4. DISCUSSION
Our data suggest that riluzole is the first member of an entirely new class of sodium channel inhibitor compounds. Its primary binding site is the well-characterized ”local anesthetic receptor,” with F1579 as the key residue. However, the location of the bound ligand is different from that of local anesthetics, and the effect exerted by it is also radically different. As for the location, in silico docking revealed that unlike other sodium channel inhibitor molecules, riluzole was preferentially located within the fenestrations, and tended to avoid the central part of the pore or the vicinity of the selectivity filter, where it could interfere with conduction. Localization within the fenestration has been previously observed in molecular dynamics simulations investigating the general anesthetic isoflurane (Raju et al., 2013) and propofol (Wang et al., 2018), as well as the local anesthetic benzocaine (Boiteux et al., 2014; Buyan et al., 2018; Martin & Corry, 2014) and lidocaine (neutral form) (Nguyen et al., 2019), but it has never been found to be the dominant position. The unusual location predicted by in silico docking is supported by the way riluzole acted in experiments. It stabilized inactivated conformation without interference with conduction. Inactivated state stabilization affected both the equilibrium and the recovery rate, as evidenced by the shift of the SSI and the RFI curves, respectively (Fig. 1; see also Fig. 1 in (Lukacs et al., 2018)).
The non-instantaneous onset (as seen in the SDO protocol) and the delayed recovery (see the RFI protocol) together ensure that effective inhibition is confined within a strict temporal window. Single pulse-evoked currents or synaptic activity-evoked action potentials are ”missed” by riluzole because inhibition becomes effective only after the channel has reached inactivated conformation, and the conformation of the ligand-channel complex has been stabilized. After this, the inhibition is effective, but only for a limited time – even in the case of the non-therapeutic concentration of 100 µM it was only effective for approximately 10-20 ms. At therapeutic concentrations the effect of riluzole is no more than a fine modulation: the refractory time after an action potential is prolonged by a few milliseconds, and a small fraction of the channel population is kept in inactivated conformation. The exact extent of prolongation and the magnitude of the fraction kept in inactivated conformation is dependent on the membrane potential during the interspike interval. This temporal window of effectiveness is the basis of both persistent (or ”late”) sodium current selectivity and the low pass filtering effect. The contribution of the persistent component of sodium currents (INaP) is significant during gradual depolarizations preceding action potentials, especially during burst firing (Del Negro et al., 2010; Taddese & Bean, 2002). In these cases, there is enough time for the onset of inhibition, and the inhibition remains effective until the membrane becomes hyperpolarized for a sufficient time.
An upregulation of the persistent sodium current has been shown to be involved in a number of pathologies, including epilepsies (Oyrer et al., 2018; Stafstrom, 2007), cardiac pathologies (Antzelevitch et al., 2014; Chadda et al., 2017; Makielski, 2016), neurodegenerative disorders (van Zundert et al., 2012; Waxman, 2008), pain disorders (Lampert et al., 2006; Misawa et al., 2009; Xie et al., 2011), even cancer metastasis (Djamgoz & Onkal, 2013), and type II diabetes (Rizzetto et al., 2015). Selective inhibitors of INaP, including riluzole, have been found to be effective in cardiac diseases (Antzelevitch et al., 2014; Belardinelli et al., 2013; Karagueuzian et al., 2017), epilepsies (Anderson et al., 2014), pain syndromes (Xie et al., 2011), and neurodegenerative diseases. Riluzole was especially effective in preventing damage in spinal cord and peripheral nerve injury(Fehlings et al., 2016; Ghayour et al., 2017; Gloviczki et al., 2017).
Riluzole represents a new class among INaP selective inhibitors because the non-blocking modulation mechanism endows it with an ”ultrafast” offset kinetics, which, however, is not based on actual dissociation. A similar mechanism has been assumed for inhibition by riluzole in a recent computational study (Phillips & Rubin, 2019). To illustrate how this compares to offset kinetics of other well-known sodium channel inhibitors, we re-plotted the data from a comparative study of nine inhibitor compounds (El-Bizri et al., 2018), which included six antiarrhythmic drugs, and the three best known INaP selective inhibitors. We supplemented these data with our results on the effects of riluzole (Fig. 8). Binding rate constants were plotted against unbinding rate constants. The latter were calculated from the time constants of the offset (koff = 1 / τoff ), then the former was calculated from koff and theIC50 value (kon = koff / IC50 ). We calculatedkon and koff values the same way for riluzole using time constants of both the fast and slow recovery processes (τfast = 2.25 ms, and τslow = 823 ms), and the IC50value 3.98 µM. As for the slow recovery process (which reflects true dissociation) riluzole did not differ from the other compounds, but when we used data from fast recovery process, it was in a completely different range. It seems that by staying in the fenestration, riluzole can ”pretend” to have been dissociated, which practically increases its speed 366-fold. This allows it to be effective in frequency ranges that are unavailable for conventional drugs. If there was genuine dissociation during the fast recovery process, the value ofkon would be 112 s-1µM-1. This would require an onset time constant of 0.086 ms at 100 µM concentration (from the equationτon = 1/(cc *kon +koff )), which is clearly much faster than the experimentally observed onset.
In summary, our data indicate that riluzole is unique in being able to combine fast onset kinetics with high affinity to the conventional local anesthetic binding site of sodium channels. This high-affinity binding occurs only in depolarized (open or inactivated) conformations. The binding itself does not prevent conduction, only slightly alters the gating of channels, making them more likely to stay in inactivated conformation. This produces a preferential inhibition of i) INaP, ii) cells with depolarized membrane potential, and iii) cells firing at high frequencies. The combination of increased persistent component, compromised ability to keep resting membrane potential, and abnormally high firing frequency almost always signifies pathology, which is the reason why riluzole can selectively inhibit this type of pathological activity. The basis of this selectivity is the non-blocking modulation mechanism. We anticipate that compounds of this class (fast-acting non-blocking modulators) may be favorable for the treatment of a number of pathologies, such as neuropathic pain syndromes, certain neuromuscular diseases, epilepsies, and cardiac fibrillations, which are characterized by an enlarged persistent sodium current and high-frequency firing.