Neural Modulation Of Salt Secretion In Teleost Opercular Epithelium By α₂-Adrenergic Receptors And Inositol 1,4,5-Trisphosphate

Euryhaline teleost fish in sea water excrete NaCl from chloride cells in the skin and gill epithelia. On entry to fresh water or dilute brackish water, these salt secretory systems are rapidly turned off during the initial phase of acclimation to the hypotonic environment. Estuarine euryhaline teleosts, such as the killifish Fundulus heteroclitus, normally experience rapid salinity changes and thus are good models for the regulation of salt transport in teleosts. The ‘instantaneous’ reduction in Cl⁻ secretion seen in euryhaline teleosts on entry into fresh water (Motais et al. 1966; Pic, 1978) cannot be explained on the basis of a change in trans-gill electrical potential, while the parallel decrease in Na⁺ efflux is in part explained by a reduction in electrical potential (e.g. Potts and Evans, 1973; Pic, 1978). The reduction in Cl⁻ efflux could be controlled in a variety of ways, for instance by the actions of hormones, by direct cellular effects or by the nervous system. This study investigates the last possibility.

Early nerve sectioning experiments in fish, in which cranial nerves IX and X innervating the gill were severed, induced changes in ion fluxes and water permeability (Mayer-Gostan and Hirano, 1976), thus implicating the autonomic nervous system in control of the gill ion transport. A general review of autonomic function is given by Donald (1997). Nilsson and Petterson (1981) demonstrated that autonomic nerve stimulation of Atlantic cod Gadus morhua primarily caused vasoconstriction of arteriovenous pathways in the gill vasculature, but no transport parameters were measured. Exposure of isolated perfused head preparations and studies in vivo indicate that systemically applied catecholamines universally dilate the gill vasculature (examples include Nilsson, 1984; Perry et al. 1984; for reviews, see Nilsson and Holmgren, 1993; Donald, 1997). Donald (1987), using catecholamine fluorescence histochemistry, observed that adrenergic neurons innervate various parts of the gill vasculature of several species, including one marine species (the toadfish Tetractenos glaber), and that nerve endings are found in the core of the gill filaments between the surface epithelium and the wall of the central venous sinus. The nerves in this area form a subepithelial plexus in the core of the gill filament. In turn, this is the major location for gill chloride cells in both seawater and freshwater species. Donald (1987) predicted that the most likely effector for these sympathetic neurons would be the chloride cells of the filament epithelium and that these neurons, because of their location, were clearly not vasomotor. Adrenaline has been shown to affect the solute permeabilities of gills in several species (for a review, see Isaia, 1984; Mayer-Gostan et al. 1987) and to affect the functional surface area for transport (Nilsson, 1984), although it is difficult to distinguish vasomotor from more direct transport functions in isolated perfused systems or in vivo. Perry et al. (1984) observed increases in rates of Cl⁻ influx in isolated perfused heads from freshwater trout and concluded that the effect was mediated by α-adrenergic receptors and was probably not a purely vasoactive response. The anatomical patterns of innervation are well-described for rainbow trout (Oncorhynchus mykiss) and perch (Perca fluviatilis). Sympathetic neurons routinely track along the cranial nerves supplying the gill (Dunel-Erb et al. 1993); hence, stimulation of (mixed) branchial nerves should activate sympathetic fibers. However, there are very few studies that have actually stimulated the nerves innervating teleost gills to investigate their function. In one such study, adrenaline and branchial nerve stimulation of freshwater rainbow trout gills in a ‘nonperfused’ preparation that was insensitive to vasoactive responses inhibited Ca²⁺ influx (Donald, 1989), suggesting that adrenergic nerves can directly affect ion transport in the gills of freshwater fish. However, there is no direct evidence to indicate autonomic nervous control of NaCl transport in freshwater or seawater fish.

The opercular epithelium and skin of marine teleosts has been a good model for the operation and control of chloride cells, which may be multifunctional and transport NaCl as well as other ions, such as K⁺, Ca²⁺ and NH₄⁺ (for a review, see Marshall and Bryson, 1998). The preparation has demonstrated sensitivity to a variety of neurotransmitters and hormones (for a review, see Marshall, 1995) and it lends itself conveniently to an examination of the effects of these substances in a system that lacks the vasomotor responses typical of the gill. This research was undertaken to provide evidence for at least partial control of salt secretion in fish via the autonomic nervous system using experiments on isolated opercular epithelia of killifish in which the nerve supply was dissected intact and stimulated directly. The sympathetic action is shown to be direct (not via cholinergic pathways) and consistent pharmacologically with the previously demonstrated α₂-adrenoceptors, and the response is present in fish that are wholly and partly adapted to sea water. These results have appeared in part in abstract form (Marshall et al. 1995).

