Sodium Channels and Neuronal Hyperexcitability

Na + CHANNELS AND EPILEPSY

183

of currents. This shows up as a null opening for a series of sweeps looking at single channels. This is consistent with the general idea that Ted Cummins suggested that these might be modulations rather than occlusions. In fact, one sees blockade for very brief depolarizations which really only permit open channels but not inactivated ones to be around for any length of time. The dissociation and subsequent availability data from prepulses indicate that the channels are now behaving as if they are in the inactivated state. You don’t need an inactivated channel to get an ‘inactivated’-type stabilization . you can get that through an open channel blockade. It seems that they all degenerate down to that common state.

Catterall: One system that we have studied, the type 2 channel, which is a fastgating, fast-inactivated brain channel, has a persistent current of about 2% or so when it is transfected into mammalian cells. This can be dramatically increased by co-transfecting G protein bc subunits, creating a situation with 20% persistent current on average. So with one (admittedly unphysiological) manipulation, you can generate huge persistent currents from a Na+ channel that we know is present in all of the cells that generate them. We haven’t been able to show that activation of G protein-coupled receptors in a neuron, or even in a transfected cell, can reproduce this e¡ect. So far this is an e¡ect of overexpression of Gbc. Nevertheless, I think that this is a likely mechanism in neurons. We just don’t know how to stimulate the neurons to get this kind of e¡ect. Certainly, in this case you can modulate the level of persistent current in transfected cells with an intracellular transduction protein.

Raman: When you are calling it persistent current, do you mean that there is some late steady state current with some step depolarization to some potential?

Catterall: The current is sustained as long as you depolarize.

Raman: And it will not inactivate if you depolarize to a more positive value? Catterall: You can inactivate all of the channels if you predepolarize for a very

long time. In that case you have no transient current left, and hence no persistent current.

Raman: So you would de¢ne ‘persistent current’ as a component of the current that passes through a channel that is capable of inactivation at a more positive potential.

Catterall: Although we haven’t done much single channel recording, we think it’s a gating mode change of the kind that Wayne Crill described in cortical neurons and other people have described elsewhere. The only di¡erence in these experiments is that we have been able to make a manipulation that greatly changes the amount of persistent current. We think somehow that the G protein puts the channel in a state in which it doesn’t inactivate, and we suspect this is due to direct binding of the G protein bg subunits to the channel.

From the mutagenesis experiments I will discuss in my paper, the binding of the drugs that we think are anticonvulsants seems to be a¡ected by the same

184 DISCUSSION

mutations in the same way as the binding of local anaesthetics. We think the same receptor site is involved in binding the anticonvulsant and local anaesthetic drugs and although binding there may also inhibit gating currents, as Gary Strichartz mentioned earlier, this drug binding probably also blocks the current.

Cummins: This doesn’t rule out the possibility that phenytoin might also have an e¡ect somewhere else as a mode shifter.

Catterall: It doesn’t rule that out, but it does say that at the concentrations at which it is anticonvulsant it is binding to the site that blocks Na+ channels.

Crill: With the great diversity if channels, one can do paper physiology and get bursts many di¡erent ways. It is quite possible that the anticonvulsants might a¡ect the Na+ channel currents and have an e¡ect by decreasing excitability, but that doesn’t say much about what is causing the epilepsy.

Segal: Although there is a type of epilepsy in humans caused by a Na+ channel mutation (Wallace et al 1998) it is clear that Na+ channel mutations will account for only a small fraction of epilepsy. Of the causes of epilepsy that are understood at the genetic level, most are due to cell migrational problems, defects in energy metabolism or accumulation of material that disrupts neuronal function. However, the situation is reversed when we look at epilepsy treatment instead of epilepsy causes. Most of the antiepileptic drugs have their major action at Na+ channels (Rogawski & Porter 1990). Our trial and error discovery of anticonvulsant drugs has arrived at the Na+ channel again and again since the Na+ current is one of the ¢nal common pathways in expressing the seizure phenotype. As we learn more about the ¢nal common pathways such as the Na+ channel it will be easier to work back and look at preceding steps involving protein kinases and other intracellular modulators.

