Epigenetic regulation of oligodendrocyte identity

Abstract

The interplay of transcription factors and epigenetic modifiers, including histone modifications, DNA methylation and microRNAs during development is essential for the acquisition of specific cell fates. Here we review the epigenetic “programming” of stem cells into oligodendrocytes, by analyzing three sequential stages of lineage progression. The first transition from pluripotent stem cell to neural precursor is characterized by repression of pluripotency genes and restriction of the lineage potential to the neural fate. The second transition from multipotential precursor to oligodendrocyte progenitor is associated with the progressive loss of plasticity and the repression of neuronal and astrocytic genes. The last step of differentiation of oligodendrocyte progenitors into myelin-forming cells is defined by a model of de-repression of myelin genes.

Epigenetic Modulation of Gene Expression Defines Cell Identity

One of the most exciting findings of the past few years has been the discovery that developmental processes are regulated by the crosstalk between transcription factors and epigenetic modulators of gene expression, including post-translational modifications of nucleosomal histones, changes in histone variants, chromatin remodeling enzymes, DNA methylation and microRNAs ^1–6. Collectively, these other factors have been defined as “epigenetic regulators” (epi = above), because they control gene expression by regulating transcript levels, independently of changes in DNA sequence. The global effects of chromatin modifications on gene expression and the transmissibility of the information during cell division further delineate the importance of epigenetics in building a “memory” or “signature” of cell identity ⁷. It has been proposed that the identity of each cell is dependent on “heritable memory” that is traced by epigenetic regulators and is associated with the activation of a lineage-specific program of gene expression ⁸. The acquisition of this “memory” during development has been defined as “epigenetic programming” ⁸. In this review we discuss these concepts within the framework of the molecular events characterizing the transition from pluripotent stem cells to myelin-forming oligodendrocytes.

Epigenetic programming and reprogramming

The concept of differentiation was originally envisioned as unidirectional progression through sequential states, each characterized by precise “epigenetic marks”. The discovery of enzymatic activities with the ability to “erase” these epigenetic marks ^{9, 10}, however, revolutionized the field and introduced the idea of differentiation as “bidirectional” event. An original breakthrough study ¹¹, provided the first experimental evidence that the process leading from ESCs to somatic cells could be reversed by over-expression of four transcription factors, including the Pou-family factor Oct4, the SRY (sex determining region Y)-box 2 factor Sox2, the homeodomain protein Nanog and either the Kruppel-like factor Klf4 or c-Myc. After transfer of four of these genes, somatic cells exhibited almost indistinguishable patterns of histone modifications and DNA methylation as ESCs ^12–14. They acquired the ability to produce viable chimeric mice when injected into blastocysts ¹² and were named induced pluripotent stem (iPS) cells¹¹. Although it is not completely understood how these transcription factors are able to induce pluripotency, it is clear that reprogramming entails several modifications of histone marks and changes in DNA methylation. Indeed treatment with pharmacological agents affecting histone modifications or inhibiting DNA methylation, facilitate the induction of iPS cells ^15–17. Together, these data suggest that the efficiency of reprogramming is dependent on histone modifications and DNA methylation.

This review introduces the epigenetic regulators of cell identity and then focuses on the steps characterizing the progression from ESCs to multipotential neural precursors to oligodendrocyte progenitors (OPC) and from OPCs to myelinating oligodendrocytes. Due to space constraints we apologize to colleagues whose work could not be cited ^{4, 18,19}.

