Neurodegeneration and Demyelination in Multiple Sclerosis

Summary

Progressive multiple sclerosis (PMS) is an immune-initiated neurodegenerative condition that lacks effective therapies. While peripheral immune infiltration is a hallmark of relapsing-remitting MS (RRMS), PMS is associated with chronic, tissue-restricted inflammation and disease associated reactive glial states. The effector functions of disease associated microglia, astrocytes, and oligodendrocyte lineage cells are beginning to be defined, and recent studies have made significant progress in uncovering their pathologic implications. In this review, we discuss the immune-glia interactions that underlie demyelination, failed remyelination, and neurodegeneration with a focus on PMS. We highlight the common and divergent immune mechanisms by which glial cells acquire disease associated phenotypes. Finally, we discuss recent advances that have revealed promising novel therapeutic targets for the treatment of PMS and other neurodegenerative diseases.

Graphical Abstract

Progressive multiple sclerosis (PMS) is a neurodegenerative condition with chronic inflammation and glial cell pathology. In this review, Garton et al. discuss immune-glia interactions in PMS, highlighting potential therapeutic targets for PMS and other neurodegenerative diseases.

Introduction

Multiple sclerosis (MS) is a disease characterized by autoimmune neuroinflammation resulting in demyelination and axonal injury and culminating in neurodegeneration and progressive disability. Discrete relapses, driven by peripheral immune cells migrating into the CNS, tend to occur in the early phases of MS, termed ‘relapsing-remitting’ MS (RRMS). RRMS often occurs over the background of a slow progression of symptoms in the absence of disease flares (‘secondarily progressive’; SPMS). In the primary progressive form (PPMS), patients develop a gradually worsening disease without any remissions. Despite recent discussion about whether this triad system of classification should be transitioned into a more diffuse spectrum of disease¹, it is clear that the two ends of the spectrum are not treatable with similar strategies. As we will discuss below, there is a paucity of treatments for the two progressive forms of MS (hereafter termed PMS) in part due to the compartmentalized nature of pathology in PMS. To uncover new therapeutic strategies for PMS, it is essential to investigate the resident CNS glial cells and their roles in perpetuating the chronic inflammatory environment that characterizes PMS. It should be noted that adaptive immunity has recently been suggested to also contribute to the inflammatory environment in PMS, with resident effector memory T cells and B cell infiltration being associated with disease severity^2–6; a comprehensive discussion of adaptive immunity falls outside the scope of this review, however. Here we will discuss current findings with regards to demyelination, glial activation and neurotoxicity, and the challenges the field has faced when it comes to treating PMS.

Immune-glia interactions contribute to demyelination and failed remyelination in MS

Myelin is a lipoprotein sheath produced by oligodendrocytes (OLs) which wraps axons, speeding conduction and providing metabolic and trophic support to neurons. The immune-mediated destruction of myelin, a hallmark of MS, therefore has direct consequences on neuronal function and integrity. Depending on the anatomic structures involved, demyelinating lesions can be asymptomatic or lead to diverse and disabling focal symptoms of MS. Over time, chronically demyelinated neurons are vulnerable to axonal injury and metabolic stress, contributing to neurodegeneration in PMS^7,8. Oligodendrocyte precursor cells (OPCs) are widely distributed glia with the capacity to differentiate into myelinating OLs in addition to having intrinsic homeostatic functions such as interactions with neurons and the vasculature^9–13. However, while some OPC-mediated remyelination occurs, especially in the early stages of MS, it is limited in later PMS^14–17. Furthermore, there is substantial heterogeneity in remyelination between individuals and even between lesions in a single person with MS (PwMS), suggesting that dynamic features of the local environment enhance or inhibit remyelination. Myelination-promoting therapies could profoundly improve patient quality of life and impact disease progression, and to this end significant effort has been devoted to defining mechanisms behind demyelination and failed remyelination in MS. Immune-glia interactions, initiated by peripheral immune infiltration and perpetuated by reactive glial phenotypes, drive demyelination and failed remyelination¹⁸. In this section we will discuss how peripheral immune molecules act on OL lineage cells and contribute to their death and dysfunction. We also review recent literature describing inflammatory OLs (iOLs) and OPCs (iOPCs) in MS and models of inflammatory demyelination.

Demyelination with failed or limited remyelination is a common feature of MS lesions, which otherwise have variable amounts of inflammation, axonal injury, OL loss, and reactive gliosis based on lesion age and activity^19–21. Based on these characteristics, lesions can be subdivided into three categories. The first are acute active lesions, which have breakdown of the BBB, perivascular lymphocyte infiltration (mostly cytotoxic T cells), abundant microglia and macrophages throughout the lesion, and variable axonal injury. Chronic active lesions, in contrast, have a hypocellular and gliotic core with a rim of activated myeloid cells. Finally, inactive lesions are sparsely cellular throughout and are characterized by accumulation of chondroitin sulfate proteoglycans (CSPGs). A comprehensive discussion of the different biological markers that define each lesion type as well as useful clinical markers used to categorize them in patients has been recently reviewed elsewhere (see ^22–24). Recently, sub-domains of these lesions have been identified using spatial transcriptomics, further adding complexity to these three main types²⁵. Active lesions are more commonly seen in early RRMS or SPMS, while inactive and chronic active lesions dominate in the non-relapsing phase of SPMS and PPMS^19,26,27. Myelin debris engulfed within microglia and macrophages, evident in some active and chronic active lesions, suggests ongoing myelin destruction^19,20. Interestingly, despite consistent demyelination OL numbers tend to be preserved in early active lesions, though are increasingly lost in chronic active/inactive lesions²⁸. Whether this ‘dying back’ is the initial step in OL death, or if there are distinct pathologic mechanisms underlying demyelination and OL death remains unclear.

Humoral factors mediating demyelination: immunoglobulins and complement

Humoral factors, including antibodies and complement proteins, are leading candidates in the destruction and clearance of myelin. Antibodies (mainly IgG and to a lesser extent IgM) are deposited in a majority of active, chronic active, and inactive MS lesions^21,29–31. While the source of CNS-deposited antibodies is difficult to ascertain, the presence of cerebrospinal fluid (CSF)-restricted oligoclonal bands in most PwMS is evidence of antibody production in or around the CNS likely related to clonally expanded B cells and plasmablasts^32,33. Indeed, activated B cells with somatic hypermutations of V(H) genes can be found in the CSF and form follicle-like structures in the subarachnoid space^33–35. Antibodies derived from B cells within the subarachnoid space are sufficient to induce demyelination in spinal cord explants³⁶, but despite this, no single dominant autoantigen has been found in MS. A recent study identified a clone of IgG1 antibodies that targeted the proteolipid protein 1 (PLP1), a component of myelin, that could induce MS-like pathology in mice; however, these PLP1-specific antibodies were present in only ~57% of MS patients examined³⁷. A few additional candidate antigens have emerged, with several reports detailing antibody responses at the axo-glial interface to neurofascin and contactin-2^38–40. One study found that CSF-derived antibodies from MS patients bind to neurofilament light, an axonal protein⁴¹, and myelin-reactive “2d2” T cells can recognize neurofilament in experimental autoimmune encephalomyelitis (EAE), an animal model of autoimmune neuroinflammation⁴². Other reports, however, found similar amounts of neurofilament light chain autoantibodies in control samples or reactivity of CSF-derived antibodies to ubiquitous, non-CNS antigens^43,44. More recent studies have confirmed common antigenic motifs seen in Epstein-Barr Virus (EBV) infection and MS, including peptide sequences from the EBV proteins BRRF2 and EBNA^45–48. The difficulty in defining the target antigen may be related to both diversity of targets between individuals and epitope spreading, which occurs during the relapsing phase of the disease and broadens the pool of antigen targets.

