Oligodendrocyte progenitor cell recruitment and remyelination in multiple sclerosis: the more, the merrier?

Abstract

Promoting remyelination to prevent/reduce neurodegeneration in patients with multiple sclerosis (MS) is a major therapeutic goal. The longstanding view that the block of oligodendrocyte progenitor cell (OPC) differentiation in MS lesions is the leading cause of remyelination failure has inspired the scientific community to focus primarily on OPC differentiation-promoting compounds as pro-remyelinating agents. Yet, these strategies have been challenged by findings that active MS lesions contain surviving oligodendrocytes that may contribute to remyelination, while many chronic lesions contain low numbers of oligodendroglial cells. In addition, clinical trials using differentiation-stimulating drugs have shown limited efficacy. Thus, a strategic shift in the design of potential remyelination-promoting therapies may be required to achieve significant clinical benefits, which calls for a careful reconsideration of the mechanisms underlying remyelination failure in MS. Here, we argue that both the rate and the efficacy of OPC recruitment are fundamental determinants of remyelination, and that stimulating this process in MS may be crucial to achieve myelin regeneration.

We first review different types of MS lesions in early and chronic MS, with a particular focus on OPCs and surviving oligodendrocytes. Based on the neuropathological findings and results obtained using models of demyelination, we make the case that OPC differentiation block in chronic MS is likely the consequence of defective OPC recruitment during earlier phases of the disease, because (i) if the recruitment is too slow, OPCs reach the axons after what we define as ‘remyelination-permissive window’, and thus remain undifferentiated; and (ii) if the recruitment is inefficient, OPC density in the lesions remains below the threshold required for differentiation.

Importantly, we highlight that OPC proliferation in MS lesions is scarce, which strongly suggests that repeated episodes of demyelination/remyelination (OPC differentiation) will deplete the lesional OPC pool unless perilesional OPCs are recruited. We also point out that surviving mature oligodendrocytes in a subtype of early MS lesions may actually prevent the recruitment of OPCs. Because it has been suggested that OPC-mediated remyelination may be more efficient than that by surviving oligodendrocytes, we suggest that stimulating OPC recruitment during active disease should benefit remyelination in multiple types of lesions, including those with spared oligodendrocytes. Finally, we review molecular determinants of OPC recruitment and suggest a potential therapeutically-relevant strategy to increase this process in patients with MS.

multiple sclerosis, myelin, oligodendrocyte progenitor cell, regeneration, remyelination

Introduction

Despite recent advances in immunotherapies for patients with multiple sclerosis (MS), preventing the development of irreversible neurological handicap remains an unmet clinical need.¹ Because neurological disability is directly related to axonal and neuronal loss in the disease,² the development of neuroprotective strategies is crucial to prevent disease progression. Remyelination of demyelinated axons is associated with improved axonal survival both in experimental models^3,4 and MS lesions.⁵ Thus, stimulating remyelination in patients with MS is considered as a potential neuroprotective strategy.

The longstanding view that impaired differentiation of oligodendrocyte progenitor cells (OPCs), observed in MS, is the main explanation for inefficient remyelination has been recently challenged by observations that active demyelination in early MS is associated with preservation of mature oligodendrocytes in a subset of patients, and that chronic lesions show oligodendroglial depletion (reviewed and discussed by Franklin et al.⁶). These observations question the potential efficacy of pro-remyelinating strategies based on drugs that stimulate OPC differentiation into oligodendrocytes, and may explain modest functional benefits and scarce impact on the clinical outcome in clinical trials testing these compounds (reviewed by Lubetzki et al.¹) Thus, achieving clinical benefits in patients with MS may require complementary/alternative approaches to stimulate remyelination.

This review makes the case that oligodendroglial depletion is an important determinant of remyelination failure in a large subset of white matter lesions in patients with MS. In addition, we emphasize that slow repopulation of demyelinated areas by OPCs can result in lesions that contain OPCs and oligodendrocytes but remain demyelinated because the remyelination-permissive time window has passed. We describe that stimulating OPC recruitment may benefit remyelination in multiple ways, which might also be relevant for increasing remyelination in demyelinating lesions with spared mature oligodendrocytes. Importantly, we suggest a therapeutically-relevant paradigm to increase OPC recruitment in patients with MS.

Oligodendroglial numbers in multiple sclerosis lesions

In principle, demyelination can occur either because oligodendrocytes die, or because myelin is targeted by a pathological process while oligodendrocytes are spared, or because oligodendrocytes survive but stop synthesizing myelin. Therefore, the question of oligodendroglial persistence/loss in demyelinated MS lesions has been a subject of numerous studies. These have investigated the numbers of both OPCs and mature oligodendrocytes in early and/or chronic MS, with the ultimate goal of elucidating the mechanisms underlying the failure of remyelination in chronic lesions.

Early multiple sclerosis lesions

An important insight into the extent of oligodendroglial loss in early MS was provided in 1999 by Lucchinetti and colleagues.⁷ These authors analysed 395 lesions (biopsies and autopsies) from 113 patients with less than 5 years of disease duration in most cases and performed quantification of oligodendroglial lineage cells to determine the extent of destruction of these in MS lesions. Mature oligodendrocytes were identified using immunolabelling for myelin oligodendrocyte glycoprotein (MOG), while oligodendroglia (oligodendrocytes and OPCs) were identified using in situ hybridization for proteolipid protein (PLP) mRNA.

The authors identified two main types of lesions in early MS:

• In type I lesions, oligodendroglial cells were present. These represented around 70% of all lesions and could be subdivided in three subcategories: those with loss of mature oligodendrocytes and increased numbers of PLP mRNA+ (presumably immature) cells as compared to early active areas (41% of all lesions), those with minimal oligodendrocyte loss (13%), and those with extensive oligodendrocyte loss not accompanied by increase in numbers of PLP mRNA+ oligodendroglia (compared to early active areas). Type 1 lesions were characterized by remyelination, and remyelinated areas showed increased oligodendroglial numbers.

• In type 2 lesions, there was evidence of general oligodendroglial loss (both oligodendrocytes and immature oligodendroglia), that was either total (16% of all lesions) or gradual, from the plaque edge to the centre (14%). Type 2 lesions showed no evidence of remyelination.

Thus, as long as oligodendroglial cells were present, there was evidence of remyelination of spared axons, given that axonal loss could be extensive (maximal in inactive demyelinated regions, and reduced in remyelinated lesions). Notably, these different lesion types were characteristic to the patient, so that a single patient consistently presented a particular type of the lesion. In general, type 1 lesions were observed in patients with relapsing-remitting or secondary progressive MS, while type 2 lesions were observed predominantly in patients diagnosed with primary progressive MS or acute MS. Indeed, later studies highlighted that patients showing predominant oligodendroglial loss in the lesions had higher disability and were more likely to have progressive disease.⁸ This concept of lesion heterogeneity in terms of oligodendroglial pathology was later further elaborated into a description of four distinct pathogenic mechanisms of early lesion formation.^9,10

Preservation of mature oligodendrocytes in a subset of active multiple sclerosis lesions

Importantly, the above-mentioned study⁷ showed that loss of PLP mRNA could be more extensive than that of MOG+ mature cells in early active lesions. This suggests that either OPCs are also lost in these areas or that mature oligodendrocytes are preserved but stop synthesizing PLP, thus myelin. Bruck and colleagues¹¹ analysed oligodendroglial preservation in active MS lesions using immunohistochemistry for MOG. They observed that a subset of early lesions shows preservation of mature oligodendrocytes. In addition, Wolswijk¹² noted that lesions with evidence of recent demyelinating activity contained mature oligodendrocytes positive for GalC and MOG, but that lacked processes suggesting these were not engaged in myelination. Macrophage presence (indicating ongoing demyelination) was positively correlated with numbers of ‘demyelinated oligodendrocytes’,¹² so that lesions lacking macrophages also lacked oligodendrocytes. In older lesions, these cells were present only around the borders. Similarly, a study by Heß and colleagues¹³ showed that oligodendrocytes labelled with TPPP/p25 or NogoA are preserved in active lesions but lost in the centre of chronic lesions.

