Remyelination in the CNS: from biology to therapy

Transcription

1 NeuroN Glia interactions Remyelination in the CNS: from biology to therapy Robin J. M. Franklin* and Charles ffrench-constant Abstract Remyelination involves reinvesting demyelinated axons with new myelin sheaths. In stark contrast to the situation that follows loss of neurons or axonal damage, remyelination in the CNS can be a highly effective regenerative process. It is mediated by a population of precursor cells called oligodendrocyte precursor cells (OPCs), which are widely distributed throughout the adult CNS. However, despite its efficiency in experimental models and in some clinical diseases, remyelination is often inadequate in demyelinating diseases such as multiple sclerosis (MS), the most common demyelinating disease and a cause of neurological disability in young adults. The failure of remyelination has profound consequences for the health of axons, the progressive and irreversible loss of which accounts for the progressive nature of these diseases. The mechanisms of remyelination therefore provide critical clues for regeneration biologists that help them to determine why remyelination fails in MS and in other demyelinating diseases and how it might be enhanced therapeutically. Oligodendrocyte The myelin-forming cell of the CNS. *Department of Veterinary Medicine and Cambridge Centre for Brain Repair, University of Cambridge, Madingley Road, Cambridge, CB3 0ES, UK. MRC Centre for Regenerative Medicine and MS Society/ University of Edinburgh Centre for Translational Research, Centre for Inflammation Research, The Queen s Medical Research Institute, 47 Little France Crescent, Edinburgh, EH16 4TJ, UK. s: rjf1000@cam.ac.uk; cffc@ed.ac.uk doi: /nrn2480 Demyelination is the pathological process in which myelin sheaths are lost from around axons (Fig.1). In the CNS demyelination is usually the consequence of a direct insult targeted at the oligodendrocyte, the cell that makes and maintains the myelin sheath. This type of demyelination is sometimes referred to as primary demyelination to distinguish it from secondary demyelin ation (or Wallerian degeneration), in which myelin degenerates as a consequence of primary axonal loss. From a clinical perspective there are two major causes of primary demyelination in the CNS: genetic abnormalities that affect glia (leukodystrophies), and inflammatory damage to myelin and oligodendrocytes (with multiple sclerosis (MS) being by far the most significant of these inflammatory diseases) (BOX 1). Regardless of its cause, demyelination impairs function: the acute loss of a myelin internode is associated with conduction block. This block can be resolved by redistributing and inserting Na + channels along the denuded axolemma, to allow non saltatory conduction along the demyelinated segment 1. In addition to this adaptive response, and in striking contrast to the generally inadequate attempts at regeneration that follow damage to neuronal elements, the sequela to CNS demyelination is often the robust regenerative process of remyelination. In this article we review current know ledge of the biology of remyelination, including the cells and molecular signals that are involved. We also describe when remyelination occurs, when and why it fails and the consequences of its failure. Lastly we discuss approaches by which it might be enhanced therapeutically in demyelin ating diseases. For other recent reviews that consider some of these issues, see REFS 2 4. What is remyelination? Remyelination is the process in which entire myelin sheaths are restored to demyelinated axons, reinstating saltatory conduction 5 and resolving functional deficits 6,7 (Fig. 1). The term myelin repair is also sometimes used; however, this term evokes thoughts of a damaged but otherwise intact myelin internode being patched up, a process for which there is no evidence. It thus does not emphasize the truly regenerative nature of remyelination, in which there is restoration of the pre lesion cytoarchitecture. The tissue reconstruction in remyelination is complete except for one caveat: the correlation between axon diameter and myelin sheath thickness and length that is established during developmental myelination is less apparent in remyelination. Remyelination results in a thinner and shorter myelin sheath than would be expected for a given diameter of axon 8 the original dimensions are never attained 9. The relationship between axon diameter and myelin sheath thickness is expressed as a fraction: the circumference of the axon divided by the circumference of the myelin sheath, called the g ratio. The identification of abnormally thin myelin sheaths (greater than normal NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

2 Oligodendroglia Cells of the oligodendrocyte lineage, such as precursor cells and more-differentiated myelin-forming cells. g ratio) remains the most reliable means of identifying remyelination. This abnormality is obvious when large diameter axons are remyelinated (Fig. 1), but is less clear with smaller diameter axons such as those in the corpus callosum, where the g ratios of remyelinated axons are indistinguishable from those of normally myelinated axons 10. How is the relationship between myelin parameters and axon size established in myelination, and why is it disengaged in remyelination? In the PNS, axonally expressed neuregulin 1 (NRg1) type III has a pivotal role: reduced expression of type III NRg1 leads to a thinner than normal myelin sheath (increased g ratio), whereas overexpression leads to a thicker than normal myelin sheath (decreased g ratio) 11. In the CNS, however, the role of neuregulins is less clear: although type III Nrg1 +/ mice show reduced myelin sheath thickness in the corpus callosum 12, reports of normal myelination in mice with oligodendroglia that lack NRg1 or its receptors erbb3 and erbb4 suggest that NRg1 is not necessary for CNS myelination and that other signals must also contribute to the precise relationship between axon and oligodendrocyte 13. Likewise, the mechanistic basis of the increased g ratio in remyelination is not known and seems to be different from that which controls developmental myelination, as overexpression of NRg1 leads to hypermyelination in the developing CNS but not during remyelination 13. one hypothesis is that the difference results from the degree of change that the axon is undergoing. Whereas the myelinating oligodendrocyte a Neuron Oligodendrocyte Myelin sheath Axon Demyelination Remyelination No remyelination Functional recovery Progressive decline b Myelination Demyelination Remyelination Figure 1 The fate of demyelinated axons. a Following demyelination in the CNS, a demyelinated axon has two possible fates. The normal response to demyelination, at least in most experimental models, is spontaneous remyelination involving the generation of new oligodendrocytes. The myelin sheaths that are generated in remyelination are typically thinner and shorter than those that are generated during developmental myelination. Nevertheless, they are associated with recovery of function. In some circumstances, however, and notably in multiple sclerosis, remyelination fails, leaving the axons and even the entire neuron vulnerable to degeneration that largely accounts for the progressive clinical decline that is associated with demyelinating diseases. For this reason, therapies that increase the chances of the regenerative outcome of demyelination are keenly sought. b A well-established and effective means of identifying remyelination is to embed well-fixed tissue in resin and examine semi-thin sections by light microscopy. The images in this series are transverse sections from the adult rat cerebellar white matter, showing normally myelinated axons of various diameters in the left-hand panel, demyelinated axons (plus debris-filled macrophages) following injection of ethidium bromide in the middle panel and remyelinated axons with typically thin myelin sheaths four weeks after the induction of remyelination in the right-hand panel. 840 NoveMbeR 2008 volume 9