Ptacek: I have always been struck by the fact that most anticonvulsants (perhaps all) are dirty drugs. The way I have been thinking about the tight regulation of neuronal excitability is that even though there are variations, if you have enough variables, the ¢t of net excitability around the mean will be tight. Perhaps the same is true when we are talking about anticonvulsants. It is aesthetically pleasing to think we could have a very selective drug, but perhaps this would have too strong an e¡ect and toxic consequences.

Segal: One of the fascinating things about phenytoin is that it is not just a weaker TTX: it acts more on the persistent current and acts more with bigger depolarizations. One of the NMDA antagonists, memantine, also has similar properties: it hardly blocks at the beginning of NMDA application, but has increasing e¡ect with increasing stimulation (Chen et al 1998). It may be that the best drugs are ones that have these moderate e¡ects on initial currents and then become stronger as the pathological activity continues.

Noebels: Relevant to that, Ken Courtney was the ¢rst to describe this frequencydependent reduction of phenytoin on the action potential. Later, Bob MacDonald

Na + CHANNELS AND EPILEPSY

185

showed that this might even be a general property of many e¡ective anticonvulsants. I thought I understood from your presentation that you thought this was an artefact of the assay.

Segal: MacDonald’s ‘sustained repetitive ¢ring’ model shows a use-dependent e¡ect of phenytoin on the Na+ channel (McLean & MacDonald 1986) as well as a use-dependent e¡ect of memantine (McLean 1987, Netzer & Bigalke 1990). The memantine e¡ect turns out to be on the NMDA receptor (Chen & Lipton 1997) and not on the Na+ channel (Lundquist 1999). I do not think that the e¡ects uncovered by the sustained repetitive ¢ring model are artefacts. The point I am making is that one cannot conclude that drugs that reduce sustained repetitive ¢ring are drugs that have e¡ects on Na+ channels rather than on glutamate receptors.

Noebels: Could you describe the model?

Segal: The way the sustained repetitive ¢ring model operates is that they put an electrode into a cell, depolarize the cell and it ¢res a lot. This is done with and without an anticonvulsant. By the fact that they are depolarizing the cell by current injection, they are getting a depolarization that in many senses is similar to the ones that we are studying as these plateau depolarizations. In contrast, we are looking at the action potentials at resting potential, and it seems that there is virtually no e¡ect of phenytoin, as you would expect from a situation where you are dealing with an action potential that has a huge safety factor. When we study plateau depolarizations, there is no safety factor for the plateau because there is a close balance between Na+ and K+ currents. Any slight change is going to push the voltage one way or another.

Bean: When you see spontaneous activity after putting in the synaptic blockers, do you generally see this long-lasting plateau, even after the cell stops spiking? The plateau was clearly present in some recordings. In other cases, the cell spiked, and once it stopped spiking the voltage came down again. The plateaus that outlast the spiking seem really interesting. What I wonder is, where are the K+ channels in these cells, and are these important in regulating that activity? Intuitively, I would think that the size of the K+ currents in the hippocampal neurons is huge. This also raises the issue of the role of the K+ channels in epileptic activity in vivo. How often do you see that plateau outlasting the spiking?

Segal: The plateau outlasted the spiking about half the time. This impressed us very much for the reason you mentioned . this suggested an equal balance between Na+ currents and K+ currents that we knew could be large. This is what led us to consider the possibility that persistent Na+ currents were large, something that seemed hard to believe 10 years ago.

Bean: In most of the experiments you seemed to be using zero Ca2 + and high Mg2 + to completely block transmission. Is it the same if you just put on APV and CNQX to block transmission but leave Ca2 + at normal levels?

186	DISCUSSION

Segal: Presumably by blocking the Ca2 + current we are also blocking the Ca2 + - modulated K+ , which is probably one of the major inhibitory currents that would be occurring in vivo. But, on the other hand, in vivo you would also be having another excitatory stimulus from glutamate excitation. In our model we are knocking out one of the endogenous inhibitory actions, as well as knocking out the neurotransmitter excitation. It is probably balancing out, and we are getting some re£ection of the in vivo voltages that would occur in such an event, but by having knocked out one excitatory and one inhibitory current we are looking at it in a simpli¢ed situation.