Therefore, oligodendrocytes seem to survive demyelination in a subset of early MS lesions, but become progressively depleted during the course of the disease.

Chronic multiple sclerosis lesions

Early studies by Gus Wolswijk^14,15 were the first to report presence of OPCs in chronic lesions of patients with MS. These studies showed highly variable OPC densities between the lesions, as well as within different regions of large lesions. The same author also compared brain versus spinal cord lesions using O4 as a marker of oligodendroglial lineage and GalC as a marker of mature oligodendrocytes, so that O4+ GalC− cells were considered as OPCs and O4+ GalC+ cells as oligodendrocytes.¹⁶ Several points were highlighted: (i) OPCs were found in both brain and spinal cord lesions,^12,15,16 but their numbers decreased with disease progression and the age of the patients¹⁶; (ii) differentiated oligodendrocytes were found in lesions that contain higher OPC densities; and (iii) spinal cord lesions contained fewer OPCs compared to brain lesions from the same patient, which was attributed to the fact that spinal lesions appeared older (contained fewer macrophages). These observations thus suggest progressive oligodendroglial loss with the disease duration, attributed, at least in part, to defective OPC recruitment and proliferation.

Chang and colleagues^17,18 used NG2 as a marker to investigate OPC numbers, and immunohistochemistry for PLP to label pre-myelinating oligodendrocytes. They observed that while NG2 cell density was lower in the demyelinating white matter lesions as compared to normal tissue, gray matter lesions did not show changes in NG2 cell density. In active and chronic active white matter lesions, NG2+ cells were found on the edges but not in the lesion, which suggested that active inflammation leads to loss of NG2+ cells, or the loss of NG2 immunoreactivity. The centre of chronic inactive lesions, however, contained NG2+ cells, but the density of these was significantly reduced compared to the normal appearing white matter (NAWM). In addition, the authors identified two different morphologies of NG2+ cells: stellate shaped cells, present both in controls and in MS patients, and elongated cells, present only in MS patients, exclusively in those lesions that also contained pre-myelinating oligodendrocytes. Pre-myelinating oligodendrocytes were detected in 70% of the lesions analysed, but these were present only in certain areas, in clusters, while other areas showed depletion, so that the overall lesional density was very low. Importantly, lesions derived from patients with longest duration of the disease were free of pre-myelinating oligodendrocytes and contained very low numbers of stellate-shaped NG2+ cells.

Of note, the above-mentioned studies by Wolswijk^12,15,16 and Chang and colleagues^17,18 are frequently cited as an argument that OPCs are indeed detected in chronic MS lesions, but the fact that their density is significantly affected is usually overlooked.

More recent studies have analysed oligodendroglial numbers using antibodies specific for oligodendroglia-specific transcription factors such as Olig1¹⁹ and Olig2.^13,20,21 These showed that, while active MS lesions contain OPCs, chronic MS lesions show a significant decrease in the density of oligodendroglial cells. Moreover, an increase in OPC density in the periplaque white matter (PPWM) compared to NAWM was reported,²¹ and one study showed increased oligodendroglial density in the shadow plaques.¹⁹ These studies also attempted to distinguish between undifferentiated OPCs and OPCs undergoing differentiation^19,20 or differentiated oligodendrocytes.²¹ In the study by Kuhlmann and colleagues,²¹ a distinction was made between cells strongly positive for Olig2 (OPCs) and those weakly positive (oligodendrocytes). OPCs were classified either as differentiating (NogoA+) or non-differentiating (NogoA−). Differentiating OPCs were found in patients with early multiple sclerosis only, in PPWM, actively demyelinating lesions and shadow plaques, but not the demyelinated ones. Non-differentiating OPCs in chronic MS were found in PPWM, and a strong decrease in OPC numbers was found in chronic lesions. Similar results were obtained by Wegener et al.,²⁰ who observed high densities of Olig2+ cells in PPWM and shadow plaques, but a drastic reduction in chronic lesions. The study by Boyd and colleagues²² showed OPC loss in a subset of active (3/9) and chronic inactive (2/6), and the majority of chronic active lesions (5/6). Different results were reported by a more recent (also more extensive) study¹³ using similar markers that showed that Olig2 densities are not significantly affected in active lesions, but become negligible in the centre of inactive and mixed lesions. The study by Tepavčević and colleagues¹⁹ used Olig1 as an oligodendroglial marker. The Olig1+ NogoA+ cells (differentiating/differentiated oligodendroglia) were found to be the predominant oligodendroglial population in actively demyelinating lesions, but were not observed in chronic lesions. The density of Olig1+ cells was drastically reduced in chronic lesions and increased in remyelinated shadow plaques. One of the differences between Olig1 and Olig2 labelling is that, as mentioned by Wegener et al.²⁰ and Kuhlmann et al.,²¹ levels of Olig2 are high in OPCs, particularly reactive OPCs, and low in mature oligodendrocytes, while Olig1 is expressed in both populations.²³

Interestingly, a study by Jakel and colleagues²⁴ observed a general reduction in OPCs in both NAWM and lesions in patients with progressive MS as compared to the controls. These observations were made for total OPC nuclei and, using immunohistochemistry for Sox6 and in situ hybridization for Brevican (BCAN), markers found to be enriched in OPCs. Moreover, RNA sequencing performed in this study highlighted altered oligodendroglial heterogeneity in the patients with MS, with the preferential loss of an intermediate oligodendroglial population (likely differentiating OPCs).

Presence of differentiating OPCs in MS lesions was also investigated using markers specific for this stage of oligodendroglial differentiation. A study by Moll and colleagues²⁵ identified Sox17 as a marker of differentiating OPCs. In MS lesions, numbers of Sox17+ cells were significantly decreased in chronically demyelinated lesion core. In regions undergoing active demyelination, numbers of these cells were comparable to those in the control tissue, while the PPWM showed an important increase in Sox17+ cell density. A breast carcinoma amplified sequence 1 (BCAS-1) was identified as another marker of newly generated oligodendrocytes, expressed only transiently during oligodendroglial differentiation.²⁶ In chronically demyelinated MS lesions, BCAS-1 expression was rare. In chronic active lesions, BCAS-1 cells were detected in remyelinated areas around the border but not in the lesion core. Moreover, BCAS-1 density was increased in the PPWM next to the lesions showing remyelination.

