Remyelination in animal models of multiple sclerosis: finding the elusive grail of regeneration

Remyelination biology and the therapeutic potential of restoring myelin sheaths to prevent neurodegeneration and disability in multiple sclerosis (MS) has made considerable gains over the past decade with many regeneration strategies undergoing tested in MS clinical trials. Animal models used to investigate oligodendroglial responses and regeneration of myelin vary considerably in the mechanism of demyelination, involvement of inflammatory cells, neurodegeneration and capacity for remyelination. The investigation of remyelination in the context of aging and an inflammatory environment are of considerable interest for the potential translation to progressive multiple sclerosis. Here we review how remyelination is assessed in mouse models of demyelination, differences and advantages of these models, therapeutic strategies that have emerged and current pro-remyelination clinical trials.


Introduction
Multiple sclerosis (MS) is a chronic demyelinating, inflammatory and neurodegenerative disease of the central nervous system (CNS). Remyelination is a regenerative process by which oligodendrocytes restore myelin sheaths to demyelinated axons. Evidence from animal models indicate that remyelination can restore neuronal conduction (Smith et al., 1979(Smith et al., , 1981, promote functional recovery (Jeffery and Blakemore, 1997;Jeffery et al., 1999;Liebetanz and Merkler, 2006;Duncan et al., 2009;Mei et al., 2016a) and protect axons from degeneration (Irvine and Blakemore, 2008;Mei et al., 2016a). Myelination and connection of the oligodendrocyte-axonal unit provides metabolic support and reciprocal signaling that promotes axonal function and survival (Nave, 2010;Simons and Nave, 2015;Thornton and Hughes, 2020;Duncan et al., 2021).
In animal models, aging influences oligodendroglial properties (Koutsoudaki et al., 2020;Perdaens and van Pesch, 2021;Zhang et al., 2021;Rawji et al., 2023) and progenitor-driven remyelination declines with aging (Shields et al., 1999;Sim et al., 2002) and can vary depending on regional origin of OPCs (Crawford et al., 2016). Impaired remyelination with aging can be rejuvenated with young myeloid-derived macrophages (Ruckh et al., 2012) and with fasting or metformin treatment . Chronological aging is strongly associated with the development of clinical and pathological features of progressive MS (Zuo et al., 2022;Graves et al., 2023). Longer disease duration and older age is associated with higher numbers of inactive lesions and smoldering lesions (Frischer et al., 2015). Microglial activation and cortical demyelination found in progressive MS tissue (Kutzelnigg et al., 2005;Howell et al., 2011) as well as B cell follicular structures and meningeal inflammation (Howell et al., 2011) are all factors that may influence remyelination capacity in progressive MS and are not well recapitulated in mouse models.
Animal studies have revealed that newly born oligodendrocytes efficiently generate myelin sheaths (Bacmeister et al., 2020;Neely et al., 2022) and while mature surviving oligodendrocytes can generate myelin sheaths they are less efficient (Bacmeister et al., 2020;Neely et al., 2022;Mezydlo et al., 2023) and rarely restore internodes (Mezydlo et al., 2023). The differences between remyelination capacity of surviving mature oligodendrocytes and newly generated oligodendrocytes in response to demyelination likely has implications in which subsets of oligodendroglia are capable of remyelinating in human MS (Franklin et al., 2021). Radiocarbon dating with the genomic integration of 14 C has been used to investigate the age of oligodendroglia within MS lesions and this study indicated limited production of new oligodendrocytes within shadow plaques that may have undergone remyelination suggesting that mature surviving oligodendrocytes contribute to subsequent remyelination in human MS lesions (Yeung et al., 2019). Post-mortem tissue analysis is limited by the inability to determine lesion age and extent of remyelination, and assumptions made about human oligodendrocyte progenitor properties, such as for carbon dating studies that OPCs must divide (and incorporate 14 C) prior to differentiation despite rodent studies indicating that OPCs can directly differentiate into mature oligodendrocytes without cell division (Hughes et al., 2013). The contribution of newly born oligodendrocytes compared to surviving mature oligodendrocytes to remyelination in human MS will be difficult to determine definitively based on human post-mortem tissue analysis.
Remyelination and pro-regenerative strategies remain a major unmet need in the treatment of MS. Studying remyelination in human MS patients has been impaired by the limited array of tools to measure remyelination in humans and improved clinical measures need to be developed for incorporation into clinical trials to facilitate testing of remyelination therapies (Hill et al., 2022). While many animal models exist for investigating remyelination, this review will focus primarily on in vivo rodent models in which remyelination has been clearly demonstrated while highlighting limitations and advantages to consider in relation to human MS and a discussion of remyelinating therapies in clinical trials for MS.

