Voltage change-induced gating transitions of the rabbit skeletal muscle Ca²⁺ release channel

Abstract

1. We used the planar lipid bilayer method to study single ryanodine receptor Ca²⁺ release channels (RyRCs) from fast skeletal muscle of the rabbit. We found that changes in membrane voltage directly induced gating transitions of the RyRC: (i) in the steady state, even at activating Ca²⁺ concentrations (20 μm), at a constant membrane potential the channels resided in a low open probability (P_o) state (inactivated-, I-mode), and (ii) upon abrupt changes of voltage, the apparent inactivation of the RyRCs was relieved, resulting in a rapid and transient increase in P_o.

2. The magnitude of the P_o increase was a function of both the duration and the amplitude of the applied prepulse, but was independent of the channel activity during the prepulse.

3. The voltage-induced P_o increase probably involved major conformational changes of the channel, as it resulted in substantial alterations in the gating pattern of the channels: the voltage change-induced increase in P_o was accompanied by the rapid appearance of two types of channel activity (high (H) and low (L) open probability modes).

4. The response of the RyRC to voltage changes raises the interesting possibility that the activation of RyRC in situ might involve electrical events, i.e. a possible dipole-dipole coupling between the release channel and the voltage sensor.

Excitation-contraction (E-C) coupling in skeletal muscle is thought to involve the concerted action of at least two membrane proteins: (i) the dihydropyridine receptor (DHPR) of the transverse tubule membrane, which is assumed to function as both a voltage sensor and a signal transducer that receives and communicates the signal of transverse tubule depolarization towards the sarcoplasmic reticulum (SR), and (ii) the ryanodine receptor Ca²⁺ release channel (RyRC), which on the SR side is activated upon receipt of the signal from the DHPR (Campbell et al. 1987; Lai, Erickson, Rousseau, Liu & Meissner, 1988). Although the nature of the signal communicated between the two proteins is not yet known, many observations from electrophysiological (Ríos & Brum, 1987), morphological (Franzini-Armstrong, 1970; Block, Imagawa, Campbell & Franzini-Armstrong, 1988) and biochemical (Smith, Imagawa, Ma, Fill, Campbell & Coronado, 1988; Lai et al. 1988) studies point to the possibility that it might involve a direct mechanical interaction between the two receptor proteins. Thus, it is hypothesized (Ríos & Pizarró, 1988; Ríos, Karhánek, Ma & González, 1993) that skeletal muscle excitation-contraction coupling is founded on a sequence of protein conformational changes which first occur in the DHPR (voltage sensor function), then spread to the RyRC (signal transduction) and, finally, result in the opening of the RyRC. Recently, we have obtained evidence suggesting that such coupled conformational changes between the two receptors in triadic preparations can, in fact, occur (Minarovic & Mészáros, 1996).

There is also evidence indicating that this mechanical mode of RyRC activation is complicated by other events in E-C coupling. The RyRCs (both the skeletal and the cardiac isoforms) are activated by Ca²⁺ itself, to whose explosive increase during E-C coupling the skeletal channel must be exposed (Escobar, Monck, Fernandez & Vergara, 1994; Tsugorka, Rios & Blatter, 1995). Skeletal Ca²⁺ release channel activity, measured in isolated SR vesicle experiments (Meissner, 1984; Chu, Fill, Stefani & Entmann, 1993) or in planar lipid bilayers (Smith et al. 1988; Chu et al. 1993), is at least partially inactivated by Ca²⁺ at high Ca²⁺ concentrations (that might, however, not be attained during E-C coupling in vivo), which might result from the binding of Ca²⁺ to the Mg²⁺-inactivation site (Lamb, 1993). However, Ca²⁺ release measured in cut fibres (Schneider & Simon, 1988) displays Ca²⁺-dependent inactivation at modestly elevated Ca²⁺ concentrations. Furthermore, the phenomenon of adaptation of single release channels, i.e. only a transient increase of channel open probability (P_o) after activation (Györke & Fill, 1993; Valdivia, Kaplan, Ellies-Davies & Lederer, 1995), also suggests that following activation RyRCs undergo a decrease in open probability, which might reflect the occurrence of either a Ca²⁺-dependent or a spontaneous channel inactivation.