Adult killifish (mummichogs, Fundulus heteroclitus L.) were captured in Antigonish estuary, transferred to indoor holding facilities and adapted to full-strength sea water (salinity 30 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 (Tetramarine, Tetra Werke, Germany) at a rate of 1.0 g 100 g⁻¹ body mass day⁻¹ supplemented twice weekly with frozen brine shrimp. Animals were killed by pithing, and the opercular epithelium was dissected. Opercular skin pieces were dissected from the underlying opercular bone so that the epidermis and dermis (with chromatophores and nerves) were included. Two small branches of the anterior main branch of the trigeminal nerve innervate the opercular area. A third branch of the nerve trunk, one that tracks anteriorly and innervates the roof of the buccal cavity, was severed so that the remaining portion could be dissected with the skin. The nerve trunk was severed at the point where it exits the cranium, and the nerve was gently drawn through a foramen in the bones of the roof to the mouth so that it came away cleanly with the skin and was approximately 7 mm long.

Freshwater adaptation involved transfer of killifish from brackish-water (salinity 3.0 g l⁻¹) holding facilities (where the animals were kept for at least 1 week) to dechlorinated, ultraviolet-sterilized Antigonish tap water (NaCl 0.15–0.30 mmol l⁻¹, Ca²⁺ 0.08–0.12 mmol l⁻¹, pH 5.5–6.5) in glass aquaria. The acclimation period was at least 2 weeks. The transfer of freshwater-adapted animals back to sea water was accomplished by isolating a single freshwater-adapted animal in a 5 l bucket for 2 days, then introducing a flow of sea water at the same temperature over a period of 10 min to complete the change to full-strength sea water. Animals were allowed to acclimate for 48 h in full-strength sea water before experimentation; this provided a partially acclimated group to examine the transition between freshwater and seawater acclimation.

The epithelium with the nerve was mounted in a modified Ussing chamber so that transmembrane electrophysiological variables could be monitored (V_t, transepithelial potential in mV; R_t, transmembrane resistance in Ω cm²; and I_sc, short-circuit current μA cm⁻²) whilst a separate circuit could be used to stimulate the nerve. Epithelia were clamped to 0 mV except for short periods to record V_t. The epithelium was supported by a nylon mesh and pinned out over the aperture (area 0.238 cm²), with the rim area lightly greased and beveled to minimize edge damage. Epithelial variables were measured using a current voltage-clamp (D. Lee Co., Sunnyvale, CA, USA) and nerve stimulation was applied using square-wave stimulators (Grass Instruments, models SD9 and S44, Quincy, MA, USA) arranged in series to allow various pulse trains to be applied. The nerve was placed across the electrodes which, in turn, were embedded in the bottom of a groove in the membrane-mounting insert, adjacent to the central aperture in which the epithelial surface was exposed in the Ussing chamber. The groove ensured that the nerve was not damaged when the two halves of the membrane insert were assembled. In this way, epithelial transport electrophysiology could be measured while the nerve trunk leading to the epithelium was stimulated separately.

In the series where stimulation frequency is assessed (see Fig. 1), pulses were applied at 1, 3.3, 5, 10, 33, 60 and 100 Hz for 30 s to all membranes, but the different frequencies were applied in random order, each frequency being tested once per preparation with a recovery period between tests of at least 5 min. The transient reductions in I_sc were quantified by taking the integral of the curve over time to the point when the I_sc returned to the initial level. Inhibition is reported as the integral of the current change (μA cm⁻²) over time (min), so the units are μA min cm⁻². In the pulse train series, each preparation was exposed to a train of pulses of 10–25 min total duration until a steady-state level of inhibition had been reached. The trains (2 s of 1 ms pulses at 10 Hz with an 8 s rest) were continued for the entire test period, followed by a recovery period of at least 20 min.

