Pore Properties of Lymnaea Stagnalis Neuron Stretch-Activated K⁺ Channels

In many neurons, resting K⁺ conductance is modulated by neurotransmitters via second messengers; for the well-studied Aplysia californica mechanosensory neurons, the S-channel operates in this way (Siegelbaum et al. 1982). We have shown that an apparently adventitious characteristic of this channel allows it to be activated by stretch under patch-clamp conditions (Vandorpe and Morris, 1992; Vandorpe et al. 1994). Stretch-activated (SA) K⁺-selective channels similar to the S-channel are found in all molluscan neurons (Morris and Sigurdson, 1989; Bedard and Morris, 1992), in Drosophila melanogaster somatic muscle (Zagotta et al. 1988; Gorczyca and Wu, 1991), in rat heart (Kim and Duff, 1990; Kim, 1992) and in fish embryos (Medina and Bregestovski, 1991). The observations that SA K⁺ channels are insensitive to intracellular Ca²⁺ (Sigurdson and Morris, 1989), insensitive to physiological membrane voltages (Small and Morris, 1994), often persistently active with a low open probability (e.g. Vandorpe et al. 1994) and poorly activated by membrane tension in situ (Morris and Horn, 1991; Wan et al. 1995) are consistent with a conductance-modulating role for these channels. Second-messenger modulation of SA K⁺ channels in early teleost embryos has also been shown to regulate resting conductance (Medina and Bregestovski, 1988, 1991; see Morris, 1992, 1995, for a discussion of possible roles of stretch-sensitive channels).

Some SA channels exhibit mechanosensitivity only after the plasma membrane is decoupled from the cortical cytoskeleton (Small and Morris, 1994). Though stretch-inactivated cation-selective channels in rat supraoptic neurons appear to have mechanotransduction as their primary physiological function (Oliet and Bourque, 1993a,b), membrane tension in molluscan neurons is normally buffered by the cortical cytoskeleton in a way that protects SA K⁺ channels from mechanical stimuli. Although the mechanosensitivity of the channels is of biophysical interest, and although the channels may yet be shown to exhibit pathophysiological or developmentally regulated mechanosensitivity, the critical physiological issues for molluscan SA K⁺ channels in neurons relate to their permeation properties and the manner of their modulation.

Some aspects of the pore properties of Aplysia californica S-channels from identified mechanosensory neurons (Shuster et al. 1991) and of SA K⁺ channels from Cepaea nemoralis neurons (Bedard and Morris, 1992) have been examined. It is the focus of this paper to describe the pore properties of a related channel, the K⁺-selective SA channel of Lymnaea stagnalis neurons. The results also provide an opportunity to compare the permeation characteristics, as seen at the single-channel level, of what are probably closely related channels in freshwater, terrestrial and marine gastropod neurons.

The circumesophageal ring of ganglia was dissected from Lymnaea stagnalis and gently agitated for 60 min in normal saline (NS) (in mmol l⁻¹, 55 NaCl, 1.6 KCl, 2 MgCl₂, 3.5 CaCl₂ and 5 Hepes–NaOH adjusted to pH 7.6) with CaCl₂ removed and 0.25 % Protease type XIV (Sigma) added. The ganglia were then washed in NS, and individual neurons were plated in NS with 5 mmol l⁻¹ glucose and 0.1 mg ml⁻¹ gentamicin sulfate (Sigma). Cells were kept at room temperature for 1–6 days in culture before patch-clamp recordings (Sigurdson and Morris, 1989) were made.

Cell-attached recordings were made in a bath solution of NS. Pipette solutions contained only KCl and/or the test cation salt, as indicated, and did not contain any MgCl₂, CaCl₂, Hepes or glucose in order to minimize interference by these agents with test cation permeation while measuring channel selectivity and unitary conductance. Experiments depicted in Fig. 5 did, however, contain a high-K⁺ version of NS (no NaCl, and 50 mmol l⁻¹ KCl instead of 1.6 mmol l⁻¹ KCl). Excised inside-out patches also had only KCl and/or the test cation indicated. In the case of thallium (Tl⁺), the chloride salt is insoluble so we used thallium acetate (TlAc) and substituted potassium acetate (KAc) for KCl. The resulting large offsets at the silver/silver chloride electrode (Raynauld, 1994) could, however, be zeroed out. With symmetrical TlAc/KAc conditions, we assumed a reversal potential (V_rev) of 0 mV.