The results we present here suggest that in the steady state (even at activating Ca²⁺ concentrations) the P_o of the RyRC is low, which is probably due to a high degree of inactivation. However, upon an abrupt change in the membrane voltage, the channel can rapidly convert into active states of elevated P_o. This response of the release channel to voltage changes raises the possibility that the activation of RyRCs in situ might involve electrical events, such as a dipole-dipole interaction between the release channel and the voltage sensor.

METHODS

Bilayer experiments

Heavy SR vesicles were prepared from fast skeletal muscle of the rabbit according to the procedure of Meissner (1986). In brief, rabbit hindlegs were bought from a butcher's shop (Speciality Meat, Aiken, SC, USA), and the white muscle processed within 45 min of the rabbit's death as follows. Muscles were homogenized in the presence of protease inhibitors, a microsomal fraction was obtained by differential centrifugation, then further purified by isopycnic centrifugation in a discontinuous sucrose gradient. The fraction from the 35–41 % interface was collected, resuspended, frozen in liquid nitrogen and stored at -80°C.

Synthetic lipids (Avanti Polar Lipids, Birmingham, AL, USA, or Sigma) were used to prepare bilayers of composition 5 : 3 : 2 phosphatidylserine (PS) : phosphatidylethanolamine (PE) : phosphatidylcholine (PC), or 7 : 3 PE : PC (5 mg ml⁻¹ in decane; Aldrich). The bilayer was painted over a Teflon partition with a 0.15 mm aperture. Solutions of 250 mm (cis, the SR application side, corresponding to the cytoplasmic side) and 50 or 250 mm CsCH₃SO₃ (trans) buffered with 5 mm Cs-MOPS to pH 7.4 were used to isolate Ca²⁺ release channel activity from that of other channel types (Fill, Stefani & Nelson, 1991). The desired cis Ca²⁺ concentration was adjusted by adding calcium methanesulphonate and verified with a calcium-selective electrode (Orion Research Inc., Cambridge, MA, USA). The Ca²⁺ release channel was identified on the basis of its conductance (408 ± 12 pS), sensitivity to activation by Ca²⁺, and responsiveness to agonists (1 mm ATP, 10 μm ryanodine (Calbiochem)) as well as to inhibitors (e.g. 2 μm Ruthenium Red (Aldrich) and 3 mm MgCl₂). Unless otherwise stated, all chemicals were obtained from Sigma.

Data acquisition and analysis

Single-channel currents were measured with a BLM-120 bilayer amplifier (Bio-LOGIC Science Instruments, France). A 386 AT computer with pCLAMP software (CLAMPEX, Axon Instruments) and a Labmaster DMA (Scientific Solutions, Solon, OH, USA) interface (TL-125) were used for data collection. The recordings were low-pass filtered with an 8-pole Bessel filter at 2000 Hz (filter dead time, 91 μs) and digitized with a 200 μs sampling rate. Under these conditions, openings that are shorter than 0.2 ms are missed completely, and openings that last 0.2–0.4 ms are missed partially (Zahradníková & Zahradník, 1995).

The channels were held at a constant membrane potential, at which the channel activity was measured in 409.6 ms segments either repeatedly in succession, or at given time intervals (0, 2.5 and 3.5 s) after a voltage prepulse of varying amplitude (V_m= 0 or −50 mV) and duration (0–2000 ms). The membrane potential was expressed as cis vs. trans. The passive current components were electronically compensated on-line. The single-channel activity was characterized by open probability (P_o), average open time (t_o) and frequency of channel openings (n_o), which were calculated using TRANSIT software (Baylor College of Medicine, Houston, TX, USA) implemented on a Pentium computer. Transitions between open and closed states were detected using the 50 % amplitude threshold criterion. This was always more than 2.5 times, but usually 3.5 times the standard deviation of noise.

