Control Of Epithelial Cl⁻ Secretion By Basolateral Osmolality In The Euryhaline Teleost Fundulus Heteroclitus

Euryhaline teleosts that enter fresh water respond by rapidly reducing the rate of ion secretion and the rate of passive ion loss across the gill and skin epithelia, but the mechanisms by which these events occur are not well understood. It is well known that blood ion levels and osmolality rise when euryhaline teleosts are transferred to higher salinities and that they fall in animals transferred to more dilute solutions. Early experiments involving salt depletion and salt loading demonstrated appropriate increases in Na⁺ efflux with salt loading and decreases with salt depletion (for a review, see Maetz, 1974), but these effects were not connected to blood osmolality per se. Salinity transfer experiments that examined the acclimation of killifish to sea water demonstrated an elevation of serum osmolality in the first 2 days, after which plasma ion levels and osmolality were corrected (Jacob and Taylor, 1983; Zadunaisky et al., 1995; Marshall et al., 1999).

An increase in [NaCl] and/or hypertonicity on the basolateral side of opercular epithelia in vitro (to mimic early events during the adaptation to hypersaline conditions) evokes an immediate increase in the rate of secretion of Cl⁻ (Zadunaisky et al., 1995). This effect cannot be mimicked by a similar change in the apical bathing solution, suggesting that the cells are responding specifically to a blood parameter. However, because the osmolality of the control Ringer used in the previous experiments was 250 mosmol kg⁻¹ and the normal osmolality of the plasma of seawater mummichog is 290–308 mosmol kg⁻¹ (Zadunaisky et al., 1995; Marshall et al., 1999), the ‘hypertonic’ state may actually have been a recovery from an induced hypotonic state. This makes the results difficult to interpret physiologically. Nevertheless, changes in plasma osmolality may be an important early cue that directly affects ion-transporting cells and can affect Cl⁻ secretion. In turn, augmentation of Cl⁻ secretion may depend in part on the degree of phosphorylation of the basolateral Na⁺/K⁺/2Cl⁻ cotransporter (Flemmer and Forbush, 1999) and/or the apically located cystic fibrosis transmembrane conductance regulator (CFTR) anion channel (Singer et al., 1998). These findings are consistent with observations of the effects of hypertonicity in mammalian renal medullary cells (for a review, see Burg et al., 1997) that connect hypertonic shock with alterations in cell function and gene expression.

However, euryhaline teleosts also adapt rapidly to hypotonic solutions. In sea water, there is a well-established NaCl secretion mechanism involving a basolateral Na⁺/K⁺/2Cl⁻ cotransporter and Na⁺/K⁺-ATPase in series with anion channels that are similar to the CFTR anion channel in the apical membrane, and Cl⁻ secretion follows this transcellular pathway while Na⁺ is secreted passively via a paracellular route (for reviews, see Wood and Marshall, 1994; Marshall and Bryson, 1998). Transfer of mummichogs to fresh water produces a sharp decrease in the rate of Na⁺ efflux that starts in the first 30 min and declines more slowly over 4–8 h (Potts and Evans, 1967; Motais et al., 1966). The ‘instantaneous’ reduction that occurs in the first 30 min is followed by a more protracted reduction in the rate of ion loss and augmentation of ion uptake during full adaptation to fresh water (Wood and Marshall, 1994). We have shown that a portion of the rapid reduction in the rate of ion secretion may be mediated by α₂-adrenergic receptors that are activated by the sympathetic nervous system (Marshall et al., 1998c) and that these α₂-receptors act via phospholipase C and inositol trisphosphate and intracellular Ca²⁺ (Marshall et al., 1993, 1998c). Neural stimulation in vitro did not completely inhibit Cl⁻ secretion (Marshall et al., 1998c), so other inhibitory mechanisms presumably also contribute.

To examine the possible role of hypotonic shock in the shutdown of Cl⁻ secretion when teleosts enter fresh water, whole animals were transferred to fresh water and their blood ion levels monitored (to determine the extent of hypotonicity). Opercular epithelia isolated in vitro were exposed to basolateral hypotonic shock while electrophysiological aspects of ion transport were measured. Because cell swelling and the phosphorylation state of ion transporters are probably involved, the protein tyrosine kinase inhibitor genistein was applied to initiate a mechanistic investigation of the reduction in the rate of ion secretion in response to hypotonic shock. Preliminary results have been presented elsewhere (Marshall and Bryson, 1997; Marshall et al., 1998a).

