TET1-mediated Oligodendrocyte Homeostasis Regulates Myelination and Synaptic Functions

1 Department of Neurobiology and Collaborative Innovation Center for Brain Science, School of Basic Medicine, Fourth Medical University, Xi'an, Shaanxi 710032, China 2 Division of Experimental Hematology and Cancer Biology, Department of Pediatrics, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229, USA 3 Shaanxi Key Laboratory of Brain Disorders, Xi’an, Shaanxi 710021 China 4 School of Electronic, Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China 5 School of Neuroscience & Department of Biological Sciences, Virginia Tech, Blacksburg, VA 24061, USA 6 These authors contributed equally to this work. 7 Lead contact


Introduction
Myelination by oligodendrocytes (OLs) enables saltatory conduction of action potentials and provides long-term trophic support for axons, maintaining integrity throughout the central nervous system (CNS) (Nave and Werner, 2014). The formation of mature myelinating OLs is a complex process that is tightly coordinated by genetic and epigenetic events in a spatially and temporally specific manner (Emery and Lu, 2015;Zuchero and Barres, 2013). Epigenetic regulation such as DNA methylation, histone modifications, and chromatin remodeling are critical for multiple aspects of OL development, function and regeneration (Gregath and Lu, 2018;Liu et al., 2016;Lu et al., 2019). For instance, proper maintenance of genomic cytosine methylation (5mC) is essential for the normal development, homeostasis, and function of mammalian cells (Iurlaro et al., 2017;Smith and Meissner, 2013).
Genetic ablation of DNA methyltransferases 1 (DNMT1), which maintains DNA methylation after replication, resulted in impaired OL precursor cell (OPC) expansion and differentiation during early development (Moyon et al., 2016). In adults, the role of DNMT1 appears diminished as reduction in Dnmt3a expression, but not Dnmt1, leads to defective OPC differentiation and remyelination after injury (Moyon et al., 2017).
Although DNA methylation in terminally differentiated cells was largely irreversible, the modified nucleotide 5-hydroxymethylcytosine (5hmC) was recently discovered as an intermediate product generated during cytosine demethylation (Kriaucionis and Heintz, 2009;Tahiliani et al., 2009). DNA demethylation like methylation is a highly regulated process mediated by specialized enzymes, the Ten-Eleven Translocation (TET) family of dioxygenases. TET enzymes oxidize 5mC into 5hmC and initiate the active DNA demethylation process (Ito et al., 2010;Tahiliani et al., 2009). Dynamic cytosine methylation or demethylation has been established as common epigenetic modification regulating various processes from development to diseases in a cell-type and context-dependent manner (Graff et al., 2011;Sun et al., 2014;Szulwach et al., 2011b). Although TET enzymes are present in OL lineage cells (Zhao et al., 2014), at present, whether and how DNA demethylation contribute to OL lineage development, myelination and remyelination after injury remain poorly defined.
In this study, we demonstrate a genome-wide shift in 5hmC landscape during OL specification and identify an age-dependent function of TET1 in OL lineage development and homeostasis. A TET1-regulated epigenetic program is also required for efficient remyelination as depletion of Tet1 in adult animals impaires recovery after demyelinating injury. Tet1 depletion results in genome-wide alterations in 5hmC and transcriptome profiles in OPCs. One pathway that is disproportionally altered in this process is a set of calcium transporter genes, such as Cav1.2, Cav2.1 and Itpr2. Our results further show that TET1-mediated Ca 2+ homeostasis is critical for OL differentiation in vitro. Strikingly, we find that Tet1 depletion in OL lineage cells leads to reduced expression of secreted factors such as GDNF (glial cell line derived neurotrophic factor) and an impairment of synaptic transmission.
These data suggest that TET1 regulates normal OL lineage differentiation, proper oligodendrocyteneuron communication and neuronal activities by both cell-autonomous and cell non-autonomous signaling pathways.

DNA hydroxymethylation landscape dynamics during OL lineage differentiation
To investigate the 5hmC landscape marking DNA demethylation during the OL lineage transition, we carried out antibody-based 5hmC immunoprecipitation combined with Illumina sequencing (hMeDIPseq) (Ficz et al., 2011;Xu et al., 2011) to analyze the 5hmC distribution across the genome. We compared 5hmC distribution within OPCs to those in neural progenitor cells (NPCs) (Li et al., 2017) and identified 1237 genes that were specifically hydroxymethylated in the promoter-TSS region of OPCs but not in NPCs ( Figure 1A). Gene ontology (GO) analysis identified gene terms that are highly associated with OPC differentiation, such as cell projection organization, fatty acid transport, regulation of cytosolic calcium ion concentration, as well as signaling pathways that are essential for OL development, such as the G-protein coupled receptor pathway (Chen et al., 2009;Mogha et al., 2016) ( Figure 1B). Further comparisons with a neural cell-type transcriptome dataset (Zhang et al., 2014) revealed a significant increase in the 5hmC signal among OL-associated gene loci during OPC differentiation, including Cspg4 (chondroitin sulfate proteoglycan 4), Mag (myelin-associated glycoprotein) and Elovl7 (elongation of very long chain fatty acids protein 7) ( Figure 1C and Figure   S1). In contrast, the genes with 5hmC peaks enriched in promoter-TSS regions from NPCs were associated with negative regulation of OL differentiation including Id2, Ngf and Zfp28 ( Figure 1D).
These data suggested a unique distribution pattern of genomic 5hmCs in the gene loci associated with OL lineage progression during the transition of OPCs from NPCs.

