INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

IT WAS PREVIOUSLY DEMONSTRATED that primary cultures of respiratory cells of the sea bass (Dicentrarchus labrax) gill serve as a good model for studying Cl secreting epithelia (2, 10). This model could also help to clarify the mechanism of ion balance regulation in marine teleosts. Taken together, the results of previous studies (2, 10) have permitted functional characteristics of this cell model to be more precisely determined. In this way, the Na⁺-K⁺-ATPase pump, by maintaining low intracellular Na⁺ and Cl levels, provides the driving force for Cl entry into the cell through the basolateral membrane via Na⁺-K⁺-2Cl cotransport and the Cl/HCO₃ exchange mechanism. Intracellular Cl accumulation above its equilibrium leads to Cl transport across the apical membrane via small conductance Cl channels sensitive to cAMP. It is now well established that transepithelial Cl secretion is also dependent on K⁺ channel activity. These channels are required for the control of membrane potential and also to allow K⁺ ion recycling across the basolateral membrane. Moreover, K⁺ channels could also be involved in cell volume regulation. It was therefore postulated in the present study that K⁺ channels in gill cells could be of significant physiological importance in the control of ionic homeostasis in fish. For these reasons we have undertaken the study of K⁺ channels in primary cultures of sea bass gill cells in an attempt to elucidate mechanisms of transepithelial ion transport.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals

Primary Culture of Gill Cells

The following procedures were performed in sterile conditions under a laminar flow hood. The gill arches (cartilaginous part) were removed, and the remaining filaments (primary lamellae) were dipped in the washing medium L15 (see Solutions and Chemicals). The filaments were washed under gentle automatic shaking (5 × 10 min, 100 agitations/min) and then teased into small tissue fragments (explants). Each explant was inoculated in a plastic culture dish (Nunc; 35 mm diameter, multiwell plates). The culture growth medium L15 was then added gradually to the explants. Explants were then maintained in a precision-controlled low-temperature incubator (Jouan) at 17°C in a humidified air atmosphere (atmospheric PCO₂). The growth medium was changed every second day.

Patch-Clamp Experiments

Single-Channel Experiments

All experiments were performed at room temperature (22 ± 2°C).

Data Analysis

Reversal potentials (E_rev) were estimated by interpolation after a polynomial function of the I-V curve was fitted using Sigma Plot and Sigma Stat software (Jandel Scientific).

The K⁺-to-Na⁺ permeability ratio was calculated from E_rev in excised patches according to the modified equation of Goldman-Hodgkin-Katz
where P_K and P_Na are K⁺ and Na⁺ permeabilities, respectively, and R is the gas constant, T is the absolute temperature, and F is Faraday's constant. The use of this equation to evaluate selectivity requires the tacit assumption that the channel being studied has a negligible selectivity for anions.

Solutions and Chemicals

Primary cultures. All solutions were prepared using tissue-culture quality chemicals.

Solutions for electrophysiological studies. The standard bath and pipette solutions contained (in mM) 140 KCl, 5 HEPES, and 0.1 CaCl₂ (pH 7.8 adjusted with 1M KOH) for control conditions. In some experiments, cells were also bathed in an NaCl solution containing (in mM) 140 NaCl, 5 HEPES, and 0.1 CaCl₂ (pH 7.8 adjusted with 1N NaOH). For determination of the channel ionic selectivity, solutions containing (in mM) 60 KCl, 80 NaCl, 5 HEPES, and 0.1 CaCl₂ or with 100 mM CaCl₂ and 5 HEPES were also used.

9

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Gigaseals were obtained in 20.5% of the cells tested (in all, 521 cells from 56 monolayers). A total of 107 patches exhibited channel activity. However, this activity depended on the age of the culture in that single-channel activity was never recorded from cells aged <7 days, whereas the maximum activity was found in cells from monolayers aged 10 ± 3 days.

Patch-clamp studies performed with 140 KCl in the pipette and bath solutions on the same cell preparation showed a Cl channel with same biophysical characteristics (small conductance <10 pS, potential dependence, etc.) than the Cl channel described in our last paper (10). Because the interest of this small conductance Cl channel was minor in this study dealing with K⁺ conductance present in the gill cells in primary culture, we eliminated their recordings from the data described here. Despite our search to find other types of K⁺ channels, we never recorded, out of 200 single patch-clamp experiments, any other K⁺ channel than the stretch-activated one shown in this study.

