INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Sodium channels are voltage-sensitive membrane proteins that produce the action potentials in excitable tissues such as nerve, skeletal muscle, and cardiac muscle. With sufficient depolarization from the resting potential, sodium channels activate, open, and allow sodium ion flux. Within a few milliseconds of opening, the channels inactivate and return to a nonconducting state (Hodgkin and Huxley, 1952). The general structure of the sodium channel isoforms from different excitable tissues appears to be conserved (Catterall, 1995; Fozzard and Hanck, 1996). Sodium channel  subunits consist of four homologous domains (D1-D4), and each domain contains six transmembrane segments (S1-S6). The arrangement of the four domain regions in the membrane forms a pore for conducting sodium ions.

Although the structures of the different tissue isoforms of sodium channel are generally similar, there are marked differences among the isoforms in kinetic behavior and pharmacology. Compared to skeletal muscle sodium channels, cardiac sodium channels activate and inactivate at more negative membrane potentials and have a slower time constant of macroscopic current decay when expressed in mammalian cells (Wang et al., 1996a; Wright et al., 1997). Pharmacological differences between cardiac and skeletal muscle sodium channels include a relatively greater sensitivity of the cardiac muscle isoform to Cd²⁺ ions and a lower sensitivity to tetrodotoxin (Frelin et al., 1986). While the residues responsible for Cd²⁺ and tetrodotoxin block have been delineated (Tomaselli et al., 1995), sodium channel block by local anesthetics is less clearly understood because channel affinity profoundly varies depending on channel state.

The two principal models used to explain the state-dependent modulation of receptor affinity are the Modulated Receptor hypothesis (Hille, 1977) and the Guarded Receptor hypothesis (Starmer et al., 1984). In Hille's (1977) modulated receptor scheme, the inactivation particle (h gate) accounted for the state-dependent alterations in receptor affinity. In contrast, Starmer et al. (1984) attributed state-dependent alterations in receptor affinity to the activation particle (m gate). A clear determination of which mechanism is responsible for receptor modulation might help explain why some local anesthetics, such as cocaine or lidocaine, strongly affect cardiac physiology at concentrations that have little obvious effect on skeletal muscle physiology.

The purpose of the present study was to examine the mechanisms that influence channel affinity and state-dependent cocaine block of skeletal muscle and cardiac sodium channels. Two previous studies suggested that the anesthetic receptor in cardiac sodium channels has a higher affinity for lidocaine than does the receptor in skeletal muscle sodium channels, and that this difference in receptor affinity accounts for the difference in lidocaine sensitivity between cardiac and skeletal muscle tissue (Nuss et al., 1995b; Wang et al., 1996b). In contrast, we have recently shown that mammalian cardiac (hH1) and skeletal muscle sodium channels (µ1) have very similar affinities for cocaine and for lidocaine (Wright et al., 1997). At intermediate voltages (120 mV to 90 mV), however, cocaine blocked hH1 channels with much greater potency, suggesting that the modulation of receptor affinity differs at the two isoforms. Although we lacked direct evidence to support our hypothesis, we suggested that cocaine blocked a larger proportion of hH1 channels than µ1 channels at intermediate voltages because of greater steady-state inactivation of the hH1 isoform (Wright et al., 1997). To test this hypothesis, we induced shifts in steady-state inactivation by creating µ1/hH1 channel chimeras and also by coexpressing the rat brain 1 subunit with the  subunits of hH1 or µ1. In stark contrast to data from other expression systems (Isom et al., 1992, 1995; Nuss et al., 1995a), we found that 1 subunit coexpression induced strong positive shifts in the steady-state inactivation of both hH1 and µ1 channels. When we plotted the midpoint voltages of steady-state cocaine block as a function of the midpoint voltages of either steady-state activation or inactivation, we found that the relationship between block and steady-state inactivation was linear, and that block was better correlated with steady-state inactivation than with activation. In simulations of block, steady-state cocaine block of each channel could be fairly well predicted by using the steady-state inactivation curve of each channel and the K_d values of resting and inactivated channels. The fit by the model was further improved by imposing an equilibrium shift in the steady-state inactivation curve. These data and the model suggested that the modulation of local anesthetic receptor affinity depends heavily on steady-state inactivation and perhaps to some extent on channel activation resulting from the coupling between inactivation and activation. Some of the presented data have appeared in an abstract (Wright et al., 1998).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Solutions and chemicals

The extracellular solution used to perfuse HEK cells contained (in mM): 65 NaCl, 85 choline chloride, 2 CaCl₂, and 10 HEPES (titrated with tetramethyl ammonium hydroxide to pH 7.4). The pipette solution contained (in mM) 100 NaF, 30 NaCl, 10 EGTA, and 10 HEPES (titrated with cesium hydroxide to pH 7.2). Cocaine hydrochloride was purchased from Mallinckrodt, Inc. (St. Louis, MO), and was stored at 20°C as a 200 mM solution in distilled water. Final anesthetic concentrations were obtained by serial dilution from a 10 mM stock solution prepared in external bathing solution.

Construction of µ1/hH1 channel chimeras

For µ1_(1-3)hH1₍₄₎, site-directed mutagenesis (Wang and Wang, 1997) was used to create a ClaI site at positions 3862-3867 of the µ1-cDNA1/AMP vector (the translation initiation site was at +1n) by converting C into T at position 3867n. The µ1-ClaI-3862 was further modified into µ1-ClaI-ClaI by introducing another ClaI site in the 3' polylinker. This was achieved by the digestion of µ1-ClaI-3862 with SalI, followed by a blunting reaction and ligation to the ClaI linker. hH1-ClaI was created by mutating the ATTGAC (4416-4421n) in the hH1 clone (Gellens et al., 1992) into a ClaI site: ATCGAT. The translation initiation site in the hH1 clone is at position +1n. The µ1_(1-3)hH1₍₄₎ channel was cloned by ligating the large µ1 ClaI fragment (domains 1-3) to the small hH1 ClaI fragment (domain 3 and 4 cytoplasmic linker and domain 4).