Identifying remyelination
Remyelination in animal models is determined through assessment of myelin sheath thickness and internode length, which is based on the early observations from animal models that remyelinated myelin sheaths are thinner than expected for axonal diameter (Blakemore, 1973(Blakemore, , 1974Bunge et al., 1961; Figure 1A) and internode length is shorter (Gledhill and McDonald, 1977). Myelin sheath thickness is best quantified by transmission electron microscopy (TEM) of 50-90 nm resin embedded sections and g-ratio analysis (ratio of myelinated axon to the axon alone) (Blakemore and Franklin, 2008). Remyelination of small diameter axons is difficult to determine as small axons with thinner myelin sheaths at baseline are not discernable from remyelinated axons (Stidworthy et al., 2003;Bai et al., 2016). Utilization of 3D-TEM techniques have demonstrated that reduced internodal length can be used as a readout of early remyelination in the corpus callosum, however, with time internodal length is restored to distances seen pre-demyelinating insult (Bai et al., 2016). Areas of remyelination can also be difficult to determine over time as myelin sheath thickness and remodeling closely resembles the surrounding normal appearing white matter (Neumann et al., 2020). Longitudinal intravital microscopy of oligodendroglia (Hughes et al., 2018;El Waly et al., 2020;Orthmann-Murphy et al., 2020;Bottes and Jessberger, 2021;Call and Bergles, 2021; Figure 1B) and genetic labeling strategies for lineage tracing of oligodendrocyte progenitors and subsequent myelin sheath generation (Mei et al., 2016a;Figure 1C) are alternative methods that have been utilized for assessment of remyelination and dynamics and patterns of myelin sheath formation. The complexities of accurately assessing remyelination need to be carefully considered when using animal models.

Toxin-mediated demyelination models
Experimental models of demyelination based on the use of toxins, while these models may not recapitulate the autoimmune pathobiology of MS, they offer the advantage of stereotyped demyelination and remyelination process which has been invaluable in the investigation of the molecular mechanisms involved in remyelination.

