Multiplex Enhancer Interference Reveals Collaborative Control of Gene Regulation by Estrogen Receptor α-Bound Enhancers  Julia B. Carleton, Kristofer C. Berrett, Jason Gertz  Cell Systems  Volume 5, Issue 4, Pages 333-344.e5 (October 2017) DOI: 10.1016/j.cels.2017.08.011 Copyright © 2017 Elsevier Inc. Terms and Conditions

Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 1 Genes Upregulated by Estrogens Likely Receive Input from Multiple ER-Bound Enhancers (A and B) The number of ER binding sites within 100 kb of the transcription start sites (TSS) of genes upregulated, downregulated, or not regulated by estrogen following an E2 induction in Ishikawa cells (A) and T-47D cells (B) shows that upregulated genes are enriched for multiple ER binding sites in the vicinity. See also Figure S1. (C and D) The relationship between fold change in response to E2 and the number of ER binding sites within 100 kb of the TSS for genes upregulated by ER in Ishikawa cells (C) and T-47D cells (D) demonstrates that having more ER binding sites nearby leads to larger gene expression changes. (E) Many genes upregulated by E2, in comparison with genes downregulated and not regulated, have multiple ER binding sites that are co-occupied by RNAP2 (left) or marked by H3K27ac (right). Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 2 Enhancer-i Can Simultaneously Target Multiple ER Binding Sites (A–C) MMP17 (A), CISH (B), and FHL2 (C) are upregulated by E2 in Ishikawa cells and harbor multiple ER binding sites within 100 kb of their TSS. Gray bars indicate ERBS targeted by Enhancer-i. (D) The gene expression response to E2 (white, control guide RNAs, n = 8 replicates from 4 independent experiments) is significantly reduced by targeting of 10 ER binding sites (green, Enhancer-i, n = 8 replicates from 4 independent experiments), or the 4 promoters of G0S2, MMP17, CISH, and FHL2 (blue, Promoter-i, n = 4 replicates from 2 independent experiments) as measured by qPCR. Error bars represent the SEM; ∗∗p < 0.01 and ∗p < 0.05, paired t test. (E) Enhancer-i specifically reduces the transcriptional response to E2 of targeted genes without dramatically altering the transcriptome as measured by RNA-seq. See also Figure S2. Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 3 Enhancer-i Is Specific to Targeted Sites and Reduces ER Occupancy as well as Histone Acetylation (A) The specificity of targeting for Enhancer-i can be seen by ChIP-seq of FLAG-tagged SID4X-dCas9-KRAB when looking at all ER binding sites; targeted sites are shown in red. (B) ER occupancy is reduced by Enhancer-i at all targeted sites as shown by the fold reduction in ER occupancy, quantified here by comparing normalized ChIP-seq signals of ER following an E2 induction in control cells and Enhancer-i cells (2 independent replicates per condition). The untargeted ERBS near DOK7 (gray bar) does not show a reduction in ER occupancy following Enhancer-i treatment. Error bars represent the SEM; ∗p < 0.05, paired t test. (C) Enhancer-i (x axis) reduces H3K27ac at most targeted sites (purple) following E2 treatment in comparison with control guide RNAs (y axis). See also Figures S3–S6. (D) Example of Enhancer-i effects on ER binding (blue), H3K27ac (purple), and dCas9 binding (red), at FHL2-1 as measured by ChIP-seq. Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 4 Enhancer-i Reveals Necessary ERBS and Suggests that Sites Collaborate (A–C) The impact on estrogen-induced gene expression after targeting individual and combinations of ERBS nearby MMP17 (A), CISH (B), and FHL2 (C), is shown, with the left schematic depicting the targeted sites with black hexagons. Untargeted ERBS are indicated by empty rectangles. Horizontal bars represent 4 biological replicates from 2 independent experiments, except C2+3 and C1+3, which represent 3 biological replicates from 2 independent experiments. (D and E) CRISPR/Cas9-mediated genetic deletions of 3 ERBS near MMP17 (D) and CISH (E) confirm the necessity of individual sites. Bars labeled WT, ΔM1–ΔM3, and ΔC2 and ΔC3 represent 4 biological replicates from 2 individual clones (n = 8, 2 independent experiments), while ΔC1 represents 4 biological replicates from 3 individual clones (n = 12, 2 independent experiments). The bar labeled “ΔCISH controls” (D) represents MMP17 expression in CISH ERBS deletions (n = 3 independent clones). The bar labeled “ΔMMP17 controls” (E) represents CISH expression in MMP17 ERBS deletions (n = 3 independent clones). See also Figure S7. (F) The percent reduction in the estrogen response achieved with Enhancer-i and genetic deletions is highly correlated, as shown by scatterplot. Error bars represent SEM; ∗∗p < 0.01 and ∗p < 0.05, paired t test. Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 5 Necessary ER Binding Sites Have Distinct Genomic Features (A) Area under the receiver-operating characteristic curve (ROC AUC) values for genomic features indicate that ERBS location relative to transcription start sites is the best predictor of necessity. See also Figure S8 and Data S1. (B) The relationship between necessity and distance can be seen by scatterplot, with nearby ERBS affecting gene expression more than distal ERBS. (C) ROC AUC analysis after removing 2 ERBS that are more than 70 kb from the TSS of the target genes shows that ERE strength is most predictive. (D) The relationship between necessity and ERE strength is shown as a scatterplot, with more positive ERE scores indicating a better match to the canonical ERE motif. Cell Systems 2017 5, 333-344.e5DOI: (10.1016/j.cels.2017.08.011) Copyright © 2017 Elsevier Inc. Terms and Conditions