Sodium Channels and Multiple Sclerosis: Roles in Symptom Production, Damage and Therapy

Abstract

Our understanding of the potential role of sodium channels in multiple sclerosis (MS) has grown substantially in recent years. The channels have long had a recognized role in the symptomatology of the disease, but now also have suspected roles in causing permanent axonal destruction, and a potential role in modulating the intensity of immune activity. Sodium channels might also provide an avenue to achieve axonal and neuronal protection in MS, thereby impeding the otherwise relentless advance of permanent neurological deficit. The symptoms of MS are largely determined by the conduction properties of axons and these, in turn, are largely determined by sodium channels. The number, subtype and distribution of the sodium channels are all important, together with the way that channel function is modified by local factors, such as those resulting from inflammation (eg, nitric oxide). Suspicion is growing that sodium channels may also contribute to the axonal degeneration primarily responsible for permanent neurological deficits. The proposed mechanism involves intra‐axonal sodium accumulation which promotes reverse action of the sodium/calcium exchanger and thereby a lethal rise in intra‐axonal calcium. Partial blockade of sodium channels protects axons from degeneration in experimental models of MS, and therapy based on this approach is currently under investigation in clinical trials. Some recent findings suggest that such systemic inhibition of sodium channels may also promote axonal protection by suppressing inflammation within the brain.

INTRODUCTION

The course of multiple sclerosis (MS) is very variable between patients, but there is commonly an initial relapsing/remitting period characterized by “attacks” from which a full recovery may be made, followed after a decade or so by a more progressive course when there is a steady accumulation of permanent neurological deficit. The neurological deficits in the relapsing/remitting and progressive phases tend to have different underlying mechanisms, and sodium channels may contribute to each in different ways. Our knowledge of how sodium channels contribute to the production of neurological deficits far outweighs our knowledge of how (and indeed whether) they contribute to the immunological mechanisms of MS, but some recent findings suggest they may affect the properties of immunological cells.

To understand the role of sodium channels in MS it may be helpful briefly to summarize their elementary properties. The sodium channels of interest in MS are mainly voltage gated and located in the cell membranes of neurons and their axons, or in the membranes of immune cells. Voltage‐gated sodium channels are normally closed, but they open transiently in response to depolarization of the membrane in which they are embedded. A small depolarization may only open a small percentage of the channels but although the channels may often open for only less than 1 ms, while they are open a stream of sodium ions flows through the channels into the cell. Sodium ions are positively charged and so they carry a current across the membrane and this serves to depolarize the cell still further, promoting the opening of additional sodium channels. In this way many sodium channels can open almost simultaneously, and, in excitable cells such as neurons, this can create the “explosion” of inward current that characterizes an action potential. Once open, sodium channels close automatically, often within a millisecond, although, as we shall see later, under pathological conditions the behavior of the channels changes and then the sodium current may become more persistent.

SODIUM CHANNELS AND THE SYMPTOMATOLOGY OF MULTIPLE SCLEROSIS

Loss of function—relapses

Demyelination.  Relapses in MS are typically characterized by a temporary (weeks or months) loss of function resulting in symptoms such as paralysis, blindness and numbness. Such deficits are primarily attributable to axonal conduction block, and probably the most important cause of this is demyelination. (Inflammation might also be able to block conduction in otherwise normal axons, as discussed below.) Demyelination is a potent cause of axonal dysfunction (135), especially if whole internodes of myelin are lost (ie, segmental demyelination), as commonly occurs in MS. Several factors play a role in the conduction block, but dominant is the fact that although sodium channels are aggregated in very high density (∼1000/µm²) precisely at the nodes of Ranvier, there is only a low density (<25/µm²) normally located beneath the myelin sheath (116, 143). This pattern is ideal for saltatory conduction, but if the myelin is removed by demyelination, as occurs in MS, it exposes a long length of axonal membrane that appears to be inexcitable. Some computer simulations predict that conduction may be possible along such demyelinated axolemma (23, 145, 149), but this has yet to be shown experimentally. The (relative) inexcitability of the freshly demyelinated membrane is devastating for conduction because the gap exposed by the loss of a whole myelinated internode is far too long for the current generated by the last active node, or hemi‐node (ie, the “node” at the start of the demyelinated stretch), to be able to excite the next hemi‐node, in the absence of myelin to guide the current and reduce capacitance. Conduction therefore fails. This conduction failure is arguably the most important cause of neurological deficit in MS. The failure was predicted by Charcot (37), and it was the first deficit to be detected experimentally, first in the peripheral nervous system (49, 92) and later in the central (94, 95, 96). The conduction block is not necessarily permanent however, as discussed below.

Conduction can also fail even when demyelination is less extensive and confined only to the paranodes resulting in nodal widening (63, 87, 139), a pathology known to occur in MS (153). Failure in this case arises from the increase in membrane capacitance at the widened node, which can require more current to depolarize it than is available. The widened nodal gap also means that the exciting local current flowing out at the node is no longer funneled to flow specifically across the nodal membrane, but rather it becomes dissipated over a larger area.