The chimera, µ1₍₁₎hH1_(2-4), was created by using the BsiWI site (at positions 1328-1333) of µ1-cDNA1/AMP. This restriction site in µ1 is located at the 3' end junction of the domain 1 S6 segment. Because hH1 lacks this BsiWI site, we performed site-directed mutagenesis at the 3' end junction of the domain 1 S6 segment of hH1 (at positions 1252-1257) to create the clone hH1-BsiWI (hH1 also has a BsiWI restriction site in the 3' polylinker). Three-way ligation was used to join DNA fragments: 1.7 Kb-HindIII-BsiWI from µ1-cDNA1/AMP, 4 Kb-BsiWI from hH1-BsiWI, and 5 Kb-HindIII-BsiWI from hH1-BsiWI. The orientations of µ1_(1-3)hH1₍₄₎, µ1₍₁₎hH1_(2-4), and of two other channel chimeras (hH1_(1-3)µ1₍₄₎ and hH1₍₁₎µ1_(2-4)) were determined by restriction mapping and sequencing. Introduction of the restriction sites for chimera formation did not produce point mutations in either the µ1 or the hH1 portion of channel chimeras.

Channel expression in HEK 293t cells

The methods used for maintaining transformed human embryonic kidney (HEK 293t) cells and for transiently expressing µ1 (Trimmer et al., 1989) and hH1 (Gellens et al., 1992) were described in a previous paper (Wright et al., 1997). For transient expression of the  subunits of cloned channels in HEK cells, we prepared the following DNA solution (Cannon and Strittmatter, 1993): 1 µg CD8 (cell surface antigen) and 2-10 µg sodium channel cDNA clone in the pcDNA1/amp vector (Invitrogen, San Diego, CA) were prepared in 250 mM CaCl₂ and added to a test tube containing 0.36 ml Hanks' balanced salt (2×) solution (in mM: 274 NaCl, 40 HEPES, 12 dextrose, 10 KCl, 1.4 Na₂HPO₄, pH 7.05). After a 20-min incubation at 22°C, the DNA solution was added to a cell culture (in a TI-25 flask) that was 30-50% confluent. After 15 h at 37°C, the transfected cells were replated onto 35-mm culture dishes (which also served as recording chambers) containing 2 ml fresh DMEM supplemented with taurine (1%), penicillin/streptomycin (1%), and heat-inactivated fetal bovine serum (10%). For coexpression of the rat brain 1 subunit (Isom et al., 1992) with µ1 or hH1  subunits, saturating levels (>10-fold molar excess) of 1 cDNA were used to ensure that the resulting macroscopic currents were produced by channels consisting of both  and 1 subunits.

Electrophysiology procedures and data analysis

Whole-cell voltage clamp (Hamill et al., 1981) of HEK cells was used to study macroscopic sodium currents at room temperature (23 ± 2°C). Electrode resistances ranged from 0.4 to 1.0 M. Command voltages were programmed by pCLAMP software and delivered by a List EPC7 voltage clamp. Data were sampled at 50 kHz and were filtered at 5 kHz. After gigaohm seal formation and establishment of whole-cell voltage clamp, the cells were always dialyzed for 25-30 min before acquiring data. Time-dependent shifts in the midpoint voltage of sodium channel availability during our experiments (~30-60 min after membrane rupture) would have been ~5-7 mV (Wang et al., 1996a). The holding potential for all experiments was 140 mV. Most of the capacitative current was canceled by the EPC7 circuitry. The remaining capacitative artifact and the leakage current were subtracted by the P/4 method. The P/4 method was not used for studies of use-dependent cocaine block. Voltage errors at +30 mV were 5 mV after 30-50% compensation. Least-squares curve fitting was performed with Microcal Origin software. Depending on the experiment, data were fitted by an empirical Boltzmann function {1/[1 + exp((V_0.5  V)/k)]}, where V_0.5 is the midpoint voltage of the function and k is the slope factor (in mV/e-fold change in current); by a single exponential function {y₀ + A1*[1exp(x/1)]}; or by the sum of two exponential functions {y₀ + A1*[1exp(x/1)] + A2*[1exp(x/2)]}. Data are presented as mean ± SE.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Role of domain regions in sodium channel activation and inactivation

Previous studies indicated that steady-state inactivation of µ1/hH1 channel chimeras differed from the steady-state inactivation of the wild-type isoforms (Makita et al., 1996; Chahine et al., 1996; Benzinger et al., 1997). Our goal was to induce voltage-dependent shifts in steady-state channel activation or inactivation in an attempt to determine how shifts in channel kinetics might affect steady-state cocaine block. Fig. 1 A shows current records obtained from the two wild-type sodium channels, µ1 and hH1, and from two channel chimeras. The µ1 panel shows the four domains (D1-D4) of the sodium channel  subunit, the charged (+) S4 segments in each domain, and the amino and carboxyl termini. The channel chimera, µ1_(1-3)hH1₍₄₎, contained domains 1-3 from the µ1  subunit and domain 4 from the hH1  subunit, whereas µ1₍₁₎hH1_(2-4) contained domain 1 from µ1 and domains 2-4 from hH1. Despite repeated attempts, two other channel chimeras, hH1₍₁₎µ1_(2-4) and hH1_(1-3)µ1₍₄₎, failed to express current. To evoke the currents shown in Fig. 1 A, we delivered 10-ms step depolarizations from a holding potential of 140 mV. The inward currents for µ1, µ1_(1-3)hH1₍₄₎, and µ1₍₁₎hH1_(2-4) channels peaked at 30 mV, whereas the peak inward current for hH1 channels occurred at 40 mV. The time dependence of macroscopic current decay for both µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels clearly resembled the decay of hH1 currents, which decayed more slowly than the µ1 current. At +30 mV, µ1 current decayed with a time constant of 0.28 ± 0.01 ms (n = 7). In contrast, hH1, µ1_(1-3)hH1₍₄₎, and µ1₍₁₎hH1_(2-4) currents decayed with time constants of 0.49 ± 0.03 ms (n = 6), 0.46 ± 0.02 ms (n = 11), and 0.41 ± 0.01 ms (n = 13), respectively.