Chromatin remodeling, DNA methylation and microRNAs

Histone modifications

Chromatin modifiers comprise two large groups of enzymatic activities. One group is responsible for secondary modifications of the histones, while the other group includes enzymes using ATP to unwind nucleosomes and affect DNA/histone interactions. Post-translational modifications include the deposition or removal of acetyl, methyl, phosphoryl, sumoyl, ubiquitin, ADP-ribosyl groups from amino acids in the histone tails (Box 1). The most studied histone modifications are acetylation/deacetylation (Box 2) and methylation (Table 1). The functional consequences of these modifications on transcription depend on the type of change and on the position of the targeted amino acid residue within the tail of nucleosomal histones. Methylation of lysine residues on histone H3 have been particularly studied due to their important role for transcriptional activation or repression. For example, deposition of methyl groups on lysine 4 of histone H3 (H3K4me3) is associated with gene activation, while methylation of lysine 9 (H3K9me3) or 27 (H3K27me3) results in repression ²⁰. Removal of methyl groups from the same residues has opposite effects ^{1, 10, 21}. The second group of enzymes includes components of the SWI/SNF complex which mobilize the nucleosomes and provide transcription factors (either inhibitors or activators of gene expression) access to DNA ^{22, 23}. Two principal forms have been identified: BAFs (i.e. BRG1/hBRM associated factors) and PBAF (Polybromo associated factors), although several additional denominations have been used to describe complexes formed by association with other subunits.

BOX 1 . Nucleosomal Histones.

The basic unit of chromatin is the nucleosome, which is composed of a core octamer of histones (H2A, H2B, H3, and H4) surrounded by 146 base pairs of DNA. The post-translational modifications on specific amino acid residues on the tails of the histones modulate the transcriptional activity of genes and the access of the transcriptional machinery to DNA. These modulations include acetylation/deacetylation, methylation, sumoylation and ubiquitination of lysine residues, methylation and citrullination of arginine residues, phosphorylation of serines and threonines, ADP ribosylation and isomerization of prolines (reviewed by ⁸⁰). The addition or removal of chemical groups on specific residues of histones contributes to transcriptional activation or repression. The presence of histone modifications or specific transcription factor binding to defined chromatin regions can be assessed using a technique called ChIP (Chromatin ImmunoPrecipitation). The advantage of this technique is that proteins and DNA are cross-linked in cells prior to chromatin extraction and this allows the detection of the precise histone modifications or transcription factors present in a given chromatin region in a specific cell type at a precise developmental time point. The proteins in chromatin can be immunoprecipitated with antibodies that recognize post-translational modifications of histones, transcription factors or chromatin modifying enzymes. The DNA-protein complex can be then recovered, reverse-cross-linked and subject to further analysis.

BOX 12. Histone acetylation and deacetylation.

Histone acetylation is functionally correlated with transcriptionally competent chromatin and is the result of a balance between the activity of histone acetyltransferases (HATs) and histone deacetylases (HDACs). HATs are categorized into two families: type A, which are nuclear and type B, which are cytoplasmic and contribute to acetylation of newly synthesized histones ⁸¹. Based on structural and functional differences, type A HATs are subdivided into five families. The GNAT (GCN5, PCAF and ELP3) and the p300/CBP family (p300 and CBP) acetylate various substrates, including histone and non-histone proteins, such as p53 and E1A. The MYST family (TIP60, MOZ, MOF, MORF and HBO1) has a MYST domain with the ability to interact with DNA. Other families of HATs include transcription factors with HAT activity (ATF2, TAF1 and TFIIIC90) and the nuclear hormone-related HATs (SRC4 and ACTR). Most histone acetylation occurs at the N-terminal histone tails, with the exception of H3K56 ⁵⁰. HDACs have been categorized into four groups: Class I (HDAC-1, -2, -3, and -8), Class II (HDAC-4, -5, -6, -7, -9 and -10), Class III (SIRT 1-7) and Class IV (HDAC11), based on subcellular localization, tissue distribution and sequence homology ⁸². Class I and IV are mainly nuclear. Class II HDACs are cytoplasmic, with the exception of HDAC10, but have the ability to shuttle to the nucleus. Class III includes nuclear and cytosolic NAD-dependent enzymes called Sirtuins, of which only Sirt2 has been shown to regulate oligodendrocyte differentiation ⁸³. Class IV HDACs include only HDAC11, which has been recently shown to positively modulate myelin gene expression ⁷⁸.

Table 1.

