills are involved in Cl⁻ uptake, but not in Na⁺ uptake (Goss et al. 1992a,b). The freshwater chloride cell has also been implicated in active Ca²⁺ uptake (Payan et al. 1981; Perry and Wood, 1985; Isihara and Mugiya, 1987; Perry and Flik, 1988; Flik and Perry, 1989).

In general, this evidence about the possible ionoregulatory role(s) of the MR cell in fresh water has been circumstantial and correlational. An important hindrance to progress has been that, in contrast to seawater teleosts, surrogate ‘models’ for the freshwater gill epithelium are not available - i.e. there are no related flat epithelial structures containing numerous MR cells and suitable for mounting in vitro. For example, the opercular epithelia of freshwater-adapted brook trout (Salvelinus fontinalis;Marshall, 1985) and rainbow trout (Onco-rhynchus my kiss; present study) completely lack MR cells, based on the absence of fluorescent cells when they are stained with the mitochondrial fluorophore DASPEI. Opercular epithelia of freshwater-adapted tilapia (Oreochromis mos-sambicus;Foskett et al. 1981; Wendelaar Bonga et al. 1990) and killifish (Fundulus heteroclitus’, Degnan et al. 1977) and goby (Gillichthys mirabilis’, Marshall, 1977) adapted to 1–5 % sea water do contain some MR cells. However, when these epithelia are mounted in vitro with balanced saline on both sides, they either fail to transport Cl⁻ (tilapia) or actively secrete Cl⁻ (killifish, goby). The failure of these experiments to demonstrate Cl⁻ uptake by freshwater MR cells could be associated with the use of high mucosal NaCl concentrations in vitro that more closely approximate seawater conditions. Alternatively, in the case of killifish and goby, these euryhaline animals are normally seawater-resident and they may retain hormonally inhibited MR cells while in brackish water. Ca²⁺ fluxes in these preparations have not been examined except for tilapia (see note added in proof).

In the present study, we wished to examine Na⁺, Cl⁻ and Ca²⁺ transport in a flat preparation of MR cells from a teleost that is normally resident in fresh water in an attempt to clarify the role of the freshwater chloride cell. We found that the skin overlying the cleithrum bone of rainbow trout contains numerous MR cells; this paper reports initial experiments with this preparation. The tests carefully mimicked in vivo freshwater ionic and osmotic conditions, hence possible artefacts associated with high mucosal NaCl concentrations were eliminated.

Adult 100–300 g rainbow trout [Oncorhynchus my kiss (Walbaum)] were obtained from Fraser’s Mills Hatchery, Antigonish, Nova Scotia, and were maintained under natural photoperiod in flowing dechlorinated tapwater at 10–12°C for at least 10 days prior to use. This water was soft (pH 6.8–7.2) with an average composition (in mmol I⁻¹) of Na⁺, 0.16; K⁺, 0.02; Ca²⁺, 0.10; Mg²⁺, 0.06; Cl⁻, 0.17; SO₄²⁻, 0.14; and titratable alkalinity (to pH 4.0), 0.20. At least 7 days prior to use, the trout were transferred to individual fish boxes served with flowing water of the same composition; they were not fed during this period. This confinement facilitated cortisol injections (see below) and minimized disturbance. The fish were killed by a blow to the head, and a terminal blood sample was taken by heparinized syringe from the caudal vessels for analysis of plasma Na⁺, Cl⁻, Ca²⁺ and cortisol. The cleithrum membranes were then immediately dissected.

In one series, fish were treated with cortisol by the method of Perry and Wood (1985) in an attempt to cause proliferation of MR cells on the cleithrum skin. Trout were injected (intramuscularly) once a day for 9–11 days with 4 mg kg⁻¹ hydrocortisone hemisuccinate sodium salt (Sigma) in 140 mmol I⁻¹ NaCl (0.4ml kg⁻¹). The results were compared with those of an uninjected control group. A saline-injected control was not performed as we wished to avoid the elevation of endogenous cortisol levels that accompanies this procedure.