Bean: What happens if you put on APV and CNQX in normal Ca2 + and Mg2 + ? Segal: I don’t remember any plateau depolarizations under these conditions, but we only did a few experiments of this sort before switching to using the synaptic

blocking solution with zero Ca2 + . Waxman: What ends the plateau?

Segal: You can end it by injecting current, but in our experiments typically we would allow it to end spontaneously, usually in several seconds.

Waxman: Do you have a sense of the events underlying this?

Segal: One would expect that the Na+ current would decrease over time. One can’t blame this on Ca2 + accumulation. There might be other mechanisms such as peptide secretion.

Catterall: I wanted to follow up the question that Je¡ Noebels asked, which is what is the hallmark of an anticonvulsant drug as opposed to some other kind of Na+ -blocking drug? It seemed to me from our own work and reading the literature that it is preferential inactivated state block or voltage-dependent block, which is common for phenytoin, carbamazepene and lamotrigine. For example, anti-arrhythmic drugs mostly don’t behave that way: they are frequency-dependent blockers. In your paradigm, do you get the same selective block of the late channel openings with carbamazepine and lamotrigine?

Segal: We tried lamotrigine, but our experiments were complicated by a DMSO solvent artefact that made it hard to get a good baseline (Douglas 1996). At some point we have to go back and do this with the newer water-soluble form of lamotrigine. We haven’t tried carbamazepine. These are incredibly tedious experiments. Holding a patch over a period of 100 min is not the sort of thing you can do every day, and each experiment takes a long time to analyse.

Catterall: You could probably do this with macroscopic currents. Knowing the answer for phenytoin, you could see what happens with macroscopic currents.

Segal: We did our studies using single channel recording in part because we worried that any persistent inward current that we would see by macroscopic current recording could be dismissed as an artefact of poor clamp control of the neuron. However, Chao & Alzheimer (1995) did such a study of macroscopic currents while we were doing our single-channel studies and they found a similar

Na + CHANNELS AND EPILEPSY

187

result to ours. But it is hard to know how much of their quantitative result was due to direct e¡ects on Na+ channels or better voltage control after blockade of Na+ channels. There are real concerns about voltage control because of the demonstration of regenerative Na+ channel potentials in dendrites (Stuart & Sakmann 1994).

Ptacek: I presume the set of 1000 genes involved in causing epilepsy includes metabolic disorders. I’m not sure that it matters from a therapeutic point of view, but in terms of getting at the molecular mechanisms, there are two fundamentally di¡erent types of epilepsy. First, there are the asymptomatic ones where everything seems to be normal except that the membrane excitability is teetering and you get pushed over the edge. These epilepsies are much more likely to be due to variations in ion channels. However, if you ¢ll cells up with sphingomyelin or other compounds that can’t be processed normally because of an enzyme de¢ciency, or there is head trauma, developmental disruption or a brain tumour, I can imagine how these could give epilepsy, but those kinds of problems seem fundamentally di¡erent.

Segal: Since epilepsy is a disease of pathological excitation of neurons it is useful to consider the six ¢nal common pathways for making a neuron more excitable: changes in endogenous excitatory currents, endogenous inhibitory currents, neurotransmitter excitatory currents, neurotransmitter inhibitory currents, gap junctions and ion pumps. Changes in one of these six ¢nal common pathways could produce a simple form of epilepsy without other e¡ects on neurons. However, a simple change such as an increase in an ionic current could also have more complicated e¡ects if this produces changes in cell migration or cell survival. So the boundary between simple ¢nal common pathway, epilepsy, and complicated epilepsy may turn out to be very fuzzy. Even causes of epilepsy such as energy metabolism defects that clearly are not acting directly on the six ¢nal common pathways may turn out to have simple e¡ects mediated by their e¡ects on pumps and channels. I agree that there will be some distinction between mutations acting directly on the six ¢nal common pathways and those that do not, but the distinctions are not likely to be very crisp.

