A Potassium-Selective Channel in Isolated Lymnaea Stagnalis Heart Muscle Cells

The resting and active conductances of excitable cell membranes are mediated by protein channels which are specific to varying degrees (Latorre & Miller, 1983) for particular ions. The resting conductance constitutes the framework against which transiently activated conductances must operate to alter the membrane potential. For molluscan muscle cells, the nature of the resting conductance mechanisms is largely unknown, although some studies have indicated that the resting membrane is permeable to potassium ions (e.g. Burnstock, Greenberg, Kirby & Willis, 1967; Twarog, 1967; Hill, Greenberg, Irisawa & Nomura, 1970). A more detailed examination of the ionic basis of the resting potential in the heart ventricle muscle cells of the freshwater snail Lymnaea stagnalis has suggested that the resting conductance has potassium, chloride and sodium components and can be adequately described by a form of the Goldman-Hodgkin-Katz equation. The ratio of potassium to sodium permeability (P_K/P_Na) was found to average about 14, and the slope over a wide range of potassium concentrations was about 50mV/10-fold change (Brezden & Gardner, 1984).

In this study, we have examined the conductances of these Lymnaea cells using the gigaohm seal patch clamp technique (Hamill et al. 1981). To facilitate the investigations, enzymatic dispersion was used to isolate the heart ventricle cells. The isolated cells are suitable for patch clamping (and, therefore, presumably free of basement membrane) and can be maintained for several days in vitro.

Single-channel recording from the ventricle cells has allowed us to begin addressing the following questions. (1) Are there ion channels which are spontaneously active at the cell’s resting voltage, and, if so, to which ions are these channels permeable? (2) What factors modulate the kinetics or conductance of these channels? (3) How are these channels distributed in the cell membrane? Answers to these questions are particularly important in understanding the behaviour of myogenic tissue such as the Lymnaea heart ventricle.

Hearts were removed from adult Lymnaea stagnalis (see Brezden & Gardner, 1984, for details). The atria and aortae were cut away and the ventricles were placed in normal Lymnaea saline (Table 1) containing 5 mmol I⁻¹ glucose, 200 i.u. ml⁻¹ of penicillin and 2i.u.ml⁻¹ of bacitracin (antibiotic/glucose saline, AGS). The ventricles were cut into approximately 1 mm pieces with Vannas spring scissors and placed in 15 ml conical centrifuge tubes containing 5 ml of 0·25% trypsin (Sigma, type XII-S) in AGS (2–4 ventricles per tube). The tissue fragments were treated for 30 min at room temperature and subsequently centrifuged at 115 g for 10 min. The trypsin was replaced with a 04 % solution of collagenase (Sigma, type II-S) in AGS and digested for a further 2h. Throughout these procedures, the tubes were gently agitated in a vortex mixer to facilitate dispersion of the cells. Following collagenase treatment, the tubes were centrifuged and the dispersed cells were resuspended in 5 ml of AGS. The cells were plated on sterile glass coverslips in 35 mm Petri dishes. The coverslips had been previously treated with 3 % (v/v) HC1 for 30 min and 0·1 % (w/v) Na₂CO₃ for 30 min. They were sterilized by exposure to dry heat (200°C) for 12h (Jacobson, 1977). The Petri dishes were plated at a density of about two hearts per dish. The isolated cells were kept at ambient room conditions.

The resting potential of 2-day-old isolated cells was measured using 3 mol l⁻¹ KCl-filled glass microelectrodes and standard electrophysiological techniques (WPI model KS700 microelectrode amplifier). Prior to recording, the AGS saline was replaced with normal saline.

Patch microelectrodes were fabricated from Corning 7052 (Kovar sealing) glass (1·65mm o.d., 1·15mm i.d.). After coating with Sylgard 184 (Dow Corning) the tips were fire-polished. It was necessary to apply a few cmHg of negative pressure to obtain gigaohm seals (gigaseals). We had much less success in obtaining gigaseals using Pyrex or soda glass pipettes. Recordings were made from cell-attached patches using standard techniques (Hamill et al. 1981) with a List EPC-5 patch clamp amplifier at a gain of 50 mV pA⁻¹. This amplifier was also used to establish and maintain transpatch potentials. Data were filtered at 4 kHz with a four-pole Bessel filter and recorded on a Racal FM tape recorder for subsequent digitization and analysis on a Digital PDP-11 computer using threshold detection software. Event selection and measurement were done using semi-automated, user-inspected routines.