Adult mummichogs (common killifish, Fundulus heteroclitus) of both genders were captured in Antigonish estuary, transferred to indoor holding facilities and adapted to full-strength sea water (salinity 3 g l⁻¹) for at least 10 days at 20–25 °C and ambient photoperiod under artificial light. Fish were fed marine fish food blend (Nutrafin flakes and tubifex worms, R. C. Hagen, Montreal) at a rate of 1.0 g 100 g⁻¹ body mass day⁻¹, supplemented twice weekly with frozen brine shrimp.

Transfer to fresh water involved the transfer of seawater-acclimated mummichog to dechlorinated, ultraviolet-sterilized Antigonish tapwater ([NaCl], 0.15–0.30 mmol l⁻¹; [Ca²⁺], 0.04–0.08 mmol l⁻¹; pH 5.5–6.5) in glass aquaria. Pairs of seawater animals were placed into 10 l aquaria of full-strength sea water at 20 °C for 7 days (to acclimate to the new surroundings), and fresh water was then introduced as a flow of water at the same temperature over a period of 10–12 min to complete the salinity transfer. Animals were removed after 6, 12, 18 and 24 h and anaesthetized in buffered (pH 7.1) 1:5000 tricaine methane sulphonate (MS222) in 150 mmol l⁻¹ NaCl. Blood was collected from the severed caudal peduncle into heparinized haematocrit tubes. Plasma [Na⁺] was measured on triplicate dilutions using atomic absorption spectrophotometry (Varian AA-375).

Animals were anaesthetized in 0.2 g l⁻¹ MS222 buffered to pH 7 and killed by decapitation. The paired branchial epithelia were dissected from the medial surface of the left and right opercular bones using fine forceps while the epithelium was kept moist with Cortland’s saline (composition given below). The membranes were mounted over a round aperture 0.125 cm² in diameter held in place by five insect pins and supported by an open-weave nylon mesh. The 2 mm rim adjacent to the aperture was bevelled and coated with a thin layer of high-vacuum grease to minimize edge damage to the preparation. The insert was mounted in a modified Ussing chamber surrounded by a water jacket controlled at 22±1 °C, stirred vigorously by magnetic fleas and aerated with 99 % O₂/1 % CO₂. Hemichamber volume was 4.0 ml. Voltage asymmetries were zeroed, and solution resistance (approximately 100 Ω) was compensated in the voltage clamp. Transepithelial potential (V_t in mV), transmembrane resistance (R_t, in Ω cm², calculated from voltage responses to transepithelial current pulses of 5.0 μA) and short-circuit current (I_sc, μA cm⁻²) were measured using current/voltage clamps (D. Lee Co., Sunnyvale, CA, USA, or WP Instruments DVC 1000). I_sc was expressed as positive for secretion of anions, and epithelia were clamped to 0 mV except for short periods to record V_t. Seawater opercular epithelia were deemed acceptable if they developed a steady-state I_sc greater than 50 μA cm⁻² and had an R_t of at least 80 Ω cm² to eliminate damaged membranes. Typical values were a steady-state I_sc of 125 μA cm⁻² and an R_t of 100 Ω cm².

In each case, paired membranes from one animal were mounted, with one serving as a parallel running control, and allowed to come to a steady-state I_sc over approximately 30–40 min. The treatments were then applied, and the membrane was allowed to come to a new steady-state I_sc, usually within 30–40 min. The control membrane received the drug vehicle or, in the case of genistein, equimolar treatment with daidzein. In some cases, subsequent treatments were applied and the membranes were again allowed to come to steady state.

