Oligodendroglial cells are the myelinating cells of the central nervous system. From neural stem cells to mature oligodendrocytes, their specification and then differentiation are regulated by the dynamic expression of transcription factors, which control the expression of lineage-specific genes (e.g., Ascl1, Olig1, Sox10) or myelinating genes (e.g., Yy1, Myrf). While these transcriptional events are tightly orchestrated, environmental cues are also critical in this process. The integration of external cues, such as neuronal activity, into intrinsic signals is mediated by epigenetic modifications, which are known to control chromatin organization and, in turn, regulate gene expression. In particular, chromatin condensation and accessibility are regulated by DNA methylation, histone modifications, and chromatin remodelers, which interact with long non-coding RNA (lncRNA) and microRNA (miRNA), as well as nuclear organization via lamins. Recently, the methylation of mRNA has also been described as an epigenetic modification, resulting in gene expression regulation at the translational level.
Neural stem cells (NSCs) in the developing and adult mammalian brain harbor the ability to self-renew and to generate neurons, astrocytes, and oligodendrocytes (OL) [1]. The differentiation of NSCs occurs in response to extracellular signals, along with the interplay between dynamic epigenetic modifications and lineage gene expression regulation [2][3]. Lineage specification requires both the activation of lineage genes and the repression of alternative lineage genes. In this section, we will focus on how the transition from NSCs to OLs is regulated by successive waves of DNA methylation and hydroxymethylation, as well as by histone modifications, chromatin remodelers, microRNAs, and lncRNA.[1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80][81][82][83][84][85][86][87][88][89][90][91][92][93][94][95][96][97][98][99][100][101][102][103][104]
The specification of NSCs from embryonic stem cells (ESCs) toward neurogenesis and gliogenesis occurs through specific methylation patterns. A study from Sanosaka et al. suggests that the demethylation of neuron-specific genes first occurs as ESCs transition to NSCs [4], while methylation on glial promoters, such as glial fibrillary acidic protein (GFAP), is maintained to suppress its expression [5]. Gliogenic transition is then favored and occurs via the demethylation of glial gene promoters and genes involved in gliogenic pathways, as well as the de novo methylation of neuronal genes [4][5][6][7][8][9][10]. Subsequently, the de novo methylation of astrocytic genes occurs during the specification of oligodendrocytes [6][11], along with the hydroxymethylation of oligodendrocyte genes, such as Olig1, Sox10, and Id2/4 [12][13].
Lineage determination is also regulated by histone modifications, such as H3K27me3, which is catalyzed by the polycomb repressive complex 2 (PRC2) and is constituted of the enhancer of the zeste homolog 2 (EZH2) and embryonic ectoderm development (EED). EZH2 and EED are highly expressed in proliferating NSCs and target genes related to developmental processes and neurogenesis [14][15]. As NSCs differentiate into OPCs, EZH2 and EED expression remain high, while their levels decrease in cells transitioning to neuronal and astrocyte lineages [15][16]. Moreover, the ChIP-sequencing of H3K27me3 in postnatal-day-1 rat OPCs highlights its genomic regulation of genes involved in the “global alternative lineage choice” [17]. Indeed, conditional ablation of Eed favors the differentiation into astroglial cells, to the detriment of the oligodendroglial lineage [15]. OPC lineage progression also depends on arginine methylation, as the conditional knockdown of Prmt1 in NSCs drastically reduces the number of OLs in mice [18].