Focal toxins-lysolecithin and ethidium bromide
Focal demyelinating agents can be injected into the spinal cord dorsal and ventrolateral funiculi, caudal cerebellar peduncle, corpus callosum, optic nerve and subcortical white matter to create focal demyelinated lesions. Focal toxin models offer the advantage of a synchronized short demyelinating process compared to systemic toxins such as cuprizone in which demyelination is protracted and occurring in an environment that may be influenced by continued toxin exposure.
Lysolecithin or lysophosphatidylcholine (LPC) is an endogenous lysophospholipid that can be used to generate a focal demyelinating lesion by injection into white matter tracts and demyelination results through disruption of oligodendroglial cell membranes (Hall, 1972;Plemel et al., 2018) leading to oligodendroglial cell loss (McKay et al., 1998;Plemel et al., 2018). Lysolecithin-mediated demyelination is not selective for oligodendroglia. Astrocyte loss and markers of axonal injury are notable within the lesion core (Plemel et al., 2018). Generation of larger lesions or poor surgical technique can result in axonal degeneration (Blakemore and Franklin, 2008). In young animals remyelination occurs rapidly and the majority of axons are remyelinated by oligodendrocytes and Schwann cells derived from OPCs (Zawadzka et al., 2010). After the primary myelinopathy injury a secondary inflammatory response occurs which is characterized by microglial and macrophage infiltration and activation and reactive astrogliosis (Plemel et al., 2018). T cells also appear to play a role in lysolecithin-induced remyelination with Rag1 knockout and depletion of CD4 and CD8 T cells demonstrating impaired remyelination (Bieber et al., 2003).
In the lysolecithin model, myelin (Kotter et al., 2006) and extracellular matrix components such as chondroitin sulfate proteoglycans (CSPGs) (Lau et al., 2012) within the lesion can impair oligodendrocyte differentiation. Infiltrating macrophages play an important role in influencing OPC recruitment into lesions (Kotter et al., 2005) and depletion of macrophages during early stages of remyelination impairs remyelination (Kotter et al., 2001). Oligodendroglia in remyelinating lesions express transforming growth factor (TGF) β member activin-A and activin-A released from inflammatory microglia and macrophages may promote oligodendrocyte differentiation and remyelination (Miron et al., 2013). Ethidium bromide, a DNA-intercalating agent, injected into white matter tracts creates a focal demyelinating lesion with a larger area of demyelination compared to lysolecithin (Blakemore and Franklin, 2008). Ethidium bromide, opposed to a primary myelinopathy observed in lysolecithin, is directly cytotoxic and induces cell death of oligodendrocytes and astrocytes (Blakemore, 1982). Vacuoles and splitting within the myelin sheath lamellae can occur (Yajima and Suzuki, 1979). Schwann cells have a notable contribution to remyelination in ethidium bromide lesions (Blakemore, 1982;Graça and Blakemore, 1986;Reynolds and Wilkin, 1993;Woodruff and Franklin, 1999). Ethidium bromide injections in the rat cervical spinal cord have been useful in functional studies assessing the role of remyelination on neuronal function (Jeffery and Blakemore, 1997) and cerebellar peduncle injections have facilitated investigation of repeated demyelinating events within the same site (Penderis et al., 2003).
Manipulation of astrocyte and microglial responses after cuprizone-mediated demyelination can influence remyelination. Ablation of astrocytes after chronic cuprizone treatment results in improved oligodendrocyte density, remyelination and motor functional outcomes (Madadi et al., 2019). Astrocyte secretion of cytokines such as tumor necrosis factor alpha and lymphotoxins can influence demyelination and remyelination (Arnett et al., 2001;Plant et al., 2005Plant et al., , 2007. Astrocytes can also promote microglial accumulation and activation (Skripuletz et al., 2013). Microglial depletion reduces cuprizone-mediated demyelination and injection of colony stimulating factor 1 (CSF1) induces focal demyelination (Marzan et al., 2021) suggesting an important role for microglia in mediating demyelination during cuprizone ingestion. Microglial MER proto-oncogene tyrosine kinase (MERTK) signaling (Shen et al., 2021), triggering receptor expressed on myeloid cells 2 (TREM2) signaling (Cantoni et al., 2015;Poliani et al., 2015;Cignarella et al., 2020) and colony stimulating factor 1 (CSF1) signaling (Laflamme et al., 2018) facilitate myelin debris clearance Frontiers in Molecular Neuroscience 05 frontiersin.org and remyelination after cuprizone-mediated demyelination. Trem2 deficiency results in reduction of a subset of oligodendroglia that is induced in response to demyelination which may be due to delay in myelin debris clearance and induction of this subset of oligodendroglia (Hou et al., 2023). The role of T cells in cuprizone-mediated demyelination is unclear. T cells are present in cuprizone ingestion (Remington et al., 2007) and CD8 T cells accumulate in the corpus callosum during cuprizone treatment and express activation markers (Kaddatz et al., 2021). Interleukin-17 (IL-17) secreted by T cells during cuprizone may play a role in activating microglia and mediating demyelination (Kang et al., 2012;Zimmermann et al., 2018).
The major advantages of toxin models are the robust remyelination response with stereotyped kinetics, separation of the demyelinating process from the regenerative process, minimal axonal degeneration, and decline of remyelination with aging that have allowed for discovery of targets that accelerate repair in an aged environment. While focal toxin models have these advantages, the short demyelinating insult and robust remyelination response are limitations. Diffuse CNS demyelination and white matter injury induced by cuprizone ingestion offers the advantage of an environment with prolonged oligodendroglial loss, neuronal stress and subsequent neurodegeneration, which may allow for the investigation of pathways that prevent neuronal degeneration and allow for the ability to assess motor (Manrique-Hoyos et al., 2012;Lubrich et al., 2022) and physiological outcomes (Bando et al., 2008;Cordano et al., 2022;Marenna et al., 2022). One of the potential limitations of toxin-induced demyelinating models is the absence of inflammatory niches and microenvironments found in human MS lesions such as ectopic lymphoid follicles (Negron et al., 2020) and absence of robust T and B cell responses in these toxinmediated mouse models. However, progressive MS lesions have a paucity of inflammatory infiltrates (Frischer et al., 2009) and mechanisms involved in neurodegeneration and remyelination in progressive MS may be independent of ongoing inflammatory activity that may be best modeled in chronic demyelination models such as cuprizone with the presence of late neurodegeneration and motor deficits (Manrique-Hoyos et al., 2012). Glial heterogeneity present in toxin-mediated models (Hou et al., 2023) should be compared to glial transcriptional heterogeneity found in MS tissue (Jäkel et al., 2019;Schirmer et al., 2019;Absinta et al., 2021) to better delineate whether toxin-mediated models recapitulate glial populations present in MS lesions.