A modified Cortland’s saline (305 mosmol kg⁻¹, pH 7.8) was used to bathe both membrane surfaces symmetrically; its composition was (in mmol l⁻¹): NaCl, 160; KCl, 2.55; CaCl₂, 1.56; MgSO₄, 0.93; NaHCO₃, 17.85; NaH₂PO₄, 2.97; glucose, 5.55. The saline had a pH of 7.8 when equilibrated with a 99 % O₂/1 % CO₂ gas mixture. Hypotonic shock was applied to the basolateral bathing solution by flow-through of 7–10 times the chamber volume using a dilution of the modified Cortland’s saline. The standard hypotonic shock applied was a step from 305 to 234 mosmol kg⁻¹ (a change of 71 mosmol kg⁻¹), accomplished by flow-through of diluted Cortland’s solution (three parts normal Cortland and one part deionized water). V_t measurements made under these asymmetrical conditions (dilute Cortland on the basal side) were not corrected for junction potentials because these were less than 1 mV, as measured against a flowing 3 mol l⁻¹ KCl half-cell. The change in solution resistance due to the dilution was also negligible (approximately 2 Ω cm²). The hypotonic treatment is approximately the same reduction in plasma ion content and osmolality as that 6 h after direct transfer to fresh water.

To test for effects of ion depletion versus hypotonicity, an ion-depleted isotonic solution was prepared by the addition of mannitol (71 mmol l⁻¹) to the diluted Cortland to bring the osmolality back to 305 mosmol kg⁻¹. The isotonic ion-depleted Cortland and hypotonic Cortland were applied to paired membranes using the same flow-through protocol, with the latter serving as the positive control.

The low-[Ca²⁺] regime involved a nominally Ca²⁺-free basolateral solution and an apical solution with enough Ca²⁺ to maintain the integrity of intercellular junctions in the epithelium (as judged by the maintenance of transepithelial resistance). The basolateral solution was Cortland’s saline with 10 μmol l⁻¹ CaCl₂ and 70 μmol l⁻¹ EDTA added instead of the normal Ca²⁺ content; this produced a calculated Ca²⁺ activity of less than 1.0 μmol l⁻¹. The apical solution was Cortland’s saline with a reduced Ca²⁺ content, 0.1 mmol l⁻¹ Ca²⁺ added instead of the normal 1.56 mmol l⁻¹. These solutions were introduced by flow-through of 7–10 times the chamber volume.

Samples of opercular epithelia were dissected, cut into four (approximately 2 mg wet mass) and incubated in 400 μl of Cortland’s saline (300 μl for hypotonic treatment) to equilibrate for 30 min. Control membranes were not disturbed, while test membranes were treated with 1.0 μmol l⁻¹ clonidine, 1.0 μmol l⁻¹ isoproterenol, 10 μmol l⁻¹ forskolin, hypotonic shock (addition of deionized water to 400 μl) or 10 μmol l⁻¹ forskolin plus hypotonic shock. After incubation for 10 min at 20 °C, tissue samples were placed in boiling water for 5 min and processed according to the low-range acetylation protocol for cyclic AMP radioimmunoassay, but with all volumes halved (Biotrak RPA509, Amersham International). The tissue pellets were assayed for protein content using a modified Lowry technique (Lowry micromethod kit 690-A, Sigma Scientific). The cyclic AMP content was expressed as pmol mg⁻¹ protein.

The phytoestrogen genistein, 5,7-dihydroxy-3-(4-hydroxyphenyl)-4H-1-penzopyran-4-one, which inhibits protein tyrosine kinases, and its inactive analogue daidzein, 7-hydroxy-3-(4-hydroxyphenyl)-4H-1-penzopyran-4-one (both from Sigma), were dissolved in dimethylsulphoxide (DMSO) and added at a final concentration of 100 μmol l⁻¹. The final DMSO content was less than 0.2 μl ml⁻¹ and had no effect on electrophysiological variables. Isoproterenol and forskolin (Sigma) were dissolved in saline.

Data are presented as the mean ±1 S.E.M. Comparisons between treatments were performed using paired or unpaired t-tests (as appropriate) with an α value of 0.05 to indicate the minimum statistical significance level.

Mummichogs transferred from full-strength sea water to Antigonish tapwater showed a significant reduction in plasma [Na⁺] at 6 h (Fig. 1), but this had recovered by 12 h and then remained stable until 24 h after transfer. The reduction was approximately 29 % or 59 mmol l⁻¹ Na⁺. Assuming equal reductions in anion complement and taking the activity coefficient to be 150–200 mmol l⁻¹, the reduction represents a decrease of approximately 94 mosmol kg⁻¹. Clearly, there is sufficient reduction in plasma ionic constituents to produce a hypotonic shock to the cells.