Analysis of modal behaviour was performed as described in Zahradníková & Zahradník (1995). In brief, the frequency of sweeps with specific values of P_o, t_o and n_o was statistically analysed, on which basis the modal states of individual segments were identified. All data were fitted to theoretical curves using the minimum χ² method. This and other secondary analyses, calculations and statistics were performed using Origin (version 4.0, Microcal Software, Northampton, MA, USA). Statistical significance was evaluated by Student's t test. The results are presented as means ±s.e.m.

The data used for quantitative analysis were obtained from twenty-three channels. A qualitatively similar effect was observed in twenty-one other membranes, which were not analysed quantitatively because of the presence of more than one channel.

RESULTS

The reported values of maximal steady-state open probability of skeletal muscle Ca²⁺ release channels activated solely by Ca²⁺ vary over quite a wide range, from about 0.02 to 0.2 (Fill et al. 1991; Chu et al. 1993; Ma & Zhao, 1994). We found that even at optimal activating Ca²⁺ concentrations (for instance, at 20 μm), the steady-state activity of single skeletal muscle Ca²⁺ release channels at a constant membrane potential was consistently very low (P_o= 0.006 ± 0.003 at +30 mV with 250/50 mm Cs⁺ cis/trans). However, the channel activity immediately after an abrupt change in membrane potential substantially increased. In order to elucidate the mechanism of activity changes in response to alterations in membrane potential, we applied three different voltage protocols. First, we investigated this effect in symmetrical 250/250 mm Cs⁺ solutions by measuring open probability immediately before and after a voltage change from −40 to +40 mV. As shown by the examples in Fig. 1A, openings of the channel at a membrane potential of −40 mV were prominent if the channel was held between the pulses at a holding potential of +40 mV. If the channels were held at −40 mV for a period of 3 s, the open events at this potential were much less frequent (Fig. 1B), but upon stepping up the voltage to +40 mV, the open probability of the channels significantly increased. We have collected data immediately before and after the voltage steps from −40 mV to +40 mV (according to a protocol similar to that shown in Fig. 1B). As summarized in Table 1, the P_o immediately after switching from a holding potential of +40 mV to the −40 mV test potential was 0.21 ± 0.04. Therefore, the channels seemingly did not desensitize at +40 mV. There was a significant decrease of P_o during a sojourn at −40 mV, as well as an increase in P_o after stepping back to +40 mV (Table 1, row 2). Moreover, the increase in P_o at +40 mV in those episodes, in which the open probability during the last 1 s of the prepulse was less than 20 % of the average (P_pre < 0.01; Table 1, row 3), was not significantly different from the overall increase. Therefore, the increase in P_o at +40 mV induced by the −40 mV prepulse cannot be interpreted as channel deactivation. We also found no correlation between P_o at −40 and +40 mV (r= 0.11± 0.15, n= 10) and between the increase in P_o at +40 mV and P_o at −40 mV (r= −0.26± 0.20, n= 10). Based on the above findings we interpret the increase in P_o induced by the voltage change from −40 to +40 mV as a voltage change-induced removal of steady-state inactivation.

Figure 1. The activity of a Ca²⁺ release channel under different voltage protocols.

A, single-channel currents in symmetrical ionic conditions and in the presence of 20 μm cis Ca²⁺, recorded immediately after switching to a membrane potential of −40 mV from a holding potential of +40 mV. B, channel activity recorded immediately before and after a voltage step from −40 mV to +40 mV (see top for voltage protocol). The lipid mixture was 70 % PE and 30 % PC (see Methods).

Table 1.