The in vitro bathing solution for the inside (serosal surface) of the cleithrum skin was Cortland saline (Wolf, 1963) modified to duplicate more closely the measured composition of trout plasma. The formulation (in mmol I⁻¹) was NaCl, 129.9; KC1, 2.55; CaCl₂.2H₂O, 1.56; MgSO₄.7H₂O, 0.93; NaHCO₃, 13.00; NaH₂PO₄.H₂O, 2.97; glucose, 5.55; NH₄C1, 0.30 (all salts from Sigma Chemical Co.); with 20 mg ml⁻¹ bovine serum albumin (Sigma grade III, Fraction V) as a substitute for plasma protein. After equilibration with a nominal gas mixture of 0.3% CO₂, balance O₂, the measured acid-base status (pH7.8–7.9, $PCO2$ =0-25–0.32 kPa, HCO₃⁻=5–6mmol I⁻¹) was typical of blood in vivo and the was greater than 65 kPa.

The bathing solution for the outside (mucosal surface) of the preparation was fresh water taken from the acclimation system; it was gassed with 100% O₂, resulting in a greater than 65 kPa. The mucosal solution was changed at hourly intervals and its composition analysed at the start and end of each 60 min experimental period. Typical values were Na⁺=0.15–0.30mmoll⁻¹, Cl⁻=0.15–0.30 mmol I⁻¹ and Ca²⁺=0.08–0.12 mmol I⁻¹. Ion levels in this very dilute medium tended to change as a result of fluxes and release of bound ions from the mucosal surface (see Results). The addition of ²²Na and ⁴⁵Ca radioisotopes did not appreciably alter net mucosal concentrations, but addition of ³⁶C1, which was of lower specific activity, typically raised the net Cl⁻ level to 0.40–0.60 mmol I⁻¹.

The portion of skin taken for the flux experiments overlies the cleithrum bone directly behind the posterior-ventral margin of the gill filaments. It is therefore normally irrigated with expired ventilatory water. The ‘cleithrum skin’ is roughly rectangular, about 2–3 cm² in total area, and extends dorsally and anteriorly from the pectoral fin to the ventral margin of the gill arches. It separates easily from underlying tissues. Cleithrum skin pieces from each side of the animal were moistened with Cortland saline, dissected and mounted in removable apertures of Ussing-style membrane chambers. After mounting, the mucosal surfaces were washed vigorously with fresh water to remove saline and to displace the mucus that tended to accumulate during dissection. The membranes were then placed in the chambers with saline on the serosal surface and fresh water on the mucosal surface. An initial 30 min period was allowed for the tissue to adjust to the in vitro conditions. At the start and end of this period, the mucosal surface was rinsed twice by the flow-through of 30 ml of fresh water (ten times the half-chamber volume) to ensure the removal of all traces of saline. Throughout the experiment, the mucosal and serosal media were continually mixed by magnetic stirrers, and the appropriate gases were passed across the surface of each media.

Experiments followed previously published methods (Marshall, 1986). In brief, the membranes were housed in water-jacketed chambers (11±1°C) of 3.0ml volume per side with membrane apertures of 0.238 cm² exposed surface area. The epithelium was open-circuited throughout; current-passing and voltage-measuring bridges were of agar/150mmoll⁻¹ KC1. Transepithelial voltage was measured throughout the experiment, and resistance was estimated every 5 min from the voltage response to 1 pA current pulses from the current/voltage clamps (D. Lee Co. or WP Instruments DVC-1000). The reported membrane voltages in asymmetrical solutions were corrected for junction potentials (measured against a free-flowing 3 mol I⁻¹ KC1 half-cell). The membrane resistances were corrected for solution resistances. In experiments where Cl⁻ fluxes were measured, the bridges were removed during flux determinations to prevent alterations in mucosal Cl⁻ concentration resulting from bridge leaching.

The two membranes from each fish were generally set up in pairs for unidirectional flux measurements, one being used for the mucosal-to-serosal unidirectional influx (J^ms), the other for the serosal-to-mucosal unidirectional influx (J^sm). After the mucosal rinse at the end of the initial 30 min settling period, ³⁶C1 (H³⁶C1 from New England Nuclear, neutralized with KOH), ²²Na (²²NaCl from Amersham) or ⁴⁵Ca (⁴⁵CaC12 from Amersham) was added to a final specific activity of at least 25 000ctsmin⁻¹ μ mol⁻¹ on the saline (serosal) side and 4000000ctsmin⁻¹μmol⁻¹ on the freshwater (mucosal) side. In Ca²⁺ flux experiments, 10–100-fold higher specific activities were used. A further 30–60min was allowed to elapse to ensure complete radioisotopic equilibration. Experiments typically consisted of three 60 min periods, separated by 20 min intervals used for changing over the mucosal solution and adding drugs. The first period served as a control, and the second and third were used for experimental treatments or for time controls. Within each period, samples were taken at 0 and 60min for ionic analyses and at 0, 20, 40 and 60 min for radioactivity analyses. Unidirectional fluxes were calculated in the standard fashion from the specific activity on the labelled side and the appearance of radioactivity on the unlabelled side. Flux values for the three 20 min intervals were averaged to produce a mean for each experimental period.

