Modulation of human dorsal root ganglion neuron firing by the Nav1.8 inhibitor suzetrigine

Significance

Nav1.8 sodium channels are highly expressed in primary pain-sensing neurons. Suzetrigine is a potent and highly selective Nav1.8 inhibitor that has recently been approved by the Food and Drug Administration for treating acute pain. We find that suzetrigine reduces but does not completely block electrical excitability of human dorsal root ganglion neurons, in part because robust action potentials can be generated by other types of sodium channels in the neurons, including Nav1.7 channels. The results suggest that inhibition of Nav1.8 channels alone may produce only limited reduction of pain signaling by primary pain-sensing neurons.

Abstract

Nav1.8 voltage-gated sodium channels are strongly expressed in human primary pain-sensing neurons (nociceptors) and a selective Nav1.8 inhibitor VX-548 (suzetrigine) has shown efficacy for treating acute pain in clinical trials. Nociceptors also express other sodium channels, notably Nav1.7, raising the question of how effectively excitability of the neurons is reduced by inhibition of Nav1.8 channels alone. We used VX-548 to explore this question, recording from dissociated human dorsal root ganglion neurons at 37 °C. Applying VX-548 at 10 nM (about 25 times the IC₅₀ determined using cloned human Nav1.8 channels at 37 °C) had only small effects on action potential threshold and upstroke velocity but substantially reduced the peak and shoulder. Counterintuitively, VX-548 shortened the refractory period—likely reflecting reduced potassium channel activation by the smaller, narrower action potential—sometimes resulting in faster firing. Generally, repetitive firing during depolarizations was diminished but not eliminated by VX-548. Voltage clamp analysis suggested two reasons that repetitive firing often remains in 10 to 100 nM VX-548. First, many neurons had such large Nav1.8 currents that even 99% inhibition leaves nA-level Nav1.8 current that could help drive repetitive firing. Second, Nav1.7 current dominated during initial spikes and could also contribute to repetitive firing. The ability of human neurons to fire repetitively even with >99% inhibition of Nav1.8 channels may help explain the incomplete analgesia produced by even the largest concentrations of VX-548 in clinical studies.

Primary pain-sensing neurons (nociceptors) have distinctive electrophysiological properties, notably the expression of multiple types of voltage-dependent sodium channels with different properties. Like most neurons, nociceptors possess fast-activating sodium channels that are sensitive to tetrodotoxin. However, different from most other types of neurons, nociceptors also possess a distinct slower-activating sodium current that is resistant to tetrodotoxin (1–6), reviewed in (7–11). Cloning and studies of knock-out mice showed that the major TTX-resistant component of current in mouse small DRG neurons is carried by Nav1.8 channels [12–14, reviewed in (11)].

Relative to the TTX-sensitive component of sodium current, Nav1.8-mediated current requires larger depolarization for activation, activates and inactivates more slowly, and requires less hyperpolarized holding voltages for full availability compared to the TTX-sensitive currents (7–10). Results using knock-out mice, modeling, action potential clamp, and dynamic clamp have shown how the distinctive voltage dependence and kinetics of Nav1.8 channels underlie particular functional roles in action potential generation in rat and mouse small DRG neurons: although Nav1.8 current activates more slowly than TTX-sensitive current, it nevertheless carries the majority of current during the upstroke of the action potential and the slow inactivation results in maintained current that contributes to the wide shoulder typical of the action potential of these neurons (12–17). In addition, Nav1.8 channels can carry a “resurgent” component of current that likely contributes to the wide shoulder of nociceptor action potentials (18–20). The more depolarized voltage dependence of inactivation results in more availability at more depolarized voltages than TTX-sensitive current, so that repetitive firing by prolonged depolarizations depends mainly on Nav1.8 current, at least in rat and mouse neurons (21, 22).

The overall excitability of nociceptors depends in a complex manner on the ratio of TTX-sensitive and TTX-resistant current (7, 14, 15, 17, 23). A further complexity is that the TTX-sensitive current in rat and mouse neurons likely represents a combination of channels, with Nav1.7 channels predominating but other TTX-sensitive channels also present (3, 23–26). The major contribution of both Nav1.8 and Nav1.7 channels to overall nociceptor excitability is reflected in the discovery of pain-related mutations in both Nav1.7 and Nav1.8 that produce hyperexcitability by shifting activation to less depolarized voltages or by reducing inactivation (27–30).

Because Nav1.8 channels are expressed in nociceptors with little or no expression in most other kinds of neurons, they are a particularly attractive target for developing inhibitors to treat pain by selectively reducing excitability of nociceptors (11). This approach was validated in rodent models of pain with early Nav1.8 inhibitors, including A-803467 (31) and A-887826 (32) and in rodent and nonhuman primate models of pain with newer Nav1.8 inhibitors (33–35). Most recently, several highly potent and highly selective Nav1.8 inhibitors, VX-150 and VX-548, have been advanced to clinical trials and following positive clinical trials (36), VX-548 (suzetrigine) has recently been approved by the U. S. Food and Drug Administration for the treatment of moderate to severe acute pain.

Although VX-548 produced relief from acute pain in clinical trials, the effects were far from complete analgesia and were significantly different from placebo only at the highest doses, even though VX-548 is an exceptionally potent inhibitor of Nav1.8 channels, with an IC₅₀ < 1 nM (36, 37). The presumed mechanism of action of VX-548 is to reduce excitability of primary nociceptors (36, 37). Because the neurons express other sodium channels in addition to Nav1.8, an obvious question is the extent to which excitability is reduced by selective inhibition of only Nav1.8 channels. We explored how VX-548 modifies the electrical excitability of sensory neurons dissociated from human dorsal root ganglia (DRG).