The membranes were dissected and bathed in a modified Cortland’s saline composed of (in mmol l⁻¹): NaCl, 130; KCl, 2.55; CaCl₂, 1.56; MgSO₄, 0.93; NaHCO₃, 17.85; NaH₂PO₄, 2.97; and glucose, 5.55. The saline had a pH of 7.8 when equilibrated with a 99 % O₂/1 % CO₂ gas mixture. The experimental temperature matched that of the acclimation conditions (20–25 °C). Serosal and mucosal baths were 4.0 ml and the membranes were mounted in symmetrical saline that was continuously aerated during the experiments. Yohimbine, propranolol, isoproterenol, clonidine, atropine and tubocurarine (all from Sigma Chemical Co., St Louis, MO, USA) were dissolved freshly in saline in a stock solution at 100–500 times the final concentration and were added to the serosal bathing solution.

Opercular epithelia were dissected and incubated for 10 min in Cortland’s saline substituted with 10.0 mmol l⁻¹ LiCl (replacing part of the NaCl) to inhibit a terminal monophosphodiesterase (Liedtke, 1992, 1994). The total time of 10 min included periods following addition of clonidine in test membranes. Clonidine (1.0 μmol l⁻¹) was added to well-stirred vials of the LiCl-pretreated epithelia with approximately 5 mg of tissue per 2.0 ml vial. Control (LiCl) and test (LiCl plus clonidine) vials were well aerated and stirred at 22 °C. The reaction was stopped by the addition of 80 μl of ice-cold perchloric acid (2 g perchloric acid:8 ml water) 0.5, 2.0 or 5.0 min after clonidine addition, and the tubes were then centrifuged for 1.0 min in a microcentrifuge to pellet the tissue. Duplicate samples of neutralized supernatant and InsP₃ standards were incubated with [³H]InsP₃/bovine cerebellar InsP₃ receptor (DuPont NEN NEK-064) for 1.0 h at 5 °C, centrifuged at 5 °C (10.0 min at 2000 g), and the supernatant was drained. The pellet was solubilized in strong base and counted in a liquid scintillation counter (Canberra Packard CA3000, Downer’s Grove, IL, USA). The previously pelleted tissue was analyzed for protein content by the modified Lowry method (Lowry MicroMethod Kit 690-A, Sigma). InsP₃ content was expressed as pmol InsP₃ mg⁻¹ protein.

Values are expressed as the mean ± 1 S.E.M. Comparisons between means dealing with percentage data used the Mann–Whitney U-test, whereas those between directly measured variables used t-tests that were generally paired, since two opercular membranes were mounted simultaneously from the same animal, one serving as the test and the other as a parallel running control. In drug additions, t-tests were paired as control versus treatment on single preparations.

The nerve trunk that supplies the opercular epithelium was traced back to the brain and was identified as the second branch of the trigeminal nerve (cranial V). Initially, bursts of 200 pulses of 1 ms duration were applied at different frequencies to determine an effective rate of stimulation. The pulse voltage for each preparation was increased until the threshold was exceeded, as indicated by a clear reduction in membrane current; the modal voltage was 10 V. A firing rate of 10 Hz proved to be optimal to produce a reduction in membrane current (Fig. 1). The standard protocol for the pharmacological testing was a single 40 s burst of pulses at 10 Hz (400 pulses each of 1 ms duration). This protocol, however, produced transient inhibitions typically lasting 4–6 min.

Pulses of 1 ms duration at 10 Hz applied in trains of 2 s on, 8 s off, produced more sustained inhibitions of membrane current (Fig. 2). There was a delay of 40–60 s before the onset of the reduction in I_sc. The mean reduction in membrane current was 27.3±3.3 % (from 177.5±11.9 to 129.0±11.5 μA cm⁻², N=22) and the maximal effect occurred approximately 5 min after the onset of stimulation. The steady-state level of inhibition after continuous stimulation of the nerve for 25 min was 9.3±1.9 % inhibition.

The nicotinic and muscarinic cholinergic blocking agents tubocurarine (10 μmol l⁻¹) and atropine (10 μmol l⁻¹), respectively, had no effect on the inhibition of membrane current produced by electrical stimulation of the nerve (Fig. 3). The β-adrenergic antagonist propranolol, at 1.0 and 10 μmol l⁻¹, also did not affect the inhibition by the nerve. However, yohimbine, an α₂-adrenergic antagonist, significantly reduced the effect at 1.0 μmol l⁻¹, and the reduction in the effect was more marked at 10 μmol l⁻¹ yohimbine (Fig. 3). At 100 μmol l⁻¹ yohimbine, the effects on transmembrane current during neural stimulation were also markedly blocked (Fig. 2).

