Lidocaine induces a slow inactivated state in rat skeletal muscle sodium channels

Abstract

1. Local anaesthetics such as lidocaine (lignocaine) interact with sodium channels in a manner that is exquisitely sensitive to the voltage-dependent conformational state of the ion channel. When depolarized in the presence of lidocaine, sodium channels assume a long-lived quiescent state. Although studies over the last decade have localized the lidocaine receptor to the inner aspect of the aqueous pore, the mechanistic basis of depolarization-induced ‘use-dependent’ lidocaine block remains uncertain.

2. Recent studies have shown that lowering the extracellular Na⁺ concentration ([Na⁺]_o) and mutations in the sodium channel outer P-loop modulate occupancy of a quiescent ‘slow’ inactivated state with intermediate kinetics (termed I_M) that involves structural rearrangements in the outer pore.

3. Site-directed mutagenesis and ion-replacement experiments were performed using voltage-clamped Xenopus oocytes and cultured (HEK-293) cells expressing wild-type and mutant rat skeletal muscle (μ1) sodium channels.

4. Our results show that lowering [Na⁺]_o potentiates use-dependent lidocaine block. The effect of [Na⁺]_o is maintained despite a III-IV linker mutation that partially disrupts fast inactivation (F1304Q). In contrast, the effect of lowering [Na⁺]_o on lidocaine block is reduced by a P-loop mutation (W402A) that limits occupancy of I_M.

5. Our findings are consistent with a simple allosteric model where lidocaine binding induces channels to occupy a native slow inactivated state that is inhibited by [Na⁺]_o.

Lidocaine (lignocaine) and its analogues, compounds that suppress Na⁺ current (I_Na), are used as local anaesthetics and as therapies for cardiac arrhythmias and epilepsy. When depolarized, Na⁺ channels only transiently conduct ions due to rapidly developing (fast) inactivation (Hodgkin & Huxley, 1952), a conformational change involving amino acid residues in the III-IV interdomain linker (Stühmer et al. 1989; West et al. 1992) and vicinal sites near the inner surface of the channel (McPhee et al. 1994, 1995, 1998; Smith & Goldin, 1997; Lerche et al. 1997; Filatov et al. 1998; Tang et al. 1998). While drug-free Na⁺ channels quickly recover from fast inactivation between depolarizing stimuli, lidocaine delays such recovery, inducing a stable non-conducting state that results in cumulative reduction in I_Na with brief, repetitive stimuli (so-called ‘use-dependent block’) (Courtney, 1975). Lidocaine binds to amino acids positioned on the intracellular side of the putative selectivity filter (Ragsdale et al. 1994, 1996; Sunami et al. 1997), near (and perhaps overlapping) the fast-inactivation gating structures (Bennett et al. 1995).

A high-affinity binding interaction between lidocaine and the fast inactivated conformational state is a structural hypothesis for use-dependent local anaesthetic block supported by compelling, but circumstantial, experimental evidence (Hille, 1977; Hondeghem & Katzung, 1977; Cahalan, 1978; Yeh, 1978; Bean et al. 1983; Balser et al. 1996b). Surprisingly, a recent study found that amino acid residues intimately involved in fast inactivation recover to their pre-inactivated conformation rapidly, even while conduction is fully suppressed by lidocaine (Vedantham & Cannon, 1999), casting doubt on the prevailing notion that lidocaine arrests Na⁺ channels in the fast inactivated conformational state. In rare circumstances such as prolonged depolarization, Na⁺ channels may assume slow inactivated conformations from which recovery is considerably delayed (Adelman & Palti, 1969; Chandler & Meves, 1970; Rudy, 1978). One such slow inactivated state with ‘intermediate’ kinetics (Cummins & Sigworth, 1996; Hayward et al. 1996) is influenced by externally positioned residues in the outer pore region (μ1-W402) (Balser et al. 1996a; Kambouris et al. 1998; Benitah et al. 1999).

Earlier studies (Cahalan & Almers, 1979) suggested that external Na⁺ opposed the use-dependent action of charged (quaternary) lidocaine derivatives through a competitive ‘knock-off’ mechanism. However, recent studies have shown that external Na⁺ also prevents Na⁺ channels from entering slow inactivated states (Townsend & Horn, 1997), including an inactivated state of ‘intermediate’ stability (denoted I_M) (Benitah et al. 1999). Moreover, earlier studies proposed that local anaesthetic action involves a gating conformational change linked to a stable inactivated state (Khodorov et al. 1976; Zilberter et al. 1991), and we recently showed that an outer pore (P-loop) mutation that opposes I_M also reduces use-dependent lidocaine action (Kambouris et al. 1998). Here we examine the possibility that this gating effect of external Na⁺, inhibition of I_M, plays a role in Na⁺ antagonism of use-dependent lidocaine action. Our data show that augmenting slow inactivation by lowering the extracellular Na⁺ concentration ([Na⁺]_o) sensitizes the Na⁺ channel to use-dependent lidocaine action, even when fast inactivation is disabled (μ1-F1304Q). Moreover, the effects of lowering [Na⁺]_o on slow inactivation and lidocaine action are decreased by the P-loop mutation that opposes I_M (W402A). Our findings are consistent with a simple allosteric model where lidocaine binding induces channels to occupy a native slow inactivated state that is inhibited by [Na⁺]_o.

METHODS

Molecular biology

Site-directed mutagenesis of the μ1 Na⁺ channel α subunit (Trimmer et al. 1989) was performed on a full-length cDNA clone encoding the wild-type μ1 channel using a PCR-based method (Weiner et al. 1993). Both sense and antisense oligonucleotides (30-34mer) containing the desired mutation(s) were used as primers for amplification with Pfu DNA polymerase (Stratagene). The PCR product was treated with DpnI to select for the mutant plasmid and then transformed into E. coli for subsequent isolation and purification using standard protocols. Both modifications (W402A, F1403Q) were confirmed by di-deoxy sequencing of the region of the α subunit containing the mutation and by electrophysiological characterization of, at least, two independent mutagenic clones (Lawrence et al. 1996; Kambouris et al. 1998). In vitro mRNA transcription was performed using the SP6 promoter region and commercially available reagents (Gibco, Gaithersburg, MD, USA).