Opercular epithelia mounted in vitro in membrane chambers were selectively exposed to hypotonic solutions with stepwise decreases in osmolality by flow-through of diluted samples of the serosal bathing solution. The mucosal bathing solution was unchanged. A reduction in serosal bathing solution osmolality of 20 mosmol kg⁻¹ significantly decreased Cl⁻ secretion rate (measured as I_sc) by approximately 25 % compared with control flow-through of iso-osmotic bathing solution (Fig. 2A). Further decreases of 40 and 60 mosmol kg⁻¹ inhibited I_sc further. A decrease of 60 mosmol kg⁻¹ reduced I_sc by approximately 65 %. The example shown in Fig. 2B demonstrates the reversibility of the hypo-osmotic effect when flow-through of dilute Cortland’s saline (Hypo; a step change of 60 mosmol kg⁻¹) was followed by flow-through of full-strength saline (Iso).

To ensure that the effect was not the result of a reduced concentration of Na⁺ and Cl⁻ on the basolateral side, paired membranes were exposed either to iso-osmotic/iso-ionic Cortland’s saline or to an iso-osmotic/hypo-ionic Cortland’s saline with reduced [NaCl] but supplemented with mannitol (Fig. 3). A reduction in [NaCl] from 160 to 120 mmol l⁻¹ with mannitol added to maintain osmolality had no effect on I_sc (Fig. 3). However, hypotonic shock given after this pretreatment (by flow-through of the diluted bathing solution for the control and test, diluted to 75 % of normal) produced significant reductions in I_sc that were the same for the mannitol-supplemented and control membranes. Hence, the reduction in I_sc is connected to the osmolality of the serosal bathing solution and not to similar changes in NaCl content.

The reduction in I_sc was reversible by reintroduction of isotonic solutions. In a separate group of membranes, I_sc before hypotonic shock was 183.3±11.9 μA cm⁻² (N=6), decreased after hypotonic shock to 112.1±9.3 μA cm⁻² (P<0.001 compared with controls) and recovered to 85 % of the original current 155.8±10.7 μA cm⁻² when isotonic solutions were reintroduced. The decrease in I_sc was accompanied by a significant decrease in transepithelial conductance G_t, in this case from 9.77±0.7 to 6.41±0.47 mS cm⁻² (P<0.002, paired t-test N=6), which recovered to 92 % of initial levels (8.97±0.52 mS cm⁻²) when isotonic solutions were reintroduced. There was no significant change in V_t after hypotonic shock; V_t was 18.9±0.96 mV initially, 17.98±1.88 mV (P>0.70, paired t-test compared with the control period, N=6) after hypotonic shock and 17.42±0.81 mV (P>0.05 compared with control period) on return to isotonic bathing solutions.

Assays of cyclic AMP after hypotonic shock indicated no change in cyclic AMP content compared with controls or with tissues exposed to the α₂-adrenergic agonist clonidine. Both these treatments reduced Cl⁻ secretion rate (measured as I_sc; Fig. 4). As a positive control, the β-adrenergic agonist isoproterenol increased cyclic AMP content approximately 14-fold, accompanied by an increase in I_sc compared with the clonidine and hypotonic pretreatments. Forskolin (10 μmol l⁻¹) by itself increased cyclic AMP content to 362.2±43.8 pmol mg⁻¹ protein (N=6), which did not differ (P>0.5, unpaired t-test) from the value for membranes treated with forskolin plus hypotonic shock (389.1± 34.2 pmol mg⁻¹ protein; N=6); hence, hypotonic shock does not reduce resting or forskolin-stimulated levels of cyclic AMP in opercular epithelia.

To test for possible mediation of the hypotonic effect by intracellular Ca²⁺, the Ca²⁺ ionophore ionomycin (1.0 μmol l⁻¹) was added followed by hypotonic shock and compared with a parallel running control membrane from the same animal (Fig. 5). DMSO had no detectable effect, while ionomycin reduced I_sc to 65±3.9 %. Hypotonic shock applied after ionomycin, however, produced a significant reduction in I_sc (Fig. 5), and the percentage reduction was similar to that produced in control membranes. Subsequent restoration of normal osmolality produced a rebound increase in I_sc both in ionomycin-treated membranes and in controls (Fig. 5).