View larger version (44K): [in this window] [in a new window]   FIGURE 1   Activation, steady-state inactivation, and cocaine block of µ1, hH1, and µ1/hH1 channel chimeras. (A) The sodium channel  subunit structures and associated currents recorded from the two wild-type sodium channel isoforms, µ1 and hH1, and from two µ1/hH1 channel chimeras, µ1(1-3)hH1(4) and µ1(1)hH1(2-4), are shown. In the µ1 panel, the domain regions, the charged (+) voltage sensing regions, and the amino and carboxyl termini are labeled. The channel chimera, µ1(1-3)hH1(4), consisted of domains 1-3 from µ1 and the interdomain linker between domains 3 and 4 and domain 4 from hH1, whereas µ1(1)hH1(2-4) consisted of domain 1 from µ1 and domains 2-4 from hH1. The currents were evoked by 10-ms pulses from 140 mV to voltages ranging from 100 mV to +50 mV. The peak inward current and its corresponding voltage are labeled. (B) Normalized membrane conductance (gm) plotted versus the amplitude of the 10-ms voltage step; gm was determined from the equation gm = INa/(Em  ENa), and the plot was fitted with a standard Boltzmann function. For clarity, only the fits to the µ1 and hH1 data are shown. For µ1, the average midpoint voltage (V0.5) and slope (k) of the function were 32.8 ± 2.8 mV and 9.3 ± 1.5 mV, respectively. V0.5 and k were: 48.0 ± 1.8 mV and 9.4 ± 0.9 mV, respectively, for hH1; 27.8 ± 1.2 mV and 9.8 ± 0.3 mV, respectively, for µ1(1-3)hH1(4); and 32.4 ± 0.6 mV and 9.7 ± 0.6 mV, respectively, for µ1(1)hH1(2-4). (C) Normalized sodium current availability function (h) for the four channels. The pulse protocol is shown in the inset. The dotted lines are the fits to the averaged µ1 or hH1 data, as labeled. The average V0.5 values (50% availability) and k values for the fitted Boltzmann functions were: 78.5 ± 1.0 mV and 6.1 ± 0.3 mV, respectively for µ1; 94.1 ± 1.4 mV and 7.7 ± 0.1 mV, respectively, for hH1; 82.2 ± 1.0 mV and 6.5 ± 0.2 mV, respectively, for µ1(1-3)hH1(4); and 90.7 ± 1.0 mV and 6.1 ± 0.1 mV, respectively, for µ1(1)hH1(2-4). (D) Steady-state cocaine block of the four channels. The pulse protocol is shown in the inset. The filled symbols are the averaged data for µ1(1-3)hH1(4) and µ1(1)hH1(2-4) measured in control saline. A Boltzmann function was used to fit the data obtained in the presence of 30 µm cocaine. The dotted lines are the fits to the averaged µ1 or hH1 data, and the open symbols and solid lines are the averaged data and fits to the channel chimera data. The average V0.5 and k values for the fitted functions were: 92.1 ± 0.8 mV and 5.0 ± 0.2 mV, respectively for µ1; 115.4 ± 1.2 mV and 6.1 ± 0.2 mV, respectively, for hH1; 96.7 ± 0.7 mV and 5.5 ± 0.2 mV, respectively, for µ1(1-3)hH1(4); and 106.1 ± 0.9 mV and 7.6 ± 0.2 mV, respectively, for µ1(1)hH1(2-4).

Parts B and C in Fig. 1 show the normalized membrane conductance and steady-state inactivation curves, respectively, for µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels. These data were fitted with an empirical Boltzmann function (solid lines) to determine the midpoint voltage (V_0.5) and the slope factor (k). For comparison, the fitted Boltzmann functions (dotted lines) for hH1 and µ1 channels (Wright et al., 1997) are also shown in parts B and C. The activation (i.e., the conductance-voltage relationship) of both µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels more closely resembled the activation of µ1 channels (Fig. 1 B). The V_0.5 value of channel activation for hH1 channels was significantly (p < 0.05; t-test) more negative than were the V_0.5 values of activation for µ1, µ1_(1-3)hH1₍₄₎, or µ1₍₁₎hH1_(2-4) channels, whereas the V_0.5 values of activation for µ1_(1-3)hH1₍₄₎ or µ1₍₁₎hH1_(2-4) were not significantly different (p > 0.05) from that of µ1 (Table 1).

We examined the steady-state inactivation properties of these four sodium channel isoforms by using a standard h pulse protocol. We delivered 100 ms conditioning pulses of various amplitudes and measured the available sodium current during a test pulse to +30 mV (Fig. 1 C, inset). The plot in Fig. 1 C shows the averaged data for the channel chimeras and the solid lines represent the fits by a Boltzmann function. The dotted lines are the fits to the averaged hH1 and µ1 data. The V_0.5 values of steady-state inactivation of µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels fell between the V_0.5 values of inactivation for hH1 and µ1, with µ1_(1-3)hH1₍₄₎ more closely resembling µ1 and µ1₍₁₎hH1_(2-4) more closely resembling hH1 (Table 1). Consistent with a previous report that used the oocyte expression system to examine inactivation (Makita et al., 1996), the data in Fig. 1 C suggest that all four domain regions contribute, perhaps equally, to the voltage dependence of steady-state inactivation.

Cocaine block of µ1, hH1, and chimeras

As described above, chimera formation had distinct effects on channel activation and steady-state inactivation. The V_0.5 values of activation for both channel chimeras resembled that of µ1, but the V_0.5 values of steady-state inactivation for the chimeras were intermediate to those of µ1 and hH1. We compared the steady-state cocaine block of the four channels to perhaps distinguish whether the inactivation phenotype or the activation phenotype has the more prominent role in determining steady-state cocaine block. Cocaine binds to the channels with a stoichiometry of 1:1, so we measured 30 µM cocaine block of the channels over a 110 mV range of conditioning voltages. By using this approach, we were able to examine block of resting channels at the most negative conditioning voltages, block of inactivated channels at the least negative voltages, and block of varying proportions of resting and inactivated channels at intermediate voltages. The pulse protocol (Fig. 1 D, inset) consisted of a 10-s conditioning pulse ranging from 180 mV to 70 mV followed by a 100-ms interval at the holding potential and a subsequent test pulse to +30 mV. As previously described (Wright et al., 1997), a conditioning pulse of 10 s was necessary and sufficient for cocaine block to reach steady state at all of the channels in the study (n = 2-3 cells; data not shown). A 100-ms interval inserted between the conditioning pulse and the test pulse allowed drug-free channels to recover from fast inactivation. To normalize the data at each conditioning voltage, we divided the peak current amplitude at the test pulse in the presence of cocaine by the peak current amplitude elicited by the test pulse in control saline. At strongly negative conditioning pulses, 30 µM cocaine blocked ~10% of the resting µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels, which was similar to the block of resting µ1 and hH1 channels. When the channel chimeras were inactivated by a 10-s conditioning pulse to 70 mV, 30 µM cocaine blocked ~75-80% of the channels, which was similar to the block of inactivated µ1 and hH1 channels. As with the wild-type channels, 300 µM cocaine blocked ~55% of the resting channels and ~97% of the inactivated channels (Wright et al., 1997). At 180 mV, 300 µM cocaine blocked 55.5 ± 1.7% (n = 5) of resting µ1_(1-3)hH1₍₄₎ channels and 57.6 ± 1.0% (n = 6) of resting µ1₍₁₎hH1_(2-4) channels. After a 10-s conditioning pulse to 70 mV, 300 µM cocaine blocked 96.4 ± 1.0% (n = 5) and 98.2 ± 2.8% (n = 6) of inactivated µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels, respectively. There was no significant difference (p > 0.05) in block between the two chimeras; nor was there a difference between the block of the chimeras and block of the wild-type channels. At intermediate voltages, where there was a mixture of resting and inactivated channels, the relative differences in 30 µM cocaine block among the four channels (Fig. 1 D) generally resembled the differences among the channels in steady-state inactivation (Fig. 1 C). Note that cocaine block of the µ1_(1-3)hH1₍₄₎ chimera, which has domain 4 from hH1 and thus the putative local anesthetic receptor for hH1 (Ragsdale et al., 1994), more closely resembled cocaine block of µ1 channels. The differences in steady-state inactivation therefore provided a reasonable correlation for the observed differences in steady-state cocaine block.

