Veratridine (VTD) is an alkaloid toxin found in Liliaceae plants. VTD causes persistent opening of the voltage-gated Na⁺ channel and reduces its single-channel conductance by 75 %. The mechanisms for these different VTD actions are unknown. Recent reports indicate that the VTD receptor aligns closely with the local anaesthetic (LA) receptor, which resides at D1S6, D3S6 and D4S6 of the Na⁺ channel α-subunit. To study this alignment, we created a mutant with cysteine substitutions at three S6 residues (rNav1.4-N434C/L1280C/F1579C). Under voltage-clamp conditions, amitriptyline and bupivacaine remained as potent blockers of this mutant channel when expressed in human embryonic kidney cells, whereas VTD completely failed to cause persistent opening. Unexpectedly, VTD at 100 μm progressively blocked mutant currents by 90.4 ± 1.6 % (n = 5), as assayed at 0.1 Hz for 15 min. This VTD block was reversed little during wash-off: ~70 % of mutant currents did not return in 30 min. An increase in channel opening either by repetitive pulses at 1 Hz or by the inhibition of the fast inactivation hastened the VTD block. Co-application of amitriptyline or bupivacaine, which targeted the LA receptor, prevented this VTD block. Our data suggest that (a) the VTD receptor and the LA receptor overlap extensively, (b) receptor-bound VTD lies in the inner vestibule, and (c) VTD blocks this mutant channel as a bona fide Na⁺ channel blocker. We propose that VTD likewise blocks the wild-type open Na⁺ channel, albeit partially, to decrease the unitary conductance and to stabilize the open conformation for persistent opening.

Voltage-gated Na⁺ channels are responsible for the generation of action potentials in excitable membranes. Many therapeutic drugs and natural neurotoxins target Na⁺ channels as their primary actions. For example, local anaesthetics (LAs) are clinical drugs that block Na⁺ channels, whereas veratridine (VTD; Fig. 1A, pK_a = 9.5) and batrachotoxin (BTX) are Na⁺ channel activators that promote persistent Na⁺ channel opening (Hille, 2001). VTD is found in Liliaceae plants (e.g. sabadilla seed of Mexican Schoenocaulon officinale), and BTX is found in the skin of South American frogs (e.g. Phyllobates terribilis). The profound effects of VTD and BTX on Na⁺ channel gating have long been recognized; both toxins have been used widely as pharmacological tools (Catterall, 1980). The receptor for VTD and BTX is termed the Na⁺ channel type-2 receptor.

Mammalian Na⁺ channel isoforms consist of one large α-subunit and one or two smaller β-subunits (β₁-β₃) (Catterall, 2000). The α-subunit protein, which alone forms a functional Na⁺ channel, contains four homologous repeats (domains D1–D4), each with six transmembrane α-helical segments (S1–S6). Several reports indicate that the LA receptor and the type-2 receptor for BTX and VTD align closely in situ and perhaps even overlap (Wang et al. 1998; Linford et al. 1998; G. K. Wang et al. 2000; S.-Y. Wang et al. 2000; Yarov-Yarovovy et al. 2001; Li et al. 2002). These two receptors share common amino acid residues at S6 α-helical segments derived from different domains (Fig. 1B). The LA receptor is supposedly located at three of four S6 segments, which may jointly form the inner vestibule of the Na⁺ channel permeation pathway. Accumulated evidence indicates that LAs block the Na⁺ channel from the inner vestibule (Catterall & Mackie, 1996; Wang, 2002). In contrast, the mechanisms for VTD actions remain largely unknown (Strichartz et al. 1987). It is perplexing that the VTD-modified channel carries only ~25 % of the normal single-channel conductance, while the channel remains persistently open (e.g. Barnes & Hille, 1988). To date, no reports indicate that VTD blocks the Na⁺ permeation pathway directly from the inner vestibule. It remains plausible (Hille, 2001) that receptor-bound VTD does not lie in the inner vestibule and that lipophilic VTD does not arrive and exit directly via the permeation pathway, since VTD is perhaps too large to enter the pore.