Results

We examined the effect of VX-548 on the excitability of neurons isolated from the DRG of adult human donors. Because both excitability and drug effects are likely to vary with temperature, we did experiments at 37 °C. We first defined the potency of VX-548 at 37 °C using a cell line expressing human Nav1.8 channels. Interestingly, we found that the potency of VX-548 is reduced about fivefold at 37 °C compared to room temperature (Fig. 1 and SI Appendix, Fig. S2). At both 21 °C and 37 °C, the dose dependence of VX-548 inhibition could be well-fit by assuming a simple 1:1 binding reaction with a K_d of 0.07 nM at 21 °C and 0.38 nM at 37 °C. Similar to tetrodotoxin (38), weaker binding of VX-548 at higher temperature may reflect strong temperature dependence of the unbinding rate constant (SI Appendix, Fig. S2).

Fig. 1.

VX-548 is less potent on human Nav1.8 channels at 37 °C than 21 °C. (A) Dose–response for inhibition of human Nav1.8 currents at 21 °C, mean ± SD. Measurements from 49 cells, 40 exposed to a single concentration and 9 to first a low (0.03, 0.3, or 1 nM) then a high (3, 10, or 30 nM) concentration; n = 6 for 0 (relative current measured after 10 min of exposure to control drug-free solution containing 0.1% DMSO), 4 for 0.003 nM, 7 for 0.01 nM, 8 for 0.03 nM, 9 for 0.1 nM, 5 for 0.3 nM, 5 for 1 nM, 5 for 3 nM, 5 for 10 nM, 4 for 30 nM. (B) Dose–response at 37 °C, mean ± SD. Measurements from 43 cells, 24 exposed to a single concentration and 19 to first a lower (0.03 or 1 nM) then a higher (3, 10, or 30 nM) concentration; n = 6 for 0 (relative current measured after 10 min of exposure to control drug-free solution containing 0.1% DMSO), 6 for 0.03 nM, 7 for 0.1 nM, 15 for 0.3 nM, 8 for 1 nM, 8 for 3 nM, 9 for 10 nM, 3 for 30 nM. Fitted curves: I_max/[1 + (Drug)/K_d], where both I_max and K_d were allowed to vary and the fit included weighting of data points by SD. 21 °C: I_max = 1.00, K_d = 0.072 nM. 37 °C: I_max = 0.99, K_d = 0.38 nM.

Based on these results, we tested the effects of inhibiting Nav1.8 channels on generation of action potentials by human DRG neurons at 37 °C using VX-548 at a concentration of 10 nM, which should nearly completely inhibit Nav1.8 channels (~25 times the K_d). Fig. 2A shows a typical example of the effect of VX-548 on an action potential evoked by a short (0.5 ms) current injection. There was almost no effect of inhibiting Nav1.8 channels on the initial rising phase of the action potential, but the peak of the action potential was decreased. The most dramatic effect was to produce a narrowing of the action potential by reducing the “shoulder” on the falling phase of the action potential, a distinctive characteristic of the shape of action potentials in C-fiber nociceptors in both rodents (39) and humans (40, 41). Fig. 2 C–F show collected results for the effect of 10 nM VX-548 on action potential parameters determined by protocols with short current injections like that in Fig. 2A. With this protocol, the action potential shape is not affected by the stimulating current and the voltage threshold for the action potential can be determined with precision (SI Appendix, Fig. S3). In most neurons, VX-548 produced very little change in the voltage threshold; on average, the threshold depolarized by only 1.2 ± 0.4 mV from an average control value of −47.3 ± 1.4 mV (mean ± SEM, n = 18; P = 0.017, two-tailed Wilcoxon signed rank test). The effect of inhibiting Nav1.8 channels on the rising phase of the action potential was seen most clearly in phase-plane plots of dV/dt versus voltage (Fig. 2B). VX-548 had a small effect on the maximum upstroke of the action potential, which decreased by an average of 15 ± 7% from an average initial value of 502 ± 48 mV/ms (mean ± SEM, n = 18; P = 0.095, two-tailed Wilcoxon signed rank test). The peak of the action potential was decreased by VX-548 in 17 out of 18 neurons, decreasing by an average of 12 ± 2 mV from the average control value of +44 ± 1 mV (mean ± SEM, n = 18; P = 0.0002, two-tailed Wilcoxon signed rank test). In 16 out of 17 neurons, VX-548 produced a narrowing of the action potential, by an average of 36 ± 5% from an average initial value of 1.8 ± 0.2 ms (mean ± SEM, n = 17, width measured at half-maximal amplitude; P = 0.0004, two-tailed Wilcoxon signed rank test). The narrowing was typically accompanied by a reduction of the shoulder of the action potential, as in Fig. 2A. In the cell in which the action potential in VX-548 became wider at half-maximum amplitude, it was because the peak was greatly reduced so that half-maximal amplitude was reached at a much more negative voltage than in control.

Fig. 2.