Whereas seawater-adapted animals showed a clear inhibition of membrane current in response to neural stimulation (Figs 1–3), freshwater-adapted preparations, which have very low resting membrane currents, responded to neural stimulation with only a slight increase in transepithelial current (Fig. 4). The lack of responsiveness in freshwater-adapted animals could reflect an absence of target cell receptors and/or of responsive transport systems. To clarify this, freshwater-adapted animals were transferred to sea water for 48 h; animals so treated had a resting I_sc intermediate between those of the fully acclimated freshwater and seawater animals (Fig. 4). There was a significant inhibition of transepithelial current upon neural stimulation in these animals (18±8.6 %, N=14), but the percentage (and absolute) inhibition was significantly smaller than that of fully seawater-adapted animals.

Previous work has indicated that α₂-adrenergic receptors act via intracellular Ca²⁺, mostly derived from intracellular Ca²⁺ pools (Marshall et al. 1993), suggesting an effect mediated by the phosphoinositol cascade pathway. To confirm this, we examined InsP₃ content during adrenergic stimulation (Fig. 5). In a pairwise comparison of InsP₃ content in tissues that were treated with LiCl (10.0 mmol l⁻¹), to inhibit InsP₃ degradation, or with LiCl and a maximal dose of clonidine (1.0 μmol l⁻¹), to mimic the adrenergic response, there was a significant increase in InsP₃ levels at 30 s, but no significant change in InsP₃ levels at other times after clonidine treatment (Fig. 5). The InsP₃ peak corresponds to a time just before detectable inhibition of transmembrane current by clonidine, as would be expected if InsP₃ were in the second messenger pathway leading to transport inhibition.

The existence of adrenoceptors on chloride cells of killifish opercular epithelium was demonstrated by Degnan et al. (1977), who observed rapid, profound inhibition of Cl⁻ secretion in response to the addition of noradrenaline to isolated opercular epithelia. Similar responses and a dose-dependency of catecholamine action were observed in seawater goby (Gillichthys mirabilis) skin (Marshall and Bern, 1980). β-Adrenergic receptors are also present, and isoproterenol rapidly stimulates Cl⁻ secretion, but nonspecific agonists evoke primarily the inhibitory response (Degnan et al. 1977; Marshall and Bern, 1980), so the dominant physiological response appears to be an inhibition of salt secretion. The present work establishes that the response may be initiated by a sympathetic reflex.

Results using a novel nerve–epithelial preparation indicate that sympathetic nerve fibers in the trigeminal nerve that innervates the opercular epithelium, when stimulated electrically, rapidly inhibit Cl⁻ secretion (measured as epithelial current) in seawater-acclimated Fundulus heteroclitus. The optimal stimulation frequency (10 Hz) seen in the present work is the same as that observed by (Nilsson, 1984) when measuring changes in gill vascular resistance and by Donald (1989) when measuring Ca²⁺ transport by the gill. Our preparation seemed to be less responsive at lower and higher stimulation frequencies (Fig. 1). At lower stimulation frequencies, neurotransmitter is presumably degraded before it reaches the receptors, whereas at more rapid stimulation rates, there could be fusion of action potentials or exhaustion of neurotransmitter reserves in the nerve endings that would make the stimulation less effective. If the pattern of epithelial innervation is similar to that in other transporting epithelia of teleosts (Donald, 1987; Dunel-Erb et al. 1993), then the nerve endings probably terminate below the epithelial basal lamina, so that the neurotransmitter must diffuse across this barrier to reach the target cells. The onset of the response is therefore delayed by 40–50 s (see Fig. 2).

Because yohimbine blocked the response while the β-antagonist propranolol was without a blocking effect, we can conclude that the receptor is of the α₂-adrenergic subtype. This is consistent with previous pharmacological identification of α₂-adrenergic receptors in this system (May and Degnan, 1985; Marshall et al. 1993). Since the cholinergic antagonists were also ineffective, the adrenergic receptors must be on the target chloride cells rather than on cholinergic interneurons. The lack of a cholinergic response to stimulation of the nerve is unexpected because of the well-described convergence of muscarinic cholinergic and adrenergic pathways in this epithelium (Rowing and Zadunaisky, 1978; May and Degnan, 1985). The function of the muscarinic receptors in the epithelium remains unclear. Yohimbine at high doses almost completely inhibited the effects of nerve stimulation; hence, it is unlikely that other putative neurotransmitters of the autonomic nervous system of fish are involved, notably 5-hydroxytryptamine and vasoactive intestinal polypeptide (Nilsson and Holmgren, 1993).