To address the relative location of the VTD receptor with respect to the inner vestibule, we created a mutant Na⁺ channel, replacing three putative LA-binding residues (N434 at D1S6, L1280 at D3S6 and F1579 at D4S6; Fig. 1B) with cysteine residues. We selected these three loci because they were also critical for BTX and VTD binding (Wang & Wang, 1998; G. K. Wang et al. 2000; S.-Y. Wang et al. 2000). We show that VTD blocks the permeation pathway of this mutant channel and that this VTD block can be reversed only slowly, and incompletely, on the time scale of our experiments.

We used the QuikChange XL site-directed mutagenesis kit (Stratagene) to create rat skeletal muscle Nav1.4 mutant clones as described previously (Wang & Malcolm, 1999). DNA sequencing near the mutated site confirmed these mutations. Preliminary results showed that rNav1.4-N434C, L1280C and F1579C mutants were reversibly inhibited by VTD to a level similar to that of the wild-type currents. VTD also elicited maintained currents and tail currents in rNav1.4-L1280C and rNav1.4-F1579C mutants but only minimally in the rNav1.4-N434C mutant. We then created the rNav1.4-N434C/L1280C/F1579C (3-C) mutant channel and used it exclusively for our study because this channel exhibited a unique VTD block with limited reversibility. To minimize the possibility of unwanted substitutions at positions other than the targeted sites, we selected a second independent 3-C clone from separate QuikChange synthesis reactions. We found that its VTD block was indistinguishable from that of the first 3-C mutant, indicating that the VTD block was not due to unwanted substitutions.

Human embryonic kidney (HEK293t) cells were grown to ~50 % confluence in Dulbecco's modified Eagle's medium (DMEM; Gibco) containing 10 % fetal bovine serum (HyClone), 1 % penicillin and streptomycin solution (Sigma), 3 mm taurine and 25 mm Hepes (Gibco), and then transfected by a calcium phosphate precipitation method in a Ti25 flask (Cannon & Strittmatter, 1993). Transfection of wild-type rNav1.4-pcDNA1/Amp or mutant clones (5–10 μg) along with β₁-pcDNA1/Amp (10–20 μg) and reporter CD8-pih3m (1 μg) was adequate for later current recording. Cells were re-plated 15 h after transfection in 35 mm dishes, maintained at 37 °C in a 5 % CO₂ incubator, and used after 1–4 days. Transfection-positive cells were identified with immunobeads (CD8-Dynabeads, Lake Success, NY, USA).

The whole-cell configuration was used to record Na⁺ currents (Hamill et al. 1981). Borosilicate micropipettes (Drummond Scientific Company, Broomall, PA, USA) were pulled with a puller (P-87, Sutter Instrument Company, Novato, CA, USA) and heat polished. Pipette electrodes contained 100 mm NaF, 30 mm NaCl, 10 mm EGTA and 10 mm Hepes adjusted to pH 7.2 with CsOH. The pipette electrodes had a tip resistance of 0.5–1.0 MΩ. Access resistance was 1–2 MΩ and was further reduced by series resistance compensation (≥ 80 %). All experiments were performed at room temperature (22–24 °C) under a Na⁺-containing bath solution with 65 mm NaCl, 85 mm choline chloride, 2 mm CaCl₂ and 10 mm Hepes adjusted to pH 7.4 with tetramethylammonium hydroxide. Residual outward currents were evident in some cells at voltages ≥+30 mV (Wang & Wang, 1998); these currents were present in untransfected cells and were insensitive to tetrodotoxin. Stock solution of VTD base (Sigma, St Louis, MO, USA) was prepared in dimethyl sulphoxide at 20 mm. Whole-cell currents were measured by an Axopatch 200B (Axon Instruments, Foster City, CA, USA), filtered at 3 kHz, collected and analysed with pClamp8 software (Axon Instruments). Leak and capacitance were subtracted by the Axopatch device and further by the leak subtraction protocol (P/-4). Control and drug solutions were applied near the cell with a multibarrel perfusion system. Cells were held either at −100 mV for VTD experiments or at −140 mV for functional characterizations. Voltage error was < 3 mV after series resistance compensation. Student's unpaired t test was used to evaluate estimated parameters (means ±s.e.m. or fitted value ±s.e.m. of the fit); P values of < 0.05 were considered statistically significant.