Effects of 1 subunit coexpression on channel kinetics and cocaine block

In other expression systems, coexpression of the rat brain 1 subunit with the  subunit of sodium channels most often shifted channel kinetics toward more negative voltages. In Xenopus oocytes, coexpression of the 1 subunit with the  subunits of rat brain IIA (Isom et al., 1992) or µ1 (Nuss et al., 1995a) sodium channels increased current amplitude, speeded the rate of current decay, and shifted the voltage dependence of steady-state inactivation to more negative voltages. In Chinese hamster cells, 1 subunit coexpression with the rat brain IIa  subunit caused negative shifts in both activation and inactivation, but did not obviously affect the rate of current decay (Isom et al., 1995). Furthermore, in the oocyte expression system 1 subunit coexpression with hH1 channels reduced the resting affinity for lidocaine by twofold (Makielski et al., 1996).

Our first objective was to determine what effect, if any, 1 subunit coexpression had on the channel kinetics of µ1 or hH1 when expressed in HEK cells. Although 1 coexpression with µ1 and hH1 in HEK cells increased current amplitude as in other expression systems, there were surprisingly different effects on channel activation and inactivation. The V_0.5 values of activation for µ1-1 and hH1-1 were 2 mV and 5 mV, respectively, more positive than the V_0.5 values of activation for the  subunits alone (Fig. 2 A). Compared to the V_0.5 values of activation for µ1 and hH1  subunits, the positive shift was modest (µ1 versus µ1-1, p > 0.05; hH1 versus hH1-1, p = 0.05). 1 subunit coexpression had virtually no effect on the time constant of macroscopic current decay during depolarizations to between 0 and +50 mV (Fig. 2 B). Coexpression of the 1 subunit with µ1 and hH1 channels induced strong positive shifts in the V_0.5 values of steady-state inactivation, as compared to expression of the  subunits alone (Fig. 2 C). For µ1-1 and hH1-1, the V_0.5 values of inactivation were, respectively, 10 mV (p < 0.05) and 13 mV (p < 0.05) more positive than the V_0.5 values of µ1 and hH1 (Table 1).

View larger version (32K): [in this window] [in a new window]   FIGURE 2   Effects of rat brain 1 subunit coexpression on µ1 and hH1 channel kinetics and cocaine block. (A) Normalized conductance of the  subunits of µ1 and hH1 channels with and without coexpression of the 1 subunit. The average V0.5 and k values of the fitted Boltzmann functions to the µ1-1 data were 31.1 ± 0.8 mV and 7.1 ± 0.2 mV, respectively, and these values for hH1-1 were 43.3 ± 1.0 mV and 9.2 ± 0.7 mV, respectively. (B) Time constants of macroscopic current decay plotted versus the amplitude of 10-ms depolarizations. (C) Normalized sodium current availability (h) function plotted versus the amplitude of the 100-ms conditioning pulse. Coexpression of the 1 subunit with µ1 or hH1 channels produced rightward shifts in the voltage dependence of channel inactivation. The average V0.5 and k values of the Boltzmann function fitted to the µ1-1 data were 69.0 ± 0.7 mV and 4.9 ± 0.2 mV, respectively, and these values for hH1-1 were 81.2 ± 1.7 mV and 6.8 ± 0.2 mV, respectively. (D) Steady-state cocaine block. The filled symbols are the averaged data for µ1-1 and hH1-1 measured in control saline. The open symbols and solid lines are the averaged data and Boltzmann fits to the µ1-1 and hH1-1 data, and the dotted lines are the fits to the averaged data from µ1 and hH1  subunits expressed alone (as in Fig. 1 D). The average V0.5 and k values of the µ1-1 data were 79.8 ± 0.7 mV and 5.1 ± 0.4 mV, respectively, and these values for hH1-1 were 100.4 ± 1.4 mV and 6.2 ± 0.3 mV, respectively.

Because  subunit coexpression with µ1 and hH1 elicited strong positive shifts in steady-state inactivation and modest shifts in channel activation, we examined whether the 1 subunit altered the relationship between the conditioning voltage and cocaine block. Consistent with the effect on steady-state inactivation, 1 subunit coexpression with µ1 and hH1 induced strong positive shifts in the V_0.5 values of cocaine block. (Fig. 2 D). We changed the voltage range over which we examined cocaine block by 20 mV in the positive direction. Because the change in pulse protocol elicited more slow inactivation from µ1-1 channels than from hH1-1 channels, we normalized the block at each conditioning voltage by dividing the amplitude of the test current evoked in 30 µM cocaine by the amplitude of the test current evoked in control saline. The V_0.5 values of steady-state cocaine block at µ1-1 and at hH1-1 were 12 mV and 15 mV, respectively, more positive than the V_0.5 values of cocaine block at µ1 and hH1. Compared to block of the  subunits of µ1 and hH1, coexpression of the 1 subunit did not significantly (p > 0.05) alter the percentages of resting or inactivated channels blocked by either 30 µM (Fig. 2 D) or 300 µM cocaine. At 160 mV, 300 µM cocaine blocked 54.5 ± 1.5% (n = 4) of resting µ1-1 channels and 54.9 ± 2.3% (n = 4) of resting hH1-1 channels. After a 10-s conditioning pulse to 60 mV, 300 µM cocaine blocked 98.0 ± 1.0% (n = 4) and 97.1 ± 0.3% (n = 4) of inactivated µ1-1 and hH1-1 channels, respectively. Thus, coexpression of the 1 subunit with µ1 or hH1 did not alter the cocaine affinities of either resting or inactivated channels. For the six sodium channels listed in Table 1, we plotted the V_0.5 values of cocaine block as a function of their respective V_0.5 values of activation and inactivation (Fig. 3). Regression analyses showed that the voltage dependence of steady-state cocaine block was better correlated with the voltage dependence of steady-state inactivation (R = 0.98) than with channel activation (R = 0.66).