Ion channel expression

All procedures surrounding the care and use of Xenopus laevis conformed with our institution animal care committee guidelines. Oocytes were harvested from human chorionic gonadotropin-primed adult females (Nasco, Fort Atkinson, WI, USA) after 30 min of submersion anaesthesia in an aqueous solution containing 0.17 % 3-aminobenzoic acid ethyl ester (methanesulphonate salt, Sigma). Following survival surgery, the animals were allowed to recover in shallow water to maintain head out position until awake (∼30 min). Oocytes were isolated by digestion in a 2 mg ml⁻¹ solution of collagenase (Type IA; Sigma) and stored in a modified Barth's solution containing (mM): NaCl, 88; KCl, 1; NaHCO₃, 2.4; Tris base, 15; Ca(NO₃)₂.4H₂O, 0.3; CaCl₂.6H₂O, 0.41; MgSO₄.7H₂O, 0.82; sodium pyruvate, 5; theophylline, 0.5; supplemented with penicillin, 100 U ml⁻¹; streptomycin, 100 μg ml⁻¹; fungizone, 250 ng ml⁻¹ and gentamicin, 50 μg ml⁻¹. Oocytes were injected with 50 nl of a 0.1–0.25 μg μl⁻¹ solution of cRNA transcribed from wild-type or mutant cDNA. In all oocyte experiments, an equimolar ratio of similarly transcribed rat brain β₁ subunit RNA was co-injected with the α subunit cRNA.

For expression in human embryonic kidney (HEK-293) cells, the wild-type μ1 α subunit was subcloned into the vector HindIII- XbaI site of the vector GFPIRS for bicistronic expression of the channel protein and GFP reporter as previously described (Johns et al. 1997). Cells were transfected with this plasmid using lipofectamine (Gibco), and were cultured in modified Eagle's medium (MEM; Gibco) supplemented with 10 % fetal bovine serum and 1 % penicillin-streptomycin in a 5 % CO₂ incubator at 37°C for 1–3 days. The HEK cells were not transfected with the Na⁺ channel β₁ subunit. Cells exhibiting green fluorescence were chosen for electrophysiological analysis.

Electrophysiology and data analysis

Whole-cell Na⁺ current (I_Na) was recorded from Xenopus oocytes using a two-electrode voltage clamp, OC-725B (Warner Instrument Corp., Hamden, CT, USA), 1–3 days after cRNA injection. Currents were measured using electrodes of 1–3 MΩ resistance when filled with 3 M KCl. Currents were recorded in frog Ringer (ND-96) solution containing (mM): NaCl, 96; KCl, 2; MgCl₂, 1; Hepes, 5; pH 7.6, and [Na⁺]_o was reduced to 25 or 10 mM by equimolar replacement with N-methyl-D-glucamine-chloride (NMGCl). I_Na was recorded from HEK-293 cells using an Axopatch 200B amplifier (Axon Instruments, Foster City, CA, USA). Electrode resistances were 1–3 MΩ when filled with a pipette solution containing (mM): NaF, 140; NaCl, 10; EGTA, 5; Hepes, 10 (pH 7.40). Series resistance was ∼85 % compensated. The 150 mM Na⁺ bath solution contained (mmol l⁻¹): NaCl, 150; KCl, 4.5; CaCl₂, 1.5; MgCl₂, 1; Hepes, 10 (titrated to pH 7.40 with NaOH). [Na⁺]_o was reduced to 10 mM by equimolar replacement with NMGCl. In both oocyte and HEK cell experiments, junction potentials were minimized during bath solution changes by using a 3 M KCl-agar bridge. Inactivation gating kinetics and use-dependent block were assessed using voltage-clamp protocols described in the text and figure legends. In all cases, lidocaine HCl (2 % preservative free; Abbott Laboratories, North Chicago, IL, USA) was added to the bathing solution at concentrations indicated in the text. In experiments where lidocaine was used, a recovery period at −100 mV of at least 20 s was allowed prior to each set of measurements in order to avoid cumulative use-dependent block. All electrophysiological recordings were made at room temperature (20–22°C).

Multi-exponential functions were fitted to the data using non-linear least-squares methods (Microcal Origin version 4.10, Microcal Software Inc., Northampton, MA, USA) to determine the amplitudes and time constants associated with development or recovery from slow inactivation or use-dependent block. Pooled data are expressed as the means ± s.e.m., and statistical comparisons were made using one-way ANOVA (Origin) with P ≤ 0.05 indicating significance. A Markov gating scheme (Fig. 5) was evaluated using numerical integration methods tailored for stiff sets of differential equations (Hindmarsh, 1983; Petzold, 1983) and linked to custom-written software (Balser et al. 1990). Model parameters were minimized by fitting data from four conditions simultaneously (high/low [Na⁺]_o, ± lidocaine) using a simplex-based global fitting method described previously (Tomaselli et al. 1995) and as discussed in the legend of Fig. 5.

Figure 5. A model for lidocaine block that invokes [Na⁺]_o-dependent slow inactivation.

