TET1 Modulates H4K16 Acetylation by Interacting with hMOF to Regulate Expression of DNA Repair Genes and Oncogenic Transformation

The Ten Eleven Translocation 1 (TET1) protein is a DNA demethylase that regulates gene expression through alteration of DNA methylation. Recent studies have demonstrated that TET1 could modulate transcriptional expression independent of its DNA demethylation activity; however, the detailed mechanisms underlying TET1’s role in such transcriptional regulation remain not well understood. Here, we uncovered that Tet1 formed a chromatin complex with histone acetyltransferase Mof and scaffold protein Sin3a in mouse embryonic stem cells by integrative genomic analysis using publicly available ChIP-seq data sets. Specifically, the TET1/SIN3A/hMOF complex mediates acetylation of histone H4 at lysine 16, via facilitating the binding of hMOF on chromatin, to regulate expression of important DNA repair genes in DNA double strand breaks, including TP53BP1, RAD50, RAD51, and BRCA1, for homologous recombination and non-homologous end joining repairs. Under hydrogen peroxide-induced DNA damage, dissociation of TET1 and hMOF from chromatin, concurrent with increased binding of SIRT1 on chromatin, led to hypo-acetylation of H4K16, reduced expression of these DNA repair genes, and DNA repair defects in a DNA methylation independent manner. A similar epigenetic dynamic alteration was also observed in H-RASV12 oncogenic-transformed cells, supporting the notion that suppression of TET1 downregulates DNA repair genes through modifying H4K16ac, instead of its demethylation function, and therefore contribute to tumorigenesis. Taken together, our results suggested a mechanistic link between a novel TET1 complex and H4K16ac, DNA repair genes expression, and genomic instability.