Histone methylation modifying enzymes

Mark	Methyltransferase	Refs	Demethylase	Refs
H3K4	Mll-1	21, 85	LSD1	86, 87
	Mll-2		JRAID1A
	Mll-3		JARID1B
	Mll-4		JARID1C
	Mll-5		JARID1D
	SET1A		JHDM1B
	SET1B
	ASH1
	SMYD3
	SET7/9
	PRDM9
H3K9	Suv39h1	80, 88	LSD1	87, 89
	Suv39h2		JHDM2A
	G9a		JHDM2B
	EuHMTase		JHDM2C
	ESET		JMJD2A
	RIZ1		JMJD2B
			JMJD2C
			JMJD2D
H3K27	EZH2	88	JMJD3	87, 89
			UTX
H3K36	SET2	88, 90, 91	JHDM1A	87, 89, 92
	NSD1		JHDM1B
	SMYD2		JMJD2A
	ASH1		JMJD2B
			JMJD2C
H3K79	DOT1L	84, 93
H4K20	Pr-SET7/8	94, 95
	SUV4-20H1
	SUV4-20H2

= transcription activation; = transcription repression, or heterochromatin formation and silencing. For the enzymes in grey boxes, please refer to the references for further information.

DNA methylation

DNA methylation is an important epigenetic mechanism regulating gene expression. It has been thoroughly characterized in X-chromosome inactivation in females ²⁴ and genomic imprinting ²⁵. DNA methylation occurs at the C-5 position of cytosine residues at CpG dinucleotides, due to the activity of the DNA methyltransferases Dnmt3a and Dnmt3b for de novo establishment of DNA methylation and Dnmt1 for its maintenance. These enzymes are expressed at high levels in undifferentiated ESCs ²⁶. However, knockdown of all three enzymes in ESCs does not affect self-renewal ²⁷, but prevents differentiation ²⁸. These results are consistent with the reports of unmethylated DNA in undifferentiated ESCs and with the correlation of DNA methylation with lineage commitment ^{29, 30}. Therefore DNA methylation is a mechanism that cells employ to induce a stable, although not irreversible repression of gene expression ³¹. DNA demethylation, for instance, occurs during cell division and during neural activity-induced neurogenesis in dentate gyrus ³². Few DNA demethylase candidates have been proposed to date ^31,32 and it is likely that their characterization and the discovery of new candidates will further our understanding of reprogramming.

MicroRNAs (miRNAs)

MicroRNAs are small (19–25nt) endogenous non-coding RNAs that play a role in post-transcriptional regulation of gene expression. They decrease transcripts by base-pairing to target mRNAs, and either inhibiting mRNA translation or facilitating its degradation ³³. In humans and rodents, miRNAs can be present in high or low copy number, with each miRNA having hundreds or even thousands of predicted mRNA targets.

An important relationship exists between microRNAs and DNA methylation, possibly mediated by the regulation of the levels of Dnmt1, 3a and 3b ^{34, 35}. Characterization of miRNA in undifferentiated and differentiated ESCs revealed the presence of clusters enriched in pluripotent cells³⁶ and others increasing during differentiation ³⁷. These studies and others conducted in neural stem cells have begun to shed some light on their function ^{38, 39}.

