DNA (de)methylation in embryonic stem cells controls CTCF-dependent chromatin boundaries
Ontology highlight
ABSTRACT: The interplay of DNA (de)methylation and the architectural chromatin protein CTCF is an important regulator of cellular differentiation, but molecular mechanisms determining differential CTCF binding are not well understood. Mouse embryonic stem cells (ESCs) contain up to 10% of CpG residues in 5-hydroxymethylated form (5hmC) and smaller amounts in 5-formylated or 5-carboxylated forms (5fC, 5caC). Here, we studied double knock out ESCs (DKO) deficient for TET1 and TET2 enzymes which are responsible for the conversion of 5mC to 5hmC, 5fC and 5caC. For both wild type (WT) and DKO cells, we measured changes in nucleosome positions, CTCF binding, DNA methylation and gene expression. Our integrative data analysis combined with biophysical modelling suggests that both differential (de)methylation and CTCF binding are to a large degree predictable from the DNA sequence. Common CTCF sites that remained bound in DKO cells were significantly enriched at unmethylated CpG islands, while CTCF loss was associated with differential DNA methylation. 5hmCs and 5fCs had the opposite effects in terms of CTCF priming: CTCF was preferentially lost from sites that were marked in WT cells by 5hmCs but not 5fCs. Interestingly, the losses of CTCF at TAD boundaries lead to aberrant spreading of DNA methylation, as well as downregulation of neighbouring genes. Overall, our study clarifies the fundamental mechanism of differential CTCF binding and provides evidence of DNA sequence-encoded temporal chromatin changes during cell differentiation.
Project description:Coordinated changes of DNA (de)methylation, nucleosome positioning and chromatin binding of the architectural protein CTCF play an important role for establishing cell type specific chromatin states during differentiation. To elucidate molecular mechanisms that link these processes we studied the perturbed DNA modification landscape in mouse embryonic stem cells (ESCs) carrying a double knockout (DKO) of the Tet1 and Tet2 dioxygenases. These enzymes are responsible for the conversion of 5-methylcytosine (5mC) into its hydroxymethylated (5hmC), formylated (5fC) or carboxylated (5caC) forms. We determined changes in nucleosome positioning, CTCF binding, DNA methylation and gene expression in DKO ESCs, and developed biophysical models to predict differential CTCF binding. Methylation-sensitive nucleosome repositioning accounted for a significant portion of CTCF binding loss in DKO ESCs, while unmethylated and nucleosome-depleted CpG islands were enriched for CTCF sites that remained occupied. A number of CTCF sites also displayed direct correlations with the CpG modification state: CTCF was preferentially lost from sites that were marked with 5hmC in wild type cells but not from 5fC enriched sites. In addition, we found that some CTCF sites can act as bifurcation points defining the differential methylation landscape. CTCF loss from such sites, e.g. at promoters, boundaries of chromatin loops and topologically associated domains (TADs), was correlated with DNA methylation/demethylation spreading and can be linked to downregulation of neighbouring genes. Our results reveal a hierarchical interplay between cytosine modifications, nucleosome positions and DNA sequence that determines differential CTCF binding and regulates gene expression.