Experimental autoimmune encephalomyelitis models
Experimental autoimmune encephalomyelitis (EAE) models are one of the most commonly used models to investigate the immunopathogenesis of MS (Gold et al., 2006;Baxter, 2007;Denic et al., 2011;Rangachari and Kuchroo, 2013;Simmons et al., 2013;Lassmann and Bradl, 2017). Immunization with emulsions of CNS tissue evolved into immunization with encephalitogenic antigens or adoptive transfer of myelin specific T cells (Baxter, 2007). EAE models can be generally classified as active (immunization of CNS peptides) or passive (adoptive transfer of encephalitogenic T cells) and they offer different advantages for the investigation of disease mechanisms of repair and remyelination.

Active immunization
Immunization with a CNS antigen and adjuvant is used to induce active EAE in rodents and the combination of peptide and mouse strain used influences the disease course and pathology (Gold et al., 2006;Rangachari and Kuchroo, 2013;Simmons et al., 2013;Lassmann and Bradl, 2017). C57BL/6 mice were found to be susceptible to EAE with immunization with myelin oligodendrocyte glycoprotein (MOG) 35-55 peptide (Mendel et al., 1995) which has facilitated the use of transgenic lines in investigation of disease mechanisms. In the MOG 35-55 EAE model, demyelination is secondary to axonal injury and degeneration mediated by the adaptive and innate immune system (Kim et al., 2006;Soulika et al., 2009;Nikić et al., 2011) and does not involve a cytolytic auto-antibody response (Bourquin et al., 2003) which needs to be considered when utilizing this model for investigating mechanisms of demyelination and neurodegeneration. For modeling primary antibody-mediated demyelination in immunization models, C57BL/6 mice immunized with human MOG (Oliver et al., 2003) or rats with MOG 1-125 peptide can be used to generate lesions with primary demyelination and axonal sparing (Storch et al., 1998). Cortical subpial demyelination can be modeled in rats with MOG immunization (Storch et al., 2006) or sub-clinical MOG immunization followed by injection of tumor necrosis factor (TNF) and IFN-γ overlying or in superficial cortical layers Gardner et al., 2013).
A limited degree of remyelination has been demonstrated in active EAE by the use of genetic lines labeling myelin sheaths generated from OPCs (Cspg4-CreER™; Mapt-mGFP) (Mei et al., 2016a) and transmission electron microscopy g-ratio analysis ( Figures 1A,D). Significant axonal loss occurs during EAE and a large proportion of axons remain demyelinated during late EAE, 26% in spinal cord ventral white matter analyzed in this study (Mei et al., 2016a). Despite the significant degree of neurodegeneration and modest amount of remyelination, conditional knockout of M1 muscarinic receptor in oligodendrocytes ( Figure 1D; Mei et al., 2016a), pharmacological antagonism of muscarinic receptors (Cordano et al., 2022) and K-opioid receptor agonism (Du et al., 2016) were able to enhance remyelination in MOG 35-55 EAE. Visual evoked potentials can be used as biomarker for remyelination in EAE and cuprizone (Cordano et al., 2022). Immunization with proteolipid protein (PLP) 139-151 peptide in SJL/J mice results in an acute phase of clinical disability similar to MOG 35-55 but with partial clinical remission followed periods of Frontiers in Molecular Neuroscience 06 frontiersin.org fluctuating clinical scores in what has been termed "relapsing remitting" clinical course (Rangachari and Kuchroo 2013). Remyelination has been demonstrated by electronic microscopy at acute phase peak disease course with treatment with muscarinic antagonist benztropine in this EAE model (Deshmukh et al., 2013). The relapsing remitting clinical course of the SJL/J PLP 139-151 EAE model may be useful for modeling remyelination in relapsing remitting MS however the degree of axons with high g-ratios, either demyelinated or remyelinated, normalizes during remission phase (Deshmukh et al., 2013) and whether this is due to a combination of loss of demyelinated axons or subsequent remyelination is unclear. Electrophysiological analysis of visual evoked potentials could be used as a biomarker for remyelinating in different clinical phases of the SJL/J PLP 139-151 EAE model but ongoing inflammatory activity and neurodegeneration may limit this assessment. Active immunization models have many features that differ from human MS pathobiology and clinical course. The robust inflammatory response and resulting secondary demyelination, early neurodegeneration (Jones et al., 2008), sparing of cerebrum, and predominance of CD4 T cells and lack of significant CD8 T cell infiltrate are all features of active EAE that do not correlate well with human MS pathobiology. The monophasic chronic disease course of MOG 35-55 EAE does not resemble the clinical course of relapsing remitting or progressive MS. The high degree of neurodegeneration and minimal remyelination after the acute EAE phase represent challenges for the utilization of these models in assessing remyelination strategies.