Deletion of Tet1 in OL lineage causes myelination deficits at early postnatal stages
Given that DNA hydroxymethylation is catalyzed by the TET family of dioxygenases and that TET enzymes are present in OL lineage cells (Zhao et al., 2014), we assessed the function of TET family enzymes TET1 and TET3 in OL development since TET2 had no detectable effects in OL lineage development (personal communication with Dr. P. Casaccia). We crossed the Tet1 f/f (Zhang et al., 2013) and Tet3 f/f (Gu et al., 2011) mice with Olig1-Cre line (Xin et al., 2005) to knockout the catalytic domains of these TETs early in the OL lineage ( Figure 1E and Figure S2A). The resulting Tet1 flox/flox ;Olig1Cre +/-(Tet1 cKO) and Tet3 flox/flox ;Olig1Cre +/-(Tet3 cKO) mice were born at Mendelian ratios and appeared normal at birth.
We did not detect significant differences in either the number of mature OLs or myelin protein expression between heterozygous Tet1 floxed mice (Tet1 flox/+ ;Olig1Cre +/-), Cre control (Tet1 +/+ ;Olig1Cre +/-), or wild-type mice ( Figure S3), heterozygous littermates were then used as controls. To assess Cre-mediated Tet1 depletion, we compared Tet1 expression between groups in OPCs, which have been previously demonstrated to have the highest Tet1 levels during OL differentiation (Zhao et al., 2014). Immunostaining and quantitative real-time PCR revealed decreased expression of TET1 in the corpus callosum at P4 ( Figure 1F-G) and in purified OPC cultures ( Figure 1H).
To investigate the effects of TET1 on OL development, we examined the expression of SOX10 (an OL lineage marker), and mature OL markers CC1 and MBP (myelin basic protein). The number of CC1 + mature OLs was significantly reduced in juvenile Tet1 cKO mice ( Figure 1I-J), but these developmental defects were no longer apparent in adult mutants ( Figure 1J and Figure S4A).
Expression of MBP was also substantially decreased in both cortex (gray matter) and corpus callosum (white matter) from P16 Tet1 cKO mice ( Figure 1K), but was recovered to control levels in older animals ( Figure S4A-B). Additionally, the purified oligodendrocyte precursor cells from Tet1 cKO mice showed prominently decreased numbers of CNPase-or MBP-positive cells during differentiation compared to control (Figure 2A-B). Together, these observations indicate that Tet1 loss causes a significant delay in OL maturation both in vivo and in vitro. In contrast, the same experiments in the Tet3 cKO animals did not show any significant differences between mutants and controls ( Figure S2). Therefore, we focused our efforts onto examining the processes underlying the observed myelination defects in Tet1 cKO mice.
The ultrastructure observation of optic nerve by electron microscopy (EM) revealed that the number of myelinated axons was significantly reduced in Tet1 mutants ( Figure 2C-D), and a few myelinated axons in Tet1 cKO optic nerves were characterized by higher G ratio and thinner myelin sheaths ( Figure 2C, E). The myelination deficits were further examined in the corpus callosum from Tet1 cKO mice at P14 and P27 ( Figure 2F-J). However, the myelin ultrastructure defects were no longer identifiable in two-month old adult Tet1 cKO animals ( Figure 2F, K-L). Notably, this myelin defect was not recapitulated in spinal cord. Although the numbers of CC1 + mature OLs were slightly decreased in mutant spinal cord at P8 ( Figure S5A-C), we did not detect a substantial alteration in MBP expression at this age ( Figure S5D). In addition, similar ultrastructure of myelin was detected under EM at P14 ( Figure S5E-F). These observations suggest a stage-dependent and regionspecific function of TET1 in CNS myelination.
Given the recovery seen in adults from the earlier developmental defects in myelination, we asked whether this recovery was due to the compensation from cells that escaped Cre mediated recombination. In adulthood, TET1 expression in OLs was low ( Figure S4C) as previously reported (Zhao et al., 2014), but the percentage of TET1 + /Sox10 + OL lineage cells reached ~80% in Tet1 cKO mice ( Figure S4C-D). These results suggest that the adult OLs expressing TET1 may escape from Cre-mediated Tet1 depletion and compensate for the myelin deficiency earlier in the Tet1 cKO animals.
To evaluate the neurological significance of hypomyelination in Tet1 cKO mice, we analyzed stimulus-evoked compound action potential (CAP) in optic nerves (Devaux et al., 2002;Devaux and Gow, 2008;Stys et al., 1991). Suction electrodes back filled with artificial cerebrospinal fluid (aCSF) were used for stimulation and recording. In P14 optic nerves, both the peak amplitude and the CAP area, which are indexes of excited myelinated axon numbers and nerve function (Devaux et al., 2002;Devaux and Gow, 2008), showed significant reduction under different stimulating currents in

Ablation of Tet1 results in defects in OPC cell cycle progression
Concomitant with the myelin deficiency, we observed a marked reduction of Olig2-positive cells from the embryonic stage in Tet1 cKO animals at E15.5 ( Figure S4E-F). Moreover, the number of plateletderived growth factor receptor α (PDGFRα) + cells in the mutant cortex was reduced to ~63% compared to the controls in P1 mutant, suggesting a downsized OPC pool ( Figure 3A-B).
To determine the underlying defects leading to a reduction in the OPC and OL population in Tet1 juvenile mutants, we first tested the possibility of apoptosis in OPCs with the TUNEL assay.
Combined with Olig2 staining, brain sections from E14.5, E17.5 and P1 mice revealed no distinguishable changes in the number of TUNEL + apoptotic cells among Olig2 + OL lineage cells between Tet1 cKO animals and control littermates ( Figure S7A-B).
Next, a BrdU incorporation assay was performed to examine the proliferation of OPCs. Tet1 cKO mice at P1 were pulse-chased with BrdU and sacrificed 2 hours later. Compared to controls, the percentage of BrdU and Sox10 double-positive cells in the OL lineage showed a slight but significant increase in the Tet1 cKO cortex ( Figure 3C-D). Similar results were also observed with another cell proliferation marker Ki67 ( Figure S7C-D). In addition, we validated the proliferation changes in purified OPC cultures. Both Ki67 and BrdU staining showed increased proliferating numbers of OPCs from Tet1 mutants ( Figure 3E-G).
To further investigate the stage of cell cycle blockage in Tet1-deficient OPCs, we performed flow cytometry with propidium iodide (PI) DNA staining in purified OPCs. A significant increase in the percentage of cells in S phase (23.21 % ± 3.12) and G2/M phase (13.83 % ± 1.91) was observed in OPCs from Tet1 mutants compared to the controls (11.84% ± 2.46 for S phase and 7.09 % ± 1.17 for G2/M phase), and there was a concomitant reduction in the number of cells in G1 phase in Tet1 mutants ( Figure 3H-I). These results suggested that the proliferation of OPCs from Tet1 cKO brain might be blocked at the transition from G2/M to G1 phase, which could impair OPC cell cycle progression, leading to a reduction in OPC numbers.
In addition, we noted that Tet1 deletion did not appear to substantially alter the number of other neural cell types in the brain. Western blot and immunostaining with antibodies against DCX, a marker for newly generated neurons, NeuN, a mature neuron marker, and ALDHL1, an astrocyte marker, revealed that there was comparable neuron and astrocyte pools between controls and Tet1 mutants ( Figure S8). Taken together, our data suggest that the abnormal cell cycle progression of OPCs and delayed OPC differentiation may contribute to the reduced OL numbers and hypomyelination in Tet1-deficient juvenile mice.