Disruption of Ca²⁺ responses was also attempted in a low-[Ca²⁺] regime (100 μmol l⁻¹ Ca²⁺ on the apical side and approximately 1.0 μmol l⁻¹ on the serosal side) with and without added thapsigargin (1.0 μmol l⁻¹) to deplete intracellular pools of Ca²⁺. The low-[Ca²⁺] regime by itself had no effect on I_sc (Fig. 6), indicating that the integrity of tight junctions was maintained and that the rate of ion secretion was unaffected. An hour of exposure to thapsigargin and the DMSO vehicle had little effect on I_sc; in both cases, there was a non-significant reduction of 10–15 % from the starting I_sc. Basolateral hypotonic shock inhibited I_sc in control and thapsigargin-treated membranes to the same extent (45 % for the thapsigargin-treated membranes and 49 % for the DMSO control), and there was no significant difference between the control and treated membranes in responsiveness to hypotonic shock.

The protein tyrosine kinase inhibitor genistein (100 μmol l⁻¹, basolateral side) was applied to membranes in isotonic and hypotonic conditions. Genistein in isotonic conditions significantly inhibited I_sc by 31 % (P<0.001 compared with the DMSO control), while the DMSO control did not change (Fig. 7). Subsequent hypo-osmotic shock inhibited I_sc further to 63 % of the initial I_sc. Hypotonic shock inhibited I_sc by 54 % (P<0.005 compared with genistein), but there was no additional effect with the subsequent addition of genistein (100 μmol l⁻¹, serosal side). The net effect was that hypotonic shock before or after genistein treatment produced approximately the same total inhibition of I_sc, i.e. the effects of hypotonic shock and genistein were not additive. Genistein did not block the inhibition of I_sc by the α₂-adrenergic agonist clonidine (1.0 μmol l⁻¹, serosal side), with clonidine producing more than 80 % inhibition compared with the initial I_sc. Interestingly, when genistein was added to the DMSO control membranes after clonidine inhibition, there was a small but statistically significant increase in I_sc from 35.4±15.6 to 53.7±19.4 μA cm⁻² (P<0.01, paired t-test). Finally, addition of forskolin (10 μmol l⁻¹, basolateral side) significantly increased I_sc in all three groups to approximately the same extent (Fig. 7).

Daidzein is a phytoestrogen closely related to genistein, differing only at carbon 7 in having an additional hydroxyl group. It is used as a control substance for the action of genistein on protein tyrosine kinase and is inactive in this respect (Lavens et al., 1992). Genistein, but not daidzein, inhibited I_sc in a separate set of membranes (Fig. 8), but neither drug blocked the subsequent inhibition of I_sc by clonidine or stimulation of I_sc by forskolin.

When mummichogs are transferred directly from sea water to fresh water there is a decrease in plasma [Na⁺] (Fig. 1) and in osmolality, the decrease averaging 29 % at 6 h after transfer. Hypotonic shock applied to the basolateral side of the isolated opercular epithelium, to mimic this decrease in plasma osmolality, reversibly (see Results and Fig. 5) inhibited ion secretion in a dose-dependent fashion with the maximal effect being approximately 70 % inhibition of I_sc (Fig. 2A). The effect is osmotic because a similar reduction in the NaCl content of the bathing solution but with mannitol added to maintain osmolality eliminated the effect (Fig. 3).