Adoptive transfer
Encephalitogenic T cells from CNS antigen immunized mice or myelin-specific T cells isolated from T cell receptor transgenic lines can be used to induce neuroinflammatory disease upon adoptive transfer into naive hosts (Paterson, 1960;Ben-Nun et al., 1981). A major advantage of adoptive transfer models is the avoidance of the priming and immune expansion phase that occurs in the periphery in immunization models that can contribute to the neuroinflammatory response. Adoptive transfer of myelin-specific CD4 T cells has been the most widely used model to generate EAE. Adoptive transfer of myelin basic protein (MBP)-specific CD8 T cells from C3H mice can induce EAE (Huseby et al., 2001). Through the utilization of transgenic lines in which CNS resident cells express foreign antigens, EAE can also be induced by adoptive transfer of CD8 T cells specific for the foreign antigen (Cabarrocas et al., 2003;Na et al., 2008;Saxena et al., 2008). These models may offer the advantage of induction of primary demyelination directly targeting oligodendrocytes (Na et al., 2008;Saxena et al., 2008) which may facilitate the exploration of mechanisms involved in CD8-mediated pathogenesis in MS, as CD8 T cells outnumber CD4 T cells in MS lesions (Booss et al., 1983;Hauser et al., 1986;Babbe et al., 2000) and CD8 infiltrates correlate with the degree of axonal degeneration in MS lesions (Bitsch et al., 2000;Kuhlmann et al., 2002).
Adoptive transfer of myelin-reactive Th17 cells after acute cuprizone ingestion has been used as a model to investigate oligodendroglial responses in the setting of T cell mediated inflammation (Baxi et al., 2015(Baxi et al., , 2017Kirby et al., 2019) and has the advantage of perivascular inflammatory and corpus callosum infiltrates and may allow for investigation of therapeutics that target pathways activated in oligodendroglia in the context of inflammation. Whether adoptive transfer models can be used to investigate remyelination remains to be determined, but these models offer the advantage of the ability to bypass the priming phase of active EAE and investigate therapeutic effects of compounds or genetic manipulations in the absence of a priming process.