Single-Channel K⁺ Current Recorded in Cell-Attached Patch

View larger version (17K): [in this window] [in a new window]   Fig. 1.   A: single-channel current recordings of a K+ channel at different holding potentials (Vp) in a cell-attached patch. Pipette and bath contained 140 mM KCl solution. B: single-channel current recordings of K+ channels after replacing the 140 mM KCl solution in the pipette with 140 mM NaCl solution. C: current-voltage (I-V) relationship of K+ channel in cell- attached configuration: , 140 mM KCl in pipette and bath solutions (n = 24 cells from 20 monolayers); , 140 mM NaCl in pipette and 140 mM KCl in bath solution (n = 3 cells from 2 monolayers); and , 140 mM RbCl in the pipette and 140 mM KCl in the bath solution (n = 4 cells from 2 monolayers).

The maximal channel conductance, calculated as the slope of the straight line between 60 and +60 mV, was 123 ± 3 pS (n = 24). The zero-current potential (1.6 ± 0.5 mV, n = 24) was close to the expected E_rev for K⁺ (assuming an intracellular K⁺ concentration ranging between 120 and 140 mM). This indicates that the observed current was carried by K⁺. The analysis of four independent seals presenting only one open state revealed that the probability of the channel being in the open-state was independent of the membrane potential (P_o at 40 mV = 0.24 ± 0.03; P_o at +40 mV = 0.25 ± 0.01; n = 4). The kinetic analysis of recordings made at 40 mV revealed that the channel showed one open state with a time constant of 27 ± 5 ms (n = 4).

Cell-attached recordings were also performed with the NaCl solution in the pipette and the standard KCl solution in the bath (Fig. 1B). Under these conditions, positive currents were observed from 60 to +40 mV. The corresponding I-V curve for these experiments is shown in Fig. 1C. The relationship was nonlinear over the range 60 to +40 mV, whereas the extrapolated E_rev (78 mV) was close to the expected E_rev for K⁺.

The Goldman-Hodgkin-Katz constant field equation was applied to the data on the assumption that the channel was selective for K⁺. As shown in Fig. 1C, the experimental data are described accurately by the theoretical curve. Assuming the cell-attached configuration, internal potassium concentration is 140 mM and the calculated mean permeability is 2.24 ± 0.07 × 10¹³ cm³/s (n = 3). The maximal channel conductance, calculated as the slope of the curve between 0 and +40 mV, was 96 ± 12 pS (n = 3). The I-V curve in Fig. 1C also illustrates the effect of Rb⁺ when the pipette was filled with an RbCl solution. The substitution of KCl with RbCl did not modify the slope conductance (114 ± 7 pS, n = 4) nor the E_rev (1.3 ± 5.4 mV, n = 4), indicating that the channel was equally permeable to K⁺ and Rb⁺ ions.

Channel activity rapidly declined with time in 100% of the cell-attached recordings, irrespective of the ionic composition of the bath and pipette solutions. This spontaneous run-down phenomenon is illustrated in the recording shown in Fig. 2A. In this experiment, the pipette contained the standard KCl solution and the V_h was maintained at 40 mV. One minute after seal formation, the channel activity was so low that it was often very difficult to obtain a reliable estimate of the mean NP_o. In fact, the time necessary to obtain complete inactivation of channel activity varied from 1 to 3 min. The curve in Fig. 2B shows average results for six cell-attached patches in which inactivation occurred within 1 min.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   A: inactivation of K+ channel activity in cell-attached configuration. Pipette and bath contained 140 mM KCl solution. The pipette potential was held at Vp = 40 mV. C and the dotted line indicate the closed state of the channel. B: mean number of open channel (NPo) plotted against time (NPo was calculated by computer analysis of periods of 5-s duration). Values are means ± SE for 6 experiments.