We detected sufficient 3-C mutant Na⁺ currents in HEK293t cells when cotransfected with the β₁-subunit. Figure 2A shows the family of Na⁺ currents at various voltages. Both wild-type and mutant currents were completely blocked by external 1 μm tetrodotoxin. The 3-C mutant channels were 50 % activated at −14.5 ± 1.2 mV (n = 5; Fig. 2B), which is 16.9 mV more positive than wild-type channels (−31.4 ± 1.0 mV, n = 5; P < 0.001). The 3-C Na⁺ currents decayed more rapidly than wild-type currents, with time constants at +30 mV of 190 ± 7 μs (n = 7) and 336 ± 4 μs (n = 6), respectively (P < 0.001). The steady-state fast inactivation was measured by a conventional two-pulse protocol (Fig. 2C), and the inactivation parameters were determined. The value obtained from the 3-C mutant channels differed by ~4 mV in h_0.5 (the prepulse voltage at which 50 % of the peak current was fast inactivated) from that of the wild-type counterparts (P < 0.001; Fig. 2D).

The sensitivity of the wild-type and 3-C mutant channels to bupivacaine and amitriptyline at various concentrations was measured at −160 mV for the resting block (K_R) and at −70 mV for the inactivated block (K_I) (Fig. 3A and B), as described previously (Nau et al. 2000). The K_R and the K_I values were estimated from these plots (Fig. 3C). The Hill coefficient values were between 0.7 and 1.3 except for that of amitriptyline, which yielded a value of 1.5 for wild-type and 2.0 for the 3-C mutant at −160 mV (filled circles). The reason for this apparent deviation is unclear but could be due to an incomplete approach to the steady state, and/or a shift in steady-state inactivation by the high concentrations of amitriptyline, which may result in a fraction of inactivated Na⁺ channels present at −160 mV. It is noteworthy that the resting block for both drugs was significantly enhanced in the 3-C channel (Fig. 3C; P < 0.001), whereas the inactivated block remained comparable to the wild-type.

The effects of VTD in wild-type channels reached steady state within 5 min when the cell was held at −100 mV. The inhibition of wild-type peak currents by external 100 μm VTD was 14.2 ± 0.9 % (n = 6; Fig. 4A; Before vs. After trace). After attaining the peak amplitude, the Na⁺ currents decayed normally but this did not reach completion during depolarization. A fraction of the Na⁺ currents were maintained (dashed arrow), while the untreated current trace reached the base line. In addition, a significant amount of tail current (dotted arrow) appeared upon repolarization to the holding potential (−100 mV) after the VTD application. This tail current decayed slowly, with a time constant of 0.73 ± 0.06 s (n = 4; Fig. 4A, inset). All VTD-induced tail currents and maintained currents were reversed readily during wash-off and usually reached completion within 5 min. The inhibition of the peak current by VTD was also reversible by washing. These VTD phenotypes in the wild-type Na⁺ channel with β₁ were comparable to those described previously without β₁ (Wang & Wang, 1998; G. K. Wang et al. 2000).

Under identical conditions, VTD at 100 μm inhibited the peak 3-C mutant Na⁺ current continuously (Fig. 4B). None of current traces showed any maintained currents or tail currents induced by VTD. These current traces were generated every 10 s at +30 mV, and the VTD inhibition of peak currents progressed slowly, with a time constant of 217.6 ± 26.0 s (Fig. 4C; n = 5). After a period of 15 min during which 90.4 ± 1.6 % of 3-C currents were inhibited (n = 5), the cell was superfused with VTD-free solution. Most of the 3-C currents (70.5 ± 5.4 %, n = 4) did not reappear during the wash-off, as continuously assayed every 10 s for 30 min (Fig. 4C). In two separate cells, we applied VTD at 100 μm for 5 min and then washed with external solution without drug. Using the same pulse protocol we found that the peak current was reduced to ~47 % of the control value before wash-off and to ~50 % after 5 min of wash-off. This slow wash-off phenotype in 3-C mutant channels, therefore, occurred even after a brief, 5 min, exposure to VTD. Thus, a fraction of the VTD block was not reversed within our experimental time frame. This VTD block in 3-C mutant channels is distinct from that of wild-type channels in its reversibility and magnitude.