3 Box 1 Demyelinating diseases Primary demyelination in the CNS can be caused by genetic abnormalities that affect glia or by inflammatory damage. Although they are rare, the genetic diseases have provided valuable insights into myelin biology. They usually present in childhood with generalized neurological symptoms, such as changes in gait, muscle tone and cognition, and can be subdivided into those that result from defects of lysosomal function (for example, metachromatic leukodystrophy and Krabbe s disease) or perioxsomal function (adrenoleukodystrophy), those that result from abnormal oligodendrocyte myelinogenesis that is due to deficiencies in or misfolding of key myelin proteins (hypomyelinating leukodystrophies, such as Pelizaeus Merzbacher disease) and those that result from defects in the astrocytes that provide trophic support for myelinating oligodendrocytes (Alexander s disease and, probably, vanishing white matter disease) 237. As discussed at the end of the Review, the hypomyelinating leukodystrophies may provide the best initial targets for remyelination therapies by virtue of their cell-autonomous pathology. Although the cause of the demyelination in multiple sclerosis (MS) is inflammation, the clinical phenotype reflects a combination of inflammatory damage and neurodegeneration 238,239. MS differs from the leukodystrophies in that it presents focal neurological lesions and in that the triggers for the inflammation remain unknown. BOX 3 discusses the multifaceted nature of this disease. It is thought that both genetic and environmental factors contribute: there is a pronounced latitude effect in MS, with the UK frequency of approximately 1 affected individual in 1000 rising to 1 in 500 in Scotland, and an identical twin of an affected individual has 10 times the normal risk of developing MS. In the search for alleles that confer this risk, the HLA-DRB1 gene and, more recently, those for interleukin 2 receptor alpha and the interleukin 7 receptor have been identified 240. The impact on society of this common and disabling disease is considerable: it has been estimated that the total cost to the European Union is 9 billion Euros every year 241. Another important myelin disease in the CNS is periventricular leukomalacia. This is a major cause of cerebral palsy and is thought to result from oligodendroglial loss or, at later stages of fetal development, damage secondary to ischaemia or infection 242. However, it cannot be classified as a disease that causes primary demyelination (that is, loss of pre-existing myelin sheaths) for two reasons. First, the time of greatest risk (24 32 weeks of gestation) corresponds to the time of myelin formation, with a predominance of progenitor and pre-oligodendrocytes seen in the CNS 243. Second, there may be significant axonal damage, with secondary rather than primary oligodendroglial defects resulting from abnormal axo glial communication 200. It is more accurate therefore to classify this as a disease of hypomyelination. Cuprizone-induced demyelination model A model of demyelination induced by chronic oral administration of the copper chelator cuprizone. Cuprizone is toxic for oligodendrocytes, and the model, like the ethidium bromide model of demyelination, partly mimics a form of MS in which oligodendrocyte apoptosis predominates. Remyelination occurs following removal of the toxin from the diet. associates with a dynamically changing axon that is yet to achieve its full length and diameter, the remyelinating oligodendrocyte engages an axon that is undergoing less change, having already reached its mature size 14,15. Therefore, the remyelinating oligodendrocyte is not subjected to the same degree of dynamic changes as the myelinating oligodendrocyte during development. Such changes may, by analogy with other cell types 16, regulate protein synthesis and a number of intracellular signalling pathways and make an essential contribution to the elaboration of the normal myelin sheath. When does remyelination occur? a useful conceptual framework in which to study remyelin ation is to view it as a regenerative process that has many features in common with regenerative pro cesses occurring in other tissues of the body, and as the default response to demyelination (Fig. 1). This viewpoint is based on evidence from both experimentally induced and clinical demyelination. When demyelination is induced by toxins injurious to oligodendrocytes (for example, dietary cuprizone, which results in the demyelin ation of specific CNS tracts in a dose dependent manner (cuprizone-induced demyelination model) 17, or direct delivery of lysolecithin or ethidium bromide) then remyelin ation usually proceeds to completion, albeit in an age dependent manner (discussed below) Similarly, there is evidence that axons undergoing primary demyelin ation in experimental or clinical traumatic injury undergo complete remyelination, and that the persistence of chronically demyelinated axons is unusual 22,23. on the other hand, when demyelination is induced by or associated with the adaptive immune response, such as occurs in the auto immunity mediated condition MS and in experimental autoimmune encephalomyelitis (eae), an animal model of MS, remyelination takes place in an environment that contains elements that are intrinsically hostile to the oligodendrocyte lineage. Thus, the remyelination failure that is associated with MS (and eae) can be seen as a feature of specific disease states rather than a generic feature of remyelination. However, even in MS, a disease that is normally associated with failed or inadequate remyelination, there is evidence that in some patients complete remyelination occurs in a significant proportion of lesions 24,25 (Fig. 2). Similarly, remyelination can be extensive in eae, and models with significant persistent demyelination are unusual 26. Remyelination seems to be especially efficient following demyelination of cerebral cortical grey matter in experimental models 27,28 and clinical disease 29, although the reason for this is unclear. Why is remyelination important? although the regenerative process that underlies remyelin ation may have evolved to combat a range of infectious, metabolic, immune and other perturbations that cause demyelination, the current focus of remyelination research is on it s role as an important determinant of the outcome of demyelinating diseases such as MS. Demyelination undoubtedly forms a major part of the pathology of MS, but it has become increasingly apparent in recent years that there is also substantial axonal and neuronal loss. This is thought to be the major cause of this chronic progressive disease 30 and occurs as a secondary consequence of demyelination in addition to any primary effect of inflammation. Such a hypothesis explains why patients taking immunosuppressive therapies or with apparently quiescent disease still show increasing disability and progression: they will have persistent demyelination as a result of the failure of remyelination even in the absence of active disease. a key function of remyelination in MS is therefore axon survival, a view supported by recent experimental data 31. It should be noted that remyelination is not required for the resolution of symptoms in an acute relapse: the resolution of inflammation and the adaptive responses, including the insertion of Na + channels along the denuded axolemma to allow non saltatory conduction along the demyelinated segment 1, are sufficient (although the insertion of Na + channels can be compromised in chronically demyelinated axons owing to inadequate axolemmal expression of the Na + K + atpase, the enzyme that is responsible for rapidly correcting Na + and K + levels following an action potential and thereby for allowing repetitive axonal firing 32 ). NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

4 a LFB and HLA-DR b LFB and HLA-DR c Neurofilament d Active NAWM NAWM NAWM Active Inactive Active Active NAWM Inactive Inactive Inactive Figure 2 Ms an inflammatory demyelinating disease with variable degrees of remyelination. a A low-power image of the brain of a patient with multiple sclerosis (MS) stained with a myelin stain (luxol fast blue (LFB)) and immunostained with an antibody against HLA-DR (brown), a marker of major histocompatibility complex (MHC)-expressing inflammatory cells. Within the normal-appearing white matter (NAWM), three separate types of lesions (or plaques) can be identified: active lesions containing abundant inflammatory cells (left), lesions with an active rim and an inactive centre containing few inflammatory cells (centre), and inactive lesions with sparse myelin staining indicative of demyelination (right). b A higher-power image of inflammatory cells in the NAWM (left-hand panel), in an active lesion, where they have the appearance of phagocytic macrophages (middle panel), and in an inactive lesion with few inflammatory cells (right-hand panel). c In all three of the lesion types shown in part b, intact axons can be identified by neurofilament immunostaining. d The proportion of lesions that remain demyelinated is highly variable. These schematic diagrams are from four separate patients. The red areas are areas of demyelination whereas the blue areas are areas of remyelination. Images in part c kindly provided by Dr Chao Zhao. Part d reproduced, with permission, from REF. 24 (2006) Oxford University Press. The evidence that myelin is required for axon survival comes from observations of genetic mouse models and studies of human