Transition 1: from pluripotent ESCs to multipotent neural stem cells

Histone modifications

ESCs are characterized by the properties of self-renewal and pluripotency (i.e. the ability to generate all cell types of an organism), which depend on the interplay between the pluripotency-associated factors (i.e. Oct4, Sox2 and Nanog) and epigenetic modulators. To gain a mechanistic insight on the chromatin structure in ESC, genome-wide studies were conducted using chromatin immunoprecipitation (see Box 1) and antibodies specific for pluripotency factors (i.e. Oct4, Sox2 or Nanog), for trimethylated K27 histone H3 (H3K27me3) and for the enzymatic activities responsible for this modification ^40–42. The repressive histone marks were detected in chromatin regions that were also characterized by the presence of Oct4, Sox2 or Nanog and that included several hundreds of genes, important for lineage commitment and differentiation ^{40, 41}. These repressive marks co-localize with proteins possessing enzymatic activities specific for H3K27, such as histone methyltransferases of the “polycomb group” (PcG) proteins. The PcG group includes a family of transcriptional repressors regulating body patterning. This group was originally discovered in Drosophila ^43–45 and consists of two complexes: PRC2, with methyltransferase activity forH3K27 (i.e. Ezh2, Eed), and PRC1, which includes Ring-domain proteins with E3 monoubiquitin ligase activity (i.e. Ring1A, Ring1B, Bmi)⁴⁶. In agreement with the importance of K27 trimethylation, ESCs lacking one component of the PRC2 complex displayed inappropriate expression of genes associated with lineage commitment and cellular differentiation ⁴². Therefore the undifferentiated state of ESCs is the result of a complex interplay between activating events and repressive mechanisms involving transcription factors and PcG-mediated deposition of repressive histone marks (i.e. H3K27me3).

The next challenge was to understand whether unique or similar mechanisms of repression of specific genes (for instance mesodermal genes), were shared by pluripotent cells (i.e. ESCs) and cells committed to a different lineage (i.e. neural precursors). The answer was provided by an interesting genome-wide ChIP study conducted in chromatin derived from ESCs or from lineage committed precursors, using antibodies specific for histone marks of transcriptional activation (i.e. H3K4me3) and of transcriptional repression (i.e. H3K27me3). In ESCs, the chromatin regions containing transcription factors involved in the regulation of developmental processes were characterized by the presence of both activating and repressive marks. These regions were therefore characterized as “transcriptionally ambivalent” in ESCs ⁴⁷. The authors named bivalent domains the co-existence of activating and repressive marks on precise chromatin regions. They also proposed that these domains could explain the repression of differentiation genes in ESCs (due to the presence of repressive H3K27me3), and their prompt activation (due to the presence of activating H3K4me3) during differentiation along distinct lineages ⁴⁷. One suggestion was that ESCs retain pluripotency by maintaining lineage commitment and differentiation genes repressed but “poised” for transcription (i.e. ready to be transcribed). In contrast committed precursors actively transcribe genes related to their lineage and silence those related to other lineages. In other words, the progression from ESC to neural precursors is characterized by changes in the mode of regulation of gene expression. Neural precursors remove the H3K4me3 activation mark and retain the repressive H3K27me3 mark on mesodermal or endodermal genes thereby resulting in silencing of genes associated with mesodermal or endodermal lineage choice (Fig. 1). At the same time, they retain the activating H3K4me3, and remove H3K27me3 in regions containing neural genes ⁴⁷ (Fig. 1). These studies suggested that the transition from pluripotent ES cell, (i.e. with the ability to generate endodermal, mesodermal and ectodermal derivatives) to multipotential neural “stem” cells (i.e. with the ability to generate only neuroectodermal derivatives), is characterized by the resolution of “bivalent domains” on lineage commitment and differentiation genes. However, “bivalent domains” on transcriptionally poised genes were detected not only in pluripotent ESCs, but also in neural multipotential precursors, embryonic fibroblasts and other undifferentiated cell lines ^{48, 49} and therefore, could not represent a distinctive feature for ESCs. The recent discovery of acetylated lysine residue 56 of histone H3 on Sox2, Oct4 and Nanog target genes in ESCs ⁵⁰, identifies this modification as prominent in ESCs.

Figure 1. Epigenetic mechanisms during the transition from embryonic stem cells (ESC) to neural stem cells (NSCs).