The effect of clonidine on α₂-adrenergic receptors is mediated by intracellular Ca²⁺ from intracellular sources (Marshall et al. 1993), and it was hypothesized that the phosphoinositide pathway could be involved. To test this hypothesis, we measured InsP₃ levels directly. To improve the resolution of transient and possibly small changes in InsP₃ content, LiCl was used to inhibit the breakdown of InsP₃ and the time courses were observed carefully. After clonidine had been added, there was a transient peak in InsP₃ level slightly before the onset of inhibition of the current by clonidine. The time resolution is such that the InsP₃ would have adequate time (approximately 30 s) to release Ca²⁺ and for the Ca²⁺ to act. These transient peaks in inositol phosphate levels are typical of α-adrenergic action in ion-transporting epithelia, such as the α₂-adrenergic responses of rabbit tracheal epithelial cells (Liedtke, 1992, 1994), which show a peak in InsP₃concentration at 10–40 s, and renal epithelial cells (Gesek, 1996), which show a peak in the first 2 min. These results indicate that the mediation of the α₂-adrenergic response is via second messengers generated by the production of phospholipase C, by augmented InsP₃ levels (Fig. 5) and by release of Ca²⁺ from thapsigargin-sensitive intracellular stores (Marshall et al. 1993).

The neural response appears to be most well developed in seawater-acclimated animals where the unstimulated I_sc is maximal. Partially adapted animals (48 hours in sea water) had lower resting I_sc and had less marked inhibition of I_sc on neural stimulation (Fig. 4). The 48-hour transfer animals had a large capacity for transport, as indicated by the high I_sc that develops on stimulation of β-adrenoceptors using isoproterenol. Maximal I_sc in the 48-hour group was approximately equivalent to fully adapted seawater animals. Therefore the seawater acclimation process includes development of the inhibitory autonomic response, probably by upregulation of α₂-adrenoceptors.

Whereas adrenaline and noradrenaline are both active pharmacologically and rapidly inhibit opercular epithelium Cl⁻ secretion in vitro (Degnan et al. 1977; Marshall and Bern, 1980; May and Degnan, 1985; Marshall et al. 1993), it is not clear whether the inhibition of Cl⁻ secretion is connected to a systemic release of adrenaline into the blood in a stress-like response or whether the control of Cl⁻ secretion could be via the autonomic nervous system in a ‘normal’ reflex less associated with stress. There is some evidence to suggest that changes in circulating catecholamine levels alter ventilation rates in fish (Randall and Taylor, 1991), while others have shown that trout experiencing mild stresses (hypoxia and hypercapnia) can change their ventilation rates without changes in plasma catecholamine levels (Perry et al. 1992). The latter work is consistent with an autonomic reflex pathway for neural control of ventilation distinct from systemic stress responses. Pic (1978) noted that surgically stressed Fundulus heteroclitus showed a significantly smaller adaptive decrease in Cl⁻ efflux upon entry into fresh water and an increased overall permeability to Cl⁻ compared with animals that were allowed a long recovery period from surgery. Clearly, stress responses in this situation would not be an advantage to survival on entry into fresh water. Chloride cell ion transport is more sensitive to adrenaline than to noradrenaline (e.g. Marshall et al. 1993), so alarm responses to adrenaline would probably inhibit Cl⁻ secretion but, because of the detrimental effects of stress responses, those animals that routinely experience salinity changes may use the sympathetic reflex as a significant part of their adaptation. Conceptually, then, the adaptation of a euryhaline teleost to salinity change could be as natural and unstressful as human sweating in response to heat.

Supported by NSERC Canada grant to W.S.M., by a James Chair Fellowship to C.M.L. and by the St.F.X. University Council for Research. Many thanks to A. L. MacDonald and J. Gilfoy for prize-winning animal care (best small A.C.F. in Canada, Canadian Council for Animal Care 1995).