A summary of the changes in open probability induced by voltage changes in symmetrical 250/250 mM Cs⁺ solutions and 20 μm cis Ca²⁺

	Vh (mV)	Vpre (mV)	Vtest (mV)	Ppre	Ptest	n
1	+40	—	—40	—	0.21 ± 0.04	6
2	+40	−40	+40	0.05 ± 0.01 a	0.11 ± 0.03b	10
3 *	+40	−40	+40	0.002 ± 0.0001	0.07 ± 0.02 c	10

Abbreviations: V_h, holding potential (3 s duration); V_pre, prepulse potential (3 s duration); V_test, test pulse potential (1 s duration); P_pre, open probability during the last 1 s of the prepulse; P_test, open probability during the test pulse; n, number of independent experiments.

^*Episodes in which P_pre < 0.01 selected from data in row 2.

^aSignificantly less than P_test in row 1 (P < 0.0001)

^bSignificantly more than P_pre (P < 0.05)

^cNot significantly different from P_test in row 2 (P > 0.25), but significantly more than P_pre (P < 0.005).

Second, to characterize the dependence of the voltage change-induced P_o increase on prepulse amplitude, we applied 1 s-long prepulses to 0 and −50 mV, then compared single-channel activity (recorded at +30 mV immediately after the prepulse) with control values that were obtained at constant potential (i.e. at +30 mV). We used asymmetrical ionic conditions (250/50 mm Cs⁺ cis/trans) to be able to distinguish between the effects of voltage per se and those of current polarity. In order to avoid the results being affected by time variations of channel activity at either voltage (see Armisén, Sierralta, Vélez, Naranjo & Suárez-Isla, 1996), the prepulses were applied in a repetitive, cyclical fashion (control, 0 mV, −50 mV). As shown by the records in Fig. 2A, the channel activity after the prepulse monotonically increased with the amplitude of the voltage difference between the prepulse and the test pulse, despite the fact that the single-channel current during the prepulse reversed in polarity (−6.0 ± 0.8 pA at −50 mV; +11.8 ± 0.7 pA at 0 mV; and +20.9 ± 0.8 pA at +30 mV). To test whether the observed effect is a consequence of the voltage change, or whether it is mediated (or modulated) by possible local changes in Ca²⁺ concentration induced by the effect of voltage on Ca²⁺ binding to phospholipids, we applied identical pulse protocols to channels incorporated into lipid bilayers containing either 0 or 30 % phosphatidylserine, an acidic phospholipid that is prominently involved in non-specific Ca²⁺ binding (McLaughlin, Mulrine, Gresalfi, Vaig & McLaughlin, 1981). Figure 2B demonstrates that the increase in open probability induced by prepulses to either 0 or −50 mV was independent of the lipid composition of the bilayer. Figure 2C summarizes the changes in the single channel parameters (i.e. P_o, frequency of opening (n_o) and average open time (t_o)), that were obtained from the test segments recorded immediately after the prepulse, as a function of prepulse amplitude. The results shown in Fig. 2C also indicate that the rise in P_o resulted from a combined augmentation of n_o and t_o, the latter occurring upon the increase in the prepulse amplitude.

Figure 2. The effect of voltage prepulses on channel activity at +30 mV.

A, representative records of channel activity measured either in control conditions (top), or immediately after a 3 s prepulse to 0 mV (ΔV= −30 mV; middle) or to −50 mV (ΔV= −80 mV; bottom). The lipid mixture in this experiment was 70 % PE and 30 % PC. B, comparison of the effect of prepulse amplitude on channel open probability in channels incorporated into a bilayer made of 70 % PE and 30 % PC (□) or 50 % PE, 30 % PS and 20 % PC (). C, the values of open probability (top), average open time (middle), and frequency of openings (bottom) at the three conditions (•), and the values recorded at +30 mV, 2.5 s after a 1 s prepulse to −50 mV (○). The data represent means ±s.e.m. of 6 experiments, 3 each of either lipid composition.