Plasma osmolality changes in euryhaline species have been observed for many years, but they have only recently been connected with the control of rates of transport. Zadunaisky et al. (1995) observed significant increases in the I_sc of mummichog opercular epithelium when hypertonic bathing solutions were applied to the basolateral side. Similar hypertonicity of the apical bath was without effect. Since then, preliminary reports have suggested that the reverse effect occurs in response to hypotonic bathing solutions (Zadunaisky et al., 1997). When mummichogs are transferred directly from fresh water to sea water, we found that our normal saline was in fact hypotonic (because the plasma osmolality rises after transfer), and the full transport capacity was observed only when the osmolality of the bathing solution was matched to the measured osmolality of the plasma at that particular stage in the acclimation (Marshall et al., 1999). The fact that these osmotic effects can be evoked in vitro is a demonstration that they are not mediated by neural or hormonal signals and are probably direct effects on individual cells or possibly parahormonal effects within the tissue. However, two common parahormonal effectors, ATP and adenosine, have no effect on isolated opercular epithelial transport (W. S. Marshall, unpublished observations). The hypotonic effect does not completely eliminate the I_sc, but in vivo Cl⁻ secretion stops completely, so it is likely that there are additive effects with other regulatory systems. The inhibition of Cl⁻ secretion by the α₂-agonist clonidine was unaffected by hypotonic shock (Figs 7, 8); hence, this independent adrenergic action could be additive to the osmotic response. The adrenergic inhibition is probably neural (autonomic) rather than hormonal because firing of the autonomic nerves innervating the tissue produces substantial inhibition by α₂-adrenoceptors (Marshall et al., 1998c). The combined action of adrenergic stimulation and direct hypotonic shock effects could account for the full elimination of Cl⁻ secretion within an hour of exposure to fresh water.

The reduction in I_sc is accompanied by a decrease in G_t but little, if any, change in V_t (see Results), suggesting that the inhibition is caused primarily by a reduction in the activity of a conductive pathway in the epithelium. This could represent a shutdown of either the apically located CFTR anion channel (Marshall et al., 1995; Singer et al., 1998) or the basolaterally located Na⁺/K⁺/2Cl⁻ cotransporter. There is a preliminary report of large changes, with salinity acclimation of mummichogs, in the absolute amount and degree of phosphorylation of the Na⁺/K⁺/2Cl⁻ cotransporter in the gills (Flemmer and Forbusch, 1999). In addition, CFTR activity is known to vary with its level of phosphorylation (Gadsby and Nairn, 1999) and with membrane trafficking of CFTR to the plasma membrane (Lehrich et al., 1998). At present, it is not clear which transporter is affected by hypotonic shock.

In fully adapted freshwater mummichog, the opercular epithelium does not secrete ions; instead, there is evidence for active Cl⁻ uptake (Marshall et al., 1998b). In whole-animal flux experiments with freshwater-adapted mummichog, rates of Cl⁻ uptake appear to be low while rates of Na⁺ uptake are greater, and the affinity for the uptake of both ions is low compared with that of freshwater species (Patrick et al., 1997). While the mechanisms involved in freshwater acclimation remain to be resolved for mummichog, it is clear that elimination of Cl⁻ secretion is a major early response during freshwater acclimation and that this decrease is mediated in part by plasma hypotonicity.

In a related paper (K. Daborn and W. S. Marshall, in preparation), data are presented to implicate the active closure of apical crypts by pavement cells, an effect initiated by hypotonic shock and mediated by actin microfilaments that may account for the reduction in passive ion loss by the opercular epithelium shortly after transfer from sea water to fresh water. In this combination of events, both active ion secretion and diffusional ion loss are minimized to the benefit of the animal in its hypotonic environment.

Genistein is a well-known inhibitor of protein tyrosine kinase (Lavens et al., 1992). When added to the basolateral side of the opercular epithelium, it inhibited I_sc in a manner that was not additive to the hypotonic response, suggesting that the hypotonic response involves inhibition of tyrosine kinase (Fig. 7). The inhibition of I_sc was not evoked by an analogue of genistein, daidzein (Fig. 8), which does not inhibit protein tyrosine kinase (Lavens et al., 1992).

Because genistein has some unrelated actions, these must also be considered. Genistein, but not daidzein, binds to and directly activates CFTR anion channels in a manner additive to cyclic AMP and protein kinase A (PKA) activation of CFTR (Illek et al., 1996; Hwang et al., 1997; Weinreich et al., 1997; for a review, see Gadsby and Nairn, 1999). The action may be associated with inhibition of serine/threonine protein phosphatase attached to CFTR (Reenstra et al., 1996). This effect puts into question reports of stimulation of rates of Cl⁻ transport by genistein that were attributed to inhibition of protein tyrosine kinase (Illek et al., 1995; Lehrich and Forrest, 1995; Lehrich et al., 1998; Shuba et al., 1996). Recently, genistein has also been reported to inhibit cyclic AMP phosphodiesterase in a neural cell line (Nichols and Morimoto, 1998), an effect that would also tend to activate cyclic-AMP-mediated stimulatory pathways. Genistein does have stimulatory effects on opercular epithelia as well, but only if the ion transport has been previously inhibited. Of interest here is the observation that genistein inhibited Cl⁻ secretion, and did not stimulate it.