pathology. Transgenic mice lacking the oligodendrocyte expressed gene Cnp, which encodes a phosphodiesterase, or Plp, which encodes an integral myelin sheath protein, show long term axonal degeneration, even in the presence of myelin sheaths that are either ultrastructurally normal or that show only minor abnormalities 33,34. Further analysis of mutant mice revealed a disturbance in axoplasmic transport in the absence of PLP 35 and led to the identification of a myelin associated NaD dependent deacetylase, sirtuin 2, as a potential mediator of long term axonal stability, although the precise mechanisms remain uncertain 36. Myelin is also important for axon survival in humans: patients with Pelizaeus Merzbacher disease (PMD; caused by mutations in PLP) show axon loss 37, and studies of MS autopsy tissue showing that axon preservation is seen in those areas where remyelination has occurred are consistent with this supportive role and suggest that even the thin myelin sheaths that are characteristic of remyelination are sufficient to maintain axon survival 38. Therefore, PLP and probably also other oligodendrocyte factors may have a primary role in axonal stability rather than in myelin formation per se. These observations might be interpreted as suggesting that remyelination therapies will promote axon sparing by producing an oligodendrocyte derived trophic factor signal to the axon. In keeping with the existence of such factors, insulin like growth factor 1 (IgF1) and glialcell derived neurotrophic factor (gdnf) produced by oligodendrocytes in cell culture promote survival and axon outgrowth by embryonic neurons, respectively 39. However, the transgenic experiments could also be interpreted as revealing a toxic gain of function effect of myelin caused by the loss of key proteins: certainly the presence of a myelin sheath in the Cnp and Plpknockout mice means that these mice do not model the chronic demyelination that is seen in MS. There are at least two alternative explanations that need to be considered. First, axon degeneration might reflect the loss of an inflammation protective effect of myelin during an ongoing inflammatory process. Inflammatory mediators are released by microglia or are generated by hypoxia that is secondary to tissue damage and result in a malfunction of oxidative metabolism in the exposed (demyelinated) axon. These mediators will deplete atp and perturb mitochondrial function, causing failure of the Na + K + atpase that is required for extruding Na + from the axon and preventing a pathological influx of Na + in both resting and active axons 40. as shown in studies of anoxia, the high intra axonal Na + concentration that results from this failure will cause increased activity of the Na + Ca 2+ exchange channel, with the efflux of Na + requiring a higher degree of Ca 2+ influx 41. This in turn will activate intra axonal proteases, resulting in neurofilament fragmentation and perturbation of axon transport and integrity 42. Second, a key function of the myelin sheath is that it organizes, at the node and paranode, the different voltagedependent channels that are required for saltatory conduction 43,44. Demyelination will result in the loss of saltatory conduction, with continuous conduction now made possible by an adaptive increase in voltage gated Na + (Nav) channel density on the previously internodal axolemma, as discussed above 1. This is achieved by increasing the expression and re distribution of Nav 1.2 channels, which are normally expressed on neurons only before myelination and which allow action potentials 842 NoveMbeR 2008 volume 9

5 Box 2 are opcs stem cells? Oligodendrocyte precursor/progenitor cells (OPCs) were first identified by Martin Raff and colleagues in the early 1980s as proliferating cells that could differentiate into either oligodendrocytes or glial-fibrillary acidic protein (GFAP)-expressing cells called type 2 astrocytes (hence their original name: O-2A cells). Subsequent developmental studies indicated that perinatal O-2A cells were in almost all cases committed to an oligodendrocyte fate (hence the switch to the OPC terminology) and derived from cells with potential oligodendrocyte and neuronal differentiation fates. However, an emerging picture suggests that in both these circumstances the adult OPC might be less restricted in its differentiation potential than its perinatal counterpart. In adult white matter, NG2-expressing OPCs seem to be responsible for producing the new myelin sheaths that are required in the normal aging primate brain 244, whereas in the hippocampus (a neurogenic region of the adult brain) NG2-expressing cells can give rise to new neurons 245, a phenomenon that is mirrored by OPCs in vitro and by transplanted OPCs 73,246,247. There is some evidence that in response to injury these cells might also generate astrocytes 248,249 or even Schwann cells 250, the myelinating cells of the PNS that can remyelinate CNS axons when astrocytes are absent 251. However, these alternative differentiation fates need to be confirmed by lineage-tracing studies. It may therefore be appropriate to abandon the term OPC in favour of glial precursor/progenitor cell or even neural precursor/progenitor cell, terms that better reflect the cells differentiation potential. However, this will be determined by future studies; in the main text, we refer to the cells that are involved in remyelination in the CNS as OPCs. Is there a case for referring to these cells as adult neural stem cells? They are self-renewing and multipotent, criteria that are used by some to assign the term stem cell. Set against this, the absence of asymmetric cellular division in these cells and their rapid proliferative response to injury indicate that they have much in common with progenitor cells that amplify in transit in other stem-cell-containing tissues, such as skin or bone marrow. Experimental autoimmune encephalomyelitis (EAE). An inflammatory disease of the CNS that is generated by inducing an immune response to myelin components such as myelin-oligodendrocyte glycoprotein (MOg) and myelin basic protein (MBP). This induction protocol is most commonly applied in rodents and used as a model to study MS. Oligodendrocyte precursor cell (OPC). The precursor cell that generates oligodendrocytes in the CNS. They themselves are generated in restricted, stemcell-containing regions of the CNS, from where they migrate extensively to the axon tracts that become myelinated. They persist into adulthood and are the cells that are responsible for remyelination. because they have a relatively small Na + conductance 45,46. However, Nav 1.6 channels, which are normally expressed after myelination at the nodes of Ranvier, also become much more widely distributed on demyelinated axons in MS 47. The much greater conductance of these channels causes a pathological increase in intra axonal Na +, leading to protease activation and axon loss. Clearly these differing models (trophic, protective and organizational) of the beneficial effects of remyelin ation are not mutually exclusive, but whether any oligodendrocyte derived trophic factors, such as IgF1 and gdnf, are present is an important question for those interested in preventing axon loss in MS. Drugs that mimic such factors could provide an effective approach for preventing chronic progressive demyelinating diseases such as MS, and an organizational and protective role of the sheath would point to the necessity of developing therapies that target the damaged neuron directly to prevent the rise of intra axonal Na + and/or the subsequent activation of proteases. How does remyelination occur? Oligodendrocyte precursor cells. Remyelination involves the generation of new mature oligodendrocytes. This is evident from two observations: first, that there is a greater number of oligodendrocytes in an area of remyelin ation than in an equivalent area that is not undergoing remyelination 48 ; and second, that remyelination occurs in areas that have been experimentally depleted of oligodendrocytes 49. The question of where new oligodendrocytes come from has been a central theme of remyelination research over several decades. The consensus (reached largely on the basis of data from rodent studies) is now that in most cases remyelination is mediated by new oligodendrocytes derived from a population of adult CNS stem/precursor cells, most often referred to as adult oligodendrocyte precursor cells (or progenitor cells) (opcs) (BOX 2). These multiprocessed proliferating cells are widespread throughout the CNS, occurring in both white matter and grey matter at a density similar to that of microglial cells (5 8% of the cell population) 53,54. various markers are used to identify opcs in their normal physiological state, including, most often, the proteoglycan Ng2 and platelet