The effects of variations in the external (intra-pipette) potassium, chloride and sodium concentrations on channel currents and conductances were examined. The saline compositions are given in Table 1. The 100 mmol l⁻¹ potassium saline necessarily produces a higher ionic strength than the other salines [the ionic concentrations of 100 mmoll⁻¹ potassium saline (zero sodium) was 221·5 mmoll⁻¹ whilst that of the other salines was 124·7 mmol l⁻¹]. 30 mmol l⁻¹ potassium saline contained 21·6 mmol l⁻¹ sodium (vs 50 mmol l⁻¹ in normal saline). In all experiments, the bath solution was normal Lymnaea saline. Note that this saline has a low ionic strength (i.e. about one-tenth that of sea water), and, therefore has a relatively low conductance.

The effects of 4-aminopyridine (4-AP, 1–10 mmol l⁻¹, Sigma), tetraethylammonium (TEA, 1–10 mmol l⁻¹, Sigma) and quinidine (0·25–1 mmol l⁻¹, Sigma) on channel currents were also investigated. Prior to patch clamping, the AGS saline was replaced with normal saline. The experiments were performed at room temperature (20–23 °C).

Following the formation of a gigaseal, the cell membrane was stretched by applying suction through the patch electrode via a side-port on the pipette holder; suction was created with a syringe attached in series to a mercury manometer for monitoring pressure.

The isolated cells attached to the glass substrate within minutes of plating and remained viable for up to 5 days at room temperature with no interim changes of the medium. Within the first day in vitro, they formed projections which attached to the glass substrate. The cells, as they appeared on the first and third day following isolation, are shown in Fig. 1. Cells of this morphology were selected for patch clamping.

The resting potentials of L. stagnalis cells 2 days after isolation ranged from —50mV to —67mV [N = 30, mean =—59 ± 5 (s.D.) mV], This value is almost identical to the resting potential of these cells in situ (—61 ±5 mV, Brezden & Gardner, 1984).

In the following results, the ‘transpatch’ potential (membrane potential = V_m) is given as the mean resting potential (we used a rounded value of —60 mV) minus the pipette potential. Generally, patching was more successful on 2-day-old or older cultures than on freshly isolated cells.

With normal Lymnaea saline (1·6 mmol l⁻¹ potassium) in the pipette, channel currents were not evident at the resting potential, but were detected upon depolarization of the membrane potential (Fig. 2). The predominant channel species, which was observed in every patch, was characterized by a conductance of 33 pS (ranging from 29 to 36pS, N = 5 patches). The channel openings often appeared in bursts with the current magnitude increasing steadily with increased depolarization. There was no apparent inactivation during the sustained voltages used to clamp the membrane potential.

Only about 20% of the patches examined exhibited inward currents with normal saline (1·6 mmol l⁻¹ potassium) in the patch pipette, even with the membrane hyperpolarized to —120 mV. Generation of l/V relationships was restricted to data from these patches. Both inward and outward channel currents were, however, readily seen with elevated extracellular potassium (Fig. 3). Fig. 4 shows data for the patches where current reversals were seen with normal saline in the pipette. The reversal potential of these channels (as determined from the interpolated zerocrossing point of the l/V relation) ranged from —50 to —74mV [mean — 60·4 ± 10·1 (S.D.) mV, assuming a mean resting potential of —60mV].

The potassium and probably chloride equilibrium potentials in these cells are close to the resting potential (Brezden & Gardner, 1984). To distinguish which of these ionic species is the predominant charge carrier through this channel, the potassium concentration in the pipette was varied while maintaining a constant chloride concentration.