The normal differentiation of cortical precursors in vitro and in vivo in mice is accompanied by sequential histone acetylation and the subsequent activation of promoters of neuronal and astrocyte genes, then postnatally in oligodendrocyte genes, such as Mbp and Plp. The key histone acetyltransferase in this process is the CREB binding protein (CBP), which regulates H3K9/K14 acetylation [19]. Indeed, blocking histone deacetylation in rats reduces oligodendrogenesis and favors the differentiation of cells along alternative lineage choices [20]. However, histone deacetylation has also been shown to be involved in the regulation of neural progenitors [21]. In particular, HDAC2 and HDAC3 associate with key genes regulating the differentiation of NSCs [22][23]. In NSCs, HDAC3 acts first as a repressor of neuronal differentiation in cortical NSCs [22], then antagonizes astrogliogenesis by inhibiting the acetylation levels of Stat3, which is a core effector in the Janus kinase (JAK)–STAT pathway [24]. Astrocytic genes are also repressed by HDAC1 and HDAC2 in vitro during Sonic Hedgehog (Shh)-induced oligodendrogenesis [25]. Finally, the cooperation of HDAC3 with the histone acetyltransferase p300 targets and activates enhancers of the OPC genes, such as Olig2 [24]. Inversely, the differentiation to OPC is prevented by HDAC2 and sirtuin 1 (SIRT1), which, at least in part, can inhibit specific OL differentiation genes, such as the key transcription factor Sox10 [22][26].
Chromatin structure and organization in NSCs changes during their specification throughout brain development to generate the major cell types of the CNS. These dynamic chromatin states are modulated by chromatin architectural proteins, such as the HMG proteins, and chromatin modifiers, such as BRG1. In the early stages of NSCs characterized by a neurogenic potential, HMGA1 and 2 proteins are highly expressed and mediate the global chromatin opening. As the levels of HMGA proteins decrease, chromatin becomes more condensed in a stage-dependent manner that allows for an astrogenic transition [27]. All four mammalian forms of HMGB (HMGB1, 2, 3, and 4) are expressed in proliferating NSCs [28] with specific roles described for HMGB2 and HMGB4 in the neurogenic-to-gliogenic fate transition [29][30]. Members of the HMGN family (HMGN1, 2 and 3) also positively regulate the neuron–glia fate switch [31].
BRG1, within the SWI/SNF-related chromatin remodeling complex, is a critical regulator of NSC specification by repressing neuronal differentiation, while favoring gliogenesis and differentiation in mammalian neural development. In mice, the ablation of Brg1 specifically in NSCs does not impact the initial neuronal differentiation but abolishes glial generation and differentiation, as seen by a dramatic decrease in astrocyte, oligodendrocyte progenitor, and myelin protein markers in late embryonic stages [32]. However, the function of BRG1 appears different in lower vertebrates, as the inhibition of BRG1 blocks neuronal differentiation in Xenopus [33] and has restricted effects in retinal ganglion cells in zebrafish [34].
Several miRNAs play key roles in neural lineage development. For instance, miR-124 is necessary for fate specification into neurons by reducing the expression of EZH2 [35]. Other miRNAs are necessary for the neurogenic-to-astrogenic transition, such as miR-153, which regulates the acquisition of gliogenic competence [36], or miR-17/106 [37]. To our knowledge, no miRNA has been described as being specifically required for the differentiation from an NSC to an OL.
Dong and collaborators have investigated whether lncRNAs are involved in the regulation of NSC differentiation into OPCs. They identified lnc-OPC, a specific and highly expressed lncRNA in OPCs, which is critical for cell fate determination. In vitro loss- and gain-of-functions experiments that targeted lnc-OPC, as well as Sox8OT, Neat1 and lnc-158, have highlighted the positive prominent role of lncRNAs for oligodendroglial specification [38][39][40][41].
In the CNS, oligodendroglial cells are highly sensitive to their environment, which influences their survival and proliferation rate as immature OPCs and/or their differentiation capacities into myelinating OLs. Overall, this tight regulation maintains a relative homeostatic pool of OPCs throughout time and space in both the developmental and adult CNS [42]. In particular, OPCs proliferate in response to chemical cues, such as mitogens, growth factors, and cytokines [43][44][45][46][47][48], and to neuronal cues, such as electrical activity [49][50][51][52][53]. OPCs are now thought to translate these specific extracellular cues into intrinsic signals that affect survival and proliferation via the regulation of epigenetic modifications.
