Single-Channel Studies of the Action of ( + )-Tubocurarine on Locust Muscle Glutamate Receptors

( + )-Tubocurarine (TC) has been classically regarded as a competitive antagonist at nicotinic cholinergic synapses (Jenkinson, 1960). However, recent studies of the effects of this agent on channels gated by acetylcholine receptors indicate two additional properties, open channel block and agonist-like channel gating. TC blocks channels gated by acetylcholine receptors on extrajunctional membrane of Aplysia neurones (Marty, Neild & Ascher, 1976; Ascher, Marty & Neild, 1978), at frog neuromuscular junctions (Manalis, 1977; Katz & Miledi, 1978; Colquhoun, Dreyer & Sheridan, 1979; Shaker et al. 1982), rat ganglion cells (Ascher, Large & Rang, 1979) and rat muscle (Gibb & Marshall, 1983, 1984). It also gates channels associated with acetylcholine receptors in rat myotubes (Morris, Jackson, Lecar & Wong, 1982; Takeda & Trautmann, 1984), bovine chromaffin cells (Lambert, Nooney & Peters, 1984) and human myotubes (Jackson, Lecar, Askanas & Engel, 1982).

Yamamoto & Washio (1979, 1983) showed that TC blocks transmission postsynaptically at an insect (beetle; Tenebrio molitor) nerve—muscle junction, where glutamate is the likely transmitter (Yamamoto & Washio, 1979), through open channel block, and this has since been shown to be true also for the glutamatergic neuromuscular junction of locust leg muscle (Cull-Candy & Miledi, 1983). In this paper we describe the effects of TC on the ionic channel gated by extrajunctional glutamate D-receptors (Cull-Candy & Usherwood, 1973) on leg muscle of adult locusts which have been investigated mainly using the patch-clamp technique (Neher, Sakmann & Steinbach, 1978; Patlak, Gration & Usherwood, 1979; Gration & Usherwood, 1980; Cull-Candy, Miledi & Parker, 1981).

Experiments were performed at room temperature (21–23 °C) on the extrajunctional membrane of extensor tibiae muscle from metathoracic legs of adult locusts (Schistocercagregaria, ♀) pretreated with 1–2× 10⁻⁶moll⁻¹ concanavalin A (Sigma, Grade IV) for 30min to block receptor desensitization (Mathers & Usherwood, 1976, 1978; Patlak et al. 1979). The muscle was bathed in standard locust saline (in mmoll⁻¹: NaCl, 180; KCl, 10; CaCl₂, 2; Hepes, 3; pH adjusted to 6·8 with NaOH) and fibres were voltage-clamped using a conventional two-electrode technique (Anwyl & Usherwood, 1974). Patch pipettes, with fire-polished tips of 1–2 μm diameter and containing 10⁻⁵–10⁻³ moll⁻¹ sodium L-glutamate (Sigma) either alone or with 10⁻⁵–5× 10⁻⁴mol l⁻¹ TC chloride (Koch-Light), were pressed onto the muscle fibre surface to form mega-ohm seals. Recordings of currents passing through the area of membrane under the tip of the patch pipette were stored on an analogue FM tape recorder. The patch amplifier was similar to that described by Sigworth (1983) and recordings were made with a bandwidth of d.c. to 3 kHz. Only sites giving good signal-to-noise ratios (⩾3:1) were studied, thus allowing resolution of channel openings and closings as brief as 100μs. Channel open-closed kinetics were analysed using procedures that have already been described (Gration, Lambert, Ramsey, Rand & Usherwood, 1982). Unless otherwise stated, the data were recorded under voltage-clamp at a membrane potential (V_m) of —100 mV. A new patch pipette was used for each concentration of TC and/or L-glutamate. The presence of two or more channels under the patch electrode was observed in some recordings by the occurrence of two simultaneous open events. Such recordings were not included in our analyses. If it is assumed that the glutamate receptor channel complexes are homogeneous, it is possible to predict the frequency of occurrence of simultaneous events at a recording site which would be expected if more than one receptor channel were isolated by a patch pipette (Patlak et al. 1979). Such predictions suggest that for L-glutamate concentrations ⩾10⁻⁴moll⁻¹ and for low concentrations of TC at these glutamate concentrations the frequency of occurrence of simultaneous open events would be sufficiently high to enable unequivocal identification of patches containing more than one receptor channel. However, with concentrations of L-glutamate <10^{− 4}moll^{− 1}, when channel opening rate is low, and with high concentrations (⩾10⁻⁴moll⁻¹) of TC, when channel openings are brief, the frequency of simultaneous events falls below the level at which identification of sites containing more than one glutamate receptor channel can readily be made. We cannot be certain, therefore, that recordings which did not contain simultaneous open events, obtained under these conditions, resulted from the activity of only a single receptor channel complex.