An increase in the extracellular potassium concentration from 1·6 to 30 mmol l⁻¹ (constant chloride and osmolarity) resulted in a 35 mV shift of the equilibrium potential in a depolarizing direction from —60 to —25 mV (25·2 ± 7·6mV). With 100 mmol l⁻¹ potassium in the pipette (elevated chloride and osmolarity, zero sodium), the reversal potential was found to be at about +4mV (4 ±3 mV) (Figs 4, 5, 6). These changes represent a 27 mV slope per decade change in the external potassium concentration between 1·6 and 30 mmol l⁻¹, and a slope of 57 mV per decade change between 30 and 100 mmoll⁻¹ potassium. The channel conductances (as determined from the slope of the 1/V relationships) in 1 · 6 and 30 mmol l⁻¹ potassium were 33 pS (ranging from 29 to 36 pS, N = 5 patches) and 30 pS (ranging from 20 to 37pS, N = 4 patches), respectively, whereas the conductance in 100mmoll⁻¹ potassium was 84pS (ranging from 64 to 90pS, N = 4 patches). In changing from 1·6 to 100mmoll⁻¹ potassium, the chloride concentration was increased from 62·6 to lllmmoll⁻¹. Consequently, if the channel was chlorideselective, a hyperpolarizing shift of the reversal potential would be expected. However, the shift in the reversal potential was in a depolarizing direction, as expected for a potassium-selective channel. Furthermore, a change in the potassium concentration from 1-6 to 30 mmol l⁻¹ (keeping the chloride concentration constant) also shifted the reversal potential in a depolarizing direction; no shift would have been expected for a chloride channel. These results preclude the possibility that the channel is chloride-selective.

Elevated external potassium concentration also resulted in the appearance of inward currents at the resting potential (Fig. 7), showing directly that gating can occur in the absence of membrane depolarization.

The substitution of bis(2-hydroxyethyl) dimethylammonium chloride (BDAC) for sodium (potassium = 1·6 mmol l⁻¹) resulted in an increase in channel conductance from 33 to 37 pS and a shift of the equilibrium potential from —60 to —68 mV (Fig. 8). An analysis of the residual mean squares about the fitted regression lines shows that these differences in the conductance and reversal potential are significant at the 0·1% level (F-distribution). The significance of these results is further supported by the observation that under zero sodium conditions outward channel currents were detected at the resting potential, whereas they were never detected with normal saline (50 mmol l⁻¹ sodium) in the pipette.

The application of 4–12cmHg suction to the patch through the patch electrode increased the probability of the channel being open, resulting in an increased frequency of simultaneous openings (Fig. 9). Suction had no effect on the amplitude of the channel currents. The channel activity persisted during sustained suction and returned to baseline levels upon the release of suction. Increased channel activity in response to stretch was seen at all membrane potentials and at all potassium concentrations used, inclusive of physiological saline.

An important characteristic of these stretch-activated potassium (SAK) channels is their apparently homogeneous distribution in the sarcolemma. Every patch on which we obtained a gigaseal contained more than one SAK channel. A minimum number of channels in a patch is given by the maximum number of simultaneous openings observed, which, in the patches obtained most recently, for which we have a measure of the pipette tip dimensions, was usually four or more (stretch was applied to activate as many channels as possible). Given a patch area of 10μm² (a generous estimate based on measurements of our pipette tip sizes; cf. Guharay & Sachs, 1984) and an average of four channels per patch, the minimum SAK channel density is one per 2·5 μm² of membrane. At this density a cylindrical cell of 10μm× 100μm (a typical size for Lymnaea ventricle cells) would have about 1200 SAK channels.

The potassium channel blockers TEA (Stanfield, 1983) and 4-AP (cf. Adams, Smith & Thompson, 1980) had no effect on the channel currents at concentrations up to and including 10 mmoll⁻¹. Conversely, 1 mmoll⁻¹ quinidine completely abolished the channel conductance in about 75% of the cells examined. It is unlikely that the failure to observe currents in these cases was due to the absence of SAK channels as they were always detected in control patches (no quinidine) obtained on cells in the same culture dish. In the instances where Immoll⁻¹ quinidine did not completely abolish the channel currents, they were reduced by up to 90% of normal within a few minutes of exposure. At a quinidine concentration of 0·25–0·5 mmol l⁻¹, channel currents were reduced by 30–70% within about 5 min of exposure.

Quinidine (0·25 mmol l⁻¹) gradually reduced the probability of the channel being open. After several minutes of exposure to this drug, it was possible to obtain sufficient channel openings for current measurements only by applying suction to the patch. When the patched membrane was stretched after about 4 min of exposure to 0·25 mmol l⁻¹ quinidine, the channel current magnitude declined within several seconds to about 20% of the value measured just after the onset of suction (Fig. 10). Under these conditions the decline in the current magnitude was much more rapid than the decline observed in the unstretched membrane.