Thus, it can be concluded that early MS lesions undergoing active demyelination contain variable amounts of preserved oligodendrocytes and OPCs, ranging from complete destruction to normal densities. In those lesions that contain OPCs, ongoing OPC differentiation is documented. Chronic lesions, however, generally show decreased OPC densities, combined with lack of OPC differentiation and loss of mature oligodendrocytes.

Importantly, the extent of oligodendroglial loss, as shown by Lucchinetti and colleagues,⁷ is likely to be patient-dependent. This is usually not reflected in the results presented in the above-mentioned neuropathological studies, in which lesions from different blocks and patients are pooled for comparison with NAWM etc. Thus while the pooled results obtained still reflect OPC loss in chronic lesions compared to NAWM and non-neurological controls, this depletion is likely even more extensive in specific subset of patients or in specific regions that show increased oligodendroglial loss, such as the spinal cord.¹⁶

Another recurrent observation by the studies cited above is that the density of OPCs and pre-myelinating oligodendrocytes is increased in the PPWM.^20,21,25,26 This suggests that oligodendroglial recruitment to the lesions, observed in animal models of demyelination, may be blocked at the borders of MS lesions.

The summary of the main findings described above can be found in Table 1.

Table 1

OPCs and spared oligodendrocytes in MS lesions

Study	Samples	Markers	Types of lesion	OPCs (% lesions examined)	Mature oligodendrocytes (/total cases)
Lucchinetti et al.7	113 cases (early MS); 395 lesions	MOG, PLP mRNA	Type 1: variable oligodendrocyte loss ± OPC repopulation; remyelination Type 2: total or gradual OPC/oligodendrocyte loss; no remyelination	↑(54%) ↓ (46%) Appears patient-specific; RM areas = ↑OPCs	Preserved in 13% of lesions (subset of Type 1)
Bruck et al.11	Six cases (early MS); young patients, 15–28 years of age	MOG, PLP mRNA	Early active A I Early remyelinating SP	—	Preserved 3/6 Reduction 1/6 Absent 2/6
Wolswijk12,16	C progressive; 15 cases,12 26 cases16	O4, GalC, MOG	CA CI	↓ in density with patient’s age and disease progression	Present in areas showing activity (macrophages), lack processes; Absent in inactive core
Chang et al.17,18	12 cases (11 progressive MS)17; 10 cases (nine progressive MS)18	NG2, PLP	A CA CI	WM lesions: ↓OPCs GM lesions: NC ↑ in pre-myelinating oligos in regions with ↑OPC density	—
Kuhlmann et al.21	43 cases (early and chronic MS)	Olig2, NogoA	A DM RM	Early MS: ↑OPCs in PPWM>RM>A>DM = NAWM; diff. in PPWM, RM and A Chronic DM: ↓OPCs; no diff.	—
Boyd et al.22	10 cases (progressive MS)	Olig2, NogoA	A CA CI	A: ↑5/9; ↓3/9; NA 1/9 CA: ↓5/6; 1 NA CI: ↑1/6; ↓2/6; 3/6 NA	—
Tepavcevic et al.19	Seven cases (progressive MS)	Olig1, NogoA	A CA border CI core SP	A: diff. OPCs CI: ↓OPCs SP: ↑OPCs	—
Wegener et al.20	Eight patients (progressive MS)	Olig2, NogoA	A CA border CI core SP	A: ↑OPCs (diff.) CA: ↓OPCs; CI: ↓OPCs SP: >CA and CI; < NAWM and A	—
Heß et al.13	62 cases (early and chronic MS)	Olig2, NogoA, TPPP/p25	Mixed (A core, C border) I	OPC diff is comparable between A lesions with or without REM	Present in A and border of I (subset)
Jakel et al.24	Four cases (progressive MS)	Single cell RNA seq; IHC: BCAN, Sox6	A CA CI SP	↓OPCs from progressive MS compared to controls (both NAWM and lesions); loss of diff. OPCs	Altered gene expression: ↑ in genes related to myelination
Moll et al.25	Three cases, five tissue blocks (progressive MS)	Sox17 (diff. OPCs)	A lesion border C lesion core	PPWM:↑ diff. OPCs A: NA C: ↓diff. OPCs	—
Fard et al.26	24 cases (progressive MS)	BCAS-1 (early myelinating oligos)	RM border of C lesions CI core SPs (fully RM lesions)	Remyelinating lesion border: ↑ Core: absent	—

A decrease in OPC density is consistently observed in chronic lesions, some of which show OPC depletion. A = active; C = chronic; CA = chronic active; CI = chronic inactive; diff. = differentiating; DM = demyelinated; GM = grey matter; I = inactive; NA = not affected; RM = remyelinated; SP = shadow plaque; WM = white matter.

Why are numbers of OPCs reduced in chronic multiple sclerosis lesions?

There are several potential explanations for low OPC densities in MS lesions. These include OPC destruction that can be mediated by several mechanisms, as well as the exhaustion of OPC pool following repeated episodes of demyelination/remyelination.

OPC destruction

Several reports highlight that OPCs can be directly targeted in MS, sometimes even at a higher extent than mature oligodendrocytes.²⁷ The mechanisms underlying OPC loss in MS lesions include antibody binding to OPCs, OPC targeting by cytotoxic T cells and OPC death due to mitochondrial damage.

Antibodies to OPCs

It was reported that sera from patients with active MS (relapsing-remitting) contain antibodies to OPC antigen AN2 (NG2), and that these antibodies prevent OPC differentiation in vitro.²⁸ A study by Lily and colleagues²⁹ investigated binding of antibodies from MS patient sera to neuronal and oligodendroglial surface antigens. Significant binding of IgMs and IgGs to oligodendrocyte progenitor cell-derived cell lines was found in 50% of sera from both relapsing-remitting and secondary progressive disease. Interestingly, the binding was lost when OPC lines were induced to differentiate, which is consistent with an earlier report showing that CSF immunoglobulins from patients with MS do not bind to GalC-positive oligodendrocytes.³⁰

Cytotoxic T cells

OPCs could also be targeted by cytotoxic T-cell response in MS lesions. Cytotoxicity by CD8+ T lymphocytes requires antigen presentation by the target cell via major histocompatibility complex I (MHC I) molecules. While under basal conditions in vitro OPCs do not express detectable levels of MHC I protein, exposure of these cells to inflammatory cytokines in vitro leads to MHC I upregulation.^31–33 Moreover, haplotype-mismatched OPCs undergoing graft rejection were found to upregulate MHC I and interact with CD8+ cells, which was followed by failure of remyelination, suggesting that OPCs become targeted by cytotoxic response.³² In addition, Kirby and colleagues³³ showed that interferon gamma (IFNγ) both in vitro and in vivo induces MHC I expression and antigen presentation to T cells, which resulted in decreased numbers of OPCs. Importantly, in a mouse model, upregulation of MHC class I expression by IFN gamma was coincident with the switch by OPCs from normal proteasome to immunopreoteasome.³³ This switch was also observed in the tissue of patients with MS, specifically in areas of chronic demyelination.³³ Thus, these data strongly suggest that under inflammatory demyelinating conditions, OPCs become targets of cytotoxic response by CD8+ T cells via MHC I. In addition, MHC I can also play roles other than antigen presentation.^34–36 One of these is negative regulation of proliferation via inhibition of fibroblast growth factor receptor (FGFR) signalling, as observed in neural progenitor cells,³⁶ which could lead to impaired OPC proliferation in response to demyelination, thus failure to expand oligodendroglial pool in the disease.

