Importance of oligodendrocyte protection, blood brain barrier breakdown and inflammation for remyelination

Abstract

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system (CNS). A better understanding why remyelination fails in MS is necessary to improve remyelination strategies. Remyelination is mediated by oligodendrocyte precursor cells (OPCs), which are widely distributed throughout the adult CNS. However, it is still unclear whether 1) OPCs detectable in MS lesions survived the inflammatory response and are unable to myelinate or 2) OPC and oligodendrocyte death is primarily responsible for remyelination failure and detectable OPCs entered demyelinated areas from adjacent tissue as the lesion evolves. Remyelination strategies should therefore focus on stimulation of differentiation or prevention of apoptosis as well as establishment of a supportive environment for OPC-mediated remyelination, which may be especially important in chronically demyelinated lesions.

1. Introduction

Demyelination in the central nervous system (CNS) is the pathological hallmark in multiple sclerosis (MS), which is a primary inflammatory process that leads to CNS damage and neurological deficits [1, 2]. Demyelination disables saltatory conduction and leads to loss of neural functions. During demyelination ensheathed axons are unwrapped becoming more vulnerable to environmental stressors. The initial trigger(s) inducing demyelination are still unclear. Acute lesions are believed to begin with phagocytosis of normal myelin sheaths by macrophages in the presence of infiltrating T lymphocytes [3]. This concept is supported by studies of an animal model for MS, experimental allergic encephalomyelitis (EAE), where the myelin sheath is targeted by autoreactive CD4⁺ T cells and lymphocyte infiltration is a primary event [4, 5]. However, there is evidence that at least in some cases of MS oligodendrocyte death is the initial event in MS lesion formation with intact myelin sheaths. Apoptotic oligodendrocytes and activated microglia are found in the absence of inflammation with few or no myelin phagocytes or lymphocytes [6-9]. Death of oligodendrocytes and the formation of new MS lesions could be triggered by viral infection [10], followed by microglial activation and recruitment of virus-specific CD4+ and CD8+ T cells and macrophages into lesions.

2. Oligodendrocyte cell death in multiple sclerosis

In MS, demyelination is accompanied by extensive destruction and loss of oligodendrocytes [11]. Failure to remyelinate during the recovery phase after an acute attack occurs at least partially because of oligodendrocytes death and the inability of OPCs to differentiate into mature myelinating oligodendrocytes [12-15]. From the study of MS tissue we know that some active demyelinated lesions contain no OPCs [16] or only at the margins of the demyelinated area and very few or no OPCs in the center [16, 17]. OPC numbers are also reduced in chronically demyelinated MS lesions often with a change in morphology [15, 17]. Some chronically active demyelinated lesions have a similar OPC distribution compared to active demyelinated lesions with some OPCs at the margins and no or very few OPCs in the center [17].

In acute MS, apoptosis of oligodendrocytes has been documented [8, 9, 18] but has only rarely been seen in chronic MS lesions [19, 20]. Apoptotic cell death of oligodendrocytes has been proposed to be the initial event in new lesion formation and the primary cause of inflammation in acute MS [8]. Barnett and co-workers examined a very early stage in lesion formation in a 14-year-old patient with relapsing-remitting MS within 17 hours of the onset of a new symptomatic and fatal brainstem lesion and concluded that oligodendrocyte death precedes the activation of complement and microglia. This was followed by demyelination and phagocytosis of complement-opsonized myelin sheets by macrophages [8]. In contrast, death of oligodendrocytes by apoptosis in chronic MS lesions and controls is an infrequent event, indicating that the major apoptotic loss of oligodendrocytes is an early event associated with acute lesion formation [21] and less prominent in chronic lesions [22]. Caspase 1 mRNA and protein levels are significantly elevated in OLs in acute and chronic MS, suggesting that caspase-1 may play a role in the inflammatory and apoptotic processes associated with MS pathogenesis [23]. In experimental demyelination models the number of apoptotic oligodendrocytes and severity of EAE are reduced in caspase-1- or caspase-11-deficient animals [24, 25] or in transgenic animals over-expressing the caspase-inhibitory protein p35 [26].

These data clearly indicate the impact of oligodendrocyte apoptosis for the disease and therapeutic approaches in MS. Therefore rescuing mature myelinating oligodendrocytes from apoptosis is likely to be the major goal in preventing MS, and protecting OPCs against apoptotic triggers seems to be a major goal to increase remyelination.

2.1 Molecular triggers of oligodendrocyte death

A. Cellular effectors

Cytotoxic T lymphocytes (CTLs) with a CD8⁺ phenotype are able to recognize and attack major histocompatibility complex (MHC) class I-expressing CNS cells. Most CNS cells can be stimulated to present peptides to CD8⁺ CTLs by MHC class I molecules, and are susceptible to CTL-mediated cytotoxicity in culture [27]. In disease-affected CNS parenchyma, CD8⁺ CTLs outnumber other T-cell subtypes and show clonal expansion in several inflammatory and degenerative CNS diseases, such as multiple sclerosis (MS). The extent of axonal damage in primary MS lesions is correlated to the number of activated microglia/macrophages and cytotoxic CD8⁺ T lymphocytes [28]. MBP-reactive CD8⁺ T cells induce MHC class I restricted cytotoxicity of OLs [29]. CD8⁺ T cells have the potential to induce OL lysis in vivo as a likely consequence of direct antigen recognition [30].