Ablation of Tet1 impairs adult remyelination after demyelination
Given the critical role of TET1 in early oligodendrocyte development, we hypothesized that TET1 is also required for remyelination after demyelination in the adult brain. We induced demyelinated lesions in the corpus callosum via stereotaxic guided lysolecithin (LPC) injections ( Figure 4A). LPC induces rapid myelin breakdown followed by myelin regeneration through an OPC recruitment phase at 7 days post-lesion (7 dpl) and a remyelination phase at 14 dpl in vivo. TET1 + cell numbers were increased substantially in the lesion site at 7 dpl relative to vehicle controls ( Figure 4B-C). In particular, the expression levels of TET1 in Olig2 + cells appeared much more robust after LPC treatment ( Figure 4B-C).
To determine the potential role of TET1 in remyelination, we bred Tet1 flox/flox mice with NG2-CreER T (Zhu et al., 2011), an OPC-specific tamoxifen-inducible Cre line to generate Tet1 OPC-iKO (NG2-CreER T :Tet1 flox/flox ) animals. In the Tet1 OPC-iKO mice, which were treated with tamoxifen injection from P2-P5 daily, we observed a reduction in CC1 + and MBP + OLs at P7 and P14 ( Figure   S9). To induce recombination in adult mice, 8-week-old Tet1 OPC-iKO mice were injected daily with tamoxifen for 8 days, starting 3 days prior to LPC injection in the corpus callosum ( Figure 4D).
Heterozygous littermates (NG2-CreER T :Tet1 flox/+ ) served as controls. Brains were harvested at 7, 14 and 21 dpl. To determine the extent of remyelination, we examined the expression of OPC markers and myelin related genes. Loss of Tet1 did not appear to impair the recruitment of PDGFRα + OPCs and the number of OPCs in the lesion was comparable between control and Tet1 OPC-iKO mice during the recruitment phase at dpl 7 ( Figure 4E-F). In contrast, there were significantly fewer GSTpi + differentiating OLs in the lesion site during the remyelination phase at dpl 14 and dpl 21 relative to control ( Figure 4E, G). Consistent with a reduction in the number of differentiating OLs, MBP was also reduced in Tet1-iKO lesions compared to control at dpl 21 ( Figure 4H). Notably, far fewer remyelinated axons were detected in the lesions of Tet1-iKO mice than controls at dpl 21 ( Figure 4I-J). The percentage of myelinating axons and the thickness of newly generated myelin sheaths around axons assessed by g-ratios were significantly reduced in Tet1-iKO mutants ( Figure 4J-K).
These observations indicate that TET1 is required for the proper remyelination in the context of white matter injury.

Transcriptome alterations and a genome-wide decrease of 5hmC in Tet1-ablated OPCs
To investigate the molecular underpinnings of the observed defects in early oligodendrocyte development, we compared the RNA transcriptome in purified OPC cultures from control and Tet1 cKO neonates (Figures 5A). There were approximately 1880 genes downregulated and 881 genes upregulated in Tet1 mutants compared with controls (FDR<0.05, Log2>1 or <-1) ( Table S1) Since TET1 mediates DNA hydroxymethylation/demethylation, we next tested the level of 5hmC in oligodendrocytes from Tet1 mutants. In P27 brain sections, immunostaining of 5hmC along with OL marker CC1, revealed a striking reduction in 5hmC intensity in oligodendrocytes ( Figure 6A-B), which strongly suggested that 5hmC is involved in TET1-regulated OL differentiation.
To further compare the genome-wide 5hmC distribution changes, we performed hMeDIP-seq in OPC cultures from controls and Tet1 mutants. Tet1 cKO OPCs showed a dramatic reduction in 5hmC peak signals compared to controls ( Figure S10A), which was consistent with the immunostaining results. In both groups, most 5hmC peaks resided in the intergenic region, and less than 40% of peaks were found to be within the gene bodies of annotated RefSeq genes ( Figure S10B), which is different from the distribution pattern in mouse embryonic stem cells (Xu et al., 2011) andneurons (Hahn et al., 2013). After plotting the distribution of 5hmC peaks over RefSeq genes, we found that 5hmC was reduced near the transcription start site (TSS) and transcription terminal site (TTS) in control OPCs, and Tet1 depletion caused a global reduction of 5hmC, especially at the intragenic region, promoter, and TTS ( Figure 6C and Figure S10B). For the 5hmC peak distribution around TSS (5-kb upstream and downstream of TSS), heatmap clustering revealed that there were five distributed groups, and there was an overall decrease in the level among all five groups in Tet1 cKO samples ( Figure 6D). Further analysis showed that most differentially hydroxymethylated regions (DhMRs) between the Tet1 cKO and control groups were associated with low CpG density, which was less than 1 CpG per 100 bp ( Figure S10C). We next examined the effects of TET1 mediated DNA hydroxymethylome on gene expression.
By integrating RNA-seq data with hMeDIP-seq data, we observed that among the decreased lower hydroxymethylated gene peaks in gene body regions from Tet1 cKO mice, 12.83% genes expression were downregulated (1026 of 7998) and 4.89% genes were upregulated (391 of 7998) ( Figure 6E).
The correlation percentage is comparable to DhMRs in promoter regions, 13.46% and 6.42% in downregulated and upregulated genes, respectively (Figure S10D-E). Therefore, these observations indicated that DNA hydroxymethylation is positively correlated to gene expression.
Among the genes that showed both downregulated mRNA expression and decreased gene body 5hmC levels in Tet1 cKO mice, many are associated with myelination, cell division and calcium transportation ( Figure 6F When searching for TET1-5hmc regulated factors that may involve in OL development and homeostasis, we noticed that there was a cluster of calcium transporter genes among the downregulated genes in Tet1 cKO group, which can be further confirmed by qRT-PCR assays (Figures 5I-J). CACNA1a, CACNA1c, CACNA2d1, CACNB4, and CACNG5 are all well-known plasma membrane voltage-operated Ca 2+ channels (VOCCs) that are expressed in OPCs and contribute to Ca 2+ dynamics in these cells (Larson et al., 2016;Zhang et al., 2014). In particular, Ca 2+ influx mediated by CACNA1c, also known as Cav1.2, is required for oligodendrocyte differentiation (Cheli et al., 2016;Cheli et al., 2015). Another gene Itpr2, which encodes an endoplasmic reticulum (ER) localized type 2 IP3 receptor (IP3R2), a specific Ca 2+ channel in postmitotic OLs (Zeisel et al., 2015), also showed decreased mRNA expression in OPCs from Tet1 cKO mice ( Figure 5J). In addition, a set of the genes encoding secreted factors (e.g. GDNF) and extracellular matrix proteins (e.g. Tenascin-R and Reelin), were also significantly decreased in Tet1 cKO OPCs ( Figure 5K-L).
Especially, the level of GDNF from both the cell extraction and supernatant of OPC cultures were significantly reduced in Tet1 cKO mice ( Figure S11A-B). These observations indicated that TET1-5hmc regulates calcium homeostasis in OL lineage cells and the secretion of regulators of neuronal development for the intracellular communication between OPCs and their neighbors.