Despite equipoise around specific targets, the robust deposition of antibodies in MS lesions is suspected to contribute to demyelination. One likely mechanism is the classical complement pathway, an enzymatic chain reaction triggered by antibody deposition that results in formation of opsonins (such as C3b), pro-inflammatory molecules (including the anaphylatoxins C3a and C5a), and the cytolytic C5b-9 complex termed the membrane attack complex (MAC) (Figure 1)⁴⁹. Large amounts of complement proteins that contribute to the cascade are produced in the liver and released into circulation, though complement is also produced locally by reactive astrocytes and microglia^50–53. Several histologic studies have found deposition of antibody alongside activated complement components, including C1q, C3b, and C5b-9, in MS lesions^{21,29,31,42,54}. In addition to activating the classical complement cascade, antibodies are opsonins themselves and affect local immune cell functioning directly through the Fc receptor (FcR)⁵⁵. Immunoglobulin and complement proteins often colocalize with the myelin component PLP in vesicular structures within macrophages, suggesting PLP is opsonized by these molecules prior to engulfment²⁹. One study provided evidence for the relative importance of complement over FcR signaling in experimental demyelination, finding that MOG IgG–induced demyelination was blocked in C1q^−/− but not FcRɣ^−/− animals⁵⁶. The role of complement was complicated by studies finding that complement component C5 deficient mice, which are unable to generate the C5b-9 complex, still had impaired remyelination and enhanced OL death, supporting the idea that earlier complement components such as C3 may be driving pathology⁵⁷. Despite these findings, it should be noted that in other neurodegenerative disorders, complement inhibition has had mixed results; while most studies report improved outcomes following complement inhibition⁵⁸, several older studies identified worsening of pathology in Alzheimer’s disease models^59,60. The possibility that complement serves a protective function by opsonizing myelin debris and dead OLs should be noted. Without the removal of debris, a persistent inflammatory environment could lead to impaired regeneration.

Figure 1: Complement in inflammatory demyelination.

A) The classical complement cascade. B) Macroscopic deposition of antibody and complement in demyelinating white matter lesions. Antibodies in the perivascular CSF-filled space escape into the brain parenchyma, allowing for deposition of microglial (MG)-secreted C1q and activation of astrocyte (Ast)-secreted C3 into C3a and C3b at the chronic active edge. C) Complement locally influences glia towards reactive phenotypes. Deposition of C3b onto myelin can be recognized by microglial CD11b, resulting in phagocytosis. Stimulation of C3aR with C3a results in increases of HIF-1a as well as alterations in phagocytosis depending on the duration of the stimulation. Made in BioRender.

Cytokine signaling impacts OL survival and function

Cytokines are effector molecules, produced and sensed by infiltrating immune cells and CNS resident cells, that orchestrate immune responses. OLs and OPCs express receptors for many cytokines, and several have direct impacts on survival, proliferation, and differentiation.

TNF is a multi-faceted molecule produced mainly by peripheral immune cells and glia in MS. TNF inhibition is therapeutic in numerous autoimmune conditions, and based on early evidence that it is elevated in MS CSF and CNS tissues, TNF and TNFR antagonists were evaluated as effective therapies in MS. Surprisingly, however, patients treated with the TNF inhibitor lenercept had increased MS exacerbations⁶¹. To explain this outcome, subsequent studies explored the nuances of TNF signaling and found opposing effects of signaling through the two TNF receptors, TNFR1 and TNFR2. TNF has two distinct forms, transmembrane (tmTNF) and soluble (solTNF), and TNFR1 binds solTNF or tmTNF, while TNFR2 is activated only by tmTNF⁶². TNFR1 is closely linked to cell death, harboring an intracellular death domain, and activation of TNFR1 on OLs promotes FAS-mediated apoptosis^63,64. TNFR2, in contrast, promotes OL survival and myelination in EAE⁶⁵. Additionally, TNF signaling alters the inflammatory state of OLs: TNFR1 knockdown in OLs results in a decrease in inflammatory signaling in EAE while knocking down TNFR2 enhances BBB breakdown, microglial activation, and subsequent inflammation⁶⁵. TNFR2 is also expressed on OPCs, where its activation promotes differentiation, survival, and remyelination^65–68. TNFR2^−/− animals have impaired progenitor expansion and remyelination after cuprizone⁶⁶, and TNFR2^−/− OPC cultures have impaired survival and differentiation⁶⁵. Like OLs, OPCs become active participants in neuroinflammation, secreting cytokines and chemokines that attract and activate immune cells. Here too, TNF has opposing effects depending on the receptor activated, with TNFR2 promoting protective and reparative functions of OPCs in vitro⁶⁹.

Interferon gamma (IFNγ), classically regarded as a master regulator of anti-viral responses, has dose and disease stage dependent effects on myelination. IFNγ induces apoptosis of OLs in vitro⁷⁰, and overexpression of IFNγ in the mouse brain results in developmental hypomyelination, primary demyelination in early adulthood, and CD8+ T cell-associated OL loss in the aged brain^71–73. However, low concentrations of IFNγ actually protect OLs from oxidative stress, and transgenic expression of low levels of IFNγ protects mice against cuprizone-induced demyelination^74,75. Lin et al. established the importance of timing in this system, employing a mouse model where IFNγ is inducible downstream of the GFAP promoter, resulting in astrocyte-specific expression of IFNγ. When IFNγ was expressed during the acute phase of EAE, it reduced disease severity, demyelination, and oligodendrocyte loss. However, when IFNγ was expressed during the recovery phase of cuprizone or EAE, remyelination was markedly impaired⁷⁶. Interestingly, endoplasmic reticulum (ER) stress appears to underlie the impact of IFNγ on OL lineage cells, affecting both demyelination and remyelination^76,77. Genetically or pharmacologically increasing the integrated stress response to appropriate levels in OLs increases remyelination in both EAE and cuprizone models⁷⁸. IFNγ also directly acts on OPCs, reducing proliferation, differentiation, and myelin production in vitro, though its impact on OPCs in vivo remains an area of investigation^79,80.