Cultured OLs are not susceptible to MHC class II restricted cell lysis, but expression of CD56 in CD4⁺ effector cells stimulates non-MHC-restricted cell lysis of human OLs [31]. CD4⁺ T cell mediated toxicity requires contact between OLs and T cells and is independent of the cytokine profile of the T cells and can be inhibited by pan-caspase inhibitors and Fas-targeting antibodies [32].

NK cells and γδ T cells are members of the innate immune system using non-MHC-restricted lytic mechanisms to mediate target cell injury and are detectable in MS lesions [33, 34]. Interleukin-2 (IL-2)-mediated activation of NK cells may occur in the inflammatory environment of active MS lesions and could bypass protective effects of self-MHC class I molecules that may be expressed on OLs [35]. γδ T cells have been implicated in the pathogenesis of MS [36] and are cytotoxic toward OL via perforin [37]. γδ T and NK cells interact with OL via NKG2D receptor and a corresponding ligand expressed by OL and mediate cell death via cell lysis or by induction of apoptosis. The effector molecule perforin can induce cell lysis in OL [38] or apoptosis depending on the molecular environment [39].

B. Nitric oxide (NO)

Nitric oxide (NO) is a diffusible and reactive free radical involved in a variety of physiological functions, including control of the vascular tone, platelet aggregation, and the immune response [40, 41] and a major factor contributing to the loss of neurons in ischemic stroke, demyelinating diseases, and other neurodegenerative disorders [42]. NO is involved in glutamate induced excitation with direct effects on BBB permeability, clearence of CNS inflammation, possibly via induction of encephalitogenic T cell apoptosis, and mediating demyelination, oligodendrocyte destruction and injury to axons [43]. NO is synthesized by three different types of NO synthases (NOS): the constitutively expressed nitric oxide synthase (cNOS), the inducible NO synthase (iNOS) and the mitochondrial NO synthase (mtNOS). Within the CNS cNOS is found in neurons and endothelial cells. cNOS is regulated in a calcium-dependent manner via calmodulin [44, 45]. iNOS is expressed in various cell types in response to cytokines IL-1β, IFN-γ and TNF-α [46-49]. NO synthesis is increased during CNS inflammation in ependymal cells, macrophages and astrocytes to levels that are neurotoxic to the CNS environment. Detection of nitrotyrosine is indicative for oxidative cell membrane damage including NO-mediated changes. iNOS is abundantly expressed in astrocytes in MS lesions [50] and in brain biopsies of fulminant MS [51], but not detected in normal human CNS. Increased iNOS expression levels and nitrotyrosine were also detected in chronic active MS lesions in phagocytic macrophages and at the active lesion edge, in and around perivascular lesions, and in periventricular ependymal cells [52]. Labeling with iNOS decreases as lesion age and become less active [53]. In brain biopsies from two acute cases of MS iNOS was detected in both reactive astrocytes and macrophages, whereas no signal was detected in chronic inactive MS lesions [54].

High NO concentration induce apoptosis by activating mainly the mitochondrial apoptotic pathway including modulation of apoptosis-associated Bax and Bcl-2 expression [55, 56], induction of DNA damage and activation of tumor suppressor p53 protein, which in turn induces regulatory gene expression, allowing DNA repair or programmed cell death [57, 58]. NO can function as a direct neurotoxin or may react with superoxide (O₂⁻) by a diffusion-controlled reaction to form peroxynitrite (ONOO⁻), a species that also contributes to oxidative signaling and cellular apoptosis [42]. However, the mechanism by which ONOO⁻ induces apoptosis remains unclear, although subsequent formation of reactive oxygen species (ROS) such as O2⁻, H2O2, OH⁻, and ONOO⁻ has been suggested [42]. These observations support a potential role of iNOS in initiation of MS lesions.

C. Reactive oxygen species (ROS)

In the past few years more evidence has been emerged that reactive oxygen species (ROS) contribute to several mechanisms underlying the pathogenesis of MS lesions. ROS are produced upon interaction of monocytes with brain endothelium, leading to alterations of tight-junction [59], cytoskeleton rearrangement, enhancing the migration of monocytes into MS lesions by reduction of BBB integrity [60] and subsequent leukocyte infiltration into the CNS. Infiltrated leukocytes produce high amounts of ROS leading to macrophage-mediated myelin degradation [61], hydrogen-peroxide- and NO-induced apoptotic cell death of oligodendrocytes [62, 63] and neuronal and axonal damage [64, 65]. Additionally, various antioxidants have beneficial effects in MS therapy, suggesting that ROS play an important role in MS pathogenisis [65, 66]. Extensive oxidative damage occurs to proteins, lipids, and nucleotides in active demyelinating MS lesions, predominantly in reactive astrocytes and myelin-laden macrophages [67]. Counteracting the oxidative stress hypertrophic astrocytes and myelin-laden macrophages express antioxidant enzymes superoxide dismutase 1 and 2, catalase, and heme oxygenase 1, which are markedly up-regulated in active MS lesions compared to white matter from control brains [67]. Enhanced expression of antioxidant enzymes may reflect an adaptive defense mechanism against ROS-mediated cell damage in MS lesions.