A potential cause of conduction failure in MS is the loss of sodium channels at nodes of Ranvier, as has been reported in animals with peripheral autoimmune inflammatory demyelinating disease (102). However, whether such loss occurs in MS is not known.

Inflammation/immunology.  There is a growing realization that inflammation may be an important cause of temporary neurological deficits in MS (14, 155). The case can be made most clearly when the deficits are quite brief (days), because then they are not easily explained by demyelination. Perhaps the most convincing evidence has arisen from observations made during therapy with Campath‐1H, a humanized monoclonal antibody directed against the CD52 antigen expressed by lymphocytes and monocytes (99). Within hours of the intravenous administration of the antibody, patients with MS experienced a resurrection of symptoms that had been expressed earlier in the course of their disease, but from which they had made a recovery. As old symptoms were resurrected rather than new ones appearing, the deficits presumably arose from conduction block in axons that had been previously damaged by the disease process, but the mechanism remained unclear. The resurrected symptoms were expressed for less than a day, ruling out any mechanisms based on structural changes such as demyelination with repair by remyelination. Conduction block caused by pyrexia (Uhthoff’s phenomenon, see below) was also ruled out by other observations. The administration of the Campath‐1H provoked a surge in the circulating concentration of cytokines, particularly tumor necrosis factor‐α, interferon‐γ and interleukin‐6 (99, 152), suggesting that cytokines may have directly impaired function in (presumably) demyelinated axons, although subsequent observations (38) indicated that the prime suspect, tumor necrosis factor‐α, was not responsible. Consistent with this interpretation, pretreatment of the patients with intravenous methylprednisolone was found to prevent both the cytokine surge and the expression of symptoms, suggesting that the cytokine/inflammatory response was directly involved. However, experimental study did not provide confirmation that these cytokines had direct effects on conduction (113). It now seems that the best explanation for the puzzle involves nitric oxide. Although not affecting axons directly, the cytokine surge would be expected to provoke the production of nitric oxide by the microglial and inflammatory cells located in the existing lesions, and nitric oxide can block axonal conduction, particularly in demyelinated axons (see below). Thus, at low concentrations of nitric oxide demyelinated axons would be preferentially affected, providing a plausible explanation for the transient exacerbation provoked by Campath‐1H administration.

It is also possible that other factors associated with inflammation may cause more prolonged neurological deficits by affecting sodium channels, although the evidence is weak and confusing. The concept of circulating factors that could block neurological function was introduced in 1965 (22), but despite a number of studies inspired by these early observations the identity of the factors was, and has remained, obscure [see discussion in Smith (127)]. A pentapeptide named endocaine, or QYNAD (an acronym standing for the sequence of amino acids) was more recently reported to be present specifically in the cerebrospinal fluid of patients with the inflammatory neurological disorders MS or Guillain‐Barré syndrome and to have a local anesthetic‐like effect on sodium channels, like lignocaine (6, 30). The implication that QYNAD might be responsible for neurological deficits in MS was greeted with interest, particularly when QYNAD was reported to block conduction in sciatic nerve (151). However, initial excitement was dampened by the failure of several laboratories to reproduce the findings on sodium channels (44, 112) or to find effects on the electrical activity of a neuronal network (105). Disturbingly, some of the original observations have been retracted (31), but not the main effects on sodium channels, and these have also received some support from other studies (97, 106). At this time, QYNAD’s properties remain potentially interesting, but extraordinarily elusive.

It has long been conjectured that antibodies might affect axonal conduction in MS, perhaps by binding to sodium channels (146, 147), but until recently there was little convincing evidence in support. However, preliminary findings have now been presented that anti‐neurofascin antibodies are present in MS; and that monoclonal antibodies to neurofascin colocalize with sodium channels at central nodes of Ranvier (90). Moreover, administration of the monoclonal antibodies to animals with experimental autoimmune encephalomyelitis (EAE; a model of MS) exacerbated the severity of disease, raising the possibility that, in MS, anti‐neurofascin antibodies may have deleterious electrophysiological consequences promoting loss of function.

Another study has recently described conduction block in the rat optic nerve induced by cerebrospinal fluid from some patients with MS (36). Whether this effect is mediated by effects on sodium channels is not yet clear. Similarly, there is equivocal evidence that T cells can block axonal conduction (50, 154), but as the mechanism may not involve sodium channels it will not be discussed further here.

Other factors associated with the inflammatory response that can affect sodium channel function, and so conceivably the expression of symptoms, are some cytokines and prostaglandins. For example, interleukin‐2 can affect sodium currents, at least in muscle (77), and tumor necrosis factor‐α affects sodium conductance, at least in Aplysia neurons (122), although this evidence must be weighed against a preliminary study which found no clear effects of tumor necrosis factor‐α on demyelinated axons (113). Prostaglandin E₂ affects the tetrodotoxin‐resistant sodium current in rat dorsal root ganglion, and colonic, neurons (55, 64) resulting in increased excitability (55). Whether these effects would occur within the central nervous system in MS is not clear.