View larger version (16K): [in this window] [in a new window]   FIGURE 3   Dependence of cocaine block on channel kinetics. The V0.5 values of 30 µM cocaine block of µ1, hH1, µ1(1)hH1(2-4), µ1(1-3)hH1(4), µ1-1, and hH1-1 channels were plotted versus the V0.5 values of steady-state inactivation and activation of the channels. The error bars for the x-axis correspond to the standard errors of the averaged V0.5 values of inactivation or activation. Linear regression analyses of the plots showed that the V0.5 values of cocaine block were better correlated with the V0.5 values of steady-state inactivation (R = 0.98) than with the V0.5 values of steady state activation (R = 0.66).

Comparison of block recovery and use-dependent block at µ1 and hH1 channels

While the steady-state interactions between cocaine and cardiac sodium channels appear to be important in cocaine-induced cardiotoxicity, other interactions between cocaine and cardiac sodium channels may augment the cardiotoxic effects of cocaine. The fact that cardiac excitability is rhythmic and that cardiac cells spend more time than other sodium channel isoforms in the inactivated state suggests that the recovery time course from inactivated channel block and/or use-dependent block of cardiac sodium channels by local anesthetics may influence the net effects of cocaine. We therefore examined the recovery time course from cocaine block or in the extent of use-dependent block.

To determine the unbinding rate of cocaine from inactivated µ1 and hH1 channels, we delivered a 10-s conditioning pulse to 70 mV and measured the time-dependent recovery of current at 140 mV (Fig. 4, inset). The data obtained in control saline (filled symbols) showed that, in addition to eliciting fast inactivation, the conditioning pulse produced a slight amount of slow inactivation of both µ1 and hH1 channels. For both µ1 and hH1, the recovery from inactivation was best fitted by the sum of two exponentials. In control saline, the fast time constant of recovery was 1.0 ± 0.1 ms (n = 5) for µ1 channels and was 6.3 ± 0.4 ms (n = 6) for hH1 channels. These recovery rates were similar to the recovery rates for µ1 (1.5 ± 0.1 ms; n = 5) and hH1 (4.3 ± 0.7 ms; n = 6) after delivery of a 10-ms conditioning pulse to +30 mV. In control saline (see Fig. 6, filled symbols), the fast phase of recovery from inactivation for µ1 and hH1 channels comprised 81.2 ± 1.3% and 87.9 ± 1.9%, respectively, of the fractional amplitudes of recovery. The slow time constants of recovery for µ1 and hH1 channels were 279.1 ± 40.5 ms and 724.9 ± 190.0 ms (p = 0.07), respectively.

View larger version (28K): [in this window] [in a new window]   FIGURE 4   Recovery from cocaine block of inactivated channels. From a holding potential of 140 mV, µ1 (squares) and hH1 (circles) channels were stepped to 70 mV for 10 s. After a variable recovery interval at 140 mV, the channels were stepped to +30 mV to measure the available current. The recovery time courses in control saline (filled symbols) and in 30 µM cocaine (open symbols) were best fitted by the sum of two exponentials. In control saline, µ1 channels recovered with fast and slow time constants of 1.0 ± 0.1 ms and 279.1 ± 40.5 ms, respectively, and the fractional amplitude of the fast time constant comprised 81.2 ± 1.3% of the recovery; hH1 channels had fast and slow time constants of 6.3 ± 0.4 ms and 724.9 ± 190.0 ms, respectively, and the fractional amplitude of the fast time constant was 87.9 ± 1.9% of the recovery. In 30 µM cocaine, µ1 channels recovered with fast and slow time constants of 2.0 ± 0.2 ms and 6.5 ± 0.4 s, respectively, and the fractional amplitude of the fast time constant was 26.3 ± 1.4% of the recovery; hH1 channels recovered with fast and slow time constants of 11.5 ± 1.5 ms and 6.8 ± 0.5 s, respectively, and the fractional amplitude of the fast time constant was 27.5 ± 1.4% of the recovery.

The recovery from 30 µM cocaine block also was best described as the sum of two exponentials. The fractional amplitudes of the fast time constants of recovery from cocaine block of µ1 and hH1 channels were 26.3 ± 1.4% and 27.5 ± 1.4% (p > 0.05), respectively. As with recovery from lidocaine block (Wright et al., 1997), we assumed that the fast time constants of recovery in the presence of 30 µM cocaine represented the recovery of drug-free channels and that the fractional amplitude of block at the conclusion of the fast phase of recovery represented block of inactivated channels. The fast time constants of recovery for µ1 and hH1 were 2.0 ± 0.2 ms and 11.5 ± 1.5 ms, respectively. This slowing of the fast time constant of recovery in the presence of cocaine also occurs in the presence of lidocaine and its quaternary derivatives (Yeh and Tanguy, 1985) and suggests that local anesthetics might affect the recovery from inactivation at "drug-free" channels by binding to them at various time points of the recovery phase. The time constant of recovery from cocaine block was similar for the two channels and required several seconds. The time constants of recovery from 30 µM cocaine block of µ1 and hH1 channels were 6.5 ± 0.4 s (n = 5) and 6.8 ± 0.5 s (n = 6; p > 0.5), respectively. These values were consistent with the time constant of cocaine unbinding from ventricular myocyte sodium channels (Crumb and Clarkson, 1990).

We compared use-dependent cocaine block of µ1 and hH1 channels using 1 and 2 Hz stimulation rates (Fig. 5) but found no difference between the channels in the extent of use-dependent block. Cells were held at 140 mV and received 25 ms pulses to +30 mV. Little use-dependent decrease in µ1 or hH1 sodium current occurred during 1- or 2-Hz stimulation in control saline, whereas repetitive pulses delivered in 30 µM cocaine produced potent use-dependent block. Use-dependent block of µ1 and hH1 channels by 30 µM cocaine was fitted by a single exponential function. The percentages of steady-state block at 1- and 2-Hz stimulation were 39.4 ± 1.8% and 50.1 ± 1.6%, respectively, for µ1 channels and were 41.5 ± 1.7% and 53.2 ± 1.3%, respectively, for hH1 channels. The difference in the percentage of steady-state block between µ1 and hH1 at either 1- or 2-Hz stimulation was not statistically significant (p > 0.05). Although increasing the duration of the depolarization would have increased the percentage of block at steady state, the relative similarities in use-dependent block of µ1 and hH1 channels would not likely have changed. When we delivered a single conditioning pulse to +30 mV for 300 ms followed by a 100-ms interval at the holding potential and a test pulse to +30 mV, cocaine blocked a similar percentage of µ1 channels (29.7 ± 2.0%; n = 7) and hH1 channels (24.6 ± 1.0%; n = 5; p > 0.05). Although there were no significant differences between µ1 and hH1 in the time course of recovery from block or in use-dependent block, the extremely slow nature of the recovery from block (Fig. 4) and the extent of use-dependent block (Fig. 5) at hH1 channels suggest that these two parameters, in conjunction with the rhythmic cardiac action potential, may play significant roles in cocaine-induced cardiotoxicity.