D. TNFα

Tumor necrosis factor α (TNFα) is a cytokine involved in systemic inflammation stimulates acute phase reactions and is an important modulator of the immune response. TNF-α is produced by cells of the macrophage lineage, which includes microglia in the central nervous system [68]. TNFα is a well known trigger of apoptosis and necrosis of oligodendrocytes in experimental animal models [69, 70] and in cell culture [71-73]. Inhibition of caspase-1 and caspase-3 protects oligodendrocytes against TNFα-induced apoptosis [71, 72, 74, 75]. More recently, non-caspase dependent mechanisms of TNF-induced cell apoptosis have been proposed. These mechanisms involve mitochondria-derived factors such as apoptosis inducing factor (AIF) [73, 76] and endonuclease G [77, 78]. Due to high TNFα concentrations in active MS lesions vs controls and the fact that TNFα induces apoptosis and necrosis of oligodendrocytes in experimental animal models [69, 70] and in cell culture [71-73] it was suggested that TNFα is a major player to induce OL cell death also in active MS lesions [79]. Under cell culture conditions TNFα is associated with the activation of c-Jun NH2-terminal kinase (JNK) 3 [80, 81] and activation of NFκB [79].

Microglia play a role in mediating the death of oligodendrocytes through the T-lymphocyte stimulated release of TNFα [82], which can be antagonized by interferon β-1b [68]. Since TNF- α neutralizing agents provoke inflammatory demyelination [83], TNF-α homoeostasis is a highly regulated process during inflammation and its maintenance may protect against demyelination.

Despite these results, attempts to treat MS with monoclonal antibodies targeting TNF- α showed worsening of the disease [83, 84]. In addition treatment with monoclonal antibodies against TNF-α caused demyelinating lesions in patients with rheumatoid arthritis, juvenile rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, and Crohn disease [85].

E. Fas ligand

Fas ligand (FasL or CD95L) is a type II transmembrane protein that belongs to the tumor necrosis factor (TNF) family. Fas ligand (FasL) is a homotrimeric protein and signals through trimerization of the Fas receptor (FasR), which spans the membrane of the “target” cell. This trimerization leads to apoptosis, or cell death [86]. The Fas death system has been proposed to contribute to plaque pathogenesis in chronic MS lesions [87, 88]. Microglia and lymphocytes show intense staining for FasL in MS lesions [87, 88]. FasR expression is increased in chronic MS lesions [88] and is restricted to oligodendrocytes [20, 87]. However, Bonetti and co-workers showed in an EAE animal model, that the FasR-FasL system is used to eliminate infiltrating cells and microglia, but not oligodendrocytes in the CNS [20, 89]. In vitro the FasL-FasR system induced caspase activation and cell death in human oligodendrocyte hybrid cells [90, 91]. Supporting the role of caspases in Fas-signaling, mice over-expressing the caspase inhibitor p35 [26] or transgenic animals lacking caspase 11 [24] are resistant to Fas-mediated apoptosis. Overexpression of tumor protein 53 (p53), a transcription factor involved in several cellular processes like apoptosis and cell cycle arrest, resulted in up-regulation of the death receptors FasR, death receptor 4 (DR4) and death receptor 5 (DR5) with subsequent caspase-mediated apoptosis of oligodendrocytes [92]. In addition to its complex functions as a nuclear transcription factor, p53 can act in the cytosol and mitochondria to promote apoptosis through transcription-independent mechanisms [93]. Oligodendrocytes were protected from p53-induced cell death by blocking signaling through FasR and/or tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) receptors [92].

F. TRAIL

TRAIL, also termed Apo2L, is a member of the TNF superfamily and expressed as a type II transmembrane protein (mTRAIL). TRAIL is released from the cell membrane after cleavage by metalloproteinases, yielding soluble TRAIL (sTRAIL). Active TRAIL forms a homotrimer and mTRAIL as well as an engineered soluble form induce apoptosis in several transformed cell lines [94]. The death receptors (DR), TNF-R1 (DR1), and Fas (DR2), are well-characterized members of the group and have been studied previously in MS [20, 87, 88]. DR3 is preferentially expressed by lymphocytes and is induced after T-cell activation [95]. DR4 and DR5 (TRAIL-R1 and TRAIL-R2), are two of five cloned receptors of the TNF-related apoptosis–inducing ligand, TRAIL [96-99]. Two other receptors of TRAIL, DcR1 and DcR2 (TRAIL-R3 and TRAIL-R4), are protective and act as decoy receptors (DcRs) [100, 101]. RT-PCR data indicate that DR6 is abundant in normal human CNS [102]. Oligodendrocytes obtained from human fetal spinal cord expressed DR3, DR4, DR5, DR6, and DcR1 and DcR2 [102] and adult oligodendrocytes from human tissue expressed TRAIL-R1-4 [103]. More recently DR4 and DR6 was not detected in adult CNS tissue [21]. DR3 (TRAIL-R1) is one of the receptors inducing apoptosis in human oligodendrocytes in vitro [103, 104] which is mediated by activation of JNK, but not p38 kinase or ERK1/2 kinases [104]. These data indicate that TRAIL and its receptors TRAIL-R1-4 are responsible in part for apoptosis or oligodendrocyte survival in MS tissue.

G. Interferon γ (IFNγ)

Interferon-gamma (IFN-γ) is a dimerized soluble cytokine that is the only member of the type II class of interferons. IFN-γ is produced predominantly by natural killer (NK) and natural killer T (NKT) cells as part of the innate immune response, and by CD4+ and CD8+ cytotoxic T lymphocyte (CTL) effector T cells [105]. Production of IFN-γ by CD4+T cells results in activation of microglia, which causes these cells to function as antigen presenting cells (APCs) and to produce and secrete IFN-γ by themselves [106].