Nitric oxide.  There is abundant evidence that nitric oxide is produced within the lesions of MS (128) and so it may be important that nitric oxide is also capable of blocking axonal conduction (113, 126), especially in demyelinated axons (113) (Figure 1; see also Figure 7). Experimentally, the block can be imposed within minutes of exposure to nitric oxide, maintained for hours, and relieved within minutes of removing the nitric oxide (113). The mechanism(s) is not known, but nitric oxide can affect the function of sodium channels (3, 66, 114, 123) and the sodium pump (65, 121), in addition to the function of other ion channels, perhaps by acting on channel thiols (3, 85, 125). Nitric oxide may alternatively block conduction by inhibiting mitochondrial metabolism (21, 32) and causing membrane depolarization (61, 62)

Figure 1.

A series of compound action potentials showing conduction through a demyelinating ethidium bromide lesion induced in the rat dorsal columns 23 days previously. The records were obtained 2 minutes apart and are plotted in three‐dimensional perspective with the earliest records at the front. The compound action potential has two peaks: the first shows conduction in normal axons, and the second shows delayed conduction in demyelinated axons. After the first 12 records were obtained an injection of a nitric oxide donor, spermine NONOate, was made into the lesion. The reduced amplitude of the second peak alone shows that the nitric oxide selectively blocked conduction in the demyelinated axons. A later, control injection of vehicle alone at the same site had no appreciable effect on conduction. Reproduced from Redford et al (113).

Figure 7.

Four series of records showing compound action potentials obtained simultaneously from four dorsal spinal roots in an anesthetized rat. The individual records were taken 4 minutes apart and so span several hours: the earliest records are plotted at the front. Initially the roots were stimulated at 1 Hz (ie, once per second) and after 30 minutes the medium bathing the roots was exchanged either for fresh control medium (top plots) or for control medium with the addition of the sodium channel‐blocking agent flecainide (bottom plots). After an additional 15 minutes the stimulation rate for all the roots was increased to 100 Hz, and this frequency was maintained for 5 h, after which the rate was reduced back to 1 Hz. For 2 h (bar) during the 5 h period the roots were exposed to nitric oxide which promptly blocked conduction: a higher concentration of nitric oxide was used here in comparison with Figure 1. When the nitric oxide was washed away, conduction was restored to the roots treated with flecainide, but not in the control roots. The failure of conduction in the control roots is because the axons degenerated in response to the impulse activity in conjunction with exposure to nitric oxide, whereas the inclusion of flecainide provided almost complete protection. Modified from Kapoor et al (76).

SODIUM CHANNELS AND THE RESTORATION OF FUNCTION—REMISSION

Neurological function can be restored in MS by at least four mechanisms, any of which can contribute to the remissions that characterize relapsing/remitting disease. One mechanism is simply the resolution of any inflammation that may have held normal function in abeyance, and another involves “plastic” gray matter adaptations to compensate for functional loss, but the remaining two mechanisms involve the reorganization of axonal sodium channels and will be considered further.

Restoration of conduction to demyelinated axons.  The low sodium channel density along freshly exposed demyelinated axolemma initially prevents conduction because the axolemma is inexcitable, as described above, but one of the adaptations of demyelinated axons to the demyelinated state is the acquisition of sodium channels distributed along the demyelinated axolemma. The first evidence for this was the observation in the peripheral nervous system that conduction could occur along demyelinated axons and that it adopted a continuous, rather than saltatory, mode of conduction across the demyelinated segment (26, 27). Later experiments revealed that conduction could also occur in peripheral axons by micro‐saltation between new node‐like foci of sodium channels, termed phi‐nodes, that form along the demyelinated axolemma every 100–400 (mean 255) µm (133). However, the techniques employed to prove successful conduction in demyelinated peripheral axons could not be applied in the central nervous system and so proof here awaited a study in which conduction was demonstrated by intra‐axonal recordings from individual dorsal column axons that were then labeled iontophoretically and reconstructed in three‐dimensions through the lesion at the electron microscope level (58). Conduction was proven to occur over demyelinated regions several internodes in length (2500 µm), and could be restored within 2–3 weeks of the induction of the demyelinating lesion. Although successful conduction is theoretically favored by a small axon diameter, it was observed in demyelinated axons as large as 5.5 µm in diameter (58), suggesting that most demyelinated central axons will be physically capable of conduction under ideal conditions (even though many fail to conduct in practice).