Effect of VX-548 on action potentials in human DRG neurons. (A) Action potential evoked by a short (0.5-ms) current injection before and after application of 10 nM VX-548. (B) Phase-plane plot of dV/dt versus V showing modest reduction of maximum upstroke velocity and substantial reduction of peak and shoulder. Upper dashed lines: maximum upstroke velocity in control (black) and with 10 nM VX-548 (red). (C) Collected results for effect of 10 nM VX-548 on action potential threshold determined as in SI Appendix, Fig. S3. (D) Collected results for effect of 10 nM VX-548 on peak of the action potential evoked by a short current injection. (E) Collected results for effect of 10 nM VX-548 on the maximal upstroke velocity of the action potential. (F) Collected results for effect of 10 nM VX-548 on the width of the action potential measured at half-maximum amplitude.

These effects suggest that the initial rising phase of the action potential in human DRG neurons is due mainly to Nav1.7 channels, which activate with smaller depolarizations and more rapidly than Nav1.8 channels, and that Nav1.8 channels contribute to the later phase of the rising phase of the action potential and, because they inactivate more slowly, contribute to the shoulder of the action potential.

Fig. 3 shows the effect of VX-548 when repetitive firing was evoked by a 1-s long current injection. Fig. 3A shows a typical example. In control, the neuron fired repetitively at a frequency of about 16 Hz and firing was maintained in a regular manner throughout the 1-s current injection for moderate current injections. After application of VX-548, there was no change in the early rising phase of the first action potential, but the peak was decreased from +26 mV to +9 mV and the shoulder was narrowed. In the presence of VX-548, subsequent action potentials became progressively smaller until firing stopped before the end of the current injection. These results suggest that Nav1.8 current is more important for repetitive firing during a maintained current injection than for generating the first action potential. Consistent with this, the effect of VX-548 on the peak of the 4th action potential when firing was evoked by a long current injection was much greater than the effect on the first action potential (Fig. 3D).

Fig. 3.

Effect of 10 nM VX-548 on action potentials during repetitive firing. (A) Firing evoked by a 1-s 300-pA current injection before and after application of 10 nM VX-548. (B) First 100 ms on an expanded time scale. (C) Phase-plane plot of dV/dt versus V for the first action potential. (D) Collected results for effect of 10 nM VX-548 on peak of the first action potential (Left) and 4th action potential (Right). Peak of first action potential was +54 ± 2 mV in control and +43 ± 4 mV in VX-548 (mean ± SEM, n = 21; P = 0.0001, two-tailed Wilcoxon signed rank test) with an average decrease of 11 ± 2 mV. Peak of 4th action potential was +47 ± 4 mV in control and +20 ± 4 mV in VX-548 (mean ± SEM, n = 11; P = 0.0036, two-tailed Wilcoxon signed rank test) with an average decrease of 28 ± 3 mV. (E) Collected results for effect of 10 nM VX-548 on width of first action potential, measured at half-maximal amplitude. In 16 of 21 cells, the width decreased with VX-548; in the 5 cells where width at half-maximal amplitude increased, peak was greatly reduced so that half-maximal amplitude occurred much more negative than in control. Average widths were 2.92 ± 0.33 ms in control and 2.66 ± 0.48 ms in VX-548 (mean ± SEM, n = 21; P = 0.197, two-tailed Wilcoxon signed rank test).

Fig. 4 shows an example of the effects of VX-548 on neuronal firing over a range of current injections. The effects of the example in Fig. 4A were typical of most neurons: First, the repetitive firing of the neuron was reduced but not eliminated, and second, with the largest depolarizations, the neuron failed to fire throughout the entire 1-s current injection in the presence of VX-548, going into “depolarization block” earlier than in control. However, the most striking result was that the majority of neurons (14 of 21) still fired multiple action potentials in the presence of 10 nM VX-548 (Fig. 4C and SI Appendix, Fig. S4).

Fig. 4.

Effect of 10 nM VX-548 on maximal repetitive firing. (A) Firing evoked by 1-s injections of current of increasing magnitude in control and after application of 10 nM VX-548. (B) Number of action potentials as a function of the injected current. (C) Collected results for the effect of 10 nM VX-548 on the maximal number of action potentials during 1-s current injections over a range of magnitudes. Dashed line drawn at 1 action potential.

An unexpected observation was that in 5 of the 21 neurons studied with this protocol, the maximum number of spikes fired during the 1-s stimulus (by any size of depolarizing stimulus) actually increased. This seemed related to an effect seen in many neurons: the instantaneous firing frequency of the first few spikes often increased after VX-548 (e.g., Figs. 3A and 6B). A likely explanation of this effect comes from the fact that VX-548 decreased the peak of action potential and usually also narrowed the action potential. These changes in action potential shape would result in less complete activation of voltage-dependent potassium channels during the action potential. This would reduce the potassium conductance underlying the afterhyperpolarization that follows action potentials, which might speed the depolarization toward a second action potential.

Fig. 6.

Effect of 30 nM and 100 nM VX-548 on action potential firing. (A) Example of a neuron testing successive application of 10 nM, 30 nM, and 100 nM VX-548 on firing evoked by a 1-s current pulse. (B) The beginning of the record on a faster time scale, showing the earlier firing of the second spike in 10 nM, 30 nM, and 100 nM VX-548. (C) Number of action potentials evoked by 1-s current injections of different magnitudes in this neuron. Note increased firing for current injections below 500 pA. (D) Collected data for the maximum number of action potentials evoked by a series of 1-s current injections in 8 neurons in which concentrations of 10 nM, 30 nM, and 100 nM VX-548 were applied successively. Dashed line drawn at 1 action potential.