A, in this two-affinity modulated receptor model, I_M represents the slow inactivated state for which lidocaine has high affinity, while NI_M represents all other states (closed, open, fast inactivated) for which lidocaine has lower affinity. The forward rate constant controlling slow inactivation in drug-free conditions (α) was allowed to differ in 96 mM vs. 10 mM [Na⁺]_o, and the forward rate constant for slow inactivation with lidocaine bound (α′) was determined by α and the lidocaine off-rates (k_2off, k_1off) from detailed balances (eqn (1)). The lidocaine k_on was fixed at 10⁸ m⁻¹ s⁻¹ (see text). The remaining rate constants (β, k_2off, k_1off) were derived from fitting the model to the data, but were constrained to be the same in high or low [Na⁺]_o. In the model, both lidocaine-bound states (Lido-NI_M, I_M-Lido) were considered to be non-conducting, so that NI_M is the only state where channels are available to open and conduct Na⁺. B, the data shown were replotted from Fig. 1B and C. The ordinate (P₂/P₁) represents the probability that channels are not slow inactivated, and are therefore available to open. Using a numerical integration method (Methods), the model was evaluated at each time point (abscissa, Fig. 5B) and the calculated probability of occupying the NI_M state was compared directly to the P₂/P₁ data value. At t = 0, the probability of occupying the NI_M state was set to 1.0 (P₂/P₁ = 1.0). The model was fitted to the data in four conditions (high/low [Na⁺]_o, ± lidocaine) simultaneously using a simplex-based global fitting approach (Nelder & Mead, 1965) that updates the parameters to minimize the sum-of-squares error. Continuous lines show the non-linear least-squares fit of the model to the mean 10 mM and 96 mM [Na⁺]_o data. Fitted parameters for the rate constants controlling slow inactivation were: α in 96 mM [Na⁺]_o = 0.063 s⁻¹, α in 10 mM [Na⁺]_o = 0.92 s⁻¹, and β = 1.41 s⁻¹. Fitted values for lidocaine k_1off and k_2off were 127 × 10³ s⁻¹ and 386 s⁻¹, giving K_d values for the NI_M and I_M states of 1.27 mM and 3.86 μM, respectively.

RESULTS

Effects of lowering [Na⁺]_o on use-dependent lidocaine block

During brief depolarizations, μ1 Na⁺ channels recover from inactivation rapidly in physiological (96 mM) [Na⁺]_o when the β₁ subunit is coexpressed (Cannon et al. 1993; Chang et al. 1996). Figure 1A shows pairs of Na⁺ currents elicited at the beginning of a 1 s depolarization (P1), and a second current (superimposed) recorded at the same membrane potential (P2) after a brief (20 ms) recovery interval at −100 mV. Under these conditions, fast inactivated channels recover in < 10 ms; hence, any reduction in I_Na during the P2 pulse reflects slow inactivation (I_M). In 96 mM [Na⁺]_o, (Fig. 1A, left top) peak I_Na during the P2 pulse is only slightly decreased relative to P1 (92 ± 2 %, n = 7), confirming that little I_M occurs normally. However, lowering [Na⁺]_o to 10 mM (Fig. 1A, right top) increased entry into I_M (P2: 65 ± 6 % of P1, n = 8). While reducing [Na⁺]_o decreased the magnitude of I_Na, the current was within an acceptable range for voltage control in either [Na⁺]_o. In Fig. 1B, the duration of the P1 pulse was varied from 10 ms to 1 s to examine the rate at which I_M develops. Summary data (shown for 10, 24, and 96 mM [Na⁺]_o) are well-fitted by a single exponential function (continuous lines, fit parameters in Table 1), suggesting occupancy of a single slow inactivated state. The development of a single slow inactivated state with intermediate kinetics under these conditions (with β₁ subunit coexpression) is consistent with previous results in oocytes (Featherstone et al. 1996; Kambouris et al. 1998). However, depolarizations of similar duration in mammalian expression systems (Hayward et al. 1997) or longer depolarizations in oocytes (minutes; Todt et al. 1999) recruit additional slow kinetic components with longer time constants for recovery. Previous studies suggest that these more stable inactivated states do not influence lidocaine block (Kambouris et al. 1998).

Figure 1. The effects of [Na⁺]_o on slow inactivation and use-dependent lidocaine block.

A, whole-cell currents recorded from Xenopus oocytes in the absence (top) and presence (bottom) of 50 μM lidocaine. Oocytes were bathed in solutions containing either 96 mM (left) or 10 mM [Na⁺]_o (right). The superimposed current pairs were recorded at the beginning of a pulse of 1000 ms duration to −40 mV (holding potential −100 mV, bar indicates zero current level) and then during a second pulse to −40 mV which followed a 20 ms of recovery period at −100 mV. B, symbols indicate peak I_Na (means ± s.e.m.) during the P2 pulse relative to the P1 pulse with varying P1 pulse durations (protocol inset). Oocytes were bathed in either 96 mM [Na⁺]_o (n = 7), 24 mM [Na⁺]_o (n = 13), or 10 mM [Na⁺]_o (n = 8). Continuous lines are single exponential fits (y = y₀ + Aexp(-t/τ)) to the mean data; kinetic parameters derived from fitting data measured in individual oocytes are summarized in Table 1. C, the symbols show data obtained in an identical manner in solutions containing 50 μM lidocaine and either 96 mM [Na⁺]_o (n = 10), 24 mM [Na⁺]_o (n = 7), or 10 mM [Na⁺]_o (n = 8). Continuous lines show single exponential fits to the mean data (parameters summarized in Table 1). Dotted lines show the results in drug-free conditions (from panel B) for comparison. D, voltage-dependent availability is plotted for oocytes bathed in 10 mM [Na⁺]_o (n = 9) or 96 mM [Na⁺]_o (n = 5). The continuous lines are Boltzman equations (1 + exp(V–V_½/δ))⁻¹ fitted to the mean data. Summary values for the mid-point (V_½) and slope factor (δ) were, respectively, −62.0 ± 1.5 and 3.4 ± 0.2 mV in 10 mM [Na⁺]_o, and −55.4 ± 0.7 and 3.5 ± 0.2 mV in 96 mM [Na⁺]_o. V_½ values in 96 and 10 mM [Na⁺]_o differed significantly (P = 0.01). E, slow time-dependent recovery of availability following a 50 ms depolarization to −20 mV is plotted over an 8–5000 ms range. Recovery components more rapid than 8 ms (fast inactivation, deactivation) are not considered. Data are from oocytes bathed in 10 mM [Na⁺]_o (n = 10), 96 mM [Na⁺]_o + 100 μM lidocaine (n = 12), 10 mM [Na⁺]_o + 50 μM lidocaine (n = 7) and 10 mM [Na⁺]_o + 100 μM lidocaine (n = 6). Continuous lines are single exponential fits (y = y₀ + A(1 – exp(-t/τ))) to the mean data; kinetic parameters are given in Table 2.