Initially it was not known whether the internodal sodium channels upon which conduction depended were original ones that had spread from the positions of the old nodes, but more recently it has become clear that at least some of the channels must be inserted de novo because the new channels can be of a different subtype than is usually expressed at adult nodes. During development, before ensheathment with glial cells, axons have a diffuse distribution of Na_v1.2 along the axolemma (20, 74), and it is presumably these that support impulse conduction prior to myelination (150). As myelination proceeds, Na_v1.2 channels become clustered at the new nodes, but they are gradually replaced by Na_v1.6 channels so that adult nodes are populated by Na_v1.6 channels (20, 35, 74). Interestingly, axons demyelinated caused by EAE show an increased expression of not only Na_v1.6 along the demyelinated axolemma, but also of Na_v1.2 (40). The Na_v1.2 channels were presumably inserted in response to the demyelination. Sodium channel immunoreactivity could occur over regions of more than 30 µm in optic nerve axons, and was associated with up‐regulation of mRNA for Na_v1.2 in retinal ganglion neurons. Notably, the same pattern of changes in Na_v1.2 and Na_v1.6 has also been observed along demyelinated axons within MS lesions (42) (Figure 2).

Figure 2.

Immunohistochemical labeling of Na_v1.2 (B,D,F,H) and Na_v1.6 (A,C,E,G) sodium channels in the spinal cord and optic nerve in multiple sclerosis (MS). Myelin basic protein (MBP; green) is plentiful in the normal tissue (A,B) but almost absent from the demyelinated lesions. In the normal tissue the sodium channel labeling is punctate (nodal) for Na_v1.6, but absent for Na_v1.2. In contrast, both types of channel show almost continuous labeling along demyelinated axons (eg, C,D,E,F; white arrows). Reproduced from Craner et al (42).

The central demyelinated axons studied by Felts and colleagues were found to conduct even when at least 88% of their surface area was entirely devoid of any glial contacts across several internodes (58). This last point is noted because, where present, astrocytes may play a role in the location of sodium channels as astrocytic processes make intimate contacts with demyelinated axons specifically at sites that exhibit a node‐like, electron‐dense undercoating deemed to indicate the location of high densities of sodium channels (15, 19, 118). The presence of conduction in the absence of substantial astroglial ensheathment may be related to islands of sodium channels that have been detected at the electron microscope level along naked demyelinated dorsal column axons in the rat, using well‐characterized anti‐sodium channel antibodies (Figure 3: T. Deerinck, M.H. Ellisman, P.A. Felts, S.R. Levinson and K.J. Smith, unpub. obs.). The islands have the longitudinal dimension of nodes of Ranvier, that is, approximately 1 µm, but they sometimes appear on just one side of the axon, rather than circumferentially as would be expected if they represented the locations of the old nodes. If the aggregations represent nascent nodes, it is interesting that they form in the apparent absence of the processes of astroglial, or any other, cells. The appearance of these axons is reminiscent of the clusters of sodium channels found in the absence of any ensheathing cells along peripheral, dystrophic axons (47).

Figure 3.

Electron micrographs showing longitudinal sections through central demyelinated axons in the rat dorsal columns 1 month after the intraspinal injection of ethidium bromide accompanied by β‐irradiation (to prevent repair by remyelination). The tissue is unstained but labeled with an anti‐sodium channel antibody. A. A hemi‐node at the edge of a demyelinated segment shows labeling along the axolemma at either side of the axon at the location of the old node. B. Node‐like labeling along a naked demyelinated axon. C,D. Unilateral labeling of sodium channels along naked demyelinated axons: insets show higher magnification. T. Deerinck, M.H. Ellisman, P.A. Felts, S.R. Levinson and K.J. Smith, unpublished observations.

The restoration of conduction in demyelinated axons can explain the clinical observations of apparently “silent” demyelinated lesions, namely large demyelinated lesions in pathways (eg, the optic nerves) where conduction block would have been expected to produce symptoms, but where no, or only minimal, deficit was detected (88, 103, 108, 141).

The conduction velocity along the demyelinated portion of central axons is very much reduced, as it is in peripheral axons (26), and it is associated with a refractory period for transmission prolonged from the normal range of 0.5–1.4 ms, to 1.0–6.0 ms, with one axon requiring a gap of 27 ms between impulses before the second was faithfully transmitted (58). This insecurity of conduction in demyelinated axons, coupled with the properties of sodium channels, results in a range of curious phenomena related to body temperature: in an extreme example, one patient temporarily went blind if she used a hair dryer (29). The exacerbation of symptoms upon body warming, known as Uhthoff’s sign (124, 136), is a common phenomenon in MS and it is attributed to the fact that the temperature coefficient for sodium inactivation is larger than that for activation (45). The consequence is that the duration of action potentials gets fractionally shorter upon warming and this reduces the current that is available to depolarize the demyelinated membrane to its firing threshold. In axons on the verge of conduction failure, as central demyelinated axons often are, this additional compromise can result in conduction block. Thus a patient with demyelination of the optic pathway, for example, can suffer temporary conduction block and blindness upon head warming resulting from the use of a hair dryer.