Impaired Ca 2+ dynamics in Tet1 cKO leads to OPC differentiation defects
Calcium signaling and homeostasis have been shown to be important for OPC migration, differentiation and initiation of myelin formation (Butt, 2006;Friess et al., 2016;Li et al., 2018;Zhang et al., 2019). The downregulation of multiple calcium transporter genes in OPCs from Tet1 cKO mice prompt us to test the [Ca 2+ ]i oscillations in culture. A Fluo4-based Ca 2+ -imaging approach was employed to compare the influx of Ca 2+ between groups in combination with Bay K 8644, a L-type calcium channel agonist (Greenberg et al., 1984), or verapamil, an inhibitor of L-type voltage operated Ca 2+ channels (VOCCs) (Cheli et al., 2015). In OPC cultures, we noticed that bath To confirm, we tested the effects of another trigger for Ca 2+ entry, adenosine triphosphate (ATP), on [Ca 2+ ]i oscillations in OPCs. ATP can mobilize Ca 2+ from intracellular ER stores or trigger Ca 2+ influx across the plasma membrane (Kirischuk et al., 1995;Stevens et al., 2002). Similar to the results of Bay K 8644 application, the amplitude of 100 μM ATP-induced [Ca 2+ ]i transients was significantly higher in control cultures (7.061±0.2118 AU) than in the Tet1 mutant cultures (4.417±0.1649 AU, Figure 7E-F), and the kinetics of onset were slower in the mutants as well (3.647±0.2791 in control and 2.968±0.1595 in mutant, Figure 7G). Since ATP evokes large and prolonged increases in [Ca 2+ ]i from intracellular stores mainly via P2Y purinoceptors, especially P2Y1 in OL (Agresti et al., 2005;Kirischuk et al., 1995), we applied a specific antagonist for the P2Y1 receptor, MRS2179 (100 μM), prior to ATP stimulation. In the presence of MRS2179, the [Ca 2+ ]i response was significantly repressed in both groups ( Figure 7E), which is consistent with the findings of other study (Agresti et al., 2005), and indicated that ATP-induced [Ca 2+ ]i oscillations in OPCs are mainly attributable to an action on P2Y1 receptors. These results suggested that the reduced expression of calcium transporters on both the cell membrane and the ER membrane in Tet1 cKO group is responsible for the reduced [Ca 2+ ]i oscillations in OPCs from Tet1 cKO mice.
As for the consequences of [Ca 2+ ]i oscillation changes in Tet1 cKO OPCs, we further examined the effects of calcium channel agonist Bay K 8644 on the differentiation and proliferation of OPCs.
Consistent with the results of high K + application (Cheli et al., 2015), daily three consecutive pulses Reelin (Ventruti et al., 2011;Wasser and Herz, 2017), are also critical for shaping and strengthening neuronal synapses. We therefore determined whether the epigenetic alterations due to Tet1 ablation in OL lineage cells can affect synaptic density and strength in a non-cell autonomous manner.
For sEPSCs, there is a chance of action potential-driven events due to intrinsic properties of presynaptic cell or network activity. All the mEPSCs, in turn, were recorded in the presence of TTX, which blocks action potential generation and propagation via blockage of TTX-sensitive sodium channels. Thus, mEPSCs are spontaneous in the sense of not being evoked by action potentials sensitive to the myelination status of axons, but rather reflect intrinsic synaptic properties (Pinheiro and Mulle, 2008). We found that the frequency of both sEPSCs and mEPSCs were significantly decreased in Tet1 cKO neurons ( Figure 8A-B, D-E), but the amplitude was comparable between groups for either sEPSCs or mEPSCs ( Figure 8C, F). These observations suggest Tet1-deletion in the OL lineage leads to either a decrease in synaptic density or alterations in presynaptic strength.
To determine what processes may underly the defects in sEPSC or mEPSC frequency, we first performed electron microscopy to look for alterations in synapse numbers. The ultrastructural analysis indicated that synaptic density in P21 M2 cortex was similar between both groups ( Figure   S12A-B). To further examine potential alterations on the presynaptic side, we analyzed the density of puncta positive for the presynaptic marker synaptophysin and the excitatory presynaptic terminal markers vesicular glutamate transporter 1 (vGlut1) and 2 (vGlut2). These markers were significantly decreased in M2 cortex of P21 Tet1 cKO brain (Figures 8G-H and Figure S12C). However, the number of puncta displaying the postsynaptic marker Homer1 did not obviously different between groups (Figures 8H and Figure S12C). Immunoblotting for the pre-and post-synaptic markers was consistent to the histological results ( Figure 8I-J), suggesting a downregulation of presynaptic protein expression in the Tet1 cKO cortex. Therefore, our results indicate that the loss of TET1 in oligodendrocyte lineage cells can cause a disruption of pre-synaptic strength without altering the number of synapses in neurons.

Stage-specific TET1 functions in OL development and remyelination
DNA methylation, which mainly deposits 5mC mark on CpG islands in the genome, enables stable but reversible transcription repression and is critical for mammalian development (Iurlaro et al., 2017;Smith and Meissner, 2013). Defects in the regulation of DNA methylation have been associated with various neurological diseases (Gopalakrishnan et al., 2008;Lu et al., 2013). TET enzymes reverse the methylation actions of DNMTs through oxidizing 5mC into 5hmC and promote DNA demethylation (Ito et al., 2010;Kriaucionis and Heintz, 2009;Tahiliani et al., 2009). Strikingly, we detected a genome-wide change in the DNA demethylation landscape marked by 5hmC during OPC differentiation from neural progenitor cells, suggesting a role of TET-mediated DNA demethylation in regulation of OL lineage progression.
In the present study, we found that TET1, but not TET2 or TET3 ( Figure S2, and personal communication with Dr. P. Casaccia), is critical for OPC proliferation, differentiation and myelination during early animal development, suggesting a unique function of TET1 among the TET family in oligodendrogenesis and subsequent myelinogenesis during development. In contrast to early developmental defects, we found that the developmental myelin deficiencies recovered in adult Tet1 cKO mice. We noticed that the percentage of TET1 and Sox10 double positive cells recovered to ~80% in adult Tet1 cKO mice, therefore we speculate that these OLs escaped from Cre-mediated Tet1 depletion in OPCs stage may gradually expand and replace the mutant cells in later stage to some extent. Additionally, the remyelination capacity after injury was compromised in mature Tet1 OPC-iKO brains, indicating that TET1 remains critical for the myelin regeneration process. Similar results were observed by Dr. P. Casaccia and colleagues using another Tet1 f/f line (personal communication), indicating a stage-dependent function of TET1 in OL myelination and remyelination.
Although the OL differentiation defect was not due to increased apoptosis in the Tet1 cKO mutant brain, we found that OPC cell cycle progression was impaired in Tet1 cKO animals. TET1 has been implicated in regulation of cell-cycle progression. For instance, TET1 is critical to maintain the stability of cyclin B1 and acts as a facilitator of mitotic cell-cycle progression (Chrysanthou et al., 2018;Huang et al., 2013). In line with these studies, we found that a cluster of 5hmC-enriched cell cycle genes, e.g. Ccng2, Cdca3, Ccna2, Ccnb2 and Cdc25b (Bennin et al., 2002;Boutros et al., 2013;Kalaszczynska et al., 2009;Nagi Ayad, 2003;Timofeev et al., 2010), were down-regulated in the Tet1 cKO OPCs, suggesting that TET1-5hmC is also critical for cell cycle progression in OPCs.