To test this idea, we explored directly whether inhibiting Nav1.8 channels might counterintuitively enhance excitability of the neuron after an action potential. We used a protocol in which an action potential was evoked by a short (0.5 ms) depolarization and a second short depolarization then tested the excitability of the neuron as a function of time after the first action potential, during the afterhyperpolarization. Setting the size of the stimulus to 1.5-times the threshold size, the protocol essentially measures the relative refractory period with this stimulus size. The result was striking: in most neurons, the relative refractory period was reduced by application of VX-548. Fig. 5 shows an example, where in control, the second stimulus evoked an action potential with a spacing of 42 ms but not with 35 ms, while after VX-548, a spacing of 21 ms evoked an action potential. The shortening of the refractory period by 10 nM VX-548 was seen in 11 of 14 neurons tested with this protocol (Fig. 5B). In collected results, the refractory period decreased by an average of 6 ± 2 ms from an average control value of 38 ± 13 ms to 32 ± 14 ms with 10 nM VX-548 (mean ± SEM, n = 14; P = 0.0054, two-tailed Wilcoxon signed rank test).

Fig. 5.

Reduction of refractory period by VX-548. (A) Action potentials were evoked by a pair of 0.5-ms current injections with a variable time between them, with magnitude of both at 1.5-times the threshold current determined in control. The time between the two current injections was varied from longer to shorter to determine the refractory period with this stimulus. The figure shows superimposed sweeps from 8 different sets of times in control (black) and with 10 nM VX-548 (red). In control, the second stimulus evoked an action potential with a spacing (start to start) of 42 ms but not with 35 ms, while after VX-548, a spacing of 21 ms evoked an action potential using the same stimuli. Note faster decay of afterhyperpolarization in VX-548. (B) Collected results in 14 neurons, plotting the maximum spacing between the stimuli (with magnitude set at 1.5-times the threshold in control) in which the second stimulus failed to evoke a spike.

Because of the striking ability of most neurons to retain repetitive firing in a concentration of VX-548 expected to inhibit ~96% of the Nav1.8 current, we wondered how much this depends on the ~4% of the channels expected to remain unblocked in 10 nM VX-548. We tested this by increasing the concentration of VX-548 to 30 nM, and in some neurons, to 100 nM. Consistent with the idea that Nav1.8 current is so large that even 4% of the Nav1.8 channels can contribute to repetitive firing, the maximum number of spikes was generally reduced in 30 nM and 100 nM compared to 10 nM VX-548 (Fig. 6). Yet, 7 of 15 neurons could still fire more than one spike in both 30 nM and 100 nM VX-548 (Fig. 6D and SI Appendix, Fig. S4).

To further explore the contributions of Nav1.8 and other sodium channels to action potentials and repetitive firing, we used the action potential clamp technique, recording repetitive firing in control conditions and then using that waveform of firing as a voltage command with the amplifier switched to voltage clamp mode. Fig. 7 shows a typical example in which VX-548 was applied to identify the magnitude and time-course of Nav1.8 current during repetitive firing. In the example in Fig. 7, which was typical, the first action potential evoked substantial VX-548-sensitive sodium current (10.4 nA) but it was a relatively small fraction of the total sodium current (34.2 nA). During the second action potential, the peak VX-548-sensitive current was actually larger (13.4 nA) than during the first action potential, and it was a larger fraction of the total sodium current (21.5 nA). VX-548-sensitive current during subsequent action potentials were very similar to that evoked by the second action potential.

Fig. 7.

Action potential clamp determination of Nav1.8 and Nav1.7 components of current during repetitive firing. (A) The record of repetitive firing evoked by 1-s current injection was used as command waveform after switching the amplifier to voltage clamp mode and currents evoked by the waveform were recorded in control (black) and after application of 30 nM VX-548 (red). (B) Top, current during 1st, 2nd, and 5th (last) action potentials before and after application of VX-548. Middle, subtraction yielding Nav1.8 (VX-548-sensitive) current. Bottom, subtraction yielding Nav1.7 current by application of 30 nM GsAF-1 in the continuing presence of VX-548. (C) Collected results of peak current sensitive to 30 nM VX-548 during the 1st, 2nd, 3rd, 4th, and last action potentials in 10 neurons. Each cell used a command voltage from its own repetitive firing. (D) Collected results of peak current sensitive to 30 nM GsAF-1 (in the presence of VX-548) in these neurons.

This experiment shows that after the first action potential, Nav1.8 current has a dominant role in generating action potentials during repetitive firing. However, substantial current remained during later action potentials after inhibition of Nav1.8 channels (about 9 nA in the second action potential and about 7 nA in the last action potential for the experiment shown Fig. 7 A and B). To test whether this current comes from Nav1.7 channels, we applied the Nav1.7 inhibitor GsAF-1 (42, 43), in the continuing presence of VX-548. This showed that Nav1.7 current was very large during the first action potential (ranging from 8 nA to 48 nA in 10 neurons tested) and smaller in subsequent action potentials. However, though much smaller in later action potentials than the first, consistent with reduced availability during the maintained depolarization, Nav1.7 current was still > 1nA in the 3^rd action potential in 7 of 10 neurons. These results suggest that Nav1.7 current can contribute to repetitive firing as well as the first action potential and can help account for the incomplete inhibition of repetitive firing by VX-548.

