Abstract
The human genome is folded into regulatory units termed ‘topologically-associated domains’ (TADs). Genome-wide studies support a global role for the insulator protein CTCF in mediating chromosomal looping and the topological constraint of TAD boundaries. However, the impact of individual insulators on enhancer-gene interactions and transcription remains poorly understood. Here, we investigate epigenome editing strategies for perturbing individual CTCF insulators and evaluating consequent effects on genome topology and transcription. We show that fusions of catalytically-inactive Cas9 (dCas9) to transcriptional repressors (dCas9-KRAB) and DNA methyltransferases (dCas9-DNMT3A, dCas9-DNMT3A3L) can selectively displace CTCF from specific insulators, but only when precisely targeted to the cognate motif. We further demonstrate that stable, partially-heritable insulator disruption can be achieved through combinatorial hit-and-run epigenome editing. Finally, we apply these strategies to simulate an insulator loss mechanism implicated in brain tumorigenesis. Our study provides strategies for stably modifying genome organization and gene activity without altering the underlying DNA sequence.
Subject terms: Genetic engineering, Gene regulation, Chromatin, Chromatin structure, CRISPR-Cas9 genome editing
The role of CTCF-bound insulator elements in enhancer-gene interactions and transcriptional regulation remains poorly understood. Here, the authors investigate multiple epigenome editing strategies for perturbing individual CTCF-bound insulators, and evaluate their effects on genome topology and transcription.
Introduction
TAD structure is conserved across mammalian cell types and tissues1,2, and forms the foundational structure within which cell type-specific enhancers control cell state3,4. Indeed, the majority of gene-enhancer interactions appear to occur within TADS. Within these conserved topological structures, dynamic interactions between enhancers and promoters modulate lineage-specific gene expression. The DNA binding insulator protein CTCF plays important roles in controlling enhancer-promoter interactions, and directing TAD boundary formation4–8. In certain cancers, CTCF motifs are frequently compromised by DNA mutations or methylation events that prevent binding9,10. Despite broad relevance to development and disease, CTCF insulators and TAD boundaries remain poorly understood.
The complexity of genome topology, gene-enhancer connectivity and cell type-specific enhancer activity suggests that the functional significance of individual CTCF-bound insulators are likely to be highly context-specific5,6,11,12. In particular, the consequence of CTCF loss in a given cell type will depend on the ensuing topological alteration, as well as the state of nearby genes, enhancers and sequence elements. Adding to this complexity, TAD boundaries often contain multiple CTCF binding sites1. Accordingly, we need better strategies to test the context-specific functions of individual CTCF sites, and ultimately to predict the consequences of disease-associated insulator disruptions. While genome and epigenome editing tools have been used to disrupt a small number of CTCF sites9,13, generalizable strategies for specific, and stable (mitotically heritable) epigenetic disruption of insulators are needed.
Here, we systematically test different combinations of CRISPR-based epigenome editors for their ability to disrupt individual topological insulators in their endogenous context, with specificity and stability, and without the limitations associated with DNA editing14 (Fig. 1a). We show that fusions of dCas9 to transcriptional repressors (dCas9-KRAB) and DNA methyltransferases (dCas9-DNMT3A, dCas9-DNMT3A3L) can be used to selectively displace CTCF from specific insulators, but only when precisely targeted to the cognate motif. We further demonstrate that stable, partially heritable insulator disruption can be achieved through combinatorial hit-and-run epigenome editing with dCas9-KRAB and dCas9-DNMT3A3L. Finally, we apply these strategies to activate PDGFRA expression in glioblastoma stem cells, thus simulating an insulator loss mechanism implicated in brain tumorigenesis.
