Differentiation of the Chloride Extrusion Mechanism During Seawater Adaptation of A Teleost Fish, the Cichlid Sarotherodon Mossambicus

The branchial epithelium is important to euryhaline fish for both salt uptake in fresh water (Evans, 1980) and salt secretion in seawater (Maetz, 1974; Maetz & Bornancin, 1975). ‘Chloride cells’, large, granular, mitochondria-rich cells first observed in the gills of the seawater-adapted eel (Keys & Willmer, 1932), have been implicated in many ultrastructural and biochemical studies as the cell-type responsible for the salt secretion (Maetz, 1974; Maetz & Bornancin, 1975; Girard, Thomson & Sargen 1977; Karnaky, 1980; Philpott, 1980). In many species, the cells have been observed to proliferate and/or hypertrophy in response to increased salinity (see Karnaky. 1980). However, definitive proof that the chloride cell is the salt-transporting celltype is still lacking, as pointed out in several studies (Doyle & Gorecki, 1961 ; Holliday & Parry, 1962; Fleming & Kamemoto, 1963; Strauss, 1963).

Recently, two in vitro preparations were introduced - the isolated opercular membrane from Fundulus heteroclitus (Karnaky, Degnan & Zadunaisky, 1977) and the isolated skin from Gillichthys mirabilis (Marshall, 1977)-which possess typical chloride cells (Karnaky & Kinter, 1977; Marshall & Nishioka, 1980) and are flat epithelia which can be removed from the fish and mounted in vitro in Ussing-style chambers, thereby permitting a more rigorous biophysical approach to study the mechanisms of ion transport. Both preparations, when removed from seawater-adapted fish, develop transepithelial potentials of 10–40 mV (basal positive) and generate short-circuit currents which appear to be completely accounted for by electrogenic chloride secretion from serosa to mucosa (Degnan, Karnaky & Zadunaisky, 1977; Karnaky et al. 1977; Marshall & Bern, 1980). The responses of these tissues to epinephrine, SCN⁻ and ouabain (Karnaky et al. 1977Marshall, 1977; Marshall & Bern, 1980) are very similar to the branchial responses (Maetz, 1974; Maetz & Bornancin, 1975). Other similarities in the salt-secretory behaviour of gills and the in vitro preparations (Degnan et al. 1977; Mayer-Gostan & Zadunaisky, 1978; Degnan & Zadunaisky, 1979) also exist and it has been suggested that these isolated epithelia are valuable models for the study of the ion-secretory mechanisms utilized by the gills in seawater-adapted fish (Karnaky, 1980). However, when these epithelia are removed from freshwater-adapted F. heteroclitus and G. mirabilis, they still possess typical chloride cells and continue to secrete chloride (Degnan et al. 1977; Karnaky & Kinter, 1977; Marshall, 1977). Therefore, definitive proof of the role of the chloride cell in this process is still lacking since changes in chloride cell morphology and/or density have not been correlated with parallel ion transport alterations. In addition, these preparations have not proven useful for studying the adaptive changes that occur during movement of euryhaline fish from one environment to another.

We chose the tilapia, Sarotherodon mossambicus (formerly Tilapia mossambica), in an attempt to use the isolated opercular membrane to study these processes. This tilapia is a highly euryhaline species, tolerating salinities ranging from fresh water to twice seawater (Potts, Foster, Rudy & Howells, 1967), but, unlike Fundulus and Gillichthys, normally lives in fresh water.

Tilapia, Sarotherodon mossambicus, of both sexes weighing 20–90 g were obtained from a heated outdoor freshwater pond at the University of California, Berkeley, The fish were acclimated in indoor aquaria maintained at 25–27 °C under a 10L:14D photoperiod. The water was continuously aerated and recirculated through charcoalglasswool filters. In addition, half the water was renewed every 4 days. Fish were for very other day with tropical fish food (Tetramin; Kordon Corp., Hayward, CA). Seawater was prepared from Instant Ocean synthetic sea salts ; fresh water used in the experiments was dechlorinated tap water. The experiments were performed during August 1979.