IFN-γ is a pro-inflammatory cytokine that plays an important role in many inflammatory processes, including demyelination. However, when IFN-γ is released in low levels, it supports the survival of neurons. More recently, IFNγ has been identified to activate the pancreatic endoplasmic reticulum kinase (PERK) thereby activating the ER stress response in OLs [107]. PERK coordinates an adaptive program known as the integrated stress response (ISR) by phosphorylating the α-subunit of eukaryotic translation initiation factor 2 (eIF2α). CNS delivery of IFN-γ before EAE onset ameliorated the disease course and prevented demyelination, axonal damage, and oligodendrocyte loss. However no IFN-γ-mediated beneficial effect could be observed in PERK-deficient animals [107]. Therefore, ISR activation may be considered as a potential therapeutic approach in MS.

Induction of EAE in mice lacking IFN-γ or the IFN-γ receptor as well as administration of anti–IFN-γ antibodies all resulted in more severe disease, suggesting that IFN-γ may limit disease [108-110].

In contrast, increased amounts of IFN-γ interfere with neuronal survival by activating microglia that damage CNS cells via the release of cytotoxic factors such as glutamate, nitric oxide, superoxide and pro-inflammatory cytokines [111]. Systemic administration of IFNγ causes exacerbations in MS patients associated with activation of the immune system [112, 113]. IFNγ also modulates cellular responses in oligodendrocytes by increasing expression levels of FasR in vitro, thereby enhancing their susceptibility to FasL-induced apoptosis [87].

Additionally, TNFα increases the IFNγ-induced death of OPCs, which can be partially suppressed by caspase inhibitors [114]. Oligodendrocytes from caspase-11 knockout mice are less sensitive to IFNγ-induced cell death, supporting the requirement for caspase activation in oligodendrocyte cell death [24]. Moreover, the induction of IFN-expression after demyelinating insults significantly inhibits the remyelination process in EAE [115].

These data demonstrate that IFNγ has both the potential to ameliorate or exacerbate the severity of the disease, which indicates the complexity of different signaling pathways affected by the cytokine.

H. Excitotoxicity

Glutamate is the primary excitatory neurotransmitter in the CNS and is synthesized in all CNS cells as a key metabolite in the citric acid cycle. Glutamate is produced in high concentrations in neurons, packed into vesicles and stored in synapses, from where it can be released into the CNS. This glutamate release is the major extracellular source in the normal CNS. In gray matter astrocytes are major player in glutamate uptake, whereas glutamate homeostasis in white matter is primarily regulated by oligodendrocytes [116]. In MS lesions, microglia expresses increased levels of the glutamate-synthesizing enzyme glutaminase, whereas glutamate transporter expression in OLs is down-regulated [116, 117]. Alterations in the glutamate homeostasis may contribute to OL death both through caspase-dependent and –independent mechanisms [118]. Excessive activation of the ionotropic glutamate receptors expressed by cultured rat oligodendrocytes may induce cell death thereby regulating the size of the cell population. Kainate induced toxicity suggested the involvement of both AMPA and kainite receptors [119]. In an EAE model, blocking of AMPA/kainite receptors improved time course of the disease and increased OL survival [117, 120]. In addition, glutamate sensitizes OLs to complement attack and may contribute to OL death [121]. Excess glutamate may result in cytoskeletal degradation of axons and neurons via increased production of NOS, ROS, arachidonic acid, phospholipase A2, Ca²⁺ -influx and activation of calcium-dependent proteases such as calpain [122]. Thus, imbalanced glutamate homeostasis contributes to the OLGs cell death in MS.

3. What is remyelination?

Remyelination involves reinvesting demyelinated axons with new myelin sheaths resulting in recovery of axon function. Resulting myelin sheaths are often thinner and shorter on remyelinated compared to non-remyelinated axons [123, 124]. The circumference of the axon divided by the circumference of the myelin sheath is called the g ratio. Unusually thin myelin sheaths with high g ratios are the most reliable factor for the identification of remyelination. Remyelination on large diameter axons with high g ratios is much easier to identify than for example remyelination in the corpus callosum, where remyelination on small diameter axons is almost indistinguishable to normally myelinated axons [125].

3.1 Extent of remyelination in MS and animal based MS models

Studies of experimental models of demyelination, induced either by immunological, viral or chemical stimuli, demonstrated rapid and efficient remyelination [123, 126-129] with improved neurological functions [130, 131].

The existences of shadow plaques, which represent fully remyelinated MS lesions, demonstrate that complete repair of MS lesions is possible [132]. However, it is more common to observe a rather limited repair at the edge of lesions [133]. Remyelination seems to be extensive weeks or month after the clinical onset of the disease as detected in biopsies and cases coming to autopsy at these early time points [132, 134-136]. With progression of the disease the extent of remyelination decreases [11], suggesting that chronically demyelinated lesions typically display limited remyelination restricted to the periphery of the MS plaque [133, 137]. A more recent study investigated the extent of remyelination in cerebral tissue from two MS cases. From 168 white matter lesions they identified 22 % shadow plaques, 73 % partially remyelinated lesions and only 5 % completely demyelinated lesions with an average extent of lesion remyelination of 47 % for all white matter lesions [138], suggesting that the extent of remyelination may be more extensive than previously thought.