Results
Epigenome editing can disrupt CTCF at topological insulators
Fusions of catalytically dead Cas9 (dCas9) to the Krüppel-associated box (KRAB) repressor (dCas9-KRAB) can be targeted to specific loci by gRNAs, where they catalyze histone H3 lysine 9 methylation (H3K9me3) and can repress target gene expression and enhancer function12,15–20. We therefore tested the capacity of dCas9-KRAB to displace CTCF from its binding motif. We initially focused on CTCF-bound insulators that demarcate a TAD boundary upstream of a known glioma oncogene, PDGFRA (Fig. 1b). This locus shows the hallmarks of a TAD boundary by Hi–C and contains two CTCF sites ~20 kb apart, both of which are strongly bound in HEK293 cells (Fig. 1b, Supplementary Fig. 1A). We designed a guide RNA (gRNA) targeting the CTCF motif closer to the TAD interior (annotated as site P1 in Fig. 1c), and also incorporated 8 bases of proximal genomic sequence to ensure specificity (Fig. 2a). We expressed dCas9-KRAB and the CTCF targeting gRNA in HEK293 cells by lentiviral transduction and mapped CTCF binding and H3K9me3 enrichment by genome-wide chromatin immunoprecipitation and sequencing (ChIP-seq). Targeting dCas9-KRAB to this single CTCF site achieved an 83% reduction in CTCF binding, with concomitant enrichment of H3K9me3 across a 3 kb region around the targeted site (Fig. 1c, Supplementary Fig. 1G). The observed 3 kb spreading of the histone modification is consistent with previous studies that have localized dCas9-KRAB to other regulatory elements (Supplementary Fig. 1I, 1J)15. Importantly, CTCF binding at the non-targeted proximal CTCF site within the TAD boundary region was unchanged (Supplementary Fig. 1E).
We next benchmarked dCas9-KRAB-mediated disruption of CTCF binding against Cas9-mediated genome editing of the underlying motif sequence. We targeted Cas9 to the P1 CTCF site in the PDGFRA TAD boundary using the same gRNA as the experiments above. The reduction in CTCF binding at the Cas9 genome edited CTCF site was comparable to the dCas9-KRAB epigenome edited cells (81% vs. 83%, respectively) (Fig. 1c, Supplementary Fig. 1F). We conclude that epigenome editing using dCas9-KRAB can disrupt CTCF binding with similar efficacy to genome editing.
To evaluate the specificity of epigenome editing systematically, we used ChIP-seq to map H3K9me3 and CTCF genome-wide in HEK293 cells after CTCF disruption by dCas9-KRAB (Fig. 1d, e). The P1 CTCF site (on-target) showed significant reduction in CTCF binding compared to non-targeted controls. We also observed concomitant H3K9me3 enrichment at the target site. Remarkably, we did not detect significant alteration at any of the other ~60,000 CTCF binding sites. These data demonstrate that this epigenome editing strategy can effectively disrupt this CTCF insulator with exquisite specificity.
We next expanded our study to test dCas9-KRAB-mediated disruption of CTCF binding at ten individual CTCF binding sites. This set included TAD boundaries known to insulate the oncogenes OLIG2, TAL1, and LMO2, in addition to PDGFRA9,10. As with the PDGFRA-proximal TAD boundary, the selected loci also show features of TAD boundaries by Hi–C and CTCF binding (Supplementary Fig. 1B-D). In order to target a biologically representative set, we selected TAD boundaries with between one and three CTCF binding sites (Fig. 1f, Supplementary Fig. 1G, 1H). We transiently expressed dCas9-KRAB in separate experiments with gRNAs targeting 10 different CTCF motifs. These experiments succeeded in significantly reducing CTCF binding at 8 out of 10 sites tested (Fig. 1f). We also tested integrated expression constructs encoding dCas9-KRAB and gRNA targeting a subset of these loci. These stable expression constructs reduced CTCF binding at all tested loci, including the GIRK4 proximal TAD boundary that was refractory to the transient expression strategy (Supplementary Fig. 1H). These collective results demonstrate that the epigenome editor dCas9-KRAB can selectively disrupt individual CTCF binding sites, despite the high genomic prevalence of the binding motif.
Locus-specific DNA methylation confers stable disruption
We next considered the mechanistic basis of CTCF displacement by dCas9-KRAB. We specifically sought to evaluate the relative contributions of binding site occlusion by dCas9 versus the repressive chromatin state mediated by the KRAB domain. To decouple these effects, we targeted dCas9 lacking the KRAB domain to CTCF motifs in two of the TAD boundaries evaluated above. We found that transient expression of dCas9 alone with gRNA reduced CTCF binding at both sites, although to a lesser degree than the fusion (Fig. 2a, b). We also tested whether CTCF displacement required gRNAs that directly target the CTCF motif, or whether this could also be achieved by targeting nearby sequences. We designed gRNAs to target dCas9-KRAB to sequences adjacent to CTCF motifs in the PDGFRA and OLIG2 TAD boundaries (Fig. 2a, b). Although these experiments increased H3K9me3 levels at the target loci (Supplementary Fig. 2A), their impact on CTCF binding was minimal (Fig. 2b). These data suggest that this epigenome editing fusion acts primarily through binding site occlusion and secondarily through chromatin repression to displace CTCF from TAD boundaries.