derived growth factor receptor α (PDgFRα) 49,55 57 and, less often, the transcription factors olig1 (nuclear expression) 58,59, acsl1 (REF. 60) and MYT1 (REFS 49,61). other markers such as o4 and olig2 have also been used, but as these are also expressed by cells in later differentiation stages, their use to unequivocally identify opcs also requires the exclusion of later expressed markers 59,62. adult opcs are derived from their developmental forebears 63, and the two cells share many similarities, although the adult cell has a longer basal cell cycle time and a slower rate of migration 64. Relevant to remyelination, the adult opc can be induced to proliferate and migrate like perinatal cells in vitro by the growth factors PDgF and fibroblast growth factor (FgF) 65, both of which are significantly upregulated during remyelination 66,67. The evidence that opcs are the major source of remyelin ating oligodendrocytes is compelling but indirect: first, retroviral and autoradiographic tracing indicate that dividing cells in normal adult white matter (which are likely but not proven to be adult opcs) give rise to remyelinating oligodendrocytes 68,69 ; second, transplanted opcs remyelinate areas of demyelination with great efficiency ; third, focal areas of demyelin ation in which both oligodendrocytes and opcs die are repopulated by opcs before new oligodendrocytes appear with a temporal and spatial pattern that is highly suggestive of the opcs being the source of the remyelin ating cells 49,74,75 ; and fourth, cells with transitional expression of opc and oligodendrocyte markers can be identified at the onset of remyelination 76. Remyelinating oligodendrocytes can come from the stem and precursor cells of the adult subventricular zone (SvZ), either from the precursor cells that contribute to the rostral migratory stream (RMS) 60,77,78 or from the type b, intermediate filament glial fibrillary acidic protein (gfap) expressing stem cells of the SvZ per se 79. These cells contribute to the remyelination of demyelinated cells in the corpus callosum, a large transverse tract that resides above the SvZ, and their recruitment can be enhanced by environmental enrichment and intranasal delivery of growth factors 80,81. Recent evidence suggests that this region may also be a source of remyelinating cells in MS: examination of post mortem tissue from patients with MS reveals an increased number of, and increased activation of, stem/precursor cells in and around the SvZ 82. This is a phenomenon of considerable biological interest. However, its relevance to demyelinating diseases that produce effects throughout the neuraxis is difficult to assess, first because the contribution of SvZ derived cells may be small relative to that of local NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

6 Box 3 Ms heterogeneity and remyelination Regardless of the perspective from which it is viewed (clinical, immunological, imaging, neurobiological and so on), multiple sclerosis (MS) is a complex disease that varies from patient to patient. One of the more influential classifications of MS types in recent years is based on pathological examination of biopsy and autopsy material 83. This classification identifies four patterns of lesion: pattern I, in which the inflammatory response is dominated by T lymphocytes and macrophages; pattern II, which is characterized by a B-lymphocyte- or antibody-driven pathology; pattern III, which is suggestive of a primary oligodendrocyte dystrophy (for example, Balo s concentric sclerosis); and pattern IV, which is characterized by substantial oligodendrocyte loss that is suggestive of primary, perhaps genetically determined, oligodendrocyte pathology. In patterns I and II there is relative preservation of oligodendrocyte cell bodies and relatively extensive remyelination. Conversely, in patterns III and IV there is extensive loss of oligodendrocytes and remyelination is rare or absent. What are the implications of this classification for understanding myelin regeneration in MS? The correlation between the preservation of oligodendrocytes, albeit shorn of their processes and myelin sheaths, and remyelination suggests that the key to successful remyelination in MS is a disease pathogenesis in which oligodendrocyte cell bodies are preserved. However, as discussed in the main text, experimental data suggest that remyelination is mediated not by surviving oligodendrocytes but by their precursors, the OPCs. This implies that the survival of oligodendrocytes is something of a red herring, and that the critical and thus-far missing component of the classification is the status of the OPCs in the four patterns of lesion. opcs, and second because it seems likely that the SvZderived cells contribution to the repair of white matter tracts, most of which are not as close to the SvZ as the corpus callosum, will be negligible. In contrast to the situation regarding opcs, there is little evidence indicating that intact myelinating oligodendrocytes residing outside areas of demyelination contribute to the formation of new oligodendrocytes, as they seem to be unable to divide or migrate. However, could oligodendrocytes in demyelinated areas, shorn of their processes and myelin sheaths, regenerate these and so remyelinate demyelinated axons? Such cells occur in some MS lesions and in some experimental models 83,84 (BOX 3). However, two lines of evidence suggest that they cannot contribute to remyelination. The first is that there is no remyelination when mature oligodendrocytes are transplanted into experimental models of demyelination 85. The second derives from a relatively complex experimental design in which demyelination is induced by galactocerebroside antibodies and complement, in which many oligodendrocytes cell bodies survive and in which the precursor contribution to remyelination is removed by exposing the white matter to X irradiation 84. In this situation there is no remyelination despite the retention of oligodendrocyte cell bodies in the lesion. In both situations the oligodendrocytes can extend processes, some of which are multi layered and compacted, between but not around the demyelinated axons. both lines of evidence are compelling but carry caveats: it is unknown whether oligodendrocytes that have been maintained in tissue culture or that have been exposed to a very heavy dose of X irradiation, and that therefore are likely to have incurred significant disturbances in cell function, are representative of surviving oligodendrocytes in an MS lesion. although unlikely, the question of whether a surviving oligodendrocyte can contribute to remyelination remains to be resolved. Physiological functions of OPCs. Why are adult opcs so abundant? It seems unlikely that it is simply to provide a population of cells that are available for activation in the case of demyelination. In recent years it has become apparent that at least some quiescent Ng2 positive opcs might have key physiological roles. For example, opcs in grey matter 86,87, and some in white matter 88 90, receive synaptic inputs and, at least in the case of white matter opcs, have been shown to generate action potentials 91. These data reveal a hitherto unexpected degree of crosstalk between opcs and neurons; however, the nature of the function that is subserved by this interaction remains largely unknown. Nevertheless, these studies point to a heterogeneity of normal physiological roles for opcs that echoes heterogeneity in growth factor responsiveness 92 and the developmental heterogeneity of the oligodendrocyte lineage The identification of normal physiological roles of opcs in the adult CNS is a key aim for future studies. Activation, recruitment and differentiation of adult OPCs. In response to injury, local opcs undergo a switch from an essentially quiescent state to a regenerative phenotype (Fig. 3). This activation is the first step in the remyelination process and involves not only changes in morphology 74,96 but also the upregulation of several genes, many of which are associated with the generation of oligodendrocytes during development for example, those that encode the transcription factors olig2, NKX2.2, MYT1 and SoX2 (REFS 76,97 100). opcs are likely to be activated by acute injury induced changes in microglia and astrocytes, two cell types that are exquisitely sensitive to disturbances in tissue homeostasis 101,102, and not necessarily by primary demyelination 103. Microglia and astrocytes, which are themselves activated by injury, are the major source of the factors that induce the rapid proliferative response of opcs to demyelinating injury 49,55, This response is modulated by the endogenous levels of the cell cycle regulatory protein p27kip1 (REF. 109) and is promoted by the growth factors PDgF and FgF and, doubtless, by other factors that are associated with acute inflammatory lesions and that have been demonstrated to have mitogenic effects on opcs in tissue culture. The subsequent population of demyelinated areas by opcs is referred to as the recruitment phase of remyelination; it involves opc migration in addition to the ongoing proliferation 113,114 (Fig. 3). Following recruitment, the opcs must differentiate into remyelinating oligodendrocytes (Fig. 3). This differentiation phase encompasses three distinct steps: establishing contact with the axon that is to be remyelin ated, expressing myelin genes and generating a myelin membrane, and finally wrapping and compacting the membrane to form the sheath. Despite these actions being fundamental capabilities of oligodendrocytes, we still have an incomplete understanding of how axo glial contact is established and how this interaction then regulates, in each individual cell process, the morphological changes that constitute myelination. That said, some molecules have been shown to contribute to the regulation of differentiation, and it is clear that there are 844 NoveMbeR 2008 volume 9