View larger version (19K): [in this window] [in a new window]   FIGURE 5   Use-dependent cocaine block of µ1 and hH1 channels. Sixty pulses to +30 mV were delivered to µ1 channels (A) and to hH1 channels (B) at stimulation frequencies of 1 and 2 Hz. In control saline (filled symbols in A and B), the pulse protocol did not elicit use-dependent decreases in current amplitude at either 1 or 2 Hz. The development of use-dependent block in 30 µM cocaine was best fitted by a single exponential function. (A) Use-dependent cocaine block of µ1 channels developed with a time constant of 4.1 ± 0.3 pulses and reached steady state at 39.4 ± 1.8% block during 1 Hz stimulation, and had a time constant of 4.9 ± 0.2 pulses and reached steady state at 50.1 ± 1.6% block during 2 Hz stimulation. (B) Use-dependent cocaine block of hH1 channels developed with a time constant of 4.5 ± 0.3 pulses and reached steady state at 41.5 ± 1.7% block during 1 Hz stimulation, and had a time constant of 5.4 ± 0.3 pulses and reached steady state at 53.2 ± 1.3% block during 2 Hz stimulation.

Simulation of cocaine block using a modulated receptor model

In a previous report we attributed the differences in steady-state cocaine block between µ1 and hH1 to the differences in their steady-state inactivation curves (Wright et al., 1997). The data in the present study appear to support our claim because the V_0.5 values of steady-state cocaine block were better correlated with the V_0.5 values of steady-state inactivation than with the V_0.5 values of activation. Furthermore, the shifts in steady-state inactivation did not alter the affinities of resting and inactivated channels. To determine whether we could predict the voltage dependence of cocaine block, we used the steady-state inactivation data from each channel, as well as the K_R and K_I values to simulate steady-state cocaine block. For the simulation, we first determined the apparent K_d using the h curve of each channel. That is,

(1)

where K_app is the apparent K_d, and K_R and K_I are the K_d values at resting and inactivated channels, respectively. The parameters h and 1  h are the fractional distributions of resting and inactivated channels, respectively, at a given conditioning voltage (Bean et al., 1983). We used 250 µM as the K_R value and 9 µM as the K_I value. Note that in Eq. 1 the fraction of inactivated channels (1  h) has a larger influence on the apparent K_d than does the fraction of available channels (h), and thus imposes a leftward shift on the simulation. For example, at the conditioning voltage where h = 0.9 and 1  h = 0.1, the apparent K_d is 68 µM; when both h and 1  h = 0.5, the apparent K_d is 17 µM. Fig. 6 A shows the K_app curves for µ1 and hH1 channels.

View larger version (16K): [in this window] [in a new window]   FIGURE 6   Simulations of cocaine block of µ1 and hH1 sodium channels. (A) Plot of the apparent Kd versus the conditioning voltage using Eq. 1. (B) and (C) The solid lines show the predictions of cocaine block of µ1 and hH1 channels using Eqs. 1 and 3. The filled symbols are the actual percentages of block by 30 µM cocaine, and the open symbols are the percentages of block by 300 µM cocaine. The dotted lines are the improved fits to the data after imposing shifts of 4.5 mV and 10 mV on the h curves of µ1 and hH1, respectively. See text for details.

We then used the K_app values from Eq. 1 and the Langmuir isotherm (Hille, 1992) to predict the percentage of available channels at a given cocaine concentration:

(2)

(3)

where I_Na is the peak current measured during the test pulse and [LA] is the cocaine concentration.

We applied the simulation to µ1 and hH1 channels and used the 30 µM and 300 µM cocaine data from Wright et al. (1997) to see how well the model fit the data. The solid lines in Figs. 6 B and C) are the results of the simulation using Eqs. 1 and 3, and the symbols are the percentages of available channels in 30 µM cocaine (filled symbols) and in 300 µM cocaine (open symbols). The simulation predicted steady-state cocaine block of µ1 channels within a few millivolts at the midpoint of block (Fig. 6 B), whereas steady-state cocaine block of hH1 channels was several millivolts more negative than predicted by the simulation (Fig. 6 C). Scheme A in Diagram D1 depicts the simplified model (Bean et al., 1983) that uses Eqs. 1 and 3 to predict cocaine block. Thus, even though Eq. 1 imposed a significant leftward shift on the K_app value, the simulation failed to account for an equilibrium shift toward the inactivated and blocked state. One of the defining characteristics of a modulated receptor (Hille, 1977) is that local anesthetic shifts the equilibrium from the resting and drug-bound state (R*) toward the inactivated and drug-bound state (I*) as depicted in Scheme B. The difference between the amount of block predicted by the initial simulation (Scheme A) and the actual amount of block may be indicative of an equilibrium shift in the R*  I* portion of Scheme B (Courtney, 1975). Note that we could not directly measure this equilibrium shift because channels in the R* and I* states do not conduct. To account for the additional equilibrium shift at the drug-bound states and to improve the fit by the model, we adjusted the h curves of µ1 and hH1 channels by trial and error. Adjusting the h curves of µ1 and hH1 channels by 4.5 mV and 10 mV, respectively, improved how well the model predicted cocaine block (dotted lines in Fig. 6, B and C).

