Transcriptional Dysregulation of MYC Reveals Common Enhancer-Docking Mechanism

Abstract

Transcriptional dysregulation of the MYC oncogene is among the most frequent events in aggressive tumor cells, and this is generally accomplished by acquisition of a super-enhancer somewhere within the 2.8 Mb TAD where MYC resides. We find that these diverse cancer-specific super-enhancers, differing in size and location, interact with the MYC gene through a common and conserved CTCF binding site located 2 kb upstream of the MYC promoter. Genetic perturbation of this enhancer-docking site in tumor cells reduces CTCF binding, super-enhancer interaction, MYC gene expression, and cell proliferation. CTCF binding is highly sensitive to DNA methylation, and this enhancer-docking site, which is hypomethylated in diverse cancers, can be inactivated through epigenetic editing with dCas9-DNMT. Similar enhancer-docking sites occur at other genes, including genes with prominent roles in multiple cancers, suggesting a mechanism by which tumor cell oncogenes can generally hijack enhancers. These results provide insights into mechanisms that allow a single target gene to be regulated by diverse enhancer elements in different cell types.

Keywords: chromosome structure; enhancer docking; gene regulation; super-enhancers.

Figure 1. Cell-Type-Specific Super-Enhancers in the MYC Locus Loop to a Common CTCF Site

(A) The 4.5 Mb region surrounding the MYC gene. The 2.8 Mb TAD containing MYC and portions of the two adjacent TADs are indicated with thick black horizontal lines. Super-enhancers (data from Becket et al., 2016; Frietze et al., 2012; Lin et al., 2012; Pope et al., 2014; Wang et al., 2011) are shown in colored boxes for a panel of tumor cell lines that express MYC. (B) Chromosome interaction data at the ~3 Mb MYC locus. For HCT-116, SMC1 HiChIP interactions with an origami score of at least 0.9 and a minimum PET count of 9 are shown as purple arcs; the insulated neighborhood spanning interaction, which encompasses the TAD, is shown as a blue arc (data from this study). For MCF7, Pol II ChIA-PET interactions with an origami score of at least 0.9 are shown as purple arcs; the insulated neighborhood spanning interactions from CTCF ChIA-PET data are shown in blue (data from ENCODE and Li et al., 2012). For K562, RAD21 ChIA-PET interactions with an origami score of at least 0.9 are shown as purple arcs; the insulated neighborhood spanning interaction is shown in blue and has an origami score of 0.44 (data from Heidari et al., 2014). For Jurkat, SMC1 ChIA-PET interactions with an origami score of at least 0.99 are shown as purple arcs; the insulated neighborhood spanning interactions are shown in blue (data from Hnisz et al., 2016a). CTCF ChIP-seq peaks are depicted as purple rectangles, super-enhancers are depicted as red rectangles, and typical enhancers are depicted as gray rectangles (data from this study; Hnisz et al., 2016a; Pope et al., 2014). (C) CTCF ChIP-seq and SMC1 ChIA-PET read counts in the MYC promoter regions. Purple tracks display CTCF ChIP-seq signal in the four cell lines from (B). Light blue track displays the read counts from read pileup of Jurkat SMC1 ChIA-PET data, showing that the major peak of SMC1 ChIA-PET reads occurs at the −2 kb CTCF site. Dark blue bars indicate CpG islands. ChIP-seq read counts are shown in reads per million sequenced reads per base pair. ChIA-PET reads are shown as read counts per base pair. (D) The top panel depicts all putative CTCF binding motifs as blue arrows, which indicate the orientation of the motif. The CTCF motif depicted in dark blue occurs in the most conserved region and shows the best match with consensus CTCF motif. 100 vertebrate conservation from the UCSC genome browser is depicted in the middle panel. The JASPAR score for all the motifs is indicated with blue bars. The position weight matrix for the canonical JASPAR CTCF motif and the actual sequence is shown below. See also Figure S1

Figure 2. Perturbation of the Core CTCF Motif in the MYC CTCF Loop-Anchor Reduces CTCF Occupancy and MYC Expression

(A) Schematic representation of the experiment. HCT-116, K562, Jurkat, and MCF7 cells were transduced with a construct expressing MYC under a PGK promoter and selected for successful integration. These cells were then transiently transfected with plasmid carrying Cas9 and a gRNA targeting the CTCF binding motif. Positive cells were identified and selected using fluorescence-activated cell sorting (FACS). These cells were multiplied, and clonal populations were characterized. (B) The DNA sequences in the vicinity of the core CTCF motif and the mutations generated in clonal populations of K562, HCT-116, Jurkat, and MCF7 cell lines. The reference (WT, wild-type) sequence highlighted in blue is complementary to the gRNA sequence targeting the most prominent CTCF motif (shown here in bold colored sequence). For the aneuploid MCF7 cell line, the two most common mutations are depicted. (C) ChIP-qPCR showing reduction of the CTCF occupancy in Δ-CTCF K562, HCT-116, Jurkat, and MCF7 cells. p values were generated with a Student’s t test. Error bars represent the SD of the mean from three technical replicates. (D) qPCR showing reduction of endogenous MYC mRNA levels in Δ-CTCF K562, HCT-116, Jurkat, and MCF7 cells. p values were generated with a Student’s t test. Error bars represent the SD of the mean from three biological replicates. See also Figure S2