Locus-specific alteration of DNA hydroxymethylation landscape in OPCs from Tet1 mutants
Genome-wide mapping of 5hmC from Tet1 cKO animals indicated that TET1 is required for maintaining the proper level of 5hmC in a locus-specific manner during OL differentiation. Despite On the other hand, 5hmC may regulate gene expression through associations with various regulatory elements and processes (Szulwach et al., 2011a;Szulwach et al., 2011b), including histone modification and chromatin configuration at the specific loci during cell growth and differentiation. For example, in ES cells, 5hmC is accumulated at the "bivalent domains" in promoters or enhancers that are decorated with dual histone markers, e.g. activating (H3K4me3) and repressive (H3K27me3), for gene transcription (Bernstein et al., 2006;Pastor et al., 2011). Whether and how 5hmCs cooperate with other epigenetic regulators for OPC differentiation remain to be determined.
Intriguingly, ablation of Tet1 also led to upregulation of a set of genes in OPCs, indicating that TET1-mediated 5hmC deposition may also function as a transcription repressor. Consistently, knockdown of Tet1 has been shown to increase a set of targeted genes in ESCs (Williams et al., 2011;Wu et al., 2011). TET1-mediated repression might involve in a recruitment of the MBD3/NuRD repressor complex, which was shown to co-localize with TET1 in ESCs (Yildirim et al., 2011). TET1 may also coordinate with Sin3A co-repressor complex, which has a similar binding profile to TET1 and is required for a subset of TET1-repressed target genes (Neri et al., 2013;Williams et al., 2011).
A recent study indicates that TET1 mediated transcriptional repression could channel through JMJD8 demethylase transcriptional repressor and is independent of TET1 catalytic activity during epiblast differentiation (Khoueiry et al., 2017). Therefore, the mechanisms of locus-specific transcriptional regulation by TET1 during OL development remain to be further defined.

TET1-mediated Ca 2+ homeostasis is critical for oligodendrocyte differentiation
"Calcium signaling" is critical for the response of glial cells to multiple extracellular stimuli. For instance, blocking of the voltage-operated Ca 2+ entry in OPCs inhibits their maturation and the myelin formation ability (Cheli et al., 2015). Similarly, an increase of the resting intracellular [Ca 2+ ] through membrane depolarization could facilitate MBP synthesis in OPCs (Friess et al., 2016).
We found that Tet1 deletion led to the downregulation of multiple calcium transporter genes in OPCs. Treatment of calcium channel agonist significantly restored the differentiation defects in Tet1dificient OPCs. Ca 2+ transients in OLs have been shown to regulate retraction and elongation of the developing myelin sheath (Baraban et al., 2018;Krasnow et al., 2018). Our data indicated that the [Ca 2+ ]i oscillations from both extracellular influx and intracellular stores were greatly reduced in OPCs isolated from Tet1 cKO mice, and that inhibition of VOCC activation impaired OPC differentiation.
These observations suggest that TET1-regulated calcium signaling and transport is important for OL differentiation.
Calcium homeostasis can be another layer of regulation for OPC cell cycle. Silencing Cav1.2 calcium channels mediating voltage-gated Ca 2+ influx was shown to extend cell cycle phases in OPCs to block differentiation (Cheli et al., 2015). In line with the observation, we showed that activating Ca 2+ transportation with the calcium channel agonist Bay K 8644 reduced the number of proliferating OPCs. Taken together, the reduced expression of cell cycle regulators and Ca 2+ ion channels contribute to OPC proliferation defects in developing Tet1 cKO animals.

TET1-deficient OL lineage cells alter neuronal properties in a non-cell autonomous manner
Our results revealed a downregulation of a set of genes encoding neuronal regulatory factors, including GDNF, Reelin and Tenascin R, in Tet1 cKO OPCs. GDNF can regulate neuronal synaptogenesis, especially the clustering of vesicular proteins and neurotransmitter transporters in presynaptic differentiation (Ledda et al., 2007). GDNF infusion has been shown to improve Parkinson's diseases potentially via strengthening of presynaptic synapses or numbers (Kirkeby and Barker, 2019;Patel et al., 2013). Consistent with its role in glia-neuron crosstalk, an electrophysiology study revealed that Tet1 cKO brain exhibited an impaired postsynaptic transduction in excitatory neurons. The reduced frequency of EPSC in Tet1 cKO combined with both the lack of any effect on synapse densities and the reduced expression of presynaptic components suggests that the ablation of Tet1 in OL lineage cells may lead to defects in presynaptic release.
These observations suggest that ablation of Tet1 in OL lineage cells could lead to a defect in synaptic functions through a non-cell autonomous manner. This is in keeping with the observation that loss of methyl-CpG-binding protein 2 (MeCP2) in glial cells including oligodendrocytes impair neuronal properties and cause neuropathology in Rett syndrome (Ballas et al., 2009;Nguyen et al., 2013).
Altogether, our studies reveal critical epigenetic functions for TET1-5hmC in the regulation of stage-specific OL development and remyelination. We also identify a novel TET-mediated function in OL homeostasis to maintain proper communication between OLs and neurons as well as neuronal properties and connectivity. For immunohistochemistry, cryosections (14μm) of brains or spinal cords were short-fixed 30 min in 4% paraformaldehyde and processed for antigen retrieval. Sections were treated with 10mM sodium citrate (pH6.0) at ~90°C for 10min in a microwave and cooled down in room temperature. Then sections were washed three times in PBS, blocked in 3% BSA with 0.03% Triton X-100 (blocking buffer) for 1 hour (hr) at RT, and incubated with primary antibodies in blocking buffer overnight at RT. Next day, sections were washed three times in PBS, incubated with secondary antibodies at RT for 2 hr, and then counter stained with DAPI for 5min. Finally, sections were washed three times in PBS and mounted. Images were taken on Olympus FV1200 Confocal microscope. We For BrdU pulse labeling, animals were injected intraperitoneally with 100 mg BrdU/kg body weight 2 hr prior to sacrifice. For the staining of BrdU, as well as 5hmC, before permeabilization, sections were subjected to DNA denaturation with 2N hydrochloric acid at 37 °C for 20 min and then neutralized with 0.1 M sodium borate at pH 8.5 for 2X 10 min. The G3G4 monoclonal antibody (anti BrdUrd) was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa.
Histological myelin staining was performed with Black Gold II Kit (Millipore, AG105) according to the user manual.
For electron microscopy, tissues were dissected and fixed in 2% glutaraldehyde and 4% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.2) for 24 hr and processed as previously described (Zhao et al., 2010). Ultrathin sections were examined under the JEM-1230 electron microscope (JEOL LTD, Tokyo, Japan) equipped with CCD camera and its application software (832 SC1000, Gatan, Warrendale, PA).

RNA Extraction and qRT-PCR
Total RNAs were purified from tissues or cell cultures using TRIzol reagent according to the manufacturer's instruction (Invitrogen). For qRT-PCR, RNA was transcribed to cDNA with the PrimeScript ™ RT reagent Kit (Perfect Real Time, Takara) and reactions were performed with SYBR ® Premix Ex Taq ™ (Takara) in CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Relative gene expression was normalized to internal control β-actin. Primer sequences for SybrGreen probes of target genes are listed below.