Genetic demyelination models
Genetic models that trigger oligodendrocyte ablation or ectopic expression of interferon gamma (IFN-γ) have been used to investigate demyelination and remyelination. Initial studies of transgenic mice with targeted expression of IFN-γ from the myelin basic protein (MBP) promotor (Corbin et al., 1996) and glial acidic fibrillary protein (GFAP) promoter (LaFerla et al., 2000) demonstrated forced expression of IFN-γ during development results in CNS hypomyelination and abnormal cerebellar development. Combining GFAP promotor-driven expression of a tetracycline-controlled transactivator (tTA) (Gfap-tTA) with tetracycline response element (TRE) upstream of IFN-γ sequence (TRE-IFN-γ) in a double transgenic line allows for temporal regulation of CNS IFN-γ expression upon removal of doxycycline (Lin et al., 2004). CNS IFN-γ expression during cuprizone-mediated demyelination with removal of doxycycline during cuprizone exposure in Gfap-tTA; TRE-IFN-γ mice resulted in reduced differentiated oligodendrocytes and impaired remyelination (Lin et al., 2006). Prolongation of the integrated stress response (ISR) during cuprizone-mediated demyelination with CNS IFN-γ expression enhanced mature oligodendrocyte generation and remyelination (Chen et al., 2021) suggesting a beneficial role of oligodendrocyte ISR signaling in promoting remyelination.
Oligodendrocyte ablation through the combination of an inducible conditional mature oligodendrocyte Cre line (Plp-CreER™) with a diphtheria toxin subunit A (DTA) floxed stop reporter line (ROSA26-eGFP-DTA) results in oligodendrocyte apoptosis upon exposure to tamoxifen (Traka et al., 2010). Rapid oligodendrocyte loss occurs within the first week after tamoxifen injection and demyelination peaks at 5 weeks. Mice develop clinical symptoms of ataxia, tremor, hind-limb paralysis with some degree of lethality. At 10 weeks post-tamoxifen oligodendrocytes regenerate, remyelination and axonal numbers are comparable to controls, and axonal conduction assessed by spinal somatosensory evoked potentials are restored (Traka et al., 2010). Recovered animals develop a secondary fatal immune-mediated phase of demyelination that occurs around 40 weeks post-tamoxifen accompanied by focal inflammatory lesions, extensive myelin and axonal loss and the presence of MOG-specific T cells in lymphoid organs that are encephalitogenic when transferred into naïve recipients (Traka et al., 2016). This secondary immunemediated demyelination was inhibited by tolerization with MOG 35-55 after the initial remyelination phase. The DTA model may be useful for investigation of strategies to prevent development of a secondary adaptive immune response and investigate pathways to accelerate OPC maturation and subsequent remyelination in the setting of complete loss of mature oligodendrocytes.
Loss of function of transcription factor myelin gene regulatory factor (MYRF) that induces expression of mature myelin genes (Emery et al., 2009) (Koenning et al., 2012). Subsequent development of a secondary immune-mediated demyelination has not been reported in Myrf conditional knockout, possibly due to CNS restriction of Myrf expression to oligodendrocytes whereas Proteolipid protein 1 (Plp1) is expressed in Schwann cells which may contribute to the development of secondary autoimmunity. CNS penetrant thyroid hormone receptor mimetic, sobetirome, promoted OPC proliferation, remyelination and motor recovery in Myrf conditional knockouts (Hartley et al., 2019). While the severity of oligodendrocyte loss and demyelination in genetic oligodendrocyte ablation models may be a limitation, these models have the advantage of quantifiable motor recovery outcomes and secondary robust OPC response that can be modulated.

Viral models
Chronic encephalomyelitis viral mouse models share pathogenic features similar to MS (Oleszak et al., 2004;Lane and Hosking, 2010;Pike et al., 2022) and may provide insight into the induction of CNS autoimmunity and pathways that are activated in CNS resident populations in response to anti-viral inflammatory responses.

Theiler's murine encephalomyelitis virus
Theiler's murine encephalomyelitis virus (TMEV) is a singlestranded RNA picornavirus that causes flaccid myelitis in mice (Theiler, 1934) and can be used to induce an acute encephalomyelitis demyelinating disease through intracerebral infection. BeAn and Daniel's TMEV strains can cause a biphasic disease process characterized by an acute infectious phase and viral clearance followed by a secondary progressive chronic demyelinating myelitis phase in susceptible mouse strains (Oleszak et al., 2004). Oligodendrocytes, astrocytes and macrophages are viral reservoirs during chronic infection (Rodriguez et al., 1983;Lipton et al., 1995) and demyelination occurs within areas of microglia and macrophage activation generating demyelinating lesions with variable axonal injury and remyelination that varies depending on mouse strain (Bieber et al., 2005). Macrophages and microglia play a critical role in immunemediated demyelination in the TMEV model and mouse strain differences in susceptibility to demyelination may be due to differences in macrophages and microglia (Dal Canto et al., 1996). The TMEV model offers the advantage of a cytotoxic axonal injury mechanism (Rivera-Quiñones et al., 1998;Howe et al., 2007) which may facilitate the investigation of immune-mediated mechanisms of axonal injury and neurodegeneration.

