- Top of page
Lamotrigine (LTG, 3,5-diamino-6-(2,3-dichlorophenyl)-1,2,4-triazine) is a new antiepileptic drug which shows the same efficacy as carbamazepine (CBZ) and diphenylhydantoin (DPH) when used as monotherapy in newly diagnosed epilepsy (Steiner et al., 1994; Brodie et al., 1995). LTD also has a similar antiepileptic profile to CBZ and DPH in animal seizure models (Miller, 1986), and shows a similar use-dependent block of neuronal discharges at the cellular level (McLean & McDonald, 1983; 1986; Lees & Leach, 1993; Xie et al., 1995). The use-dependent block of discharges is of great mechanistic interest as it readily explains why these nonsedative antiepileptics may effectively inhibit seizure discharges, yet spare most normal activities. The molecular basis of this interesting block has been ascribed to the voltage-dependent inhibition of Na+ currents in various preparations by DPH, CBZ and LTG (Matsuki et al., 1984; Willow et al., 1985; Lang et al., 1993; Kuo & Bean, 1994a; Xie et al., 1995). As in the case of many local anaesthetics (for review see Butterworth & Strichartz, 1990), the inhibition of Na+ channels by these antiepileptics is more pronounced at more depolarized potentials, and has been similarly envisaged with the ‘modulated receptor hypothesis’, which proposes that the inactivated state of the channel has a much higher affinity for the drug than the resting state (Hille, 1977; 1993).
However, the steady-state consideration (different binding affinities between the drug and various gating states of the Na+ channel) cannot completely explain why there should be use-dependent inhibition of neuronal discharges. For example, many Na+ channels in a neurone would be transiently inactivated after just one single action potential. Obviously DPH, CBZ and LTG do not significantly bind to and inhibit Na+ channels in this situation, otherwise normal neuronal activities would not be spared with these drugs. Based on the necessity of high-frequency discharges for the inhibition to happen, one may reason that with one single action potential the period of depolarization may be too short for the drug to act on the Na+ channel. Consistent with this idea, it has been directly demonstrated by voltage clamp studies that long (∼ seconds) depolarizations are needed for DPH, CBZ and LTG to exert their inhibitory effect on Na+ currents (Matsuki et al., 1984; Lang et al., 1993; Kuo & Bean, 1994a,b; Xie et al., 1995). Thus probably both the steady-state effect and the kinetics of the development of the effect are important to understand the use-dependent inhibition of Na+ channels and neuronal discharges by these antiepileptics.
Why, then, would long depolarizations be needed for potent block of Na+ currents by these anticonvulsants? There are at least two possibilities. Na+ channels are voltage-gated molecules whose conformational change is controlled by the membrane potential. At hyperpolarized membrane potentials most Na+ channels are in the resting (closed or deactivated) state. Upon depolarization the channel quickly opens (activated) and then is quickly inactivated within a few milliseconds (ms). This is the ‘fast’ inactivated state which has been explained by the ‘ball-and-chain’ model (Armstrong & Bezanilla, 1977; Armstrong, 1981). If the depolarization is sustained for, say, a few seconds, some Na+ channels would be driven into another inactivated conformation, the slow inactivated state (Adelman & Palti, 1969; Schauf et al., 1976; Rudy, 1978). As the development of slow inactivation and the development of the inhibitory effect of antiepileptics on Na+ channels occur on a similar time scale, one possibility is that the drug selectively binds to the slow inactivated state rather than the fast inactivated state. This is what has been proposed for the molecular action of LTG (Xie et al., 1995). On the other hand, the drug may still act on the fast inactivated state of the channel, but the binding rate is slow. A prolonged depolarization to keep the channel in the high-affinity fast inactivated state is therefore necessary for the inhibition to develop. The molecular action of DPH seems to fall into this latter category (Kuo & Bean, 1994a).
To elucidate whether the molecular mechanisms of action are really different between LTG and DPH, we studied the affinity and kinetics of LTG binding to the Na+ channels in mammalian central neurones. We found that similar to DPH, LTG also inhibited Na+ currents by slow binding onto the fast inactivated state of the channel. This process can be characterized by simple one-to-one binding with a rate constant of ∼ 10,000 M−1 s−1 and a dissociation constant of 7–9 μm, well within the clinical therapeutic concentration range of LTG.
