Gabapentin Is a Potent Activator of KCNQ3 and KCNQ5 Potassium Channels

Abstract

Synthetic gabapentinoids, exemplified by gapapentin and pregabalin, are in extensive clinical use for indications including epilepsy, neuropathic pain, anxiety, and alcohol withdrawal. Their mechanisms of action are incompletely understood, but are thought to involve inhibition of α₂δ subunit-containing voltage-gated calcium channels. Here, we report that gabapentin is a potent activator of the heteromeric KCNQ2/3 voltage-gated potassium channel, the primary molecular correlate of the neuronal M-current, and also homomeric KCNQ3 and KCNQ5 channels. In contrast, the structurally related gabapentinoid, pregabalin, does not activate KCNQ2/3, and at higher concentrations (≥10 µM) is inhibitory. Gabapentin activation of KCNQ2/3 (EC₅₀ = 4.2 nM) or homomeric KCNQ3* (EC₅₀ = 5.3 nM) channels requires KCNQ3-W265, a conserved tryptophan in KCNQ3 transmembrane segment 5. Homomeric KCNQ2 or KCNQ4 channels are insensitive to gabapentin, whereas KCNQ5 is highly sensitive (EC₅₀ = 1.9 nM). Given the potent effects and the known anticonvulsant, antinociceptive, and anxiolytic effects of M-channel activation, our findings suggest the possibility of an unexpected role for M-channel activation in the mechanism of action of gabapentin.

Fig. 1.

KCNQ3 contains a conserved neurotransmitter binding pocket. (A) Topological representation of KCNQ3 showing two of the four subunits, without domain swapping for clarity. Pentagon, approximate position of KCNQ3-W265; VSD, voltage-sensing domain. (B) Chimeric KCNQ1/KCNQ3structural model (red, KCNQ3-W265). Domain coloring as in (A). (C and D) Close-up side views of KCNQ structure as in (B), showing results of SwissDock.

Fig. 2.

Gabapentin is predicted to bind to KCNQ3-W265. (A) Electrostatic surface potentials (red, electron-dense; blue, electron-poor; green, neutral) and structures calculated and plotted using Jmol. (B and C) Long-range (B) and close-up (C) side views of KCNQ1/3 chimera model structure showing results of SwissDock unguided in silico docking of gabapentin. Domain colors as in Fig. 1.

Fig. 3.

Gabapentin is a potent activator of heteromeric KCNQ2/3 potassium channels. (A) Left, mean two-electrode voltage-clamp (TEVC) traces for KCNQ2/3 expressed in Xenopus oocytes in the absence (control) or presence of 10 nM gabapentin (n = 7 to 8). Dashed line here and throughout, zero current level. Right, voltage protocol. (B) Mean tail current and normalized tail currents (G/G_max) vs. prepulse voltage relationships recorded by TEVC in Xenopus oocytes expressing KCNQ2/3 channels in the absence (black) or presence (red) of 10 nM or 1 µM gabapentin as indicated (n = 7 to 8). Error bars indicate S.D. Voltage protocol as in (A). (C) Mean TEVC traces for KCNQ2/3 expressed in Xenopus oocytes in the absence (control) or presence of 1 µM pregabalin (n = 7 to 8). Lower inset, voltage protocol. (D) Mean tail current and normalized tail currents (G/G_max) vs. prepulse voltage relationships recorded by TEVC in Xenopus oocytes expressing KCNQ2/3 channels in the absence (black) or presence (blue) of 1 µM pregabalin as indicated (n = 5). Error bars indicate S.D. Voltage protocol as in (C). (E) Voltage dependence of KCNQ2/3 current fold increase by gabapentin vs. pregabalin (10 nM), plotted from traces as in (A and C) (n = 5–8). Error bars indicate S.D. *P < 0.05 vs. pregabalin current at −60 mV. (F) Gabapentin and pregabalin dose responses at −60 mV for KCNQ2/3 activation, quantified from data as in (A–E) (n = 7 to 8). Error bars indicate S.D. (G) Dose response for gabapentin and pregabalin effects on resting membrane potential (E_M) of unclamped oocytes expressing KCNQ2/3 (n = 7 to 8). Error bars indicate S.D. (H) Mean tail current vs. prepulse voltage relationships recorded by TEVC in Xenopus oocytes expressing KCNQ2/3 channels in the absence (black) or presence (green) of 30 µM retigabine as indicated (n = 4). Error bars indicate S.D. Voltage protocol as in (A). (I) Voltage dependence of KCNQ2/3 current fold increase by retigabine (30 µM), n = 4). Error bars indicate S.D. (J) Retigabine dose responses at −60 mV for KCNQ2/3 activation, quantified from data as in (A–E) (n = 4). Ctrl, control. Error bars indicate S.D.