The progression of pluripotent ESCs to multipotent neural precursors or “neural stem” cells (NSCs) is characterized by the repression of critical transcription factors involved in pluripotency, such as Oct4, Sox2 and Nanog. (a) In ESCs, these genes are actively transcribed (green light) and this correlates with the presence of active histone marks, such as methylation of lysine residue K4 (yellow flag) and acetylation of lysine K9 and lysine 14 (blue triangle). (b) Lineage-specific genes (neural, endodermal or mesodermal genes), in contrast are maintained in a transcriptionally competent, but inactive state (“poised state”), due to the presence of “bivalent domains” that are characterized by the simultaneous presence of histone marks of activation (yellow flag) and repression (orange flag). (c)Upon neural lineage choice, a massive rearrangement of chromatin occurs. Regulatory regions of genes involved in pluripotency and genes related to other lineages become repressed. The repression is achieved by a combination of DNA methylation (white circles), through the recruitment of de novo DNA methyltransferase DNMT3a/3b and repressive histone methylation (red and orange flags). ⁴², ⁴⁷, ⁸⁴ Neural genes, in contrast, become “active” by removing the repressive marks and retaining activating methylation of K4.

DNA methylation and microRNAs

Genome-wide studies of changes in DNA methylation during the differentiation of ESCs to glutamatergic neurons identified the occurrence of de novo DNA methylation at the transition from undifferentiated pluripotent cells to multipotential precursors ³⁰. There is now general consensus regarding the presence of hypomethylated DNA in undifferentiated ESCs and the progressive increase in DNA methylation occurring at lineage commitment, but not during terminal differentiation ^{29, 30}. In molecular terms, de novo DNA methylation occurred more frequently on promoters characterized by the presence of H3K27me3 mark and it was frequently associated with the resolution of the “bivalent domains” due to loss of the activating mark H3K4me3. Based on these results, a model was proposed, suggesting that DNA methylation may serve as a “protective” function in repressing genes that are incompatible with the chosen lineage. In addition, methylated DNA was also found on the promoters of differentiation genes lacking the “bivalent domains” ^{29, 30}. Together these studies suggest that histone modifications and DNA methylation are part of the epigenetic mechanisms underlying pluripotency and therefore are essential for the establishment and maintenance of the “stem cell identity”. Finally it is important to add an additional level of regulation, which includes the function of microRNAs. A recent study has identified the microRNA miR-145, as a target of Oct4 with the ability to repress Sox2, Oct4 and Klf4 ⁵¹. The complex layers of regulation of the undifferentiated state by microRNAs have been nicely presented in other reviews ⁵².

In summary, the ES state is characterized by the ability to generate all the cell types necessary to the organism. This elaborate task is accomplished by a combination of transcription factors, DNA methylation and histone modifications that suppress the expression of genes involved in lineage commitment and differentiation, while retaining them in a chromatin conformation that is poised for transcription. The progression towards multipotential precursors (also named “neural stem cells”) is characterized by more stable silencing of genes involved in other lineages (i.e. endoderm and mesoderm) and expression of neural genes.

Transition 2: From multipotent NSC to oligodendrocyte progenitors (OPC)

Histone modifications

The concept of progressive restriction of cell lineage potential during differentiation, applies not only to the transition from pluripotent ES cells to multipotent neural precursors, but also to the restriction of multipotential neural precursors to lineage committed oligodendrocyte progenitors (OPC). This choice is characterized by the progressive decrease of genes like Sox2 and chromatin modifications on astrocytic and neuronal genes that is initiated by the activity of histone deacetylases ⁵³ and is antagonized by Brca1 and Brm ⁵⁴. Inhibiting HDAC activity in OPCs, using pharmacological inhibitors ⁵⁵, or culturing the cells for prolonged periods of time in the presence of mitogens ⁵⁴, or keeping the cells in the presence of both HDAC inhibitors and mitogens ⁵⁶ reverted these committed progenitors to multipotent cells characterized by Sox2 expression.

The idea that commitment is associated with remodeling of chromatin on astrocytic and neuronal genes has not yet been formally demonstrated. However, it is important to mention here that several studies have reported the generation of neurons and astrocytes from oligodendrocyte progenitors ^{53, 57–59}. If OPC identity requires the repression of neuronal genes, it is likely that this event may require the presence of chromatin remodeling complexes containing the RE1 silencing transcription factor (REST)/neuron restrictive silencing factor (NRSF). REST is a powerful repressor with the ability to recruit the co-repressors Sin3A and Co-REST and the histone deacetylases HDAC1/2 ^{60, 61}. This explanation would account for the increased number of cells expressing neuronal markers after treatment with HDAC inhibitors ⁵³.