Increasingly, evidence has revealed the heterogeneity of the OPC properties between the developmental and adult OPC pools, but also according to their CNS localization. OPC survival has been shown to be variably associated with DNA and histone modifications and chromatin remodeling, depending on time and space.
During development, the ablation of Dnmt1 or Dnmt3a in neonatal OPCs tends to induce DNA damage and decreases OPC survival in vitro [11][54]. However, these effects were not observed after the ablation of Dnmt1 and/or Dnmt3a in adult OPCs during remyelination experiments [55]. OPC survival during development has also been associated with the PRMT5-dependent H4R3me2s mark, as the ablation of PRMT5 in OPCs could activate p53 pathways and increase apoptosis [56]. In addition, neonatal non-proliferative OPCs are protected from apoptosis by the chromatin remodelers EP400 and CHD7, the latest known remodelers to control chromatin closing and p53 transcriptional repression [57][58]. In adult OPCs, cell survival is partially regulated by the CHD8, which shares many common binding sites with CHD7 [58]. Indeed, the global or oligodendroglial-specific ablation of Chd8 in OPCs results in increased apoptosis, in particular, in adult spinal cord tissues, but not during development or in the brain [59].
The nuclei of proliferative OPCs are mostly euchromatic and characterized by a relaxed and transcriptionally competent chromatin structure that is enriched for permissive marks, such as histone acetylation (e.g., H3K9ac and H3K14ac) [26][60][61]. The OPC cell cycle is also regulated by the oncogene transcription factor cMyc in response to platelet-derived growth factor (PDGF) and fibroblast growth factor (FGF) mitogens [60][62]. cMyc recruits histone acetyltransferases and the ablation of cMyc in vitro can directly decrease H3K9ac and H3K14ac in OPCs [60][63][65]. This suggests that, on top of acting as a DNA binding transcription factor, cMyc is capable of translating mitogenic extracellular stimuli into epigenetic signals in proliferative OPCs.
Evidence suggests that the epigenetic regulation of OPC proliferation might differ depending on the CNS region. For example, differences in chromatin remodeling have been noted in brain and spinal cord tissues. The ablation of Chd7 or Chd8 in OPCs does not lead to an increased number of OPCs, since, despite an increase in cell cycle and proliferative genes, this also leads to an increase in apoptosis, at least in brain tissues [58][59][66]. Surprisingly, following Chd7 or Chd8 ablation, OPC proliferation is not increased but reduced in spinal cord tissues, where apoptosis is not perturbed [59][67].
OPC proliferation and cell cycle exits appear to be regulated by several epigenetic marks, including repressive ones, such as DNA and histone methylation. For example, the ablation of Dnmt1 in neonatal OPCs revealed reduced DNA methylation and defective gene repression, in particular, at cell cycle genes [11]. This result was not replicated in vivo in adult OPCs, suggesting once again that there is an age-dependent epigenetic regulation of transcriptome in OPCs [55]. However, the presence of DNA modification alone is not sufficient to induce precocious proliferation. Ablating DNMT1- or DNMT3A-mediated DNA methylation or EED-mediated histone methylation in OPCs does not result in ectopic proliferation in vitro or in vivo, and even slightly negatively perturbates OPC proliferation during development [11][15][55]. Inversely, TET1-mediated DNA hydroxymethylation does not seem to affect neonatal or adult OPC proliferation in vitro or in vivo [12][68].
One of the main and most studied functions of neonatal and adult OPCs is their ability to differentiate into mature OLs. OPC differentiation is in part an intrinsic propensity but is also regulated by environmental and neuronal cues that can influence epigenetic modifications. Recent studies highlighted the inter-neuronal (e.g., myelinated or non-myelinated axons) and intra-neuronal (e.g., variable size of internodes) heterogeneity of myelination in the CNS, which appeared to be essential for global brain connectivity and function, and which could reflect the heterogeneity of the OPC population and/or their environment [69][74].