The time course of VTD block in 3-C mutant channels was rather slow during wash-in. One possible reason for this slowness is a dependence of the VTD block on the opening of 3-C mutant channels. Repetitive pulses at a higher frequency that favours frequent channel opening should enhance the rate of VTD block. We found that with repetitive pulses of +30 mV at 1 Hz, the rate of VTD block was indeed accelerated significantly, with a time constant of 36.0 ± 3.6 s (n = 5; P < 0.001; Fig. 5). This value is ~sixfold faster than that at 0.1 Hz. Repetitive pulses also inhibited the peak current further and enhanced the magnitude of the maintained currents as well as the tail currents in VTD-treated wild-type channels (Leibowitz et al. 1986; Sutro, 1986; Wang & Wang, 1998). However, in wild-type channels these effects reached steady state rapidly, usually within 20 pulses, whereas in 3-C mutant channels the VTD block required many more pulses (~150 pulses at 1 Hz; 100 μm) to reach steady state. We estimated the association rate constant for VTD binding with the open 3-C channel to be ~3 × 10⁶m⁻¹ s⁻¹. This value was comparable to that obtained by Leibowitz et al. (1986) with the native Na⁺ channel and based on similar assumptions. Our estimate was derived as 0.03/(0.5 × 0.0002 × 0.0001), or 3 % of open channels blocked by VTD at 100 μm during each pulse (i.e. 1/36 ~ 0.03) for channels staying open for ~0.2 ms with 50 % probability of being open at the peak time. Upon wash-off, the majority of the current again remained blocked. It is interesting that VTD even at a concentration as low as 3 μm blocked more than 50 % of the 3-C mutant currents after 800 pulses at 1 Hz, whereas VTD at a concentration of 3 μm elicited minimal or no effects on wild-type currents. Therefore, the VTD affinity appeared greatly increased in the 3-C mutant channel, probably due to its diminished dissociation rate constant. Furthermore, VTD dissociated readily from wild-type channels upon repolarization to −100 mV, as indicated by the decay of tail currents (τ = 0.73 s; Fig. 4A, inset). In contrast, the VTD block was reversed rather slowly during wash-off even when channels were under repetitive pulses (Fig. 5B). Clearly, VTD did not dissociate from its receptor upon repolarization to −100 mV.

To determine whether the fast inactivation gating also modulates the VTD binding, we treated the cell with chloramine-T for 5 min. Chloramine-T at 0.5 mm readily impaired the fast inactivation of native Na⁺ channels (Wang, 1984; Niemann et al. 1991). Likewise, chloramine-T slowed the decaying phase of the 3-C mutant currents and produced a substantial amount of maintained currents near the end of the 20 ms pulse (Fig. 6A, Before). We limited the treatment with chloramine-T to 5 min and then washed the cell with normal external solution, since prolonged chloramine-T application reduced the peak 3-C mutant current precipitously. VTD blocked the chloramine-T-treated 3-C mutant peak currents potently, as measured at +30 mV at 0.1 Hz during the VTD wash-in (Fig. 6B, open circles). There was an apparent time-dependent VTD block of the non-inactivating 3-C mutant current; the maintained currents were clearly more sensitive than the peak currents to VTD block (Fig. 6A and B, open triangles). The time course of VTD block of peak currents at 0.1 Hz was comparatively fast and well fitted by an exponential function, with a time constant of 51.6 ± 2.6 s (n = 7). This value is fourfold faster than that of untreated 3-C mutant channels (Fig. 4; P < 0.001). This result again suggests that VTD block occurs more readily with the open mutant channel. Evidently, the inactivation gate limited the access of VTD to its receptor site. Wash-off of the VTD block in 3-C mutant channels that were pretreated with chloramine-T remained slow (Fig. 6B, open circles); 72.1 ± 2.0 % of the peak currents did not reappear during a 30 min period (n = 5). This value is not significantly different from that of untreated 3-C mutant channels (Fig. 4C).