Fig. 4.

Gabapentin-activated current is XE991 sensitive and exhibits altered gating kinetics. (A) Exemplar −60 mV KCNQ2/3 current before (left, black), during wash-in of gabapentin (red) and partial washout with bath solution in the absence of drug (black), and then wash-in of XE991 (blue). (B and C) Mean activation at +40 mV (B) and deactivation at −80 mV (C) rates for KCNQ2/3 before (control) and after wash-in of 1 µM gabapentin (GABAP) (n = 7). Activation rate was quantified using the voltage protocol as in Fig. 3A. Deactivation rate was quantified using the voltage protocol shown here. Error bars indicate S.D.

Fig. 5.

Gabapentin is a potent activator of homomeric KCNQ3 and KCNQ5 potassium channels. (A) Mean two-electrode voltage-clamp (TEVC) traces for homomeric KCNQ2, 3*, 4, or 5 channels (as indicated) expressed in Xenopus oocytes in the absence (control) or presence of 1 µM gabapentin (n = 4–8). Voltage protocol, upper inset. (B) Mean tail current (left) and normalized tail currents (G/G_max; right) vs. prepulse voltage relationships recorded by TEVC in Xenopus oocytes expressing homomeric KCNQ2, 3*, 4, or 5 channels (as indicated) in the absence (black) or presence (red) of 1 µM gabapentin as indicated (n = 4–8). Error bars indicate S.D. (C) Voltage dependence of current fold increase by gabapentin (1 µM) for homomeric KCNQ2, 3*, 4, or 5 channels, plotted from traces as in (A) (n = 4–8). Error bars indicate S.D. (D) Gabapentin dose responses at −60 mV for homomeric KCNQ2, 3*, 4, or 5 channel activation, quantified from data as in (A) (n = 4–8). Ctrl, control. Error bars indicate S.D.

Fig. 6.

Gabapentin activation of KCNQ2/3 requires KCNQ3-W265. (A–C) Two-electrode voltage-clamp (TEVC) of water-injected Xenopus laevis oocytes showing no effect of gabapentin (10 nM) on endogenous currents or membrane potential (E_M) (n = 5). (A) Mean traces; (B) mean peak current; (C) mean E_M, in the absence (control) or presence of 10 nM gabapentin. Voltage protocol (A, upper inset). Error bars indicate S.D. (D and E) TEVC of Xenopus laevis oocytes showing effects of gabapentin (10 nM) on heteromeric KCNQ2/KCNQ3-W265L channels. (D) Mean traces; (E) mean tail current (left) and mean normalized tail current (G/G_max; right). n = 5. Error bars indicate S.D. (F and G) TEVC of Xenopus laevis oocytes showing effects of gabapentin (10 nM) on heteromeric KCNQ2-W236L/KCNQ3-W265L channels. (F) Mean traces; (G) mean tail current (left) and mean normalized tail current (G/G_max; right); n = 5. Error bars indicate S.D. (H) Mean tail current fold changes vs. prepulse voltages for channels as indicated; KCNQ2/KCNQ3 results (black line) from Fig. 3E shown for comparison; n = 5. Error bars indicate S.D. *P < 0.05 vs. other groups at −60 mV. (I) Mean dose responses for channels as indicated; KCNQ2/KCNQ3 results (black line) from Fig. 3F shown for comparison; n = 5. Error bars indicate S.D.