7 a Normal adult white matter d OPC differentiation (axon engagement and myelin shealth formation) b Demyelination and OPC activation c OPC recruitment (proliferation and migration) OPC Activated OPC Oligodendrocyte Microglial cell Astrocyte Macrophage Remyelination failure 1 (OPCs absent) Recruitment factors enhance remyelination Differentiation factors inhibit remyelination Remyelination failure 2 (OPCs present ) Recruitment factors inhibit remyelination Differentiation factors enhance remyelination Figure 3 The phases of remyelination. a Normal adult white matter contains astrocytes, microglia and oligodendrocyte precursor cells (OPCs), in addition to myelinating oligodendrocytes. b OPC activation. Following demyelination (in which oligodendrocytes and myelin are lost) the microglia and astrocytes become activated, which in turn leads to the activation of any OPCs in the vicinity. c Recruitment phase. The activated OPCs respond to mitogens and pro-migratory factors that are generated predominantly by reactive astrocytes and inflammatory cells. The proliferation and migration of the OPCs results in the demyelinated area becoming populated by an abundance of OPCs. Macrophages also start to remove the myelin debris. d Differentiation phase. In the final phase of remyelination the recruited OPCs differentiate into remyelinating oligodendrocytes, a process that involves axon engagement and the formation of a myelin sheath. Remyelination failure can occur because of a failure of OPC recruitment, in which case therapeutic targets should be recruitment factors and not differentiation factors, which would inhibit remyelination. Alternatively it can occur because of a failure of differentiation, in which case differentiation promoters are desired and recruitment promoters would counteract remyelination. Experimental and clinical data suggest that differentiation is the most vulnerable phase of remyelination, and it is during this phase that remyelination generally fails. many similarities between the differentiation of opcs into myelinating oligodendrocytes in development and during the regenerative process. FgF plays a key part in inhibiting differentiation as well as in promoting recruitment, and it thereby regulates the transition from the recruitment to the differentiation phases 112,115. IgF1 is another factor that has major roles in both processes 116. However, differences in the regulation of development and regeneration of myelin do occur: the transcription factor olig1, which is essential for developmental myelination 117, has a less redundant role in opc differentiation during remyelination 58. another example is the Notch signalling pathway: this negative or positive 121 (depending on the ligand) regulator of differentiation in development is redundant during remyelination, as conditional knockout of the NOTCH1 gene from opcs has no effect on remyelination 122. In seeking to identify the sources of the various regenerative factors that promote opc proliferation, recruitment and differentiation, several studies have provided compelling evidence for a key role for the inflammatory response to demyelination. The relationship between inflammation and regeneration in many other tissues is well recognized. However, inflammation s involvement in myelin regeneration has been obscured in a field that has been dominated by the immune system mediated pathology of MS and its various animal models (such as eae) it is unquestionably true that the adaptive NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

8 Heterochronic parabiosis The anastomosis of the circulation of two animals of different ages, to determine whether the presence or absence of factors in the bloodstream is responsible for any changes associated with aging, such as diminished repair capacity. immune response mediates tissue damage in these systems. Nevertheless, several descriptive studies that use experimental models 123,124 and MS tissue 125,126 point to an association between inflammation and remyelination. The role of the innate immune response to demyelination in the context of remyelination has become apparent in part through the use of non immunity mediated, toxin induced models of demyelination. Depletion or pharmacological inhibition of macrophages following toxin induced demyelination leads to an impairment of remyelination , as does an absence of T cells 131. Conversely, growth factor combinations that enhance remyelination following cuprizone induced demyelination enhance the expression of inflammatory cytokines and chemokines 132. Specifically, the pro inflammatory cytokines interleukin 1β (IL1β) 133 and tumour necrosis factor α (TNFα) 134, along with lymphotoxin β receptor (LTβR) 135 and major histocompatibility complex class II (MHCII) molecules 136, have all been implicated as mediators of remyelination in this model. on the other hand, another cytokine, interferon γ (IFN γ), inhibits remyelination 137. Phagocytic macrophages play a critical part in the removal of the myelin debris that is generated during demyelination. This is important because CNS myelin contains proteins that inhibit opc differentiation both in vitro and during remyelination The observation that macrophage activation enhances myelination by transplanted opcs in the myelin free retinal nerve fibre layer points to there being additional, as yet undefined regenerative factors that are produced by macrophages 141. Why does remyelination fail? Non-disease-related factors for remyelination failure. The efficiency of remyelination is affected by the nondisease related factors age 21, sex 142 and genetic background 143. These generic factors will have a bearing on the efficiency of remyelination regardless of the disease process that is involved. Like all regenerative processes, the efficiency of remyelination decreases with age 21,144. This manifests as a decrease in the rate at which it occurs and is likely to have a profound bearing on disease progression (which, in the case of MS, can occur over many decades). The consequences of slow remyelination are compounded by an age associated increase in the vulnerability of demyelin ated axons to atrophy 145. The age associated effects on remyelination are due to a decrease in the efficiency of both opc recruitment and opc differentiation 49. of these two events, the impairment of differentiation is rate determining: following demyelination in old mice, increasing the provision of opcs by overexpressing the opc mitogen and recruitment factor PDgF does not accelerate remyelination 110. The impairment of opc differentiation with age mirrors the failure of oligodendrocyte lineage differentiation that is associated with many chronically demyelinated MS plaques 96, The decline in remyelination efficiency occurs more rapidly in males than in females, but the basis of this sex divergence is not clear 142. The basis of the aging effect is likely to lie in ageassociated changes in both the extrinsic environmental signals to which opcs are exposed in remyelinating lesions and intrinsic determinants of opc behaviour. an impaired macrophage response in aging, associated with a delay in the expression of inflammatory cytokines and chemokines 150, leads to poor clearance of myelin debris and therefore to the persistence of myelinassociated differentiation inhibitory proteins 21,139. Changes also occur in the expression of remyelinationassociated growth factors following toxin induced demyelination that are in accordance with delays in opc activation, recruitment and differentiation and that are illustrative of age associated environmental changes in the remyelination 150,151. both in vitro studies, revealing ageassociated changes in