A distinguishing feature of epigenome editing is its potential to induce changes that are transient or, alternatively, are stably maintained through cell division. We therefore investigated the temporal stability of H3K9me3 enrichment and CTCF depletion incurred by dCas9-KRAB. We transiently expressed dCas9-KRAB targeted to the PDGFRA or OLIG2 insulators in HEK293 cells. We then serially passaged the cells for 12 days, at which point dCas9-KRAB protein could not be detected (Fig. 2c). At early time points, H3K9me3 enrichment and reduced CTCF binding were readily detected (Fig. 2d). However, as dCas9-KRAB expression was lost, histone methylation was also lost and CTCF binding largely restored to control levels (Fig. 2d). Our results indicate that CTCF displacement requires continuous dCas9-KRAB expression, as is consistent with prior studies that have used this fusion to repress chromatin and enhancer activity12,15–19.
We therefore explored whether other epigenome editing strategies could confer stable, mitotically heritable (or at least partially heritable) displacement of CTCF insulators. Fusions of dCas9 to DNA methyltransferase 3A (dCas9-DNMT3A) and optimized derivatives (e.g., dCas9-DNMT3A3L) have been used to direct DNA methylation to specific loci and silence promoters, with varying degrees of stability13,21–25. CTCF binding is sensitive to DNA methylation status7,8, and can be disrupted by targeted dCas9-DNMT3A13. We tested the ability of dCas9-DNMT3A and dCas9-DNMT3A3L to disrupt CTCF binding at the PDGFRA insulator in HEK293 cells. We transiently expressed dCas9-DNMT3A or dCas9-DNMT3A3L for 3 days, and measured DNA methylation by bisulfite sequencing. We detected methylation over the targeted CTCF motif on 20 to 40% of alleles, with DNMT3A3L more effective than DNMT3A (Supplementary Fig. 2B). Importantly, DNA methylation over the CTCF motif persisted after the epigenome edited cells were serially passaged. We observed ~20% DNA methylation and a congruent ~20% reduction in CTCF binding 12 days after the transient transfection, even though the dCas9 fusions were no longer detected (Fig. 2e, Supplementary Fig. 2F).
Although the DNA methyltransferase fusions induce more stable changes, they were relatively less effective at displacing CTCF. We therefore tested whether the combination of dCas9-targeted H3K9me3 and DNA methylation, used previously for ‘hit-and-run’ epigenetic gene silencing26,27, could disrupt CTCF insulators with greater efficacy and stability. We transiently co-expressed dCas9-KRAB with either dCas9-DNMT3A or dCas9-DNMT3A3L, along with gRNA targeting the CTCF motif in the PDGFRA insulator. We found in particular that the dCas9-KRAB/dCas9-DNMT3A3L combination markedly increased DNA methylation and reduced CTCF binding. Remarkably, robust CTCF disruption was still evident 12 days after transient expression of this fusion, despite lack of continued protein expression (Fig. 2e, f, Supplementary Fig. 2F). To confirm that the altered insulator state induced by dCas9-KRAB/dCas9-DNMT3A3L is epigenetically maintained through mitosis, we continued the time course out to 27 days (Fig. 2g, h, Supplementary Fig. 2C). Methylation at the target locus peaked at 12 days, but was still maintained at 27 days (Fig. 2g, h). At this extended time point, the CpG within the CTCF motif (CpG ‘356) retained >20% DNA methylation while a nearby CpG (CpG ‘388, 32 bases from the motif) retained >50% methylation, near its peak level of ~64% (Fig. 2g, h). This rate of decline is considerably slower than expected from passive demethylation associated with cell divisions (Supplementary Fig. 2D), supporting that the DNA methylated state is epigenetically maintained through mitosis. To evaluate specificity of editing, we performed ChIP-seq on HEK293 cells 12 days after transient transfection with the dCas9-KRAB/dCas9-DNMT3A3L combination. This confirmed robust H3K9me3 enrichment and CTCF depletion at the target insulator, with few significant alterations at other CTCF binding sites in the genome (Fig. 2f, Supplementary Fig. 2E). We conclude that the dCas9-KRAB/dCas9-DNMT3A3L combination can potently disrupt CTCF insulators and initiate a DNA methylated state that is stably maintained after withdrawal of the epigenome editors.