Patch pipettes had an outside diameter of 2–3 μm before fire-polishing (2–2.5 μm after polishing) and a resistance of 1–5 MΩ with high-K⁺ solution in the pipette and NS in the bath. Pipettes were made from borosilicate glass (N51A, i.d. 1.15 mm, o.d. 1.65 mm; Garner Glass, Claremont, CA, USA), using a List L/M-3P-A pipette puller (Darmstadt, Germany). They were coated with Sylgard 184 (Dow Corning, Midland, MI, USA) then fire-polished.

To select for SA K⁺ channels over other K⁺ channels, namely Ca²⁺-activated K⁺ channels, we routinely applied stretch to increase the open probability (P_open) of the SA K⁺ channel in a stable manner. The amount of suction applied to a given patch was fixed during any single run, but varied between –6.7 ×10² and –8.0 ×10³ Pa depending on the patch and its history; stretch sensitivity of SA K⁺ channels varies with patch history (Small and Morris, 1994). For further selection of SA K⁺ channels, we used 1 mmol l⁻¹ tetraethylammonium (TEA⁺) in the pipette (Shuster et al. 1991). External TEA⁺ reduces the SA K⁺ single-channel current by half at approximately 50 mmol l⁻¹ but has no discernible effect at 1 mmol l⁻¹ (Small and Morris, 1995). Bath solution changes were made by perfusing 9 ml of test solution at approximately 4 ml min⁻¹ into a 2 ml recording chamber (excess was aspirated from the top of the chamber).

Channel currents were recorded using an Axopatch 1D (Axon Instruments, Foster City, CA, USA) connected to a PC microcomputer via a TL-1 interface (Axon Instruments, Claremont, CA, USA). pClamp 5.5 (Axon Instruments, Claremont, CA, USA) was used to digitize and export records to Sigmaplot 4.1 (Jandel Scientific, Corte Madera, CA, USA). Once in Sigmaplot 4.1, the records were analyzed then exported to Coreldraw 3.0 (Corel, Ottawa, Ontario, Canada).

Unitary chord conductance was measured and current–voltage (I/V) relationships constructed by measuring the single SA K⁺ channel amplitude in response to 2 s voltage ramps from -100 to 120 mV (V_m) and 2 s voltage steps at the voltage extremes. Leak current was determined empirically for each ramp and subtracted from the total current. Following this subtraction, voltage ramp responses from three or four patches were superimposed and the open channel level fitted by eye to obtain I/V relationships. The end points of these I/V relationships, shown in the figures as symbols (means ± S.E.M.), were obtained separately by measuring the current amplitude of responses to 2 s voltage steps to -100 and 120 mV (V_m). Paired Student’s t-tests were carried out to determine the statistical significance of the difference between these means. To test the hypothesis that the K_m values obtained from the hyperbolic fits to inward and outward conductances with increasing symmetrical K⁺ concentrations in Fig. 1 were different, the hyperbolas were linearized using:

Single-channel current amplitude was measured over a range of voltages with symmetrical [K⁺]. The effect of [K⁺] on SA K⁺ channel activity is illustrated in Fig. 1. At different concentrations there was no change in channel kinetics as judged by the appearance of SA K⁺ channel responses to voltage ramps (Fig. 1A). At [K⁺]⩾e50 mmol l⁻¹, the I/V relationship flattened at extreme membrane depolarizations. Chord conductances were determined between 0 and 120 mV (outward), 0 and –100 mV (inward). Applying the Michaelis–Menten equation, hyperbolic fits to the mean conductances using γ = γ_max/(1 + K_m/[K⁺]) (Fig. 1B) yielded Michaelis–Menten constants (K_m) of the channel for K⁺ of 28±4 mmol l⁻¹ for inward currents obtained by measuring chord conductance from 0 to -100 mV, and 91±35 mmol l⁻¹ for outward currents (chord conductance measured from 0 to 120 mV). These hyperbolic fits yielded asymptotic maxima (γ_max) representing concentrations at which the K⁺ binding site(s) within the pore saturates. For inward currents, γ_max was 180±9 pS, and for outward currents, 160±35 pS. The observation that different K_m values were obtained for inward and outward currents suggests that the SA K⁺ channel pore contains more than one binding site for K⁺. Likewise, the Michaelis–Menten formulation applies to a single association–dissociation reaction, but the K_m values here represent one or more sites with which K⁺ interacts as it permeates the pore.