The mean NP_o decreased with time. After channel activity had been lost, it was possible to restore it by application of a negative pressure to the patch pipette. A typical experiment is shown in Fig. 3A. Increasing the negative pressure >20 mmH₂O produced a rapid and reversible increase in NP_o (Fig. 3C). This stretch activation of the channel was effective only up to a negative pressure of 40 mmH₂O, after which patches usually ruptured. The I-V curve of the reactivated channel (with a negative pressure ranging between 30 and 40 mmH₂O) was linear, with a slope conductance of 132 ± 6 pS (n = 7) and a E_rev unchanged from 0 mV (Fig. 3B). Finally, the main characteristics of the channel before and after stretch activation were very similar.

View larger version (11K): [in this window] [in a new window]   Fig. 3.   A: negative pressure dependence of K+ channel activity for cell-attached configuration. Up to 3 channels are activated simultaneously by increasing negative hydrostatic pressure (applied as suction to the patch pipette) as indicated on left. The pipette potential was held at Vp = 20 mV. Pipette and bath contained 140 mM KCl solution. B: I-V relationship of K+ channel in cell-attached configuration during negative pressure stimulation of between 30 and 40 mmH2O. C: mean NPo as a function of pipette pressure (suction).

To investigate whether these stretch-activated channels could also be activated by osmotic pressure, the effect of cell swelling, induced by exposure of cells to a hypotonic solution, was then studied. The perfusion of six different cell-attached patches with a 200 mosmol/kgH₂O solution brought about an increase in channel activity. A typical recording obtained at a V_h of 20 mV and with the KCl solution in the pipette, is given in Fig. 4A. The mean NP_o increased from 0.05 ± 0.01 to 0.53 ± 0.06, n = 6 (Fig. 4B) within ~25 s of the onset of perfusion of the hypotonic solution. This effect was reversible after washout with an isotonic solution. The NP_o returned to control levels with a similar delay (NP_o = 0.08 ± 0.03, n = 6).

View larger version (31K): [in this window] [in a new window]   Fig. 4.   A: activation of K+ channel in cell-attached patch during change of external hypertonic saline solution to hypotonic solution. Fifty seconds before time zero, the external hypertonic bath medium (350 osmol/kgH2O NaCl) was replaced by a hypotonic medium (200 osmol/kgH2O NaCl). Note the increase in channel activity during the "hypotonic" period. B: mean NPo during hypotonic shock (NPo was calculated by computer analysis of periods of 5-s duration) (n = 6).

Single Channel K⁺ Current Recorded in Excised Inside-Out Configuration

View larger version (26K): [in this window] [in a new window]   Fig. 5.   A: single-channel current recordings of a K+ channel at different holding potentials (Vp) in excised inside-out configuration. Pipette and bath contained the 140 mM KCl solution. C on the right of the recordings indicates the closed state of the channel. Same excised patch after replacing the bath solution with the 140 mM NaCl solution (B) or by the 60 mM KCl + 80 mM NaCl solution (C). D: single K+ channel current trace in excised configuration with 100 mM CaCl2 solution in the pipette and 140 mM KCl solution in the bath. E: I-V relationship of the channel in excised inside-out configuration under symmetrical 140 mM KCl solution in the bath and in the pipette (, n = 13), after replacing the bath solution with 140 mM NaCl (, n = 8) or 60 mM KCl + 80 mM NaCl (, n = 7 ) or after replacing the pipette solution with 100 mM CaCl2 (, n = 7).

Under these experimental conditions, the ion selectivity of this channel could not be accurately determined because K⁺ or Cl was able to induce the same current. To verify the K⁺ selectivity of the channel, inside-out excised patches were perfused on the cytoplasmic side with a K⁺-free solution or a solution containing only 60 mM K⁺. Figure 5B illustrates the single-channel currents recorded when the pipette solution contained 140 mM KCl and the bath medium 140 mM NaCl. At a V_h of 0 mV, channel openings showed downward deflections with an average unitary current amplitude of 2.3 ± 0.1 pA (n = 8). By definition, this had to correspond to the passage of K⁺ ions, because Na⁺ currents would have produced upward deflections and Cl is at electrochemical equilibrium. The mean I-V relationship of eight experiments performed under these conditions is shown in Fig. 5E. From the current E_rev estimated by extrapolation (+74 ± 6 mV), a K⁺-to-Na⁺ permeability ratio (P_K/P_Na) equal to 19 was calculated. The maximal slope conductance was 84 ± 6 pS.