If the LA receptor and the VTD receptor overlap considerably, binding of a potent LA drug should prevent the VTD block. To test this possibility, we added a potent Na⁺ channel blocker, amitriptyline at 300 μm, to the solution containing 100 μm VTD. This mixture rapidly and completely blocked the 3-C Na⁺ currents within 10–15 s during wash-in, as expected from amitriptyline block alone (Fig. 7A). After ~80 pulses, each applied at an interval of 1 s, we switched to a drug-free solution. Following this switch, the 3-C Na⁺ currents quickly reappeared, as measured at 0.2 Hz (Fig. 7A and B) and almost reached the level before drug treatment (95.8 ± 3.3 %, n = 4). With 100 μm VTD alone, the majority of the VTD block would not be reversed under the same conditions (Fig. 4 and Fig. 5). Similar results were found with co-application of bupivacaine at 300 μm and VTD at 100 μm. A major fraction of the Na⁺ currents (76.3 ± 1.9 %, n = 4) recovered after a drug-free wash. This result indicates that, at the holding potential of −100 mV, the 3-C mutant interacts with amitriptyline and bupivacaine readily but not with VTD, which interacts primarily with the open channel. We also found that the VTD block could not be reversed by 300 μm bupivacaine if it was applied after the VTD block had already been achieved. In fact, bupivacaine rapidly blocked the remaining 3-C currents after the VTD block, and only this additional block was reversed readily during subsequent wash-off. This phenomenon suggested that the remaining currents were carried by unmodified 3-C mutant channels and that receptor-bound VTD could not be displaced by bupivacaine in this mutant.

We created a mutant Na⁺ channel with cysteine substitutions at three separate S6 loci to study the mechanisms of VTD action. VTD binds to this mutant channel tightly with diminished reversibility and we were able to demonstrate explicitly that (a) the VTD receptor and the LA receptor overlap extensively, (b) receptor-bound VTD lies in the inner vestibule, and (c) VTD blocks this mutant Na⁺ channel. Furthermore, VTD neither elicits maintained currents in this mutant during depolarization nor induces tail currents following repolarization. A discussion of the implications of these findings follows.

What is the evidence supporting the overlapping of the VTD receptor and the LA receptor? First, VTD blocks the 3-C mutant with diminished reversibility. Previous reports indicate that these three cysteine-substituted amino acids (N434, L1280 and F1579) are probably LA-binding residues but may also form a part of the putative type-2 receptor for Na⁺ channel activators (Ragsdale et al. 1994; Wang et al. 1998; Linford et al. 1998; S. Y. Wang et al. 2000; Yarov-Yarovoy et al. 2001). Since LAs block the Na⁺ channel via the inner vestibule, VTD should behave the same as LAs if it interacts with these residues directly. The reason for the diminished reversibility of the effects of VTD in the 3-C mutant channel is not clear, but it may be a consequence of stronger interactions among three sulphhydryl groups (cysteine) and many functional groups found in VTD (Fig. 1A). Second, co-application of activator VTD and amitriptyline/bupivacaine prevents the VTD block. This result demonstrates that LA-occupied Na⁺ channels cannot bind VTD. Furthermore, LAs cannot reverse the VTD block if applied subsequently. Thus, LA amitriptyline /bupivacaine and activator VTD are mutually exclusive in binding under our experimental conditions. It is interesting that benzocaine also fails to block native VTD-modified Na⁺ channels but readily blocks native channels pretreated with chloramine-T (Ulbricht & Stoye-Herzog, 1984). In comparison, BTX-activated rNav1.4 Na⁺ channels in their open conformation can be further blocked by cocaine (Wang & Wang, 1999), suggesting that the VTD receptor has much larger overlapping regions with the LA receptor than the BTX receptor. On the basis of these results, we conclude that the VTD receptor and the LA receptor overlap extensively and that, when bound, VTD and LAs lie in the inner vestibule. The alternative, that these two receptors are far apart and that the VTD block of the 3-C mutant is an indirect effect of cysteine mutations, seems unlikely.

Our results show that repetitive pulses at 1 Hz hasten the VTD block. This use-dependent block phenotype is commonly found for many organic compounds. One example is a quaternary ammonium (QA) derivative of lidocaine (lignocaine), QX-314, which blocks Na⁺ channels in a use-dependent manner (Strichartz, 1973). Internal QX-314 cations become accessible to the LA receptor when the channel is activated, as if the activation gate limits its access. Other examples include various internal QA cations that block voltage-gated K⁺ channels (Armstrong, 1975). Such QA block is also strongly dependent on channel activation.