4. How does remyelination occur?

4.1 Which cells are primarily responsible for remyelination?

Based on data from experimentally induced demyelination in rodents adult oligodendrocyte progenitor cells (adult OPCs) but not surviving mature oligodendrocytes are the major source of remyelinating oligodendrocytes. This view is based on several lines of indirect experimental evidence. First, remyelinating oligodendrocytes in normal adult white matter originate from proliferating cells, which are identified by injecting LacZ expressing retrovirus into normal white matter or by labelling with tritiated thymidine or BrdU. These proliferating cells are likely adult OPCs which has not been proven yet [139-142]. Second, transplanted OPCs are efficiently able to remyelinate demyelinated brain areas [143, 144]. In focal areas of demyelination, devoid of OPCs and oligodendrocytes, OPCs repopulate these areas before mature oligodendrocytes are detectable, indicating that OPCs may be the source of remyelinating cells [127, 142]. In adult tissue OPCs have a characteristic multipolar morphology and express several markers, of which NG2 and PDGFRa are the most commonly used [145-147].

Mature oligodendrocytes outside areas of demyelination do not divide or migrate and therefore are unlikely to contribute to remyelination. Surviving mature oligodendrocytes have been detected in MS lesions and experimentally induced demyelination in rodents [9, 148], some with shorn off processes and myelin sheaths, suggesting that these cells could recover and contribute to remyelination by growing out new processes and myelin sheaths. However, no remyelination could be detected after transplanting mature oligodendrocytes into experimentally demyelinated brain areas [149] or after triggering demyelination with the selective sparing of mature oligodendrocytes; anti-galactocerebroside antibodies and complement in combination with irradiation induced depletion of OPCs [148]. These studies indicate that the contribution of mature oligodendrocytes to remyelination is rather low compared to OPCs. However, anti-galactosylceramide antibody treatment together with irradiation of oligodendrocytes likely causes a very different damage in mature oligodendrocytes compared to conditions in MS lesions. Therefore it is unlikely that mature oligodendrocytes contribute to remyelination as well, but it can not be excluded.

4.2 Activation of adult OPCs

In response to injury, OPCs undergo a switch from an essentially quiescent state to a regenerative phenotype before they differentiate into oligodendrocytes and repair regions of demyelination. This activation step is disturbed in some chronically demyelinated MS lesions that contain quiescent OPCs [13]. Demyelinating lesions open the blood brain barrier (BBB), and recent findings suggest that this might be an important factor in activating the OPC population to induce remyelination [150].

A classical point of view is that microglia and astrocytes become reactive in case of CNS injury [150-153], proliferate [154] and become the major source of factors that stimulate OPC-mediated remyelination. However, the injury response of OPCs differs from that of other glia [150]. In “closed” lesion models, where the blood brain barrier (BBB) is not disrupted, like facial nerve injury or sciatic nerve crush, microglia and astrocytes are rapidly activated [155-162], but no reactive NG2-positive OPCs are found around the lesioned neurons [150]. In contrast, after injury or infection of the CNS with BBB breakdown, OPCs show rapid responses [126, 127, 163-168], displaying an increased expression of chondroitin sulfate proteoglycans (CSPGs), including NG2 and transcription factors Olig2, NKX2.2, MYT1 and SoX2 [169-173] combined with a reactive morphology with hypertrophy of the cell body and processes at the site of tissue damage [152, 167, 174]. This may indicate that BBB disruption enhances OPC activation and therefore facilitates remyelination [150]. However, it is important to mentioned that remyelination is also efficient in cuprizone-induced demyelination models with a completely intact BBB [175], suggesting that OPCs activation occurs via different pathways or might be different in various animal based models.

In cuprizone-induced demyelination remyelination and OL density improved significantly in PDGF-A-overexpressing transgenic mice compared to wild-type mice [170]. OPC density and proliferation were increased in the corpus callosum during acute demyelination but not during chronic demyelination or the subsequent recovery period compared to controls [170]. Additionally, in cuprizone-induced demyelination the intracranially injection of growth factors PDGF-AA, FGF-2, NT3 and IGF-1 enhanced remyelination [239]. However, in GFAP-PDGF-A transgenic mice, where demyelination was induced by intraspinal injection of lysolecithin or dietary administration of cuprizone, morphological analysis of lysolecithin lesions did not reveal any difference in the time course or extent of remyelination between GFAP-PDGF-A and wild-type mice [176], whereas OPC density within lesions was significantly increased with both demyelinating triggers compared to wild-type mice. The extent of remyelination in both lysolecithin-induced and cuprizone-induced demyelination models was investigated because quantitative analysis during early stages of remyelination in the corpus callosum after cuprizone-induced demyelination have been shown to be unreliable [125, 176]. Additionally, factors that have been shown to be mitogenic and chemotactic for OPCs under cell culture conditions (PDGF, FGF-2, EGF, TGF) were unable to activate OPCs in experimental animal models, whereas various injury response-related cytokines (TNFα, TGFβ1, IL-1α, IFNγ), which may come from platelets or macrophages provoked the OPC injury response and increased expression of CSPGs (particularly NG2) [150].