References

Baker MD, Bostock H 1998 Inactivation of macroscopic late Na + current and characteristics of unitary late Na + currents in sensory neurons. J Neurophysiol 80:2538^2549

Baker MD, Bostock H 1999 The pH dependence of late sodium current in large sensory neurones. Neuroscience 92:1119^1130

Chao TI, Alzheimer C 1995 E¡ects of phenytoin on the persistent Na + current of mammalian CNS neurons. NeuroReport 6:1778^1780

Chen HS, Lipton SA 1997 Mechanism of memantine block of NMDA-activated channels in rat retinal ganglion cells: uncompetitive antagonism. J Physiol 499:27^46

188	DISCUSSION

Chen, HS, Wang YF, Rayudu PV et al 1998 Neuroprotective concentrations of the N-methyl-D- aspartate open-channel blocker memantine are e¡ective without cytoplasmic vacuolation following post-ischemic administration and do not block maze learning or long-term potentiation. Neuroscience 86:1121^1132

Douglas AF 1996 The e¡ects of the anticonvulsant lamotrigine on the sodium channel activity of rat hippocampal neurons in microculture. Undergraduate Thesis, Harvard University, Cambridge, MA, USA

Lundquist ER 1999 Memantine as a possible anti-epileptic drug: examining its mechanism of action using models that test sodium current. Undergraduate Thesis, Harvard University, Cambridge, MA, USA

McLean MJ 1987 In vitro electrophysiological evidence predicting anticonvulsant e⁄cacy of memantine and £unarizine. Pol J Pharmacol Pharm 39:513^525

McLean MJ, MacDonald RL 1986 Carbamazepine and 10,11-epoxycarbamazepine produce useand voltage-dependent limitations of rapidly ¢ring action potentials of mouse central neurons in cell culture. J Pharm Exp Ther 238:727^738

Moorman JR, Kirsch GE, VanDongen AMJ, Joho RH, Brown AM 1990 Fast and slow gating of sodium channels encoded by a single mRNA. Neuron 4:243^252

Netzer R, Bigalke H 1990 Memantine reduces repetitive action potential ¢ring in spinal cord nerve cell cultures. Eur J Pharmacol 186:149^155

Rogawski MA, Porter RJ 1990 Antiepileptic drugs: pharmacological mechanisms and clinical e⁄cacy with consideration of promising developmental stage compounds. Pharmacol Rev 42:223^286

Stuart GJ, Sakmann B 1994 Active propagation of somatic action potentials into neocortical pyramidal cell dendrites. Nature 367:69^72

Wallace RH, Wang DW, Singh R et al 1998 Febrile seizures and generalized epilepsy associated with a mutation in the Na + -channel b1 subunit gene SCN1B. Nat Genet 19:366^370

Weiss RE, Roberts WM, Stˇhmer W, Almers W 1986 Mobility of voltage-dependent ion channels and lectin receptors in the sarcolemma of frog skeletal muscle. J Gen Physiol 87:955^983

Sodium Channels and Neuronal Hyperexcitability.

Novartis 241

Copyright & 2002 JohnWiley & Sons Ltd

Print ISBN 0-471-48530-6 Online ISBN 0-470-84668-2

Therapeutic concentrations of local anaesthetics unveil the

potential role of sodium channels in neuropathic pain

Gary R. Strichartz, Zhongren Zhou, Catherine Sinnott and Alla Khodorova

Pain Research Center, Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115-6110, USA

Abstract. Neuropathic pain is frequently associated with hyperexcitability of primary a¡erents, characterized by spontaneous impulses and repetitive ¢ring. Electrophysiology and molecular biology reveal changes in dorsal root ganglion Na + channels under conditions of neuropathic pain, but the manner by which these changes alter the physiology of sensory a¡erents remains unknown. Equally mysterious is the mechanism by which i.v. local anaesthetic-like Na + channel blockers suppress neuropathic pain behaviour at concentrations well below those reported for channel inhibition. We have compared the anti-allodynic actions of i.v. lidocaine (L) and stereoisomers of mexiletine (R-M, S-M), in rats after spinal nerve ligation, with their ability: (1) to inhibit fast, tetrodotoxin-sensitive neuronal Na + currents, elicited by brief (1 ms) pulses, at 10 Hz, from ‘resting’ potentials (780, 760 mV) and (2) to suppress the seconds long plateau and the repetitive ¢ring produced in axons by slowing of Na + channel inactivation (e.g. using scorpion a-toxins). Both L and R-M at 5^10 mM relieved allodynia; S-M was ine¡ective. Na+ currents also were inhibited by M, with a⁄nities that were increased by both repetitive ‘¢ring’ (KR,S = 5 mM) and depolarization of the ‘resting’ membrane (KR = 15 mM; KS = 30 mM). Stereopotency ratios depended on the manner in which di¡erent states of the channel were inducted. Both L and M shortened the action potential’s ‘plateau’ in a-toxin treated axons, without reducing the spike, and suppressed repetitive ¢ring with IC50s = 5 mM, and no stereoselectivity. These ¢ndings together demonstrate that Na+ channel blockers, at ‘therapeutic’ concentrations, can inhibit neuronal hyperexcitability.