In addition to the activation gate, the inactivation gate also limits the VTD access during depolarization. Inhibition of the fast inactivation by chloramine-T produces a significant amount of maintained 3-C mutant currents. VTD blocks this maintained current preferentially and elicits an apparent time-dependent block during depolarization. Such time-dependent VTD block of maintained currents is commonly found for organic compounds that block the voltage-gated channel via the inner vestibule (e.g. Armstrong, 1975; Zhou et al. 2001). Together, our results support a hypothesis that VTD binds with the open Na⁺ channel (Sutro, 1986) and that both the activation and inactivation gates limit the VTD access to the inner vestibule.

If the VTD block occurs in the inner vestibule of the channel, could VTD enter and exit the inner vestibule through the cytoplasmic permeation pathway, as envisaged for internal QX-314 cations? A recent model for the opening of K⁺ channels (Jiang et al. 2002) may provide an answer. Pore-lining inner helices of the K⁺ channel contain a gating hinge (a flexible glycine residue) that in a straight relaxed conformation forms an inverted teepee structure to close the pore near its intracellular surface. In a bent conformation of about 30 deg, the inner helices splay open, creating a wide entryway of about 12 Å (1.2 nm). This open conformation allows large organic cations and the ‘inactivation gate’ to enter the K⁺ pore from the intracellular solution (Zhou et al. 2001). Three S6 helices in the rNav1.4 Na⁺ channel also contain a conserved glycine residue at position 12 (Fig. 1B). Assuming that Na⁺ channel S6 helices splay open in a similar fashion, it is feasible that VTD, probably in its protonated form, enters and exits the inner vestibule via the cytoplasmic permeation pathway. In fact, QA ions as large as 15 Å (1.5 nm) in diameter enter and block this Na⁺ permeation pathway (O'Leary et al. 1994; Huang et al. 2000). Other activators, such as BTX, may also take this pathway (Li et al. 2002). Nonetheless, we cannot rule out the possibility that the VTD receptor is also accessible to lipophilic VTD via the membrane phase (Ulbricht, 1998; Hille, 2001), particularly when VTD is in its neutral form.

How VTD causes persistent Na⁺ channel opening remains unresolved. Hille et al. (1987) suggested that VTD acts as a wedge that stabilizes the open state and impedes channel closing. As VTD is first confined in the inner vestibule during its initial binding, it could block the channel, as found in the 3-C mutant. Barnes & Hille (1988) reported that in some single-channel events, a gap of a few milliseconds indeed appears to separate the normal opening from the VTD-modified opening, suggesting that VTD may close the channel initially. However, because the gap is found in < 25 % of total events, it remains uncertain whether VTD fully blocks the native channel initially.

Assuming that a full initial VTD block occurs, we previously suggested that a second-step trapping transition of the VTD receptor is present (G. K. Wang et al. 2000). This trapping transition relieves the channel from an initial full VTD block to a partial one, allowing ~25 % of the current to pass through during depolarization. If this is true, the 3-C mutant channel may interact with VTD too tightly to enter the second transition for channel reopening. Alternatively, VTD binds to the open Na⁺ channel in the inner vestibule and this open pore becomes ‘stuck’ in its conducting state because of the rigid VTD structure. In this second model no initial full VTD block occurs and the VTD-bound open channel remains ‘leaky’, allowing ~25 % of the current to pass through. For the 3-C mutant, the tight binding of VTD completely eliminates the leakage so that no current passes by VTD.

In any event, we propose that VTD blocks the native open channel, although imperfectly, to reduce its unitary conductance by ~75 % and to stabilize its open conformation for persistent opening. We notice that putative gating hinges are adjacent to VTD- and BTX-binding residues (Fig. 1B; position 12 vs. 13). This close in situ association between the gating hinges and bound VTD/BTX in the inner vestibule may explain why VTD and BTX facilitate channel opening. Our model also implies that VTD somehow restrains the inactivation gate from closing the channel during prolonged depolarization as in bilayer studies (Garber & Miller, 1987); otherwise, the VTD-induced channel opening would be shut off rapidly. Such restraint seems possible if S6 helices also form the docking site for the inactivation gate (Zhou et al. 2001; Yarov-Yarovy et al. 2002). VTD in the inner vestibule could modulate this nearby docking site directly or indirectly.