The single-channel permeability coefficient, P_K (cm s⁻¹), was estimated using the ‘limiting conductance’ (I_K/V) for symmetrical 50 mmol l⁻¹ K⁺. Ideally, I_K(V)/V is equivalent to the limiting conductance and the limiting conductance is a constant. In practice, even with symmetrical solutions, some rectification occurred (Fig. 1), so the limiting conductance (110 ± 2 pS, N=7) was obtained from the steepest linear region (inward current), yielding P_K=5.8 ×10⁻¹³ cm s⁻¹.

When an ion channel contains more than one ion at a time, ion–ion repulsion can occur inside the pore and destabilize ion binding, resulting in rapid transmembrane passage despite high-affinity binding (Tsien et al. 1987). Convincing evidence for a multi-ion pore with such ion–ion repulsions can be obtained by presenting a channel with mixtures of two permeant ions. If the channel is a multi-ion pore, the unitary conductance will not increase monotonically with an increasing mole fraction of the more permeant ion species; instead, it goes through a minimum and then increases. This phenomenon is called the anomalous mole fraction effect (Hille, 1984). The two permeant ions used to test for this effect on the SA K⁺ channel were K⁺ and Rb⁺. The chord conductance of 100 % symmetrical K⁺ measured from –100 to 120 mV was 79±2 pS. Although SA K⁺ channel unitary chord conductance for 100 % Rb⁺ as measured from –100 to 120 mV was small (16±0.1 pS), when a solution containing a mixture of 25 % K⁺ and 75 % Rb⁺ was used, it decreased even further (by approximately 40 %) to a minimum of 9.5±0.2 pS (Fig. 2).

Reversal potentials reveal the relative affinity of a site(s) for a range of cations. This issue of binding is one aspect of selectivity. Conductance provides information about another aspect, the rate of net movement through the channel (Hille, 1984; Tsien et al. 1987). The selectivity of a channel as determined by conductance ratios may be different from that determined by relative permeability. Single SA K⁺ channel I/V responses were obtained with 2 s voltage ramps from –100 to 120 mV (V_m) in symmetrical solutions of K⁺, Rb⁺, Tl⁺ and NH₄⁺ (Fig. 4). There were no detectable inward or outward currents with Na⁺, Li⁺ or Cs⁺. An upper limit for conductance of these three monovalent ions was obtained by assuming that the amplitude of events was less than the noise level. Values of conductance for K⁺, Rb⁺ and NH₄⁺ were obtained by measuring the chord conductances from –100 to 120 mV. Values of conductance for Tl⁺ were obtained by measuring the chord conductances from –50 to 50 mV. Large inward and outward Tl⁺ currents were observed and both appeared to flicker. Rb⁺ and NH₄⁺ currents were small in amplitude. The selectivity sequence from these measurements is Tl⁺=K⁺>Rb⁺>NH₄^{+ ⪢}Na⁺=Li⁺=Cs⁺ (Table 1; g_X/g_K, column 3).

Because some channels are pH-sensitive, we sought to determine whether H⁺ or OH⁻ concentrations had any substantial effect on SA K⁺ channel unitary conductance. We used KOH and HCl to obtain the desired pH values for the recording pipette solutions (Fig. 5.). We chose pH values as far from pH 7.6 (normal) as practicable. The solutions with more extreme pH values hindered seal formation, presumably through membrane destabilizing effects. Surprisingly, pH extremes had no detectable effects on SA K⁺ channel responses to voltage steps or ramps in cell-attached patches (Fig. 5). Neither SA K⁺ channel kinetics, judged by the appearance of single-channel events, nor unitary current amplitude was affected.

