3.3 The two recovery processes in binding site mutant channels.
We repeated the same experiments using the RFI and 3PTprotocols in cells expressing binding site mutant (F1579A) channels. We found that both the fast and the slow processes were radically altered by the mutation. In mutant channels fitting the recovery required a biexponential function, the major fast component had a time constant of 0.30 ± 0.02 ms, and it contributed 85.5 ± 2.5% to the amplitude. The time constant of the minor, slow component was 4.36 ± 0.62 ms. Riluzole only slightly delayed fast component, to 0.43 ± 0.02 ms (p = 0.012, paired t-test, n = 7) (Fig. 3). The minor, slow component was not changed significantly.
The effect of the mutation on the fast recovery was also indicated by the decreased inhibition of 2nd pulse-evoked currents in the 3PT protocol (Fig. 4), see the minimal difference between blue and red traces in Fig. 4D. Interestingly, 3rd pulse-evoked currents were still considerably inhibited (by 37.2 ± 6.6%). Apparent affinity (IC50 ) values, calculated from the extent of inhibition, were 1590 ± 180 µM, 895 ± 126 µM, and 114 ± 20.0 µM for 1st, 2nd, and 3rdpulse-evoked currents, respectively. Compared to the apparent affinity values of WT channels, 1st pulse-evoked responses were least sensitive to the mutation (2.62-fold decrease in affinity; p = 0.012, paired t-test, n = 9), while 2nd and 3rd pulse-evoked responses much more sensitive (26.7-fold and 28.6-fold decrease in affinity, respectively; p < 1*10-7 for both).
The slow recovery process was also accelerated, (compare Fig. 4E to Fig. 2F). For a detailed analysis of time constants see supporting information Fig. S1.
3.4 Conformation-selective photolabeling. The problem of interrelated binding and gating .
To summarize, we observed two distinct recovery processes, and they were both accelerated by mutation of the local anesthetic binding site. What physical processes may underlie them?
In the presence of riluzole, the rate of the fast recovery process may be determined either by modulated gating (i.e., recovery from inactivation is slowed down by the bound drug) or by the dissociation of the drug – which is the conventional explanation.
The modulated receptor hypothesis (Hille, 1977) predicts that, since the drug has different affinities to different conformational states, drug binding must alter the energetics of conformational transitions, making higher affinity states more stable. Any mutation that changes binding affinities to different states must also change the effect of the drug on rates of transitions between these states (i.e., the gating of drug-bound channels). In other words, decreased affinity must necessarily cause decreased modulation of gating. Mutation-induced changes in the fast and slow recovery processes, therefore, could both be manifestations of decreased affinity, on the level of gating and binding, respectively. Mutation-induced acceleration of the fast recovery process (compare Fig. 1 and Fig. 3) may be due to decreased modulation of channel gating, while acceleration of the slow recovery process (compare Fig. 2F and 4E) may be caused (at least partly) by faster ligand unbinding. If we accept this explanation, we must suppose that channel gating alone (without dissociation of the inhibitor) can make the channel available for activation and conduction, in other words, drug-bound channels can conduct, therefore ”non-blocking modulation” (Lukacs et al., 2018) is possible.
The complex problem of interrelated gating and binding/unbinding can be simplified using photolabeling-coupled electrophysiology, as we have previously demonstrated (Lukacs et al., 2018). By binding the photoreactive riluzole analog, azido-riluzole, covalently to the channel, we could exclude the processes of binding and unbinding. We found that covalently bound channels were still able to conduct ions, but with modulated gating. In the 3PT protocol modulated gating was reflected as the difference between 1st and 2nd pulse-evoked currents (Fig. 2). This difference almost disappeared when the key residue of the local anesthetic binding site was mutated (Fig. 4). It was logical to investigate if mutation of the same residue would cause the same lack of modulation in photolabeling-coupled electrophysiology experiments. In addition, we also investigated whether the binding was conformation-dependent. We have improved the method of photolabeling-coupled electrophysiology by synchronizing UV light pulses with the voltage protocol, instead of the continuous illumination we had used previously (Lukacs et al., 2018). This allowed us to target specific conformations of the channel, and also to verify that the channel-ligand complex indeed assumes a drug-bound resting conformation before unbinding occurs. By precise timing of the UV light pulse, it was possible to test how soon this unfavorable drug-bound resting conformation was ended by unbinding (see supporting information). These data helped us identify the physical processes underlying both fast and slow recovery processes.