Tilapia were maintained for 2 weeks in indoor freshwater aquaria. On day 0 the animals were transferred to 33% seawater for 48 h and subsequently to 100% seawater. The sojourn into 33 % seawater was necessary to prevent mortalities during the seawater transfer process. Four fish were sacrificed on day 0 and eight fish each day were sacrificed at 1, 3, 5, 7, 14 and 21 days following transfer to seawater. Following decapitation, the lower jaw was bisected and the head pinned, dorsal side down, to a dissection board. Each opercular membrane was exposed by holding the operculum away from the head with the aid of dissecting pins between two of the branchiostegal rays and, with frequent perfusion with Ringer solution, was gently peeled away from the opercular bone by teasing the underlying connective tissue with a small, blunt-tipped glass probe. The tissue was lifted away from the operculum, placed in a Petri dish filled with fresh Ringer solution and cut away from the branchiostegal rays. The tissue was then floated above the aperture (area = 0·181 cm²) of a lucite chip coated with silicone grease (Dow Corning Corp., Midland, MI). It was then secured to the chip by eight small pins. A second chip, greased and similarly designed except for possessing eight holes to receive the pins from the other chip, was secured on top of the first chip so that the membrane was sandwiched between the two. After the tissue was mounted, the chips were slid down a central groove in a lucite chamber filled with Ringer solution, effectively partitioning the chamber into two halves. The volume of each half could be varied from 3 to 10 ml. Fluid circulation and gassing were accomplished via bubble lift. The tissue was allowed to equilibrate in the chamber until the electrical parameters stabilized, usually within 1 h. The steady-state open-circuit transepithelial voltage, resistance and current were determined and the data from fish from the same day of acclimation were averaged to determine mean ± standard error (⁠⁠). The second opercular membrane was also dissected and either incubated in DASPEI (2-(dimethylaminostyryl)-i-aethylpyridiniumiodine; ICN Pharmaceuticals, Plainview, NY ; see DASPEI experiments) or similarly mounted in a matching chamber if isotope-flux experiments were being conducted.

Each opercular membrane surface was bathed with tilapia Ringer solution of the following composition (in HIM): 155 Na, 151 Cl, 10 HCO₃, 3 K, 2 Ca, 1 Mg, 1 SO₄, 1 PO₄, 10 Tris (hydroxymethylaminomethane), 10 glucose. The solutions were gassed with 100% O₂. The final pH, 7·6 for freshwater-and 7·9 for seawater-adapted fish, was achieved by HCl titration, except for chloride-flux experiments in which H₂SO₄ was used. The experiments were performed at room temperature (24–26 °C). When epinephrine or SCN⁻ (Sigma; St Louis, MO) was used, stock solutions were prepared immediately before use by dissolving the compound in fresh Ringer solution. A small volume of the stock solution was added to the Ringer bathing the tissue to achieve the desired concentration (epinephrine, 10⁻⁶M; SCN⁻, 10⁻²M) following removel of an equal amount to maintain the bath volume.

Transepithelial potential difference (PD, mucosa ground) was measured via polyethylene 4 % agar-3 M KCl bridges (located within 2 mm of each surface of the tissue) which connected each half of the chamber to a calomel electrode (Corning ; Medfield, MA). Current was passed across the tissue via 4 % agar-3 M-KCl bridges connected to Ag/AgCl half cells which were in turn connected to an automatic dual-channel voltage-clamping device (D. Lee, Co.; Sunnyvale, CA). All bridges were pre-equilibrated for a minimum of 24 h in Ringer solution to minimize KCl leakage during the experiments. Electrode asymmetry was nullified and solution resistance compensated for by the voltage-clamp device. Current or PD was continuously monitored on a chart recorder (Linear Instruments; Irvine, CA). Since preliminary work demonstrated that the voltage deflexions across the tissue responded to current pulses in a linear fashion between –30 and +100 mV, the transepithelial resistance (R) was calculated using Ohm’s law from the voltage deflexion caused by a known brief pulse of current during open-circuit conditions or from the current deflexion caused by rapidly clamping the tissue to + 10 mV during short-circuit conditions. Except for isotope flux experiments, in which the tissue was continuously short-circuited, the open circuit PD and R were used to calculate the current. Although we have found the calculated current (Z_c) is somewhat larger than the short-circuit current (I_sc; generally I_SC/I_C = 0·7), for the purposes of the present study in which several tissues were examined each day, we felt that I_c was a valid way to assess the relative capacity of these tissues to transport chloride actively.