Interestingly, the combination of 30 nM VX-548 and 30 nM GsAF-1 generally left some sodium current unblocked, especially during the first action potential. In experiments on 8 neurons, we added 1 µM tetrodotoxin on top of VX-548 and GsAF-1 and found inhibition of an additional component of sodium current (Fig. 8). This component of current was only substantial during the first action potential and was generally much smaller than either Nav1.7 or Nav1.8 current during the first action potential. Strikingly, however, in 2 of the 8 neurons, this component was actually larger than either Nav1.7 or Nav1.8 current (Fig. 8B). Further experiments will be needed to identify the channels underlying this current.

Fig. 8.

Additional sodium current component with both Nav1.7 and Nav1.8 inhibited. Action potential clamp experiments were performed as in Fig. 7. (A) Example of a neuron in which a large sodium current remained in the first action potential in the presence of 30 nM VX-548 and 30 nM GsAF-1 and was inhibited by application of 1 µM TTX. (B) Collected results for the different components of sodium current evoked during the 1st, 2nd, and last action potentials in 10 neurons. Asterisks indicate neurons in which only VX-548 and GsAF-1 were applied.

In these action potential clamp experiments, the control of voltage is undoubtedly imperfect, because of the very large currents (often greater than 30 nA) which would be expected to produce errors in membrane voltage because of imperfect series resistance compensation. As a result, the magnitude and exact timing of the components of current identified by sensitivity to VX-548, GsAF-1, and tetrodotoxin are imperfectly defined. Nevertheless, the experiments seem adequate to support several key conclusions: that both Nav1.8 and Nav1.7 make major contributions to the first action potential, that Nav1.8 current during action potentials shows little decline during repetitive firing, that Nav1.7 current is much smaller in later action potentials during repetitive firing but can still contribute, and that many neurons have a component of tetrodotoxin-sensitive current in addition to Nav1.7.

Discussion

The general characteristics of action potentials in our recordings were similar to those in previous recordings from human neurons (14, 40, 41, 44–46). In agreement with previous work, virtually all of the neurons we recorded from had obvious shoulders on the falling phase of the action potential, consistent with representing cell bodies of C-type fibers (40, 41, 47). Almost all the neurons (191 of 195) expressed Nav1.8 channels, as evidenced by clear effects of VX-548 on action potential shape or by the kinetics and voltage dependence of sodium currents in voltage clamp experiments (48). The great majority of neurons that survived long enough to be tested produced inward currents in response to capsaicin (141 of 149 neurons tested). These properties suggest that most of the neurons likely correspond to nociceptors (49, 50).

VX-548 is an ideal tool for elucidating the functional role of Nav1.8 channels because of its exceptional potency and selectivity (36, 37). [Interestingly, A-803467, a widely used tool for studying roles of Nav1.8 channels in rat and mouse neurons, was found to be only weakly effective on the TTX-resistant component of current in human DRG neurons (48).] At the level of single action potentials evoked from the resting potential, the effects of VX-548 show that Nav1.8 channels make very little contribution to the sodium current underlying the action potential threshold, usually near −45 mV, which changed only slightly, and also make relatively little contribution to the rising phase up to the maximum upstroke, reached around −10 mV to +20 mV and reduced by only ~15%. However, there is substantial activation of Nav1.8 conductance by the time of the action potential peak, which was typically reduced by 10 to 15 mV. The reduction of action potential amplitude in human neurons by inhibiting Nav1.8 channels agrees with previous results with a different Nav1.8 inhibitor, PF-01247324 (33).

In addition to reducing the action potential peak, VX-548 inhibition of Nav1.8 current almost always narrowed the action potential and reduced the shoulder. The contribution of Nav1.8 channels to the wide shoulder of action potentials in small DRG neurons is consistent with previous observations in rat, mouse, and human DRG neurons (12, 13, 47). The Nav1.8 current during the shoulder could result simply from the relatively slow inactivation of Nav1.8 channels, but it may also represent flow of “resurgent” sodium current that flows when channels pass through an open state during recovery from inactivation (51). In mouse DRG neurons, resurgent current from Nav1.8 channels seems well-suited for helping support the shoulder of the action potential (18–20). Block of resurgent Nav1.8 current could help explain why VX-548 is so effective in narrowing the action potential. However, although greatly reduced, a distinct shoulder was usually still present after block of Nav1.8 channels, most clearly evident by an inflection in the falling phase of the action potential in phase-plane plots (e.g., Figs. 2B and 3C). This is consistent with analysis of action potentials in small capsaicin-sensitive rat DRG neurons, where voltage-dependent calcium current as well as Nav1.8 current contributes to the shoulder (12).

The narrowing of the action potential by VX-548 may contribute to its ability to inhibit pain signaling. If it occurred in nerve terminals of C-fibers in the spinal cord, the reduction of both action potential height and width would reduce calcium entry and therefore transmitter release at the spinal cord synapses, which could combine with reduction of nociceptor excitability to reduce pain signaling. In rats, propagation of action potentials in the dorsal roots of C-fibers is not completely inhibited by tetrodotoxin, suggesting at least some expression of Nav1.8 channels in this region of the neurons (52).

In most neurons, inhibiting Nav1.8 channels reduced but did not completely inhibit repetitive firing evoked by 1-s long current injections. The earlier cessation of repetitive firing when Nav1.8 channels are inhibited by VX-548 is consistent with results in rodent DRG neurons suggesting that Nav1.8 channels are particularly important for supporting repetitive firing during a maintained depolarizing stimulus because they require more depolarized voltages to inactivate than other sodium channels—and thus retain more availability during maintained depolarization—and also because they have faster kinetics of recovery from inactivation at any given voltage (15, 21–23, 53, 54).