2002 Sodium channels and neuronal hyperexcitability. Wiley, Chichester (Novartis Foundation Symposium 241) p 189^205

Traditional local anaesthetics (lidocaine), antiarrhythmics (mexiletine) and anticonvulsants (phenytoin), antipsychotics (carbamazepine) and anti-depressants (amitriptyline) have all been used clinically, with substantial interpatient

189

190	STICHARTZ ET AL

variability, to treat neuropathic pain (Ackerman et al 1991, Boas et al 1982, Chabal et al 1992). One common action of all of these diverse drugs is their ability to inhibit voltage-gated Na + channels (Tanelian &Victory 1995, Mao & Chen 2000). Many amphipathic amines, with a modicum of structural similarity and, at best, modest speci¢city, are nevertheless able to inhibit these channels, whose activity is essential for almost all membrane excitability. The question remains, however, as to whether this inhibitory action is the one most salient for neuropathic pain relief, since all of these drugs are known to have several other direct e¡ects in the peripheral and central nervous systems.

The pathophysiological mechanism(s) for neuropathic pain is not known. Investigators have regularly detected abnormal, additional impulse activity in injured, as well as in non-injured peripheral nerves from animals with induced neuropathic pain (Kajander & Bennet 1992, Chabal et al 1989, Devor et al 1992, Study & Kral 1996, Ali et al 1999). Such ‘hyperexcitability’ has often been ascribed to a change in voltage-gated Na + channels (Matzner & Devor 1994), although it is unclear whether the salient change is an overall increase in channel density, a spatial redistribution of normal channels (Novakovic et al 1998), a shift in the expression pro¢le of channel isoforms with di¡erent gating kinetics (Cummins et al 1999), or changes in other currents, for example K + , that also modulate excitability.

‘Hyperexcitability’ in nerve injury models manifests as a reduction in electrical stimulation threshold, often to the point of generating (apparently) spontaneous impulses, and the appearance of high frequency bursts of impulses (Chabal et al 1992, Devor et al 1992, Kajander & Bennet 1992, Study & Kral 1996) which, if present in normal nociceptors, would convey a pain-like percept to the CNS. The Na + channel called PN3 (or SNS), detected primarily in small diameter sensory neurons (Novakovic et al 1998, Rabert et al 1998), appears to play a critical role in neuropathic pain of several aetiologies, since prevention of its expression by gene knockout (Akopian et al 1999) or by antisense oligonucleotides delivered into the spinal CSF (Porreca et al 1999) can reverse or prevent the behavioural signs of neuropathic pain. Na + currents through PN3 channels are longer lasting than those of other ‘fast inactivating’ channels (Elliott & Elliott 1993, Rush et al 1998) and can thereby support a long, post-stimulus depolarization with a period of sustained impulse ¢ring (Parri & Crunelli 1998, Kapoor et al 1997). In opposition to a critical role of PN3 upregulation are voltage-clamp and molecular biological assays that indicate a reduction in PN3-mediated Na + current and messenger RNA after certain types of pain-producing procedures and suggest that a di¡erent Na + channel, NaN, having even more prolonged currents, may be increased (Cummins & Waxman 1997, Cummins et al 1999, Tate et al 1998). Furthermore, the reported drop in K + current density will also favour prolonged depolarization as well as both spontaneous and induced repetitive ¢ring.