Dissecting topological mechanisms of gene control
We next sought to use this epigenome editing strategy to model a disease-associated insulator loss event. We previously reported that disruption of an insulator in the PDGFRA locus by aberrant DNA methylation is associated with increased expression of the oncogene in IDH-mutant gliomas. We showed that disruption of the CTCF motif by genome editing increased PDGFRA expression in patient-derived gliomasphere models9. This prompted us to explore whether epigenome editing could alter PDGFRA locus topology and expression without changing underlying DNA sequence. We used the GSC6 gliomasphere model for this purpose, which retains an intact PDGFRA insulator and expresses the oncogene at low levels. We used lentiviral transduction to achieve consistent expression of dCas9-KRAB and gRNA in GSC6 cells. We then mapped CTCF, H3K9me3, and the enhancer-associated marker H3K27 acetylation (H3K27ac) by ChIP-seq. We detected a 95% reduction in CTCF binding at the PDGFRA insulator, while adjacent CTCF sites were unchanged (Fig. 3a, Supplementary Fig. 3A, 3B).
In IDH-mutant gliomas, PDGFRA insulator loss is associated with an altered locus topology that allows the PDGFRA promoter to interact with an enhancer in an adjacent TAD9. We therefore investigated how PDGFRA promoter interactions were altered by epigenome editing, using 4C-seq. We found that CTCF displacement by epigenome editing markedly reduced the interaction between the promoter and insulator, which reside at opposite sides of the TAD, consistent with loss of a long-range chromosomal loop that normally connects these loci (Fig. 3b). Displacement of the insulator also caused a qualitative increase in long-range interactions between the PGDFRA promoter and several sites in the adjacent TADs. These changes were less striking than the loss of interactions with the target insulator, but were also consistent with TAD boundary loss. Putative sites of interaction include the LNX1 proximal CTCF site, and the previously described FIP1L1-proximal enhancer9 (Fig. 3a, b, Supplementary Fig. 3C, 3D). We conclude that epigenome editing can disrupt this TAD boundary, alter locus topology and allow aberrant cross-boundary enhancer-promoter interactions.
We next investigated how PDGFRA insulator disruption altered gene expression in GSC6 cells using RNA-seq. Insulator disruption increased expression of PDGFRA and several downstream PDGFR pathway genes (Fig. 3c). We did not detect significant changes in the expression of other genes in the PDGFRA TAD or the adjacent TAD, although several genes in a more distal upstream TAD were modestly induced (Supplementary Fig. 3E). Overall, PDGFRA scored as the most significant differentially expressed gene in this locus (Supplementary Fig. 3F).
Finally, we investigated how insulator disruption might alter intra-TAD interactions of an inactive promoter. We designed a 4C viewpoint to the promoter of GSX2, which is neither expressed in GSC6 gliomaspheres nor activated by insulator disruption (Supplementary Fig. 4A, 4B). We found that the interaction between the GSX2 promoter and the PDGFRA insulator was reduced by disruption of the latter, albeit to a lesser extent than that observed for the PDGFRA promoter viewpoint. We found that insulator disruption also impacted long-range interactions made by the GSX2 promoter, particularly with H3K27ac-marked sites in the adjacent TAD. This suggests that CTCF displacement by epigenome editing alters the overall folding of this region, with context-specific and promoter-specific effects. Taken together, our results demonstrate the potential of epigenome editing to model insulator loss events and evaluate consequential changes in genome topology and gene expression (Fig. 3d).
Discussion
In summary, we present epigenome editing strategies for potent, specific, and stable disruption of CTCF insulators (Fig. 1a). We demonstrate the applicability of a combinatorial ‘hit-and-run’ strategy to robustly disrupt insulators and establish a DNA methylated state that is epigenetically maintained after withdrawal of the epigenome editors. Alternate strategies may be employed to achieve either transient or stable insulator disruption. Another advantageous feature of epigenome editing, relative to genome editing, is the remarkable specificity with which an insulator can be disrupted, as evidenced by lack of detectable off-target effects in genome-wide analyses. Our data also provide mechanistic insight into an insulator loss event associated with oncogene activation by simulating the disease-associated lesion without altering the underlying genetic sequence. Our data faithfully simulate a cancer-associated insulator loss and confirm that it allows the PDGFRA oncogene to engage in multiple long-range interactions with a neighboring TAD, including with a super-enhancer element that likely drives its expression (Fig. 3d). Looking forward, these strategies can provide powerful and general means for dissecting the roles of insulators in endogenous contexts, for simulating and evaluating causality of disease associated lesions, and for precisely controlled modulation of genome topology in experimental, and ultimately, therapeutic settings.