The efflux of ions across the cleithrum skin and the release of bound ions from the mucosal surface (see Results) significantly altered the ionic concentrations in the dilute mucosal solution. If left uncorrected, mucosal concentrations, especially of Na⁺ and Cl⁻, would have increased greatly over the course of a 5 h experiment. To minimize this effect and to maintain concentrations as close to true freshwater values as possible, the mucosal surface was extensively rinsed initially (see above), and the mucosal solution was renewed at the end of each 60 min period. Water samples were taken immediately before and after each renewal, and the concentration of the ion of interest (Na⁺, Cl⁻ or Ca²⁺) was determined so that the average concentration was known for flux calculations. These measurements of change in total mucosal concentration also allowed calculation of the net loss (or gain) of ions by the membrane over each period.

At the end of some experiments, especially time controls, samples of serosal saline and mucosal fresh water were drawn from the membrane chambers into Hamilton gas-tight syringes and analyzed for ⁠, pH and total CO₂ content. This served to check the maintenance of correct acid-base status and O₂ levels.

At the end of each experiment, the density of MR cells on the membranes was assessed using the mitochondrial fluorophore DASPEI [2– (4-dimethylamino-styryl)-N-methylpyridinium iodide; ICN Pharmaceutics; Marshall and Nishioka, 1980]. The stock was 0.2 mg DASPEI per millilitre of distilled water; this was diluted to 0.008 mg ml⁻¹ final concentration with Cortland saline. The membranes were left in the chamber apertures and were incubated in oxygenated DASPEI solution for a minimum of 20min prior to microscopic examination. A Zeiss photomicroscope III equipped with epifluorescence and a mercury/mercury vapour burner was used to observe and count the MR cells in situ. Cell density estimates were made by averaging the numbers of cells in 10 randomly chosen half fields at a magnification of 200 (total counted area=3.25 mm²). Cell counts could not be made after ionomycin treatments as DASPEI fluorescence became nonspecific.

The adrenergic agonists adrenaline (L-epinephrine bitartrate; Sigma), clonidine (clonidine hydrochloride; Sigma) and isoprénaline (L-isoproterenol bitartrate; Sigma) were added to the serosal saline at a concentration of 10⁻⁵moll⁻¹. Cyclic AMP mobilization was duplicated by the addition of 10⁻³moll⁻¹ db-cAMP (dibutyryl-cyclic-adenosine monophosphate, Na⁺ salt; Sigma) plus 10⁻⁴moll⁻¹ IBMX (3-isobutyl-l-methylxanthine; Sigma) to the serosal side. Lanthanum chloride (10⁻⁴moll^−l; Sigma) was added to the mucosal fresh water as a putative Ca²⁺ channel blocker. The Ca²⁺ ionophore ionomycin (ionomycin free acid; Calbiochem) was dissolved in a minimum of dimethylsulphoxide (DMSO) and added to the mucosal fresh water at concentrations of 1.0, 3.2 or 10μmoll⁻¹. The resulting maximum DMSO concentration was 0.18%. In these experiments, the water pH was adjusted to the pK (8.3) of ionomycin with NaHCO₃, and the same DMSO concentration and water pH were employed in the control period prior to the addition of ionomycin.