The expression of astrocytic specific genes, such as glial fibrillary acidic protein (Gfap) and S100beta, could similarly be down-regulated by deacetylation of nucleosomal histones (in this review, we shall consider Gfap expression as an astrocytic marker, even though it is acknowledged that it can also be expressed by adult neural stem cells ⁶²). The involvement of histone acetylation/deacetylation in regulating astrocytic gene expression is supported by multiple lines of evidence. Differentiation of neural stem cells into astrocytes appears to be associated with the recruitment of the co-activator complexes including STAT1/3, histone acetyltransferases and Smad1 to specific regions of the Gfap promoter ⁶³. Treatment with HDAC inhibitors increases the expression of S100 and Gfap in cultured oligodendrocyte progenitors and in neonatal rats after transplantation ⁵³. Finally, in developing astrocytes Gfap is repressed by complexes containing HDAC, the nuclear receptor co-repressor (N-CoR) and adaptor proteins ⁶⁴, and N-CoR knockout mice are characterized by precocious expression of Gfap ^{64, 65}. Similar mechanisms could occur in the cell after oligodendrocyte lineage choice.

The involvement of histone methylation in the transition from NSC to OPC is also suggested by experimental evidence on the role of PcG proteins, especially Ezh2 ^{66, 67}. Ezh2 was down-regulated during commitment to neuronal and astrocytic lineage and up-regulated during commitment to the oligodendrocytic lineage. Overexpression of Ezh2 in mouse neural stem cells (NSCs) increased the number of oligodendrocytes at the expenses of astrocytes, while silencing of Ezh2 yielded opposite results ⁶⁶. Another PcG protein, member of the PRC1 complex, the histone H2A E3 monoubiquitin ligase Ring1B, seems to be essential for oligodendrocyte differentiation ⁶⁷. Ring1B deletion through Cre-mediated recombination in vitro selectively resulted in decreased number of oligodendrocytes from NSCs isolated from mouse embryonic olfactory bulb ⁶⁷. Thus, components from both PRC2 and PRC1 complexes are critical for oligodendrogliogenesis from NSCs. One intriguing possibility is that these PcG proteins could play an essential role in down-regulating genes involved in neurogenesis and astrogenesis, thereby allowing oligodendrogliogenesis to take place. Another possibility is that oligodendrogliogenesis occurs only after PcG-mediated repression of the neurogenic basic HLH transcription factor neurogenin ⁶⁸, similar to what is described for astrogliogenesis ⁶⁹. Indeed components of the PRC1 complex modulate timing of astrogenesis in the developing brain and were shown to repress the neurogenin promoter in a developmental stage–related manner ⁶⁹.

Role of microRNAs in modulating neuronal and non-neuronal transcripts

Finally, recent reports have underlined the importance of miRNAs in modulating astrocytic and neuronal differentiation. Krichevsky et al. demonstrated that over-expression of miR-124a and miR-9 in mouse ESCs inhibited astrogliogenesis, possibly by interfering with STAT3 phosphorylation, which is critical for STAT3 activation. This was achieved by microRNA targeting of STAT3 upstream factors ⁷⁰. MiR-124a also negatively regulates non-neuronal gene expression in neural progenitor cells and is regulated by REST/NRSF. During neural differentiation, the expression of REST decreases and leads to de-repression of miR-124a, thereby allowing the degradation of non-neuronal transcripts ⁷¹.

Based on the experimental evidence discussed above, we propose that the transition from NSC to OPC requires a series of repressive events resulting from the cross-talk of histone modifications, DNA methylation and miRNAs.