View larger version (9K): [in this window] [in a new window]   Diagram 1

We also used the model to simulate 30 µM cocaine block of the channel chimeras, µ1-1 channels, and hH1-1 channels (Fig. 7). The solid lines in Fig. 7, A-D show the predicted results using Eqs. 1 and 3 for 30 µM cocaine block of µ1_(1-3)hH1₍₄₎, µ1₍₁₎hH1_(2-4), µ1-1, and hH1-1 channels. Introducing equilibrium shifts of 5 mV and 6 mV, respectively, to the h curves of µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) channels improved the prediction (Fig. 7, A and B, dotted lines). As with µ1 and hH1 channels, applying equilibrium shifts of 4.5 mV and 10 mV to the h curves of µ1-1 and hH1-1 channels, respectively, improved the prediction by the model (Fig. 7, C and D, dotted lines). Fig. 8 plots the average V_0.5 values of cocaine block versus the average V_0.5 values of steady-state inactivation. Each symbol corresponds to the averaged data from one of the isoforms in the study (see Fig. 8 legend). The cross symbol directly above each data point is the corresponding V_0.5 value of cocaine block as predicted by the simple model (Scheme A). The simulation results in Figs. 6-8 show that the equilibrium shift was larger for hH1 and hH1-1 channels than for any of the other channels, suggesting that a larger equilibrium shift from R* to I* at the hH1 isoform may be an important factor in cocaine-induced cardiotoxicity (see Discussion).

View larger version (26K): [in this window] [in a new window]   FIGURE 7   Simulations of cocaine block of channel chimeras and channels coexpressed with the 1 subunit. As described in Fig. 6, the apparent Kd at each channel was first determined using Eq. 1. (A-D) The solid lines show the predictions of cocaine block by 30 µM cocaine, and the dotted lines show the improved fits after imposing negative shifts in the h curves. The shifts for µ1(1-3)hH1(4), µ1(1)hH1(2-4), µ1-1, and hH1-1 channels were 5 mV, 6 mV, 4.5 mV, and 10 mV, respectively.

View larger version (14K): [in this window] [in a new window]   FIGURE 8   Plot of V0.5 values of cocaine block versus the V0.5 values of steady-state inactivation for the six channels in the study. As in Fig. 3, the solid line is the regression line through the plot of average V0.5 values of cocaine block versus the average V0.5 values of steady-state inactivation. In this figure, the individual channels are distinguished by a symbol (filled square, µ1; open square, µ1-1; filled circle, hH1; open circle, hH1-1; open triangle, µ1(1-3)hH1(4); and open diamond, µ1(1)hH1(2-4)). The crosses are the predicted V0.5 values of cocaine block using Eqs. 1 and 3 and the dotted line is the regression through the points. The cross symbol located directly above each data point corresponds to the predicted V0.5 value of cocaine block for that particular channel. As discussed in the text, the predicted values for hH1 channels and hH1-1 channels had the largest deviations from the measured data (filled circle and open circle, respectively).

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

In a previous paper (Wright et al., 1997), we showed that mammalian isoforms of cardiac and skeletal muscle sodium channel had similar resting and inactivated affinities for cocaine and lidocaine. This was in contrast to other studies (Nuss et al., 1995b; Wang et al., 1996b) which suggested that the sodium channels in cardiac tissue have a greater affinity for lidocaine compared to other sodium channel isoforms, and that this greater affinity might explain why cardiac tissue is relatively more sensitive than skeletal muscle to certain local anesthetics. We suggested that the differences in steady-state inactivation between hH1 channels and µ1 channels may in part explain the cardiotoxic effects of cocaine.

In the present study we further addressed state-dependent cocaine block of µ1 and hH1 sodium channels by focusing on the mechanisms responsible for the greater cocaine block of the hH1 isoform at intermediate voltages. First, we used shifts in the steady-state inactivation of µ1 and hH1 channels as a tool for investigating steady-state cocaine block. We induced shifts in steady-state inactivation by creating µ1/hH1 channel chimeras and also by coexpressing the rat brain 1 subunit with the  subunits of µ1 or hH1. The shifts in the midpoint voltage of steady-state inactivation produced by these methods elicited directionally similar shifts in the midpoint voltage of cocaine block. Second, unlike several previous studies conducted with other expression systems, 1 subunit coexpression with the µ1 or hH1  subunits in HEK cells shifted the voltage dependence of steady-state inactivation to more positive voltages, rather than toward more negative voltages, and did not affect the time constant of macroscopic current decay. In addition, coexpression of the 1 subunit did not alter the cocaine affinities of either resting or inactivated channels. And third, we used the h curves, the K_d values at resting and inactivated channels, and the Langmuir isotherm to simulate steady-state cocaine block.

Effects of channel chimera structure and 1 subunit coexpression on channel kinetics

We found it interesting that channel activation for both µ1_(1-3)hH1₍₄₎ and µ1₍₁₎hH1_(2-4) resembled the activation kinetics of µ1 channels. Because µ1₍₁₎hH1_(2-4) contains only domain 1 from µ1 channels, it would be tempting to speculate that domain 1 is the crucial domain for determining channel activation. However, Chahine et al. (1996) and Benzinger et al. (1997) have also shown that the activation of several different µ1/hH1 channel chimeras each resembled the activation of µ1. Recently, Mitrovic et al. (1998) used cysteine point mutation within D2S4 and subsequent treatment with a cysteine modifying agent to demonstrate that domain 2 plays a significant role in channel activation. The findings by Mitrovic et al. (1998) and the data in Fig. 1 B argue against the hypothesis (Marcotte et al., 1997) that activation begins by outward movement of the S4 segment containing the most charged residues (domain 4) and concludes with the outward movement of the S4 segment with the fewest charged residues (domain 1). If this type of sequential movement of S4 segments occurred, then the activation phenotype of µ1₍₁₎hH1_(2-4) should have resembled the activation of hH1 channels rather than µ1 channels. These data suggest that channel activation is much more complex than can be deduced from the number of charged residues in the S4 segments. One possible explanation for the discrepancy between our channel chimera data and the apparent importance of domain 2 in activation (Mitrovic et al., 1998) may be that domain regions from the µ1 isoform dominate the activation phenotype of channel chimeras constructed from µ1 and hH1.

The steady-state inactivation data obtained for µ1, hH1, and the channel chimeras indicate that the four domain regions have an evenly distributed role in determining the h phenotype (Fig. 1 C). The V_0.5 of steady-state inactivation for µ1₍₁₎hH1_(2-4) channels was ~4 mV less negative than that of hH1, and the V_0.5 of inactivation for µ1_(1-3)hH1₍₄₎ channels was ~4 mV more negative than that of µ1. In another chimera study (Benzinger et al., 1997), the effects of a single domain substitution on steady-state inactivation phenotype seemed less clear because the V_0.5 values of steady-state inactivation for all chimeras were actually less negative than that of µ1. Although we were unable to test this phenomenon further using comparable channel chimeras (hH1₍₁₎µ1_(2-4) and hH1_(1-3)µ1₍₄₎), the present data suggest that each domain contributes, perhaps equally, to the steady-state inactivation phenotype.