Na⁺ and Ca²⁺ in water, saline and plasma samples were determined by atomic absorption (Varían 375 AA or 1275 AA). Samples and standards for Ca²⁺ were made up in a 5000 mg I⁻¹ K⁺ diluent to minimize interference effects. Cl~ in water was determined colorimetrically (Zall et al. 1956) and in saline and plasma by coulometric titration (Radiometer CMT10). Cortisol was determined by a commercial ¹²⁵I radioimmunoassay (Immuchem Cortisol kit, ICN Biomedicals Inc.) using standards diluted to the protein concentrations found in trout plasma; ¹²⁵I radioactivity was measured on a Packard Minaxi 5000 autogamma counter. ²²Na, ³⁶C1 and ⁴⁵Ca radioactivities were determined by counting to 1 % error on a Fig. 1A). In contrast, the Cl⁻ net loss rate stabilized at about –300 nmol cm⁻² h⁻¹ by period 1, and was never significantly different from J^net for Cl⁻ (Fig. IB). J^net for both Na⁺ and Cl⁻ did not vary significantly over the three periods. The Ca²⁺ net loss rate [not shown; overall average=–7.1±7.5nmolcm⁻²h⁻¹ (13)] could not be distinguished from zero or from the J^net for Ca²⁺ [–0.243± 0.063 nmol cm⁻² h⁻¹ (6)] throughout the experiment. These results suggest that increases in Na⁺ (and initially in Cl⁻) concentration in the mucosal bath do not come entirely from transepithelial fluxes but instead represent washout of ions adsorbed to the epithelial surface. For this reason, the increase in mucosal ion concentration is not a good indication of net ion flux.

The of the serosal bathing solution of the time controls at the end of the experiment (total incubation time approximately 5h) was 49.5±2.6kPa (19), the pH was 7.844±0.019 (22), the total CO₂ was 5.51 ±0.08 mmol I⁻¹ (20), the HCO₃⁻ concentration was 5.40±0.06mmoll⁻¹ (20) and the HCO₃⁻ was 0.27±0.01 kPa (20). Except for a small fall in ⁠, these values were unchanged from those at the start of the experiment and indicated that the in vitro conditions provided appropriate acid-base status and ample O₂ supply to the tissues.

To establish which, if any, ionic species might be actively transported by the epithelium, we compared the observed unidirectional flux ratio under open-circuit conditions for each pair of membranes (J^ms/J^sm) with that predicted by the Ussing equation from the prevailing V_t and ionic activities. Table 2 reports only data from period 1 (in the absence of drug treatments) but data from periods 2 and 3 yield the same conclusions. The analysis indicated that the observed flux ratio for Ca²⁺ was threefold greater (P<0.01) than that predicted on the basis of passive diffusion, i.e. there was active Ca²⁺ uptake from the mucosal to serosal sides. In contrast, the flux ratios observed for Na⁺ and Cl⁻ fluxes were in agreement with the predicted ratios (P=0.866 and 0.433, respectively), indicating that the movement of these two ions across the epithelium could be explained by passive diffusion alone. The much higher flux ratios for Cl⁻ than for Na⁺ resulted both from the positive V_t and from the higher mucosal Cl⁻ concentration. All observed flux ratios were far less than 1.0 in this in vitro preparation, reflecting the large chemical gradients favouring J^sm and opposing J^ms.

Ca²⁺ unidirectional influx (J^ms) was positively correlated (r=0.63, P<0.01, N=30) with the number of MR cells in the preparation (Fig. 2), suggesting that the MR cells are associated with Ca²⁺ uptake. There was no significant correlation of MR cell density with Ca²⁺ efflux, or with Na⁺ and Cl⁻ influxes or Cl⁻ efflux, suggesting that these fluxes are not associated with MR cells. Na⁺ efflux (J^sm) was negatively correlated (r=−0.65, P<0.05,N=8) with MR cell density; the meaning of this is unclear. There was no relationship between MR cell numbers and either V_t or membrane resistance.

To ascertain whether MR cell density affected the observed Ca²⁺ influx and Na⁺ efflux, we applied a hormone treatment known to increase MR cell numbers on the gills of trout (Perry and Wood, 1985; Perry and Laurent, 1989; Laurent and Perry, 1990). Ten days of cortisol injections caused about a 40-fold elevation in measured plasma cortisol levels without altering plasma Na⁺, Cl⁻ or Ca²⁺ concentrations (Table 3). The appearance of the skin changed; it became darker, more mucified and more fragile during dissection. However, contrary to expectation, this treatment caused a highly significant (P<0.001) 50% decrease in MR cell density on the cleithrum skin (Table 3). In accordance with the previous correlation analyses, the decrease in MR cell numbers was accompanied by trends for decreased Ca²⁺ influx (J^ms) and increased Na⁺ efflux (J^sm), though neither was significant. Cortisol treatment also significantly increased V_t without any effect on membrane resistance (Table 3).