Further experiments were performed by replacing the 140 mM NaCl bathing solution with a solution containing 60 mM KCl and 80 mM NaCl. Figure 5C illustrates single-channel currents recorded under these conditions. The corresponding I-V relationship reported in Fig. 5E displayed an inward rectification over the applied potential range from 50 to +40 mV. In these experimental conditions, the K⁺ Nernst E_rev was calculated at +21.3 mV. The experimental E_rev was determined at +19.8 ± 1.2 mV (n = 7), leading to a P_K/P_Na = 20.8. The average conductance of the channel was 76 ± 4 pS (n = 7).

To examine the permeability of the channel to Ca²⁺ ions, experiments were performed with 100 mM CaCl₂ solution in the pipette and 140 mM standard KCl solution in the bath, respectively. Figure 5D illustrates the kinetic properties of the channel, while the I-V curve in Fig. 5E shows that the currents were always positive for V_h ranging from +20 to 40 mV. The E_rev tended toward that of K⁺; however, it was not possible to record channel activity at potentials more negative than 40 mV because larger hyperpolarizations ruptured the patch seal. Nevertheless, the recorded current was mainly due to K⁺, indicating an absence of Ca²⁺ permeation through the channel. Under these conditions, the maximal slope conductance of the channel was 104 ± 11 pS (n = 7).

As observed in cell-attached experiments, channel activity decreased within 1 to 3 min after the patch was excised. In 17 of the 28 inside-out excised patches, channel activity was completely lost. The application of negative pressure to the patch pipette produced an increase in channel activity that was rapid and reversible on removal of the suction. Such stretch-activation was observed in 11 of 15 patches studied. Figure 6A illustrates this phenomenon for a V_h of 40 mV, when the bath and the pipette solutions contained standard 140 mM KCl. In this typical recording, the application of suction (46 mmH₂O) to the pipette induced the opening of three channels. As illustrated in Fig. 6B, this stretch activation did not depend on the Ca²⁺ concentration at the inside surface of the membrane. In fact, lowering Ca²⁺ to 10⁹ M in the bath (see MATERIALS AND METHODS) did not suppress the stretch activation of K⁺ channels in these cultured cells.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Activation of K+ channels by application of negative pressure to the patch in excised inside-out configuration for 2 different concentrations of free Ca2+ in the bath solution. A: pipette contained the 140 mM KCl solution with 0.1 mM CaCl2 and bath contained the 140 mM KCl solution with 1 mM CaCl2. The holding potential was held constant at Vp = 40 mV. B: pipette was filled with the same 140 mM KCl solution but bath contained 140 mM KCl solution with low concentration of free Ca2+ (1 nM) by addition of EGTA. The holding potential was held constant at Vp = 20 mV. Note that stretch activation of the K+ channel was independent of the cytosolic Ca2+ concentration.

Pharmacological Properties

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Effect of 5 mM BaCl2 on the K+ channel activity in an excised inside-out patch. A: in the control recording, pipette and bath contained the 140 mM KCl solution. C on the right of the recordings indicates the closed state of the channel; "1, 2, 3" and the dotted lines indicate the different open states. In the lower recording, 5 mM BaCl2 was added to the bath solution. B: amplitude histograms of the 2 recordings shown in A. C: effect of 500 µM GdCl3 on the K+ channel activity in an excised inside-out patch. The pipette contained the 140 mM KCl solution and after a control period was perfused with the 140 mM KCl containing 500 µM GdCl3. The bath solution contained 140 mM KCl. D: mean NPo calculated for control conditions, after 140 mM KCl perfusion, and after 140 mM KCl + 500 µM gadolinium perfusion. Values are means ± SE of 5 experiments.