INTRODUCTION
The Ten Eleven Translocation 1 (TET1) protein, a member of TET family, is a key player in DNA demethylation (Veron and Peters 2011). However, a recent study revealed that Tet1, in addition to its transcriptional regulatory function through its catalytic activity in DNA demethylation, possesses both activation and repressor functions in the regulation of a certain subset of genes in mouse embryonic stem cells (mESCs) (Williams et al. 2011). This observation was further supported by a study in which changes of transcriptional expression induced by overexpression of TET1 were highly similar to those induced by its demethylation-enzymatically-dead mutant in differentiated cell lines, suggesting that TET1 mainly regulates gene expression through a DNA methylation independent manner (Jin et al. 2014). The repressive role of TET1 in transcriptional regulation has been proposed to derive from its interaction with polycomb repressive complex 2 (PRC2) to form a histone modifying complex, thereby modifying chromatin repressive mark (H3K27me3) in mESCs (Wu et al. 2011).
However, the interaction between TET1 and PRC2 complex is, so far, exclusively presented in embryonic stem cells (ESCs), but not in differentiated cells such as fibroblasts and HEK293T cells (Neri et al. 2013), indicating that TET1/PRC2 complex may act to repress gene expression in an ESCs-specific manner. On the other hand, SIN3A (homolog of Sin3 in yeast), a key component in multiple regulatory complexes, is involved in both transcriptional repression and activation through recruitment of diverse transcriptional factors or chromatin remodeling machinery at target promoters (Kadamb et al. 2013;Solaimani et al. 2014). A recent study has shown that TET1 interacts with SIN3A in both mESCs and HEK293T cells and presents highly overlapping binding profile on genome-wide (Williams et al. 2011), implying TET1 may associate with SIN3A to regulate gene expression in both ESCs and differentiated cells. However, the exact mechanisms underlying the functional nature of TET1 and its associated protein complexes in regulating its target gene expression remain to be unveiled.
Recently, it was demonstrated that there are dysfunctional DNA repair mechanisms and increased mutation frequencies in TET1-deficient non-Hodgkin B cell lymphoma (B-NHL), indicating that TET1 function as a tumor suppressor (Cimmino et al. 2015). This observation, in line with a previous study in which there were decreased foci of MLH1 and delayed removal of RAD51 in mouse Tet1-knockout primordial germ cells (Yamaguchi et al. 2012), indicates that TET1 plays an important role in DNA repair in mammalian cells. However, the underlying mechanisms of TET1 functions in DNA repair in response to DSBs are unclear.
Homologous recombination repair (HRR) and non-homologous end joining (NHEJ) are two categories of DNA repair pathway in response to DNA double strand breaks (DSBs). Some DNA repair genes, such as RAD50, BRCA1, RAD51, and TP53BP1, act as tumor suppressors and are frequently mutated or aberrantly downregulated in human cancers, resulting in impairments of DNA repair in response to DSBs, which is recognized as one of the hallmarks of tumorigenesis (Hanahan and Weinberg. 2011 ;Negrini et al. 2010). Whole Genome Bisulfate Sequencing (WGBS) data analysis in the Tet1-deficient primordial germ cells showed that the methylation levels of most DNA repair genes had no obvious alteration (Yamaguchi et al. 2012), indicating that Tet1 possibly affects expression of DNA repair genes through a mechanism independent of its DNA demethylation function.
H4K16ac is a well-known targeted epigenetic site for post-translational modification in transcriptional activation (Taylor et al. 2013). Human MOF (hMOF, also known as KAT8), a member of the MYST (Moz-Ybf2/Sas3-Sas2-Tip60) family of HATs, specifically modifies H4K16ac and is frequently downregulated in various types of cancers, including medullo-blastomas, breast carcinomas, colorectal carcinoma, gastric cancer, and renal cell carcinoma (Cao et al. 2014;Pfister et al. 2008). Studies have shown that depletion of hMOF renders both a global reduction of H4K16ac and DNA repair defects in budding yeast and mammal cells (Li et al. ;Sharma et al. 2010).
In addition, overexpression of hMOF reverses silencing of certain tumor suppressor genes induced by H4K16 deacetylation (Kapoor-Vazirani et al. 2008). Conversely, SIRT1 has the ability to deacetylate H4K16ac (Vaquero et al. 2007；Mishra et al. 2014, and is required for DNA repair and genomic stability in both yeast and mammals (Uhl et al. 2010;Boulton and Jackson 1998). Noteworthy, an elevated SIRT1 expression has been observed in a variety of human cancers relative to their non-transformed counterparts, including leukemia, glioblastoma, prostate, colorectal, and skin cancers (Chen et al. 2005;Huffman et al. 2007;Liu et al. 2009). Importantly, exogenous expression of SIRT1 reverses the effects of hMOF on H4K16ac and sensitization to the topoisomerase II inhibitor of cancer cells (Hajji et al. 2010), implying H4K16ac is dynamically modulated by both hMOF and SIRT1.
In this study, we first revealed, through integrative genomic analysis using publicly available ChIP-seq data sets, that significantly overlapped distribution of TET1, Sin3a, Mof, and H4K16ac were observed in mESCs. We further demonstrated that TET1, hMOF, and SIN3A interact with each other by in vitro biochemical studies in human cell lines. We next showed that the identified TET1 chromatin complex specifically modulates H4K16ac. Finally, we uncovered, under DNA damage and oncogenic-induced transformation, that dynamic recomposition of these TET1/SIN3A/hMOF chromatin complex components could cause hypo-acetylation of H4K16, thereby impairing DNA repair and ultimately involving in tumorigenesis in a DNA methylation independent manner.

Integrative genomic analysis reveals similar binding patterns between Tet1, Mof, and H4K16ac in mESCs
Previous studies have generated a considerable number of ChIP-seq data sets of DNA binding proteins (DBPs) from mESCs, which is available in GEO and ENCODE databases. We retrieved all 219 available ChIP-seq data sets corresponding to 103 different DBPs in mESCs to investigate the co-occupancy between Tet1 and the rest of the DBPs (Supplementary flanking region (defined as promoter) (Fig. 1a). Next, the same correlation analysis of each component in this sub-cluster and six available histone modifications in mESCs showed that Tet1, with other five DBPs (Kdm2a, Mof, Dpy30, Sin3a, and Lsd1), were closely related to H4K16ac, H3K4me3, H3K9ac, and H3K27ac (Fig. 1b). Given that Mof is the only histone acetyltransferase in the sub-cluster, whereas Kdm2a, Lsd1, and Dpy30 are either histone demethylases or histone methyltransferase complex component, we decided to focus on the investigation of the co-efficiency among the three histone acetylation marks and Tet1, Mof, and Sin3a by correlation analysis. Our results revealed that Tet1, Sin3a, Mof, and H4K16ac were clustered together with the highest correlation coefficients (Fig. 1c). Furthermore, ChIP-seq signals enrichment analysis revealed that Tet1, Sin3a, Mof, and H4K16ac had highly overlapping distribution patterns around promoter regions ( Fig. 1d and Supplementary Fig. 1).
Taken together, these observations imply that TET1, SIN3A, hMOF have highly similar binding patterns with H4K16ac at the genomic level.
In order to determine whether TET1/SIN3A/hMOF was a complex distinguished from either PRC2 complex or SIN3A/HDAC complex and targeted different chromatin marks, we investigated the binding profiles of the major components of these three complexes, including Suz12, Ezh2, Sin3a, Tet1, Mof, Hdac1, and Hdac2, as well as their associated histone marks H3K27me3 and H4K16ac. Initially, we determined binding states by dividing promoter regions from ChIP-seq data set of each of above proteins and histone marks into clusters M1-M7 using ChromHMM ( Supplementary   Fig. 2a, 2b). Our results showed that Tet1/Sin3a/Mof complex, PRC2 complex, and Sin3a/Hdac1/Hdac2 complex were enriched in cluster M6, M7, and M3, respectively ( Fig. 1e). Remarkably, Tet1, Mof, and Sin3a combined with H4K16ac, commonly enriched in cluster M6, was primarily related to DNA damage and repair associated pathways by KEGG pathway analysis (Fig. 1f). In addition, H4K16ac was enriched in M5, which was significantly associated with DNA repair related biological processes similar to those in M6 (Supplementary Fig. 2c). Intriguingly, we observed most of promoter regions in 177 DNA repair genes, including Brca1, Brca2, Rad51,Trp53, and Mlh1, as shown in Supplementary Fig. 3, were co-occupied with binding of Tet1, Sin3a, Mof, and H4K16ac. In summary, our integrated genomic analysis indicates that Tet1 may form a complex with Mof and Sin3a targeting H4K16ac in mESCs.

