Fig. 4. Shift of the inactivation curve by lidocaine plus an anticonvulsant. (A ) Simplified gating schemes illustrating the possible actions of lidocaine and a coexisting anticonvulsant (phenytoin, carbamazepine, or lamotrigine). R and I represent the resting and inactivated states of the Na+channel, respectively. During depolarization (+V), most channels are moved to the inactivated state. D1 and D2 represent two different drugs, both of which selectively bind to the inactivated state rather than the resting state. State RD, the drug-bound resting states, is, thus, neglected for simplicity. If D1 and D2 share the same binding site in the inactivated channel (top , the “one-site” model), then the channel can be bound by either D1 or D2 but not both. If D1 and D2 have separate binding sites (bottom , the “two-site” model), then the inactivated Na+channels may be significantly occupied by both D1 and D2 (i.e. , state ID1D2), especially when high concentrations of D1 and D2 are present. (B -D ) The experimental data on the inactivation curve shift in different drug combinations and the predictions based on different models in A (n = 3 for each part). The predicted values of shift by the one-site model are 17.8, 14.2, and 17.8 mV, which always fall into the 95% CI of the experimental data (12.2-18.0, 10.9-16.7, and 10.9-18.7 mV, respectively), for 100 μm lidocaine + 100 μm carbamazepine (B ), 100 μm lidocaine + 100 μm lamotrigine (C ), and 100 μm lidocaine + 100 μm phenytoin (D ), respectively. Conversely, the predicted values of shift based on the two-site model are 26.5, 20.8, and 26.8 mV for B , C , and D , respectively, and are all outside the 95% CI of the experimental data.