Transition 3: From oligodendrocyte progenitors to myelin-forming cells

Histone modifications

Ultrastructural studies have described the characteristic appearance of compared chromatin in myelinating and non-myelinating oligodendrocytes ⁷². Because compaction is associated with transcriptional repression, we propose a model of oligodendrocyte differentiation characterized by the progressive decrease of transcriptional inhibitors, followed by the up-regulation of activators and then myelin gene expression (Fig. 2). This model is supported by the detection of transcriptional inhibitors in proliferating OPCs followed by their gradual decrease during maturation into myelin-forming cells ^{55, 73}. An example is the expression of Hes5, a transcriptional inhibitor that not only inhibits myelin gene expression by direct recruitment of histone deacetylases to myelin gene promoters, but also decreases the effectiveness of the activators by direct repression of their promoters and protein sequestration ⁷³. The expression of this inhibitor is regulated by histone deacetylase activity ⁷⁴ and the progressive decline of its levels modulates myelin gene expression by decreasing inhibition on the myelin gene promoters and releasing the inhibition on the activators (such as Mash1 and Sox10) ⁷³.

Figure 2. Epigenetic mechanisms modulating the transition from neural stem cell (NSCs) to oligodendrocyte progenitor (OPCs) and from OPC to myelin-forming cells.

Schematic diagram of oligodendrocyte lineage choice in two steps. The top panel (a) depicts a central role of HDAC in oligodendrogliogenesis, due to repression of alternative fates (i.e. repression of neuronal or astrocytic lineage genes). The bottom panel (b) shows a schematic representation of the process of differentiation of OPCs into myelin-forming cells. OPCs express high levels of transcriptional inhibitors of myelin genes. Differentiation is characterized by the increased expression of transcriptional activators and by the progressive down-regulation of the inhibitors due to the removal of active histone marks (pink triangles) and addition of repressive histone marks on their promoters, including deacetylation ⁵⁵ and repressive methylation (red flag). For myelin genes, two scenarios are presented. The first one is a hypothetical model showing acetylated histones during active transcription of myelin genes. The second one refers to a study identifying HDAC11 as a positive regulator of myelin gene expression ⁷⁸.

Consistent with the model described above, multiple studies have demonstrated the requirement of HDAC activity for oligodendrocyte differentiation in vitro and in vivo. Systemic administration of rat pups with HDAC inhibitors prevented the differentiation of OPCs ⁷⁵. Several members of class I HDAC, including HDAC1 and HDAC2, were shown to be expressed in oligodendrocyte cells in vivo and to be critical for oligodendrocyte differentiation in development and repair ^{55, 75}. Silencing of HDAC1 and HDAC2 resulted in defective OPC differentiation in vitro, and was associated with high levels of Sox2 ⁵⁵. Hdac1 and Hdac2 were also recruited to the promoter of Sox2 during repair of demyelinating lesions in adult rodents ⁵⁵. Microinjections of hdac1 antisense oligonucleotides in zebrafish and analysis of hdac1 mutants revealed the absence of oligodendrocytes ⁷⁶. Further supporting evidence was provided by the phenotypic analysis of the conditional Hdac1 and Hdac2 double knockout mice, generated by crossing floxed Hdac1 and Hdac2 mice with Olig1-Cre transgenic lines. The double knockout mice, developed severe hypomyelination and died during the second postnatal week ⁷⁷.

In addition, HDAC1 and HDAC2 form complexes and repress the expression of transcription factors that inhibit myelin gene expression and oligodendrocyte differentiation, such as Hes5 and Tcf4 ^{55, 77}. A zinc-finger protein, Yin Yang 1 (YY1), recruits HDAC1 to the promoter region of Tcf4 and Id4, thereby decreasing their levels due to transcriptional repression. An additional model of down-regulation of Id2/4 transcript levels is proposed and associated with competition of Hdac1/2 with the downstream Wnt signaling molecule β-catenin for the binding of Tcf4 ⁷⁷. Together these results suggest that myelin gene expression requires the repression of transcriptional inhibitors and subsequent surge of transcriptional activators. Additional regulators of myelin genes are summarized in Table 2.

Table 2.