Culture of OL precursor cells, immunocytochemistry and ELISA assay
Mouse OPCs were isolated from P6 cortices of mice by immunopanning with antibodies against Ran-2, GalC and O4 sequentially as previously described (Chan et al., 2004). Briefly, cerebral hemispheres were diced and digested with papain at 37°C. Following gentle trituration, cells were resuspended in a panning buffer containing insulin (5 µg/ml) and then incubated at room temperature sequentially on three immunopanning dishes: Ran-2, anti-GalC, and O4. O4 + GalC − OPCs were released from the final panning dish with trypsin (Sigma). For immunocytochemistry, cell cultures were fixed in 4% PFA. After TritonX-100 permeabilization for 15 min, samples were incubated with primary antibody for 1 h at room temperature followed by fluorescent secondary antibody for another hour. Cells were then counter-stained with DAPI and visualized with Confocal microscope. Experiments were replicated using cells from three different primary cultures.
GDNF ELISA (enzyme-linked immunosorbent assay) was performed using commercial kits (EK 0935, Boster Biological Technology, Wuhan, China) following the manufacturer's instructions. Culture supernatant from 10 5 /cm 2 OPCs were collected after 24hr incubation and sterilized by filtering through 0.22 μm filters. Synthetic mouse GDNF peptide was used to generate a standard curve for each experiment. The plates were developed using the TMB substrate, and the reaction was stopped by addition of equal volume of STOP buffer. The results were read using Infinite M200 Pro colorimetric plate reader (TECAN, Austria).
Flow cytometric analysis of cell cycle with propidium iodide DNA staining PI staining for flow cytometry was performed according to the user manual of DNA Content Quantitation Assay (Cell Cycle) from Solarbio (#CA1510). Briefly, OPCs from control or Tet1 cKO mice were harvested, washed in PBS and fixed in cold 70% ethanol for 30 min at 4°C. After wash twice in PBS, cells were treated with RNase and then stained with PI. With guava easyCyte6HT (Millipore), the forward scatter (FS) and side scatter (SS) were measured to identify single cells. For analysis, ModFit LT software was used make the PI histogram plot. Experiments were replicated three times.

Lysolecithin-induced demyelinating injury
Lysolecithin-induced demyelination was carried out in the corpus callosum of 8-week-old mice. Anesthesia was induced and maintained by peritoneal injection of a mixture of ketamine (90 mg/kg) and xylazine (10 mg/kg). The skull was exposed, and a hole was cut into the cranium. Focal demyelinating lesions were induced by stereotaxic injection of 0.8 μl 1%lysolecithin solution (L-alysophosphatidylcholine, Sigma L4129) into the corpus callosum at coordinates: 0.8 mm lateral, 0.8 mm rostral to bregma, 1.2mm deep to brain surface) using a glass-capillary connected to a 10 μl Hamilton syringe. Animals were left to recover in a warm chamber before being returned into their housing cages. LPC-induced injuries were conducted in a genotype-blinded manner.

Electrophysiology
Analyzing the compound action potential was performed according to previous protocols (Devaux and Gow, 2008;Evans et al., 2010). Tet1 cKO and control littermates were killed by cervical dislocation and then decapitated. Optic nerves were dissected free and cut between the orbit and the optic chiasm in the standard artificial cerebrospinal fluid (aCSF) containing (in mM): NaCl 126, KCl 3.0, CaCl2 2.0, MgCl2 2.0, NaH2PO4 1.2, NaHCO3 26 and glucose 10 at 37℃. Following dissection, optic nerves were equilibrated in aCSF for at least 30 min with constant aeration (95% O2/5% CO2). Then the nerve was gently placed in a customized perfusion chamber, maintained at 37℃ and perfused with aCSF at 2-3ml/min speed. Suction electrodes back-filled with aCSF were used for stimulation and recording. One electrode was attached to the rostral end of the nerve for stimulation and the second suction electrode was attached to the caudal end of the nerve to record the CAP, thus all recordings were orthodromic. Stimulus pulse strength (100µs duration, SS-201J Isolator constant current output with SEN-7203 stimulator, Nihon Kohden, Japan) was adjusted to evoke the maximum CAP possible and then increased another 25% (i.e. supramaximal stimulation). During an experiment, the supra maximal CAP was elicited every 10s and repeat 10 times. The signal was amplified 100×AC membrane potential (100mV/mV) by a Multiclamp700B amplifier, filtered at 10 kHz and acquired at 10 kHz (Digidata 1322A, Molecular Devices, USA). The average CAP amplitude and area were measured in Clampfit 10.0 software (Molecular Devices, USA) offline and performed blind to genotype. Image drawing and statistical analysis were performed in GraphPad Prism 6.
Whole-cell patch clamp recordings were conducted as previously described (Wang et al., 2017). Briefly, coronal slices (~300μm) were prepared from P21 mice with VT1200s vibrotome (Leica, Germany) and incubated at 32℃ with carbogenated aCSF for 10 to 15 minutes. For all the synaptic activity measurements, the slices were constantly perfused with carbogenated aCSF at room temperature. Whole-cell patch clamp recordings were performed in the layer 2/3 neurons of the M2 region verified with BX51WI microscope (Olympus, Japan) with infrared differential interference contrast (IR-DIC) visual guide. Neurons were held at a membrane potential of −70 mV with a Muticlamp 700B amplifier and characterized by injection of rectangular voltage pulse (5 mV, 50 ms) to monitor the whole-cell membrane capacitance, series resistance and membrane resistance. Pipettes were filled with the internal solution containing the following: 110 CsOH (50% wt), d-gluconic acid (49%-53% wt), 4 mM NaCl, 15 mM KCl, 5 mM TEA-Cl, 20 mM HEPES, 0.2 mM EGTA, 5 mM lidocaine N-ethyl chloride, 4 mM ATP magnesium salt, and 0.3 mM GTP sodium salt. pH was adjusted to 7.2-7.4 with CsOH, and osmolarity was adjusted to 298-300 mOsm with K2SO4 (approximately 8-10 mM). Cells were excluded from the study if the series resistance (Rs) changed by more than 20%. In addition, cells with Rs of more than 20 MΩ at any time during the recordings were discarded. sEPSC or mEPSCs were recorded 5 minutes after entering whole-cell patch-clamp recording mode, without or with pre-addition of 1 μM tetrodotoxin (TTX) and 100 μM picrotoxin to the aCSF. Signals were low-pass filtered at 2 kHz and sampled at 10 kHz with a Digidata 1322A. Data analysis was conducted in Clampfit 10.0 software and performed blind to genotype.

RNA-Seq and Data Analysis
RNA-seq assays were performed by RiboBio Co., Ltd. (Guangzhou, China). Briefly, libraries were prepared using Illumina RNA-Seq Preparation Kit (TruSeq RNA Sample Prep Kit) and sequenced by HiSeq 3000 sequencer. RNA-seq reads were mapped using TopHat with settings of "read mismatches=2" and "read gap length=2" (http://ccb.jhu.edu/software/tophat/ index.shtml). TopHat output data were then analyzed by DEGseq to compare the changes of gene expression between Tet1 cKO and control, based on the calculate RPKM values for known transcripts in mouse genome reference. Heatmap of gene differential expression was generated using R Package (http://www.r-project.org).