Reactive NG2-positive OPCs undergo rapid cell division within 1–2 mm of the lesion site and accumulate in number for at least 7 days post-lesion [152, 165, 167, 177]. The agent that initiates reactive changes in OPCs is not a component of serum, since freshly extracted serum injected into the CNS did not enhance OPC reactivity [150]. However, whole blood, macrophages and platelets were all shown individually to cause reactive changes in OPCs [150]. The association between reactive behavior of OPCs, BBB breakdown and inflammation, and the surveillance-like functions of OPCs previously proposed [147, 164, 178-181] imply a crucial mechanism linking OPCs with platelet-mediated healing and the inflammatory response to CNS injury.

In general, dysfunction of the BBB is a major hallmark of MS [182, 183] and BBB damage has been shown in acute MS plaques [184] and chronic-progressive MS [185, 186]. A key factor in MS progression appears to be BBB alteration in genetically predisposed individuals, leading to increased vascular permeability and leukocyte infiltration into the brain [187, 188]. Deposition of the serum protein fibrinogen in MS lesions is one of the earliest events in the formation of MS lesions [189-192]. Extravascular fibrinogen that is deposited in tissues upon vascular rupture is not merely a marker, but a mediator of diseases like MS with an inflammatory component and has been shown to activate microglia even in the absence of T cells [193, 194]. Additionally, the extracellular matrix (ECM) proteins vitronectin and fibronectin are detected in MS lesions [195-197]. Similar to fibrinogen, fibronectin and vitronectin bind to integrins and induce microglial activation and expression of matrix metalloproteinase-9 (MMP-9) ([198], which facilitates remyelination in part by processing the inhibitory NG2 proteoglycan [199].

Although the role of NG2 in remyelination has not been investigated and this proteoglycan has previously been found to have an inhibitory effect on axon elongation [200, 201], NG2-positive cells in the developing and adult brain are part of the oligodendrocyte lineage and capable of giving rise to new oligodendrocytes under both normal and demyelinating conditions [202].

4.3. Recruitment, proliferation and differentiation of adult OPCs

Once OPCs are activated, reactive microglia and astrocytes support the proliferative response of OPCs to demyelinating injury by stimulating the production of growth factors platelet derived growth factor-AA (PDGF-AA) and basic fibroblast growth factor (FGF-2) [139, 153, 154, 163, 203], which are up-regulated during remyelination [204, 205]. The number of oligodendrocyte progenitors recruited to the lesions is modulated by the levels of the cell cycle regulatory protein p27Kip-1 [206]. OPC proliferation and migration into demyelinated lesions is regulated in a similar manner to what has been identified by in vitro studies. Neonatal OPCs proliferate in vitro in response to PDGF-AA [207], FGF-2 [208], or insulin-like growth factor-1 (IGF-1) [209]. They migrate in response to PDGF-AA and, to a lesser extent, with FGF-2 treatment [210-212]. PDGF-AA also acts as a survival factor for neonatal oligodendrocyte progenitors and oligodendrocytes [213]. PDGF-AA and FGF-2 also stimulates proliferation of OPCs from adult spinal cord or optic nerve, but cells are less motile and respond at a slower rate than neonatal progenitors [214, 215]. The combination of PDGF-AA and FGF-2 has a striking effect on adult OPCs. OPCs from adult spinal cord or optic nerve migrate and proliferate to the same extent as neonatal progenitor cells [214, 215].

To have an impact on remyelination activated OPCs have to differentiate into oligodendrocytes. Within this differentiation phase the oligodendrocytes have to establish contact with the axon that is to be remyelinated, myelin gene expression has to be induced to generate a myelin membrane and this membrane has to wrap around the axon and compact to form the sheath. The understanding of how axo–glial contact is established and how this interaction regulates myelination is still uncomplete. There are many similarities between developmental myelination and remyelination and some common molecules have been shown to contribute to the regulation of both. FGF-2 seems to play a major role in the inhibition of differentiation and therefore regulates the transition from the proliferating and migrating OPCs into differentiating oligodendrocytes [216-218]. Signaling through the insulin-like growth factor receptor1 (IGF1R) plays a critical role in remyelination via effects on oligodendrocyte progenitors [219]. IGF-1 is a survival factor for oligodendrocytes [220-222] and a differentiation factor for neonatal [209, 223] and adult [224] oligodendrocyte progenitors in vitro. Its levels are elevated within demyelinating and remyelinating lesions in the adult CNS [225-227].

Remyelination is often associated with a prominent inflammatory response in experimental demyelination models and MS lesions [136, 228, 229]. These studies demonstrate an impairment of remyelination in the absence of lymphocytes, MHC Class II antigens, inflammatory cytokines, macrophages, and inhibition of microglial activation [230-237]. From the study of MS tissue we know that remyelination occurs in lesions that contain macrophages [238].

Conversely, in a cuprizone-induced demyelination model which has a minimal inflammatory response the intracranially injection of growth factors PDGF-AA, FGF-2, NT3 and IGF-1 enhanced remyelination [239] and increased the expression of proinflammatory cytokines [240]. Specifically TNFα [237, 240], IL-1β [231, 240], the lymphotoxin β receptor (LTβR) [241] and the major histocompatibility complex class II (MHCII) were elevated [236]. The increased cytokine levels correlate with increased remyelination, but it is very unlikely that they directly stimulate OPC-mediated remyelination (see chapter 2.1).