Table 1.

Kinetic parameters for development of I_M

	96 mm[Na+]o		24 mm[Na+]o		10 mm[Na+]o
Lidocaine	A	τ (ms)	A	τ (ms)	A	τ (ms)
0	0.10 ± 0.03	462 ± 119	0.15 ± 0.03	264 ± 44	0.38 ± 0.05*	452 ± 31
50 μm	0.39 ± 0.06	389 ± 39	0.68 ± 0.03*	201 ± 34*	0.83 ± 0.02*	86 ± 9*

The function y = y₀ + Aexp(-t/τ) was fitted to the data in Fig. 1, panels B and C, and the values (means ± s.e.m.) for A and τ are shown.

^*Value differed significantly from data measured under the same conditions in 96 mm [Na⁺]_o.

Figure 1A (lower, left) shows that in 96 mM [Na⁺]_o, 50 μM lidocaine caused a reduction in P2 peak I_Na (65 ± 6 % of P1, n = 7) that was similar in magnitude to lowering [Na⁺]_o to 10 mM in drug-free solutions. Lowering [Na⁺]_o to 10 mM while in lidocaine (Fig. 1A, bottom, right) dramatically augmented use-dependent block (P2: 8.5 ± 2 %, n = 10). Notably, the lidocaine-associated block was entirely dependent on the preceding depolarization (i.e. was use dependent). Even 100 μM lidocaine did not block the current elicited by the first pulse (P1) after a long hyperpolarization in either 10 mM [Na⁺]_o (peak I_Na 99 ± 11 % of control, n = 12) or 96 mM [Na⁺]_o (99 ± 9 %, n = 10). Figure 1C shows the time course for development of use-dependent lidocaine block as the length of the P1 pulse was extended from 10 ms to 1 s (note ordinate difference from Fig. 1B). The results in drug-free conditions (from Fig. 1B) are shown as dotted lines. As before, the time-dependent loss of availability was well fitted by a single exponential (continuous line). Both the rate and extent of use-dependent block were augmented by lowering [Na⁺]_o from 96 to 10 mM. The time constant for block development decreased from 389 ± 39 ms in 96 mM [Na⁺]_o to 86 ± 9 ms in 10 mM [Na⁺]_o (P < 10⁻⁶, Table 1).

If inducing slow inactivation plays a major role in use-dependent lidocaine ‘block’, then recovery from use-dependent block should recapitulate recovery from I_M. Moreover, increasing the lidocaine concentration should increase the extent of use-dependent block, but should not alter the rate at which channels recover availability. Figure 1E shows the fraction of channels available to open (P2/P1) as the recovery interval at −100 mV is lengthened following a brief depolarization to −20 mV. A single exponential was fitted to fractional recovery data spanning an 8 to 5000 ms time range (continuous lines). In lidocaine-free solutions, the time constant for recovery from I_M was 335 ± 87 ms (10 mM [Na⁺]_o, Table 2), consistent with earlier results linking varying [Na⁺]_o (Benitah et al. 1999) and outer pore mutations (Kambouris et al. 1998) to I_M. In 50 and 100 μM lidocaine, the total number of channels exhibiting slow recovery progressively increased (i.e. the y intercept decreased), but the time constant for slow recovery was unchanged (see Table 2). Hence, the rate of recovery from lidocaine block was concentration independent and was kinetically indistinguishable from recovery from I_M.

Table 2.

Kinetic parameters for recovery from I_m

Lidocaine	[Na+]o	A	τ (ms)
0	10 mm	0.09 ± 0.01	335 ± 87
50 μm	10 mm	0.38 ± 0.06*	360 ± 74
100 μm	10 mm	0.47 ± 0.04*	368 ± 58
100 μm	96 mm	0.29 ± 0.02*	252 ± 12**

The function y = y₀ + A(1 − exp(−t/τ)) was fitted to the data in Fig. 1, panel E. Fitted parameters (means ± s.e.m.) are shown. Lidocaine increased the fraction of slowly recovering channels

^*(P < 0·0001 versus 0 μm lidocaine, 10 mm [Na⁺]_o). In 10 mm [Na⁺]_o, lidocaine did not change the time constant for slow recovery (τ), although in 100 μm lidocaine raising [Na⁺]_o from 10 to 96 mm significantly decreased τ

^**(P = 0·02).

Lidocaine induces a hyperpolarizing shift in the voltage dependence of I_Na availability (Bean et al. 1983), implying lidocaine-bound channels occupy a stabilized inactivated state. We therefore examined how [Na⁺]_o influences voltage-dependent availability of I_Na. The time constant for development of I_M was consistently ≤ 500 ms (Table 1); hence, 5 s prepulses (Fig. 1D, inset) were used to achieve steady-state occupancy of I_M while minimizing recruitment of even more stable (ultra-slow) inactivated states seen with longer (∼1 min) depolarizations (Kambouris et al. 1998; Todt et al. 1999). Figure 1D shows that lowering [Na⁺]_o induced a 7 mV hyperpolarizing shift in steady-state availability, consistent with the notion that external Na_o⁺ destabilizes I_M (Benitah et al. 1999). In this case, if use-dependent lidocaine block requires stable occupancy of I_M, then raising [Na⁺]_o should speed recovery from lidocaine block. Figure 1E shows that raising [Na⁺]_o to 96 mM not only reduced the extent of slow inactivation (filled vs. open circles in 100 μM lidocaine), but also speeded the time constant for recovery from block (Table 2: 252 ± 12 ms vs. 368 ± 58 ms, P = 0.02).