Methods
Cell culture and transfection
HEK293 culture and passaging: The human embryonic kidney cell line HEK293 (ATCC, CRL-1573) was cultured in DMEM with 10% FBS, 1% Pen/Strep antibiotic mix (Gibco).
HEK293 transient transfection: For transfections, 10 cm plates were seeded 48 h in advance with 2 × 106 cells per plate. To make transfection mix 8 µg of dCas9 effector plasmid DNA and 1 µg of guide RNA expression plasmid were re-suspended in Opti-MEM media (Thermo). For combinations of effectors, total transfected DNA was kept constant at 8 µg. Fugene HD (Promega) was added to transfection mix according to the manufacturer protocol using a 3:1 ratio to DNA. Cells were harvested 72 h after transfection for ChIP, DNA, or RNA extraction.
HEK293 serial passaging: For serial passaging experiments, 90% of cells were harvested and remaining cells were passaged into fresh media in a new 10 cm dish. Cells were passaged every 72 h until final harvest.
GSC6 culture and passaging: GSC6 gliomasphere lines were derived from IDH wild-type tumors resected at Massachusetts General Hospital28. These cell lines were established under Protocol # 2005P001609, which was approved by The Partners Human Research Committee (PHRC), and complies with all relevant ethical regulations. Gliomaspheres were maintained in culture as described9,29,30. Neurosphere cultures contain Neurobasal media supplemented with 20 ng ml−1 recombinant EGF (R and D Systems), 20 ng ml−1 FGF2 (R and D Systems), 1 × B27 supplement (Invitrogen), 0.5 × N2 supplement (Invitrogen), 3 mM L-glutamine, and penicillin/streptomycin. Cultures were confirmed to be mycoplasma-free via PCR methods.
Plasmids
Plasmids were purchased from Addgene including dCas9-KRAB, dCas9-KRAB-T2a-Puro, pdCas9-DNMT3A-EGFP (Catalog numbers 50919, 71236, 71666, respectively). Expression vector for dCas9 (JDS286) was a gift from J. Keith Joung. Human codon optimized sequence for dCas9–Dnmt3a3l24 was synthesized (Genscript) and cloned into pcDNA3.1 including an in-frame P2A-eGFP fusion. All plasmids are available from Addgene: pcDNA3.1-dCas9-Dnmt3a-Dnmt3l-P2A-eGFP: ID 128424. Expression vectors including dCas9-Dnmt3a-Dnmt3l and the gRNA expression cassette are also available from Addgene: pDT304 (PuroR): ID 128800, and pDT305 (eGFP): ID 128801. SpCas9 and guide RNA expression vectors lentiCRISPR and lentiGuide were gifts from the Broad Genomic Perturbation Platform.
Guide RNA target selection and expression vector cloning
Insulator loci were identified in previous studies and by selecting CTCF binding sites near conserved Hi–C derived TAD boundaries that were co-bound by cohesin complex components (derived from GSE44267)31. CTCF motif positions were identified within ENCODE ChIP-seq peaks at target sites using JASPAR. Approximately 200–300 bp of motif flanking genomic sequence were used to identify candidate guide RNA sequences. Guide RNA sequences were scored using previously described rules32. The highest scoring gRNA sequences overlapping the CTCF motif were identified and synthesized with vector compatible overhangs (IDT).
Chromatin Immunoprecipitation (ChIP) assays
ChIP was performed as previously described33. Briefly, cells were crosslinked in 1% formaldehyde at 37 °C for 10 min, quenched with glycine, lysed on ice in buffer containing 1% SDS, diluted to a final concentration of 0.3% SDS and sonicated for 5 min total time at ~10 W power. Sonicated chromatin was incubated with the relevant antibody overnight at 4 °C with rotation after dilution to 0.1% SDS. Antibody-bound chromatin was isolated using Protein G magnetic beads (Thermo). Bound complex was washed several times and eluted from beads at 65 °C for 1 h with shaking. Eluted complex was treated with RNase (Roche) for 30 min at 37 °C and proteinase K (Thermo) for 3 h at 63 °C while crosslinks were reversed concurrently. Eluted DNA was isolated using 2× Ampure XP magnetic beads (Beckman) and quantitated using Qubit dsDNA HS (Invitrogen). Antibodies used: CTCF (D31H2), diluted 1:240 (Cell Signaling Technologies, catalog #3418), H3K9me3, diluted 1:750 (abcam, catalog #ab8898).