Third, in order to characterize the kinetics of the voltage change-induced P_o rise, we applied prepulses of different duration (with the same membrane potential of −50 mV). The duration of the prepulses was altered cyclically (0, 250, 500, 1000 and 2000 ms) in order to ensure that the measured values were not affected by long-term changes in channel activity. The activity during test segments seemingly increased with the duration of the prepulse (Fig. 3A). Furthermore, as shown in Fig. 3B and C, the rate of onset of the effect and that of the subsequent return to the steady state (i.e. the relaxation from the voltage change-induced increase in P_o, n_o and t_o) differed significantly (see also Table 2).

Figure 3. The effect of changing the duration of the prepulse on channel activity at +30 mV.

A, representative traces recorded in control conditions, and immediately after the return of the membrane potential to +30 mV from a prepulse to −50 mV lasting 250, 500, 1000 and 2000 ms. B and C, time course of changes in open probability (top), frequency of channel openings (middle), and average open time (bottom) at +30 mV with prepulses of variable duration to −50 mV (B), and after a 2 s prepulse to −50 mV (C). All data represent means ±s.e.m. from 4 experiments, in which the lipid composition was 50 % PS, 30 % PE, and 20 % PC (see Methods).

Table 2.

Time constants of onset of changes in channel activity induced by prepulses to −50 mV, and return to the steady state

	τon (ms)	τoff (s)
Open probability	231 ± 196	1.93 ± 0.26
Frequency of openings	409 ± 227	2.23 ± 0.32
Average open time	187 ± 250	2.65 ± 0.48

τ_on, onset of changes. τ_off, return to the steady state, i.e. relaxation.

The results in Fig 2C and 3B indicate that the effect of the prepulses on the open probability was not a simple upscaling of the same kind of openings. This suggests that the voltage-induced effect might be due to the occurrence of intermodal transitions, similar to those found for the cardiac channel (Zahradníková & Zahradník, 1995). Therefore, we also analysed the effects of voltage prepulses on the modal transitions of the channel. The frequency distributions of 409.6 ms data segments for average open time (Fig. 4) revealed the occurrence of three types of segments with characteristic kinetic parameters (see Table 3). Three modes of activity (I, inactivated; L, low open probability; and H, high open probability), similar to those found for the cardiac ryanodine receptor channel (Zahradníková & Zahradník, 1995), were clearly distinguished. Under the conditions used during the course of this study, the H-mode had a 4 times larger P_o than that of the L-mode, the difference seemingly stemming from a 2-fold increase in t_o and a 2.5-fold increase in n_o.

Figure 4. Frequency histogram of channel average open time.

Data were best fitted by a double Gaussian function (Aexp((x - x_i)/2w_i)/w_i√2π, where i = 1 or 2) with parameters A₁= 35.9, x_c1= 0.52 ms, w₁= 0.31 ms, A₂= 3.6, x_c2= 1.41 ms, w₂= 1.13 ms. The curve had a minimum around t_o= 1 ms, which defines the critical average open time value (t_c) for distinguishing between modes L and H.

Table 3.

Characteristics of channel gating in the three modes of activity

	H-mode	L-mode	I-mode
Open probability	0.18 ± 0.02	0.05 ± 0.004	0.0
Frequency of openings (s−1)	75 ± 9	30 ± 2	—
Average open time (ms)	1.49 ± 0.48	0.70 ± 0.08	—

The values of P_o, n_o and t_o for the individual modes did not depend on either the amplitude or the duration of the prepulse, nor they were affected by the time elapsed from the end of the prepulse. However, the relative probability of the three modes was strongly affected by the voltage prepulses. As seen in Fig. 5, the voltage prepulses induced a decrease in the probability of I-mode and a considerable increase in H-mode (from 0 to 0.25) as well as in L-mode (from 0.1 to 0.5) probabilities. The re-distribution of modes occurred with monoexponential time courses with similar time constants (see Table 4). After switching back to the holding potential of +30 mV, the proportion of each mode returned to the equilibrium distribution with slower kinetics (Table 4). The probabilities of both H- and L-mode after the prepulse increased with increasing potential difference between the prepulse and the holding voltage. However, the H-mode became more prominent with respect to the L-mode as the prepulse amplitude was increased (i.e. P_H/P_L= 0.1 and 0.4 for prepulses to 0 and −50 mV, respectively).