the growth factor responsiveness of adult opcs 152, and in vivo studies, demonstrating slower recruitment into precursor depleted white matter of transplanted old adult precursor cells than of young adult cells 153, are indicative of intrinsic changes in opcs during adult aging. a recent study confirmed these changes, by revealing a critical age associated change in the epi genetic regulation of opc differentiation during remyelination 100 (Fig. 4). Differentiation of opcs is associated with the recruitment of histone deacetylases (HDaCs) to the promoter regions of differentiation inhibitors 154,155. In old animals HDaC recruitment is impaired, resulting in prolonged expression of these inhibitors, delayed opc differentiation and, hence, slower remyelination. This effect can be replicated following the induction of demyelination in young animals with the use of the HDaC antagonist valproic acid. a key question relating to the development of remyelination therapies is the extent to which age associated changes can be reversed. Intriguing experiments on skeletal muscle regeneration using the technique of heterochronic parabiosis provide clear proofof principle that poor regeneration in old animals can be rejuvenated 156. This is supported by the observation that systemic progesterone can induce a small but significant enhancement in remyelination rate in aged male rats 157. Disease-specific factors for remyelination failure. In addition to these generic factors, remyelination could also be incomplete or fail for disease specific reasons. The strongest evidence for remyelination failure is provided by MS, and the subsequent discussion specifically relates to this disease, although the issues discussed could be relevant to other diseases with a demyelinating component. Theoretically, remyelination could fail because of a primary deficiency in precursor cells, a failure of precursor cell recruitment, or a failure of precursor cell differentiation and maturation 113. early speculation on remyelination failure focused on the first of these mechanisms. It was believed that the process of remyelination would itself deplete a CNS area of its precursor cells, so that subsequent episodes of demyelination occurring at or around the same site would fail to remyelinate owing to a lack of opcs. However, data from experimental studies indicate that opcs are remarkably efficient at repopulating regions from which they have been depleted 158, albeit in an 846 NoveMbeR 2008 volume 9

9 a b HDAC AC AC AC Differentiation inhibitors HDAC ON Differentiation inhibitors OFF Inhibitors Inhibitors Activators Inhibitors Activators Myelin genes OFF Myelin genes Figure 4 epigenetic regulation of opc differentiation during remyelination. a Acetylation (AC) of histones in the promoter region of oligodendrocyte progenitor cell (OPC) differentiation inhibitors (HES5 and SOX2) is associated with a chromatin conformation that is permissive for inhibitor expression. These inhibitors prevent the expression of myelin genes that are associated with OPC differentiation. b During remyelination, OPC differentiation is associated with the recruitment of histone deacetylases (HDACs) and the repression of inhibitor gene expression. This in turn favours the expression of myelin genes and OPC differentiation. In older animals, recruitment of HDACs to the promoters of the inhibitory molecules is impaired, resulting in a transcriptional environment that is skewed towards the inhibition of myelin genes. This results in impaired OPC differentiation and delayed remyelination (for more information see REF. 100). ON age dependent manner 153, and that repeated episodes of focal demyelination in the same area neither deplete opcs nor prevent subsequent remyelination 159. The situation might be different, however, when an area of tissue is exposed to a sustained demyelinating insult: in this case remyelination impairment seems to be due, at least in part, to a deficiency in opc availability 123, In the second mechanism, MS lesions fail to remyelinate not because of a shortage of available precursor cells but rather because of a failure of opc recruitment, involving proliferation, migration and repopulation of areas of demyelination. experimental findings from demyelinated areas in which oligodendrocyte lineage cells are absent indicate that this mechanism might account for the failure of remyelination in at least a proportion of lesions. Why lesions should become deficient in opcs is not clear, but one possibility is that they are direct targets of the disease process in the lesion. The identification of patients that generate antibodies against the opc expressed antigen Ng2 supports this possibility 163. a recent study has suggested that opc recruitment into areas of demyelination may fail owing to disturbances in the local expression of the opc migration guidance cues semaphorin 3a and 3F 164. Moreover, in situations in which opcs need to be recruited into lesions from surrounding intact tissue, the size of the lesion will clearly have a bearing on the efficiency of remyelination: larger lesions will require a greater opc recruitment impetus than smaller ones, especially in aging, when older opcs seem to be intrinsically less responsive to recruitment signals 153. The best evidence at present supports the third mechanism, a failure of differentiation and maturation, as several sets of observations based on detecting oligodendrocyte lineage cells in areas of demyelination indicate that this stage of remyelination is the stage that is most vulnerable to failure in MS. The presence of opcs that are apparently unable to differentiate in MS lesions was initially shown with the oligodendrocyte lineage marker o4 (REF. 146) and was subsequently shown with the opc marker Ng2 (REFS 96,147), with PLP (to reveal premyelinating oligodendrocytes) 148 and, most recently, with olig2 and NKX2.2 (REF. 149). even though the density of opcs in chronic lesions is on average lower than in normal white matter, the density can be as high as in normal white matter or remyelinated lesions 149. This shows that opc availability is not a limiting factor for remyelination. one possible explanation for this failure of differentiation is that chronically demyelinated lesions contain factors that inhibit precursor differentiation. First implicated was the Notch jagged pathway, a negative regulator of opc differentiation 118 : Notch and its downstream activator HeS5 were detected in opcs and jagged was detected in astrocytes in chronically demyelinated MS lesions 165. However, the expression of Notch by opcs and of jagged by other cells in lesions undergoing remyelination 122,166 and, more informatively, the lack of a remyelination phenotype in experimental models following conditional deletion of Notch in oligodendrocyte lineage cells suggest that Notch jagged signalling is not an essential negative regulator of remyelination. The ability of inhibitors of γ secretase, an enzyme that is involved in the Notch pathway, to enhance recovery following eae in mice might be indicative of an inhibitory role for Notch signalling in remyelination, but this finding is difficult to interpret given the additional expression of Notch in the inflammatory effector cells 167. Similar results that failed to reveal any essential role for other factors that have been implicated in oligodendrocyte biology, such as the chemokine receptor CXCR2 (REFS 168,169), β1 integrin 170,171 and osteopontin 172,173, have been obtained in remyelination studies using knockout mice. However, potential inhibitory factors have been identified in other experimental and pathological studies. The accumulation of the glycosaminoglycan hyaluronan, an inhibitor of opc differentiation, in MS lesions might contribute to an environment in chronic lesions that is not conducive to remyelination 174. The demyelinated axon itself was implicated when demyelinated axons were shown to express polysialylated neural cell adhesion molecule (PSa NCaM) 175, which inhibits myelination in cell culture 176. Demyelinated axons might be remyelinated less easily than healthy axons are myelinated during development as a result of pathological changes that occur during demyelination or as a consequence of being chronically demyelinated 148. Dystrophic axons in lesions are rarely remyelinated (R.J.M.F., unpublished observations), although it is unclear whether the process of remyelination can lead to the reversal of dystrophic changes in the CNS 35. Certainly, transplantation studies indicate that axons do not lose their ability to be myelinated simply as a consequence of not being myelinated 177. NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