The low security of conduction at sites of demyelination also contributes to the early loss of function upon repeated use of demyelinated axons, which can result in progressive weakness upon exercise (93), or perhaps cause the fading or blurring of vision upon fixated gaze (91, 144). These deficits have been related to the fact that repeated conduction along demyelinated axons can result in intermittent periods (eg, 0.2–2.0 s) of complete conduction block, interspersed with periods when conduction through the lesion occurs faithfully (57, 134). The periods of conduction block result from membrane hyperpolarization in response to activity of the electrogenic Na⁺/K⁺ ATPase (sodium pump) induced by the raised intracellular sodium resulting from the preceding impulse activity (24, 26).

Abnormalities in channel expression, and their consequences, are perhaps best exemplified in the Purkinje neurons of the cerebellum. Black and colleagues (16) have found that the mRNA and protein of the Na_v1.8 subtype of sodium channel are clearly expressed in the Purkinje neurons of animals with EAE and patients with progressive MS, but not at all in normal Purkinje neurons. The Na_v1.8 channels are expressed with annexin II (39), a protein that facilitates insertion of the channel into the cell membrane, which is consistent with a belief that the channels are functional. The Na_v1.8 channel is resistant to steady state inactivation (4) which means that it can open even when the membrane is depolarized: other subtypes inactivate under these conditions. The Na_v1.8 channel also recovers rapidly from inactivation (54), and thus it will, for example, support repetitive firing even in depolarized neurons. Now the electrical behavior of a neuron is largely determined by the aggregate properties of its sodium channels, and so the atypical properties of Na_v1.8 channels would be expected to confer atypical properties on the Purkinje neurons of MS patients. Consistent with this expectation, Purkinje neurons transfected to express Na_v1.8 channels display atypical properties, including the production of sustained repetitive firing in response to a sustained depolarizing current (115). Purkinje neurons from mice with EAE display similar properties (120). Such unusual behavior is particularly interesting when considered with the fact that patients with MS commonly exhibit cerebellar deficits with characteristics consistent with a sodium channel abnormality (148). For example, the deficits are frequently persistent, rather than remitting, may have a paroxysmal expression (in common with inherited channelopathies), and they typically show a favorable response to therapy with sodium channel‐blocking agents such as carbamazepine.

Hyperexcitability.  During the course of their disease, many patients with MS experience so‐called “positive” symptoms, namely the gain of phenomena such as tingling sensations and neuropathic pain. These effects can be attributed to changes in the pattern of expression and distribution of different types of sodium (and potassium) channels that can confer not only excitability to demyelinated axolemma, but also hyperexcitability. Affected axons can become spontaneously active, generating long trains of impulses that arise ectopically at the site of demyelination, conducting in both directions from it (129, 130). The impulses traveling to the brain are interpreted as tingling etc. sensations referred to the body part innervated by the axons (see also 100), or as flashes of light in the case of optic nerve axons (46). Hyperexcitable axons can generate either continuous discharges (at 10–50 Hz), or may fire in busts, with burst durations of 0.1–5 s, separated by silent periods of 0.1–100 s. Continuous discharges, in particular, have been related to the appearance of a slow, persistent inward sodium current that can appear at demyelinated membranes (75, 117, 138) (Figure 4), and this mechanism is consistent with the fact that bothersome tingling sensations respond well to sodium channel‐blocking agents such as carbamazepine. Sensory axons are more prone to hyperexcitability than their motor counterparts, perhaps because they have a greater expression of persistent sodium currents (25, 98), and so sensory manifestations of hyperexcitability are more common than motor, although phenomena such as myokymia have been reported (68).

Figure 4.

Portions from an intra‐axonal, micropipette recording from a location at or near the demyelinated portion of a central axon in an ethidium bromide lesion in the rat spinal cord. The records show oscillations in the membrane potential which either routinely (A) or periodically (B,C) initiate action potentials. The almost sinusoidal waveform is clearly observed where the depolarizations are subthreshold for initiating spikes (B,C). The oscillations are abolished by sodium channel‐blocking agents (not shown) indicating that they arise from currents flowing through sodium channels. Reproduced from Kapoor et al (75).

Hyperexcitability in demyelinated axons is normally accompanied by the acquisition of mechanosensitivity, which results in the generation of phasic and tonic bursts of ectopic impulses upon slight deformation (particularly stretching) of the site of demyelination (129, 130). Sensory axons are again more susceptible than motor axons, and demyelination affecting the sensory axons of the cervical dorsal columns can result in the perception of electric shock or tingling sensations upon stretching the cervical cord by bending the neck forwards (Lhermitte’s symptom) (73, 84).