5. Failure of remyelination

It has been difficult to clarify why remyelination fails due to a lack of suitable experimental model. However, by using novel experimental models, including OPC transplantation, the significance of the relationship between inflammation and remyelination becomes clear. As discussed earlier remyelination is often associated with an inflammatory response in experimental demyelination models and MS lesions, however remyelination is impaired in the absence of lymphocytes, MHC Class II antigens, inflammatory cytokines, macrophages, and inhibition of microglial activation (see chapter 4.3). Additionally, dysfunction of the BBB is an inflammatory response and a major hallmark of MS [182, 183]. In some MS lesions the BBB maybe restored before remyelination is complete, which might prevent further OPC responses and remyelination [150]. Supporting the importance of BBB breakdown for remyelination it has been shown that myelin debris, which is removed by phagocytic macrophages, inhibits OPC differentiation both in vitro and during remyelination [242-244].

In a chronic demyelination model the induction of inflammation is sufficient to induce remyelination [150, 245]. In the absence of inflammation OPCs survive poorly and do not migrate after transplantation into the normal adult CNS [246]. However, in x-irradiated spinal cord, implanted OPCs survive and migrate into both grey and white matter, which demonstrates that the failure of OPC survival and migration in normal adult white matter is due to a nonpermissive environment and not a property of the transplanted cells [246].

In the absence of transplanted OPCs demyelinated areas become repopulated with OPCs from adjacent non-damaged tissue at a rate that depends on the animal's age [247, 248]. Neonatal OPCs were 3-5 times more efficient than adult OPCs to colonize OPC-depleted tissue [249], indicating that the recipients age as well as the OPCs donor age are important factors for effective remyelination [247]. In cases where the demyelinated areas are large and OPCs must migrate over 1-2 months to reach the center of the lesion, Blakemore and co-workers documented a decrease in the remyelination potential of transplanted OPCs over time. Resolution of the inflammatory response limits the remyelination capacity of OPCs [249]. It is possible to increase the remyelination potential of OPCs by adding important inflammatory cytokines. When OPCs are injected into the demyelinated retina together with zymosan, a drug that induces proinflammatory cytokine secretion in immune cells e.g. TNFα-secretion in macrophages [250], remyelination is improved [251].

Chronically demyelinated lesions usually contain nonreactive, scarring astrocytes, which are unable to produce OPC supporting growth factors [163, 205, 252-254]. In a chronic demyelination model, transplanted OPCs did not repopulate areas of chronic demyelination, but they did repopulate chronically demyelinated astrocytosed tissue without remyelination [245], which is similar to the situation found in some chronically demyelinated MS lesions. Induction of acute inflammation in these demyelinated, but OPC repopulated areas resulted in remyelination [245], suggesting that it might be possible to induce remyelination in chronically demyelinated MS lesions. These results together with data from MS lesions [8] (see chapter 2) suggest that at least in some cases OPC death due to a hostile, non-supporting environment is the primary reason why MS lesions fail to remyelinate. If OPCs in acute MS lesions were protected against cell death their activation, proliferation and differentiation would be supported by the inflammatory process, thereby improving remyelination. As an alternative OPCs in chronically demyelinated lesions are the result of cells that have entered the area after the inflammatory burst. They remain inactivated and unable to remyelinate [255]. They are not OPCs that survived the primary demyelination insult and are unable to differentiate into myelinating oligodendrocytes [256].

6. Promotion of remyelination

So far no clinical therapy is available that promotes remyelination. There are at least two major approaches in animal models of demyelination to stimulate remyelination. One strategy is based on cell replacement by transplanting OPCs into areas of demyelination (exogenous therapies) while the others aim to stimulate resident precursor cell populations to repair demyelinated lesions (endogenous therapies).

A. Exogenous therapies

Concerning cell transplantation there are many different sources of cells available that can be potentially differentiated into myelinating cells including primary OPCs [143, 257, 258], Schwann cells [259-261], neural stem cell lines [262] and embryonic stem cell derived glial precursors [263]. These studies provided evidence that transplanted cells can contribute to remyelination. However, as mentioned earlier (chapter 2), at least some MS lesions already contain a considerable amount of OPCs that due to their non-supportive environment are unable to myelinate. Therefore, there may be very little benefit to transplant OPCs into these MS lesions. Especially in case of chronically demyelinated lesions it seems to be inevitable to re-establish a supportive environment for remyelination before cell transplantation is considered as a choice to repair demyelinated areas. Additionally, all potential sources of cells for transplantation have technical or ethical problems. Peripheral stem cells from bone marrow that generate peripheral glia are attractive candidates for cell transplantations into the brain, as autografting would be possible, but the incompatibility of Schwann cells with astrocyte-containing tissue limits the feasibility of using these cells in MS [264]. Multipotent neural progenitor cells with features of somatic stem cells can differentiate into neurons, astrocytes and oligodendrocytes and are found in the adult brain [265]. They support neurogenesis, can be expanded extensively in vitro, survive in the CSF and have been tested to induce recovery in EAE [266], which makes them a potential source for cell transplantation in various human CNS diseases [141, 267]. Intraventricular and intravenous delivery techniques of stem and precursor cells have been developed that allows these cells to enter the CNS and to repair demyelinated areas [266, 268]. However, this beneficial impact on remyelination in EAE seems to be an indirect immunomodulatory effect of neural stem cells on T cell populations in the CNS and peripheral lymph nodes rather than a direct contribution of these cells to remyelination [269, 270]. Sourcing and ethical problems are a major concern in the use of fetal or adult brain grafts, which contain large numbers of neural precursor cells. Pluripotent embryonic stem cells can be differentiated into oligodendrocytes via neural precursor cells [271]. In addition, oligodendrocytes from embryonic stem cells require at least a month in cell culture under strictly controlled conditions and transformed stem cell lines carry a risk of tumour formation [272].