Interestingly, it was suggested that OPCs positive for MHC I can also upregulate MHC II,^33,37 the significance of which in inflammatory demyelination remains to be determined.

Mitochondrial damage-induced susceptibility

The pathology of MS has been associated with energy metabolism impairments, such as mitochondrial pathology in various CNS cell types, including oligodendroglia.^38,39 Several studies have highlighted that OPCs appear particularly vulnerable to metabolic alterations and nutritional deficiencies, more than mature oligodendrocytes (reviewed by Tepavčević⁴⁰).

For example, exposure of rodent oligodendroglia to glucose deprivation induced significant alterations in the processes of OPCs but not those of mature oligodendrocytes.⁴¹ Such increased vulnerability of immature oligodendroglia could be explained by differences in the rate of glycolysis versus oxidative phosphorylation/mitochondrial metabolism according to the differentiation stage of oligodendroglial cells. Thus, both rat and human OPCs had higher oxygen consumption compared to oligodendrocytes.⁴² Accordingly, it was also observed that, under optimal conditions, cultured OPCs and oligodendrocytes derived from postnatal rat brain rely on oxidative phosphorylation for ATP synthesis, while oligodendrocytes isolated from the adult rat brain preferentially use glycolysis,⁴³ which suggests increased susceptibility of immature cells to oxidative damage. Interestingly, while the treatment of human cells with rotenone (complex I inhibitor) did not affect the survival of undifferentiated OPCs nor differentiated oligodendrocytes, it compromised the viability of differentiating oligodendroglia,⁴⁴ suggesting that newly generated oligodendrocytes/late OPCs are the most vulnerable differentiation stage.

Therefore, OPCs may become targeted by multiple mechanisms in MS, which can explain their depletion in a subset of active MS lesions, including those that contain mature oligodendrocytes.

Exhaustion of OPC pool following repeated episodes of demyelination/remyelination

Besides being directly targeted by the inflammatory process, OPC pool could also progressively diminish as a consequence of differentiation in response to repeated demyelinating insults. It has been documented that MS lesions undergo several, sometimes numerous episodes of demyelination.⁴⁵ If OPC differentiation in these lesions is not compensated by OPC pool renewal (proliferation/recruitment), this will eventually lead to OPC depletion/low OPC densities.

Scarce OPC proliferation/extensive differentiation in active multiple sclerosis lesions

Most of the above-mentioned studies investigating OPC numbers in MS lesions have also addressed OPC proliferation in using different markers such as Ki67^12,13,15 and proliferating cell nuclear antigen (PCNA).^19,20 The results show that OPCs in active lesions are mostly undergoing differentiation, while only a minor proportion undergoes proliferation (as judged by PCNA expression). Some of the studies show basically negligible proliferating fraction of OPCs.^12,13,15 Even though all the studies consistently show that OPC proliferation does not appear to be a significant event,^{12,13,15,19,20} the differences in the proportions detected (10–20%^19,20 versus occasional cells^12,13,15) reflect the differences in the markers used with PCNA antibody^19,20 labelling more OPCs than Ki67.^12,13 Of note, in the studies using Ki67,¹⁵ extensive labelling was observed in a tumour detected in one of the patients demonstrating that the antibody was efficient in labelling proliferating cells. Because some studies have found that PCNA is less specific than Ki67 in detecting highly proliferating cells,⁴⁶ this can explain the differences in the results reported.

Low OPC proliferation in MS lesions is in agreement with the results obtained by Hughes and colleagues,⁴⁷ who investigated NG2 cell homeostasis in the adult murine brain. These authors found that while OPC proliferation was effectively induced by local depletion of NG2 cells, OPC division was rare prior to differentiation into myelinating oligodendrocytes, recorded in only seven out of 107 differentiating events. Thus, it could be that if OPCs are exposed to differentiation-inducing stimuli (e.g. demyelinating MS lesions), they undergo differentiation without expanding the OPC pool.

As mentioned above, it appears that OPC differentiation in active MS lesions largely exceeds OPC proliferation. This means that lesional OPC pool will decrease with each subsequent episode of demyelination, unless OPCs from neighboring non-affected tissue repopulate the area. The issue of OPC depletion with successive episodes of demyelination has been addressed in a study in which demyelinating lesions were induced three consecutive times in the young rat brain, and remyelination was evaluated.⁴⁸ No decrease in remyelination was observed. However, this study does not reflect OPC response in MS for several reasons. First, the experimental model used is characterized by very efficient OPC recruitment, possibly because reactive astrocytes and surviving oligodendrocytes, that express OPC chemorepellents such as netrin-1 in MS lesions,^19,49 are not present during the phase of OPC recruitment.¹⁹ Effectively, if early netrin-1 expression is artificially induced in a similar lesion model, OPC recruitment is diminished and remyelination impaired.¹⁹ Second, episodes of demyelination in MS most likely largely exceed the three employed in this study.⁴⁵ Moreover, the lesions were induced in young rats, which means that age-associated decline in OPC repopulation was not reflected.⁵⁰ Notably, oligodendroglial depletion following chronic exposure (6–7 months) to demyelinating agent cuprizone was identified as the primary cause of remyelination failure.⁵¹

Lesional OPC senescence induced by IFNγ

A potential contributor to the failure of OPC pool to expand in MS lesions could be the pro-senescent effect of the pro-inflammatory cytokine IFNγ. This cytokine upregulates the levels of the transcription factor paired related homeobox protein 1 (PRRX1) in mouse and human OPCs.^52–54 PRRX1 expression in the mouse brain and after induction of demyelination by lysolecithin was observed predominantly in quiescent OPCs. When IFNγ was co-injected with lysolecithin, the expression of PRRX1 increased on OPCs, and the percentage of Ki67+ OPCs decreased.⁵⁴ In addition, PRRX1 decreased proliferation and tissue colonization by transplanted human OPCs, which was not accompanied by increased differentiation.⁵⁴ The inhibitory effect of IFNγ on OPC proliferation could also be explained by an alternative/complementary mechanism which is the induction of MHC I expression on OPCs.^31–33 It has been shown that MHC I expression inhibits FGFR signalling and leads to a decrease in proliferation in neural progenitor cells.³⁶ It remains to be determined whether this mechanism is also active in OPCs during inflammatory demyelination. Therefore, IFNγ signalling may contribute to the failure of lesional OPCs to proliferate during demyelinating episodes, which further highlights the importance of efficient OPC recruitment from perilesional areas in maintaining the lesional OPC pool.

Thus, repeated demyelinating insults in aging patients with MS, associated with expression of molecules that prevent OPC proliferation and recruitment, could be a factor contributing to eventual exhaustion of the OPC pool in MS lesions.