Gd³⁺ is a blocker of stretch-activated channels in a variety of tissues, acting principally at the extracellular surface of the channels. To investigate whether K⁺ channel activity could be blocked by Gd³⁺, the perfusing pipette technique of Tauc et al. (33) was used. Figure 7C illustrates the effect of Gd³⁺ on K⁺ channel activity recorded from an inside-out patch. Under control conditions, two channels were open simultaneously. The perfusion of 500 µM Gd³⁺ through the patch pipette strongly blocked channel openings, thus reducing the mean NP_o from 0.75 ± 0.20 to 0.04 ± 0.02 (n = 5, Fig. 7D). The unitary conductance of the channel, however, remained unchanged.

Quinidine applied to the internal surface of five inside-out patches also induced strong inhibition of channel activity. One example of a recording obtained with 50 and 500 µM quinidine is given in Fig. 8A. Quinidine (500 µM) decreased the current amplitude from 2.3 ± 0.2 to 0.30 ± 0.03 pA (n = 3) and reduced the mean NP_o from 0.41 ± 0.05 to 0.020 ± 0.001 (Fig. 8B). The effect of quinidine was partially reversible when the drug was washed out. The current amplitude was not significantly reduced at a low quinidine concentration (50 µM; Fig. 8, A and B).

View larger version (25K): [in this window] [in a new window]   Fig. 8.   A: effect of quinidine on K+ channel activity in an excised inside-out patch. Under control conditions, the pipette and bath contained the 140 mM KCl solution. The quinidine concentration was increased from 50 to 500 µM in the bath solution. The presence of 500 µM quinidine abolished channel activity, which was partially reversible. B: representation of unitary current (I) and mean NPo under control conditions or in the presence of 50 or 500 µM quinidine. Values are means ± SE for 5 experiments. Statistics: Student's unpaired t-test: data at rinse are significantly different from values after quinidine addition (500 µM) with P < 0.05 (*). C: K+ channel activity for control conditions (left) and with 20 mM tetraethylammonium (TEA) added to the pipette solution (right) for cell-attached conditions. Pipette and bath contained 140 mM KCl solution. No inhibition was observed with 20 mM TEA.

The effect of TEA on the K⁺ channel was studied when the blocker was applied to either the extracellular (added to the pipette solution) or the intracellular surface (added to the bath solution) of the patches. The use of TEA concentration as high as 20 mM did not significantly modify the channel activities recorded under either experimental condition (Fig. 8C).

Dependence on pH

View larger version (20K): [in this window] [in a new window]   Fig. 9.   A: effect of pH variations on K+ channel activity in an excised inside-out patch. Note the inhibition of K+ channels by acidification of the bath solution from 7.7 to 6.8 and a partial reactivation by increasing pH to 8.2. Pipette and bath contained 140 mM KCl solution. The pipette potential was held constant at Vp = 20 mV. B: representation of the mean NPo for 3 different values of pH: 7.7, 6.8, and 8.2. Values are means ± SE for 7 experiments. C: repetitive decrease (from 7.7 to 6.8) and increase (from 6.8 to 7.7) of pH over the time showing a total reversible effect of acidification. D: correlated mean NPo after the same inhibition and stimulation profile.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The present study focused on the identification of potassium channels present on gill cells in primary culture. Patch-clamp experiments were performed on the apical membrane of these cells. This single-channel analysis revealed the existence of a channel that was selective to potassium over sodium ions. In the cell-attached mode and under control conditions (KCl standard solution), the channel was spontaneously active at the resting membrane potential and could permit K⁺ efflux from the cell. The unitary conductance of the K⁺ channel under these conditions ranged between 100 (physiological conditions) and 120 pS (symmetrical K⁺ solutions).

The kinetic properties of the channel in excised configuration were not distinguishable from those seen in cell-attached patches. For both recording configurations and in the presence of identical K⁺ concentrations on both sides of the patch, the inward and outward K⁺ currents were similar. The rectification observed in the presence of asymmetrical solutions was the consequence of a Goldman-type rectification. Moreover, the open state probability of the channel did not exhibit voltage-dependence characteristics over the range of imposed potentials.

The channel was blocked by Ba²⁺ ions, with the kinetics of this inhibition being characteristic of the action of slow blocking agents. Thus Ba²⁺ ions reduced the length of the burst duration and induced long periods of inactivity between any two bursts. Moreover, Ba²⁺ reduced not only the open probability of the channel, but also the amplitude of the current flowing through the channel. Quinidine and gadolinium were also potent inhibitors of the channel, whereas TEA did not influence its activity.