Effect of lowering [Na⁺]_o on lidocaine block in a fast-inactivation mutant

The kinetic overlap between fast inactivation and entry into I_M in low [Na⁺]_o makes it difficult to rule out the possibility that the inhibitory effect of [Na⁺]_o on lidocaine block is due to an indirect effect on fast inactivation. We therefore examined the interaction between [Na⁺]_o and lidocaine block in mutant μ1 channels with impaired fast inactivation (F1304Q) (West et al. 1992; Lawrence et al. 1996). Figure 2A shows slowly decaying F1304Q currents during a 200 ms depolarization in 96 mM (left) or 10 mM [Na⁺]_o (right) before and during exposure to lidocaine (200 μM). In drug-free conditions, the slow decay of F1304Q current has been linked to slow inactivation (Townsend & Horn, 1997). The rate of this slow decay is increased either by lowering [Na⁺]_o (Fig. 2A, see also Townsend & Horn, 1997) or by lidocaine (Fig. 2A and Balser et al. 1996b). Moreover, lowering the [Na⁺]_o to 10 mM potentiated the effect of lidocaine, markedly speeding F1304Q current decay (Fig. 2A, right). In Fig. 2B these [Na⁺]_o-dependent differences in lidocaine action are summarized by plotting the fraction of peak current remaining 200 ms into the depolarizing pulse (f₂₀₀). Lowering [Na⁺]_o from 96 to 10 mM hastened the current decay of F1304Q produced by either 50 or 200 μM lidocaine (P < 0.001). Hence, in conditions where I_M is facilitated (by low [Na⁺]_o) and fast inactivation is destabilized, low concentrations of lidocaine rapidly induce the formation of a stable non-conducting state when the membrane is depolarized.

Figure 2. Interaction between [Na⁺]_o and lidocaine block in F1304Q channels.

A, whole-oocyte currents elicited at −20 mV (holding potential −100 mV, bar indicates zero current level) before and during lidocaine exposure (200 μM) in either 96 mM [Na⁺]_o (left) or 10 mM [Na⁺]_o (right). Lidocaine enhanced the decay of whole-cell current to a greater extent when [Na⁺]_o was reduced. B, summary data plotting the fractional reduction in current by 200 ms (f₂₀₀) at −20 mV in 96 mM [Na⁺]_o (control n = 8, 50 μM lidocaine n = 8, 200 μM lidocaine n = 8) or 10 mM [Na⁺]_o (control n = 7, 50 μM lidocaine n = 4, 200 μM lidocaine n = 5). At each lidocaine concentration (0, 50, 200 μM), current decay was hastened by lowering [Na⁺]_o from 96 to 10 mM (*P < 0.001).

Lidocaine block in a mutant channel with impaired slow inactivation

Previous work has shown that mutation of a P-loop tryptophan (μ1-W402) reduces an intermediate kinetic component of slow inactivation (I_M) (Balser et al. 1996a). Alanine replacement (W402A) primarily attenuates the I_M state (in preference to more long-lived slow inactivated states), and also reduces use-dependent lidocaine block (Kambouris et al. 1998). However, this correlation between reduction of I_M and reduced lidocaine block could be fortuitous: the mutation could modify slow inactivation gating and the lidocaine receptor (directly or indirectly) through independent mechanisms. Nonetheless, if augmentation of lidocaine block by low [Na⁺]_o was attenuated by the W402A mutation, the hypothesis of a mechanistic linkage between I_M occupancy and lidocaine block would be strengthened. Figure 3 shows the time course for development of slow inactivation (open symbols) and use-dependent lidocaine block (filled symbols) in W402A channels bathed in 10 mM [Na⁺]_o. The continuous lines are single exponential fits to these data. The comparable exponential fits to the wild-type data (from Fig. 1B and C) are shown by the dashed lines. In drug-free conditions, the I_M state developed more slowly (τ = 1870 ms) and less fully (A = 0.19) in the W402A channel relative to the wild-type channel (Table 1, comparable parameters for wild-type were: τ = 452 ms, A = 0.38). In 50 μM lidocaine, the rate (τ = 274 ms) and extent (A = 0.38) of use-dependent block was similarly reduced compared to wild type (Table 1, τ = 86 ms, A = 0.83). Hence, the W402A mutation reduced the slow inactivation normally induced by lowering [Na⁺]_o and at the same time inhibited the development of use-dependent lidocaine block in low [Na⁺]_o. These findings suggest that external Na⁺ and the P-loop mutation both modulate use-dependent lidocaine action through effects on the I_M state.

Figure 3. Lidocaine block of a mutant channel (W402A) with impaired slow inactivation.

The time course for development of slow inactivation (▵, n = 6) and use-dependent lidocaine block (▴, n = 6) is shown for W402A channels bathed in 10 mM [Na⁺]_o. The continuous lines are single exponential fits to these data (similar to Fig. 1, parameters given in text). Comparable results in wild-type channels at the same [Na⁺]_o (from Fig. 1B and C) are shown by the dashed lines. Compared to wild type, the W402A mutation markedly reduced the extent of slow inactivation in 10 mM [Na⁺]_o, and also significantly reduced the development of use-dependent lidocaine (lido) block in 10 mM [Na⁺]_o.