Remyelination efficiency dependent on timing of OPC arrival: notion of remyelination-permissive period

The efficacy of remyelination is likely dependent on the timing of OPC arrival to the demyelinated axons. This notion is supported by transplantation studies using a model of demyelination in which endogenous remyelination was suppressed by X-irradiation, so that all myelin repair observed could be attributed to transplanted cells. Thus, when OPCs were transplanted at a distance from acutely demyelinated lesion, the efficiency of remyelination observed was significantly lower than when cells were transplanted directly into the lesion,⁵⁵ which suggested that increased interval between myelin loss and OPC contact with demyelinated axons is associated with diminished myelin repair. This is consistent with the observation that demyelinated axons that are not quickly remyelinated undergo changes that render them refractory to remyelination, such as the expression of PSA-NCAM, the molecule that prevents remyelination by OPCs.⁵⁶ In addition, several studies using toxin-induced models of remyelination have highlighted the importance of acute inflammation in the efficacy of remyelination. For example, inflammation associated with rejection of transplanted OPCs increased remyelination by endogenous cells.⁵⁷ Moreover, induction of acute inflammation was required to induce remyelination by grafted OPCs in a chronically demyelinated TAIEP (tremor, ataxia, immobility, epilepsy and paralysis) rat.⁵⁸ Importantly, macrophage depletion by clodronate liposomes was associated with decreased efficiency of myelin repair, related to presence of non-digested myelin debris.⁵⁹ Later studies demonstrated that macrophage-mediated effect was not due solely to myelin digestion, but also to specific signalling that stimulates OPC differentiation.⁶⁰ Specific subsets of adaptive immune system have also been shown to have positive effects on the efficacy of remyelination.^61,62 Thus, while inflammation in MS has been associated with appearance of demyelinated lesions, various signals associated with acutely inflammatory environment actually stimulate myelin repair (reviewed by Hohlfeld⁶³) as apparent in patient tissue in which remyelination is frequently observed in nascent lesions.^7,64,65 Moreover, increased time interval between demyelination and remyelination can lead to the formation of astrocytic barriers, that decrease OPC recruitment and subsequently, myelin repair.⁶⁶ Thus, rapid OPC arrival to demyelinated axons, during the ‘remyelination-permissive period’ characterized by axons receptive to myelination and presence of acute inflammation, should enhance OPC differentiation and myelin repair (Fig. 1).

Figure 1

Both the rate and efficiency of OPC recruitment determine the likelihood of myelin repair. (A) If OPCs are recruited quickly and at high density, these cells reach the axons shortly after myelin loss, during the ‘remyelination-permissive period’. This means that axons remain permissive to myelination, and acute inflammatory stimuli are still present. This stimulatory environment (stars) promotes OPC differentiation and remyelination.^58,60,67 (B) If OPC recruitment is inefficient, two scenarios can occur. First, chemorepellents may prevent the OPCs from reaching the lesions,^19,22,68 so these cells will remain at lesion borders, where they may or may not differentiate. Second, if OPC recruitment is too slow, these cells will reach the lesion once the acute inflammatory stimuli (that stimulate differentiation) are no longer present. In addition, prolonged interval between demyelination and oligodendrocyte-axon contact can render the axons refractory to remyelination.⁵⁶ This second scenario is inhibitory (X) to remyelination and can explain chronically demyelinated MS lesions that contain undifferentiated OPCs and/or pre-myelinating oligodendrocytes. Importantly, lack of rapid remyelination will increase the likelihood of axonal degeneration.⁴

High OPC density increases OPC differentiation

Several analyses of OPC numbers in different types of MS lesions (acute, chronic, remyelinated shadow plaques) demonstrate that remyelinated lesions show increased oligodendroglial density as compared to demyelinated lesions, and in some cases also as compared to the control tissue. This was shown using several markers such as Olig2^13,20,21 and Olig1.¹⁹ Importantly, an in vitro study investigated the importance of the density at which the OPCs were plated on the differentiation of these cells in OPC-neuron co-cultures.⁶⁹ In co-cultures with high OPC density, differentiation started at only 5 days after plating, whereas in lower density conditions, it started at 2 weeks after plating. While it is possible that factors secreted by OPCs may have increased the differentiation of neighbouring OPCs, this appeared unlikely because the block in differentiation of OPCs plated at low densities could be overcome by adding other cell types, thus increasing the total cell number. Because higher cell density increases tissue/substrate stiffness,⁷⁰ the modification of mechanical properties of the substrate appears as a likely explanation. Importantly, increasing substrate stiffness has been shown to promote cellular differentiation.^71,72 Thus, OPCs appear more likely to differentiate when they are present at higher numbers.

The in vitro observation that cell density-induced increase in tissue stiffness triggers OPC differentiation is consistent with the in vivo observations on the role of Piezo1, the mechanosensitive ion channel protein, in inducing OPC differentiation during CNS development. Loss of Piezo1 during developmental myelination led to an increase in OPC proliferation and OPC numbers without an increase in differentiation, which suggests that Piezo1 sensing of the tissue stiffness regulates the transition from OPC proliferation to differentiation.⁷³ Thus, it appears that in cell culture and young animals, promotion of tissue stiffness by increased OPC density stimulates differentiation via Piezo1-mediated mechano-sensing. However, exaggerated tissue stiffness in aged animals, attributable to extracellular matrix components rather than OPC density, is detrimental to aged OPC function and remyelination, which can be rescued by knocking down Piezo1.⁷³

While increased tissue stiffness may be one mechanism underlying OPC differentiation when these cells are present at high density, multiple other mechanisms are likely involved, and need to be elucidated. One of these, in the context of remyelination, may be the expression of the guidance molecule netrin-1, upregulated by adult OPCs recruited to demyelinating lesions,¹⁹ that acts both as chemorepellent and differentiation inducer for adult OPCs.^19,49 Thus, a progressive increase in the lesional OPC density mediated by OPC recruitment/proliferation, increases the levels of netrin-1 in the lesions, which then contributes to termination of OPC recruitment via chemorepellent effect and initiation of OPC differentiation.

Therefore, efficient recruitment of OPCs to demyelinating lesions will increase the likelihood of remyelination by these cells for at least two reasons: (i) ensuring OPC-axon contact during the ‘remyelination-permissive period’ (axons permissive+ stimulation of differentiation by acute inflammatory environment); and (ii) increased OPC density leading to OPC differentiation. Importantly, in rodent models of demyelination in which lesions are small, OPC repopulation of the demyelinated area is quick and efficient, but this is not the case in rabbits in which the cross sectional area of the lesion is 20 times that of rodent lesions, which results in failure to recover normal oligodendrocyte densities.⁷⁴ These observations are highly relevant for MS lesions in the white matter, that can be of considerable size, which has been negatively correlated with the percentage of remyelination.⁷⁵ Thus, improving OPC recruitment in MS lesions may improve OPC differentiation and increase remyelination.

Molecular determinants of OPC recruitment

OPC migration is modulated by defined molecular cues both during CNS development and after demyelination in adult CNS. Molecular determinants of developmental OPC migration have been extensively studied, and a detailed review of these findings is provided elsewhere.^76,77 Instead, here we will focus on the molecules for which the expression in MS lesions has been demonstrated and/or their potential influence on OPC recruitment investigated in experimental models of demyelination/remyelination (Table 2). Notably, some of these major determinants of OPC recruitment following demyelination are indeed the molecules that play a crucial role in axonal extension and OPC migration during embryonic development.

