Specificity of M‐channel activators: binding or effect?

The neuronal M‐current is primarily generated by voltage‐dependent heteromeric potassium channels formed by KCNQ2 and KCNQ3 subunits (Grunnet et al. 2014). These KCNQ2/KCNQ3 channels are expressed in the central nervous system where they protect neurons from potentially harmful hyperexcitability by stabilizing the neuronal resting membrane potential. Individuals with inherited mutations that reduce function of KCNQ2/KCNQ3 channels are at higher risk of suffering from epilepsies such as benign familial neonatal seizures. KCNQ2/KCNQ3 channel activators, which would boost the M‐current, are therefore an attractive anti‐epileptic therapy for epilepsy. Retigabine, which was approved for clinical use in 2011, is the first KCNQ2/KCNQ3 channel activator used to treat epilepsy. Unfortunately, because different KCNQ subunits are expressed in different tissues and retigabine shows poor specificity between channels formed by various KCNQ subunits (cardiac KCNQ1 being the exception), patients taking retigabine suffer from side‐effects such as bladder dysfunction (Grunnet et al. 2014). To minimize side‐effects and improve life quality for those treated with KCNQ2/KCNQ3 channel activators there is therefore a need to develop KCNQ channel activators with better subtype specificity.

The study published in this issue of The Journal of Physiology by Wang and coworkers (Wang et al. 2017) brings us one important step closer to understanding how subtype specificity may be achieved. Wang and coworkers compared the effects of the retigabine analogue ICA069673 (called ICA73) on homomeric KCNQ2 and KCNQ3 channels and found that ICA73, in contrast to retigabine, is a potent activator of KCNQ2, but has basically no effect on KCNQ3. By making chimeras between the ICA73‐sensitive KCNQ2 and the ICA73‐insensitive KCNQ3, they identified the voltage‐sensing domain, transmembrane helices S1–S4, as the important structural determinant of subtype specificity of ICA73. The transmembrane helices S3–S4 of KCNQ2 were of particular importance for fully preserving the drug's effects. The pore domain on the other hand, transmembrane helices S5–S6, was not of importance for subtype specificity. They then used a clever rubidium efflux screening assay to pinpoint which residues in the voltage‐sensing domain are critical for ICA73 effects. Two residues in transmembrane helix S3 (F168 and A181 in KCNQ2) were identified as particularly important (Fig. 1). These residues are clearly distinct from the previously identified retigabine binding site in S5 (e.g. Lange et al. 2009; Kim et al. 2015) (Fig. 1).

Figure 1. Schematic illustration of channel structures critical for the activating effects of ICA73 and retigabine.

Middle panel illustrates side view of one KCNQ subunit with transmembrane helices S1–S6. Lower panel illustrates top view of a tetrameric KCNQ channel. Segments/residues important for ICA73 effects are marked in dark red and segments/residues important for retigabine effects in dark blue. Numbering refers to KCNQ2. Wang and coworkers identified S3 residues as important determinants of subtype specificity of ICA73. A181 was essential for the ICA73‐induced shift in channel opening, whereas F168 was important for both the shift and current potentiation. A tryptophan in S5, W236, has previously been identified as critical for retigabine binding (e.g. Lange et al. 2009; Kim et al. 2015). One plausible model for the subtype specificity of ICA73 is that this drug binds to a non‐conserved site in S3 (illustrated in the middle left panel), similar to other KCNQ channel openers that have previously been suggested to bind to the VSD (e.g. Padilla et al. 2009). Another possible scenario is that ICA73 binds to the retigabine binding site in S5, but causes different downstream effects to retigabine via S3 (illustrated in the middle right panel).

Interestingly, Wang and coworkers observed two parallel channel‐activating effects of ICA73 on KCNQ2, which seem to be mediated via distinct mechanisms. The most notable effect of ICA73 was a dramatic shift of the voltage dependence of channel opening towards negative voltages. The side‐chain signature at both identified residues in S3 was essential for this ‘shifting’ effect. The less notable effect of ICA73 was a potentiation of current amplitude at voltages where channel activation had already saturated. For this ‘potentiating’ effect, only the side‐chain signature at the lower S3 residue (F168) was essential. Wang and coworkers therefore speculate that certain residues in the voltage‐sensing domain (e.g. F168) might be important for ICA73 binding while other residues (e.g. A181) might be critical for how the channel responds to drug binding. It is important to note that mutation studies of the channel alone cannot distinguish between binding residues and residues important for effects downstream of drug binding. Experiments providing more evidence of drug binding (like those identifying the retigabine binding site in S5; Kim et al. 2015) are necessary to conclusively identify the binding site for ICA73. So it is possible that retigabine and ICA73 both bind to the same binding site, but induce different downstream effects (Fig. 1).

The overall message from this study is that structural differences in the voltage‐sensing domains of related KCNQ channels may offer the sought‐after subtype specificity of drugs. Thus, retigabine analogues or other activators that target the voltage‐sensing domain of specific KCNQ subunits, or act via different downstream mechanisms, would hopefully replicate the anti‐epileptic effect of retigabine, but result in minimal adverse side‐effects. Still, the findings by Wang et al. trigger a number of interesting questions. For instance, which residues form the binding site of ICA73 and what are the molecular mechanisms underlying the two distinct ICA73 effects? Also, because the related KCNQ5 channel is insensitive to ICA73 (Brueggemann et al. 2014) despite containing the ICA73‐sensitive KCNQ2 side‐chain signature at the critical S3 positions, one might wonder which additional factors determine subtype specificity. These and other questions remain to be answered before we fully understand how to cleverly design improved KCNQ channel activators with subtype specificity.

Additional information

Competing interests

None declared.

Funding

This work was partly funded by grants from CureEpilepsy (414889 M160145 to R.B.S.), The Swedish Society for Medical Research (to S.I.L.), and NIH NIGMS (R01 GM109762 to H.P.L.).