A problem inherent in physiological studies of disaggregated tissue is the difficulty of assessing the degree of cell damage incurred during enzyme treatment. This leads to uncertainty about the extent to which results obtained with isolated cells represent their behaviour in vivo. A good indicator of the integrity of the cell membrane is the transmembrane potential difference. In the isolated Lymnaea stagnalis heart cells, the resting potential was virtually identical to that reported for these cells in situ (about —60 mV in both cases) (Brezden & Gardner, 1984). Consequently, it is likely that the treatment used in this study did not result in serious disruption of cell function. Furthermore, the cells remained viable and were able to contract upon mechanical stimulation for at least 5 days following isolation. We have not examined older cultures, but it is likely that the cells would survive for considerably longer periods with periodic changes of the culture medium.

Inspection of the single-channel current records (Fig. 2) makes it immediately evident that the SAK channel has at least one open state and two closed states, since events occur as a burst of openings separated by longer closed intervals.

The 35 mV shift in the reversal potential caused by an increase of extracellular potassium from 1·6 to 30mmoll⁻¹ corresponds to a slope of 27mV per 10-fold change in the potassium concentration. Although this is considerably lower than the 58 mV slope expected for a perfect potassium-selective channel, whole cell recording also shows a deviation from an ideal potassium electrode at low extracellular potassium concentrations (Brezden & Gardner, 1984).

When the pipette potassium concentration was raised to 100mmol l⁻¹ (zero sodium, 111 mmol l⁻¹ chloride), the reversal potential shifted to about +4 mV. This is equivalent to a 57 mV slope per 10-fold change in the extracellular potassium between 30 and 100mmoll⁻¹. Again, these results are consistent with those obtained with changes in external potassium during whole-cell recording (Brezden & Gardner, 1984), and suggest that this channel contributes to the resting conductance in these cells.

The hyperpolarizing shift in the reversal potential observed upon removal of external sodium suggests that the SAK channel is permeable to sodium as well as to potassium. Under physiological conditions (1·6 mmoll⁻¹ potassium, 50mmoll⁻¹ sodium), application of the modified Goldman-Hodgkin-Katz equation (Moreton, 1968), using the measured reversal potential of —60 mV and an estimated intracellular potassium concentration of 51·5 mmol l⁻¹ (Brezden & Gardner, 1984), gives a P_K^: P_Na ratio of approximately 16. This is close to the ratio of 14 estimated for intact Lymnaea heart cells.

The effects of extracellular ion changes on the channel conductance over the range of potassium concentrations examined is not straightforward. Unavoidably, the 100 mmol l⁻¹ potassium saline had a higher ionic strength than the other salines. The effects of a generalized increase in ionic strength have not been investigated independently. Nevertheless, the channel conductance increase between 1·6 and 100 mmol l⁻¹ potassium is to be expected. Over this range, the conductance of a potassium-selective channel from sarcoplasmic reticulum, for example, increases sharply and reaches half saturation at 54 mmol l⁻¹ (Coronado, Rosenberg & Miller, 1980). The similarity of channel conductance at 1·6 and 30mmoll⁻¹ potassium for the Lymnaea cells is difficult to explain. The 30 mmol l⁻¹ potassium saline contained less sodium (21·6mmoll⁻¹ in 30mmoll⁻¹ potassium saline vs SOmmoll⁻¹ in normal ’ saline). Since sodium appears to interfere with potassium permeation through this channel (Fig. 8), sodium reduction would have been expected to increase the channel conductance. Further investigations using excised patches are required to determine the source of these anomalies. It is possible that the l/V relationships contain as yet undetected sublinearities (Yellen, 1984a).

The potassium channel blockers 4-AP and TEA, applied on the external side of the membrane, had no effect on the channel conductance at concentrations as high as 10 mmol l⁻¹. Quinidine, on the other hand, did block the channel currents at a concentration of 0·25–1 mmol l⁻¹. Hermann & Gorman (1984) reported that 0·1 mmol l⁻¹ quinidine blocks primarily the delayed potassium current in Aplysia califomica neurones. The calcium-activated potassium current was blocked only with 20- to 50-fold higher concentrations. By contrast, Fishman & Spector (1981) found that on neuroblastoma cells quinidine completely blocked calcium-activated potassium currents at 0·04 mmol l⁻¹.

Rather than producing an all-or-none block (cf. tetrodotoxin on sodium channels), quinidine (0·25 mmol 1⁻¹) reduces the SAK channel amplitude. This suggests that it may produce a fast local anaesthetic-like block (Yellen, 1984b). However, quinidine is probably membrane-permeable (Hermann & Gorman, 1984), so additional effects are also possible. The fact that the full effect of quinidine usually took several minutes to develop may reflect a slow increase in the submembrane quinidine concentration, or perhaps indirect effects induced by this drug. The interpretation of the action of quinidine is further complicated by the acceleration of channel current inhibition during stretch activation of the SAK channels. This suggests that there may be a use-dependent component in the action of quinidine.