ChIP-qPCR: Real-time PCR was performed to quantitate ChIP DNA using SYBR Select Master Mix (Applied Biosystems). Primers to quantitate IP DNA in CTCF peaks were generated using the Geneious software program (Biomatters). Each sample was internally normalized using a positive control locus that is constitutively enriched across samples. Enrichments relative to this positive control were compared between experimental and control samples to obtain relative quantitations for each epitope.
Sequencing library preparation: Library preparation was performed as previously described[9] using 5 ng (CTCF and H3K27ac), or 20 ng (H3K9me3) of ChIP DNA input. Paired-end 38 bp reads were generated on a NextSeq500 (Illumina). H3K27ac sequencing reads for HEK293 were obtained from GEO under accession GSM2439222. Reads were uniquely aligned to hg19 using BWA and peak calling was performed using HOMER.
DNA methylation
Genomic DNA was isolated from 1 × 106 cells using Quick-gDNA (Zymo) following manufacturer’s instructions. Methylation sensitive digestion: Genomic DNA was mass normalized to 100 ng and completely digested using the methylation-sensitive restriction enzyme Hin6I (Thermo) according to manufacturer’s instructions. Mock digested genomic DNA prepared from the same dilution was processed in parallel without addition of enzyme. Quantitation of digested DNA: Digested and mock digested samples were quantitated at target sites by real-time PCR as described above. DNA methylation percentage was calculated as the fraction of undigested genomic DNA retained in the digested samples.
Chromosome conformation analysis (4C-seq)
4C sample preparation was performed as previously described by Splinter et al. using NlaIII (NEB) and Csp6I (Thermo)34. Viewpoint primers were selected from a previously reported database35.
Primers sequences were concatenated with Illumina compatible adapter sequences and unique barcodes and synthesized as Ultramer oligos (IDT). Final library amplification was set up on ice using 3.2 µg of purified 4C sample DNA, 2.5 µM primers, and Ultra II Q5 Master Mix (NEB) split into 16 × 50 µl PCR reactions with extension at 65 °C. Amplification products were purified using 1.5× Ampure XP (Beckman) bead purification. Size distributions were established using a Bioanalyzer DNA 12000 Assay (Agilent).
Paired-end sequencing reads were generated as above and analyzed with the 4C-ker pipeline as previously described to identify genomic loci with differential chromosome contacts between conditions36.
Gene expression analysis
RNA extraction: RNA was extracted from 1–2 × 106 cells using the RNeasy kit (Qiagen) following manufacturer’s instructions.
RNA-seq: Sequencing libraries were generated using NEBNext Ultra II RNA Library Prep Kit (NEB) after poly(A) mRNA isolation following manufacturer’s instructions. A minimum of 30 × 106 Paired-end reads were generated on a NextSeq500 (Illumina). Sequencing reads were aligned to hg19 using STAR, transcripts counts were derived using RSEM, followed by analysis with DESeq2 to identify differentially expressed genes. Gene Ontology enrichments of differentially expressed genes was performed using PANTHER37.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Supplementary information
Acknowledgements
We thank Liz Gaskell and members of the Bernstein laboratory for constructive comments and criticisms. We thank C. Davis, Y. Drier, and the MGH and Broad Institute flow cytometry core facilities for their assistance with experiments and analysis. B.E.B. is the Bernard and Mildred Kayden Endowed MGH Research Institute Chair and an American Cancer Society Research Professor. This work was supported by a Pioneer Award from the NIH Common Fund and National Cancer Institute (DP1CA216873). W.A.F. is supported by an F32 from the National Cancer Institute.
Author contributions
D.R.T. and B.E.B. conceived the project and designed experiments. D.R.T. and W.A.F. performed experiments. D.R.T. and B.E.B. wrote the paper.