One opercular membrane from each fish was incubated in 2 μm DASPEI for 45–60 min. This fluorescent vital dye specifically stains mitochondria (Bereiter-Hahn, 1976), thereby allowing quantification of ‘mitochondria-rich’ cells. Following the incubation period, the tissue was examined by fluorescence microscopy (Zeiss, HBO 200 W/4). Since the pavement, undifferentiated and basal epithelial cells are relatively lacking in mitochondria (see Results), the mitochondria-rich cells, which concentrate the dye, can be individually identified. The entire epithelium was visually scanned before taking micrographs (Ektachrome, ASA 400) of one or two areas with the highest densities of fluorescing cells and one or two areas considered ‘typical’ or possessing what appeared to be average cell densities for the tissue. The total number of fluorescing cells within each field (0·48 mm²) was counted and expressed per cm².

Determinations of cell densities in areas with the highest density showed no changes during the seawater adaptation process nor any correlation with I_c. It was clear, however, that changes in cell density occurred in other areas. Therefore, although the choice of an ‘average density’ area was subjective, it was felt that the densities determined for these areas were more representative of the total opercular membrane fluorescing-cell population. The densities in two ‘average’ areas were determined for many tissues, and the mean difference between the two areas within each tissue was approximately 14%. It seems likely, therefore, that the error involved in these determinations is acceptable. As a result, analysis and discussion of cell density will pertain only to average-area density determinations.

To analyse the time-course of changes in the fluorescing-cell density following seawater transfer, the cell densities obtained from fish on the same day of acclimation were averaged and expressed as Since the fluorescing cells appear round when viewed from above, fluorescing-cell size was determined by measuring cell diameters to the nearest micron for cells in which a single nucleus was clearly visible following micrograph projection. The average cell diameter for each fish at a particular day of seawater acclimation was averaged with others from the same day and expressed as ⁠.

Following mounting, the tissue was allowed to achieve a stable open-circuit PD and R before being continuously short-circuited. The tissue was again allowed to stabilize before H³⁶Cl (New England Nuclear; Boston, MA) diluted with distilled water and neutralized with NaOH, was added to either side of the epithelium to a final activity of 2μCi/ml. Following a 45–60 min isotope equilibration period, successive 1 ml samples of the cold bath were removed at 15–30 min intervals, mixed with 10 ml scintillation fluid and counted in a scintillation counter for 20 min each. Samples were taken to provide at least 3 flux periods, following which a 0·1 ml sample of the ‘hot’ side was removed, mixed with 0·9 ml ‘cold’ Ringer, added to 10 ml scintillation fluid and counted. All samples removed from the baths were replaced with an equal volume of non-isotopic Ringer solution. In addition, when removing a sample, care was taken to avoid development of any hydrostatic pressure gradients by lowering the bath on the opposite side of the tissue to the same extent. ³⁶Cl-efflux (serosa to mucosa) and influx (mucosa to serosa) were determined on paired membranes from the same fish in matching chambers. I_sc was continuously monitored and R determined at least once every 15 min throughout the experiment. For each tissue pair, a net flux was calculated (efflux-influx) and compared to the average I_sc for the two membranes. The difference between the net flux and I_sc were averaged for five pairs of membranes and compared to zero (paired t-test).