Remyelination.  Repair of demyelinated axons by remyelination is common in some patients with MS (107, 110, 111). Based on observations in experimental lesions there is little doubt that the remyelinated axons will conduct impulses, although this remains unproven in MS. Central axons in which conduction had been blocked by experimental demyelination were shown to conduct when repaired by spontaneous remyelination (131, 132), even when the new myelin internodes possessed as few as five lamellae (58). Furthermore, maturation of the new sheaths was accompanied by the restoration of near‐normal conduction velocity, refractory period for transmission and ability faithfully to conduct trains of impulses at high frequency (131, 132). The type of remyelinating cell does not seem to be very important, as remyelination is effective in restoring conduction irrespective of whether it is achieved primarily by oligodendrocytes (131, 132), or endogenous Schwann cells (56), or by transplanted cells of various types (69, 70, 79, 142). At least in the case of Schwann cells, the new nodes formed by remyelination regain the normal configuration of ion channels, including the presence of Na_v1.6 channels in the nodal membrane (18) (Figure 5). Accordingly, remyelination has been associated with the restoration of function at the behavioral level in experimental lesions (72), and also probably in clinical studies (33), and where it occurs, remyelination probably makes an important contribution to remissions.

Figure 5.

Immunohistochemical labeling of mature (>1 year old) nodes of Ranvier formed on central axons in the rat dorsal columns during the process of remyelination by endogenous Schwann cells. The axons were formerly demyelinated by the intraspinal injection of ethidium bromide. Upper panel: green shows the location of Caspr, found on either side of nodes of Ranvier. The picture labeled 2A shows that Na_v1.2 sodium channels are not present at the remyelinated nodes, although the antibody recognizes them at nodes in developing optic nerve (inset). The remyelinated nodes label clearly for the presence of Na_v1.6 sodium channels bounded by Caspr labeling (B). Peripheral myelin is shown by P0 labeling (blue in C,D) and it is clear that Na_v1.6 sodium channels, but not Na_v1.2, are present at the nodes formed by remyelination. Scale bars 10 µm. Lower panel: green marks ezrin, a molecule expressed at the nodal microvilli of Schwann cells. Na_v1.2 sodium channels are not present at the central nodes formed by remyelination by Schwann cells (pictures labeled 4A, B, C), but Na_v1.6 channels are expressed at the nodal gap (D, E, F). Reproduced from Black et al (18).

AXONAL DEGENERATION AND INJURY—PROGRESSIVE DISEASE

Axonal degeneration is believed to be the main cause of the permanent neurological deficits that characterize the progressive forms of MS. A number of mechanisms have been proposed to account for the degeneration but here we focus only on the growing evidence that sodium channels may play an important role in a mechanism involving energy insufficiency.

Pathological evidence shows that axons degenerate in large numbers in acute, inflammatory demyelinating lesions where the magnitude of the damage is proportional to the intensity of inflammation (59, 140). Such studies indicate a role for factors associated with the inflammatory response, including nitric oxide. Nitric oxide is produced in large amounts within inflammatory MS lesions (128) and one of its most important potential effects is that it can potently inhibit mitochondrial function (21, 32, 51), reducing the production of ATP. Most ATP in the brain is destined to fuel the pumps that maintain ion gradients (5), and of these the most important is the Na⁺/K⁺ ATPase, or sodium pump. Thus axons passing through nitric oxide‐rich inflammatory lesions in MS may be relatively starved of the ATP required to maintain sodium homeostasis, in which case they could be particularly vulnerable to the energy demands resulting from sustained impulse activity. An experiment based on this reasoning found that, indeed, axons could be killed by physiological frequencies of impulse activity, if this occurred while the axons were exposed to nitric oxide (135). The axons will have suffered a raised entry of sodium ions because of the impulse activity, perhaps exacerbated by a nitric oxide‐induced modification of the sodium channels that makes them pass a persistent sodium current, namely a persistent drain of sodium ions into the axon (3, 66). By analogy with axons rendered experimentally ischemic, a high internal sodium ion concentration can result in a lethal influx of calcium ions via reverse action of the sodium/calcium exchanger (137). Thus impulse activity in the presence of nitric oxide could initiate a sequence of events that culminates in the calcium‐mediated digestion of the axon [described in more detail in Bechtold and Smith (8)]. In patients, impulse activity would, of course, be substantially increased in any axons that became hyperexcitable and spontaneously active, as they can generate continuous trains of impulses at frequencies up to 50 Hz (7, 75, 117, 129, 130). The realization that sodium loading may be an important cause of axonal degeneration has stimulated research into strategies for axonal protection, as discussed below.