Overall, the results suggest that the basic elements of the interplay between Nav1.7 and Nav1.8 kinetics determined by a detailed study using modeling and dynamic clamp in mouse neurons (15) apply to human DRG neurons but also show that human neurons are less dramatically affected by loss of Nav1.8 mediated sodium current than mouse neurons. In human neurons, inhibiting Nav1.8 channels had only small effects on spike threshold or maximum upstroke velocity—in striking contrast to mouse neurons, where Nav1.8 current contributes ~3 fold more current at spike threshold than Nav1.7 current (15) and carries 80% of the current during the upstroke (13), with a similar picture in rat neurons (12). The results can be explained if human neurons have a larger relative density of Nav1.7 channels than mouse or rat neurons, so that the faster activation of Nav1.7 generates the initial upstroke before there is substantial activation of Nav1.8 channels, which then activate to provide additional current by the time of the action potential peak and provide continuing current during the shoulder. In addition, however, there are differences between mouse and human neurons in the kinetics and voltage dependence of both TTX-sensitive and TTX-resistant components of current (48) that will be important to take into account, especially for future modeling studies on human DRGs.

An unexpected observation was that inhibiting Nav1.8 channels generally reduced the refractory period. This was most clearly evident using repeated short stimuli applied from the resting potential (Fig. 5) but the effect was also manifested by more rapid initial firing during long current injections, which was seen with VX-548 inhibition in a substantial fraction of neurons. It seems very likely that the reduction of the refractory period occurs because of the reduction of the peak and narrowing of the action potential when Nav1.8 channels are inhibited. A notable feature of the action potentials in all the neurons we recorded from was a prominent afterhyperpolarization (40, 44, 45), which, as in all neurons, reflects potassium conductances that are activated by either the depolarization of the action potential or entry of calcium during the action potential or both. Activation of either purely voltage-dependent potassium channels or calcium-activated potassium channels would be reduced by a smaller, narrower action potential. This could then shorten the relative refractory period, because the loss of Nav1.8 channels does not much affect the voltage threshold, which depends mainly on other types of sodium channels. The effect of VX-548 to shorten the refractory period is a counterintuitive effect of inhibiting sodium channels that can actually enhance action potential firing under some circumstances.

VX-548 binds to Nav1.8 channels in an unusual state-dependent manner, with tight binding to closed resting channels and apparent unbinding during large depolarizations (37, 55). This state dependence is also seen with A-887826 (56) and another Nav1.8 inhibitor, LTGO-33 (57), and to a weak extent with A-803467 (58, 59). Mutagenesis experiments with LTGO-33 and VX-548 suggest that the compounds bind at a site in the voltage sensor region of domain II (37, 57), suggesting a simple model whereby binding of VX-548 or LTGO-33 stabilizes the voltage sensor in the closed position so that channels cannot open with normal voltage dependence or kinetics but that sufficiently large depolarizations can force movement of the voltage sensor and cause the compound to unbind. In principle, relief of inhibition by depolarization could occur during repetitive action potentials to produce “reverse use dependence”. This does occur for A-887826 (56, 60) but not with VX-548, because unbinding of VX-548 requires larger and longer depolarizations than occur during an action potential (37, 55), even at 37 °C (60) where unbinding is faster and requires less depolarized voltages than at room temperature (55, 60). Consistent with this, in the action potential clamp experiments in Fig. 7, VX-548 inhibited equally well in the later action potentials as in the first, with no evidence of relief of inhibition during repetitive firing.

Overall, our results show that Nav1.8 channels play a major role in facilitating repetitive firing of human DRG neurons, but the most striking result is that many neurons can still fire repetitively even when exposed to concentrations of VX-548 25 to 250 times higher than the K_d. Although further experiments will need to be done to fully explain what combination of sodium channels underlie the repetitive firing remaining in high concentrations of VX-548, it seems likely that the continued firing reflects in part the presence of other types of sodium channels, including Nav1.7 channels and in part the fact that many neurons have such a high expression of Nav1.8 channels that blocking >96% of the channels still leaves enough unblocked Nav1.8 current to contribute to firing. The shortening of the refractory period by VX-548 likely also contributes to maintained repetitive firing with some stimuli.

These results showing that VX-548 reduces but does not eliminate repetitive firing of human DRG neurons has obvious parallels with the results of clinical trials with VX-548 to treat pain, where pain from some conditions is reduced but far from eliminated by VX-548 (36), suggesting that nociceptor excitability is reduced but not blocked with clinical treatment. The results also suggest that human DRG neurons have such high expression of Nav1.8 channels that even with an exceptionally potent inhibitor like VX-548, it may be necessary to achieve levels at the neuronal membrane of greater than 100 times the K_d to eliminate significant contributions of Nav1.8 channel currents to excitability. This may help explain why only the highest doses of VX-548 had clinically significant effects in some conditions (36), despite the extraordinary potency of VX-548 at the channel level (36, 37). The results also show that human DRG neurons have large currents from Nav1.7 channels as well as Nav1.8 channels, so that even total block of Nav1.8 channels does not totally block excitability, as shown by the fact that almost all neurons (14 of 15) could fire at least one spike even in 100 nM VX-548, and some could fire multiple spikes (Fig. 6 and SI Appendix, Fig. S4) - consistent with the action potential clamp experiments showing large Nav1.7 currents during the first spike and smaller but usually nonzero Nav1.7 currents in later spikes. At least at the level of neuronal excitability, these results suggest that there may be an intrinsic limit of efficacy for reducing pain by Nav1.8 inhibition even with 100% target engagement.