Project description:5-methylcytosine (5mC), the predominant epigenetic modification on DNA, plays critical roles in mammalian development and is dysregulated in various human pathologies. In mammals, the TET family of dioxygenases can oxidize 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) in a stepwise manner. 5fC and 5caC are selectively recognized and excised by mammalian thymine DNA glycosylase (TDG), and restored to normal cytosine through base excision repair (BER). Once 5mC/5hmC is converted to 5fC and/or 5caC, the modified cytosine is committed to demethylation through BER. Thus 5fC and 5caC most likely mark active demethylation in the mammalian genome. Here we introduce a genome-wide approach to obtain single-base resolution maps of 5fC and 5caC, respectively. We show that, in mouse embryonic stem cells (mESCs), 5fC and 5caC are preferentially generated at highly hypomethylated regions and more active enhancers. Moreover, 5caC-marked regions are characterized by the lowest methylation and highest enhancer activity among all modification sites associated with 5hmC, 5fC and 5caC, and are enriched adjacent to pluripotency transcription factor (TF)-binding motifs. These observations, together with the surprising lack of overlap between 5fC and 5caC sites, highlight a gradient of Tet-mediated 5mC oxidation activity at regulatory elements in tuning epigenetic dynamics11. DNA immunoprecipitation coupled chemical-modification assisted bisulfite sequencing (DIP-CAB-Seq) for Tdg fl/fl and Tdg-/- mESCs
Project description:5-methylcytosine (5mC), the predominant epigenetic modification on DNA, plays critical roles in mammalian development and is dysregulated in various human pathologies. In mammals, the TET family of dioxygenases can oxidize 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) in a stepwise manner. 5fC and 5caC are selectively recognized and excised by mammalian thymine DNA glycosylase (TDG), and restored to normal cytosine through base excision repair (BER). Once 5mC/5hmC is converted to 5fC and/or 5caC, the modified cytosine is committed to demethylation through BER. Thus 5fC and 5caC most likely mark active demethylation in the mammalian genome. Here we introduce a genome-wide approach to obtain single-base resolution maps of 5fC and 5caC, respectively. We show that, in mouse embryonic stem cells (mESCs), 5fC and 5caC are preferentially generated at highly hypomethylated regions and more active enhancers. Moreover, 5caC-marked regions are characterized by the lowest methylation and highest enhancer activity among all modification sites associated with 5hmC, 5fC and 5caC, and are enriched adjacent to pluripotency transcription factor (TF)-binding motifs. These observations, together with the surprising lack of overlap between 5fC and 5caC sites, highlight a gradient of Tet-mediated 5mC oxidation activity at regulatory elements in tuning epigenetic dynamics11.
Project description:Cytosine DNA bases can be methylated by DNA methyltransferases and subsequently oxidized by TET proteins. The resulting 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) are considered demethylation intermediates as well as stable epigenetic marks. To dissect the contribution of these cytosine modifying enzymes, we generated combinations of Tet knockout (KO) embryonic stem cells (ESCs) and systematically measured protein and DNA modification levels at the transition from naive to primed pluripotency. Whereas the increase of genomic 5-methylcytosine (5mC) levels during exit from pluripotency correlated with an upregulation of the de novo DNA methyltransferases DNMT3A and DNMT3B, the subsequent oxidation steps turned out to be far more complex. The strong increase of oxidized cytosine bases (5hmC, 5fC, and 5caC) was accompanied by a drop in TET2 levels, yet the analysis of KO cells suggested that TET2 is responsible for most 5fC formation. The comparison of modified cytosine and enzyme levels in Tet KO cells revealed distinct and differentiation-dependent contributions of TET1 and TET2 to 5hmC and 5fC formation arguing against a processive mechanism of 5mC oxidation. The apparent independent steps of 5hmC and 5fC formation suggest yet to be identified mechanisms regulating TET activity and may constitute another layer of epigenetic regulation.
Project description:Ten-eleven translocation (Tet) family of DNA dioxygenases converts 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5- carboxylcytosine (5caC) through iterative oxidation reactions. While 5mC and 5hmC are relatively abundant, 5fC and 5caC are at very low levels in the mammalian genome. Thymine DNA glycosylase (TDG) and base excision repair (BER) pathways can actively remove 5fC/5caC to regenerate unmethylated cytosine, but it is unclear to what extent and at which part of the genome such active demethylation processes take place. Here, we have performed high-throughput sequencing analysis of 5mC/5hmC/5fC/5caC- enriched DNA using modification-specific antibodies and generated genome-wide distribution maps of these cytosine modifications in wild-type and Tdg-deficient mouse embryonic stem cells (ESCs). We observe that the steady state 5fC and 5caC are preferentially detected at repetitive sequences in wild-type mouse ESCs. Depletion of TDG causes marked accumulation of 5fC and 5caC at a large number of distal gene regulatory elements and transcriptionally repressed/poised gene promoters, suggesting that Tet/TDG-dependent dynamic cycling of 5mC oxidation states may be involved in regulating the function of these regions. Thus, comprehensive mapping of 5mC oxidation and BER pathway activity in the mammalian genome provides a promising approach for better understanding of biological roles of DNA methylation and demethylation dynamics in development and diseases. In this dataset, we include the DIP-seq data of 5mC, 5hmC, 5fC and 5caC in both control and Tdg knockdown mouse embryonic stem cells.