We noted that for a given pipette [K⁺], the inward K⁺ currents in cell-attached experiments (Fig. 5) were smaller (by approximately 40 %) than in excised inside-out patches (Fig. 1). Either a cytoplasmic component not included in the bath of excised patches or a component of the recording pipette of cell-attached patches not present in excised patch pipettes (i.e. 2 mmol l⁻¹ Mg²⁺, 3.5 mmol l⁻¹ Ca²⁺ and 5 mmol l⁻¹ Hepes) could be responsible. Large inward K⁺ currents were observed in cell-attached patches in pipettes containing 1 mmol l⁻¹ Ca²⁺ and 5 mmol l⁻¹ Hepes but no Mg²⁺ (Sigurdson and Morris, 1989), suggesting that the reduced current amplitude in our excised patches was not caused by Ca²⁺ or Hepes or a cytoplasmic component. Moreover, with just 50 mmol l⁻¹ K⁺ and 1 mmol l⁻¹ TEA⁺ in the pipette (cell-attached configuration; i.e. identical to excised patches), inward K⁺ currents did not differ (not shown) from those of excised inside-out patches, suggesting that extracellular Mg²⁺ acts as a fast blocker of the inward K⁺ current.

Another divalent which has been shown to block monovalent cationic channels is Ba²⁺. Ba²⁺, whose ionic radius is 0.135 nm (that of K⁺ is 0.133 nm; Hille, 1984), blocks several K⁺ channels from both the inside and outside but generally acts from inside with much greater affinity (micromolar as opposed to tens or hundreds of millimolar; Armstrong et al. 1982; Eaton and Brodwick, 1980; Vergara and Latorre, 1983). Internal Ba²⁺ produces a slow block of Aplysia californica S-channels (Shuster and Siegelbaum, 1987). We characterized the internal Ba²⁺ block of Lymnaea stagnalis SA K⁺ channels. Channel P_open was elevated to maximal levels and maintained with –2.7 ×10³ Pa throughout the experiment to ensure that the channel remained in the open configuration. Extracellular Ba²⁺ at a concentration of 50 mmol l⁻¹ did not block outward K⁺ currents of Lymnaea stagnalis SA K⁺ channels (Sigurdson, 1990) but intracellular Ba²⁺ (5 mmol l⁻¹) results in a complete block of SA K⁺ channels (Fig. 6A). The concentration-dependence of the intracellular Ba²⁺ block of SA K⁺ channels (Fig. 6B) yields an IC₅₀ of 0.14±0.08 mmol l⁻¹. The kinetics of Ba²⁺ block were analysed in three patches held at 40 mV: 10 s records were made into events lists with Fetchan and then used to construct dwell-time histograms. These were fitted with a double exponential curve to obtain fast and slow open and closed time constants (τ, in ms) (control: τ_open1=0.71±0.01, τ_open2=4.7±0.80, τ_closed1=0.72±0.08, τ_closed2=13±5.2; 1 mmol l⁻¹ internal Ba²⁺: τ_open1=0.48±0.13, τ_open2=2.8±0.95; τ_closed1=0.49±0.12, τ_closed2=110±23; values are means ± S.E.M.). The slow closed time constant was significantly increased by more than 800 % with 1 mmol l⁻¹ Ba²⁺. A question not answered in the study of Ba²⁺ blocking of Aplysia californica S-channels (Shuster and Siegelbaum, 1987) was whether Ba²⁺ could block an already closed channel or whether the channel had to be in the open conformation to be blocked. When excised patches were stepped from 0 mV to strong depolarizing potentials [120 mV (V_m)] under conditions in which SA K⁺ channel P_open was near maximal (SA K⁺ channels do not inactivate) and the intracellular Ba²⁺ concentration was sufficient to ensure channel block (2 mmol l⁻¹), the channel began in the open state but was blocked after a delay (mean ± S.E.M. of 10 steps of one patch = 280±93 ms; four example traces are shown in Fig. 6C). This suggests that SA K⁺ channels need to be open before they can be blocked by internal Ba²⁺.

We obtained one patch from which we were able to obtain enough data to test whether the unblocking rate of Ba²⁺ is voltage-dependent. The rate of binding of a ligand to a site should be dependent on the ligand concentration, whereas the rate of dissociation should be independent of ligand concentration. Therefore, the voltage-dependence of Ba²⁺ interactions with the channel should not be complicated by the concentration-dependence of the Ba²⁺ block when only rates of dissociation are measured. With 2 mmol l⁻¹ intracellular Ba²⁺ (likelihood of block greater than 90 %), an excised patch was stepped from 70 mV (this depolarized potential favored the blocked state) to increasingly hyperpolarizing potentials, which would unblock the channel, and the delay before unblocking was measured (Fig. 7A). The delay before unblocking is plotted in Fig. 7C, which illustrates the voltage-dependence of the unblocking; the decreasing delay before unblocking with increasing hyperpolarization suggests that Ba²⁺ is electrostatically repelled by increasing electropositivity at the external face (depicted in the diagram Fig. 7B).