Opercular membranes were fixed in modified Karnovsky’s fixative containing 1 % paraformaldehyde, 3% glutaraldehyde (both Polysciences; Warrington, PA) and 3 mm-Ca⁺² in Na cacodylate buffer (pH 7 · 4) for 2 – 12 h while still attached to the branchiostegal rays. Fixation solutions were diluted 15% for membranes from freshwater-adapted fish. Central areas of the membranes were cut into small squares, postfixed for 1 h with 1 % OsO₄, stained en bloc for 1 h with 1 % uranyl acetate, and dehydrated in alcohol. Samples were embedded in Araldite. Thick sections (1 – 2 μ m) cut with glass were stained with toluidine blue, examined, and areas of interest selected before cutting thin sections with a diamond knife. Thin sections were stained with uranyl acetate and lead citrate and examined on a Jeol JEM-1005 electron microscope at 80 kV.

Table 1 compares the basic electrophysiological properties of opercular membranes isolated from freshwater-and seawater-adapted tilapia. Membranes from freshwater-adapted fish had a high resistance (R) and only small currents (I_c) and voltages (PD ; serosa positive) were generated across them. Steady-state R ranged from 1165 Ω.cm² to 6330 Ω.cm² and steady-state I_c, never greater than 3 μ A/cm², was usually less than 1 μ A/cm². Tissues isolated from fish adapted to seawater for 2 and 3 weeks had a much lower R (120 – 470 Ω.cm²) and substantially higher I_c (38 – 149 μ A/cm²) and PD. For membranes from these sea-water fish, R usually decreased during the first hour in the chamber before stabilizing at a new lower value while I_c generally increased during this period. Subsequently, these parameters remained stable for several hours.

The results of paired ³⁶Cl flux determinations on short-circuited membranes from freshwater- and seawater-adapted fish are presented in Table 2. Unidirectional chloride fluxes across membranes from freshwater fish were small and no net flux was demonstrated (P > 0·10). Both unidirectional fluxes increased for fish adapted to seawater for 10 days. The influx increased nearly threefold, while efflux increasd enfold compared to the freshwater membranes, resulting in a net chloride-extrusion not significantly different from the measured I_sc (P >0 · 10).

To assess the degree of similarity of this chloride transport process with those observed in the Fundulus opercular membrane and the Gillichthys skin, the effects of serosal exposure to SCN⁻ (10⁻²M) and epinephrine (10⁻⁶M) were examined. Both agents, inhibitory of chloride transport in these isolated membranes (Degnan et al. Marshall & Bern, 1980), similarly inhibit I_c in the opercular membranes from seawater-adapted tilapia. SCN⁻ inhibited I_c by 29 · 4 ± 2 · 1% (n = 8). Epinephrine inhibited I_c by 40 · 0 ± 6 · 9% (n = 6) at the peak response within 10 min following its addition to the serosal medium. Usually I_c spontaneously returned to pre-inhibition levels during the next 30 – 60 min. This time-course is similar to that reported for the Fundulus opercular membranes (Degnan et al. 1977). Subsequent experimentation has revealed that this recovery is due to epinephrine oxidation (Foskett & Hubbard, 1981).

The morphology of the tilapia opercular membrane is similar to that previously described for killifish, Fundulus heteroclitus, and sea raven, Hemitripterus americanus (Karnaky & Kinter, 1977). The tilapia opercular membrane is a thin, stratified epithelium with an underlying layer of loose connective tissue (Fig. 1). Within the connective tissue can be found capillaries, occasional bundles of striated muscle and pigment cells. Numerous multicellular sensory receptors are distributed throughout the epithelium which contains the four well identified cell-types previously described in gill and opercular epithelia (Karnaky, 1980): mucous, pavement, chloride and non-differentiated cells. The tilapia opercular membrane has an additional cell-type which is always found resting upon the basal lamina. These ‘basal’ cells form a continuous layer between the rest of the epithelium and the connective tissue. Contacting the external environment, pavement cells form a thin, continuous surface layer interrupted by occasional mucous and chloride cells. Underlying this surface layer, the epithelium is a stratified mixture of non-differentiated and occasional chloride and mucous cells.