Electrochemical gradient and use-dependent block

These findings point toward a mechanistic link between occupancy of the I_M state and use-dependent lidocaine action. However, all the currents measured in the oocyte experiments were inward (Fig. 1A) and lidocaine block consistently increased in parallel with a reduced electrochemical gradient for Na⁺ (Fig. 1C). We therefore re-examined the effect of lowering [Na⁺]_o on lidocaine block in μ1 Na⁺ channels expressed in HEK-293 cells using the whole-cell voltage-clamp method (see Methods). Using this approach, we could raise [Na]_i to 150 mM, and thereby induce outward currents in low [Na⁺]_o at typical depolarized membrane potentials. Figure 4A shows I_Na measured before (left) and during (right) exposure to 10 μM lidocaine in either 10 mM [Na⁺]_o (top) or 150 mM [Na⁺]_o (bottom). In each panel, the superimposed Na⁺ currents were elicited by an 800 ms depolarizing pulse to −20 mV and a subsequent brief (50 ms) pulse to the same membrane potential, separated by a 20 ms interval at −100 mV to allow recovery from fast inactivation (protocol inset, Fig. 4A). Hence, the fractional decrement in peak I_Na between the superimposed currents (pulse 2/pulse 1) reflects slow inactivation or use-dependent lidocaine block. Although the macroscopic currents were somewhat smaller in cells bathed in low [Na⁺]_o due to gating effects, the absolute driving force for Na⁺ flux through the open channel at −20 mV was 28 mV greater in 10 mM [Na⁺]_o (+48 mV vs. −20 mV). The left panels (drug free) indicate that slow inactivation was still 15 % greater in low [Na⁺]_o despite the change in direction of current flow or the increased electrochemical gradient (summarized data Fig. 4B, left). Use-dependent current reduction in lidocaine was 23 % greater in low [Na⁺]_o (right panels, Fig. 4A; summary data Fig. 4B, right). Lidocaine block increased under conditions where slow inactivation was enhanced even though the currents in low [Na⁺]_o were outward and the electrochemical gradient driving I_Na was increased.

Figure 4. Na⁺ channel α subunits expressed in HEK-293 cells.

A, Na⁺ currents were recorded from HEK cells prior to (left) and during lidocaine exposure (right, 10 μM) in either 10 mM (top) or 150 mM [Na⁺]_o (bottom). The pipette [Na]_i was 150 mM, so that I_Na at −20 mV was outward in 10 mM [Na⁺]_o (V_clamp – E_Na = +48 mV), but inward in 150 mM [Na⁺]_o (V_clamp – E_Na = −20 mV). In each case, the first (larger) I_Na was elicited by an 800 ms depolarizing pulse to −20 mV, while the second (smaller) I_Na was recorded from a brief (50 ms) step to −20 mV that followed a 20 ms recovery interval at −100 mV (see inset voltage clamp protocol). B, the bar graph plots the fractional change in peak I_Na between the superimposed currents in panel A (pulse 2/pulse 1) for each of the four bath conditions. The number of cells recorded in each condition are shown in parentheses. *Significant difference from the matched condition in 150 mM [Na⁺]_o (P < 0.05).

DISCUSSION

In a previous paper (Kambouris et al. 1998) we reported that alanine mutation of a P-loop tryptophan (W402A), external to both the selectivity (DEKA) ring (Heinemann et al. 1992) and the putative local anaesthetic receptor in domain IV S6 (Ragsdale et al. 1994), unexpectedly reduced use-dependent lidocaine block. Since the W402A mutation also reduced I_M, the results suggested the possibility that this component of slow inactivation gating and use-dependent lidocaine action are mechanistically linked. However, although these two effects of the mutation (gating and pharmacological) shared similar kinetic features, we could not exclude the possibility that the mutation modified I_M and the local anaesthetic receptor through distinct mechanisms. Our rationale for the present study was to explore this potential linkage using a different, non-mutagenic approach. Our results show that lowering [Na⁺]_o both augments entry into the I_M state and potentiates use-dependent block by lidocaine (Fig. 1). With a view to bridging this result to our prior mutagenesis work (Kambouris et al. 1998), we show (Fig. 3) that W402A reduced the effect of lowering [Na⁺]_o on both slow inactivation gating and lidocaine block, suggesting that [Na⁺]_o and the P-loop mutation both modify lidocaine block through effects on the I_M state. In addition, we find that these effects of [Na⁺]_o on lidocaine action are not abolished by mutagenic disruption of fast inactivation (Fig. 2).

An allosteric model for lidocaine-induced use dependence

We considered whether the opposing effects of Na_o⁺ and lidocaine on use-dependent channel availability could be rationalized by a modulated receptor scheme (Fig. 5A). For simplicity, all gated conformational states other than the slow inactivated state (I_M) were grouped together as ‘available’ states for Na⁺ conduction (NI_M). Included among the NI_M states is the fast inactivated state, recognizing that (1) fast-inactivation gating structures recover rapidly despite use-dependent lidocaine block (Vedantham & Cannon, 1999) and (2) under the conditions of our protocol (Fig. 1) fast inactivated channels were allowed sufficient time to recover, and were therefore ‘available’ to open.

Discrete block of unitary current cannot be observed with lidocaine (Grant et al. 1989), so the true on-rate for lidocaine binding remains uncertain. Hence, we arbitrarily set the on-rate for lidocaine binding at a fixed upper limit approaching diffusion (10⁸ m⁻¹ s⁻¹), and allowed the unbinding rates to determine the stabilities of the two (NI_M and I_M) drug-bound complexes. We further assumed that channels in the I_M state have the higher affinity for lidocaine (i.e. a lower k_off).