SA K⁺ channel conductance saturated at high K⁺ concentrations in a similar manner to other K⁺ channels (Coronado et al. 1980; Eisenman et al. 1986). Saturation indicates that there is at least one binding site in the conduction pathway. The affinity of the SA K⁺ channel for K⁺, based on the Michaelis–Menten constants (K_m, 28 mmol l⁻¹ for the inward current and 91 mmol l⁻¹ for the outward current; Fig. 1), was comparable to that of the inwardly rectifying K⁺ channel ROMK1 (conductance half-saturates in the 20 –50 mmol l⁻¹ range, based on Fig. 4 of Lu and MacKinnon, 1994), but was much greater than that of the Shaker K⁺ channel (K_m approximately 300 mmol l⁻¹; Heginbotham and MacKinnon, 1993) and the high-conductance Ca²⁺-activated K⁺ channel (conductance saturates at 2000 mmol l⁻¹; Eisenman et al. 1986).

The absolute P_K for the Lymnaea stagnalis neuron SA K⁺ channels at 50 mmol l⁻¹ K⁺ (5.8 ×10⁻¹³ cm s⁻¹) exceeded that for Aplysia californica S-channels (1.5 ×10⁻¹³ cm s⁻¹ at 360 mmol l⁻¹; Shuster et al. 1991) and that of SA K⁺ channels in fish embryos (1.3 ×10⁻¹³ cm s⁻¹ at 140 mmol l⁻¹ K⁺; Medina and Bregestovski, 1988). It is closest to Shaker K⁺ channels of Drosophila melanogaster (approximately 4 ×10⁻¹³ cm s⁻¹ at 140 mmol l⁻¹ K⁺; Heginbotham and MacKinnon, 1993) and Cepaea nemoralis SA K⁺ channels (3.4 ×10⁻¹³ cm s⁻¹ at 70 mmol l⁻¹; Bedard and Morris, 1992). Given the abundance of intracellular K⁺ in the physiological environment of S-channels (Aplysia californica is a marine snail, approximately 360 mmol l⁻¹ cytoplasmic [K⁺]) compared with the situation for Lymnaea stagnalis SA K⁺ channels (freshwater snail, approximately 50 mmol l⁻¹ cytoplasmic [K⁺]), it is not at all surprising that Lymnaea stagnalis channels have a higher K⁺ affinity. A higher-affinity site could contribute to a higher conductive permeability only for a channel capable of multi-ion occupancy. Several lines of evidence suggest that the SA K⁺ channel is a multi-ion channel, including the existence of different K_m values for conductance for inward and outward currents.

We note that the higher-affinity site (K_m of 28 mmol l⁻¹versus 91 mmol l⁻¹) is at the external face of the membrane where the [K⁺] is lowest. The conducting SA K⁺ channel will spend most of its tenure in the membrane at potentials near the equilibrium potential for K⁺, with influx and efflux balanced. The elevated affinity of the outer region of the pore for K⁺ would help ensure occupation of this region by a K⁺, and hence would ensure that influx was not diffusion-limited near the K⁺ equilibrium potential in spite of the lower extracellular [K⁺].

In Lymnaea stagnalis, Cepaea nemoralis and Aplysia californica neurons, physiological concentrations of K⁺ vary by almost an order of magnitude, yet the conductances for their SA K⁺ channels are remarkably similar. For cell-attached recordings of outward currents, conductances for SA K⁺ channels from these species are all approximately 50 pS; 58 pS for Cepaea nemoralis (Bedard and Morris, 1992), 50 pS for Aplysia californica and 44 pS for Lymnaea stagnalis (Vandorpe and Morris, 1992). One cannot argue that this 50 pS limit simply reflects saturation of conductivity at K⁺ concentrations below those of the most dilute cytoplasm, since highly selective K⁺ channels with both lower conductances, e.g. voltage-gated channels of approximately 15 pS (Zagotta et al. 1988), and higher conductances, e.g. maxi-K⁺ channel 100 pS (Pallotta et al. 1981), can and have been engineered by nature. Perhaps for those channels, like the S-channel, that perform the same roles as those undertaken by SA K⁺ channels (background modulation of excitability in response to neurotransmitters rather than fast conductance changes in response to membrane voltage) there exists an optimum conductance. If so, that optimum may be set by the fixed specific capacitance of membranes and the SA K⁺ channel density of about 1 μm⁻² that is common to all these species (see Vandorpe and Morris, 1992). If the major role of the channels was related to [K⁺] homeostasis across cell membranes rather than to the modulation of excitability, it would be reasonable to expect conductances to vary according to the prevailing ionic strength of the physiological fluids. Instead, we find conductances of approximately 50 pS for both Lymnaea stagnalis and Aplysia californica, in spite of the order of magnitude difference in the K⁺ concentration of their physiological fluids.