While not directly implicated in OL death like TNF or IFNγ, the impact of numerous other cytokines have been studied on OPCs: IL-17 blocks OPC differentiation and enhances inflammation through induction of NOTCH1⁸¹; IL-1β and IL-33 both block OPC proliferation and enhance differentiation through activation of the p38-MAPK signaling pathway^82–84; IL-4, a driver of allergic-type inflammation, however promotes OPC differentiation through a PPARγ dependent mechanism⁸⁵; and transforming growth factor beta (TGFβ) regulates OPC differentiation during development and promotes remyelination after experimental demyelination^86,87.

The emerging role for inflammatory OLs and OPCs in MS

Beyond their roles in homeostasis and myelination, OLs and OPCs are increasingly appreciated as active participants in neuroinflammation^88,89. Major histocompatibility complex type 1 (MHCI) and 2 (MHCII) are cell surface proteins that present self-antigen to cytotoxic T cells or phagocytosed antigen to helper T cells, respectively. Accumulating evidence suggests that IFNγ produces an inflammatory phenotype in OL lineage cells, particularly when combined with IL-17 or TNF. Inflammatory OLs (iOLs) or OPCs (iOPCs) perpetuate tissue inflammation and lose physiologic functions, including remyelination potential⁸⁸. Early evidence from cultured cells found that OLs upregulate MHCI in response to IFNγ, and subsequent studies confirmed increased MHCI expression in OLs from MS tissue^90–93. Interestingly, iOPCs acquire traits usually present only in immune cells, including MHCII expression and the ability to ‘cross-present’ phagocytosed antigens on MHCI⁹⁴. Recent single cell and single nucleus RNA-seq studies have confirmed OL and OPC upregulation of antigen presentation machinery in EAE and MS^95–98. Potentially through this mechanism, iOLs and iOPCs amplify neuroinflammation by activating T cells and secreting cytokines and chemokines⁸⁹. Indeed, iOLs and iOPCs play a substantial role in neuroinflammation: one study found that deleting ACT1 (a component of the IL-17R signaling cascade) specifically in PDGFRα+ OPCs markedly reduced EAE severity, highlighting the role of OPCs in perpetuating inflammatory IL-17 signaling⁹⁹. Cytotoxic lymphocytes can also target iOLs and iOPCs for destruction, contributing to demyelination and limiting the pool of cells available to remyelinate⁹⁴. Further studies are needed, however, to fully clarify their impact on de- and remyelination in MS.

Recent Developments in the Understanding of the Mechanisms of Glial Activation and Glia-Mediated Neurotoxicity in MS.

Microglia and Astrocytes in the setting of neuroinflammation and neurodegeneration

In the context of inflammation-mediated neurodegeneration, glia have diverse functions and can either promote protection and recovery or become deleterious and neurotoxic¹⁰⁰. In the healthy brain, microglia and astrocytes perform homeostatic functions that directly or indirectly support to neurons, synapses, oligodendrocytes, and each other. Microglia survey their environment, clearing debris and remodeling neuronal circuitry via pruning of unnecessary synaptic connections^101–103. Astrocytes, meanwhile, provide metabolic support to neurons and stabilize synaptic connections^104,105. In the setting of inflammation, however, these glial cells can lose their homeostatic functions and enter reactive states characterized broadly by production of inflammatory cytokines, interactions with immune cells, and even direct neurotoxicity, as will be discussed below. In the following sections, we will review what is known about how microglia and astrocytes are activated towards disease-associated states, including the similarities and differences between reactive glia in MS vs other neurodegenerative diseases, and the mechanisms by which these glia mediate neurotoxicity.

Microglia in MS vs other neurodegenerative diseases

While the historic classification of macrophages or microglia into “M1” and “M2” phenotypes is oversimplistic, myeloid cells clearly adopt distinct states in the settings of disease. Activation of microglia can be beneficial, as reparative microglia remove debris and encourage regeneration. For example, a recent study showed how in mice with defects of the PLP1 protein, microglial removal of myelin protected against CD8+ T cell-mediated axonopathy¹⁰⁶. However, research in several neurodegenerative diseases has identified subsets of “reactive microglia” as critically involved in their pathogenesis^107–109. While there is substantial overlap in microglia phenotype across diseases, resulting in the broad term “Disease-Associated Microglia” (DAM), many fields have generated terms for disease-specific reactive microglia (e.g. “human AD microglia” or HAMs)^110,111. MS has recently joined the list of disorders with a specific nomenclature for their DAMs. Sampling of microglia from the chronic active edge of paramagnetic rim lesions (PRLs) in MS has revealed the pathogenic subtype of microglia termed “microglia inflamed in MS” or MIMS which are characterized by elevations in C1Q, FTH1, CD68, and additional iron-regulatory genes¹¹². Single nucleus transcriptomics have identified that microglia have distinct genetic signatures in MS compared to both secondary demyelination and healthy controls¹¹³. In particular, while both secondary demyelination and MS microglia have decreases in classic homeostatic markers like P2RY12 and CX3CR1, MS microglia exhibit increases in FTH1, SIPA1L1, and ACSL1 whereas secondary demyelination show higher levels of MYO1E and APOE¹¹³. Therefore, it is important to consider that while most neurodegenerative or demyelinating disorders have similar microglial phenotypes, there are key distinctions that could have important implications for the development of microglia-targeting therapies.

Mechanisms of microglial activation in MS

Blood-brain barrier (BBB) disruption is a prominent feature of acute MS lesions. Gadolinium enhancing lesions have been shown to correlate with perivenular cuffs of immune cells which release matrix metalloproteases that allow them to traverse the endothelium. Demyelinating lesions radiating outward from the post-capillary venules are rapidly becoming a specific marker of MS and can be detected using magnetic resonance imaging-based acquisitions such as T2*^114,115. T and B cells that encounter cognate antigen are reactivated in the CNS where they release cytokines that mediate microglial activation. Incubation of microglia with IFNγ, TNF, and IL-1β results in activation of mTOR signaling and subsequent production of cytotoxic nitric oxide¹¹⁶. A recent paper identified additional cytokines that may selectively trigger pro-inflammatory polarization in the setting of MS: knockdown of IFIT3, largely produced by migrating macrophages, reduced microglial ability to polarize towards this reactive phenotype in the spinal cords of EAE mice possibly via attenuation of STAT1 and NF-kB signaling¹¹⁷.

Exposure of microglia to bloodborne humoral elements also results in activation of inflammatory transcriptional profiles. In EAE, focused ultrasound-induced BBB disruption resulted in focal alterations in the activation state of microglia despite no change in amount of cellular infiltrate¹¹⁸. In particular, fibrinogen (a key member of the clotting cascade) directly signals the CD11b receptor on myeloid lineage cells and induces oxidative stress pathways, redox regulation, and type 1 IFN gene families as well as clustering of microglia in the perivascular space^119–121. Fibrinogen-stimulated microglia recreate the oxidative stress signature of microglia in EAE, suggesting that exposure of microglia to fibrinogen from blood may be a key signal in their polarization towards a reactive state.