hMeDIP-Sequencing analysis
hMeDIP Sequencing service was provided by KangChen Bio-tech (Shanghai, China). hMeDIP-Sequencing library preparation was performed according to a previous study (Down et al., 2008) with minor modifications. Genomic DNA was sonicated to ~200-800bp with a Bioruptor sonicator (Diagenode). 800 ng of sonicated DNA was end-repaired, A-tailed, and ligated to single-end adapters following the standard Illumina genomic DNA protocol (FC-102-1002, Illumina). After agarose sizeselection to remove unligated adapters, the adaptor-ligated DNA was used for immunoprecipitation (IP) with a mouse monoclonal anti-5-hydroxymethylcytosine antibody (Diagenode, C15200200). For this, DNA was heat-denatured at 94°C for 10 min, rapidly cooled on ice, and immunoprecipitated with 1 μL primary antibody overnight at 4°C with rocking agitation in 400 μL IP buffer (0.5% BSA in PBS). To recover the immunoprecipitated DNA fragments, 20 μL of magnetic beads were added and incubated for an additional 2 hours at 4°C with agitation. After IP, a total of five washes were performed with ice-cold IP buffer. Washed beads were resuspended in TE buffer with 0.25% SDS and 0.25 mg/mL proteinase K for 2 hours at 65°C and then allowed to cool down to room temperature. DNA was then purified using Qiagen MinElute columns and eluted in 16 μL EB (Qiagen). 14 cycles of PCR were performed on 5 μL of the immunoprecipitated DNA using the singleend Illumina PCR primers. The resulting products were purified with Qiagen MinElute columns, after which a final size selection (300-1,000 bp) was performed by electrophoresis in 2% agarose. Libraries were quality controlled by Agilent 2100 Bioanalyzer.
For sequencing, the library was denatured with 0.1 M NaOH to generate single-stranded DNA molecules and loaded onto channels of the flow cell at 8 pM concentration, amplified in situ using TruSeq Rapid SRCluster Kit Illumina). Sequencing was carried out by running 150 cycles on Illumina HiSeq 2500 using TruSeq Rapid SBS Kit Illumina) according to the manufacturer's instructions.
After sequencing images generated, the stages of image analysis and base calling were performed using Off-Line Base caller software (OLB V1.8). After passing Solexa CHASTITY quality filter, the clean reads were aligned to Mus_musculus genome (UCSC mm10) using BOWTIE software (V2.1.0). Aligned reads were used for peak calling, both mRNA and LncRNA associated hMeDIP enriched regions (peaks) with statistically significant were identified for each sample, using a q-value threshold of 10 -4 by MACS v2. Both mRNA and LncRNA associated hMeDIP enriched regions (peaks) were annotated by the nearest gene using the newest UCSC RefSeq database. Differentially hydroxymethylated regions (DhMRs) between two groups with statistically significant were identified by diffReps (Cut-off: log2FC=1.0, p-value=10 -4 ). DhMRs were annotated by the nearest gene using the UCSC RefSeq and database of multiple databases integration.

[Ca 2+ ]i imaging
Fluo4 was used to measure [Ca 2+ ]i changes. To load cells with Ca 2+ probe, cultures were incubated in standard artificial cerebrospinal fluid (aCSF) containing (in mM): NaCl 125, KCl 3.0, CaCl2 2.0, MgCl2 2.0, NaH2PO4 1.25, NaHCO3 26 and glucose 20, supplemented with 0.03% Pluronic F-127 and 0.6μM cell-permeable form of indicator (Fluo4-AM, Invitrogen) for 20 min at 37 . After loading, cells were transferred into recording chamber on the stage of Olympus FV1000 confocal inverted microscope equipped with phase contrast optics. Measurements started after at least 15 min storage in aCSF to ensure deesterification of indicators. Fluo4 was excited with 488 nm laser and emitted light was collected at 515 nm. A series of sections were collected every 500 ms, at 500 ms intervals, for 180 s in total. Pharmacological agents were directly added in aCSF: ATP (Sigma, 100μM) or Bay K 8644 (10 μM), 30s after the first image was taken. The images were analyzed using the Olympus FV10-ASW 4.1 software. Cell bodies were selected as regions of interest (ROI) and normalized changes of Fluo4 fluorescence intensities were calculated as ΔF/F=(F−F0)/F0 (F, fluorescence intensity; F0, baseline intensity). Data were expressed as Mean ± standard error of the mean (SEM), 'N' represents the number of responding cells. Experiments were performed 3 times from different cultures and the results were pooled together for analysis. N>100 in ATP treatment group and N>70 in Bay K 8644 group. Significance was determined by unpaired t tests, using Prism 6.0 (Graphpad).

Statistical analysis
Numerical values were analyzed using Mean ± SEM and are presented as bar graphs. Group with equal variances and normal distribution were compared using Student's t test (unpaired) with SPSS 21.0. One-Way ANOVA with repeated-measures and Two-Way ANOVA were applied to the data when appropriate. Significance is denoted as *p<0.05, **p<0.01 or ***p<0.001 in the figures.

