Axonal protection in multiple sclerosis—a particular need during remyelination?

The current issue of Brain contains four papers that illuminate different aspects of inflammatory demyelinating disease, especially multiple sclerosis (Black et al., 2006; Coman et al., 2006; Howell et al., 2006; Patrikios et al., 2006). One paper focuses on axonal protection, while the others describe aspects of spontaneous remyelination in the disease. Promoting remyelination is a major therapeutic goal in multiple sclerosis, but some observations from the current papers raise the possibility that remyelination may transiently render axons vulnerable to degeneration before long-term protection is achieved, as discussed below.

In recent years the identification of strategies to protect axons from degeneration in multiple sclerosis has emerged as a major research priority. This emphasis is in response to the realisation that axonal degeneration is substantial in the disease (Trapp et al., 1998; Ganter et al., 1999), and that it is a major cause of permanent neurological deficit (De Stefano et al., 1998). The mechanisms responsible for the axonal degeneration are not certain, but inflammation and demyelination appear to be major risk factors (Raine and Cross, 1989; Ferguson et al., 1997; Trapp et al., 1998). This observation has prompted a new therapeutic approach based on the partial blockade of sodium channels (reviewed in Bechtold and Smith, 2005), and the potential value of this approach is explored in the current paper by Black et al. (2006) (see below). However, in addition to drug based approaches to axonal protection, there is widespread agreement amongst multiple sclerosis researchers that repair of the demyelinated axons by remyelination should also provide a good strategy for protection (Kornek et al., 2000; Stangel and Hartung, 2002; Dubois-Dalcq et al., 2005). Unfortunately, promoting remyelination in patients has proven to be a difficult therapeutic challenge, and this has focussed attention on the inherent ability of some lesions in multiple sclerosis to undergo spontaneous remyelination. The current papers by Coman et al. (2006), Howell et al. (2006) and Patrikios et al. (2006) explore aspects of remyelination in multiple sclerosis, using samples obtained at autopsy. Some more minor observations from these papers lead the current author to wonder whether remyelination may, surprisingly, make axons transiently more vulnerable to degeneration, even though remyelination may be an excellent approach for long-term axonal protection. A vulnerability expressed during the early stages of remyelination could help to explain why axons continue to degenerate at a ‘slow burn’ once overt inflammation has subsided (see below) (Lassmann and Wekerle, 2006). If so, the pharmacological strategy of partial sodium channel blockade might be especially valuable while remyelination is being achieved, before the newly formed nodes of Ranvier have matured.

To understand why remyelination could transiently render axons more vulnerable to degeneration, it is helpful to explore the mechanisms upon which the study by Black et al. (2006) is based. The idea that partial sodium channel blockade may provide axonal protection grew out of the observations that nitric oxide is a prominent component of inflammatory multiple sclerosis lesions (Smith and Lassmann, 2002), and that nitric oxide is also a potent inhibitor of mitochondrial metabolism and ATP production (Brown and Borutaite, 2002). Axons consume significant amounts of ATP, which is used primarily to fuel the sodium/potassium ATPase, or sodium pump. This pump functions to remove the sodium ions that enter the axon during impulse activity, and, if its function is inadequate due to nitric oxide-limited supply of ATP, the axons will become loaded with sodium ions. Observations made in the field of stroke research (Stys, 2004) have taught that an excessive intra-axonal sodium concentration can cause the sodium-calcium exchanger (situated in the axon membrane) to operate in reverse, so that calcium ions are imported into the axon. The resulting raised intracellular concentration of calcium ions can cause axonal degeneration by activating various degradative enzymes. Based on this line of reasoning, it was hypothesised that axons exposed to nitric oxide may be killed by the sodium load resulting from sustained impulse activity (Smith et al., 2001). Experiments confirmed this vulnerability, raising the possibility that axons could degenerate in multiple sclerosis lesions if they are required to conduct the physiological impulse traffic through inflammatory lesions containing nitric oxide (Smith et al., 2001).

Based on the above observations it seemed reasonable to suppose that if normal axons exposed to nitric oxide were vulnerable to sodium accumulation, demyelinated axons were probably even more vulnerable. Thus whereas in normal axons sodium entry is restricted to 1 μm lengths of axon membrane at the nodes of Ranvier, in conducting demyelinated axons the entry can occur all along the exposed, formerly internodal axolemma, an area up to a thousand times greater than is exposed at the normal node (Bostock and Sears, 1976; Smith et al., 1982). Indeed, two of the current papers (Coman et al., 2006; Howell et al., 2006) show that large portions of the demyelinated axolemma within multiple sclerosis lesions are well populated with sodium channels, in agreement with earlier observations (reviewed in Waxman et al., 2004). In normal axons, experimentation had revealed that sodium channel blocking agents (or blockers of the sodium-calcium exchanger) can protect axons from degeneration due to the combination of nitric oxide and impulse activity (Kapoor et al., 2003), and so sodium channel blocking agents have been examined for their ability to protect axons in inflammatory demyelinating animal models of multiple sclerosis and Guillain-Barré syndrome. In experiments performed in London and New Haven, the sodium channel blocking agents flecainide, lamotrigine and phenytoin were found to be very effective in protecting axons in rat and mouse models of experimental autoimmune encephalomyelitis and neuritis (Lo et al., 2003; Bechtold et al., 2004, 2005, 2006). These findings indicated that sodium channel blockade may provide an effective therapy for diseases such as multiple sclerosis and GBS, and the current paper from the Waxman laboratory continues this line of research, importantly showing that phenytoin provides good axonal protection for up to six months in both monophasic and chronic-relapsing models of multiple sclerosis, if the therapy is maintained (Black et al., 2006). This observation builds confidence that efficacy will be maintained upon chronic administration of the drug in the clinic, and, in fact, lamotrigine and phenytoin are currently being tested for axonal protection in clinical trials in London and New Haven, respectively recruiting patients with secondary and primary progressive multiple sclerosis: the results are expected in 2008/9. Chronic administration of the drugs is required because, in multiple sclerosis at least, axonal loss appears to occur continuously as a ‘slow burn’, in addition to bursts of degeneration associated with the appearance of inflammatory demyelinating lesions (Trapp et al., 1998; Lassmann and Wekerle, 2006).