Table 2

Modulators of OPC migration/recruitment that are overexpressed/downregulated in MS

	Expression in MS lesions	Effect on OPC recruitment in models of demyelination
Sema 3A	Active (mRNA);78 chronic active (protein)22	Chemorepellent22,68
Sema 3F	Active22,78	Chemoattractant22,68
Netrin-1	Active and chronic19,49	Chemorepellent, enhances differentiation19
FGF2	Active and chronic active79	Not examined; inhibits differentiation80,81
Anosmin-1	Chronic lesion core79	Not examined
Sulfatase 2	Chronic active82	Decreases migration by promoting BMP/Wnt signalling82
EGF	Decreased in CSF and spinal cord in patients with MS84,87	Chemokinesis83,86
tPA	Acute lesions88	Chemokinesis85

Semaphorins 3A and F

Guidance molecules Semaphorins 3A and 3F are important regulators of OPC migration during both development⁸⁹ and myelin regeneration.^22,68,90In vitro studies showed that Semaphorin (Sema) 3A is a chemorepellent for both embryonic⁸⁹ and adult^68,89 OPCs, while Sema 3F exerts a chemoattractive effect on both.^68,89,90 Importantly, both Sema 3A and 3F mRNA were detected in active but not chronic MS lesions, and their respective receptors neuropilin (Nrp)1 and Nrp2 were detected on OPCs in MS brains.⁷⁸ A later study²² showed that Sema 3F protein but not Sema·3A was expressed in active lesions. Sema 3A expression was observed in chronic active lesions, in addition to Sema 3F, with Sema 3A predominating on lesion borders showing demyelinating activity.²² The role of Semas 3A and 3F was investigated by gain and loss of function experiments in lysophosphatidylcholine/lysolecithin (LPC)-induced demyelinating lesions in mice.^22,68 Both lentiviral-mediated overexpression⁶⁸ and infusions of recombinant²² Sema 3A and Sema 3F modified OPC recruitment to demyelinating lesions, so that less OPCs were recruited in response to Sema3A overexpression while OPC recruitment was enhanced in response to Sema 3F overexpression. Mice deficient for Nrp-1 (Sema 3A receptor) showed increased OPC recruitment after demyelination.^22,68 Even though lentiviral-mediated Sema 3A overexpression in the spinal cord decreased OPC recruitment, remyelination was not significantly affected.⁶⁸ Yet, a different study showed that infusion of recombinant Sema 3A decreased both OPC recruitment and remyelination.²² Interestingly, infusions of recombinant Sema 3A in demyelinated rat lesions after termination of OPC recruitment also decreased remyelination, indicating that this molecule has a negative effect on OPC differentiation as well.⁹¹ Notably, three studies showed that Sema 3F-mediated increase in OPC recruitment accelerated remyelination.^22,68,90 In MS lesions, the inverse relationship between Sema3A expression and OPC number was observed in individual lesions, particularly in chronic active lesions.²² While correlation between Sema 3F and OPC numbers was less clear, active plaques that contained more Sema 3F compared to others generally contained more OPCs.²²

Netrin-1

Netrin-1 is another guidance molecule that plays a role in OPC migration during both development^89,93,94 and myelin regeneration.¹⁹ Netrin-1 can act on a variety of receptors such as DCC (deleted in colorectal cancer), neogenin and Unc5 family (A, B, C). Its action on the target cell (chemoattraction or chemorepulsion) is dependent on the receptors that netrin-1 binds to on the cell, so that activation of DCC is associated predominantly with chemoattraction, while activation of Unc5 family alone or in combination with DCC leads to chemorepulsion.^{19,89,92–96} Importantly, netrin-1 and its fragments are expressed in MS lesions,^19,49 both in actively demyelinating and the chronic ones. In contrast, in the lysolecithin mouse model of demyelination that shows efficient remyelination, netrin-1 expression was absent in early phases, and was upregulated only after OPC recruitment, once OPCs started differentiating.¹⁹ To elucidate whether netrin-1 plays a role in myelin repair, this molecule was either overexpressed or its function was blocked in experimental spinal cord demyelinating lesions.¹⁹ These experiments demonstrated that netrin-1 plays a dual role in remyelination, depending on the timing of its expression. Thus, its overexpression prior to OPC recruitment using a lentiviral vector reduced OPC arrival to the lesion and decreased remyelination. However, if netrin-1 function was blocked (using function-blocking antibodies) after the recruitment phase at the onset of OPC differentiation, OPC recruitment was prolonged in time, but these cells failed to differentiate, suggesting that netrin-1 plays an important role in the transition from recruitment to differentiation phase.¹⁹ In agreement with these results, netrin-1 expression in active MS lesions was associated with OPC differentiation, but in chronic lesions with netrin-1 expression OPC density was significantly reduced.¹⁹ These observations suggested that early netrin-1 expression in MS lesions initially contributes to OPC differentiation but reduces OPC recruitment, which with repeated episodes of demyelination, contributes to ultimate OPC depletion and remyelination failure.

FGF2 and anosmin-1

The expression of fibroblast growth factor (FGF) 2 and anosmin-1 has also been investigated in MS tissue.⁷⁹ The expression of FGF2 was observed in active lesions and at the active border of chronic active lesions, suggesting that it acts as a chemoattractant for FGFR1-expressing OPCs. On the other hand, anosmin-1, an extracellular matrix-associated molecule that induced adhesion of adult mouse OPCs, was expressed in the chronic lesion core, thus suggesting it could be acting to reduce OPC migration.⁷⁹ Yet, no studies have investigated the effect of FGF2/anosmin-1 on OPC recruitment to experimental demyelinating lesions. This is particularly difficult in the case of anosmin-1 because the ortholog of ANOS-1 gene has not been described in mice and rats.⁹⁹ Regarding FGF-2, while a study using mouse hepatitis virus- and cuprizone-induced demyelination models in FGF-2 knock-out (KO) mice showed that FGF2 may actually be acting to inhibit CNS remyelination, the effect observed was attributed to the inhibition of OPC differentiation/oligodendrocyte maturation by FGF2.⁸¹ The potential inhibitory effect on differentiation/maturation rather than recruitment is supported by the fact that functional recovery after cuprizone demyelination was accelerated in inducible oligodendrocyte-specific FGFR1 KO mice.⁸⁰

Sulfatase 2

Another recently identified inhibitor of OPC migration is sulfatase 2 (Sulf2), an extracellular endosulfatase that modulates the signalling microenvironment by editing the pattern of sulfation on heparan sulfate proteoglycans. The mRNA expression of Sulf2 was identified in activated human primary OPCs and murine OPCs recruited to lysolecithin-induced spinal cord lesions.⁸² Importantly, the expression of this gene was detected in OPCs and oligodendrocytes around demyelinated lesion border in the tissue from patients with MS.⁸² In mice, conditional ablation of Sulf2 in OPCs increased OPC recruitment, and the resulting increase in OPC density was associated with accelerated differentiation in oligodendrocytes and enhanced remyelination.⁸²