We allowed the forward rate constant for slow inactivation (α) to differ in 10 and 96 mM mM Na_o⁺, but assumed (in the simplest case) the rate constant controlling recovery from slow inactivation (β) was not influenced by [Na⁺]_o or drug binding (Fig. 1E), and was therefore constrained to be identical for drug-free (top of Fig. 5A) and drug-bound (bottom of Fig. 5A) channels. Based on these definitions and detailed balances, the forward rate constant for slow inactivation with lidocaine bound (α′) is explicitly defined as a function of α and the off-rates for lidocaine binding (k_1off, k_2off) as follows:

(1)

Since we assume lidocaine binds with greater affinity to the I_M state (i.e. k_2off < k_1off), the rate of slow inactivation with lidocaine bound must exceed the rate under drug-free conditions (α′ > α). In this sense, the lidocaine action parallels that of allosteric effector molecules (Monod et al. 1965), where drug binding allosterically promotes an intrinsic channel conformation (i.e. slow inactivation). Figure 5B shows model fits (continuous lines) to the 10 mM and 96 mM [Na⁺]_o data from Fig. 1B and C. The fitted rate constants and binding affinities (see legend) were all constrained to be identical in 10 and 96 mM [Na⁺]_o, with the exception of α which decreased 15-fold as [Na⁺]_o was raised, consistent with slowed development of I_M. By speeding entry of drug-bound channels into the I_M state, the model simulates the marked increase in the rate at which lidocaine ‘block’ develops when [Na⁺]_o is reduced. The model converged on K_d values for the NI_M and I_M states (1.27 mM and 3.86 μM, respectively) that resemble previously determined experimental K_d values for tonic and ‘inactivated-state’ lidocaine block of skeletal muscle Na⁺ channels (Nuss et al. 1995).

Processes other than I_M may influence use-dependent Na⁺ channel block

Sodium channel μ1 α subunits expressed in oocytes exhibit an unusual degree of slow (mode 2) gating and slow inactivation when the β₁ subunit is omitted (Zhou et al. 1991). In the oocyte experiments (Figs 1-3), the β₁ subunit was coexpressed in order to minimize these phenomena, and thereby more closely recapitulate the in vivo gating phenotype. However, use-dependent lidocaine block of Na⁺ channels expressed in oocytes is attenuated by β₁ coexpression, a phenomenon that may result from β₁ effects on slow gating (Balser et al. 1996c) or from a direct β₁ interaction with the lidocaine receptor (Makielski et al. 1996). Thus, our intepretation of the lidocaine-[Na⁺]_o interaction in oocytes could be confounded if there are changes in α-β₁ coupling as [Na⁺]_o changes. However, gating of the μ1 Na⁺ channel α subunit in cultured mammalian cells (in contrast to oocytes) is almost entirely rapid, regardless of whether the β₁ subunit is coexpressed (Ukomadu et al. 1992). The experiments in Fig. 4 were performed without β₁ subunit coexpression and recapitulated the results in oocytes, suggesting that these findings are not contingent on the expression system or variable α-β₁ coupling.

Although our results cannot exclude the possibility that Na_o⁺ inhibits use-dependent lidocaine action through a competitive or otherwise gating-independent mechanism, several factors support a role for slow inactivation. The wild-type Na⁺ channel occupies the open state (and conducts Na⁺ ions) for only a few milliseconds (coincident with peak I_Na), while use-dependent lidocaine ‘block’ develops over hundreds of milliseconds (Table 1, Fig. 1C), during a time period when channels are inactivated and non-conducting. Moreover, lowering the [Na⁺]_o to 10 mM still produced no first-pulse (so-called ‘tonic’) block in 100 μM lidocaine (see Results, paragraph 2), a concentration that induces substantial use-dependent block under these conditions (Fig. 1C). At the same time, lowering [Na⁺]_o enhanced I_M entry and lidocaine-use dependence, while both effects were reduced by a mutant (W402A) that inhibits the same slow inactivation gating process (Fig. 3). Finally, use-dependent lidocaine block tracks the degree of slow inactivation in a manner independent of the electrochemical gradient for Na⁺ (Fig. 4).

Nonetheless, it remains possible that a competitive interaction between Na⁺ and lidocaine influences the results to some degree, especially if the two species interact at a site where Na⁺ occupancy correlates poorly with current magnitude. Further, while Na_o⁺ probably inhibits slow inactivation through occupying a site external to the selectivity filter (Townsend & Horn, 1997), it is still conceivable that Na_o⁺ and the W402A mutation modify slow inactivation and use-dependent lidocaine action via independent, albeit parallel mechanisms. A competitive interaction between Na⁺ and lidocaine in the permeation pathway could have influenced the F1304Q results (Fig. 2) given the much longer single-channel open time and high frequency of re-openings for this μ1 mutant (Lawrence et al. 1996). This could partly explain why lowering [Na⁺]_o seemed to potentiate lidocaine suppression of F1304Q I_Na to a greater extent than wild type.

It is well known that cardiac (hH1) I_Na is more sensitive to use-dependent lidocaine action than skeletal muscle I_Na (Nuss et al. 1995; Wang et al. 1996). In light of recent reports that find slow inactivation is more prominent in the skeletal muscle isoform (Richmond et al. 1998), a positive interaction between slow inactivation and lidocaine sensitivity may seem paradoxical. However, the component of slow inactivation that is amplified in skeletal muscle has a kinetic signature (τ = ∼5–20 s) far slower than the ‘intermediate’ I_M state discussed here (τ < 500 ms). Additionally, we have recently shown that hH1 channels recover from I_M states more slowly than μ1 channels, and this may partly account for the enhanced ‘sensitivity’ of the cardiac isoform to use-dependent lidocaine action (Nuss et al. 2000).