Data availability
Data is available in GEO, accession number GSE121998. All other relevant data supporting the key findings of this study are available within the article and its Supplementary Information files or from the corresponding author upon reasonable request. The source data underlying Figs. 1f, 2b, 2d, 2e, 2g and Supplementary Figs. 1e-h, 2a-b, 3a-f are provided as a Source Data file. This study also utilized publicly available datasets GSE44267 and GSM2439222. A reporting summary for this Article is available as a Supplementary Information file.
Competing interests
B.E.B. declares outside interests in Fulcrum Therapeutics, 1CellBio, Nohla Therapeutics, HiFiBio, Arsenal BioSciences, Biomillenia and Cell Signaling Technologies. B.E.B., W.A.F, and D.R.T. are named inventors on a patent application related to the detection and modification of insulators.
Footnotes
Peer review information Nature Communications thanks Elphège Nora and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Supplementary Information accompanies this paper at 10.1038/s41467-019-12166-w.
References
- 1.Rao SSP, et al. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell. 2014;159:1665–1680. doi: 10.1016/j.cell.2014.11.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Dixon JR, et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature. 2012;485:376–380. doi: 10.1038/nature11082. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hnisz D, Day DS, Young RA. Insulated neighborhoods: structural and functional units of mammalian gene control. Cell. 2016;167:1188–1200. doi: 10.1016/j.cell.2016.10.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Schmitt AD, et al. A compendium of chromatin contact maps reveals spatially active regions in the human genome. Cell Rep. 2016;17:2042–2059. doi: 10.1016/j.celrep.2016.10.061. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rao SSP, et al. Cohesin loss eliminates all loop domains. Cell. 2017;171:305–320.e24. doi: 10.1016/j.cell.2017.09.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nora EP, et al. Targeted degradation of CTCF decouples local insulation of chromosome domains from genomic compartmentalization. Cell. 2017;169:930–944.e22. doi: 10.1016/j.cell.2017.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hark AT, et al. CTCF mediates methylation-sensitive enhancer-blocking activity at the H19/Igf2 locus. Nature. 2000;405:486–489. doi: 10.1038/35013106. [DOI] [PubMed] [Google Scholar]
- 8.Bell AC, Felsenfeld G. Methylation of a CTCF-dependent boundary controls imprinted expression of the Igf2 gene. Nature. 2000;405:482–485. doi: 10.1038/35013100. [DOI] [PubMed] [Google Scholar]
- 9.Flavahan WA, et al. Insulator dysfunction and oncogene activation in IDH mutant gliomas. Nature. 2016;529:110–114. doi: 10.1038/nature16490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hnisz D, et al. Activation of proto-oncogenes by disruption of chromosome neighborhoods. Science. 2016;351:1454–1458. doi: 10.1126/science.aad9024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ji X, et al. 3D chromosome regulatory landscape of human pluripotent cells. Cell Stem Cell. 2016;18:262–275. doi: 10.1016/j.stem.2015.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Fulco CP, et al. Systematic mapping of functional enhancer–promoter connections with CRISPR interference. Science. 2016;354:769–773. doi: 10.1126/science.aag2445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Liu XS, et al. Editing DNA methylation in the mammalian genome. Cell. 2016;167:233–247.e17. doi: 10.1016/j.cell.2016.08.056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tsai SQ, Joung JK. Defining and improving the genome-wide specificities of CRISPR–Cas9 nucleases. Nat. Rev. Genet. 2016;17:300. doi: 10.1038/nrg.2016.28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Thakore PI, et al. Highly specific epigenome editing by CRISPR-Cas9 repressors for silencing of distal regulatory elements. Nat. Methods. 2015;12:1143–1149. doi: 10.1038/nmeth.3630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kearns NA, et al. Functional annotation of native enhancers with a Cas9-histone demethylase fusion. Nat. Methods. 2015;12:401–403. doi: 10.1038/nmeth.3325. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Klann TS, et al. CRISPR-Cas9 epigenome editing enables high-throughput screening for functional regulatory elements in the human genome. Nat. Biotechnol. 2017;35:561–568. doi: 10.1038/nbt.3853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.O’Geen H, et al. Ezh2-dCas9 and KRAB-dCas9 enable engineering of epigenetic memory in a context-dependent manner. Epigenet. Chromatin. 