The opercular membranes from freshwater- and seawater-adapted fish are similar with the exception of the chloride cell ultrastructure, which is markedly different depending on the fish’s adaptation medium. These changes are described later.

Having established that the current generated by the isolated opercular membrane from the seawater fish represents a chloride-secretory process and that this mechanism seems to be very similar to that of the other isolated opercular and skin preparations, we examined the time-course of the development of this process following transfer of tilapia from fresh water to seawater (Fig. 2). The I_c developed by the epithelia isolated from the freshwater fish (2 · 1 μ A/cm²) was significantly elevated within 24 h after Kasfer of tilapia to seawater and continued to increase until 1 – 2 weeks following seawater transfer, when it reached a new steady-state level of approximately 100 μ A/ cm². Electron-microscopic observations of opercular membranes from seawater-adapted tilapia revealed the presence of large, typical chloride cells (Fig. 3). The chloride cells do not reach the basal lamina as basal cells are always located beneath them. All exhibit apical crypts which may be very deep and elaborate and which contain an amorphous, electron-dense material. These cells have a rich population of large mitochondria containing numerous cristae and are, in this regard, distinct from all other cell-types present in the epithelium. An extensive branching tubular system continuous with the basolateral surface is characteristic of these cells. This tubular system ramifies extensively throughout the cell except for a small perinuclear region associated with golgi membranes and a narrow apical ‘clear zone’ beneath the crypt. Often, the chloride cells are seen in groups of two or three sharing a common crypt. In addition, accessory cells (Hootman & Philpott, 1980) are usually associated with the seawater chloride cell, often observed as extensions projecting into the apical crypt associated with the chloride cell by shallow tight junctions. The ultrastructure of the seawater tilapia opercular chloride cell, like that of chloride cells in other opercular membranes, is essentially identical to that of the gill chloride cell (Philpott & Copeland, 1963; Karnaky & Kinter, 1977).

Chloride cells observed in opercular membranes from freshwater-adapted tilapia are small and poorly developed compared with those observed in membranes from seawater-adapted fish (Fig. 4). Although these cells contain more mitochondria than surrounding epithelial cells, the freshwater chloride cell contains fewer, smaller mitochondria with poorly developed cristae compared with the seawater chloride cell. The tubular systems appear rudimentary. Apical crypts have never been observed, but limited contact of these cells via a small extension of the cytoplasm with the surface of the epithelium is occasionally seen. Accessory cells and associations between the freshwater chloride cells have not been observed.

Since the chloride cells in the opercular membranes from both freshwater- and seawater-adapted tilapia are clearly the only cell-types with appreciable numbers of mitochondria, the number of chloride cells was estimated by counting the number of cells stained with the fluorescent dye, DASPEI, a specific mitochondrial dye. Fluorescing cell density increased in parallel to the increasing I_c during the first 3 days of seawater adaptation but subsequently reached a plateau, or even began to decline, before I_c reached a steady-state (Fig. 5). However, the fluorescing cells appeared to hypertrophy. The cells in the freshwater epithelium (Fig. 6a) had an average diameter 10 · 2 ± 0 · 4 μ m compared to 16 · 7 ± 0 · 5 μ m for 3 weeks adapted seawater fish (Fig. 6b). The time-course of this cellular hypertrophy is shown in Fig. 5. During the initial 3 days in seawater, when the increasing I_c was paralleled by increasing chloride cell density, the cell diameter did not appear to change. Following day 3 of seawater adaptation, however, when the chloride cell density was not further augmented while I_c continued to increase, the chloride cells hypertrophied dramatically and continued to do so throughout the entire 3-week time-course.

This result suggests that cellular hypertrophy is responsible for maintaining the high I_c observed between days 3 and 21 following seawater transfer and implies that larger cells can secrete salt at a higher rate than smaller ones. To explore this hypothesis, the amount of current generated by an individual chloride cell (I_c/cell) was calculated, by dividing the average I_c by the average cell density for a given day of seawater adaptation, and compared with the cell size at that time. Fig. 7 shows that during the first 3 days in seawater, when the chloride cell density and I_c were increasing, I_c/cell increased independently of an apparently unchanging cell size. After day 3, since I_c increased while chloride cell density stayed the same or decreased, I_c/cell increased and this increase was intimately related to the cell size.