TET1 forms a chromatin complex with SIN3A and hMOF to target H4K16ac
To confirm that TET1 formed a chromatin complex with hMOF and SIN3A, we first obtained different fractions of nuclear protein extracts in HEK293T cells separated by size fractionation using sucrose gradient centrifugation. As shown in our data, TET1, hMOF, and SIN3A were simultaneously enriched in pool 3, 4, and 5, suggesting they may be complexed with each other (Fig. 2a). Next, we performed immunoprecipitations (IPs) analysis of chromatin-bound protein after overexpression of Flag-TET1 or HA-SIN3A in HEK293T cells. Our data showed that both Flag-TET1 and HA-SIN3A interacted with hMOF ( Fig. 2b). In order to identify which region of TET1 interacted with hMOF and SIN3A, we constructed three fragment plasmids of TET1, described as Flag-FL1, Flag-FL2, and Flag-FL3, which respectively contained CXXC domain, Cysteine-rich domain and DSBH (double stranded -helix) conserved domain (Fig. 2c). Our Co-IP data showed that hMOF and SIN3A only interacted with Flag-FL3 (Fig. 2d). Consistently, the interactions were confirmed by His-pulldown assays using proteins overexpressed in, and purified from E. coli cells (Fig. 2e).
Taken together, our results suggest that TET1 forms a chromatin complex with hMOF and SIN3A via its C-terminal region.
To determine whether TET1 targeted H4K16ac, we performed IP assays in HEK293T cells co-transfected with Flag-FL and HA-H4K16WT, HA-H4K16Q (a mimic acetylated mutant), or HA-H4K16R (an unacetylated mutant), respectively. Compared with interaction with HA-H4K16WT, Flag-FL had increased interaction with HA-H4K16Q, but decreased association with HA-H4K16R (Fig. 2f). Next, we performed chromatin fractionation and Western blot analysis on chromatin extracts from HEK293T cells with overexpressed HA-H4K16R and HA-H4K16WT, respectively. TET1 binding was significantly decreased in cells overexpressing HA-H4K16R compared with that in cells overexpressing HA-H4K16WT (Fig. 2g). In addition, when HEK293T cells were treated with the HDAC class I and class II inhibitor trichostatin A (TSA), increased TET1 binding was observed (Fig. 2h). Taken together, our data indicates that TET1 preferentially associates with histone H4 bearing K16 acetylation mark.