The polyvalent cation lanthanum (La³⁺) inhibits Ca²⁺ uptake in intact trout (Verbost et al. 1987, 1989; Perry and Flik, 1988). We applied 10⁻⁴moll⁻¹ La³⁺ to the mucosal bath to determine whether Ca²⁺ transport by the cleithrum skin would be similarly affected. The responses to La³⁺ were followed for 2 h (periods 2 and 3) after the control period 1. There was no effect on electrophysiological variables or on unidirectional Ca²⁺ influx (J^ms; Table 4). However, parallel time controls normally showed an increase in the transepithelial resistance in periods 2 and 3 (see Table 1), which did not occur in the presence of La³⁺ (Table 4). The unidirectional efflux (J^sm) and J^net of Ca²⁺ were significantly increased by La³⁺ (Table 4), suggesting that La³⁺ increases the Ca²⁺ permeability of the cleithrum skin. The flux ratio criterion was satisfied (i.e. Ca²⁺ fluxes were indistinguishable from simple diffusion) after addition of lanthanum. The Ca²⁺ net loss rate, determined from changes in concentration in the mucosal medium, was markedly enhanced by La³⁺, while the unidirectional Ca²⁺ efflux was only modestly increased. Hence, it appeared that La³⁺ displaced bound Ca²⁺ from the mucosal surface.

The calcium ionophore ionomycin was used in an attempt to increase the Ca²⁺ permeability of the apical membrane, a treatment that should increase Ca²⁺ uptake if this step were rate-limiting. Ionomycin was dissolved in a minimum of DMSO and added to the mucosal side only. In the control period, the DMSO vehicle alone was added. Results with 1.0 and 3.2μmoll^−l ionomycin (with DMSO<0.06%) were identical and have been combined in Table 5. We used these lower doses because preliminary experiments with 10μmoll⁻¹ ionomycin (with 0.18% DMSO) showed marked progressive decreases in transepithelial resistance from 8.0±0.8kΩcm² in the control period (1) to 5.0±1.0kΩcm² and 2.4±0.4kΩcm² in the two successive periods (2 and 3) with ionomycin, respectively (P<0.00l, N=6). This large reduction in tissue resistance is indicative of disruption of epithelial integrity. At the lower concentrations of ionomycin (and DMSO), these effects did not occur although, as with La³⁺ (Table 4), there was no increase in resistance during periods 2 and 3 (Table 5).

Ionomycin significantly augmented unidirectional Ca²⁺ influx (J^ms) during periods 2 and 3 and unidirectional Ca²⁺ efflux (J^sm) during period 3 (Table 5). However, the most dramatic effect of ionomycin during both experimental periods was to cause a large net gain of Ca²⁺ by the cleithrum skin, as determined from changes in concentration in the mucosal medium. This occurred despite the fact that the J^net indicated a modest net loss of Ca²⁺across the tissue. This marked disappearance of Ca²⁺ from the mucosal solution probably represents increased apical permeability to Ca²⁺ and resultant sequestering of Ca²⁺ intracellularly in the epithelium.

Addition of either the cr₂-adrenergic agonist clonidine (10⁻⁵moll⁻¹, N=7) or the β-adrenergic agonist isoprénaline (10⁻⁵moll⁻¹,N=7) to the serosal side had no significant effect over 2 h (periods 2 and 3) on electrophysiological variables, on unidirectional Ca²⁺ influx (J^ms) or on the net loss/gain of Ca²⁺ measured from changes in mucosal concentration. The unidirectional Ca²⁺ efflux (J^sm) was not measured in these experiments. Addition of the non-selective adrenergic agonist adrenaline (10⁻⁵moll⁻¹; N=6) alone to the serosal side in period 2, and in combination with db-cAMP (10⁻³ mol I⁻¹) plus IBMX (10⁻⁴ mol I^−J) to the serosal side in period 3, again had no effect on electrophysiological variables and no effect on the unidirectional influx or efflux, on J^ms-J^sm or on the net loss/gain of Ca²⁺ by the preparation. In separate experiments, these same protocols had no effect on the comparable fluxes of Na⁺ (N=4) or Cl⁻ (N=4). As none of these treatments had significant effects, and all data were essentially similar to the time controls (Table 1, Fig. 1), individual results have not been tabulated.