Figure 5. The effects of voltage prepulses on modal behaviour of channels.

A, time course of modal changes at +30 mV induced by prepulses of variable duration to −50 mV. B, time course of equilibration of modes at +30 mV after a 2 s prepulse to −50 mV. C, voltage dependence of mode probability immediately after a 1 s prepulse to the indicated potentials (•), and 3.5 s after return to +30 mV from a −50 mV prepulse (○).

Table 4.

Time constants of onset of changes in occurrence of modes induced by prepulses to −50 mV, and return to the steady state

	τon (ms)	τoff (s)
H-mode	256 ± 240	1.12 ± 0.32
L-mode	291 ± 164	11.0 ± 4.0
I-mode	227 ± 83	5.5 ± 1.0

τ_on, onset of changes. τ_off, return to the steady state.

The modal analysis might be slightly affected by the time resolution of our data. By comparison with the cardiac RyRC, for which a detailed modal analysis has been performed (Zahradníková & Zahradník, 1995), we can estimate that a proportion of openings (about 50 % in the L-mode and only about 25 % in the H-mode) was missed. The average open time within L-mode might have been slightly overestimated, while the frequency of openings within L-mode slightly underestimated. However, the presence of unresolved openings within segments assigned as L-mode or H-mode will have no measurable effect on the correct assignment of modes, nor on the open probabilities within individual modes. Another possible source of error in our modal analysis is due to the presence of ‘ghost’ null segments, i.e. those L-mode segments that are erroneously assigned as I-mode due to the presence of unresolved openings only (Zahradníková & Zahradník, 1995). For the cardiac channel, less than 5 % of the I-segments were erroneously assigned. We estimate that ∼5–10 % of the I-mode segments might have been erroneously assigned here for the skeletal channel. The greatest probability of erroneous assignment is under conditions of low open probability (no prepulse). This might slightly affect the on-rates of the prepulse-induced I-L transitions. However, there should be no measurable effect on the observed open probabilities and their time courses. Therefore, the true values of the I to L transition rate might be slightly lower than the estimated value of 3.4 s⁻¹, but not less than 2.5 s⁻¹. On the other hand, the true value of the L to I transition rate might be higher than the estimated 0.18 s⁻¹, but not more than 0.4 s⁻¹ (see Table 2 and Table 4). We interpret the extremely slow rate of exit of the channel from the L-mode during equilibration (τ_off= 11 s) as an indication of a sequential H to L to I transition.

DISCUSSION

The main findings of this work are (i) that the skeletal muscle RyRC in the steady state, even in the presence of activating Ca²⁺ concentrations, is inactivated to a high extent, and (ii) that this inactivation can be relieved by changing the membrane voltage. We interpret the observed phenomenon as a voltage change-induced removal of a steady-state inactivation (instead of a transient activation) for the following reasons: (i) in the steady state, there is a high proportion I-mode channels; (ii) the increase in open probability does not have clear turn-on kinetics at the test potential; rather, it seems to be instantaneous; and (iii) the increase in open probability at the test potential displays a monoexponential dependence on the duration of the prepulse.