TET1 depletion causes a significant reduction in H4K16ac levels
Given that TET1 contains CXXC domain, which enable its direct DNA binding, we reason that TET1 may recruit hMOF to genomic loci for regulation of H4K16ac mark.
To test this hypothesis, we analyzed alterations of several histone modifications in TET1-depleted cells. Western blot analysis showed that H4K16 was hypo-acetylated in Tet1-knockout mice embryonic fibroblast (Tet1 -/-MEF) cells compared to in wild type Tet1 +/+ MEFs. We observed a similar hypo-acetylation status in TET1-knockdown HCT116 and HeLa cells, respectively. However, the other histone markers, such as H3K4me2, H3K4me3, and H3K27me3, demonstrated insignificant alterations (Fig.   3a). Immunofluorescence staining further confirmed a significant reduction of H4K16ac in Tet1 -/-MEF cells, TET1-knockdown HCT116, and TET1-knockdown HeLa cells compared to their respective controls. Our results indicated that depletion of TET1 resulted in a significant reduction of H4K16ac level (Fig. 3b). Agree with this observation, the level of chromatin bound hMOF was significantly decreased in TET1-knockdown HEK293T cells (Fig. 4a).
As hypo-acetylation of H4K16 was reported to facilitate 53BP1 recruitment(Hsiao and Mizzen), we also determined the number of 53BP1 foci and found a two-fold increase in the number of 53BP1 foci in Tet1 -/-MEF cells ( Fig. 3c and 3d). Our results indicate that depletion of TET1 results in a significant reduction of H4K16ac levels and promotion the recruitment of 53BP1 binding to chromatin.

TET1 switches binding of hMOF and/or SIRT1 on chromatin to regulate
H4K16ac mark on the promoters of the DNA repair genes Interestingly, using chromatin fractionation we found SIRT1 binding increased in TET1-knockdown HEK293T cells, whereas hMOF binding was decreased (Fig. 4a), indicating that TET1, as a switch, controls bindings of hMOF and SIRT1 to chromatin.
To determine whether TET1 was required for regulation of DNA repair genes through modifying H4K16ac, we measured the mRNA levels of RAD50, BRCA1, RAD51, and TP53BP1 in HA-hMOF overexpression or SIRT1-knockdown HEK393T cells, with or without TET1 knockdown, via RT-qPCR. We found that depletion of TET1 blocked the increased expression of RAD50, BRCA1, RAD51, and TP53BP1 both in hMOF-overexpressing cells (Fig. 4b) and in SIRT1-knockdown cells (Fig. 4d). Next, we further determined the effect of H4K16ac enrichment at the promoter of RAD50, BRCA1, RAD51, and TP53BP1 in HA-hMOF-overexpressing or SIRT1-knockdown HEK293T cells, with or without TET1 knockdown, via ChIP-qPCR. The results showed that depletion of TET1 blocked the increased H4K16ac enriched at the promoter of RAD50, BRCA1, RAD51, and TP53BP1 in both hMOF-overexpressing ( Fig. 4c) and SIRT1-knockdown HEK293T cells (Fig. 4e). These results suggest that TET1, as a switch, inversely modulates the bindings between hMOF and SIRT1 on chromatin, which in turn dynamically controls H4K16ac levels, and ultimately contributes to the expression of these important DNA repair genes.