Cortisol, the predominant corticoid in teleost fish (Henderson et al. 1970), has level been implicated in the seawater adaptation process (Maetz, 1974). In an attempt to induce the physiological and morphological changes in the opercular membrane of a freshwater fish observed following transfer of tilapia to seawater, a series of daily cortisol injections (5 μ g/g body weight/day for 8 – 10 days) were made into freshwater fish. Cortisol significantly increased the density of fluorescing cells in the isolated opercular membrane compared with saline-injected controls (Table 3). Based on electron micrographs, the fine structure and size of these cells (not shown) is similar to that observed in cells from opercular membranes of uninjected freshwater fish. It is interesting to note that the absolute increase in the cell density induced by cortisol is the same as the maximal change in cell density induced by seawater adaptation and that both treatments caused the cell density to increase approximately 2 · 5 times (× 2 · 25 for cortisol treated and × 2 · 85 for 3-day seawater fish). However, even though the average cell density more than doubled in the cortisol-injected fish compared with controls, there do not appear to be any differences in the electrophysiological properties of the isolated tissues between the two groups (Table 3).

The electrophysiological properties of the opercular membranes from the tilapia Sarotherodon mossambicus are radically different depending upon the state of adaptation of the animal. The opercular membranes from seawater tilapia have high conductances and generate large currents and voltages. As in the Fundulus opercular membrane (Karnaky et al. 1977) and the Gillichthys skin (Marshall & Bern, 1980) the short-circuit current appears to be carried exclusively by chloride since the net isotopic fluxes of chloride are not significantly different from the measured short-circuit currents (Table 2). In addition, epinephrine and serosal SCN⁻ inhibit the chloride current similarly in all three tissues. The average PD displayed by opercular membranes from tilapia adapted to seawater for 2 and 3 weeks (21 · 5 mV) is smaller than that measured in vivo in unstressed tilapia in seawater (35 · 2 mV; Dharmamba, Bornancin & Maetz, 1975). However, tilapia displays a significant PD in 1/3 seawater (14.7 mV ; Dharmamba et al. 1975) where diffusion potentials are minimized, suggesting that, like the opercular membrane, the branchial epithelium possesses an electrogenic pump mechanism. The trans-gill PD displayed by tilapia in seawater is probably the result, therefore, of both an electrogenic pump mechanism and a passive diffusion potential due to the large salt gradient. In contrast to the seawater condition, the isolated membrane from freshwater adapted tilapia is characterized by having a high transepithelial resistance (> 3000 Ω.cm²) and negligible current (< 3 μ A/cm²). Radioisotopic flux measurements have failed to demonstrate any net chloride transport across the freshwater tissue. This is in contrast to results from whole animal studies which have implicated the branchial/ buccal epithelium as the site of salt uptake by freshwater fish. It may be that the celltype responsible for branchial salt uptake is not present or functional in the opercular membrane. Alternatively, since sodium fluxes were not measured and the chloride fluxes were determined under short-circuit conditions, it remains possible that under open-circuit conditions chloride may be transported across the tissue, linked electrically to sodium transport. Such transport would be limited, however, since the PD maintained by the freshwater membrane is small (< 5 mV). Since both unidirectional chloride fluxes across the freshwater membrane are small, the tissue may serve primarily as a relatively impermeable barrier to salt loss from the freshwater fish. Seawater transfer results in increased unidirectional chloride fluxes across the isolated tilapia opercular membrane compared to the freshwater values. Similar increases in unidirectional fluxes following seawater transfer have been consistently observed in whole animal studies of salt turnover in euryhaline fish (see Maetz, 1974), including tilapia (Dharmamba et al. 1975), suggesting that the electrophysiological differences observed between the seawater and freshwater opercular membranes probably reflect changes for the entire branchial/buccal epithelium.