Other differences between ‘classic’ slow inactivation and the I_M state warrant discussion in this context. Most studies find that slow inactivation competes with fast inactivation (Rudy, 1978; Featherstone et al. 1996), whereas our prior studies suggest that fast inactivation and the I_M state are positively coupled. After relatively brief periods of depolarization (< 100 ms) sufficient to induce I_M but not ‘classic’ slow inactivation, μ1 channels carrying the III-IV linker IFM→QQQ mutation exhibited less slow inactivation than wild type (Balser et al. 1996a). In addition, following a 1 s depolarization, partly destabilizing fast inactivation (μ1F1304Q) hastened recovery from an intermediate (τ ∼500 ms) component of inactivation, but slowed recovery from a more stable (τ ∼5 s) inactivated state (Nuss et al. 1996). Taken together, these results support the notion that fast inactivation and I_M are positively coupled, while fast inactivation and ‘classic’ slow inactivation are inversely coupled.

The mechanistic role of fast inactivation in lidocaine action remains unresolved. Accessibility of the III-IV linker to covalent modification is reduced by inactivation, but accessibility recovers rapidly with hyperpolarization, and this recovery is not slowed by use-dependent lidocaine block (Vedantham & Cannon, 1999). While this result would suggest use-dependent lidocaine action does not involve fast-inactivation gate ‘trapping’, mutations that severely disrupt fast inactivation (i.e. III-IV linker IFM→ QQQ) markedly attenuate use-dependent block of peak I_Na (Bennett et al. 1995). At the same time, lidocaine markedly accelerates the decay of whole-cell current through F1304Q channels during sustained depolarization (Balser et al. 1996b), and this effect is potentiated by reducing [Na⁺]_o (Fig. 2). Whole-cell current decay through F1304Q channels in drug-free conditions is multiexponential, and prior studies suggest this represents entry into multiple inactivated states. Specifically, whole-cell (Nuss et al. 1996) and single-channel analyses (Lawrence et al. 1996; Townsend & Horn, 1997) suggest that the early, more rapid decay phase represents entry into a ‘destabilized’ fast inactivated state, while the slower decay that follows represents partitioning into more stable, slow inactivated states. Hence, prior studies have suggested that lidocaine may influence fast inactivation gating in III-IV linker mutants by ‘repairing’ the destabilized gating process to some degree (Balser et al. 1996b; Fan et al. 1996). Therefore, the salutary effect of lowering [Na⁺]_o on lidocaine action in Fig. 2 could be influenced by concurrent lidocaine effects on fast and slow inactivation gating.

We previously found that the W402A mutation reduced the rate of entry into the I_M state during depolarization (Kambouris et al. 1998). However, the mutation did not shift the steady-state availability curve to depolarized potentials (Kambouris et al. 1998), contrary to the expectation for an inactivation-destabilizing intervention (and in contrast to raising [Na⁺]_o in Fig. 1D). It is important to note that Fig. 1D assesses the equilibrium between inactivated states and closed states during modest (sub-threshold) depolarization, but does not consider inactivated-state stability at more depolarized membrane potentials sufficient to open the channel. Figure 1B and prior data (Kambouris et al. 1998) indicate that either raising [Na⁺]_o or the W402A mutation decrease the rate of entry into I_M at depolarized membrane potentials sufficiently to activate channels. W402A antagonizes the effects of reducing [Na⁺]_o on both I_M and use-dependent development of lidocaine block (Fig. 3), suggesting these two interventions share common mechanisms at depolarized membrane potentials. At hyperpolarized potentials, in Xenopus oocytes with the β₁ subunit coexpressed, the W402A mutation seemed to have little effect on the time constants for recovery from slow inactivation, consistent with the absent shift in steady-state availability (Kambouris et al. 1998). Conversely, raising [Na⁺]_o increases the rate of recovery from slow inactivation at hyperpolarized voltages (Townsend & Horn, 1997), and our data (Fig. 1D) show that Na_o⁺ shifts the steady-state availability curve at ‘subthreshold’ voltages in a positive direction. Both findings suggest I_M is ‘destabilized’ by Na_o⁺ at negative voltages, and Fig. 1E shows that the rate of recovery from use-dependent lidocaine ‘block’ at −100 mV is also hastened when [Na⁺]_o is raised.

The evidence that mutating residues accessible to the pore modifies slow inactivation (Balser et al. 1996a; Todt et al. 1999) suggests a relationship between the Na⁺ permeation pathway and the I_M gating mechanism. An analogous ‘slow’ inactivation process in K⁺ channels (so-called ‘C-type’ inactivation) involves conformational rearrangements in the outer pore, and is inhibited by raising the extracellular concentration of K⁺ (Lopez-Barneo et al. 1993; Liu et al. 1996). Moreover, use-dependent ‘block’ of Shaker channels by intracellular pore blockers involves induced C-type inactivation (Baukrowitz & Yellen, 1996). Hence, our results may suggest some general similarities between use-dependent drug action in Na⁺ channels and K⁺ channels. We caution, however, that these findings should not be generalized to suggest occupancy of I_M fully explains the use-dependent current suppression by all local anaesthetics. In contrast to lidocaine, which exhibits a preference for inactivated channels, high-affinity binding to the open channel may play a greater role in use-dependent block by disopyramide (Barber et al. 1992) as well as quinidine and flecainide (Ragsdale et al. 1996). Hence, future studies will determine the relative role of the I_M state in the action of other Na⁺ channel blocking compounds with diverse use-dependent kinetic properties.

Our findings support a model for use-dependent lidocaine block where drug binding induces the channel to enter an intrinsic inactivated state (I_M) with high lidocaine affinity. The available evidence suggests that lidocaine binds to a site in the aqueous pore, on the cytoplasmic side but very near to the putative selectivity filter (Ragsdale et al. 1994, 1996; Sunami et al. 1997). Site-directed mutations suggest that the I_M state also involves conformational changes in the pore, but slightly external to the selectivity filter (Balser et al. 1996a; Kambouris et al. 1998; Benitah et al. 1999). The emerging recognition that Na⁺ channel pore segments are exceptionally mobile (Benitah et al. 1997; Tsushima et al. 1997) could abridge these intramolecular distances.