The most convincing evidence for a channel possessing a multi-ion pore, short of unidirectional flux studies with radiotracers (Hille and Schwarz, 1978; Hodgkin and Keynes, 1955), is obtained by testing for anomalous mole fraction effects. Using Rb⁺ as a companion permeant ion species to K⁺, we observed the key feature of anomalous mole fraction behavior, a conductance minimum. Most K⁺ channels exhibit this behavior (for a review, see Pallotta and Wagoner, 1992). A cloned K⁺ channel, Shaker (Heginbotham and MacKinnon, 1993) and a high-conductance Ca²⁺-activated K⁺ channel (Neyton and Miller, 1988), have proved to be multi-ion pores, and the affinities of the individual binding sites within the pore have been determined for the latter. In one type of Cl⁻ channel, it has been possible, using site-directed mutagenesis, to demonstrate that a particular pH-titratable amino acid residue participates in anomalous mole fraction behaviour (Tabcharani et al. 1993).

The selectivity sequence determined using reversal potentials under bi-ionic conditions (K⁺ at the cytoplasmic face, test cation at the extracellular face) was Cs⁺>K⁺>Rb⁺>NH₄⁺>Na⁺>Li⁺, whereas that determined using relative conductance in symmetrical solutions was Tl⁺=K⁺>Rb⁺>NH₄^{+ ⪢}Na⁺=Li⁺=Cs⁺. The high bi-ionic selectivity of the SA K⁺ channel for Cs⁺, an ion that did not carry current through the channel, suggests that there was a fast non-permeant block (Yellen, 1984b) of the SA K⁺ channel lumen by Cs⁺ at an external site. Na⁺ blocks Cepaea nemoralis neuron SA K⁺ channels at an internal site (Bedard and Morris, 1992) in a similar manner, namely it reduces current but does not alter the reversal potential.

The selectivity ratio for Rb⁺ compared with K⁺ determined from bi-ionic potentials (0.63) differs from that determined from relative conductances (0.20). A difference is to be expected (Hille, 1984) since the bi-ionic selectivity (0.63) reflects only the height of potential energy barriers (relative to those for K⁺) faced by Rb⁺ as it permeates the channel, whereas the conductance or current-based selectivity (0.20) is also affected by the energy wells whose depth is essentially a measure of binding affinity (relative to that for K⁺). Comparable differences are observed in a large-conductance Ca²⁺-activated K⁺ channel (BK⁺) which has a relative bi-ionic permeability for Rb⁺ of 0.7 (Yellen, 1984a) and relative conductance permeability of 0.07 (Blatz and Magleby, 1984). For many types of K⁺ channel, the potential energy barriers facing Rb⁺ and K⁺ as they permeate are evidently quite similar but the binding sites in K⁺ channels have substantially greater affinity for Rb⁺ than for K⁺. For Shaker channels, where this situation also applies, it has been suggested that energy barriers and binding sites are associated with discrete structural features of the channel (Heginbotham and MacKinnon, 1993).

The effects of pH on ion channels are not trivial, especially in view of the results from a recently cloned voltage-independent K⁺-selective channel which is activated by protons in the pH range 6.8–7.6 (Suzuki et al. 1994). Acid pH depresses the Na⁺ conductance of nerve (Woodhull, 1973) and the outer segment of isolated retinal rods of frog (Mueller and Pugh, 1983). Extracellular protons decrease L-type Ca²⁺ channel conductance (Prod’hom et al. 1989). Cyclic-nucleotide-gated cation channels contain two identical titratable sites (pKa 7.6) near the extracellular mouth of their conduction pathway, which make their conductance pH-sensitive (Root and MacKinnon, 1994). Nevertheless, the conductance of two different mechanosensitive channels, an SA K⁺ channel found in rat heart (Kim, 1992) and an SA cation channel in chick skeletal muscle (Guharay and Sachs, 1985), is unaffected by pH even though, in both cases, there is evidence of specific pH effects on channel kinetics. It is clear that Lymnaea stagnalis SA K⁺ channel conductance, too, is insensitive to extreme extracellular pH variations.