We conclude that stimulation of ion transport in this epithelium does not appear to involve adrenergic agonists or cyclic AMP mediation. Nevertheless, it is of some interest that the measured serosal [40.9±1.3 kPa (14)] after epinephrine plus db-cAMP plus IBMX was significantly lower (P<0.05) than in the time controls [49.5±2.6kPa (19)], indicating that these agents had stimulated the O2 consumption of the tissue.

Relative to most other teleost skin preparations, the trout cleithrum skin started with a high transepithelial resistance which progressively improved during the 5 h of in vitro incubation (Table 1), suggesting a progressively improved sealing of the epithelium around the edges of the chamber. The high transmural resistance is consistent with the notion that freshwater skin and gill epithelia have a low permeability to minimize diffusive ion loss. Epithelia from seawater-acclimated fish (Degnan et al. 1977; Marshall, 1977; Foskett et al. 1981) typically have resistances only a few per cent of those reported here. Even by freshwater standards, the cleithrum skin resistance (about 12kΩcm²) was high: two-to threefold greater than the epithelial resistances observed previously across the opercular epithelia of freshwater-acclimated brook trout (Marshall, 1985) and tilapia (Foskett et al. 1981). This difference may result from the beneficial effects of serosal albumin (not used in previous studies), the use of true fresh water on the mucosal surface (cf. Marshall, 1985) and/or from the different properties of the epithelia.

The small inside-positive V_t (about +4 mV) was constant over time, indicating that ion selectivity did not change markedly during the in vitro incubation. The positive polarity of V_t was surprising, because a small inside-negative diffusion potential is usually recorded in intact trout and perfused trout heads in water of this Ca²⁺ concentration (McWilliams and Potts, 1978; Perry and Wood, 1985; Perry and Flik, 1988). Indeed the opercular epithelium of the brook trout mounted under similar conditions in vitro developed a V_t of about –9 mV (Marshall, 1985). The reason for this difference is unknown, but it may be related to the nature of skin mucus in the freshwater rainbow trout. An inside-positive diffusion potential develops when this mucus is placed in a dialysis cell with fresh water on the outside (Handy, 1989). In this regard, it is noteworthy that cortisol pretreatment, which clearly increased mucus secretion, made V_t significantly more positive in the cleithrum skin (Table 3). Paradoxically, this same pretreatment rendered V_t significantly more negative in the perfused trout head preparation (Perry and Wood, 1985).

The cleithrum skin of the rainbow trout contains MR cells which are apparently involved in the active uptake of Ca²⁺ from the water into the fish. Evidence includes the existence of DASPEI fluorescent cells in the epithelium, the significant positive correlation between the density of these MR cells in the skin and the unidirectional Ca²⁺ influx (J^ms; Fig. 2) and the significantly higher flux ratio (J^ms/J^sm) than that predicted by the Ussing flux ratio equation on the basis of passive diffusion (Table 2). Because the osmotic gradient favours net volume flow from mucosa to serosa, J^ms for Na⁺, Cl⁻ and Ca²⁺ could include a solvent drag component, although the flux ratios for the most permeant ions, Na⁺ and Cl⁻, show no evidence of this (i.e. they do not have larger than predicted flux ratios). In general, the low osmotic permeability of freshwater teleosts and the very high transmural resistance of the trout skin suggest that volume flow, and hence the solvent drag component, would be small.

The correlation between MR cell density and the rate of Ca²⁺ influx is in accord with the findings of Perry and Wood (1985) on the gills of isolated perfused trout head preparations and intact trout. The threefold higher J^ms/J^sm flux ratio than that predicted by the Ussing equation matches the fourfold difference reported by Perry and Flik (1988) for intact trout with similar water and plasma Ca²⁺ activities to those employed in the present study. These findings therefore lend considerable weight to the growing consensus that Ca²⁺ uptake is active in freshwater fish and that the MR or ‘chloride’ cells are the sites of this active uptake on the gills (Payan et al. 1981; Perry and Wood, 1985; Isihara and Mugiya, 1987; Perry and Flik, 1988; Flik and Perry, 1989).