Figure 3. Perturbation of the MYC Enhancer-Docking Site Reduces Looping to Super-Enhancers

(A) 4C analysis showing reduced looping of the MYC promoter proximal CTCF site to super-enhancers in CTCF motif deletion cells (ΔK562) versus unmodified cells (K562). H3K27Ac ChIP-seq and CTCF ChIP-seq are shown in blue and purple colors respectively. Blowups show the 4C interactions for three K562 specific super-enhancers. The 4C viewpoint is situated 112 base pairs upstream of the deleted loop-anchor region. (B) 4C analysis showing reduced looping of the MYC promoter proximal CTCF site to super-enhancers in CTCF motif deletion cells (ΔK562) versus unmodified cells (K562) using a viewpoint centered on the most distant super-enhancer downstream of the MYC gene. H3K27Ac ChIP-seq and CTCF ChIP-seq are shown in blue and purple, respectively. Blowups show the 4C interactions at the MYC promoter and distant super-enhancer near the viewpoint. Shading represents the 90% confidence interval based on three biological replicates. Peak calls from the H3K27Ac ChIP-seq were used to define the regions to be quantified and are indicated in gray boxes at the bottom of the panels. Boxplots show quantification of the reads per fragment for the indicated regions. p values were generated using Student’s t test, and data pairs with a p value < 0.05 are indicated with an asterisk. Reads are shown in reads per million sequenced reads per base pair. Typical enhancers and super-enhancers are shown as gray boxes and red boxes, respectively. See also Figure S3

Figure 4. dCas9-Mediated Methylation of the CTCF Loop-Anchor Site Reduces MYC Expression in Tumor Cells

(A) Top panel shows CTCF ChIP-seq at the MYC gene region in HCT-116 cells. ChIP-seq reads are shown in reads per million sequenced reads per base pair. Bottom panels shows a blowup of the ~700-bp region underneath the CTCF peak depicting the CTCF motifs (blue arrows) and the gRNAs (red rectangles) used to target dCas9-DNMT3A-3L to the enhancer anchor. Lollipop symbols indicate the location of CpGs that are assayed for methylation levels in (C). (B) Schematic representation of the experiment. HCT-116 or K562 cells were transfected with plasmids encoding the dCAS9-DNMT3A-3L, GFP, and a gRNA together with a plasmid encoding 2 additional gRNAs. HCT-116 or K562 cells were isolated by FACS after 2 days, and DNA and RNA were isolated. (C) Methylation at MYC promoter loop-anchor site in untreated cells and cells transfected with dCas9-DNMT3A-3L in conjunction with the 5 indicated gRNAs. (D) qPCR analysis of MYC mRNA levels and fraction of methylated CpGs for untreated and dCas9-DNMT3A-3L + 5 gRNA transfected cells. Error bars represent the SD of the mean for three biological replicates. See also Figure S4

Figure 5. Putative Enhancer-Docking Sites Occur at Additional Genes with Prominent Roles in Cancer

(A) Identification of genes with putative enhancer-docking sites. Genes were filtered for their expression status, presence of a CTCF binding site within 2.5 kb of the TSS and evidence of looping to an active enhancer, defined by H3K27Ac ChIP-seq. (B) Venn-diagram showing the overlap of TSSs with putative CTCF enhancer docking in K562, HCT-116, and Jurkat cells. (C) Venn-diagrams showing the number of TSSs from active genes, the number of these that exhibit putative CTCF enhancer-docking and how many of these have a constitutive CTCF site within 2.5 kb of the TSS. (D) Conservation analysis of the CTCF motifs in the CTCF-bound elements in putative enhancer-docking sites. The mean 46-way PhastCons score of the highest JASPAR scoring motifs in CTCF peaks within putative CTCF enhancer docking and their flanking regions are shown. (E) Examples of genes with putative CTCF enhancer-docking sites from the different cell types analyzed. CTCF ChIP-seq peaks are shown as purple rectangles, typical enhancers are shown as gray rectangles, and super-enhancers are shown as red rectangles. Black arrows indicate the CTCF sites that may facilitate enhancer docking. The insulated neighborhood loop is shown in blue and loops internal to it are shown in purple. HCT-116 HiChIP interactions internal to the neighborhood with an origami score of at least 0.9 and a minimum PET count of 15 are shown for the TGIF1 locus. Jurkat SMC1 ChIA-PET interactions internal to the neighborhood with an origami score of at least 0.97 are shown for the RUNX1 locus. K562 RAD21 ChIA-PET interactions internal to the neighborhood with an origami score of at least 0.9 and a minimum PET count of 30 are shown for the VEGFA locus. K562 RAD21 ChIA-PET interactions internal to the neighborhood with an origami score of at least 0.9 and aminimum PET count of 30 are shown for the PIM1 locus. Data are from this study and two others (Hnisz et al., 2016a; Heidari et al., 2014). See also Figure S5