- Top of page
LTG, like DPH and CBZ, inhibited repetitive discharges and Na+ currents in a use-dependent manner, which may be closely related to its mechanism of antiepileptic action. We investigated the molecular events underlying such a use-dependent block and found that the affinity of LTG toward the inactivated Na+ channels (1/KI) was more than 200 times higher than its affinity toward the resting (deactivated) channels (1/KR). The findings are quantitatively consistent with a simple scheme that LTG preferentially binds to the inactivated state of Na+ channels via a simple bimolecular, one-to-one binding reaction.
As the inactivated Na+ channels seem to be the major pharmacological ‘target’ for LTG, a key issue in understanding the molecular action of LTG would be to clarify the nature of the high-affinity inactivated state. Xie et al. (1995) proposed that LTG selectively binds to the slow inactivated state of the Na+ channel. This proposal is based on the findings that LTG showed no inhibition of Na+ currents with a train of short-duration pulses (which cause only channel opening and fast inactivation but not slow inactivation), that the inactivation curve was substantially shifted by LTG when the inactivating pulse was long (‘slow inactivation curve’) but only insignificantly shifted when the inactivating pulse was short (‘fast inactivation curve’) and that LTG did not change the decaying kinetics of macroscopic Na+ currents (‘the development of fast inactivation’). However, all these findings can be equally well explained if LTG binds to the fast inactivated state of Na+ channels with slow binding kinetics.
A detailed analysis of the kinetic data in this study revealed interesting information about the inactivated state which LTG binds to. The time constant of recovery from the fast inactivated state is ∼1 ms at –120 mV (Kuo & Bean, 1994b), and recovery from the slow inactivated state at similar hyperpolarizing potentials is very much slower (Kuo & Bean, 1994a). As a first approximation, in a time course of recovery from inactivation one may assume that all the recovery happening within 10 ms is from the fast inactivated state, and all that is happening after 10 ms is from the slow inactivated state. If LTG selectively binds to the slow inactivated conformation, then the recovery in LTG should show a major component which is even slower than the recovery from the drug-free slow inactivated state. However, the experimental findings, shown in Figure 4, did not indicate such a component. On the other hand, the exponential fits to the recovery courses after 10 ms yielded smaller time constants in the presence of LTG (Figure 4). This suggests that most Na+ channels, which are presumably bound by LTG at the end of the long (9 s) prepulse, recover faster than those drug-free slow inactivated channels. The kinetics of recovery thus support a significant binding of LTG to the fast rather than to the slow inactivated state of Na+ channels.
The binding rates of LTG onto Na+ channels are also informative in this regard. If LTG significantly binds to the fast inactivated state of Na+ channels, which develops within a few milliseconds after depolarization, then the requirement of long depolarizations for LTG action most likely indicates slow binding of LTG onto the inactivated Na+ channels. In other words, if the I state in the foregoing scheme represents the fast inactivated state, then the R to I transition is so fast that the overall slow speed of the R to I to ID must be due to a rate-limiting step I to ID. An important connotation from this argument is that if the I to ID transition is accelerated by increasing LTG concentration, the macroscopic rate of LTG action or LTG binding should be correlatively faster. On the other hand, if LTG selectively binds to the slow inactivated state (i.e. that I state in the scheme represents the slow inactivated state), then the macroscopic effect of LTG should not develop faster than the development of the slow inactivated state. Also, the slow R to I step in this case very likely would be partially or even principally responsible for the slow macroscopic action of LTG. The binding rates of LTG thus probably would not be tightly correlated to the LTG concentration, which presumably changes only the rate of the I to ID step. In Figure 5, we demonstrated that in some conditions the macroscopic binding rates of LTG are remarkably faster than the development of the slow inactivated state (e.g. the –10 s−1 binding rate observed in 1 mm LTG), and that the binding rates are linearly correlated to LTG concentrations. These findings again suggest that LTG mainly binds to the fast inactivated Na+ channels with slow binding rates rather than selectively binds to the slow inactivated state. We therefore conclude that the major ‘target’ state of Na+ channels for LTG binding is the fast but not the slow inactivated state, and the slow recovery and binding rates may represent the key molecular events underlying the use-dependent inhibitory effect of LTG on neuronal Na+ currents and discharges.