Cytosolic acidification has been shown strongly to reduce open-time probability in many K⁺ channels (8, 12). The maximal channel activity observed in the present study was with an intracellular pH ranging from 7.5 to 7.8. It is possible that a culture medium maintained at pH 7.8 represents physiological pH values in seawater fish gill cells (considering that this pH value is close to that of sea bass plasma). In the present study, the effect of acidification was partially or totally reversible. Repetitive decrease and increase of pH over the time in the same recording induced a reversible effect of acidification showing that the inhibition of the activity of the channel was a real effect and not a run-down phenomenon.

Some interesting findings made here concerned the fact that channel activity exhibited a spontaneous run-down phenomenon that could be reactivated by the application of a negative pressure via the patch pipette. The reactivated channel exhibited kinetic and conductance characteristics very similar to those determined before the inactivation process taking place. This suggests that the pipette suction probably reactivated the same channel and that the K⁺ channel in the present study was a stretch-activated channel. Ion channels activated by membrane stretch have been detected in several epithelial cells (7, 35, 36). Many of these channels are nonselective stretch-activated channels with single conductances ranging between 25 and 35 pS. In primary cultures of gill cells, it appears that the channel was at least 19 times more permeable to K⁺ than to Na⁺ and relatively impermeable to Ca²⁺.

Stretch-activated K⁺ channels with small conductances and insensitive to calcium have been identified in amphibian proximal tubules (15, 16, 27-29). However, none of these channels exhibit characteristics such as those identified here in gill cell K⁺ channels. Maxi K⁺ channels have also been shown to exhibit mechanosensitive properties in different cell preparations. For example, Pacha et al. (23) described a high-conductance K⁺ channel in the apical membrane of intercalated cells of the rabbit collecting duct that shares some of its properties with the K⁺ channel examined in the present study. In particular, the conductance was close to 100 pS and the channel could be activated by pipette suction in the cell-attached as well as inside-out patch-clamp configurations. Moreover, this mechanical activation was not dependent on cytosolic calcium. Some important differences between the two channel types, however, also exist. Notably, the activity of K⁺ channels in gill cells was not gated by voltage and was not sensitive to TEA. Therefore, the mechanical activation could definitely not be attributed to an increase in cytosolic Ca²⁺. This observation led to the conclusion that the channel was activated directly by membrane stretching. It is therefore evident that the channel is considerably different from the large conductance K⁺ channel (maxi K⁺ channels) typically found in other epithelial tissues (4, 22, 24).

Finally, because very little information exists concerning K⁺ channels in fish gill cells, it is not possible to directly compare the K⁺ channel in the present investigation with K⁺ channels described in other reports. The most relevant studies concerning our findings are those of Gögelein et al. (12) and Greger et al. (13) on the rectal gland of the dogfish and of Chang and Loretz (5, 6), who reported on goby enterocytes. The former work revealed that the basolateral membrane possesses a K⁺ channel with a unitary conductance, kinetic characteristics, and rectifying properties very different from the K⁺ channel in cultured gill cells. The latter study described a mechanosensitive cation channel of 70 pS that poorly discriminated between Na⁺ and K⁺ and which may operate rather as an Na⁺ channel. Therefore, to our knowledge, the present work is the first study describing stretch-activated K⁺ channels in epithelial gill cells.

To understand the physiological role of the K⁺ channels in the apical membrane of gill cells, it is necessary to know whether this channel is active at the resting membrane potential. This question is important because the channel inactivated rapidly after seal formation in the cell-attached and inside-out patch-clamp configurations. The inactivation process could be interpreted in light of two different hypotheses. 1) The channel was inactive at rest before seal formation and the application of suction to seal the membrane-induced stretching that activated the channel. Subsequently, when the application of negative pressure was terminated, the patch returned to its inactive state. 2) The channel was active at rest, and its spontaneous activity was, therefore, representative of the cell conductance under normal conditions. After the seal formation, the channel run down observed could have been due to mechanical constraints on the cell membrane induced by the patch pipette.