There is evidence that it may be preferentially the Na_v1.6 subtype of sodium channel that predisposes axons to damage. Thus Craner and colleagues labeled dorsal column axons from animals with EAE for myelin basic protein, Na_v1.2, Na_v1.6, the sodium/calcium exchanger, and β‐amyloid precursor protein as a marker for axonal damage (41) (Figure 6). Nearly all (92%) of the damaged axons were positive for Na_v1.6, whereas only 38% were positive for Na_v1.2, and, of these, 95% also expressed Na_v1.6. Interestingly, most (73%) of the damaged axons coexpressed Na_v1.6 with the sodium/calcium exchanger, but only 4% of apparently undamaged axons expressed both these markers. The authors concluded that co‐expression of Na_v1.6 and the sodium/calcium exchanger is associated with axonal injury. Notably, similar findings were made in acute MS lesions (42). The association of Na_v1.6 with axonal injury is potentially related to the fact that Na_v1.6 channels produce a larger persistent sodium current, than, say, Na_v1.2, and so would tend to promote sodium influx even in the absence of impulse activity (119).

Figure 6.

Images of tissue from multiple sclerosis immunohistochemically labeled for different molecules. The panels on the left show the same microscopic field, and a single, different field is shown on the right. A and B show labeling for Na_v1.6 and Na_v1.2 sodium channels respectively (red), C and D show labeling for the sodium/calcium exchanger (NCX, green), E and F show β‐amyloid precursor protein (β‐APP, blue) and G and H show the merged images. The same structure (which is presumably a damaged, ie, β‐APP⁺, axon) is labeled by all the antibodies in the left panel, but on the right the β‐APP⁺ structure is weakly positive for NCX, but it is not positive for Na_v1.2. The data indicate that damaged axons are positive for Na_v1.6 and NCX, but not for Na_v1.2. Reproduced from Craner et al (42).

The damage resulting from sodium channel activity may be augmented by a calcium‐mediated modification of channel structure. It has recently been found that sodium‐mediated calcium entry in stretch‐injured axons can result in the activation of proteases that cleave part of the α‐subunit of the sodium channel responsible for channel inactivation. The resulting failure of inactivation results in a persistent inward sodium current that results in more calcium entry, in a positive feedback mechanism (71). Working by another mechanism, calcium may also increase the sodium current amplitude of Na_v1.6 channels, and affect their inactivation kinetics, via calmodulin (67). It seems reasonable to propose that these mechanisms may occur in MS, and, if so, this would amplify the sodium and calcium entry, promoting the activation of local calcium‐dependent degradative enzymes, and resulting in lysis and degeneration of the axon.

Although, on the evidence above, demyelinated axons might appear to be especially vulnerable to sodium‐mediated degeneration, and although this seems intuitively logical, it is not inevitably true. Thus although it is true that conduction of a single impulse along a demyelinated axon will result in much more axonal sodium loading than if the axon were myelinated, it is also true that a demyelinated axon will not conduct a very heavy impulse load, that is, it will not conduct at very high frequency, and it will not conduct at moderate frequency for very long before conduction intermittently fails (134). In these ways demyelinated axons gain protection from sodium loading, whereas normal axons remain vulnerable. Indeed, it was normal axons that degenerated in response to impulse activity in the presence of nitric oxide in the study referred to earlier in which axons were stimulated in the presence of nitric oxide (135): demyelinated axons were not studied in this way, and might not, in any case, have conducted at all in the presence of nitric oxide. It is notable that a detailed pathological study of MS found little or no correlation between plaque load and axonal loss, indicating that demyelination may not be the primary cause of axonal loss (48).

The prediction of energy insufficiency in MS caused by nitric oxide, described above, is largely circumstantial, but direct evidence supporting a belief that an energy deficient state may exist in MS lesions, perhaps by a different mechanism, has recently been provided by a study of expression levels of 33 000 characterized genes in postmortem motor cortex (52). The investigators found that 26 nuclear‐encoded mitochondrial genes were decreased, as was the functional activity of mitochondrial respiratory chain complexes I and III, specifically in neurons. If, as the authors suggest, these observations mean that ATP supply is diminished in MS neurons, then these cells will be particularly vulnerable to sodium accumulation, and damage by reverse sodium/calcium exchange. It is thus possible that some observations of a hypoxia‐like state within MS tissue (1, 2) owe their origin not to true tissue hypoxia, but rather to an energy‐insufficient state arising from the inadequacy of mitochondrial function.

The mechanisms for degeneration and protection described above seem to apply most directly to the intensely inflammatory lesions of relapsing/remitting MS where expression of the inducible (high output) form of nitric oxide synthase (iNOS) is particularly prominent and axons degenerate in large numbers. However, there is also a “slow burning” axonal degeneration that continues in chronic inactive demyelinated plaques and that may contribute significantly to clinical progression of the disease (80). In this regard it is noteworthy that although intense inflammation may be lacking in such cases, the axonal degeneration nonetheless occurs on a background of residual chronic inflammation where many microglial cells are intensely positive for iNOS (83). It is therefore possible that axons in progressive MS, as well as those in relapsing/remitting disease, may be vulnerable to nitric oxide‐mediated energy insufficiency, and hence to sustained impulse activity.

The mechanisms underlying the symptomatology of MS, including those not dependent on sodium channels, have recently been reviewed in detail (136).