The steady-state block at a holding potential of –60 mV (Figure 1) and the shift of the inactivation curve with 9 s inactivating pulses (Figures 2 and 3) yielded very similar KI (dissociation constants of LTG binding to the fast inactivated Na+ channels in mammalian hippocampal neurones) of 7–9 μm. The clinical therapeutic concentrations of LTG in plasma are usually 6–40 μm, but sometimes may be as high as 70 μm, especially during combined therapy with valproic acid (Peck, 1991; Kilpatrick et al., 1996). Because 55% of total LTG is bound to plasma proteins, the free concentration (presumably representing the concentration of LTG in the cerebrospinal fluid) is about 3–18 μm, and may be as high as 30 μm. Thus the measured KI (7–9 μm) is well within the clinically relevant concentration range of LTG, and significant block of Na+ current may be expected in appropriate clinical conditions. It should be noted that normally the Na+ current in hippocampal neurones is very large, and inhibiting part of it may not jeopardize the firing of the cell. However, during ictal discharges the prolonged depolarization might already drive a large proportion of Na+ channels into the inactivated state, and a little further reduction of the available Na+ channels may abolish the generation of action potentials. Perhaps the reduction of Na+ currents needed to stop ictal discharges is different in different seizure foci, and this may be part of the reason why the effective concentrations of LTG for satisfactory seizure control vary markedly between patients (Kilpatrick et al., 1996).
The rate constant for LTG binding to the inactivated channel was estimated to be ∼10,000 m−1 s−1 (Figure 5). With the therapeutic free concentrations 3 to 30 μm, LTG would have a macroscopic binding rate (the product of drug concentration and the binding rate constant) of 0.03 to 0.3 s−1. This means that LTG requires a sustained depolarization of the neurone for at least a few seconds to approach significantly its steady-state blocking effect on Na+ channels at room temperature. At body temperature the binding should be faster. Also, as many Na+ channels have been inactivated and become unavailable during ictal depolarization, LTG may not have to reach its steady-state effect to abolish the seizure discharges. It is therefore conceivable that some seizure discharges are probably stopped by LTG before the sustained depolarization has lasted for a few hundred milliseconds. Nevertheless, as such long depolarizations are unusual in normal neuronal activities but are typical of many seizure discharges (for review see Dichter & Ayala, 1987; Lothman & Collins, 1990), the slow binding kinetics of LTG very likely play a key role in its selective effect against seizure discharges without disturbing most normal activities.
It would be interesting to make a comparison between LTG and DPH, the first nonsedative anticonvulsant still widely in use today. Like LTG, DPH also significantly binds to the fast inactivated Na+ channels with slow kinetics via a simple bimolecular reaction (Kuo & Bean, 1994a), and thus also displays use- or voltage-dependent inhibition of Na+ currents. Further comparison reveals that the similarities between these two drugs are not only qualitative but also quantitative. The 7–9 μmKI of LTG is very close to the ∼7 μmKI previously found for DPH. The binding rate constant for LTG binding to the inactivated channel (10,000 m−1 s−1) is also very close to that of DPH, which is around 9,000 to 14,000 m−1 s−1 (Kuo & Bean, 1994a). Thus LTG is very similar to DPH in its pharmacological action on Na+ channels when compared on an equimolar basis. However, the clinical efficacy of the two drugs could still be different because of differences in their therapeutic concentrations. According to a simple one-to-one binding reaction, high therapeutic concentrations (∼20 μm) of LTG (KI 7–9 μm) would exert a steady-state inhibition of the available Na+ channels by 69–74%, whereas high therapeutic concentrations (∼8 μm) of DPH (KI ∼7 μm) would inhibit Na+ currents by 53%. Thus theoretically LTG would have a higher chance of effective seizure control than DPH, at least when adequate doses are used and high therapeutic concentrations are reached. Moreover, LTG, with its very similar binding rate constant to DPH and its high therapeutic concentration ∼3 times higher than DPH, could have a macroscopic binding rate ∼3 times faster than DPH (when both are at their high therapeutic concentrations). This might make LTG more effective than DPH in abolishing ictal discharges characterized by relatively short depolarizations or less frequent repetitive bursts. However, the cost of LTG is much higher than DPH. Also, the decrease of Na+ current necessary for seizure control could be very individual and not necesarily large in a particular patient. Thus it seems reasonable still to try DPH first, and to reserve LTG for those seizures refractory to DPH or other ‘traditional’ anticonvulsants inhibiting Na+ currents in a similar fashion. Combined therapy of LTG with DPH or CBZ may also be advisable in the treatment of refractory seizures, as the binding between these antiepileptics and the inactivated neuronal Na+ channels probably is not saturated in most clinically relevant drug concentrations.
We thank the Wellcome Foundation Ltd. (Kent, U.K., and its branch in Taipei, Taiwan) for providing lamotrigine as a gift. This work was supported by a grant from National Science Council, Taiwan, R.O.C. NSC-86–2314-B-002–195.