The effects of TC on the neurally evoked twitch contraction of the locust metathoracic, femoral retractor unguis muscle were also examined, using the technique of Usherwood & Machili (1968).

All parameter fits to equations were obtained using a derivative-free, non-linear least-squares procedure (NAG subroutine E04FDF). Fitted parameters are quoted ±standard error of parameter estimate, the latter having been obtained via inversion of an approximation of the Hessian (NAG subroutine E04YCF). Open time distributions were fitted using mixtures of exponential functions by standard techniques, described in more detail in Kerry et al. (1986).

Visual inspection of single-channel current records shows that TC leads to diminished activity of the glutamate receptor (GluR) channel, involving a reduction in frequency of openings and of open channel life-time (Fig. 1). The probability of the channel being open (p_o) was estimated directly from the data. With a constant Concentration of L-glutamate (10⁻⁴moll⁻¹) in the patch electrode, p_o was 0·061 [±0·016 (S.D.)] in the absence of TC and decreased with increasing TC concentrations (Fig. 2) to 0·0031 (±0·0007) in the presence of 5×10⁻⁴moll⁻¹ TC (Fig. 2A). Although these results could be explained either by channel blocking (open and/or closed) and/or receptor antagonism by TC, the evidence presented below suggests that the latter hypothesis is unlikely.

TC depressed the neurally evoked twitch contraction of the locust retractor unguis muscle in a stimulus-frequency-dependent fashion (Fig. 3). However, when stimulation of the muscle was discontinued for even a brief period (e.g. approx. 1 min) a suprathreshold stimulus applied to the retractor unguis nerve at the end of this period evoked a twitch contraction which was similar in amplitude to that seen in the absence of TC. This observation cannot be explained by competitive antagonism. Furthermore, if, as is suggested by our single-channel data, TC blocks the channels gated by postjunctional glutamate receptors on locust muscle, then the twitch data imply that this is at the level of the open rather than the closed channel.

In our single-channel studies with 10⁻⁴moll⁻¹ L-glutamate the mean channel open time (m_o), determined directly from the data (thereby excluding brief openings and brief closings which were not resolved), was 1·13 (±0·22) ms in the absence of TC. In the presence of 5× 10⁻⁴moll⁻¹ TC, m_o was 0·26 (±0·02) ms. Values for m_o at different TC concentrations, but corrected for the absence of brief openings in the patch recordings due to the limited frequency response of the recording system (Colquhoun & Sigworth, 1983), are presented graphically in Fig. 4A. The corrected m₀ values are slightly smaller than their uncorrected counterparts (not illustrated), but otherwise exhibit the same decline with increasing TC concentration. Increasing the concentration of L-glutamate whilst maintaining a constant TC concentration did not substantially reverse the effect of TC. In the absence of TC, p_o and m_o both increased as the concentration of L-glutamate was raised. For example, p_o increased from 0·061 (±0·016) to 0·78 (±0·17) when the concentration of L-glutamate was raised from 10⁻⁴ to 10⁻³moll⁻¹. However, when 5×l0⁻⁴moll⁻¹ TC was also present in the patch pipette the same change in L-glutamate concentration only raised p_o from 0·0031 (±0·0007) to 0·0048 (single experiment). These results, albeit from a single experiment, suggest that the primary effect of TC at the GluR is not one of competitive antagonism.