Impairment of homologous recombination repair and non-homologous end joining in TET1-, but not TET2-or TET3-knockdown cells
To determine the role of TET1 in DNA repair, C57 wild type mice (WT) and Tet1 heterozygous mutated mice (Tet1 +/mice) were subjected to x-ray radiation. The coat-state rating scale results showed that there was a severe deterioration of the coat in Tet1 +/mice compared with WT mice after four months of x-ray radiation (Fig.   5a, 5b), suggesting that Te t1 +/mice had defects in DNA repair mechanisms in response to DSBs. Next, we measured the percentage of DNA present in comet tail and the tail moment in comet assay to determine the extent of DNA damage in Tet1 -/-MEF cells, respectively. Our results showed both of these parameters were increased approximately by two-folds in the Tet1 -/-MEF cells than those in WT cells (Fig. 5c, 5d, and 5e), indicating that Tet1 -/-MEF cells had a higher level of DSBs. We also measured foci formation of DSBs maker H2AX by immunofluorescence staining.
Consistent with our comet assay, we found that the number of H2AX foci increased two-fold in Tet1 -/-MEF cells compared to WT cells (Supplementary Fig4a, 4b). In addition, Western blot analysis indicated that the level of H2AX increased in TET1-knockdown cells, but not in TET2-or TET3-knockdown HEK293T cells ( Supplementary Fig.4c). DAPI staining and statistical analysis further showed that the percentage of micronuclei in Tet1 -/-MEF cells were two-folds higher than that in WT cells (Fig. 5f, 5g). These results indicate that loss of TET1 leads to severe DNA damage and the defects in DNA repair and genomic instability.
Homologous recombination repair and non-homologous end joining are two types of DNA repair mechanism in response to DSBs. To determine the extent of HRR and NHEJ repair frequencies in TET1-depleted cells, we set out to employ two types of GFP reporter systems in HEK293T cells, in which a defective GFP gene is functionally restored to WT cells upon I-SceI transfection (Mao et al. 2008). We found TET1-depletion resulted in 25% decrease of GFP positive cell numbers in HRR reporter assay (Fig. 5h), and 50% reduction in NHEJ reporter assay, and neither of these outcomes was observed in TET2-and TET3-knockdown cells (Fig. 5i). More importantly, we demonstrated that TET1-depletion induced transcriptional repression of the important genes in DNA repair, including RAD50, BRCA1, RAD51, and TP53BP1, via RT-PCR and Western blot analysis (Fig. 5j, 5k). Collectively, our results indicate that depletion of TET1, but not TET2 or TET3, results in the defects of HRR and NHEJ in response to DSBs and genomic instability through downregulation of the DNA repair genes.
Dynamic regulation of TET1, SIN3A, hMOF, and SIRT1 binding on chromatin in response to hydrogen peroxide-induced oxidative damage Next, we investigated alterations in different histone modifications by treating HEK293T cells with DNA damaging reagents including adriamycin (ADR), bleomycin (Bleo), camptothecin (CPT), and hydrogen peroxide (H 2 O 2 ). Our data showed that H 2 O 2 treatment specifically resulted in a significant decrease of H4K16ac levels, whereas H3K27me3, H3K4me2, and H3K4me3 had no obvious changes in response to DNA damage reagents (Fig. 6a). In addition, we measured the bindings of TET1, hMOF1, SIRT1, and SIN3A on chromatin after DNA damage reagent treatment as described above. We found that H 2 O 2 treatment caused a significant decrease of both TET1 and hMOF binding, and an increase in SIRT1 binding, but no obvious alteration of SIN3A binding (Fig. 6b). To support this notion, H 2 O 2 treatment caused significantly decreased interactions between Flag-FL3 with hMOF, but not with SIN3A (Fig. 6c). As shown in Fig. 6d, H 2 O 2 treatment also led to transcriptional repression of the important genes in DNA repair including RAD50, BRCA1, RAD51, and TP53BP1, while the DNA methylation level, as assessed by profiling using reduce representation bisulfite sequencing (RRBS), did not present alteration in their corresponding promoters (Fig. 6e, 6f). ChIP-qPCR assay further revealed that TET1 and H4K16ac enriched together at promoter regions of these DNA repair genes and displayed decreased enrichment after H 2 O 2 treatment (Fig. 6g). Taken together, our results suggest that oxidative damage induces dynamic recomposition of TET1, hMOF, and SIRT1 binding on chromatin, which is concurrent with the deacetylation of H4K16 and reduced expression of the important DNA repair genes.

TET1 binding on chromatin is dependent on ataxia telangiectasia mutated (ATM) protein in HEK293T cells
Our data also showed an increased TET1 binding on chromatin after wortmannin treatment, which is an inhibitor of early DNA damage response proteins PIKK family (phosphatidylinositol 3-kinase-related protein kinase) (Supplementary Fig. 5a), indicating that TET1 binding was dependent on PIKK family kinases. Further studies showed that there was increased binding of TET1 after ATM inhibition using either its inhibitor CGK733 or ATM-siRNA (Supplementary Fig. 5b, 5c), suggesting that binding of TET1 on chromatin is dependent on ATM, but not DNA-PKcs ( Supplementary Fig. 5d).

A similar epigenetic TET1/SIN3A/hMOF recomposition mechanism is verified in an H-RAS V12 transformed cell line
Previous studies have reported that oxidative damage leads to a higher risk of cancers in humans by diminishing histone acetylation, which predominantly occur at H4K16ac(Leufkens et al. ;Fraga et al. 2005;Sosa et al. 2013). Therefore, we hypothesized that dynamic recomposition of this TET1/SIN3A/hMOF/SIRT1 chromatin components might contribute to tumorigenesis. By employing an epithelial ovarian cell line T29 and its oncogenic counterpart T29H cell line which is transformed with H-RAS V12 , we simultaneously analyzed the levels of oxidation and H4K16ac, expression of DNA repair genes, and the binding of TET1, SIN3A, hMOF1, and SIRT1, as compared with those in WT cells. Consistently, transcriptional expression of RAD50, BRCA1, RAD51, and TP53BP1 was decreased in T29H cells (Fig. 7a), but the methylation levels on their promoters was not significantly altered according to our RRBS data (Supplementary Fig. 6a). As expected, compared to T29 cells, there were decreased bindings of TET1 and hMOF, and increased SIRT1 binding on chromatin in T29H cells (Fig. 7b), concurrent with the reduction of H4K16ac levels ( Fig. 7c). The bindings of both TET1 and H4K16ac on the promoter of these DNA repair genes were decreased as well (Fig. 7d, 7e). Meanwhile, we also found elevated 8-OH-dG level using a dot blot assay in T29H cell (Supplementary Fig. 6b).
These results suggested that oncogenic transformation causes hypoacetylation of H4K16 via decreased chromatin binding of TET1 and hMOF1, and increased binding of SIRT1, which lead to downregulation of the DNA repair genes, ultimately impairing DNA repair and involving tumorigenesis.