Physiological adjustments employed by euryhaline fish to cope with changing osmotic environments must be relatively rapid to prevent large-scale changes in the internal milieu of the animal. In the present study, the isolated tilapia opercular membranes were found to have an increased chloride current within 24 h after the fish were transferred into seawater (Fig. 2). This current continued to increase to fully adapted seawater levels after 1 – 2 weeks. These measurements provide the first clear demonstration of the changes in net extra-renal salt secretion during the seawater adaptation process in a teleost fish. The time-course of this change is similar to the changes in unidirectional branchial ion fluxes measured in other species in vivo (see Maetz, 1974) and suggests that the observed time-course of the development of chloride secretion by the isolated tilapia opercular membrane mirrors the time-course of net salt secretion across the branchial epithelium during seawater adaptation.

It has long been assumed that chloride cells are the source of this extra-renal salt secretion. Most convincingly, proliferation, hypertrophy and changes in subcellular structure of these cells accompany seawater adaptation (see Introduction for references). In the present study, typical chloride cells, similar to those previously described in branchial and other opercular epithelia (Karnaky, 1980), are found only in the opercular epithelium removed from the seawater-adapted tilapia. In contrast to other opercular and skin epithelia, the chloride cells in the tilapia opercular membrane do not extend to the basal lamina due to the presence of a continuous layer of basal cells which rest upon the basal lamina. It has been suggested (Conte & Lin, 1967; Shirai & Utida, 1970) that basal cells in the gill represent the source of newly differential chloride cells. It is interesting to speculate that these cells serve a similar function in the tilapia opercular membrane.

To establish the chloride cell as the salt-secretory cell in teleosts, we have attempted to correlate the magnitude of the chloride current with the number of chloride cells present in the opercular membrane. Following seawater transfer, the estimated chloride cell density increased in parallel to increasing chloride secretion only during the initial 3 days (Fig. 5). A similar time-course of the change in chloride cell density following seawater transfer has been observed in Japanese eel gills (Shirai & Utida, 1970). The increased number of chloride cells observed in the opercular epithelium appears to be due to an augmentation of the immature chloride cell population characteristic of the freshwater epithelium since the cell size does not appear to increase during this time (Figs. 5, 7). In addition, the salt-secretory mechanism becomes activated, resulting in an increased chloride secretion by tilapia opercular membranes within 24 h of seawater transfer (Fig. 2). After day 3 in seawater and the associated peak in the number of cells, the size of the chloride cells increases dramatically (Figs. 5, 7). The cell diameters agree well with diameters measured in electron microscopy and compare favourably with chloride cell diameters reported by others. Our results demonstrate that cellular hypertrophy is intimately related to the amount of chloride secretion by individual chloride cells (Fig. 7). The immature chloride cell in fresh water appears capable of secreting 3 p-equiv Cl/h, increasing upon seawater adaptation to 116 p-equiv Cl/h for 3-week-adapted fish. The observed correlation between the size of the chloride cell and the amount of chloride secretion strongly implicates the chloride cell as the chloride secretory cell-type in this epithelium.

Chloride cell hypertrophy in the tilapia opercular membrane appears to be due to increased synthesis and incorporation of plasma membrane into the basolateral tubular system (Fig. 4), as also seems to be the case for chloride cells of the branchial epithelium (see Philpott, 1980). In other chloride-secreting tissues, including the Fundulus opercular membrane (Degnan et al. increased intracellular levels of cAMP stimulate secretion, presumably by increasing apical membrane chloride conductance (Frizzell, Field & Schultz, 1979). Phosphodiesterase inhibition, while increasing the conductance of the tilapia opercular membrane, does not lead to enhanced chloride secretion (unpublished data), suggesting that the apical exit step is not rate-limiting for chloride secretion by this tissue. Since the basolateral membrane and its associated transport proteins may, therefore, be rate-limiting for chloride transport across this tissue, and if cell size is determined by the basolateral membrane area, we might expect, as observed, to find a positive correlation between the chloride secretory capacity of an individual chloride cell and its diameter.