Project description:Enzymatic erasure of DNA methylation in mammals involves iterative 5-methylcytosine (5mC) oxidation by the ten-eleven translocation (TET) family of DNA dioxygenase proteins. As the most abundant form of oxidized 5mC, the prevailing model considers 5-hydroxymethylcytosine (5hmC) as a key nexus in active DNA demethylation that can either indirectly facilitate replication-dependent depletion of 5mC by inhibiting maintenance DNA methylation machinery (UHRF1/DNMT1), or directly be iteratively oxidized to 5-formylcytosine (5fC) and 5-carboxycytosine (5caC) and restored to cytosine (C) through thymine DNA glycosylase (TDG)-mediated 5fC/5caC excision repair. In proliferative somatic cells, to what extent TET-dependent removal of 5mC entails indirect DNA demethylation via 5hmC-induced replication-dependent dilution or direct iterative conversion of 5hmC to 5fC/5caC is unclear. Here we leverage a catalytic processivity stalling variant of human TET1 (TET1.var: T1662E) to decouple the stepwise generation of 5hmC from subsequent 5fC/5caC generation, excision and repair. By using a CRISPR/dCas9-based epigenome-editing platform, we demonstrate that 5fC/5caC excision repair (by wild-type TET1, TET1.wt), but not 5hmC generation alone (by TET1.var), is requisite for robust restoration of unmodified cytosines and reversal of somatic silencing of the methylation-sensitive, germline-specific RHOXF2B gene promoter. Furthermore, integrated whole-genome multi-modal epigenetic sequencing reveals that hemi-hydroxymethylated CpG dyads predominantly resist replication-dependent depletion of 5mC on the opposing strand in TET1.var-expressing cells. Notably, TET1.var-mediated 5hmC generation is sufficient to induce similar levels of differential gene expression (compared to TET1.wt) without inducing major changes in unmodified cytosine profiles across the genome. Our study suggests 5hmC alone plays a limited role in driving replication-dependent DNA demethylation in the presence of functional DNMT1/UHRF1 mechanisms, but can regulate gene expression as a bona fide epigenetic mark in proliferative somatic cells.
Project description:Enzymatic erasure of DNA methylation in mammals involves iterative 5-methylcytosine (5mC) oxidation by the ten-eleven translocation (TET) family of DNA dioxygenase proteins. As the most abundant form of oxidized 5mC, the prevailing model considers 5-hydroxymethylcytosine (5hmC) as a key nexus in active DNA demethylation that can either indirectly facilitate replication-dependent depletion of 5mC by inhibiting maintenance DNA methylation machinery (UHRF1/DNMT1), or directly be iteratively oxidized to 5-formylcytosine (5fC) and 5-carboxycytosine (5caC) and restored to cytosine (C) through thymine DNA glycosylase (TDG)-mediated 5fC/5caC excision repair. In proliferative somatic cells, to what extent TET-dependent removal of 5mC entails indirect DNA demethylation via 5hmC-induced replication-dependent dilution or direct iterative conversion of 5hmC to 5fC/5caC is unclear. Here we leverage a catalytic processivity stalling variant of human TET1 (TET1.var: T1662E) to decouple the stepwise generation of 5hmC from subsequent 5fC/5caC generation, excision and repair. By using a CRISPR/dCas9-based epigenome-editing platform, we demonstrate that 5fC/5caC excision repair (by wild-type TET1, TET1.wt), but not 5hmC generation alone (by TET1.var), is requisite for robust restoration of unmodified cytosines and reversal of somatic silencing of the methylation-sensitive, germline-specific RHOXF2B gene promoter. Furthermore, integrated whole-genome multi-modal epigenetic sequencing reveals that hemi-hydroxymethylated CpG dyads predominantly resist replication-dependent depletion of 5mC on the opposing strand in TET1.var-expressing cells. Notably, TET1.var-mediated 5hmC generation is sufficient to induce similar levels of differential gene expression (compared to TET1.wt) without inducing major changes in unmodified cytosine profiles across the genome. Our study suggests 5hmC alone plays a limited role in driving replication-dependent DNA demethylation in the presence of functional DNMT1/UHRF1 mechanisms, but can regulate gene expression as a bona fide epigenetic mark in proliferative somatic cells.