Mouse hepatitis virus
Mouse hepatitis virus (MHV) is a positive-strand RNA virus with neurotropic strains that can be used to induce a chronic demyelinating disease through intracranial or intranasal inoculation of susceptible mouse strains. An acute encephalomyelitis phase is followed by a secondary phase of demyelination and remyelination (Bender and Weiss, 2010;Lane and Hosking, 2010). MHV-specific T cells appear to instigate demyelination (Wu et al., 2000;Dandekar et al., 2001Dandekar et al., , 2004Dandekar and Perlman, 2002). Diffuse macrophage and microglial activation and upregulation of oxidative stress pathways in the MHV model resembles changes seen in MS tissue (Schuh et al., 2014). Microglia play a critical role in the clearance of myelin debris and facilitating remyelination (Sariol et al., 2020). Oligodendrocytes that survive the acute MHV infection have prolonged MHC class I expression (Pan et al., 2020). Viral encephalomyelitis models offer the ability to investigate microglial, astrocyte and oligodendrocyte secondary responses to an inflammatory viral insult and subsequent repair mechanisms.

Remyelination therapies-from mouse to human
High-throughput screens that evaluate the ability of compounds to promote oligodendrocyte differentiation in vitro have been used to identify pathways with remyelination potential (Joubert et al., 2010;Deshmukh et al., 2013;Mei et al., 2014;Najm et al., 2015;Porcu et al., 2015;Lariosa-Willingham et al., 2016;Mei et al., 2016b). Many pathways and compounds that have demonstrated improved remyelination in animal models have moved to human MS clinical trials ( Table 1).
Assessment of remyelination in MS clinical trials has been challenging, with few validated clinical tools to assess remyelination in humans (Hill et al., 2022). Measurement of visual pathway conduction speed with visual evoked potential (VEP) has allowed for assessment of the change in P100 latency which correlates with remyelination in autoimmune encephalomyelitis (Cordano et al., 2022) and toxin (Marenna et al., 2022) mouse models ( Figure 1E). Multifocal VEP (mf-VEP) compared to full-field VEP (ff-VEP) may be a more accurate measure of remyelination after optic neuritis (Pihl-Jensen et al., 2017) and is a secondary outcome measure in NCT05131828-CCMR Two trial. Opicinumab treatment after first time acute optic neuritis significantly reduced ff-VEP latency in the per-protocol population at 32 weeks in the RENEW trial (Cadavid et al., 2017). Opicinumab combined with interferon beta-1a did not demonstrate a significant improvement in disability at 72 weeks in SYNERGY trial, however univariate analysis suggested that younger age, shorter disease duration and higher baseline brain volumes may be associated with improved disability (Cadavid et al., 2019). The ReBUILD clemastine placebo crossover trial in relapsing remitting MS (RRMS) with chronic optic neuropathy demonstrated significantly improved ff-VEP latency while on clemastine treatment (Green et al., 2017). New MRI techniques such as myelin water fraction (MWF), diffusion tensor imaging (DTI) and magnetization transfer ratio (MTR) are currently being developed to assess remyelination (Hill et al., 2022). MTR has shown promise as an imaging measure of remyelination with significant MTR lesion change in GSK239512 (Schwartzbach et al., 2017) and in gray matter and brainstem lesions in bexarotene (Brown et al., 2021). Continued development of clinical techniques to assess remyelination and thoughtful design of patient inclusion criteria and outcome measures are critical for designing clinical trials for remyelination therapies in MS. Rigorous pre-clinical testing of remyelinating pathways and compounds in animal models has

Conclusion
Animal models of MS with demonstrated remyelination capacity vary considerable in their mechanisms of demyelination, inflammatory infiltrates, degree of ongoing inflammatory activity, axonal loss and neurodegeneration and extent of remyelination. Focal toxin models offer the advantage of stereotyped remyelination after a short single demyelinating insult which has allowed for the investigation of factors that promote or inhibit this robust reparative response. A prolonged demyelinating insult predominated by corpus callosum and cortical demyelination and subsequent neurodegeneration and motor decline can be modelled with chronic cuprizone exposure and may share some features of the neurodegenerative process in progressive MS. Both cuprizone and EAE models induce inflammatory subsets of glia that are found in MS tissue and further investigation of how these subsets of glia contribute to ongoing inflammatory activity and promote or inhibit repair may offer insight into potential mechanisms to modulate remyelination in inflammatory settings. While no single animal model recapitulates the pathobiology of MS, considerations of the limitations and advantages of each model should be taken into account when investigating remyelination and translating animal model findings to human MS.

Author contributions
DP and EH contributed to writing the manuscript. EF contributed to researching and generating the clinical trial table. All authors contributed to the article and approved the submitted version.

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.