Stimulators of OPC chemokinesis

Rather than acting as chemoattractants, certain molecules increase OPC recruitment via chemokinetic effect (non-directional stimulation of migration). These include epidermal growth factor (EGF) and tissue type plasminogen activator (tPA). EGF overexpression in the subventricular zone (SVZ) enhanced OPC repopulation of proximal demyelinating lesions by SVZ-derived cells.⁸³ Moreover, EGF overexpression in cortical NG2 cells conferred these with migratory properties.⁸⁶ Interestingly, levels of EGF are decreased in the CSF and spinal cord tissue of patients with MS as compared with the controls.^84,87

Another factor that stimulates OPC chemokinesis is tPA. During the development, tPA expressed on blood vessels stimulated OPC emigration from the germinal pMN zone. Importantly, OPC recruitment to lysolecithin-induced lesions was delayed in tPA KO mice and could be enhanced by tPA infusions. The effect of tpA on OPCs was dependent on the protease-independent activation of EGF signalling in OPCs.⁸⁵ These observations may be relevant for MS as tPA is expressed in acute MS lesions.⁸⁸

Interestingly, transcription factor Sox2 was identified as an internal factor that plays an important role in OPC recruitment. When Sox2 was conditionally deleted in OPCs, LPC-induced lesions failed to remyelinate, which was primarily due to defective OPC recruitment.¹⁰⁰ While it is unlikely that Sox2 expression in OPCs could be modified in MS to increase recruitment, elucidating the mechanisms activated by this transcription factor in OPC may reveal factors that render OPC receptive to recruitment-promoting signals.

Thus, OPC recruitment following remyelination can be modulated by overexpressing/inhibiting defined molecular cues within the areas of demyelination, which has an effect on remyelination (Table 2).

Identity of the remyelinating cell in MS-OPC versus surviving oligodendrocyte

OPCs have been considered as the source of new myelin in MS lesions for a long time. Early research established that while transplantation of OPCs into demyelinating lesions resulted into successful myelin repair,¹⁰¹ engrafted post-mitotic oligodendrocytes were incapable of regenerating the myelin.⁹⁷ These studies were later confirmed in the context of endogenous remyelination as genetic tracing of OPCs⁹⁸ and mature oligodendrocytes¹⁰² showed that OPCs and not oligodendrocytes were the cellular source of regenerated myelin in lysolecithin-induced lesions. Of note, remyelination of periventricular LPC lesions mobilizes not only parenchymal OPCs but also SVZ-derived OPCs, the generation of which is stimulated by nearby demyelination.^103–108 However, the notion that the sole effectors of successful remyelination are OPCs has been challenged more recently by studies showing that CNS demyelination in the cats fed with irradiated diet and vitamin B12-deficient monkeys occurred because oligodendrocytes stopped myelinating, even though they did not die.^109,110 Importantly, remyelination in these models was carried out by mature oligodendrocytes.¹¹⁰ The differences between the studies describing mature oligodendrocyte-mediated remyelination and those using the lysolecithin model showing strictly OPC-mediated remyelination are likely due to different causes of demyelination between these models. Lysolecithin is a toxin that targets primarily myelin sheaths, but also cellular membranes, thus causing cell death in the lesion area. This means that intralesional oligodendrocytes do not survive, and because mature oligodendrocytes do not migrate, those that are spared in lesion surroundings do not contribute to remyelination.¹⁰² For this reason, more recent studies have focused on the capability of mature oligodendrocytes to remyelinate the lesions using models other than lysolecithin-induced lesions. Thus, detailed analyses of mice fed with cuprizone demonstrated that both OPCs and mature oligodendrocytes contribute to remyelination.¹¹¹ In addition, the efficiency of OPC- versus oligodendrocyte-mediated remyelination was compared in a study using a transgenic zebrafish with myelinating glia-specific expression of the rat capsaicin-inducible cation permeable transient receptor potential V1 channel (TRPV1).¹¹² Demyelination was induced by subjecting the zebrafish embryos to 10 μM capsaicin for 2 h, thus stimulating an important calcium influx into myelin sheaths. This study found that, although surviving oligodendrocytes did contribute to remyelination, they made fewer new myelin sheaths than OPCs, thus concluding that remyelination by spared oligodendrocytes is not very efficient. In addition, this study found that surviving oligodendrocytes frequently produced mistargeted new myelin around cell bodies, also detectable around the edges of MS lesions. It is important to note that the demyelination paradigm used in this study (extensive Ca⁺² overload) was quite harsh, so it is possible that, even though some oligodendrocytes survived, these were too damaged to significantly contribute to remyelination. In situations in which efficient surviving oligodendrocyte-mediated remyelination was observed, demyelination took much longer time to develop, which was 3–4 months after the ingestion of irradiated diet in cats or after 3 years of vitamin B12 deprivation in monkeys.¹¹⁰ Thus, it appears that in these cases the trigger of demyelination was a much more subtle process that was prolonged in time, possibly reflecting an arrest in myelin synthesis due to reversible diet-induced metabolic alterations, which may explain extensive remyelination by surviving oligodendrocytes described in these studies.

The observations of remyelination mediated by spared mature oligodendrocytes in specific animal models appear supported, at least in part, by several studies on MS tissue. As discussed above, in a subset of demyelinated MS lesions, mature oligodendrocytes are preserved.^7,11–13 A single cell RNA sequencing study of oligodendroglial cell subpopulations in MS tissue highlighted that mature oligodendrocytes in MS appear altered in terms of gene expression. Thus, while stable mature oligodendrocytes from control subjects show low levels of transcriptional machinery involved in myelination, these cells upregulate the expression of myelination-related genes in MS.²⁴ This observation suggests that mature oligodendrocytes stripped of their myelin sheaths/that stopped myelinating become engaged in remyelination. Yet, it cannot be excluded that these mature oligodendrocytes are actually the newly-generated ones that have higher expression of myelin gene transcripts.

Generation of new oligodendrocytes/persistence of pre-existing ones in MS was investigated by evaluating the incorporation of C¹⁴ (released in the atmosphere during the nuclear bomb tests in the 1950s) into the genomic DNA of oligodendrocytes isolated by FACS from NAWM, shadow plaques and PPWM from MS patients, as well as white matter from control subjects.¹¹³ The presence of C¹⁴ in the genomic DNA of mature oligodendrocytes indicates that these were generated after 1950s, which is when the subjects analysed developed the disease. In subjects with very aggressive disease, who therefore died at young age, higher incorporation of C¹⁴ in the NAWM oligodendrocytes as compared to the control subjects was found, thus confirming previous observations that oligodendrogenesis occurs mostly in younger patients. Interestingly, oligodendrocytes isolated from the areas identified as shadow plaques did not show higher labelling for C¹⁴, which suggested that pre-existing oligodendrocytes and not newly-generated ones contributed to remyelination. It was concluded that, if new oligodendrocytes were generated in shadow plaques, they did not survive. The authors did acknowledge, however, that the areas identified as shadow plaques could have been areas of partial myelin destruction rather than remyelinated areas. In addition, it could also be that OPCs in these areas differentiated directly, without proliferating, as shown in rodents,⁴⁷ which could explain the lack of significant C¹⁴ incorporation. In line with this possibility, the authors indeed confirmed previous observations that OPC proliferation in the lesions is basically absent, although detectable in the NAWM.