As discussed previously, the complement cascade has been studied for its contribution to demyelination in the setting of inflammation (Figure 1) and has recently been shown to be associated with MS disease severity¹²². The complement components C3a and C5a promote inflammation and leukocyte invasion^123,124. Signaling of C3a onto the C3a receptor (C3aR) on microglia results in upregulation of hypoxia-inducible factor 1a (HIF-1a) leading to depletion of ATP stores, metabolic impairment, and aberrant microglial response in a model of AD¹²⁵. C3aR signaling also has biphasic effects on phagocytosis: short term administration promotes phagocytosis whereas longer term (24 hour) stimulation stifles phagocytosis¹²⁶. The other members of the early complement cascade are gaining more attention as well. The ability of non-anaphylatoxin members of the complement cascade to directly trigger pro-inflammatory transcriptional changes in microglia has recently come to the forefront¹²¹. Stimulation of primary microglia with the opsonizing form of C3 (iC3b) results in oxidative stress responses, phospholipid metabolic processes, and reorganization of the extracellular matrix¹²¹ with specific upregulation of additional complement genes such as C1QA. This may be due to direct signaling through the complement receptor 3 (CR3; aka CD11b), which is expressed by microglia. A recent study showed that siRNA-mediated knockdown of CR3 attenuates activation of microglia in the setting of chronic inflammation caused by rotenone¹²⁷. This resulted in subsequent reduction in the production of C1q, TNF, and IL-1α, a cocktail of cytokines termed “TIC” that is known to induce neurotoxic reactive astrocytes in multiple neurodegenerative diseases⁵¹. C3 therefore has multiple mechanisms by which it can signal through microglia to promote reactive and phagocytic phenotypes.

Astrocytes in MS vs other neurodegenerative diseases

Recent studies have identified a broad heterogeneity in astrocyte subtypes in different disorders. While there are some overlapping differentially expressed genes (DEGs) in spinal cord injury, LPS-induced neuroinflammation, and EAE, a recent study only found 2.6% of all astrocyte reactivity DEGs were present in all three¹²⁸. A RiboTag approach to examining regional differences in astrocyte transcript profiles in EAE showed distinct pathways especially in spinal cord vs brain derived astrocytes¹²⁹. Clearly many subtypes of astrocytes exist but are still referred to under the broad multi-disorder banner of “reactive astrocytes”^130,131.

In MS, single cell profiling demonstrates that astrocytes decrease their expression of the nuclear factor erythroid 2-related factor 2 (NRF2), a master transcriptional regulator of the endogenous anti-oxidant response, while simultaneously upregulating MAFG, a small Maf (sMAF) transcription factor involved in promoting pro-inflammatory transcriptional profiles¹³². In contrast, in AD, MAFG has been suggested to not be upregulated whereas a different sMAF protein (MAFF) has been suggested to be the most responsive sMAF protein to AD pathology^133–135, though the cellular source of MAFF in AD has not been determined yet. While these sMAF proteins can both form heterodimers with NRF2 to initiate antioxidant and anti-inflammatory transcription, they can form homodimers which lack a transcriptional activation domain and subsequently repress transcriptional programs^136,137. Upregulation of sMAFs such as MAFG in MS and MAFF in AD without concurrent upregulation of NRF2 therefore represents a mechanism by which astrocytes can repress their antioxidant and anti-inflammatory profiles.

Phosphorylation of the ER stress sensor IRE1a following stimulation of astrocytes with pro-inflammatory cytokines such as IL-1β and TNF results in full translation of the transcription factor XBP1, which has been shown to drive a subset of astrocytes towards pathogenicity in MS^138–140. This pathogenicity was in part due to the increased ability of XBP1+ astrocytes to recruit peripheral monocytes to the CNS and promote microglial activation. IRE1a-XBP1 signaling in astrocytes is stimulated by the accumulation of unfolded proteins and activation of the unfolded protein response.

Recent work on MS patient induced pluripotent stem cell-derived astrocytes demonstrated increased processing and presentation of antigen as well as genes related to EBV response compared to healthy controls, even in the absence of inflammatory stimuli¹⁴¹. These data suggest an intrinsic factor in MS astrocytes that predisposes them towards reactive states which may contribute to development of pathology, a finding potentially corroborated by the discovery of epigenetic signatures in EAE that prime astrocytes towards reactivity¹⁴². The upregulation of EBV-related genes in MS astrocytes strongly supports the recent evidence that highlights a role for EBV infection in predisposing individuals to developing MS, a topic which has been extensively reviewed elsewhere (see ^46,143,144).

Microglia-astrocyte crosstalk contributes to astrocyte activation in MS

Astrocytes are central figures in the signaling milieu during neuroinflammation, modulating peripheral immune infiltration and chronic tissue inflammation (reviewed in ¹⁴⁵). Independent of peripheral immune signaling, microglia and astrocytes can also engage in crosstalk that results in development of reactive astrocytes. It has been well described that following NLRP3 inflammasome action on Caspase-1, microglia can stimulate reactive astrocytes via the secretion of the TIC cocktail of cytokines^51,146. Upon stimulation with TIC, astrocytes take on a phenotype characterized by increased complement production and neurotoxicity. Additional signaling between microglia and astrocytes has become evident. The MIMS discussed above have been shown to interact closely with reactive astrocytes in chronic active MS lesions largely via microglial C1Q and astrocytic C3¹¹². Intriguingly, in a cell culture model of AD, Guttikonda et al. showed that microglia require the ability to produce C3 in order to initiate reciprocal signaling with astrocytes which subsequently produced excess C3¹⁴⁷. It is possible that microglial production of C3 is therefore also vital to activate neurotoxic astrocytes in MS. The ability to investigate astrocyte signaling interactions has improved with a recently developed technique termed RABID-seq in which a viral-based barcoding system allows transmission of the virus between interacting cells so that cell-cell interactions can be tracked followed by single-cell (sc)RNA-sequencing. This technique demonstrated that microglia and astrocytes interact via the axon guidance molecules Sema4D-PlexinB2 and EphrinB3-EphB3 in EAE, and that inhibition of EphB3 suppresses reactive astrocyte formation and ameliorates EAE¹⁴⁸.

Another example of microglial-astrocyte crosstalk mediating glial activation involves the master transcriptional regulator of the inflammatory response, NF-κB, Its activation has distinct consequences in different cell types, with NF-κB activation appearing to protect OLs during EAE^149,150 but also potentially enhancing pathologically reactive astrocytes. Reactive astrocytes show increased NF-κB activation in both MS and EAE¹⁵¹. A study by Ponath et al. found a known MS-risk variant, rs7665090^G, located nearby the NFKB1 locus which increased NF-κB signaling and gene expression (e.g., CXCL10, C3) in human astrocyte cell cultures as well as in astrocytes within MS lesions, suggesting a potential astrocyte-mediated causal link to pathogenesis¹⁵². Of note, a recent study utilized a microfluidic platform to analyze cell-cell interactions and identified that astrocytic NF-κB activation (experimentally induced by stimulation with LPS) can be suppressed by signaling from microglial-derived amphiregulin (an epidermal growth factor), suggesting a potential glial-crosstalk system that regulates astrocyte NF-κB¹⁵³. Intriguingly, they found that microglial expression of amphiregulin could be induced by astrocyte-derived IL-33, completing an anti-inflammatory signaling loop between astrocytes and microglia. It may be possible that this glial interaction is altered in MS, although this has not yet been investigated. With recent observations that MS risk genes are also expressed in microglia and that 18 of the known MS risk gene variants are related to the NF-κB and AP-1 signaling pathway it is plausible that excessive glial responsiveness or failed NF-κB regulation may account for chronic CNS inflammation^154,155.