Materials and Methods

Human Nav1.8 Cell Line.

A cell line stably expressing human Nav1.8 α- and β3-subunits in Chinese Hamster Ovary cells (B’SYS GmbH) was grown in Ham’s F-12 medium (Corning) containing 10% FBS, 1% penicillin/streptomycin (Sigma), and 3.5 µg/ml Puromycin (Sigma) and 350 µg/ml Hygromycin (Sigma) under 5% CO₂ at 37 °C.

Electrophysiology with Human Nav1.8 Cell Line.

Cells were replated on coverslips for 1 to 6 h before recording. Whole-cell patch-clamp recordings were made with either an Axon Instruments Multiclamp 700B amplifier (Molecular Devices) controlled by pClamp9.2 software (Axon Instruments), filtered at 5 kHz with a low-pass Bessel filter, and digitized at 50 kHz by a Digidata 1322A data acquisition interface or a Sutter Instruments dPatch integrated amplifier/data acquisition system, with filtering at 10 kHz and digitization at 100 kHz.

Patch pipettes had resistances of 1.5 to 2.2 MΩ when filled with the internal solution, consisting of 61 mM CsF, 61 mM CsCl, 9 mM NaCl, 1.8 mM MgCl₂, 1.8 mM EGTA, 14 mM creatine phosphate (tris salt), 4 mM MgATP, and 0.3 mM GTP (tris salt), 9 mM HEPES, pH adjusted to 7.2 with CsOH. The shank of the electrode was wrapped with Parafilm to reduce capacitance and enable partial series resistance compensation (~70%) without oscillation. The whole-cell configuration was established with cells in room-temperature (21 °C) Tyrode’s solution (155 mM NaCl, 3.5 mM KCl, 1.5 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 10 mM glucose, pH adjusted to 7.4 with NaOH). After establishing whole-cell recording, the cell was lifted from the bottom of the recording chamber and placed in front of an array of quartz flow pipes (250 µm internal diameter, 350 µm external, Polymicro Technologies) attached with styrene butadiene glue (Amazing Goop, Eclectic Products) to an aluminum rod (1 × 1 cm) whose temperature was controlled by resistive heating elements and a feedback-controlled temperature controller (TC-344B; Warner Instruments). The external solution for recordings was Tyrode’s solution containing 200 nM tetrodotoxin to inhibit a small endogenous sodium current in the cell line (55) and 1 mg/mL Pluronic PF-68 (Sigma), a surfactant poloxamer without which the apparent potency of VX-548 was found to be reduced (55). Solution changes were made (in <1 s) by moving the cell between adjacent pipes. Experiments were done with temperature controlled to either 21 °C or 37 °C.

The potency of VX-548 determined at room temperature in these recordings (IC₅₀ of 0.07 nM at 21 °C) was higher than in previous studies using automated patch clamp, with values of 0.68 nM (36, 37) and 0.32 nM (55) near room temperature. We used the same hNav1.8 cell line and the same stock solutions of VX-548 as for previous experiments with automated patch clamp (55), so the difference likely reflects the different systems of solution application. In the manual patch clamp system, the cell is exposed to a constant flow of drug-containing solution while with automated patch clamp, solutions are static after application through the microfluidics, which have a high surface-to-volume ratio. Also, the flow pipes for the manual patch clamp experiments are glass (fused silica) which may minimize possible loss of compound occurring by adsorption to plastic.

In comparing potency of VX-548 at 21 and 37 °C, cells studied at the two temperatures were often interleaved on the same day, including the cells whose results are shown in SI Appendix, Fig. S2, leaving no doubt about the lower potency at 37 °C.

Preparation of Human DRG Neurons.

Neurons were obtained from the DRG of human donors. The procurement network of AnaBios Corporation includes only US-based Organ Procurement Organizations and Hospitals. Policies for donor screening and consent are those established by the United Network for Organ Sharing (UNOS). Organizations supplying human tissues to AnaBios follow the standards and procedures established by the US Centers for Disease Control (CDC) and are inspected biannually by the Department of Health and Human Services (DHHS). The distribution of donor medical information is in compliance with HIPAA regulations to protect donor privacy. All transfers of donor tissue to AnaBios are fully traceable and periodically reviewed by US Federal authorities.

Neurons were dissociated from DRG and suspended in medium as previously described (40) and shipped in suspension in a container with temperature maintained at 4 °C. The neurons were then plated on small (5 or 10 mm) round poly-D-lysine coverslips placed in 24- or 48-well plates. Coverslips were prepared by exposure to 70% ethanol and exposed to UV for approximately 15 to 30 min and then incubated overnight at 4 °C with 0.01 mg/mL poly-D-lysine diluted in sterile water. After trituration of the tissue, 10 to 18 µL of the sample was plated on each coverslip and incubated at 37 °C (5% CO₂) for 1 to 2 h to allow the cells to settle and attach to the coverslip. Each well was then gently flooded with 1 mL of culture media consisting of BrainPhys media (Stemcell technologies, Cat. 05709), 1% penicillin/streptomycin, 1% GlutaMAX, 2% NeuroCult SM1 (Stemcell technologies, Cat 05711), 1% N-2 Supplement (Thermo Scientific, Cat # 17502048), and 2% Fetal Bovine Serum. Plates were housed in a 5% CO₂ incubator set at either 37 °C for up to 5 d, with half of the media in each well exchanged every 2 to 3 d.