The steady-state open probability of the RyRC that we report here is lower than other values in the literature for the channel activated with 20 μm Ca²⁺. Many of the previously published values were obtained with a stimulation protocol during which the channel is held at 0 mV and repetitively pulsed to a non-zero membrane potential, an experimental protocol which avoids long-term polarization of the bilayer (which supposedly decreases bilayer lifetime; Fill et al. 1991). With repetitive pulsing, open probability values for the skeletal RyRC with about 10 μm Ca²⁺ (as the sole activating ligand) were in the range 0.13–0.2 for rabbit, pig or human skeletal RyRC (Fill et al. 1990, 1991; Chu et al. 1993; but see also Ma & Zhao, 1994). Therefore, the differences between our data and those of others are most probably, at least partially, due to our use of a steady holding potential. There are also other differences in conditions used by different laboratories, which could affect the experimentally observed open probability values for the RyRC: the type of anion used in the saline solution (Fruen, Kane, Mickelson & Louis, 1996) and the different phosphorylation state of the channel in different preparations (Herrmann-Frank & Varsanyi, 1993). In addition, Barg, Copello & Fleischer (1997) reported that the skeletal RyRC, when bound to the endogenous FK506 binding protein FKBP12, is quiescent, while it becomes more active upon removal of this protein. In fact, by using a polyclonal FKBP12 antibody, we detect SR membrane-bound and rapamycin-removable FKBP12 in our preparations (not shown). We can speculate that the presence of the tightly bound FKBP12 in our preparations might be another reason for the low open probability. Still another factor that has to be considered is the redox state of the RyRC, as it has been shown that the P_o of the skeletal RyRC is increased upon oxidation (Zable, Favero & Abramson, 1997).

The extent of membrane potential per se has been shown to affect the gating of the release channel at pH < 7.2 (Ma, Fill, Knudson, Campbell & Coronado, 1988) or after covalent modification (Zahradníková & Zahradník, 1993). The changes in membrane voltage have also been shown to affect channel gating; however, the data are sporadic and inconsistent. The Ca²⁺ release channel has been reported to enter a long-lived closed state either after large pulses to positive voltages ≥ 40 mV (both chicken RyRC1 and RyRC3, Percival, Williams, Kenyon, Grinsell, Airey & Sutko, 1994; sheep RyRC2, Sitsapesan, Montgomery & Williams, 1995), or at negative voltages ≤ −50 mV (rabbit skeletal RyRC1; Ma, 1995). In some cases, almost all studied channels were observed to undergo this phenomenon (Percival et al. 1994). In other instances, the release channels were more prone to this behaviour if they were activated to high P_o levels by organic ligands, but less so if activated solely by Ca²⁺ (Sitsapesan et al. 1995); or the effect was seen only in a particular subpopulation of channels and only in the presence of an activating organic ligand (Ma, 1995). The effect was interpreted as inactivation (Percival et al. 1994; Sitsapesan et al. 1995) or desensitization (Ma, 1995).

The effect we describe here is definitely not due to either desensitization or even deactivation, since the inactivation could be relieved without (preceding) activation of the channel (Fig. 1B). An additional difference between our findings and the findings of others is that, as we observed, the inactivation and its relief by a voltage change occur at moderate values of membrane potential (ΔV= 30 mV; Fig 2 and Fig 5C) and at either polarity (Table 1). The reports mentioned above, on the other hand, describe the decrease in open probability after a step to certain membrane potentials, but the protocols did not address specifically the question of whether the channel was in the inactivated state before the pulse. Ma (1995) could not distinguish between active and inactivated channels at a holding potential of 0 mV, when the current amplitude at his ionic conditions was zero. Sitsapesan et al. (1995) reported that the channel that inactivated at +40 mV did not respond to a new challenge with the agonists without being repolarized to −40 mV, but they did not study the response of the channel to repolarization to −40 mV without first removing the agonists. Therefore, we cannot conclude whether there are some common points between the effect that we report here and the findings of others.

Another finding of the present work is that rabbit skeletal muscle RyRCs have modal properties similar to those of the cardiac isoform (Zahradníková & Zahradník, 1995). The occurrence of modes of activity has been previously documented for both the neuronal and skeletal isoforms (Armisén et al. 1996), as well as for the chicken skeletal muscle β-isoform of RyRC (Percival et al. 1994). However, the methods of analysis used by these authors did not permit the estimation of intramodal open probabilities, or the description of the properties of the inactivated mode. In this work we described the most basic properties of the three modes of activity for the skeletal RyRC.