Quinidine-sensitive, stretch-sensitive channels may be present in other tissues as well. For example, Germann & Dawson (1984) have recently reported that the colon cells of turtle exhibit quinidine-sensitive, stretch-activated potassium currents.

Since the reversal potential of the SAK channel coincides with the cellular resting potential, and as both the channel and the resting membrane have similar P_K; P_Na ratios, it is likely that the SAK channel is largely responsible for the resting conductance in Lymnaea heart ventricle cells. For this conclusion to be valid, however, it is necessary to show that the channel spends some time in the open state when V_m is at the resting potential. This poses a dilemma since a channel has no net flux at its reversal potential, and hence carries no current even though it may be open and conducting. One approach to this problem is to examine whether steady-state channel activation can be detected at voltages both more depolarized and more hyperpolarized than the resting potential. If so, it would be reasonable to assume that activation also occurs at the resting potential. Patches exposed to normal saline all exhibited outward currents when depolarized 20 mV or more from rest, but only in about 20% of the patches examined were inward currents resolved when the membrane was hyperpolarized. The probability of being open is apparently low, therefore, with hyperpolarization in the presence of normal saline. To obtain more information about the gating properties of the SAK channel at V_m = — 60 mV (resting potential), we used elevated potassium concentrations in the pipette, which would be expected to produce inward currents at rest given an open channel. Under these conditions it was readily apparent that the SAK channel is capable of conduction at the resting potential. Furthermore, it was possible to see outward currents at rest with normal (1·6 mmol 1⁻¹ ) potassium when sodium was absent from the pipette. On a few occasions (usually in the presence of an elevated potassium concentration in the pipette) a smaller conductance channel was also observed, but it was not ubiquitous like the SAK channel and would be unlikely to contribute substantially to the resting conductance. This set of observations leads us to conclude that the SAK channel probably plays a major role in establishing the resting conductance.

The SAK channel does not reveal itself at rest, so its stretch-sensitive character is easily overlooked. Certainly, we discovered its nature by chance through the use of high-potassium pipette solutions which allowed for inward currents at rest: we observed that the channel became very active while suction was being applied to make a gigaseal, then became more quiescent when suction was released.

Guharay & Sachs (1984) describe a stretch-sensitive channel in cultured chick skeletal muscle. This channel, however, is less selective (P_K:P_Na = 4:l) than the channel we describe here. Like Guharay & Sachs (1984), we can only speculate about possible physiological roles for stretch-activated channels. Since the Lymnaea SAK channel is more potassium-selective than the one in chick skeletal muscle, it would be expected to produce rather different effects upon activation. It could, for example, reduce membrane excitability by increasing the sarcolemmal potassium permeability. In a myogenic tissue such as the Lymnaea heart, a reduction in membrane excitability might allow the ventricle to fill to a greater degree than would be possible in the absence of SAK channels. Alternatively (or in addition), these channels may play a role in osmoregulation in this freshwater species by releasing osmotically active ions when the cells swell during hypo-osmotic stress. We calculate that for a cell having dimensions of 10μm× 100μm, containing 50 mmol l⁻¹ potassium and having 1200 SAK channels, conditions which create a 0·5 pA current through each of the channels for 1 s would lead to the loss of about 1 % of the intracellular potassium in that time (using a probability of being open = 0·5). This ‘order-of-magnitude’ calculation suggests that it is worthwhile to investigate the possible role of the SAK channel in osmoregulation. That osmoregulating channels may exist in other tissues is suggested by Hamill (1983), who has provided preliminary data on the singlechannel activity of an erythrocyte potassium channel which is activated by hypoosmotic stress. Also, quinine, a potassium blocker and stereoisomer of quinidine, has been shown to interfere with cell swelling-induced potassium efflux in lymphocytes (Sarkadi, Mack & Rothstein, 1984).

We thank the Natural Sciences and Engineering Research Council of Canada for operating grants to DRG (A-4290) and CEM (A-8402), and the Muscular Dystrophy Association of Canada for a grant to CEM. We thank Mr Wade Sigurdson for photographing the isolated Lymnaea cells.