Early evidence showed that OPC differentiation is associated with demethylation of a specific myelin gene, Mag, during rat development [75]. A more global demethylation during OPC differentiation that is associated with permissive gene expression was further confirmed by DNA methylation whole-genome sequencing. During development, OPC differentiation into OL is correlated with the decreased DNA methylation and increased DNA hydroxymethylation of genes involved in lipid synthesis and myelin formation [11][12]. However, because of its role in OPC proliferation and survival, the ablation of DNA methyltransferases in neonatal OPCs does not induce early differentiation, but in contrast, results in global hypomyelination of the CNS [11][54]. Indeed, the ablation of DNA methyltransferases in developmental and adult post-mitotic OPCs has a slight effect on myelination and remyelination [11][55]. Recent studies also highlight an age-dependent role of demethylation on OPC differentiation, suggesting a more important role of DNA hydroxymethylation in adult oligodendroglial cells [12][68]. In particular, TET1-mediated DNA hydroxymethylation targets genes involved in the late stages of OPC differentiation, such as biosynthesis and neuroglial communication [12][13][68]. The ablation of Tet1 in adult OPCs is sufficient to reduce DNA hydroxymethylation at these specific genomic sites, downregulating their expression and blocking OL late differentiation [68]. In old mice, lower TET1 expression and decreased DNA hydroxymethylation levels could be directly associated with the delayed remyelination observed in aging [68]. While the downregulation of TET enzymes in vitro has been shown to affect neonatal OPC differentiation, this effect is still being examined in developmental studies in vivo [12][68][76]. Overall, these studies suggest a dual role for DNA modifications in oligodendroglial cells, balancing both the methylation and demethylation of specific genomic regions at different ages and different stages of the differentiation process.
At the chromatin level, OPC differentiation is mainly associated with open conformation at discrete loci, which is reflected by chromatin remodeling, histone modifications, and nuclear lamin reorganization, allowing for access of transcription factors to specific genes that are characteristic of the differentiated state. For example, the chromatin remodelers BRG1 and EP400 have been shown to be essential for OPC differentiation and myelination, at least during development when targeting early lineage [57][77][78]. However, they seem to be dispensable for later differentiation stages and during myelin maintenance [57][77]. Both BRG1 and EP400, in association with OLIG2, directly bind to differentiation genes, such as Myrf and Sox10, especially at enhancers and transcription start sites, which are characterized by the permissive histone marks H3K27ac and H3K4me3, respectively [57][59][68][77][78]. They also share similar chromatin occupancies with CHD7 and CHD8, which are essential for OPC differentiation, during developmental and adult (re)myelination [58][59][66]. In particular, the ablation of Chd8 in neonatal OPCs results in massive hypomyelination of the CNS, leading to seizures and eventually death of the mice at postnatal day 21. Interestingly, CHD8 can itself directly recruit KMT2/MLL, a histone lysine methyltransferase that is responsible for the addition of H3K4me3 and the subsequent activation of oligodendroglial genes (i.e., Olig1/Olig2, Sox10, Myrf) [59].
OPC differentiation is also dependent on marks that are generally associated with gene repression, such as histone methylation, deacetylation, and citrullination. Indeed, the ablation or inhibition of Prmt5, Ezh2, Eed, Hdac1/2, or Padi2 in OPCs results in defective differentiation and myelination [15][17][23][54][56][61][79][80][81][82][83][. These marks mainly regulate the downregulation of OPC-specific inhibitors of differentiation (such as Id2/Id4) or cell cycle (such as Cdk4/6, Cxcl2/5/10/14) genes, and therefore, are often essential for OPC cell cycle exits and early OL differentiation, but less involved in myelin maintenance [15][17][80][81]. These enzymes, such as HDACs, PRMT5, and PADI2, can also modify non-histone targets, altering the function or localization of oligodendroglial proteins or transcription factors (i.e., OLIG1, alpha-tubulin, PDGFRa, or myelin proteins) [80,85–88]. Because of steric constraints, histone modifications can also be dependent on or exclusive of each other. For example, the ablation of Prmt5 in OPC results in decreased symmetric H4R3me2s, allowing for H4K5 acetylation and preventing differentiation and myelination. The addition of histone acetylation inhibitors in vitro is sufficient to rescue OPC differentiation, even without PRMT5 [56].