The shifts in the voltage dependence of activation and/or steady-state inactivation after coexpression of sodium channel  subunits with the rat brain 1 subunit appear to vary depending on the expression system and perhaps also on the sodium channel isoform. Coexpression of the 1 subunit shifted the activation and inactivation kinetics of rat brain sodium channels to more negative voltages in Xenopus oocytes (Isom et al., 1992) and in Chinese hamster cells (Isom et al., 1995). 1 subunit coexpression with µ1 channels in oocytes also elicited a negative shift in the voltage dependence of steady-state inactivation (Nuss et al., 1995a). In oocytes, the effect of 1 subunit coexpression with cardiac sodium channels varies from no shift in the steady-state inactivation of hH1 channels (Nuss et al., 1995b) or rat heart sodium channels (rH1; Qu et al., 1995) to one report of a modest but significant positive shift for hH1 channels (Makielski et al., 1996). In contrast to these studies, we found that coexpression of the 1 subunit with sodium channel  subunits in HEK cells markedly shifted steady-state inactivation (Fig. 2 C) of µ1 and hH1 channels in the positive direction by ~10 mV and ~13 mV, respectively. We cannot presently explain the variable effects of 1 subunit coexpression on sodium channel kinetics. One possibility may be that cellular processes such as protein phosphorylation (see Cukierman, 1996 for review), which can affect channel kinetics, differ among different expression systems. Alternatively, the time-dependent negative shift in steady-state inactivation (Wang et al., 1996a) may be reduced or eliminated by 1 subunit coexpression.

The positive voltage shifts in steady-state inactivation produced by 1 subunit coexpression with µ1 and hH1 resulted in directionally similar shifts in cocaine block (Fig. 2 D). The average V_0.5 values of cocaine block for µ1-1 and hH1-1 were 12.3 mV and 15.0 mV, respectively, more positive than the V_0.5 values of cocaine block for the  subunits of µ1 and hH1. Also important, 1 subunit coexpression did not affect the cocaine affinities of either resting or inactivated channels. In contrast, 1 subunit coexpression with hH1 channels in the oocyte expression system decreased the affinity of resting channels for lidocaine by ~2-fold (Makielski et al., 1996); a result most likely due to a small (3-7 mV) positive shift in steady-state inactivation.

Steady-state cocaine block, the modulated receptor model, and cocaine-induced cardiotoxicity

The data in Figs. 1-3 indicated that shifts in the voltage dependence of steady-state inactivation induced linear shifts in the voltage dependence of steady-state cocaine block. The simple model (Scheme A in Diagram D1; Bean et al., 1983), which used the h curve and the K_d values at resting and inactivated channels, gave a fairly accurate prediction of steady-state cocaine block at µ1 channels (V_0.5 within ~5 mV), but gave a somewhat less accurate prediction at hH1 channels (V_0.5 within ~10 mV). Even though the V_0.5 values of cocaine block were shifted by ~10 mV compared to the V_0.5 values of steady-state inactivation, the simple model failed to account for the entire equilibrium shift because it did not include the transitions between the R* and I* states (Scheme B). Thus, the difference between µ1 and hH1 channels in the size of the underestimated equilibrium shift may reflect the difference in the amount of cocaine-induced leftward shift in h, which in our case becomes evident in the V_0.5 value of steady-state cocaine block. For example, the V_0.5 values of steady-state inactivation at µ1 and hH1-1 channels were similar at 79 mV and 81 mV, respectively, but the V_0.5 value of cocaine block was ~8 mV more negative for hH1-1 channels than for µ1 channels. Indeed, these data suggest that the 15 mV difference between the V_0.5 values of steady-state inactivation at µ1 and hH1 (Wright et al., 1997), as well as a larger cocaine-induced shift in the h curve of hH1 channels contribute to cocaine-induced cardiotoxicity.

Although steady-state inactivation appears to play the major role in determining steady-state cocaine block, channel activation could be one of the factors responsible for the magnitude of the underestimation in h shift. Compare the V_0.5 value of steady-state inactivation to the V_0.5 value of cocaine block for each channel (Table 1), and note the interesting quantitative difference between µ1 and hH1. For µ1 channels the difference between the V_0.5 value of cocaine block and the V_0.5 value of steady-state inactivation was ~14 mV and for hH1 channels the difference was ~19 mV. These differences were essentially the same at µ1-1 and hH1-1 channels. For both of the channel chimeras, the difference between the V_0.5 value of steady-state cocaine block and the V_0.5 value of steady-state inactivation was ~15 mV. When we adjusted the h curves to improve the fit by the simple model, the shift in the h curve that best improved the fit to the cocaine data from µ1 and µ1-1 channels was 4.5 mV, whereas the adjustment for hH1 and hH1-1 channels was 10 mV. For µ1₍₁₎hH1_(2-4) and µ1_(1-3)hH1₍₄₎, the adjustments were 5 and 6 mV, respectively, which more closely resembled the adjustment required for µ1 channels. The fact that µ1 channels and the channel chimeras had the same activation phenotype and required similar adjustments in h to improve the fit by the simple model suggests that channel activation may affect the magnitude of the underestimated shift. As the conditioning voltage becomes less negative, hH1 channels enter preopen or preactivated states, which may influence the equilibrium shift from R* to I*. However, at the µ1 isoform and at the chimeras, channel activation begins at voltages ~15 mV more positive than at the hH1 isoform, so these channels would therefore not enter the preactivated states at the same voltages. This possible role for activation in the equilibrium shift of the h curve is consistent with previous suggestions that activation is important in local anesthetic action (Starmer et al., 1984; Yeh and Tanguy, 1985).

The notion that state-dependent differences between cardiac and skeletal muscle sodium channels influence local anesthetic action in cardiac tissue has been previously reported for single batrachotoxin-modified sodium channels (Zamponi et al., 1993; Zamponi and French, 1993). In native channels, the antiarrhythmic properties of lidocaine and the cardiotoxic properties of cocaine most likely result from steady-state interactions with inactivated sodium channels at the diastolic membrane potential, as well as from use-dependent interactions with activated and inactivated channels during the cardiac action potential. Here, we present an additional mechanism wherein cocaine induces a larger shift in the h curve at cardiac sodium channels than at skeletal muscle sodium channels. This additional shift in h may arise from the coupling between channel activation and inactivation (O'Leary et al., 1995), both of which proceed at significantly more negative potentials in the cardiac isoform.