SODIUM CHANNELS AND IMMUNE CELLS

There is growing evidence that sodium channels may affect not only the properties of neuronal cells, but also the properties of some immune cells. Such effects could explain why Bechtold and colleagues found that flecainide therapy in EAE and experimental autoimmune neuritis (EAN; a model of Guillain‐Barré syndrome) reduced not only the number of degenerating axons, but also the intensity of the inflammation at the peak of disease (10, 11). Thus, in EAE, flecainide therapy significantly reduced the number of activated macrophages/microglia and also the number of cells positive for iNOS (10). Similarly, phenytoin reduced the spinal cord infiltrates in mice with EAE by 75% (43). In fact it is not entirely clear whether the protective effect of sodium channel blockade in EAE is caused by effects on axons, or on immune cells, or both, although the axonal protection achieved in models where inflammation is absent (76) indicates that at least some of the protection is mediated on axons directly.

If sodium channel inhibition truly has anti‐inflammatory effects within the CNS, there are several ways in which it could be mediated. For example, a role for sodium currents in T cell activation and co‐stimulation has been suggested (60, 78, 82), and fast transient sodium currents have been observed in several T cell and B cell lines (60). Furthermore, initial work in our laboratory has suggested that therapeutically relevant doses of lamotrigine and phenytoin can inhibit T cell activation and proliferation, for example, they reduce the proliferation of lymph node cells in response to phytohemagglutinin (89). With regard to B cells, both amiloride‐ and tetrodotoxin‐sensitive sodium channels have been reported on human and rat B cells (28, 34, 104, 109) and blockade of amiloride‐sensitive sodium channels inhibited antibody secretion by lymphocyte hybridomas (156). In this context it is notable that flecainide therapy of rats with EAN was found to cause a reduction in the serum titer of anti‐peripheral myelin antibodies (11), although it is possible that this reduction was related to the reduced axonal damage which may have limited the stimulus for B cells to produce antibodies.

Sodium channels are also present in macrophages and microglial cells, namely cells that perform effector and other functions in MS, EAE and EAN, and the expression of Na_v1.6 channels is increased in activated microglia and macrophages in MS and EAE (43). Voltage‐gated sodium channels have been described in rat (81) and human (101) microglia, and it has been suggested that the activation of a sodium current might be necessary to trigger microglial activation and facilitate a subsequent immune response (53). In support of this suggestion, the sodium channel‐blocking agent lignocaine has been found to inhibit superoxide release from rabbit alveolar macrophages (13) and Craner and colleagues (43) found that treatment of activated rat microglia with the sodium channel‐blocking agent tetrodotoxin reduced microglial phagocytic activity by 40%.

THERAPY

It is now accepted that axonal degeneration is a major cause of permanent neurological deficit in MS, and this has sharpened the need to identify therapeutic strategies for axonal protection. The realization that electrical activity and sodium loading could be an important cause of axonal degeneration raised the possibility that axons might gain protection by partial blockade of their sodium channels, or blockade of the reverse mode of action of the sodium/calcium exchanger (8). This possibility was confirmed by the observation that axons survived exposure to nitric oxide if the sodium channel‐blocking agents flecainide or lidocaine (135) or tetrodotoxin or sipatrigine (62) were included in the incubation medium, or if the sodium/calcium exchange‐blocking agent bepridil was added (76). Furthermore, axons survived sustained impulse conduction in the presence of nitric oxide if flecainide was present, even though the dose was sufficiently low that conduction continued despite the presence of the drug (76) (Figure 7). The partial blockade of sodium channels mediated by the drugs presumably limited the rise in the sodium ion concentration within axons, keeping it below the level at which damaging calcium entry occurred via reverse action of the sodium/calcium exchanger (see above).

To determine whether the pharmacological blockade of sodium channels might form the basis of a therapy in MS, additional experiments examined whether such drugs could protect axons from degeneration in animal models of the disease, namely different forms of EAE. Flecainide and lamotrigine were found to be protective in a chronic relapsing rat model of EAE (9, 10, 12) and also in a model of Guillain‐Barré syndrome (11). In EAE, flecainide therapy, from 7 days post immunization to the end of the experiment at day 28, reduced axonal degeneration from the control value of 38% degeneration in severely affected animals, to only 2% in similarly severely affected animals (9, 10). Phenytoin was also found to provide significant protection in a progressive model of EAE in the mouse (86), even when therapy was extended over a 6‐month period (17). Importantly, therapy with sodium channel‐blocking agents improved the functional outcome (9, 10, 17, 86). Based on these encouraging observations, clinical trials of lamotrigine and phenytoin have been initiated in London and New Haven, respectively, treating patients with secondary and primary progressive MS. Success in these trials will indicate that the seemingly relentless accumulation of permanent neurological deficit in progressive MS can be subdued, with the hope that patients can retain their neurological function in the long term.