The importance of the membrane geometry on mechanosensitive K⁺ channels was recently studied by Patel et al. (25). They identified a cloned mammalian two P-domain mechanogated S-like K⁺ channel and demonstrated that the opening of TREK 1 (tandem of P domains in a weak inward rectifying related K⁺ channel) was highly sensitive to the curvature of the membrane. We favor this second hypothesis over the first because, in whole cell experiments (see companion paper, Ref. 9a), a spontaneous decrease of the ionic conductance was never observed. According to the data of Patel et al. (25), it is reasonable to postulate that the cell membrane exhibits a crenated form at rest, which is compatible with channel activity. Soon after the formation of a gigaseal, this notched configuration of the membrane progressively disappears because the membrane beneath the patch becomes more rigid. The crenated form is thus replaced by a cup form (which is not conductive to channel activity). Afterward, the application of a negative pressure reinduces the crenated form and channel activity is restored. The fact that a positive pressure was unable to induce channel activity would tend to favor this explanation.

Considering that the K⁺ channel could be active in resting cells, it may be involved in K⁺ fluxes across the apical membrane of gill cells. We previously demonstrated that primary cultures of respiratory cells of the sea bass gill represent a good model for studying Cl-secreting epithelia (2, 10). In accordance with classical models for electrogenic chloride secretion, the secretion of K⁺ takes place across the basolateral membrane and serves to recycle K⁺ ions accumulated in the cytosol by the Na⁺-K⁺-ATPase and the Na⁺-K⁺-2Cl cotransporter mechanisms (13). However, K⁺ channel activity has been recorded in the apical membrane of some chloride-secreting epithelia, including lacrimal acinar cells (31) and vas deferens epithelial cells (30). Moreover, K⁺ fluxes have been also measured across the isolated skin of the marine teleost Gillichthys mirabilis (20) and seawater opercular epithelia (21).

Using an equivalent circuit model, Cook and Young (9) examined the effect of an apical K⁺ conductance on the secretion of fluid by an epithelium in which secretion was driven by the secondary active transport of Cl. They concluded that a K⁺ conductance in the apical membrane could enhance secretion with a maximal effect when the proportion is ~10-20% of the total K⁺ conductance. In the marine teleost, this effect could be of importance, because the gill cell must always excrete high quantities of Cl against a strong electrochemical gradient. Therefore, an apical K⁺ conductance could improve Cl secretion by maintaining a high potential difference across the apical membrane.

Many studies have demonstrated that stretch-activated channels are also activated by hypotonic shock and could be implicated in the cell volume regulation (see Ref. 14 for review). Different types of K⁺ channels could be involved in such regulation. First, osmotic shock could modulate Ca²⁺ influx through stretch-activated cation channels, whereas a local increase in cytosolic Ca²⁺ could activate Ca²⁺-sensitive K⁺ channels. This hypothesis was first proposed by Christensen (7), whereas a similar mechanism had already been postulated to explain regulatory volume decrease in different epithelial cell types (17, 19, 26, 32, 34, 35). Second, mechanical stress on the cell membrane induced by the osmotic shock could open two different types of truly stretch-activated K⁺ channels. These K⁺ channels are either Ca²⁺ sensitive (18, 23) or Ca²⁺ insensitive (11, 28). The K⁺ channel described in the present study probably belongs to the second category. Two types of Ca²⁺-insensitive stretch-activated K⁺ channels have been demonstrated in the basolateral membrane of Necturus proximal tubule. These channels exhibit a lower conductance than that of the channel described here, but interestingly they are activated by pipette suction and by osmotic swelling (11, 16).

In the cell-attached experiments described here, exposure of cells to a hypotonic solution induced an increase in channel activity. The kinetics of this form of channel activation were similar to that obtained when negative pressure was applied to the patch pipette. Thus, in cultured gill cells, the stretch-activated K⁺ channel is regulated by both mechanical pressure and osmotic change, suggesting that channel activation could be involved in the control of cell volume following hypotonic shock.

Perspectives

In a companion paper (9a), we investigate potassium conductances and fluxes in both apical and basolateral membranes and their behavior in response to exposure of the membranes to a hypotonic medium. For this purpose, whole cell patch-clamp and ⁸⁶Rb⁺ efflux experiments on primary cultures of gill cell were employed.