2019;12:26. doi: 10.1186/s13072-019-0275-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.O’Geen H, et al. dCas9-based epigenome editing suggests acquisition of histone methylation is not sufficient for target gene repression. Nucleic Acids Res. 2017;45:9901–9916. doi: 10.1093/nar/gkx578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mlambo T, et al. Designer epigenome modifiers enable robust and sustained gene silencing in clinically relevant human cells. Nucleic Acids Res. 2018;46:4456–4468. doi: 10.1093/nar/gky171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Vojta A, et al. Repurposing the CRISPR-Cas9 system for targeted DNA methylation. Nucleic Acids Res. 2016;44:5615–5628. doi: 10.1093/nar/gkw159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.McDonald JI, et al. Reprogrammable CRISPR/Cas9-based system for inducing site-specific DNA methylation. Biol. Open. 2016;5:866–874. doi: 10.1242/bio.019067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Galonska C, et al. Genome-wide tracking of dCas9-methyltransferase footprints. Nat. Commun. 2018;9:597. doi: 10.1038/s41467-017-02708-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Stepper P, et al. Efficient targeted DNA methylation with chimeric dCas9–Dnmt3a–Dnmt3L methyltransferase. Nucleic Acids Res. 2017;45:1703–1713. doi: 10.1093/nar/gkw1112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Huang Y-H, et al. DNA epigenome editing using CRISPR-Cas SunTag-directed DNMT3A. Genome Biol. 2017;18:176. doi: 10.1186/s13059-017-1306-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rowe HM, et al. De novo DNA methylation of endogenous retroviruses is shaped by KRAB-ZFPs/KAP1 and ESET. Development. 2013;140:519–529. doi: 10.1242/dev.087585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Amabile A, et al. Inheritable silencing of endogenous genes by hit-and-run targeted epigenetic editing. Cell. 2016;167:219–232.e14. doi: 10.1016/j.cell.2016.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Wakimoto H, et al. Maintenance of primary tumor phenotype and genotype in glioblastoma stem cells. Neuro. Oncol. 2012;14:132–144. doi: 10.1093/neuonc/nor195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Rheinbay E, et al. An aberrant transcription factor network essential for Wnt signaling and stem cell maintenance in glioblastoma. Cell Rep. 2013;3:1567–1579. doi: 10.1016/j.celrep.2013.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Suvà ML, et al. Reconstructing and reprogramming the tumor-propagating potential of glioblastoma stem-like cells. Cell. 2014;157:580–594. doi: 10.1016/j.cell.2014.02.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zuin J, et al. Cohesin and CTCF differentially affect chromatin architecture and gene expression in human cells. Proc. Natl Acad. Sci. USA. 2014;111:996–1001. doi: 10.1073/pnas.1317788111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Doench JG, et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat. Biotechnol. 2016;34:184–191. doi: 10.1038/nbt.3437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ram O, et al. Combinatorial patterning of chromatin regulators uncovered by genome-wide location analysis in human cells. Cell. 2011;147:1628–1639. doi: 10.1016/j.cell.2011.09.057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Splinter E, de Wit E, van de Werken HJG, Klous P, de Laat W. Determining long-range chromatin interactions for selected genomic sites using 4C-seq technology: from fixation to computation. Methods. 2012;58:221–230. doi: 10.1016/j.ymeth.2012.04.009. [DOI] [PubMed] [Google Scholar]
- 35.van de Werken HJG, et al. Robust 4C-seq data analysis to screen for regulatory DNA interactions. Nat. Methods. 2012;9:969–972. doi: 10.1038/nmeth.2173. [DOI] [PubMed] [Google Scholar]
- 36.Raviram R, et al. 4C-ker: a method to reproducibly identify genome-wide interactions captured by 4C-Seq experiments. PLoS Comput. Biol. 2016;12:e1004780. doi: 10.1371/journal.pcbi.1004780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Thomas PD, et al. PANTHER: a browsable database of gene products organized by biological function, using curated protein family and subfamily classification. Nucleic Acids Res. 2003;31:334–341. doi: 10.1093/nar/gkg115. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data is available in GEO, accession number GSE121998. All other relevant data supporting the key findings of this study are available within the article and its Supplementary Information files or from the corresponding author upon reasonable request. The source data underlying Figs. 1f, 2b, 2d, 2e, 2g and Supplementary Figs. 1e-h, 2a-b, 3a-f are provided as a Source Data file. This study also utilized publicly available datasets GSE44267 and GSM2439222. A reporting summary for this Article is available as a Supplementary Information file.