In vitro studies of human cells also support the possibility that, under specific circumstances, oligodendrocytes can temporarily stop myelinating and then engage in remyelination. Thus, the main response of adult human oligodendrocytes to nutrient deprivation was process retraction and low metabolic rate, which compromised myelination, but preserved survival.⁴² Interestingly, this pathology appears reversible until a specific point of no return is reached,²⁷ similarly to what was described in cats fed with irradiated diet,^109,114 suggesting that oligodendrocytes that stop myelinating in response to metabolic stress may recover their myelinating function if optimal conditions are restored.

Therefore, modelling remyelination mediated by spared oligodendrocytes is a major challenge, and will be crucial for elucidating novel therapeutic approaches to reduce demyelination/promote myelin repair in MS.

Could promoting OPC recruitment to demyelinating lesions with spared oligodendrocytes improve remyelination?

Presence of spared non-myelinating oligodendrocytes in a subset of MS lesions raises important questions on strategies to increase remyelination in these areas. The obvious strategy would be to identify factors leading to the arrest of myelination by mature cells, and develop strategies to stimulate these to re-engage in myelination. Yet, it is not clear whether some of the spared oligodendrocytes are actually beyond the ‘point of no return’ described in the context of dying-back oligodendropathy,¹¹⁵ which means dysfunctional in terms of remyelination. As mentioned above, a zebrafish study in which demyelination was induced by Ca²⁺ overload¹¹² suggested that OPCs make new myelin more efficiently than pre-existing oligodendrocytes. Similar results were obtained by a recent study comparing the myelination potential of human OPCs versus mature oligodendrocytes, both from paediatric and adult surgical resections.¹¹⁶ While paediatric cells performed generally better than adult cells in a nanofiber ensheathment assay, within each group more fibers were sheathed by OPCs than by oligodendrocytes. Thus, while the development of novel models of demyelination characterized by oligodendroglial preservation will be crucial to address remyelination strategies based on stimulation of spared oligodendrocytes, it may be that a more efficient approach would be to stimulate OPC-mediated remyelination in these areas.

It is important to highlight that in MS lesions with spared oligodendrocytes, presence of these will likely prevent remyelination by OPCs. One of the potential mechanisms is the expression of netrin-1 by mature cells that acts as chemorepellent for adult OPCs,^19,49,117 thus preventing the OPC recruitment towards demyelinated axons. It is possible that one way of stimulating remyelination in these lesions would be to overexpress OPC attractants, as it has been shown that the balance of chemoattractive versus chemorepellent cues determines the efficiency of OPC recruitment.²²

Gene therapy for OPC recruitment

As mentioned above, enhancing OPC recruitment to demyelinating lesions accelerates remyelination. While molecular candidates for increasing OPC recruitment have been identified, the translational paradigm for targeting the overexpression of these to the areas of demyelination is not clear. In experimental settings, such overexpression has been achieved by direct injections of lentiviral vectors encoding for the molecules of interest^19,68 or direct infusions of recombinant molecules.²² Yet, this paradigm is not applicable to MS, a disease characterized by disseminated lesions, many of which are not accessible by injection. We recently addressed the issue of molecular targeting to MS lesions to increase OPC recruitment. Because transplantation of genetically-engineered haematopoietic stem cells (HSCs) has been used to deliver a missing protein to the CNS via myeloid cells in both animal models of and patients with leukodystrophies,^118–123 we investigated whether Sema 3F, an OPC attractant, could be delivered to demyelinating spinal cord lesions using the same strategy.⁹⁰ We first demonstrated that transplanted genetically modified HSCs give rise to approximately 50% of macrophages in demyelinating spinal cord lesions. Notably, lesions in mice with blood cells overexpressing Sema 3F showed enhanced OPC recruitment and accelerated remyelination of demyelinated axons.⁹⁰ Thus, using myeloid cells as vehicles could be a clinically relevant paradigm for overexpressing OPC recruitment-enhancing molecules in the lesions undergoing demyelination/myelin clearance. In this manner, recruitment would be stimulated as early as possible to ensure repopulation of the oligodendroglial pool in the lesion. In clinic, this could be achieved by combining gene therapy with aHSCT, already performed in a subset of patients with MS that fulfill specific criteria and fail to respond to conventional therapies.¹²⁴ However, because of the risks associated with aHSCT,¹²⁴ a safer approach to target OPC attractants to the lesions may be to perform systemic injections of autologous transgene-carrying monocytes/macrophages, thus avoiding the preconditioning protocols in aHSCT associated with significant health risks. Future work should address the feasibility of such approach in stimulating OPC recruitment and CNS remyelination.

Concluding remarks

In conclusion, it appears clear that chronic demyelination in multiple sclerosis is consistently associated with a decreased density or sometimes even complete depletion of oligodendroglial cells. OPC differentiation appears enhanced at higher densities and in the presence of acute inflammatory stimuli. Thus, stimulating quick OPC repopulation of the lesions could be a valid strategy not only to prevent OPC depletion, but also to enhance their differentiation. OPC repopulation of MS lesions could be achieved using two strategies: (i) OPC transplantation; and (ii) stimulation of endogenous OPC recruitment. Regarding transplantation, strategies to generate human OPCs from induced pluripotent stem cells (iPSCs)^125–127 or fibroblasts¹²⁸ have been described, and the (re)myelinating potential of these cells has been validated in mouse models.^125,126,128 Importantly, it is possible to identify and select human cells committed to oligodendrocyte lineage,¹²⁹ which if performed prior to transplantation would ensure oligodendroglial differentiation of grafted cell preparations. Thus, given that haplotype matching and human OPC availability, previously described as major limiting factors for transplantation therapy^31,130 are no longer an issue, this strategic approach is clearly of interest, particularly for large, transplantation-accessible lesions. However, as acute inflammation appears required for remyelination even by transplanted OPCs,⁵⁸ this approach may need to be performed as early as possible after demyelination.

While the therapeutic potential of transplantation in MS has been reviewed in detail elsewhere,^131–133 here we have focused on a different and potentially complementary approach, which is to stimulate OPC recruitment as means of preserving the lesional OPC pool to increase remyelination. Overexpression of OPC attractants within the areas of demyelination in rodents can be achieved using genetically engineered blood myeloid cells as vehicles, which suggests novel therapeutic avenues for enhancing myelin regeneration in MS.

Acknowledgements

We would like to acknowledge all the authors cited in this article for their contributions to the field, as well as the patients and their families whose consent on the use of human tissue was crucial for the analyses of oligodendroglial changes in the disease. Regarding the financial support for our work: in Spain, we acknowledge the Ministry of Science and Innovation (Young Investigator Grant to V.T.) and Basque Government (support to Neurobiology lab in Achucarro Basque Center for Neuroscience); in France, we thank the Fondation d’Aide pour la Recherche sur la Sclerose en Plaques (ARSEP), Fondation pour la Recherche Médicale, INSERM-DHOS, Assistance Publique des Hôpitaux de Paris (APHP), the programme ‘Investissements d'Avenir’ ANR-10-IAIHU-06, NeurATRIS and the Agence Nationale de la Recherche. We also thank Bernard Zalc for his critical reading of the manuscript.

Competing interests

C.L. participates on advisory boards for Roche, Biogen, Merck-Serono, Genzyme, Vertex, and Rewind and has scientific collaborations with Vertex and Merck-Serono. V.T. reports no competing interests.