10 Time A A B B Environment C C D OPC Remyelination Figure 5 A schematic representation of the dysregulation hypothesis of remyelination failure. The schematic shows two sets of interlocking cogwheels on two shafts. The cogwheels represent specific cell cell receptor ligand interactions (A A, B B, C C and D D ) that drive remyelination (shown by the rotational movement of the shafts with time). Note that each interaction only drives the process for a defined period of time (for example, A A completes its role before D D starts) and that some interactions (for example, B B and C C ) drive the process at the same time. The model illustrates two important principles. First, that the loss of some interactions (for example, one but not both of B B and C C ) following the removal of either receptor or ligand has no effect on remyelination, as each can substitute for the other. Second, that despite this, the dysregulated expression (shown as additional teeth on cogwheel B in the inset, representing prolonged expression of this ligand) of any one of the receptors or ligands (including those that are functionally redundant) will jam the process and stall remyelination, as the necessary synchrony will be lost. OPC, oligodendrocyte precursor cell. Chronic lesions in MS: a failure of the acute lesion? although many studies in the past few years have concentrated on putative inhibitory signals in their attempt to account for the failure of opcs to undergo complete differentiation in demyelinated MS plaques, an alternative explanation is that these lesions fail to remyelinate because of a deficiency of the signals that induce differentiation. This hypothesis, based on the absence of factors, is difficult to prove but is consistent with a model of remyelination in which the acute inflammatory events have a key role in activating precursors and in creating an environment that is conducive to remyelination (see above). MS lesions are rarely devoid of any inflammatory activity; however, chronic lesions are relatively non inflammatory compared with acute lesions and constitute a less active environment in which opc differentiation might become quiescent. The contrast between inflammatory environments and non inflammatory environments that model chronic lesions in promoting regenerative behaviour has been demonstrated in two recent studies based on the myelinating potential of transplanted opcs 141,178. Chronic lesions usually contain scarring astrocytes, which are hypertrophied and non reactive, in contrast to the reactive astrocytes that are associated with acute lesions and that support remyelination through the production of, inter alia, growth factors 55,66, although these non reactive astrocytes are likely to be the consequence D and not the cause of remyelination failure, their presence might not be helpful when attempting to promote remyelination in chronic lesions. The two possibilities that remyelination failure reflects the presence of negative factors or the absence of positive factors are of course not mutually exclusive. Moreover, it has become apparent from many studies in recent years that there is a multitude of interacting factors, both environmental 182,183 and intrinsic 184,185, that guide the behaviour of oligodendrocyte lineage cells through the various stages of remyelination. efficient remyelination might depend as much on the precise timing of action as on the presence or absence of these factors. In an earlier review we articulated this in a model called the dysregulation hypothesis, in which remyelination failure reflects an inappropriate sequence of events 113 (Fig. 5). although the causes of remyelination failure in such a varied disease as MS are likely to be multiple, we still regard this hypothesis as useful for understanding remyelination failure in most cases. How can remyelination be enhanced? at present, there are no therapies in the clinic that promote remyelination. There are two major approaches that are currently being tested in animal models of demyelination: cell replacement by transplantation (exogenous therapies) and promotion of repair by the resident stem and precursor cell populations in the adult CNS (endogenous repair). We will describe both, but two general points should be considered. First, a complete distinction between the two is artificial, as an important use of cell transplantation may be the delivery of secreted factors to promote endogenous repair. Second, the effectiveness of any remyelination therapy depends on the ability to suppress the effect of any ongoing disease process on the new oligodendrocytes. In diseases such as periventricular leukomalacia (PvL) (BOX 1) this is not an issue, as the ischaemic insult (occurring in utero or at the time of birth) is transient. equally, for the leukodystrophies, such as Pelizaeus Merzbacher disease, cells delivered by transplantation will not have the pathogenic mutation and so will not be subjected to any cell autonomous effects of the genetic defect. For MS, the issue of ongoing damage is a major concern. However, the clinical application of increasingly effective anti inflammatory drugs 186 may provide the necessary disease suppressive background in which remyelination strategies could prove to be effective additions to the drug cocktails that are already used in the management of MS. Enhancing remyelination by cell therapies. The rationale for using cell transplantation to promote remyelination comes from experimental studies initiated in the early 1980s which showed that transplanted glial cells myelinate in the CNS following their introduction into the developing CNS of rodents with myelin mutations or with toxininduced focal demyelination Many studies have used these approaches, and different cell types, including primary opcs 70,72,190, Schwann cells 189,191,192, olfactory ensheathing cells , neural stem cell lines 196 and 848 NoveMbeR 2008 volume 9

11 embryonic stem cell derived glial precursors 197, were all shown to generate myelinating cells following transplantation. In addition to the considerable amount of information that these studies generated regarding the biology of myelination and remyelination, they also provided proofof principle for the clinical application of glial cell transplantation. opc transplantation into shaking pups dogs with PLP mutations that model Pelizaeus Merzbacher disease myelinates axons in the dysmyelinated CNS of these animals 198. Multiple transplantation sites enable widespread myelination throughout the CNS in mice with mutations in myelin basic protein (Mbp) (shiverer mice), resulting, in some cases, in a greatly prolonged lifespan and resolution of the symptoms of myelin dysfunction 199 (Fig. 6). It is not surprising, therefore, that cell transplantation has been suggested as a therapy for a wide range of myelination disorders. However, there are a number of significant concerns that have to be addressed. First, the logic of transplantation: there is clearly little benefit to be gained by transplanting opcs into lesions (such as those that occur in MS) that already contain abundant cells with the ability to generate new oligodendrocytes. Here, the environment is inhibiting (or failing to promote) differentiation and regeneration, and this is also likely to be the case for the exogenous cells. This could be overcome by modifying the cells before transplantation for example, to inhibit the activity of any deleterious component of the diseased environment but in this case it would be more logical to try to manipulate the resident precursor cells. Recent findings which show that the density of oligodendroglia is not reduced in PvL, even in areas of myelin abnormalities revealed by abnormal MbP staining 200, suggest that here too transplantation will provide little benefit. by contrast, the logic for transplantation in genetic myelin disorders is more compelling, as here no normal endogenous precursor cells will be available. Second, the method of delivery: for focal lesions requiring remyelination, single injections might provide a sufficient spread of cells for repair. However, the morecommon scenarios of MS or leukodystrophies multifocal or diffuse disease will require multiple injection sites. given that each implantation carries a small risk of intracerebral haemorrhage, it is clearly desirable to find other methods of generating widespread dispersion of transplanted cells. Two experimental approaches have been developed: intraventricular and intravenous delivery 201,202. Stem and precursor cells injected into the lateral ventricles will enter the CNS, and glial precursor cells will be able to generate myelin forming oligodendrocytes in widely dispersed areas. This therefore seems to be a realistic approach, although the risk of cell aggregates blocking the intraventricular flow of cerebrospinal fluid remains a theoretical concern. Intravenous administration of neural stem cells leads to the