The average MR cell density of 139 mm⁻² on the trout cleithrum skin is similar to values of 90–230 mm⁻² reported for the opercular epithelium of tilapia in fresh water (Foskett et al. 1981; Wendelaar Bonga et al. 1990). By way of comparison, the gills of freshwater trout contain about 400 MR cells per square millimetre on the surfaces of the secondary lamellae (Perry and Wood, 1985) and 500–2000 mm⁻² on the filamental surfaces (Perry and Laurent, 1989; Laurent and Perry, 1990). Total gill surface area (Wood, 1974) is somewhat greater than total body surface area (Webb, 1971) and about 100-fold greater than the cleithrum skin area. In the light of these figures, one would predict that the capacity of the skin relative to the gills for Ca²⁺ uptake would be very small if uptake were commensurate with the low density of MR cells extrabranchially.

The present data support the view that the gills are not the sole sites of Ca²⁺ uptake, and that significant uptake may also occur through the general body surface, i.e. skin and fins (Mashiko and Jozuka, 1964; Simmons, 1971; Dacke, 1979). By shielding most of the general body surface, Perry and Wood (1985) estimated that unidirectional Ca²⁺ influx through these extrabranchial routes in freshwater trout was about 7 μmolkg⁻¹ h⁻¹, or about equal to the unidirectional influx through the gills. A typical 250 g fish would have a cleithrum skin area of about 6 cm² (both sides) with a J^ms of about 0.08 nmol cm⁻² h⁻¹ (Table 2), yielding an influx of about 2 nmol kg^−lh⁻¹. This value is negligible relative to the total extrabranchial influx (Perry and Wood, 1985). However, MR cells have been reported to be widely distributed over the general body surface in other species (e.g. Korte, 1979). Although MR cells were sparse (buccal cavity, lower jaw) or absent (opercular epithelium) on other surfaces of trout, we did not perform an exhaustive search. A 250 g trout has a surface area of about 350 cm² (Webb, 1971); if the average J^ms over this surface were the same as that across the cleithrum skin it would amount to 0.1 μmol kg⁻¹ h⁻¹. If methodological differences can be ignored, the discrepancy suggests either that the in vitro preparation does not transport at its normal in vivo rate or that other extrabranchial surfaces transport at higher rates.

The movements of Na⁺ and Cl⁻ could be explained by passive diffusion alone (Table 2). Again this conclusion errs on the side of caution, because Handy (1989) measured Na⁺ and Cl⁻ activities in the skin mucus of rainbow trout which were about fivefold higher than in fresh water of comparable composition to that used here, i.e. the predicted flux ratios were probably underestimated (Table 2). Furthermore, there was no relationship between MR cell density on the epithelium and unidirectional Na⁺ or Cl⁻ fluxes, apart from a curious negative correlation with Na⁺J^sm (of unknown explanation). Neither general adrenergic stimulation, which is implicated in the control of gill Na⁺ and Cl⁻ transport (reviewed by Wood, 1991), nor cyclic AMP stimulation had any effect on unidirectional Na⁺ and Cl⁻ fluxes across the cleithrum skin. Therefore, the present data provide no support for the view that the chloride cells are the major sites for the active uptake of Na⁺ or Cl⁻ or both (see Introduction). However, they certainly do not disprove these ideas: either the in vitro conditions could have been less than optimal (e.g. lack of a key hormone or nutrient) or, as argued above, the MR cells on the cleithrum skin may differ from those on the gills. Indeed, based on ultrastructural evidence (e.g. Pisam et al. 1987), there is ample precedent for the occurrence of more than one type of MR cell in freshwater fish. In this regard, it is interesting that a cortisol injection protocol known to increase chloride cell density on the gills (Perry and Wood, 1985; Perry and Laurent, 1989; Laurent and Perry, 1990) had exactly the opposite effect on the cleithrum epithelium (Table 3). Apparently the populations of MR cells on the gills and skin are regulated differently. Alternatively, MR cells may already have been maximally proliferated in the soft-water holding conditions used in the present study (cf. Perry and Wood, 1985; Perry and Laurent, 1989).