Mechanisms of microglial and astrocyte neurotoxicity

Following pathological activation, microglia and astrocytes contribute to neurotoxicity in MS and other neurodegenerative diseases. The proposed mechanisms by which they mediate this injury are many and varied (Figure 2). One of these is glial generation of and interactions with the complement system. In the past, investigation of the role of the complement system in MS was largely focused on direct cytotoxic injury through MAC^57,156. However, more recent attention has been paid towards early factors, including C1Q and C3. Astrocytes are the primary producers of C3 in the CNS, and numerous studies have highlighted how astrocytic C3 is associated with worse pathology in MS and EAE^50,157–159. The mechanism by which C3 mediates neurodegeneration is still unclear, however. C1Q and C3 have both been identified as key contributors to the excessive elimination of synapses in AD and MS via opsonization and subsequent phagocytosis^160–164. Glia-mediated synapse loss has also been linked to the Signal regulatory protein α (SIRPα) in microglia, via its interactions with the “don’t eat me” signal CD47 (Figure 2)¹⁶⁵. The direct link between synapse loss and neurodegeneration is unclear, however. While it may be possible that excessive elimination of synapses on otherwise healthy neurons is a driving step towards cell death, it conversely is possible that already injured neurons may signal for phagocytes to remove synapses that may contribute to excessive excitatory signaling that could cause oxidative stress, and recent discussion has questioned whether aberrant synaptic “stripping” is involved in disease pathogenesis¹⁶⁶.

Figure 2: Proposed Mechanisms of Glial Neurotoxicity in PMS.

A) Normal healthy CNS environment, with CD47 (a “Don’t Eat Me” signal) signaling through microglial Sirpα to halt phagocytosis of synapses. Homeostatic astrocytes support neuronal metabolism and ensure that neurons have sufficient resources to thrive. B) The setting of neuroinflammation in the CNS. Reactive astrocytes upregulate glycolysis and translation, leading to diversion of resources away from neurons and towards glia, resulting in a state of virtual hypoxia. Astrocytes may directly mediate neurodegeneration through secretion of Long Chain Fatty Acids (LCFAs) and TNF among other cytokines, while microglial Galectin-3 enhances the neurotoxicity of oxidized phosphatidylcholine (OxPC). Meanwhile, stressed neurons downregulate CD47 concurrent with increases in local complement deposition, resulting in phagocytosis of synapses which may lead to neurodegeneration. Made in BioRender.

The search for a glia-derived directly neurotoxic factor has resulted in a few candidates. A variety of other microglia- and astrocyte-derived molecules have been previously reviewed for their potential to cause neurotoxicity^167,168. One of the more well-studied mediators of neurodegeneration is astrocyte-secreted TNF which can interact with TNF receptors on neurons to initiate Fas-mediated apoptosis^169–172. A few additional candidates have recently emerged; for example, Guttenplan et al reported that long-chain saturated lipids produced by astrocytes mediated neurotoxicity both in vitro and in vivo¹⁷³. Additionally, microglial Galectin-3, a lectin associated with pro-inflammatory conditions, exacerbates the neurotoxicity generated by oxidized phosphatidylcholines, lipids elevated in EAE and MS^174,175. In numerous neurodegenerative models, the enzyme SARM1 has been shown to play crucial roles in axonal injury with subsequent neuronal loss^176–178. Astrocytes upregulate the SARM1 in EAE, which results in reductions of glial-derived neurotrophic factor (GDNF) and subsequent demyelination and neuroinflammation¹⁷⁹. A recent study suggested that SARM1 may also mediate its pathological effects on axons by increasing NF-κB signaling following reduction of the NF-κB inhibitor IGFBP2¹⁸⁰. Deletion of SARM1 only blocks axonal degeneration in the early phases of EAE, however, and at late timepoints, SARM1 knockout mice were indistinguishable from wild types, indicating that SARM1’s contribution to axonal degeneration may vary over the course of CNS inflammation¹⁸¹. Meanwhile, in the neurodegenerative diseases amyotrophic lateral sclerosis and frontotemporal dementia, astrocytes secrete elevated levels of inorganic polyphosphate, and neutralizing this inorganic polyphosphate alleviates neurotoxicity¹⁸². Neither of these molecules have been investigated in MS yet, however, and warrant further research.

Another postulated mechanism of neurodegeneration in the setting of MS is metabolic stress and virtual hypoxia. Virtual hypoxia is a state in which neurons are deficient in metabolic resources like glucose, oxygen, and energy, potentially due to mitochondrial damage, oxidative injury, or the excessive demand of resources by reactive glia¹⁸³. Several studies have demonstrated shifts in amino acid metabolism¹⁸⁴ and increases in oxidative stress in PwMS¹⁸⁵, implicating metabolic disruption in disease pathogenesis. Microglia exposed to inflammatory stimuli such as LPS or NO shift their metabolic programming towards glycolysis, and re-focusing them towards oxidative phosphorylation via administration of dimethyl malonate generates anti-inflammatory transcriptional programs¹⁸⁶. In astrocytes, a shift towards increased glycolysis results in elevated inflammation^187,188. This suggests that when glia acquire highly energy-demanding phenotypes, there is a concomitant increase in inflammation and neuronal stress, potentially due to a paucity of available resources for neurons. The protective effects on myelin of promoting the integrated stress response in oligodendrocytes has been well described^79,189–191 although the subsequent effects on neurodegeneration have not been as well explored. Diversion of resources away from metabolically-demanding reactive glia and towards vulnerable neurons may be transiently neuroprotective as well as result in an anti-inflammatory environment that might promote downstream regeneration.

Therapeutic approaches targeting non-relapsing progressive disease and remyelination: Lessons learned and future directions

Limited Efficacy of Remyelinating and Neuroprotective Drugs in Clinical Trials in People Living with MS

Disease modifying therapeutic strategies for the treatment of PwMS are currently limited to immunomodulatory drugs that inhibit acute inflammation in the CNS. Current high-efficacy disease-modifying therapies (DMTs) for MS limit the activation or CNS infiltration of the adaptive immune system, thereby greatly reducing relapse rates¹⁹². However, these therapies have little to no impact on myelin repair or halting neurodegeneration in progressive MS. Over the last two decades, several potential pro-remyelination and neuroprotective therapies have been developed based on the concept that OPCs are abundant and proliferative cells in the CNS and can be driven to differentiate into myelinating OLs upon withdrawal of PDGFRα, induction with triiodothyronine, and a number of other pharmacological compounds¹⁹³. Indeed, exciting preclinical data support roles for anti-muscarinic agents (e.g. clemastine, benztropine), anti-histamine (GSK239512), RXR agonism (bexarotene, clobetasol), LINGO antagonists (opicinumab), and cholesterol biogenesis (emopamil binding protein (EBP) inhibitors) in vitro and in animal models leading to early phase trials of these putative remyelinating agents in PwMS¹⁹⁴. Despite these preclinical successes, however, several clinical trials of potential remyelinating and neuroprotective therapies in PwMS have yielded negative or underwhelming results, which we will review below.