At membrane potentials < — 70mV, the amplitudes of glutamate channels recorded with our mega-ohm seal technique were small, and poor signal/noise ratios precluded measurements of channel kinetics under the limited bandwidth recording conditions. Consequently, studies of the possible voltage dependence of TC action bn the glutamate receptor channel were restricted to membrane potentials between —70 mV and —120 mV. Furthermore, given the complex kinetics of the glutamate receptor channel of locust muscle (Gration et al. 1981a; Gration, Lambert, Ramsey & Usherwood, 1981b; Cull-Candy & Parker, 1982; Ball, Sansom & Usherwood, 1985; Ashford et al. 1984a,b), a full comparison of channel kinetics in the presence and in the absence of TC cannot be obtained without recording large numbers of channels from single sites at, at least, three well-separated V_m values. Such timeconsuming studies were largely incompatible with the employment of voltage-clamp to maintain muscle fibre membrane potentials at highly hyperpolarized levels. However, pooled data obtained from six recording sites (800–6000 channels for each V_m) showed that between —70mV and —120mV channel block by TC in the concentration range 10⁻⁵–2×10⁻⁴moll⁻¹ exhibited little or no voltage dependence (Table 1).

Analysis of the single-channel data in terms of a simple model for open channel block (Adams, 1976; Neher & Steinbach, 1978; Ogden, Siegelbaum & Colquhoun, 1981) has been undertaken with a view to obtaining estimates of kinetic parameters describing the blocking action.

The effect of TC concentration on p₀ is given in Fig. 2 (where the curve in Fig. 2A corresponds to equation 2). Fitting equation 2 to the data (see Materials and Methods) results in parameter estimates of K_C = 20·3 (±4·4) and K_B = 5·7×10⁵ (±4·41×10⁵) (moll⁻¹)⁻¹. The latter corresponds to a dissociation constant, K_D, of 1·75 μmoll⁻¹ (V_m = —100 mV), which should be compared to aK_D of 0-044/zmoll⁻¹ (V_m = —100 mV) for the interaction of TC with the nicotinic acetylcholine receptor (nAChR) of frog muscle (Colquhoun et al. 1979) and a K_D of 0·25μmoll⁻¹ (V_m = —80 mV) for the interaction of TC with nAChR of Aplysia neurones (Ascher et al. 1978).

Single-channel data may be employed to estimate the rate of association of blocker with channel (e.g. Neher & Steinbach, 1978; Ogden et al. 1981). When singlechannel activity in the presence of an open channel blocker occurs in bursts and blocking and unblocking are both rapid, as is the case for blocking of the Na⁺ channel by QX-222 and of the nAChR by benzocaine, both m_o and the mean block time, mb, may be determined directly from single-channel data. The former allows one to estimate k_on and the latter k_off. However, the action of TC on GluR is characterized by a relatively slow unblocking of the open channel. This means that only m_o, and hence k_on may be determined directly from single-channel data.

The influence of TC concentration on the open time probability density function (pdf) at 10⁻⁴moll⁻¹ L-glutamate was also investigated. Ashford et al. (1984a,b) and Kerry et al. (1986) have shown that three exponential functions are required to fit the open time distribution for 10⁻⁴moll⁻¹ L-glutamate, with time constants (τ) of 0·40ms (τ₁), l·2ms (τ₂) and 3·4ms (τ₃) (Kerry et al. 1986). The limited frequency response of the recording system precluded an investigation of the effects of TC on τ₁ and it was not possible to study its effects on τ₃ because of the relative paucity of long channel openings, so we have therefore concentrated our analysis on τ₂. However, in view of these constraints and in the absence of a complete description of the mechanism of channel gating it would be unwise to be dogmatic about a molecular interpretation of the action of TC on GluR based upon its effect on τ ₂. TC reduces τ ₂ in a concentration-dependent manner. A plot of τ ₂versus log TC concentration is presented in Fig. 5A accompanied by a plot of the reciprocal of τ₂versus log TC concentration in Fig. 5B. Using equation 3 to fit the TC concentration-dependence of τ₂ gives a value for k_on of 9 · 44 × 10³ (±4 · 02 × 10³) ms⁻¹ (moll⁻¹)⁻¹ (V_m = — 100mV). This association rate is comparable to that estimated by Colquhoun et al. (1979) for association of TC with the frog nAChR, i.e. k_on = l · 2 × l0⁴ms⁻¹ (moll⁻¹)⁻¹ (V_m = —100mV). Both values are in the range of those commonly held to be the diffusion-limited rate of association of a drug molecule with its receptor (Neher & Steinbach, 1978). These results suggest that TC binds to at least one of the open states of GluR.