The current model for Ca²⁺ uptake in freshwater fish is that Ca²⁺ enters the chloride cells from the water by passive diffusion through Ca²⁺-selective channels on the apical membrane (Flik et al. 1985; Perry and Flik, 1988; Lafeber et al. 1988). These channels may be regulated by agents such as hypocalcin, which could limit the rate of Ca²⁺ transport. Ca²⁺-binding proteins within the cell keep intracellular activities extremely low. At the basolateral membrane, Ca²⁺ is actively transported to the extracellular fluid by a high-affinity, calmodulin-dependent ATPase. Some of the present experiments were designed to assess whether this model is applicable to Ca²⁺ transport by the cleithrum skin.

The responses to mucosal lanthanum (10⁻⁴moll⁻¹) were tested, because this and lower concentrations of La³⁺ have been reported to inhibit immediately unidirectional Ca²⁺ influx, apparently by blocking putative apical channels in intact trout and perfused head preparations (Verbost et al. 1987, 1989; Perry and Flik, 1988). La³⁺ did not have this effect on the cleithrum skin but instead exerted its well-known general action of displacing bound Ca²⁺ and increasing diffusive permeability (Table 4), effects that have also been observed in intact fish (Eddy and Bath, 1979; Freda and McDonald, 1988). However, the flux ratio criterion was no longer exceeded after addition of lanthanum, suggesting passive distribution of Ca²⁺. The discrepancy between in vivo results and the present data could reflect a real difference in mechanism between cleithrum skin and gill Ca²⁺ transport mechanisms. Alternatively, it may be methodological. In vivo measurements have generally relied on the appearance of ⁴⁵Ca in the whole body. The action of La³⁺ to displace superficially bound Ca²⁺ could reduce apparent whole-body uptake without affecting the true transepithelial Ca²⁺ influx.

We employed the Ca²⁺ ionophore ionomycin (Liu and Hermann, 1978) to test whether apical Ca²⁺ uptake was rate-limiting on T¹¹⁸. While J”¹⁸ was stimulated, J⁸”¹ was also elevated, and there was a large gain of Ca²⁺ by the epithelium based on disappearance of total Ca²⁺ from the mucosal bathing solution (Table 5). These results (and those obtained previously by monitoring disappearance of Ca²⁺ label from fresh water) must be interpreted with caution because positive transepithelial Ca²⁺ balance was not achieved, as measured by J^net for Ca²⁺. It is likely that ionomycin promoted Ca²⁺ entry into both transport and non-transport cells. Ca²⁺ would diffuse into both cell types down the presumed favourable electrochemical gradient, but the ion would only be effectively translocated across the basolateral membranes of the transport cells (and hence contribute to J^ms). In the non transport cells, the Ca²⁺ would instead accumulate intracellularly.

General adrenergic stimulation has been variously reported to stimulate (Payan et al. 1981; perfused head), to inhibit (Donald, 1989; isolated gill arches) or to have no effect (Perry et al. 1988; intact animals) on Ca²⁺ uptake in freshwater rainbow trout. The reasons for these discrepancies are unclear. In the cleithrum skin, Ca²⁺ transport was unaffected by adrenaline, by selective β- and αadrenergic agonists and by cyclic AMP stimulation. The present results are, therefore, in accord with those of Perry et al. (1988) and with the current model of Ca²⁺ uptake in which adrenergic regulation is not involved.

In summary, the trout cleithrum skin, mounted in vitro under conditions closely duplicating those in vivo, offers the first surrogate model for the study of transport function in the freshwater gill epithelium. The present results with this model provide strong evidence for the involvement of MR cells in active Ca²⁺ uptake. Future experiments with this preparation may cast further light on the mechanism of Ca²⁺ uptake and on the role (or lack thereof) of the MR cells in the transport of other electrolytes and acid-base equivalents.

C.M.W. held the W. F. James Chair of Pure and Applied Science at St Francis Xavier University during this project. Supported by NSERC grants to W.S.M. and C.M.W. and the James Chair research stipend. We thank Drs M. J. O’Donnell and T. J. Shuttleworth for ion activity measurements and Mr R. S. Munger for excellent technical assistance.

In a simultaneous but independent study, McCormick et al. (1992) observed Ca²⁺ uptake by the isolated opercular epithelium of the cichlid Oreochromis mossambicus adapted to fresh water. As with the trout, the Ca²⁺ uptake seems to be associated with MR cells, inasmuch as proliferation of MR cells in low-calcium fresh water stimulates Ca²⁺ uptake. Circumstantially, the isolated tilapia opercular epithelium also develops the unusual positive transepithelial potential when bathed in fresh water.