DISCUSSION
In this study, we revealed that TET1 could function as a core component of TET1/SIN3A/hMOF chromatin complex, supported by their co-occupancy of common targets across the genome, association with each other by co-IP, and co-migration in size fractionation assays (Fig. 1, 2a-e). However, we still cannot rule out the possibility that TET1, hMOF, and SIN3A form sub-complexes to co-occupy the same genomic regions. We demonstrated that TET1 controlled the bindings of hMOF and promoters (data not shown); however, none of these involved in genes in the DNA repair pathway (Fig. 6) (Supplementary Fig.   7a). This supports the hypothesis that epigenetic silencing of important genes in DNA repair both in cancer cells and under oxidative damage, likely results from loss of TET1 which switches the bindings of hMOF and SIRT1 on chromatin, thus promoting aberrant hypoacetylation of H4K16, rather than the alteration of DNA methylation.
This hypothesis is further supported by the observation that a similar epigenetic dynamic alteration occurred in H-RAS V12 oncogenic transformed cells (Fig. 7).
Stephen P. Jackson's group showed that H3K4me3, H3K18ac, H4K5ac, and H4K12ac were unaffected in response to oxidative damage (Tjeertes et al. 2009). Our results also suggested that it's unlikely that hypoacetylation of H3K9ac plays a repressive role in regulation of important genes in DNA repair in response to H 2 O 2 -induced oxidative damage (Supplementary Fig. 6b). These evidence suggest that hypoacetylation of H4K16 is likely a major mechanism to downregulate these DNA repair genes under oxidative damage. However, the repressive role of other histone marks and microRNAs cannot be ruled out and need to be further explored.

Mice
Tet1 +/mice are obtained from the Jackson Laboratory (Cat# 017358). For genotyping of Tet1 +/mice, the forward primer AACTGATTCCCTTCGTGCAG, and the reverse primer TTAAAGCATGGGTGGGAGTC were used. The expected band size for homozygote mutant was 650bp, 850bp for the wild type strain, and 650bp and 850bp double bands for the heterozygote strain.

X-ray irradiation
Irradiations were performed at the Chinese Academy of Sciences (Beijing)using an x-ray machine (RS-2000 PRO Biological system). WT mice and Tet1 +/mice were irradiated with a single whole-body dose of 3Gy x-ray at 60 days of age. The irradiation was operating at 160-kV constant potential and 25 mA (0.3 mm Cu filter) at a dose rate equal to 1.136Gy min−1 for a total of 2.64 min. The cage was cleaned with 75% ethanol when each of irradiation was finished. Coat-state condition of WT mice and Tet1 +/mice with irradiation or sham-irradiated were scored after four month of irradiation.

ChIP-seq Data Preparation
For integrated genomic analysis, we collected 219 ChIP-seq data sets of 103 DNA binding proteins and 14 data sets of 8 histone modification markers in mESCs from GEO and ENCODE (See Supplementary Table 1). Further analysis details are provided in Supplemental Methods.

RRBS Library Preparation, Sequencing and Analysis
See Supplemental Methods for detailed information.            s m L y s in e d e g ra d a ti o n P ro s ta te c a n c e r P a n c re a ti c c a n c e r R e n a l c e ll c a rc in o m a E n d o m e tr ia l c a n c e r M is m a tc h re p a ir M A P K s ig n a li n g p a th w a y In s u li n s ig n a li n g p a th w a y E n d o c y to s is B a s a l tr a n s c ri p ti o n fa c to rs A c u te m y e lo id le u k e m ia F a tt y a c id m e ta b o li s m N − G ly c a n b io s y n th e s is H3K9ac M o f S in 3 a T e t 1 H 4 K 1 6 a c H 3 K 2 7 a c H 3 K 9 a c