The nature of the stimuli or effectors which induce these morphological and physiological changes following seawater transfer is unknown. Hormonal (Doyle & Epstein, 1972; Kamiya, 1972) and neural (Mayer-Gostan & Hirano, 1976) initiation have been proposed, as well as direct induction by increased blood ionic or osmotic concentrations (Mayer & Nibelle, 1970; Olivereau, 1970) or by specific ions or osmotically active substances in the external environment (Liu, 1944). In the present study, injections of cortisol, the predominant corticoid in teleosts, increased the size of the population of immature chloride cells in opercular membranes isolated from freshwater fish. Cortisol injections also augment the population size as well as the degree of differentiation (Doyle & Epstein, 1972), especially as measured by increased Na/K-ATPase levels (see Maetz, 1974), of freshwater branchial chloride cells in other species examined. In addition, stress may also induce changes in the chloride cell population in the freshwater tilapia opercular membrane. For instance, the opercular membranes from tilapia maintained in indoor glass aquaria for 2 weeks and subsequently used in the time-course experiment, while having the same fluorescing-cell density as membranes from undisturbed fish from an outdoor freshwater pond, have immature chloride cells with significantly increased cell diameters compared to the pond fish (Fig. 7). Stress results in elevated blood cortisol levels in fish (see Fryer, 1975), suggesting that stress, cortisol and seawater adaptation may act via common mechanisms to result in increased differentiation of the chloride cell population in the tilapia opercular epithelium.

Although cortisol significantly increases the size of the freshwater chloride-cell population, it is without significant effects on the electrophysiological parameters of the isolated opercular membranes from freshwater tilapia, implying that the cells are not engaged in salt transport while the fish is in freshwater. Another, as yet unknown, stimulus appears to be required to initiate salt secretion. It is possible that the cortisol-induced cells are inactive because they do not make sufficient contact with the external environment. While typical chloride cells in the seawater tilapia opercular membrane are usually observed to be in extensive contact with the surface of the epithelium via large apical pits, contact of the chloride cells in the epithelium of freshwater fish with the external milieu is minimal, whether or not they received cortisol. Alternatively, since it is clear that the freshwater chloride cells do make some contact with the external environment, it may be that the permeability properties of the apical membrane of these cells are insufficiently developed to permit chloride secretion and that the salt-secretory stimulus involves increasing the apical membrane ionic permeability.

The isolated tilapia opercular membrane provides a valuable model for the study of the processes involved in the differentiation of chloride transport by chloride cells. The sequence of events in the opercular membrane following seawater transfer of tilapia appears to involve initially increased numbers and activation of small chloride cells (which therefore possess only a poorly developed basolateral tubular system), as well as activation of quiescent immature chloride cells already present, resulting in augmented chloride secretion during the initial 3 days following seawater transfer. It is unknown at this time whether the increased cell density is due to proliferation of a precursor stem cell or simply reflects partial differentiation of undifferentiated cells already present in the epithelium. Once this population of chloride cells is established, subsequent increased chloride secretion by the tilapia opercular membrane appears to be associated with increased differentiation, growth and maturation of these cells. This differentiation would appear to primarily involve basolateral membrane elaboration (including associated transport proteins) and increased numbers and development chtochondria. As a result, the cell becomes larger and, concurrently, capable of increased chloride secretion. The seawater-adapting hormone cortisol causes chloride cell proliferation but the final activation of these cells to secrete chloride actively involves other, unknown factors.

We thank Dr B. Burnside for use of her electron microscope facilities and J. Underhill for photographic assistance. Portions of this work were presented earlier (Am. Zool. 19, 995, 1979). This research was aided by NRSA CA-09041 traineeship to J.K.F., a MARC award to T.T., and NSF Grants PCM-10348 to H.A.B. and PCM-7725205 to T.E.M.