LINGO-1 is a transmembrane cell surface glycoprotein that regulates OPC differentiation and proliferation as well as axonal growth¹⁹⁵. Genetic deletion of LINGO-1, knockdown of LINGO-1, or treatment with anti-LINGO-1 antibodies results in increased axonal myelination and neuronal survival in cell culture and animal models, including cuprizone and MOG-induced EAE^196–200. The anti-LINGO-1 antibody opicinumab initially showed improvement in visual evoked potential (VEP) testing in people with optic neuritis in the per-protocol analysis of the Phase II RENEW trial²⁰¹. Despite these encouraging results, opicinumab failed to reach its primary end points in both the SYNERGY and AFFINITY phase II trials in PwMS²⁰². After AFFINITY, clinical development of opicinumab was discontinued.

Another potential small molecule remyelination therapy, clemastine, was identified in 2014 using a micropillar array high-throughput screening platform²⁰³. Similarly to opicinumab, clemastine showed promise in several animal models of demyelination^204,205. The Phase II ReBUILD trial suggested that clemastine treatment in PwMS with chronic demyelinating optic neuropathy might promote endogenous remyelination of the optic nerve as measured by reduced P100 latency delay of 1.7ms/eye on VEP testing²⁰⁶. The small magnitude of this effect is of unclear clinical significance. A phase IIa clinical trial of RXR-agonist bexarotene in patients with RRMS failed to reach its primary efficacy outcome of improved lesional magnetization transfer ratio (MTR, a correlate of myelin density) and furthermore demonstrated poor tolerability with significantly increased adverse events in the bexarotene-treated group including central hypothyroidism (100%), hypertriglyceridemia (92%), rash (50%), and neutropenia (38%)²⁰⁷. In a post-hoc exploratory analysis of this trial, significant treatment effects were seen at the whole lesion level in gray matter but not white matter lesions, and voxel-level analyses detected significant treatment effects in white matter lesion voxels with the lowest baseline MTR, suggesting these measures may be more sensitive to myelin repair²⁰⁸. In contrast to bexarotene, a phase II study of GSK239512, a well-tolerated brain penetrant histamine H₃ receptor blocker, in PwMS showed a small but statistically significant effect on remyelination as measured by changes in lesional MTR²⁰⁹.

In addition to pro-remyelination therapies, several direct neuroprotective strategies have been tested in early phase clinical trials in both RRMS and PMS with the majority failing to reach their primary endpoints. The Phase IIb MS-SMART trial assessed the efficacy of three axonal-targeted neuroprotective drugs in a cohort of 393 patients with SPMS: amiloride that limits Na+ and Ca2+-dependent neuroaxonal injury, riluzole that reduces glutamate excitotoxicity, and fluoxetine which has pleiotropic effects on energy metabolism²¹⁰. The primary outcome of brain volume change on MRI was not met in any arms despite an adequately powered trial, suggesting that targeting these pathological mechanisms of axonal injury was insufficient to ameliorate neuroaxonal loss in SPMS over the course of the trial. Multiple other putative neuroprotective therapies have shown negative or minimally positive results in clinical trials of PwMS (reviewed in ²¹¹). In sum, recent years have yielded ineffective or minimally efficacious clinical trials with respect to remyelination or neuroprotection in PwMS. Identifying reasons why these compounds and trials have had limited success is vital to improving future progress in MS care.

Barriers to Remyelination and Neuroprotection

Many of the remyelinating strategies tested in MS clinical trials target OPC differentiation and/or proliferation directly. However, the presence of non-differentiating OPCs within chronic MS lesions, albeit at lower densities than normal white matter, suggests that OPC proliferation and differentiation may be extrinsically inhibited in these lesions^{28,96,212–214}. Over the last decade, researchers have identified several extrinsic factors that can interfere with OPC biology and may prevent therapies from effectively targeting OPCs and promoting remyelination. Similarly, many of the neuroprotective strategies for PwMS directly target neurons as opposed to reactive glia and extrinsic factors that contribute to a pro-neurodegenerative environment. As outlined above, compartmentalized chronic inflammation within the CNS (including reactive glia, immune cells, and pro-inflammatory cytokines) can interfere with effective endogenous and exogenous remyelination and neuroprotection signals. Additional extrinsic factors including changes in the extracellular matrix (e.g. chondroitin sulfate proteoglycans/CSPGs), leaky BBB, blood products including fibrinogen, axon receptivity, myelin debris, and metabolic stress may also contribute to this anti-remyelination and pro-neurodegenerative environment²¹¹.

A study using human induced pluripotent stem cell-derived OLs (hiOLs) showed that blockage of OL differentiation with activated PBMC-supernatants, an effect mediated at least in part through IFNγ, could not be restored by treatment with the OPC differentiation-promoting drugs clemastine, benztropine, or miconazole. However, OL differentiation was partially restored in this model by pre-treatment of the activated-PBMCs with the anti-inflammatory drug teriflunomide²¹⁵. Other studies have similarly showed several pro-myelinating compounds including clemastine, benztropine, miconazole, and clobetasol were unable to overcome fibrinogen’s or CSPGs’ extrinsic inhibition of OPC differentiation^216,217. These studies demonstrate the capability of MS-relevant extrinsic factors to inhibit the therapeutic efficacy of multiple remyelination therapeutic compounds that directly target OPCs.

The identification of effective novel therapeutic compounds for remyelination and neuroprotection is limited in part by the current animal models used to pre-clinically assess such compounds. Many demyelinating models such as the cuprizone chemical-demyelinating model do not accurately reflect the microenvironment and chronic inflammatory milieu associated with demyelinated plaques and progressive disease in PwMS, and inflammatory models such as EAE are accompanied by extensive axonal pathology with few demyelinated axons amenable to remyelination. Development of more relevant models is paramount for the discovery of effective therapies.

Overcoming Barriers to Remyelination and Neuroprotection

Including Extrinsic Inhibitory Factors in Pre-Clinical Screening Models

One strategy to develop drugs that overcome the barrier posed by the inhibitory microenvironment of MS lesions is to incorporate identified inhibitory signals into remyelination and neuroprotection screening assays. Several groups have developed such models that begin to address this issue. Petersen et al. added fibrinogen to primary rat OPC differentiation media prior to screening for myelin-promoting compounds in order to mimic one key component of the inhibitory lesion environment. In this screen, they found that 7 previously-identified pro-OPC-differentiation compounds failed to work in the presence of fibrinogen; however, the BMP type I receptor inhibitor DMH1 was identified as a novel molecule that could restore OPC differentiation to mature oligodendrocytes despite the presence of fibrinogen²¹⁶.