Electrophysiology with Human DRG Neurons.

For recording, coverslips were placed into the recording chamber containing about 2 mL of Tyrode’s solution at room temperature. We generally selected smaller cells for recording, assuming that many would be C-fiber nociceptors (consistent with 191 of 195 cells having Nav1.8 current and 141 of 149 tested cells having current activated by 1 µM capsaicin) and also selected cells in which glia had partially peeled off, which facilitated recording. SI Appendix, Fig. S1 shows a histogram of cell sizes used for the experiments. Prior to recording, if the cell was strongly adhered to the coverslip (as was usual 2 to 3 d after plating), an electrode with a blunt tip was used to scrape the surrounding area of the cell and gently maneuvered to ensure detachment from the coverslip. Both the “scraping” electrode and recording electrode were pulled from borosilicate capillaries (VWR International, Cat #53432-921) on a Sutter P-97 puller (Sutter Instruments).

We experimented with keeping plates at 30 °C instead of 37 °C with the hope that cells might generate fewer processes and be easier to lift but there was no obvious difference in growth of processes or success with lifting. There was no obvious difference in electrophysiological properties of the cells from plates kept at 30 °C versus those kept at 37 °C.

Whole-cell recordings were obtained using patch pipettes filled with a K-gluconate-based internal solution containing (in mM) 139.5 K-Gluconate, 1.6 MgCl₂, 1 EGTA, 0.09 CaCl₂, 9 HEPES,14 creatine phosphate (Tris salt), 4 MgATP, 0.3 GTP (Tris salt), pH adjusted to 7.2 with KOH. Membrane potentials are corrected for a liquid junction potential of −13 mV between the internal solution and the Tyrode’s solution in which the current was zeroed before recording. Seals were formed and whole-cell configuration was established in Tyrode’s solution at room temperature. Cells were then lifted off the bottom of the recording chamber and placed in front of an array of temperature-controlled quartz flow pipes, with solution maintained at 37 °C.

In current clamp experiments, cells were recorded at their natural resting membrane potential without any injection of holding current. Current steps were applied for 1 s at 0.1 to 0.33 Hz with the increments of the steps adjusted depending on the input resistance of each cell. In the protocols used for experiments like that in Fig. 3, the appropriate amount of current injection to elicit several action potentials (at least 4) was determined from the step protocol. A 1 s current step of this size was applied at 0.1 Hz during the wash-on of VX-548 for 5 min. Before and after the wash-on of VX-548, the full response of the neuron to steps of different sizes was recorded, as in Fig. 4. Action potentials were defined using a criterion of action potential peak >−20 mV and height >40 mV. Capsaicin sensitivity was tested at the end of the experiment by holding the cell at −70 mV in voltage clamp and briefly applying 1 µM capsaicin.

Voltage clamp experiments were carried out with a dPatch amplifier (Sutter Instruments) run by SutterPatch software (Sutter Instruments). Output was low-pass filtered at 10 kHz using the amplifier’s built-in Bessel and digitized at 100 kHz. Electrodes were pulled on a Sutter P-97 puller (Sutter Instruments) and shanks were wrapped with Parafilm (American National Can Company) to allow optimal series resistance compensation without oscillation. Pipette resistances were between 0.8 to 2 MΩ. Series resistance of 1.9 to 4.5 MΩ was estimated from the Sutterpatch whole-cell parameters routine. Series resistance compensation between 60 and 70% was used. For action potential clamp recordings, action potential waveforms were obtained from whole cell current clamp experiments where cells were given increasing 1-second current injection steps from their resting membrane potential to elicit repetitive firing. The action potential waveform used as the voltage clamp command corresponded to the minimum current injection that elicited repetitive firing. After switching the amplifier to voltage clamp mode, currents were then recorded using the neuron’s own action potential waveform. A series of 50-ms long −2 mV square steps from the holding potential were given preceding the action potential waveform and used for offline capacitance and leak correction. Neurons were stimulated with the action potential waveform in control solution (which contained the same DMSO concentration as the drug-containing solutions), then with 30 nM VX-548, then 30 nM VX-548 + 30 nM GsAF-1 and, in some neurons, finally 30 nM VX-548 + 30 nM GsAF-1 + 1 µM TTX. Both current clamp and voltage clamp recordings were made at 37 °C.

Drugs.

The synthesis of the active enantiomer of VX-548 is described by Vaelli et al. (55). Stocks were initially prepared at 10 mM in DMSO and further diluted to 1 or 10 µM in DMSO for the nanomolar concentration experiments. All stocks were aliquoted and frozen at −20 °C. GsAF-1 (42, 43) stock (Alomone Labs, Cat# STG-300) was prepared at 100 µM in purified water and aliquots were stored at −20 °C. TTX stock was prepared at 3 mM in water and aliquots were stored at −20 °C. Capsaicin stock was prepared at 1 mM from powder in DMSO and stored at room temperature. External control solutions contained matching concentrations of DMSO as the drug-containing solutions. All external solutions contained 1 mg/mL Pluronic PF-68 (Sigma). Solutions containing VX-548 or TTX were sonicated for 2 mins while heated to 30 °C.

Data Analysis and Statistics.

Data were analyzed using programs written in Igor Pro 6, 8, or 9 (Wavemetrics, Lake Oswego, OR), using DataAccess (Bruxton Software) to read pClamp files into Igor Pro.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.