Transcription regulators of myelin genes

Molecules	Effect	Family	References
Hes5	−	basic helix-loop-helix (bHLH)	73
Olig1/2	+	basic helix-loop-helix (bHLH)	96, 97
ASCL1 (Mash1)	+	basic helix-loop-helix (bHLH)	98
Id2, Id4	−	helix-loop-helix (HLH)	99, 100
Tcf7L2	−	high mobility group (HMG)	77, 101, 102
Sox4, Sox5, Sox6, Sox11	−	high mobility group (HMG)	103, 104
Sox8, Sox9, Sox10, Sox17	+	high mobility group (HMG)	97, 103–105
Nuclear hormone receptor	+		106, 107
YY1	+	Zinc-finger protein	101
Zfp488	+	Zinc-finger protein	108
MYT1	+/−	Zinc-finger protein	109, 110
MRF	+	NDT80 domain	111
Nkx2.2	+/−	Homeodomain	112, 113
Nkx6.1, NKx6.2	+	Homeodomain	114

An alternative model has proposed a different role for the histone deacetylase isoform, Hdac11. This histone deacetylase is highly expressed during myelination and modulates deacetylation of H3K9/K14 on myelin genes during oligodendrocyte maturation. Expression of HDAC11 increases in maturing OL-1 oligodendrocytes and primary cultured OPCs, along with its recruitment to the MBP and PLP genes. Silencing HDAC11 with siRNA increased acetylated H3K9/K14, and led to reduced myelin gene expression due to arrested oligodendrocyte maturation ⁷⁸. Although the mechanisms by which histone deacetylation actively regulates myelin gene expression remain elusive, collectively, these studies clearly demonstrate that histone deacetylation is critical for OPC differentiation.

The role of microRNAs during oligodendrocyte differentiation has been defined using specific microarrays ³⁸. This analysis identified several miRNAs and 37 of them had a specific mRNA target bias, which is computationally determined by the co-expression of a certain miRNA with its predicted targets. In particular, the study demonstrated that miR-9 negatively regulated the expression of peripheral myelin protein pmp22, a protein expressed in Schwann cells, by interacting with the 3′ untranslated region and preventing expression of this protein in oligodendrocytes ³⁸. Additionally, miR-23 negatively regulates laminB1, a structural protein expressed in the nuclear envelope, whose overexpression characterizes autosomal dominant leukodystrophy (ADLD) patients. Over-expression of lamin B1 is associated with reduced myelin gene expression in vitro and in vivo, altered subcellular localization of MBP and PLP and reduced branching of oligodendrocyte, that can be rescued by co-expression of miR-23 ⁷⁹. Together, these studies provide compelling evidence for miRNA-mediated regulation in the differentiation of progenitors to mature oligodendrocytes.

Conclusion

The recent advances in regulation of gene expression have revealed the existence of an intricate regulatory network including transcription factors, chromatin modulators DNA methylation and microRNAs. This network defines the unique identity of each cell type, via a process of “epigenetic programming”. In this review we discuss the subsequent transitions from pluripotent ESCs to myelinating oligodendrocytes. The overall concept is that repressive epigenetic regulations play a critical role for the progressive restriction of the lineage potential, due to the presence of covalent histone modifications, DNA methylation and the emerging role of microRNA. A better understanding of the molecular events characterizing each transition may provide an important framework for future therapeutic studies aimed at repair of demyelinating conditions.

BOX. OUTSTANDING QUESTIONS.

1. Oligodendrocytes are generated by a series of sequential transitions from stem cells through multipotential progenitors to oligodendrocyte progenitors that differentiate into myelin-forming oligodendrocytes. What are the main epigenetic events characterizing each of these transitions?

2. Does oligodendrogliogenesis entail the repression of transcriptional programs of neurogenesis and astrogliogenesis?

3. Does the functional state of myelinating oligodendrocyte coincide with the tail end of terminal differentiation? Is the differentiation process from stem cell to oligodendrocyte uni- or bi-directional?