The rate of dissociation can be estimated indirectly via k_off = k_on KB⁻¹. Using the estimate of k_on based on corrected m_o values, this gives a value for k_off of 1 · 53 × 10⁻²ms⁻¹, corresponding to a mean block time (k_off) of 65 · 4ms. This should be compared to a mean block time of approximately l · 4s (Colquhoun et al. 1979) for interaction of TC with frog nAChR. Clearly the lower affinity of GluR for TC is reflected in the somewhat higher rate of dissociation of the drug molecule from the GluR compared with that for the nAChR.

The blocking model may be displayed in terms of the predicted TC concentrationdependence of the mean closed time, m_c (Fig. 6). Note that, since (1 —p_o)/p_o = m_c/m_o, m_c is clearly not independent of p_o and m_o. With m_c corrected for the loss of brief closings, the fit to the m_c TC concentration data can be seen to account adequately for the long closed times observed in the presence of high concentrations of TC. For example, with 5 × 10 ^{− 4} moll⁻¹ TC, corrected m_c = 89 · 6 (±21 · 6) ms.

The results presented here support the view that the cholinergic antagonist To blocks glutamate receptor channels on locust muscle membrane when they are open. It has already been shown that TC decreases the amplitude and the decay time of excitatory postsynaptic currents and ionophoretically induced glutamate currents at locust (Cull-Candy & Miledi, 1983) and beetle (Yamamoto & Washio, 1979, 1983) neuromuscular junctions. This is similar to its observed action on cholinergic systems such as a molluscan neurone (Marty et al. 1976) and the more extensively studied frog skeletal neuromuscular junction (Manalis, 1977; Katz & Miledi, 1978; Colquhoun et al. 1979; Shaker et al. 1982). In these systems it has been proposed that TC blocks the open ionic channel in a manner similar to the action of some local anaesthetics.

The kinetic scheme shown in equation 1 for open channel block which has been used to analyse the interaction of TC with GluR is undoubtedly over-simplified. This model considers channel gatings as a pseudo-unimolecular process yet recent analysis of the gating of GluR using single-channel kinetics (Ashford et al. 1984 a,b; Kerry et al. 1986) demonstrates that the gating mechanism is more complex than this. For example, in the presence of 10 ⁻⁴moll⁻¹ L-glutamate, open time pdfs are best fitted by the sum of three exponential functions, implying that there are (at least) three kinetically significant open states of the GluR channel. The data also imply the presence at 10⁻⁴moll⁻¹ L-glutamate of four kinetically distinct closed states.

A further aspect of the kinetic complexity of GluR gating is the autocorrelation of successive open times and of successive closed times of the channel (Ashford et al. 1984 a,b;Kerry et al. 1986), a feature which accounts for the observed clusterings of openings (Gration et al. 1981b). The positive autocorrelations for the dwell times are indicative of multiple pathways linking the open and closed states of the channel (Labarca et al. 1985) and would account for the observed clustering of channel openings (Ball et al. 1985). An investigation of the transition pathways linking the open and closed states of the GluR is currently in progress, but until these and other features of this system are better understood it is not possible to analyse the mode of action of TC on GluR beyond the level of the simple model given in equation 1. However, it is of interest to note that the autocorrelation functions for channel openings and for channel closings were weakened when TC was present in the patch pipette together with L-glutamate, and approached zero with high (5 × 10⁻⁴moll⁻¹) concentrations of TC. This result is consistent with channel block of a non-linear, multistate channel, in which the blocking agent has a low unblocking rate.

Although our data argue for open channel block we have no evidence that this process is voltage-sensitive. However, our studies were restricted to the voltage range —70 to —120 mV and were undertaken after pretreatment of the muscle with concanavalin A. Both Cull-Candy & Miledi (1983) and Yamamoto & Washio (1983) showed a voltage-dependent action of TC on insect postjunctional glutamate receptors, but this was only revealed at high membrane potentials (i.e. —90 to —150 mV). Similarly, the voltage-dependent effect of TC on the frog neuromuscular junction was most obvious during strong hyperpolarizations (⩾ —130 mV) (Katz & Miledi, 1978). Magleby, Pallotta & Terrar (1981) failed to observe any voltagedependence of TC action on the rundown of end-plate current amplitudes during repetitive stimulation of the rat diaphragm muscle within the membrane potential range —60 to —120 mV. This led them to conclude that the increased rundown of end-plate current amplitudes in TC was not due to voltage-dependent open AChR channel block.