Data availability
The transcriptome and DNA hydroxymethylation profiling dataset have been deposited in the Gene Expression Omnibus database (GSE122838).  (H) Quantitative real time PCR identified the reduced expression of Tet1 mRNA in OPCs purified from Tet1 cKO mice brain. Data represent average transcript levels relative to control, after normalization±SEM for n=3 independent experiments each performed in triplicates. ***, p<0.001 compared to control by Student's t test.
(I-J) The number of mature oligodendrocytes was significantly decreased in Tet1 cKO mice after birth and recovered in adult mice. Representative images of CC1 and Sox10 immunostaining in the corpus callosum of control and Tet1 cKO groups at P8 and P27 (I). Quantification of CC1 positive cells (J) in P8, P21, P27 and 2M stages, respectively. Notably, in 2M old mice, the number of CC1 + cells were comparable between groups. *, p<0.05 compared to control by Student's t test (n=3 animals for control and n=3 to 5 for mutant group). Scale bar in I, 50μm. (K) Immunostaining of MBP in Tet1 cKO and control mice at P16. Note the decreased staining intensity of MBP in cortex and corpus collosum of Tet1 cKO mice. Scale bar, 100μm. (E) Scatter plot of the G ratio from P14 optic nerves with axonal perimeter on the x-axis and G-ratio on the y-axis. Increased G-ratio was observed in Tet1 cKO mice. (F-L) Ultrastructure of myelin in corpus collosum from Tet1 cKO mice. Representative images of electron micrographs of corpus collosum from control and Tet1 cKO mice at P14, P27 and 2M (F). Scar bar, 0.5 μm. Quantification the number of myelinated axons in defined area from P14 (G), P27 (I) and 2M (K) control and Tet1 cKO mice. Data are mean ± SEM (n=3 animals each group). ***, p<0.001 compared to control by Student's t test. Scatter plot showing the G ratio in P14 (H), P27 (J) and 2M (L) mice, with axonal perimeter on the x-axis and G-ratio on the y-axis.     Representative serial images after addition of 10 μM Bay K 8644 in OPCs from control or Tet1 cKO group (A). Note the delayed responding and weaker response in Tet1 cKO groups. Representative traces of Fluo-4 intensity in OPCs from control and Tet1 cKO mice, following application of agonist and antagonist for L-type voltage gated channel (Bay K8644 and Verapamil, respectively) (B). The average amplitude changes from Bay K 8644 stimulus was reduced in Tet1 cKO groups. Data are mean ± SEM (n=82-87 cells, from three independent experiments) (C). The average value of Fluo-4 intensity rise slope after Bay K 8644 addition was reduced in Tet1 cKO groups. Data are mean ± SEM (n=82-87 cells, from three independent cultures) (D). (E-G) [Ca 2+ ]i oscillations following the treatment with ATP in OPCs from control and Tet1 cKO mice. Representative traces of Fluo-4 intensity of OPCs from control and Tet1 cKO mice, following application of ATP and antagonist for P2Y metabotropic receptor (MRS2179) (E). The average amplitude changes from ATP stimulus was reduced in Tet1 cKO groups. Data are mean ± SEM (n=104-105 cells, from three independent cultures) (F). The average value of Fluo-4 intensity rise slope after ATP addition was reduced in Tet1 cKO groups. Data are mean ± SEM (n=104-105 cells, from three independent cultures) (G). *, p<0.05, **, p<0.01, ***, p<0.001, by Student's t test.
(H-J) Bay K 8644 treatment promoted OL differentiation in vitro and rescue the deficiency of OL differentiation from Tet1 cKO mice. Representative images of MBP immunostaining, indicating more differentiated OL in Bay K 8644 treated cell cultures in both groups (H). Quantification the percentage of MBP + cells among Olig2 + cells after 3 days application of Bay K 8644. *, p<0.05 (I). Scar bar in H, 100 μm. Quantitative real-time PCR revealed the increased mRNA expression of myelin genes, Cnp, Mbp and Plp, with daily application of Bay K 8644 in OL cultures from both control and Tet1 cKO groups, compared to blank in each group (J). Data represent average transcript levels relative to control, after normalization as mean ± SEM for three independent experiments each performed in triplicates. **, p<0.01, *, p<0.05 compared to control by two-way ANOVA. (K-L) BrdU incorporation assay revealed fewer proliferating OPCs after application of 10 μM Bay K 8644. Representative images of BrdU and Olig2 staining for OPC cultures (K). Quantification the percentage of BrdU + among Olig2 + cells. Data are mean ± SEM (n=3 different cultures each group) (L). *, p<0.05 and amplitude (C) of sEPSCs were calculated (n = 9 cells from two control and Tet1 cKO mice, respectively). Error bars represent mean ± SEM. ***, p < 0.001 significance based on Student's t test.
(D-F) Recordings of miniature EPSCs (mEPSCs) from the cortical pyramidal neurons of control and Tet1 cKO cortex at P22. Representative recording traces of mEPSCs (D). Average frequency (E) and amplitude (F) of mEPSCs were calculated (n = 10 cells from two control mice and n=12 cells from two Tet1 cKO mice, respectively.) Data represent mean ± SEM. **, p < 0.01 significance based on Student's t test.
(G-H) Expression of presynaptic markers in P21 Tet1 cKO mice. Representative images of immunostaining for Synaptophysin-1 and excitatory presynaptic terminal marker, vGluT1, in M2 cortex of control and Tet1 cKO mice. Neurons were co-labeled with MAP2. Scale bar, 10 μm (G). Quantification the puncta staining of pre-and post-synaptic markers using Imaris software. Data represent mean ± SEM. **, p < 0.01 significance based on Student's t test (H). (I) Expression of Synaptophysin, vGluT1, Homer1, PSD95 in control (C) and Tet1 mutant (M) cortex was analyzed by Western blot. β-Tubulin was used as loading control. (J) Histogram showed fold changes of synaptic markers measured by densitometry in Tet1 mutant against control after normalization with β-Tubulin. *, p<0.05 compared to control by Student's t test (n=3 animals each group).

Supplementary figure legends:
Figure S1. Examples OL lineage specific genes that showed exclusive 5hmC peaks in promoter-TSS region in OPC (A) Column diagrams show the expression of representative genes in different OL stage. Genes with OPC-specific 5hmC peak that are exclusively expressed in OL lineage. The diagrams are downloaded from http://web.stanford.edu/group/barres_lab/brain_ rnaseq.html. (B) Snapshots of 5hmC profiles for representative genes from NPC and OPC. Genomic region encompassing promoter-TSS were compared between groups. Genes highly expressed in OPC, immature OL (iOL), mature OL (OL) were shown. Cspg (chondroitin sulfate proteoglycan 4), Slc22a3 (solute carrier family 22 member 3), Atp2b3 (ATPase plasma membrane Ca 2+ transporting 3), Tmem141 (transmembrane protein 141 ), Kndc1 (kinase non-catalytic C-lobe domain containing 1), Ptprd (protein tyrosine phosphatase, receptor type D ), Mag (myelin-associated glycoprotein ), Cdc42ep2 (CDC42 effector protein 2), Elovl7 (elongation of very long chain fatty acids protein 7).   (E-F) Loss of TET1 results in less OLs from embryonic stage. Representative images of Olig2 staining at E15.5 from both groups (E). Scale bar, 100 μm. Quantification of Olig2 positive cells in cortex from E13.5 and E15.5 mice (F). *, p<0.05 compared to control by Student's t test (n=3 animals each group).  (B) Evoked CAP amplitudes of the 2 nd peak (maximal) from control and Tet1 cKO mice were plotted versus stimulus currents. (C) Total CAP area vs. stimulus current illustrating the significant decrease of CAP area in Tet1 cKO mice at all stimulus intensity compared to control. Data are mean±SEM (n=10 nerves each group). Figure S7. Normal apoptosis and more proliferating OL precursor cells in Tet1 cKO brain (A-B) No alterations in the apoptosis of OL lineage cells were observed in Tet1 cKO mice as revealed by TUNEL/Olig2 double staining. Olig2 immunostaining after TUNEL staining revealed similar number of apoptotic cells between two groups at P1 and the double labeling cells can be hardly detected (A). Quantification of TUNEL positive cells from E14.5 to P1 stage revealed no overall difference between two groups. Data are mean±SEM (n=3 animals each group) (B). Scale bar in A, 50 μm. (C-D) More proliferating OPCs was observed in Tet1 cKO mice. Representative images of Ki67/Sox10 double immunostaining in different regions as indicated from P1 control or Tet1 cKO mice (C). Quantification of the proportion of Ki67 + cells within Sox10 + population. Data are mean ± SEM (n=3 animals each group) (D). *, p<0.05 compared to control by Student's t test. Scale bar in C, 100 μm. (A-B) Normal neuronal development in Tet1 cKO mice. A Western blot assay revealed that the expression of DCX, marker for newly generated neuron, in Tet1 cKO embryonic brain was comparable to that of control at different developmental stages. Tubulin was included as loading control (A). Immunostaining of NeuN in P1 brain showed similar expression in both groups (B). Scar bar, 100 μm. (C-D) Astrocytes identified by ALDHL1 immunostaining in P1 brain showed similar number in defined area between control and Tet1 cKO mice. Immunostaining for ALDHL1 in P1 brain S1 cortex (C). Quantification of ALDHL1 + cells revealed no difference between two groups. Data are mean ± SEM (n=3 animals each group) (D). Scale bar in C, 100 μm. (E) Western blot assay for the expression of ALDHL1 and NeuN, markers for astrocyte and neuron respectively, in the brain samples from postnatal control (C) and Tet1 cKO (M) mice. β-actin was used as loading control.