Project description:Ten-eleven translocation (Tet) family of DNA dioxygenases converts 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5- carboxylcytosine (5caC) through iterative oxidation reactions. While 5mC and 5hmC are relatively abundant, 5fC and 5caC are at very low levels in the mammalian genome. Thymine DNA glycosylase (TDG) and base excision repair (BER) pathways can actively remove 5fC/5caC to regenerate unmethylated cytosine, but it is unclear to what extent and at which part of the genome such active demethylation processes take place. Here, we have performed high-throughput sequencing analysis of 5mC/5hmC/5fC/5caC- enriched DNA using modification-specific antibodies and generated genome-wide distribution maps of these cytosine modifications in wild-type and Tdg-deficient mouse embryonic stem cells (ESCs). We observe that the steady state 5fC and 5caC are preferentially detected at repetitive sequences in wild-type mouse ESCs. Depletion of TDG causes marked accumulation of 5fC and 5caC at a large number of distal gene regulatory elements and transcriptionally repressed/poised gene promoters, suggesting that Tet/TDG-dependent dynamic cycling of 5mC oxidation states may be involved in regulating the function of these regions. Thus, comprehensive mapping of 5mC oxidation and BER pathway activity in the mammalian genome provides a promising approach for better understanding of biological roles of DNA methylation and demethylation dynamics in development and diseases. Refer to individual Series
Project description:Ten-eleven translocation (Tet) family of DNA dioxygenases converts 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5- carboxylcytosine (5caC) through iterative oxidation reactions. While 5mC and 5hmC are relatively abundant, 5fC and 5caC are at very low levels in the mammalian genome. Thymine DNA glycosylase (TDG) and base excision repair (BER) pathways can actively remove 5fC/5caC to regenerate unmethylated cytosine, but it is unclear to what extent and at which part of the genome such active demethylation processes take place. Here, we have performed high-throughput sequencing analysis of 5mC/5hmC/5fC/5caC- enriched DNA using modification-specific antibodies and generated genome-wide distribution maps of these cytosine modifications in wild-type and Tdg-deficient mouse embryonic stem cells (ESCs). We observe that the steady state 5fC and 5caC are preferentially detected at repetitive sequences in wild-type mouse ESCs. Depletion of TDG causes marked accumulation of 5fC and 5caC at a large number of distal gene regulatory elements and transcriptionally repressed/poised gene promoters, suggesting that Tet/TDG-dependent dynamic cycling of 5mC oxidation states may be involved in regulating the function of these regions. Thus, comprehensive mapping of 5mC oxidation and BER pathway activity in the mammalian genome provides a promising approach for better understanding of biological roles of DNA methylation and demethylation dynamics in development and diseases. Gene expression comparison of control and Tdg knockdown mouse embryonic stem cells.
Project description:Ten-eleven translocation (Tet) family of DNA dioxygenases converts 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5- carboxylcytosine (5caC) through iterative oxidation reactions. While 5mC and 5hmC are relatively abundant, 5fC and 5caC are at very low levels in the mammalian genome. Thymine DNA glycosylase (TDG) and base excision repair (BER) pathways can actively remove 5fC/5caC to regenerate unmethylated cytosine, but it is unclear to what extent and at which part of the genome such active demethylation processes take place. Here, we have performed high-throughput sequencing analysis of 5mC/5hmC/5fC/5caC- enriched DNA using modification-specific antibodies and generated genome-wide distribution maps of these cytosine modifications in wild-type and Tdg-deficient mouse embryonic stem cells (ESCs). We observe that the steady state 5fC and 5caC are preferentially detected at repetitive sequences in wild-type mouse ESCs. Depletion of TDG causes marked accumulation of 5fC and 5caC at a large number of distal gene regulatory elements and transcriptionally repressed/poised gene promoters, suggesting that Tet/TDG-dependent dynamic cycling of 5mC oxidation states may be involved in regulating the function of these regions. Thus, comprehensive mapping of 5mC oxidation and BER pathway activity in the mammalian genome provides a promising approach for better understanding of biological roles of DNA methylation and demethylation dynamics in development and diseases.