For inwardly rectifying K⁺ channels, Mg²⁺ blocks from the inside (Rudy, 1988). The Mg²⁺ block of SA K⁺ channels was from the outside, and external Mg²⁺ has also been shown to block an SA K⁺ channel in amphibian smooth muscle (Hisada et al. 1991). The presence of 2 mmol l⁻¹ extracellular Mg²⁺ decreased the conductance of those SA K⁺ channels from 110 to 63 pS (an approximately 43 % decrease), which is nearly identical to the decrease in SA K⁺ channel conductance we obtained with 2 mmol l⁻¹ external Mg²⁺. This feature departs from most other K⁺ channels, and the fact that two SA K⁺ channels have been found to exhibit this characteristic may prove useful in differentiating them from other K⁺ channels. It is also worth noting that cation-selective cyclic-nucleotide-gated channels are blocked by external Mg²⁺, and that cation selectivity and external block by Mg²⁺ can be conferred on K⁺-selective Shaker channels by the elimination of two amino acids from the pore region (Heginbotham et al. 1992).

Several K⁺ channels are blocked from either side by Ba²⁺ but with much greater affinity from the inside (Rudy, 1988). SA K⁺ channels in Aplysia californica (i.e. S-channels; Shuster and Siegelbaum, 1987) and Lymnaea stagnalis neurons (Vandorpe and Morris, 1992) are, however, blocked by Ba²⁺ only from the inside, as are delayed rectifier K⁺ channels in squid axon (Armstrong et al. 1982). The affinites for internal Ba²⁺ block in these molluscan K⁺ channels cover a wide range with a K_d of 140 μmol l⁻¹ for the Lymnaea stagnalis channel, 20 μmol l⁻¹ for the Aplysia californica channel (Shuster and Siegelbaum, 1987) and 0.1 μmol l⁻¹ for the squid channel (Armstrong et al. 1982). The block of squid and Lymnaea stagnalis channels by internal Ba²⁺ is voltage-dependent; that for Aplysia californica S-channels has not been tested. Other SA K⁺ channels exhibit an internal Ba²⁺ sensitivity. In amphibian sympathetic neurons, I_M channels, mechanosensitive K⁺ channels reminiscent of Aplysia californica S-channels (I_M channels are so-named because they are subject to muscarinic regulation; see Rudy, 1988) are completely blocked by 4–8 mmol l⁻¹ internal Ba²⁺ (Hara and Kuba, 1993). SA K⁺ channels in rat atrial myocytes are fully blocked by 1 mmol l⁻¹ internal Ba²⁺ (Kim, 1992) and SA K⁺ channels in Drosophila melanogaster body wall muscle by 100 μmol l⁻¹ internal Ba²⁺ (Gorczyca and Wu, 1991).

The internal Ba²⁺ block of K⁺ channels is usually characterized by the following: (1) slow channel block, that is, a decreased P_open with no change in single-channel conductance; (2) voltage-dependence, with an increased blocking efficacy at depolarized potentials; (3) efficacy in the high micromolar range; and (4) blockade only in the open state (Armstrong and Taylor, 1980; Rudy, 1988). Our results are consistent with this picture: internal Ba²⁺ blocked open SA K⁺ channels, it decreased the long closed time (i.e. slow channel block) and the rate of unblocking was voltage-dependent. The observation that all SA K⁺ channels are blocked by intracellular Ba²⁺, as are non-mechanosensitive K⁺ channels, suggests that SA K⁺ channels belong to some larger family of K⁺ channels.

In summary, SA K⁺ channels are multi-ion pores with a particularly high affinity for K⁺ at an external pore region. The only monovalent ion of physiological significance to which they are detectably permeable is K⁺. These channels allow some other monovalent and divalent cations to penetrate partway into the pore, where they exhibit a blocking action but do not permeate. Most notably, at physiological levels, extracellular Mg²⁺ reduces current amplitudes by almost 50 %. The channels are insensitive to extreme fluctuations of extracellular pH.

Supported by NSERC, Canada.