In addition to in vitro screens, extrinsic inhibitory factors can be incorporated animal models of remyelination and neurodegeneration. For example, the toxin cuprizone induces demyelination, and effective remyelination occurs when it is stopped. Several models have been developed in which cuprizone is combined with CNS inflammation, which may more closely resemble an MS lesion, such as (i) adoptive transfer of myelin reactive Th17 cells into previously cuprizone-fed mice^218,219, (ii) immune boost of complete Freund’s adjuvant and pertussis toxin without exogenous myelin peptides in cuprizone-fed mice²²⁰, and (iii) cuprizone-fed mice with ectopically-expressed inducible IFNγ production in astrocytes (GFAP/tTA mice x TRE/IFNγ)⁷⁷. While improved, these models may still not adequately model the chronic gliosis, fibrinogen, or deposition of CSPGs that inhibit remyelination at later stages of lesion evolution. In the IFNγ-cuprizone model, which allows for assessment of OPC differentiation and remyelination in an IFNγ-high environment, prolonging the integrated stress response with the compound Sephin1 increased the number of remyelinating oligodendrocytes and remyelinated axons. However, the thickness of axonal myelin only returned to near pre-lesion levels when Sephin1 treatment was combined with the OPC pro-differentiation compound bazedoxifene⁷⁸. These in vitro and in vivo studies underscore the importance of screening for therapies in a more relevant environment and the potential benefit of combinatorial approaches to promote remyelination and neuroprotection in an inflammatory microenvironment.

Compartmentalized CNS Inflammation

Using the novel methodologies described above, new compounds have been identified that can promote remyelination in the setting of individual extrinsic inhibitory factors. However, a strategy that may work in one inhibitory environment (e.g. high levels of IFNγ) may not work in another (e.g. excessive fibrinogen). Many, but not all, of the identified factors that promote the perilesional microenvironment associated with inhibition of remyelination and ongoing neurodegeneration are downstream of persistently reactive and pro-inflammatory glia. Thus, a two-pronged approach of targeting OPCs in addition to reactive glia may more effectively facilitate myelin repair in PwMS. A similar approach targeting intrinsic pro-survival pathways in neurons while inhibiting pro-cell-death extrinsic signals from reactive glia may be needed for an effective neuroprotective approach in chronic inflammatory diseases.

Current high-efficacy DMTs such as anti-CD20 and anti-alpha4 integrin monoclonal antibodies are highly protective against clinical relapses and new acute inflammatory lesions in RRMS, but do not target the compartmentalized CNS inflammation and reactive glia observed in PMS. Interestingly, multiple other DMTs that are not as effective against clinical relapses may potentially target CNS compartmentalized inflammation. Dimethyl fumarate (DMF) has well-described inhibitory effects on NF-κB pathway activation, an innate immune pathway highly upregulated in disease-associated microglia and astrocytes in MS¹¹². DMF decreases susceptibility in PRLs 2.3x faster than glatiramer acetate, suggesting DMF may improve the pathology of chronic active lesions²²¹, a correlate of neurodegeneration and disease progression in MS. Identifying therapeutic targets that can effectively dampen the smoldering inflammation and innate immune activation within the CNS of PwMS is a critical unmet need.

Emerging Strategies for Limiting Neurodegeneration and Promoting Remyelination in MS

Targeting CNS Compartmentalized Inflammation

Given the increasing appreciation of CNS-compartmentalized inflammation in PwMS, several potential strategies to target CNS disease-associated microglia, astrocytes, macrophages, and B-cells are under investigation. Bruton tyrosine kinase (BTK) is an intracellular signaling protein that regulates activation, differentiation, and survival of B-cells, microglia, and possibly astrocytes²²², and the efficacy of several BTK-inhibitors (BTKIs) on MS relapses and progressive disease are currently being investigated in Phase III clinical trials. In the evolutionRMS 1 (NCT04338022) and evolutionRMS 2 (NCT04338061) trials, the only BTKI Phase III trials in PwMS with reported results, the BTKI evobrutinib failed to show superiority to oral teriflunomide in reduction of annualized relapse rate, potentially partly due to an unexpectedly low relapse rate in the teriflunomide arm of the study. While trials of other BTKIs in RMS are ongoing, these results suggest that BTKI monotherapy may not be highly efficacious against acute MS relapses. However, the inhibition of microglial and astrocytic BTK may still have benefits in targeting glial inflammation and thus may still represent a possibly beneficial drug for patients with non-relapsing PMS. Multiple BTKIs are in clinical trials for non-relapsing SPMS (NCT04411641) and PPMS (NCT04458051, NCT04544449).

Targeting the complement pathway in MS is another potentially viable therapeutic option that may modulate CNS-compartmentalized glial inflammation. Of note, a C3 inhibitor pegcetacoplan was recently FDA approved for treatment of macular degeneration, and the C5-blocking antibody eculizamab is FDA approved for neuromyelitis optica, highlighting the potential for translating complement-targeting drugs into human patients. Inhibition of signaling between the opsonizing form of C3 and its receptor CR2 with a pair of CR2-conjugated inhibitors ameliorated both acute and chronic disability in EAE²²³. The same group found that inhibition of the alternative complement pathway with a monoclonal antibody against complement factor B (mAb 1379) reduced chronic demyelination without affecting the acute phase of the disease, implying it may have promise in progressive MS²²⁴. More recently, treatment of EAE mice with a monoclonal antibody inhibitor of C1q (ANX-M1.21) reduced the concentration of MIMS in the hippocampi of EAE mice at chronic timepoints, highlighting the role that complement might play in fostering inflammatory environments¹¹². In addition, an AAV-Crry was recently shown to rescue visual function and synaptic pathology in EAE¹⁶². However, despite these preclinical successes, no clinical trial has attempted to investigate complement inhibitors in PMS.

Several other therapeutic strategies that target glial activation are under investigation in clinical and pre-clinical studies. Inhibitors of other tyrosine kinases involved in innate immune activation include mastinib which inhibits the c-Kit pathway²²⁵. A randomized, double-blind Phase III clinical trial showed that low-dose, but peculiarly not high-dose, mastinib treatment was associated with minimally less disease progression as measured by change EDSS from baseline in patients with PPMS and non-relapsing SPMS²²⁶. Astroglial toll-like receptor (TLR) signaling, including through TLR4, also have central roles in innate immune activation and have been shown to contribute to neuroinflammation and MS pathogenesis^227,228. A phase II trial in PMS has shown an association between treatment with the anti-TLR4, phosphodiesterase type 5 (PDE5)-inhibiting, and macrophage migration inhibitor factor (MIF)-inhibiting compound ibudilast and slower progression of brain atrophy in progressive MS²²⁹.