Eventually, repressive marks are also associated with the nuclear lamina, which maintains mainly repressive heterochromatin at the nuclear periphery. Recently, LMNB1 has been associated with oligodendroglial maturation genes (i.e., myelin genes and cholesterol synthesis pathways), which appeared to be sufficient to block OPC differentiation in vitro [89][90].
In addition to chromatin condensation and conformation, the OPC differentiation program also depends on post-transcriptional modifications. Recent models using the ablation of Dicer (required for the generation of functional microRNA) or the de novo analysis of oligodendroglial transcriptomic datasets identified microRNAs and lncRNAs that are essential for OPC differentiation. For example, miR-219, miR-338, miR-23, and miR-32, as well as lnc-OL1, lnc-158, and Neat1, promote differentiation, while miR-27a, miR-212, and miR-125-3p inhibit differentiation [39][40][89][91][92][93][94][95][96][97][99][100][98][101]. Interestingly, these post-transcriptional signals directly regulate some chromatin-modifying genes, suggesting feedback from miRNA and lncRNA regarding chromatin conformation. For example, miR-23 has been shown to suppress LMNB1 expression, thus rescuing OPC differentiation in vitro [89]. Similarly, lnc-OL1 can directly interact with SUZ12, a part of the histone methylation complex PRC2, with both being required for OPC differentiation during development and repair [92][102].
Recently, studies have identified the essential role of RNA methylation on OPC differentiation. The ablation of the methyltransferase Mettl14 (“writer”), the demethylase Fto (“eraser”), or the m6A “reader” Prrc2a in the oligodendroglial cell lineage all resulted in a lower number of mature OLs and global hypomyelination of the CNS [103][104]. In addition to the downregulation of several myelin genes (i.e., Mbp, Mog, Mag), many histone post-translational modification readers and writers (i.e., HMTs, HDACs) are also dysregulated when METTL14 is lacking, suggesting here again the potential feedback from m6A-modified mRNA to histone modifications [104].
Finally, in addition to timely OPC differentiation into mature OLs, recent studies have highlighted the importance of correct myelination for neuronal connectivity. This includes myelin ensheathment, compaction around the axon, and internode sizes. This is not limited to developmental myelination as myelin is constantly remodeled in adult CNS, especially during learning and when facing depressive or social isolation experiences [105–112].
A few epigenetic marks have been associated with late myelination processes. Compared to other chromatin remodelers, CHD7 binds preferentially to myelinogenesis (i.e., Mbp, Plp1, Cnp) and lipid metabolism genes (i.e., Enpp2, Nfya, Elovl7), which would suggest involvement in late OPC differentiation and myelination [66]. The genetic ablation of Tet1 in OPC induces defective remyelination, which is characterized by swellings in adult CNS after injury. TET1-mediated hydroxymethylation tends to target genes related to late myelination, as well as neuroglia communication genes involved in ion exchange and the maintenance of a tight space between the axon body and the myelin membrane (i.e., Slc12a2) [12][68]. METTL14 and m6A RNA affects the alternative splicing of some paranodal genes, such as glial neurofascin 155. In Mettl14 mutants, Xu et al. noticed increased nodal/paranodal spaces, which were associated with lower numbers of nodes, suggesting a role for mRNA methylation on internode length [104].
Because the myelination and myelin remodeling processes have not been analyzed in detail yet, additional epigenetic modifications are to be identified in this later process to finely tune the ensheathment, internodal length, and myelin compaction.
This entry is adapted from the peer-reviewed paper 10.3390/life11010062