We observed the voltage change-induced variation in channel gating when we applied prepulses to 0 and −50 mV, where the single-channel current was positive and negative, respectively. This suggests that the interaction of ions with the channel pore as a causative factor for the alterations in the voltage change-induced channel gating is unlikely. Since the magnitude of the observed effects of prepulse amplitude on the open probability did not depend on the lipid composition of the bilayer (see Fig. 2B), it seems that the voltage-induced changes in channel gating are not mediated by an effect of voltage on the local Ca²⁺ in the vicinity of the lipid bilayer (which could be due to the voltage dependence of Ca²⁺ binding to acidic phospholipids; McLaughlin et al. 1981). Furthermore, effects that are mediated by Ca²⁺ binding to lipids would occur on a much faster time scale (Langer & Peskoff, 1996).

The large effect on the opening frequency and the re-distribution of the channels between I- and L-modes seen with prepulses to 0 mV, as well as the qualitative indications that the effect can be evoked by prepulses of either polarity (see Fig. 1A and B), also suggests that it is not the absolute value of the potential, but rather the voltage change that determines the response. Therefore, our data suggest that the voltage change (i.e. a transient electric field) directly affects the channel protein, which in turn results in alterations in the behaviour of the channel. At present, our data do not allow specification of the mechanism of this effect. Further insight might be obtained by studying the effect of voltage changes with a wider range of cis Ca²⁺ concentrations and in the presence of other ligands such as ATP and Mg²⁺. However, there is a specific mechanism which should be given consideration: the voltage change might induce rapid changes in the sensitivity of the channel to Ca²⁺ ions. As the binding of Ca²⁺ to the RyRC is extremely fast (Sitsapesan et al. 1995; Schiefer, Meissner & Isenberg, 1995), the channel activity can be expected to reflect the changes in calcium affinity of the RyRC instantaneously. However, the data presented here do not seem to support this possibility, since at 20 μm Ca²⁺ the channel is expected to be fully activated in the steady state. Therefore, an increase in channel affinity would not bring about such dramatic changes in P_o as were observed.

In skeletal muscle E-C coupling, a direct or indirect mechanical interaction between the voltage sensor DHPR and the SR release site, i.e. the RyRC, is thought to be a key element of signal transduction (Ríos & Brum, 1987). This mechanical coupling is envisioned as the transmission of conformational changes between these two proteins (Ríos et al. 1993). The results we describe here raise the possibility that these coupled conformational changes might involve not only mechanical but also electrical events: the voltage sensor DHPR must behave as either a permanent or inductive dipole and, as our results seem to suggest, the same might apply to the skeletal RyRC. In this regard it is noteworthy that the II-III cytoplasmic loop of the skeletal DHPR, which has been shown to be critical in skeletal E-C coupling (Tanabe, Beam, Adams, Niidome & Numa, 1990), contains a functional subdomain which appears to be responsible for the activation of the RyRC (El-Hayek, Antoniu, Wang, Hamilton & Ikemoto, 1995), 25 % of which is composed of positively charged amino acid residues. This segment might be involved in transmitting the transverse tubule voltage change to a corresponding charged segment in the RyRC amino acid sequence. A highly acidic region has been found on the cytoplasmic surface of the RyRC that seems to be necessary for E-C coupling and was proposed to participate in the protein-protein interaction involved in this process (Yamazawa, Takeshima, Shimuta & Iino, 1997). Whether these two segments are juxtaposed and interact with each other is still to be determined. However, the results we present here raise the possibility that the communication between the DHPR and the RyRC (i.e. skeletal E-C coupling) might indeed involve a dipole-dipole interaction, since the skeletal muscle RyRC apparently possesses a voltage-sensing capacity.