The growing evidence of the role of glial activation and CNS compartmentalized-inflammation in the chronic neurodegeneration observed in progressive MS and the early, albeit minimal, positive signals of the strategies above have led to increased pre-clinical identification and testing of several other potential therapeutic targets to regulate the innate immune system and glial activation in MS. Evaluation of these targets should include determining whether they can decrease the expression or release of inhibitory factors as described above from reactive glia and in demyelination models. It should further be determined whether there are synergistic effects when these glial-targeting therapies are used in combination with remyelination and neuroprotective therapies that target OPCs and neurons, respectively.

Targeting Metabolism

Additional therapeutic strategies related to modulating metabolic pathways have shown some promise in immunomodulation, neuroprotection, and remyelination. α-lipoic acid, a naturally occurring antioxidant that has immunomodulatory effects on both the adaptive and innate immune systems²³⁰, is associated with slower progression of brain atrophy in a phase II trial in PMS²³¹. Statins have been shown in several models to have anti-inflammatory and neuroprotective effects, and the MS-STAT1 Phase II trial showed that high-dose simvastatin was well tolerated and reduced the annualized rate of whole brain atrophy by 43% in patients with SPMS²³². A multi-center Phase III trial of high-dose simvastatin verses placebo in SPMS (MS-STAT2) is currently ongoing with results expected in 2025 (NCT03387670). Circulating bile acid metabolites have been shown to be lower in PwMS compared to healthy controls and this alteration in bile acid metabolism may relate to alterations in the gut microbiome^233,234. The endogenous bile acid, tauroursodeoxycholic acid (TUDCA), has been shown to prevent neurotoxic polarization of astrocytes and pro-inflammatory polarization of microglia in vitro and reduce severity of EAE²³³. Current evidence points to a potentially protective effect of TUDCA in multiple neurodegenerative diseases, including amyotrophic lateral sclerosis, Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease^235,236. A recent observational cohort study demonstrated that higher primary bile acid levels at baseline predicted slower whole brain and retinal layer atrophy in PMS and that TUDCA supplementation was safe, increased serum levels of multiple bile acids, decreased circulating central memory CD4+ and Th1/17 cells, and increased CD4+ naive cells compared to placebo²³⁷. As expected, TUDCA supplementation was also associated with alterations in the composition and function of gut microbiota. Additional pre-clinical screens have identified multiple endogenous metabolites that promote OPC differentiation at physiological-relevant levels including taurine, an aminosulfonic acid, which was able to increase the efficacy of drug-induced OPC differentiation with benztropine and miconazole²³⁸. These studies highlight the growing appreciation of metabolism and immunometabolism and the potential of several metabolic strategies to modulate glial-activation, neuronal survival, and remyelination in PwMS.

Disinhibiting CNS Repair: Targeting Inhibitory Factors

In addition to inflammation and reactive glia, the identification of several specific extrinsic factors that may inhibit remyelination has led to pursuit of therapeutic approaches directly targeting some of these factors. Anti-fibrin immunotherapy, which can enter CNS and bind parenchymal fibrin without interfering with clotting, inhibited amyloid-associated and autoimmune demyelination-driven neurotoxicity in animal models²³⁹. The defibrinogenating agent ancrod has also been shown to enhance remyelination in the lysolecithin focal demyelination model²³⁹. Similarly, anti-CSPG compounds that competitively inhibit the synthesis of CSPGs by astrocytes were able to reduce CSPG expression and promoted remyelination in vitro and in vivo^217,240. These studies demonstrate the potential clinical benefit of targeting detrimental extrinsic factors, such as fibrinogen and CSPGs, to ameliorate neurotoxicity and facilitate CNS repair. Antagonists of these and other inhibitory signals of OPC differentiation may themselves promote remyelination and may also increase the efficacy of pro-remyelination therapies when used in combination.

Targeting Novel Cell Death Pathways

Another emerging avenue for neuroprotection and oligoprotection in MS is targeting recently-identified novel cell death pathways. Necroptosis is a form of necrotic cell death where inflammatory stimuli, including TLR4 activation and TNFR1 signaling among others, activate RIPK1, leading to formation of the necroptosome²⁴¹. Ferroptosis is another form of necrotic cell death driven by iron-dependent phospholipid peroxidation and regulated by multiple metabolic and signaling pathways²⁴². Upon extensive DNA damage, poly-ADP ribose (PAR) polymerase-1 (PARP-1) can initiate the form of cell death termed parthanatos that is ultimately executed by macrophage migration inhibitory factor (MIF) nuclease activity²⁴³.

Necroptosis, ferroptosis, and parthanatos have been implicated in neurodegeneration in PwMS^244–246, as well as in several other neurodegenerative diseases^{241,247–249}. Moreover, inhibition of necroptosis, ferroptosis, and parthanatos with RIPK1 inhibitors, ferrostatin-1 analogs, and PARP-1 or MIF nuclease inhibitors like PAANIB-1 respectively reduce neuronal loss in neurodegenerative diseases^244–246. Necroptosis, ferroptosis, and parthanatos signaling have all been shown to contribute to the generation of neurotoxic microglia and astrocytes^{244,250–252}. Additionally, glial ferroptosis and necroptosis have been shown to reduce OPC differentiation and OL myelinating capacity^244,253. Thus, inhibiting these cell death pathways may both directly prevent OL and neuronal cell death as well as mitigate disease-state associated oligotoxic and neurotoxic glia.

Conclusions

The need for new therapies that address the chronic neurodegenerative aspect of MS is growing. Despite an increasing understanding of the mechanisms by which resident glia of the CNS participate in these chronic inflammatory processes that drive this neurodegeneration, there is a paucity of novel therapeutics to address the progressive phase of MS. As we have discussed, there is a dire need for animal models that faithfully recapitulate progressive disease and remyelination in the context of inflammation. Additionally, the goal of effectively remyelinating axons in the setting of MS requires targeting both OPCs directly as well suppressing the inflammatory microenvironment presented by reactive astrocytes and microglia. Finally, in order to adequately treat PwMS, we need to specifically measure outcomes that correspond to CNS inflammation, progressive neurodegeneration, and remyelination in future clinical trials. Together, evaluation of MS as a complex, multi-faceted disease may afford future clinical trials more success.

Figure 3: Pharmacologic efforts to promote myelination and neuroprotection.

A) An oligodendroglial cell showing the targets for drugs that have been shown to promote oligodendroglial differentiation, maturation, and/or myelination. Several identified inhibitory factors of remyelination are also depicted with flathead arrows directed to the drugs that these inhibitors been shown to target. Notably, DMH1 is not inhibited by fibrinogen and Sephin1 upregulates the ISR even in the presence of IFNγ. B) Selection of key drugs and molecules that may help downregulate neurotoxic and oligotoxic reactive glia. Also displayed are pharmacologic inhibitors of the non-apoptotic cell death pathways necroptosis, ferroptosis, and parthanatos. Made in BioRender.