Na+ CHANNELS, BLOCKERS, AND NEUROPATHIC PAIN

191

In this paper we report ¢ndings from neurobehavioural and in vitro electrophysiological investigations directed to determine if potential Na+ channel blockers, at plasma concentrations known to be therapeutic, are indeed capable of signi¢cant Na + channel blockade. Using mechanical allodynia induced by spinal nerve ligation in the rat as the sign of neuropathic pain (Kim & Chung 1992), we have assessed the ability of controlled, intravenous concentrations of lidocaine and of the stereoisomers of mexiletine to restore normal force thresholds for withdrawal of the operated hindlimb. In the electrophysiological studies we have simulated nerve discharges involving non-inactivating Na+ channels by pharmacological modi¢cation of channel gating with peptide ‘a toxins’ from scorpions or sea anemone, and have examined the ability of lidocaine and mexiletine to suppress the resulting abnormal impulse behaviour. Lastly, with voltage-clamp of cells having rapidly inactivating Na + currents, we have investigated the changes in blocking potency of these drugs due to single prolonged depolarizations or to repetitive, brief, impulse-like depolarizations.

Results and discussion

Allodynia, induced by tight ligation of the spinal nerve distal to the dorsal root ganglion (DRG), is relieved by intravenous lidocaine (L) or the R-isomer of mexiletine (R-M) (Fig. 1). Steady-state plasma drug concentrations, achieved by a computer-controlled infusion pump, de¢ne the concentration-dependence for such relief. At about 1.0 mg/ml signi¢cant alleviation of allodynia is detected for both L (Chaplan et al 1995, Sinnott et al 1999) and R-M (Sinnott et al 2000); the S-M enantiomer is ine¡ective (Sinnott et al 2000). At 2^4 mg/ml L and 1.2^1.4 mg/ml R-M, allodynic thresholds reach pre-operative levels in many animals, with average values of c. 60% towards complete relief. Greater relief is not attained at higher concentrations of L or R-M, the latter limited by toxicity (pre-convulsive twitches) at 2.0 mg/ml, as is the therapeutically ine¡ective S-M.

Notably, substantial recovery from allodynia is still present days to weeks after e¡ective L infusions (Chaplan et al 1995, Sinnott et al 1999) whilst the periinfusional relief by R-M is not followed by any residual recovery (Sinnott et al 2000), implying that separate mechanisms may underlie the acute and the persistent therapeutic phases.

Some neuropathic pain may result from spontaneous or abnormally repetitive impulses in peripheral neurons. The slowly-inactivating or even more prolonged inward currents of PN3and NaN-type Na+ channels, respectively, have the theoretical ability to support long-lasting depolarizations that may be accompanied by repetitive impulse activity (Kapoor et al 1997). Modest, slow depolarizations near the resting potential may stimulate impulse initiation through activation of NaN channels whose opening occurs at potentials near threshold (Kral et al

192	STICHARTZ ET AL

FIG. 1. The e¡ects of successive intravenous lidocaine infusions at increasing concentrations on ipsilateral allodynia in rats (n = 8). Beginning on postoperative day 5, ¢ve infusions were administered at 48 h intervals which correspond to plasma concentrations of 1.11(d 5), 1.60 (d 7), 2.07 (d 9), 2.56 (d 11) and 3.03 (d 13) mg/ml. Open squares show mean paw withdrawal thresholds 5 min before administering the corresponding infusion. Filled squares show the mean paw withdrawal threshold value at the end of the infusion: *, pre-infusion threshold values on days 11, 13, 15, and 17 were signi¢cantly di¡erent from pre-infusion thresholds on day 5 (P50.02); **, post-infusion threshold values were signi¢cantly di¡erent from pre-infusion threshold values on this day (P50.02). (From Sinnott et al 1999, with permission.)

1999), much more negative than those of other Na + channels (Cummins et al 1999).

Normal, fast-inactivating Na + channels, as are found in large myelinated peripheral axons, slow their inactivation closing when bound by small peptide ‘a toxins’ (site III toxins) puri¢ed from scorpions or sea anemones (Ulbricht & Schmidtmayer 1981, Strichartz et al 1982). The usually brief spikes of action