The effect of genistein in the mummichog opercular epithelium is probably an action involving inhibition of protein tyrosine kinase, rather than other reported actions. The action of genistein in untreated membranes is an inhibition of Cl⁻ secretion, not a stimulation, as seen in other anion-secreting systems (Illek et al., 1996; Weinreich et al., 1997; Zhou et al., 1998), so direct activation of CFTR is not the action in teleosts. Also, genistein was added basolaterally, not on the apical side where the drug could have the required access to the extracellular side of CFTR for CFTR stimulation (Zhou et al., 1998). Genistein is an activator of Ca²⁺-dependent K⁺ channels in vascular smooth muscle (Xiong et al., 1995), but in the opercular epithelium Ca²⁺ mediation (e.g. by the Ca²⁺ ionophore ionomycin; Marshall et al., 1993) produces not activation of K⁺ channels (which would increase Cl⁻ secretion) but rather inhibition of Cl⁻ secretion. Finally, daidzein, an inactive analogue of genistein that does not inhibit protein tyrosine kinase (Lavens et al., 1992), was inactive in this preparation. To summarize, genistein appears to act in the opercular epithelium by inhibiting protein tyrosine kinase, possibly by reducing tyrosine phosphorylation of CFTR and thus inhibiting Cl⁻ secretion. This conclusion is supported by the reported activation of CFTR in mouse fibroblasts by tyrosine kinase p60c-src (Fischer and Machen, 1996).

Hypo-osmotic shock did not affect resting or forskolin-stimulated cyclic AMP levels in opercular epithelia (Fig. 4), so it is unlikely that the osmotic response is mediated by changes in cyclic AMP levels or the actions of protein kinase A. However, agents that increase cyclic AMP levels (forskolin and isoproterenol) can override hypotonic inhibition (Figs 4, 7), so the effects of hypotonic shock would seem to be independent of cyclic-AMP-mediated events. The lack of effect of clonidine on cyclic AMP levels is consistent with the action of α₂-adrenoceptors in this system, which inhibit ion transport via an increase in intracellular [Ca²⁺] (Marshall et al., 1993), and the release of Ca²⁺ appears to be via elevation of inositol trisphosphate levels and, presumably, phospholipase C (Marshall et al., 1998c). Similar α₂-adrenergic actions have been reported in another epithelial system, the mammalian distal convoluted tubule (Gesek, 1996).

The Ca²⁺ ionophore ionomycin (but not A23187) inhibited Cl⁻ secretion, and Ca²⁺ apparently contributed to the inhibition through α₂-adrenoceptors in the epithelium (Marshall et al., 1993). We therefore tested Ca²⁺ mediation of the hypotonic response. Mediation of the hypotonic response by intracellular Ca²⁺ is unlikely because flooding the cells in the membrane with Ca²⁺ using ionomycin had little effect on the response (Fig. 5). In addition, Ca²⁺ depletion on the basolateral side by itself or in combination with depletion of intracellular Ca²⁺ stores using thapsigargin (Fig. 6) failed to block the hypotonic effect.

The relationship between the inhibition of transmural transport described here and cell volume regulation in chloride cells is unknown. It is likely that chloride cells undergo a regulatory volume decrease in response to hypotonic shock; such responses characteristically include activation of KCl transport that is phosphorylation-dependent (Berenbrink et al., 1997; for a review, see Cossins and Gibson, 1997). Efflux of KCl and organic osmolytes such as taurine carries fluid out of the cell osmotically and restores cell volume. Many cells employ a volume-sensitive outwardly rectifying anion channel as part of the regulatory volume decrease (for a review, see Okada, 1997); however, this channel does not appear to be present in mummichog chloride cells (Marshall et al., 1995). The volume regulatory response of chloride cells, which probably involves K⁺, Cl⁻ and organic osmolytes exiting the cell across the basolateral membrane, will be an interesting area for future investigation.

The research is in part the Honours BSc of T. Luby and was supported by the Natural Sciences and Engineering Research Council of Canada through grants to W.S.M. We thank A. L. MacDonald for animal care.