B. Endogenous therapies

Enhancing endogenous remyelination is conceptually a very attractive approach with abundant cells throughout the adult brain that are capable to myelinate. Stimulating physiological repair mechanisms is an important therapeutic goal in MS and other demyelinating diseases. Several promising approaches are currently being explored including antibodies to LINGO-1, remyelination promoting IgM antibodies and hormone treatments.

LINGO-1 is a cell surface protein and a component of the Nogo-66 receptor/p75 signaling complex only expressed in neuronal tissue that has been implicated in the inhibition of myelination [273, 274] and regulated by NGF [275]. The inhibitory action is achieved through RhoA-GTP upregulation in response to the presence of MOG, MAG or Nogo-66 in the CNS [274]. LINGO-1 antagonists promote OPC differentiation and myelination in vitro and accelerate remyelination after lysophosphatidylcholine- or cuprizone-induced demyelination [276] as well as in a rat EAE model [277]. This remyelination is associated with functional recovery of conduction velocities in demyelinated axons. It is suggested that LINGO-1 functions as an inhibitor of OPC differentiation. Blocking LINGO-1 is a very encouraging approach to promote remyelination.

Several human and mouse monoclonal IgM antibodies have been identified, that promote substantial remyelination in several animal models of MS [278-283] A common feature of these antibodies is their ability to bind to the surface of oligodendrocytes and myelin, which is necessary, but not sufficient, for induction of remyelination in vivo. Although the precise molecular mechanisms for this effect remain unclear [284], a human remyelination promoting IgM prevents apoptosis in CG4 cells [81]. These autoantibodies may constitute a component of the endogenous remyelination process [279]. A recombinant form of a human IgM (rHIgM22) identified from a patient with Waldenström macroglobulinemia is under GMP development and will soon enter Phase I clinical trials.

The observation that women with MS are protected against relapses during pregnancy [285] initiated two studies showing that the lactation associated hormone prolactin enhances remyelination in toxin induced demyelination models [286]. Due to increased prolactin levels during pregnancy, the numbers of OPCs in the CNS are increased during pregnancy. Prolactin receptors are present on OPCs and their activation by prolactin stimulates the maturation of these cells into myelin-competent oligodendrocytes [286]. Another class of hormones, the thyroid hormones, has been shown to promote differentiation of OPCs and to induce remyelination after cuprizone-induced demyelination [287, 288]. Overexpression of growth factor PDGF in toxin induced demyelination models induced a strong recruitment of OPCs into demyelinated areas, but without improved remyelination [176]. Overexpressing PDGF in a chronic demyelination model resulted in reduced apoptosis in the corpus callosum during the recovery period. Therefore, PDGF may support oligodendrocyte generation and survival to promote remyelination of chronic, but not acute, lesions [170]. Antagonizing the inhibitory effect of FGF-2 on OPC differentiation might be a future therapeutic approach to enhance endogenously remyelination, because FGF2-/- mice created a sufficiently permissive lesion environment that led to effective remyelination after cuprizone-induced remyelination in mice [289].

7. Expert commentary and Five-year view

Future prospects over the coming years for repair of MS lesions are encouraging, especially with respect to remyelination. Several promising therapeutic approaches are close to clinical trials, including cell transplantation, remyelination promoting antibodies and hormone therapy. However, promoting remyelination in the adult CNS and especially in chronically demyelinated lesions with its hostile, non-permissive conditions is a very difficult task. Stimulating repair in MS lesions is challenging and the choice of animal model in which the therapy is designed will influence the choice of clinical trials in humans. Protecting OPCs and oligodendrocytes from inflammatory mediated cell death in acute MS lesions may be a powerful approach to enhance remyelination. There clearly is a need to identify additional inhibitory or missing factors responsible for the failure of remyelination. This will likely lead to combination therapies for remyelination e.g. cell transplantation along with cytokines or immune modulation. The significant advances made in remyelination biology in recent years provide reasons for optimism.

Key issues.

• Apoptotic cell death of oligodendrocytes appears to be the initial event in new MS lesion formation and the initiator of inflammation in acute MS.

• Failure of remyelination after an acute attack occurs because of oligodendrocyte and OPC death, and the inability of OPCs to differentiate into myelinating oligodendrocytes.

• OPCs in chronically demyelinated lesions are the result of cells that have entered the area after the demyelinating process is complete. They remain dormant and unable to remyelinate until activated by the inflammatory milieu.

• Transplantation of OPCs into chronically demyelinated lesions may not be necessary to induce remyelination because they are already present in the lesion.

• OPCs must be activated before they differentiate into myelinating oligodendrocytes.

• Demyelinating lesions open the blood brain barrier (BBB), which may activate the OPCs to induce remyelination.

• Invasion of macrophages or platelets into demyelinated lesions causes NG2 expression of OPCs by secretion of proinflammatory cytokines.

• Resolution of the inflammatory response limits the remyelination capacity of OPCs.