Although chronic therapy with lamotrigine and phenytoin should be safe and hopefully effective, there is also a clear need to promote the structural repair of demyelinated axons by remyelination, as noted above. Indeed, Patrikios et al. (2006) elegantly show that in many patients remyelination is only limited, even sparse, although in other patients repair can be extensive, with up to 96% of the global lesion area remyelinated. Where it occurs, the current reports provide convincing evidence that remyelination can eventually restore nodes of Ranvier that appear to have the normal pattern of molecular constituents (Coman et al., 2006; Howell et al., 2006), and so they are presumably resistant to sodium accumulation. It is known that remyelination is effective in restoring secure conduction (Smith et al., 1979), and so for these and other reasons remyelination is expected to provide the best option for effective long term axonal protection. However, although not discussed in this context by the authors, some observations reported in the current papers suggest that the process of remyelination may transiently render axons particularly vulnerable to degeneration, before the mature nodal configuration is achieved. A synthesis of the observations of Coman et al. (2006) and Howell et al. (2006), by analogy with observations during peripheral remyelination (e.g. Novakovic et al., 1996), suggest that during remyelination the sodium channels that were distributed along the demyelinated axolemma may be ‘swept’ ahead of the advancing glial plasmalemma as oligodendrocytes encircle the demyelinated axons and then extend to establish their new territories along the axons. Thus as the edges of two adjacent oligodendrocytes approach each other, two focal aggregations of sodium channels also approach each other, on their way to merging. This process seems inevitably to result in regions of particularly high sodium channel concentration, which will presumably be prone to the rapid accumulation of intra-axonal sodium ions upon repeated impulse conduction. Indeed, wide nodal gaps at some nodes formed on remyelinated axons are described in one of the current studies (Patrikios et al., 2006), together with aggregations of sodium channels 3-fold longer than normal on demyelinated axons (Coman et al., 2006). Furthermore, node-like aggregations of sodium channels, or node-like structures, can occur either side of very short (4–30 μm) internodes (Howell et al., 2006; Patrikios et al., 2006), and some aggregations are separated by only 2 μm (Howell et al., 2006). Very short (e.g. 11 μm) remyelinated internodes were also described in multiple sclerosis previously (e.g. Prineas and Connell, 1979). If sodium channels are viewed not only as the substrate for impulse conduction, but also as risk factors for degeneration, then this early stage of remyelination would seem to place axons in jeopardy. The risk will be amplified by the fact that remyelination re-establishes the ability of axons to conduct long trains of impulses at high frequency. Thus whereas demyelinated axons presumably receive some protection from sodium loading due to their inability to fire at high frequency [they have an inherently long refractory period for transmission and promptly acquire more refractoriness with repeated activation (McDonald and Sears, 1970)], this protective limitation is lost quite early in the process of remyelination, at least in experimental lesions (Smith et al., 1981). Worse still, Coman et al. (2006) draw attention to the fact that sodium channel aggregation occurs in advance of the organisation of the other nodal components. These observations raise still further the possibility that, early in remyelination in multiple sclerosis, axons may pass through a period when there is an enhanced need for the apparatus to deal with excessive sodium load (e.g. adequate provision of mitochondria and sodium pumps), but before the apparatus is acquired. In this context it is interesting to note that Kuhlmann et al. (2002) have previously found that remyelinating and remyelinated axons in multiple sclerosis can sometimes show much more evidence of axonal injury than occurs in established demyelinating lesions.

Another point worthy of consideration is that any vulnerability imposed by early remyelination might turn out to be protracted in duration. In experimental lesions it is known that remyelination can be achieved quite rapidly, within a few weeks, but the presence in multiple sclerosis of many partially remyelinated lesions suggests that the tempo in this disease may be much slower. It follows that the abnormally wide nodes, and binary nodes, may not be fleeting phenomena on the pathway to the formation of mature nodes, but rather they may evolve slowly, or even represent a stage of arrested development. If so, the period over which axons are at risk could be prolonged, increasing the chances that a series of inopportune events (such as an atypical impulse load) could, by coincidence, combine and overwhelm the ability of the axon to maintain its integrity.

In summary, the papers within this issue of Brain establish not only the potential value of remyelination in the long term in that remyelination can be widespread and confer normal nodal properties in multiple sclerosis, but also that before maturation is achieved, axons may pass through a period of enhanced risk for degeneration due to sodium accumulation. This period may be protracted in time if remyelination is only partially completed. It is possible that chance combinations of deleterious events superimposed on metabolically stressed axons could well help to explain why axons degenerate at a slow burn in progressive multiple sclerosis. If so, the administration of sodium channel blocking agents may unexpectedly be especially valuable while immature nodes formed by remyelination persist.