amelioration of symptoms in eae, suggesting that these cells can pass through the lungs and then enter the CNS from the cerebral vasculature 201. However, morerecent studies suggest that most, if not all, of this beneficial effect in eae results from the immunosuppressive effect of the neural stem cells on T cell populations in the perivascular space of the CNS 203 and in peripheral lymph nodes 204, rather than from these cells actively contributing to remyelination. The question of whether this highly attractive method of delivery of cells into the CNS is realistic remains to be answered. Third, the source of the cells: primary neural or glial precursors have the advantage of a proven ability to differentiate and myelinate following transplantation and, throughout many years of experimentation in different laboratories, no significantly increased risk of tumour formation has been shown even after growth factor expansion and maintenance before transplantation. There would be problems with the availability of these cells for human therapeutic applications, however, as surgical specimens or fetal material would be required. expansion of Schwann cells or olfactory ensheathing cells ex vivo following a peripheral nerve or peripheral olfactory system biopsy provides a more acceptable source of cells for myelination of CNS axons 195,205 and would also enable autologous transplantation. Stem cell lines represent a potential source of cells, being able to provide unlimited quantities of precursors and (with stem cell banks) the ability to partially match human leukocyte antigen types for each patient. However, the extent to which these stem cells can differentiate into oligodendrocytes remains limited: embryonic stem cell (es cell) differentiation protocols that recapitulate development are likely to be hampered by the very late appearance of oligodendrocytes in normal human neural development. Indeed, the generation of oligodendrocytes requires longer than a month in cell culture with tightly monitored conditions 206. additionally, the use of es cell derived or transformed stem cell lines carries a risk of tumour formation. Enhancing endogenous remyelination. as remyelination can be complete, and as the cells that are responsible for remyelination are abundant throughout the adult CNS (even in demyelinated lesions), a conceptually attractive approach to enhancing remyelination is to target the endogenous regenerative process. This approach is predicated on the principle that if the mechanisms of remyelination can be understood and non redundant pathways described, the causes of remyelination failure, and hence plausible therapeutic targets, will be identified. It is clear from the preceding sections that remyelination failure is associated with either insufficient opc recruitment or, more commonly, failed opc/oligodendrocyte differentiation. However, different and sometimes mutually exclusive biologies underlie these two phases of remyelination. For example, PDgF promotes opc proliferation and migration but, at least in culture, inhibits the final stages of differentiation during which the sheath is formed 207. The implication is therefore that pro recruitment therapies may not promote remyelination where the primary problem is opc differentiation, and vice versa 113 (Fig. 3). at present it is not possible to establish which of the two approaches an individual patient or lesion requires, and so there is a pressing need to develop biomarker and imaging strategies that predict whether pro recruitment or pro differentiation therapies are appropriate. NaTuRe RevIeWS neuroscience volume 9 NoveMbeR

12 a Figure 6 Transplantation of human opcs leads to widespread myelination of the brain in a mouse model of leukodystrophy. Serial sagittal images of immunodeficient shiverer mice transplanted with human oligodendrocyte precursor cells (OPCs) and labelled with antibodies to human nuclear antigen (a) and antibodies to the myelinsheath-associated protein myelin basic protein (b), which is absent in shiverer mice. Figure reproduced, with permission, from REF. 199 (2008) Elsevier Science. Theiler s virus-induced demyelination model A model of demyelination induced by infecting susceptible mouse strains with Theiler s murine encephalitis virus. Demyelination occurs after the acute phase of the disease where viral replication occurs in the CNS, and is associated with viral persistence. The mechanisms of demyelination are thought to include autoimmune destruction of myelin and bystander damage caused by the chronic inflammation in the CNS. b Animal models. a further consideration in the development of remyelination enhancement therapies is the use of appropriate animal models. In the chronically demyelin ated plaques of MS, remyelination is assumed to have failed; hence the desire to engineer an intervention that will reactivate a dormant process. by contrast, in many of the demyelination models that are used to test remyelination enhancement, such as the toxin based models, demyelination does not fail and so at best it is only possible to achieve acceleration of an already effective ongoing process 157,208,209. This problem can in part be overcome in two ways: first, by using aged animals in which the slow rate of remyelination is sub optimal, presenting an opportunity for its enhancement; and second, by modifying standard lesion models in which the endogenous process is compromised, such as the chronic cuprizone model 161. The Theiler s-virusinduced demyelination model has also proven useful for demonstrating remyelination enhancement 210,211. However, assessment of remyelination is especially complicated in eae, in which demyelination and remyelination can occur concurrently. This can make it difficult to distinguish an effect that renders the environment less hostile to remyelination and allows it to proceed at its natural rate from one in which the rate of remyelination is accelerated. For example, systemic delivery of putative remyelination enhancing factors can affect the balance of myelin damage and regeneration through effects on cells other than oligodendroglia, such as those of the immune system. This may account for the discrepancy in the studies of IgF1 (REFS ) and glial growth factor 2 (ggf2) 216,217 administered systemically in eae and delivered locally in non immunity mediated models of demyelination. on the other hand, potential proremyelinating effects of the administration of thyroid hormone that were initially identified using eae have been supported by subsequent studies in non immunitymediated models 218,219. It will be useful to determine whether the highly encouraging effects of antibodies to LINgo1, a negative regulator of opc differentiation in development 220, on remyelination and axon preservation in a rat eae model can be replicated in a non immunitymediated model of demyelination 221. Conversely, the ability of LTβR immunoglobulin (Ig) fusion decoy protein to both inhibit demyelination and apparently promote remyelination in a toxin induced model of demyelination indicates that it can now be tested in a more disease relevant model, such as eae 135. Despite the caveats regarding models and methods of analysis, several recent studies provide proof of principle for remyelination enhancement. an especially intriguing line of investigation has been the identification of polyreactive IgM autoantibodies that react with oligodendrocyte surface antigens and promote remyelination 208,209,222,223. although the mechanisms that underlie this effect remain unclear 224, autoreactive IgM antibodies may constitute a component of endogenous remyelination, and the identification of human monoclonal antibodies has clear therapeutic potential 225. The relationship between pregnancy and MS remission has prompted two studies using toxin induced demyelination that show the remyelination enhancing effects of pregnancy associated hormones 226,227. First, systemic delivery of progesterone, a steroid hormone that is associated with the maintenance of pregnancy, was shown to induce a small but nevertheless significant enhancement in remyelination in aged male rats 157. Second, systemic delivery of prolactin, an adenohypophysisderived hormone that is associated with lactation, enhances remyelination in virgin female mice 227. earlier attempts to promote remyelination in aged animals using the growth factors IgF1 and ggf2 did not prove successful, highlighting the difficulties in reversing age associated decline 214,217. overexpression of PDgF results in an increase in opc recruitment in toxininduced lesions in aged mice but does not result in an improvement in remyelination, indicating that opc differentiation rather than recruitment is the rate limiting component of the aging effect 110. However, recent studies 850 NoveMbeR 2008 volume 9