Allele-Specific Chromatin Recruitment and Therapeutic Vulnerabilities of ESR1 Activating Mutations

Abstract

Estrogen receptor α (ER) ligand-binding domain (LBD) mutations are found in a substantial number of endocrine treatment-resistant metastatic ER-positive (ER⁺) breast cancers. We investigated the chromatin recruitment, transcriptional network, and genetic vulnerabilities in breast cancer models harboring the clinically relevant ER mutations. These mutants exhibit both ligand-independent functions that mimic estradiol-bound wild-type ER as well as allele-specific neomorphic properties that promote a pro-metastatic phenotype. Analysis of the genome-wide ER binding sites identified mutant ER unique recruitment mediating the allele-specific transcriptional program. Genetic screens identified genes that are essential for the ligand-independent growth driven by the mutants. These studies provide insights into the mechanism of endocrine therapy resistance engendered by ER mutations and potential therapeutic targets.

Keywords: CDK7; breast cancer; cistrome; endocrine therapy resistance; estrogen receptor; estrogen recptor mutations.

Figure 1. Global transcriptomic analysis of the ER-mutant cell lines

(A) Pairwise spearman correlation of RNAseq between the ER-WT parental and mutant cell lines in hormone depleted (HD) conditions with vehicle (veh) treatment and HD conditions with estradiol (E2) treatment. Hierarchical clustering shows the relatedness of each sample. (B) Principal components analysis of the transcriptomes of WT-ER parental cell lines and mutant cell lines in full medium (FM), HD, and HD+E2 conditions. See also Figure S1 and Table S1-S2.

Figure 2. Transcriptomic analysis of metastatic tumors harboring the ESR1 LBD mutations

(A) Pairwise spearman correlation of RNAseq between ER⁺ metastatic tumors. (B) Heat map of the top 1,000 genes differentially expressed between D538G-mutant and WT-ER metastatic samples and gene set enrichment (GSEA) plot of the top ranked 100 genes upregulated with the induction of the D538G mutation in MCF7 cells testing for enrichment in the D538G metastatic tumor samples compared to the WT-ER metastatic samples within cluster A. (C). Heat map of the top 1,000 genes differentially expressed between Y537S-mutant and WT-ER metastatic samples and gene set enrichment plot testing the enrichment of the top ranked 100 genes upregulated with the induction of the Y537S mutation in MCF7 cells in the Y537S mutant metastatic tumor samples compared to the WT-ER metastatic samples within cluster B. See also Figure S2.

Figure 3. Distinct ER-mutant cistromes drive a unique transcriptional network

(A) Heat maps of WT-ER binding events after estradiol (E2) stimulation compared to ER-mutant binding sites in hormone depleted (HD) conditions for the Y537S, Y537N and D538G mutations in MCF7 cells and Y537S in T47D cells shown in a horizontal window of ±0.5 kb from the peak center. Heat maps depict the sites that are gained in the mutant, not different between the ER mutant and WT-ER after estradiol stimulation, and sites lost in the mutant. (B) Heat map of the ER binding intensity ratios of Y537S-mutant in HD conditions over WT-mutant after E2 stimulation and D538G-mutant in HD conditions over WT-mutant after E2 stimulation in the sites gained and lost in Y537S and D538G. (C) Heat maps of WT-ER and Y537S mutant ER binding events and corresponding FOXA1 binding in MCF7 cells showing the ER binding sites selective to the Y537S, shared with WT-ER and selective to WT-ER, shown in a horizontal window of ±0.5 kb from the peak center. (D) Heat maps of the differential H3K27ac binding sites comparing WT cells with E2 stimulation and ER-mutant binding sites in HD conditions for the Y537S MCF7 cells. Shown in a horizontal window of ±2 kb from the peak center. (E) Each box in the box plot is the cumulative H3K27ac binding overlapping with the Y537S mutant gained ER binding sites in the WT and Y537S mutant MCF7 cells. The Y axis represents the reads per kilobase per million (RPKM) reads of DNA. The line in the box represents the median, box limits indicate the first and third quartile, whiskers extend 1.5 the interquartile range from the first and third quartiles, the dots represent outliers. (F) The distribution of the regions of the ER binding sites per category, including the ER binding sites gained in the Y537S cells, sites that are not different from WT-ER and the sites lost in the Y537S cells. (G) Venn diagram showing the overlap between the super enhancer (SE) regions in Y537S MCF7 cells and the Y537S gained ER binding sites. (H) Heat map of ER mutant co-regulator binding in HD conditions compared to WT-ER co-regulator binding after E2 treatment. * p<0.05 ** p<0.005. (I) Bar graphs show binding affinity levels for peptides of the ER co-regulators EP300, MAPE, MED1 and NRIP1 in WT-ER +E2 stimulation and Y537S and D538G mutant ER in HD conditions in MCF7 cells. Error bars represent SEM, n=3. See also Figure S3 and Tables S3 and S4.

Figure 4. The mutant specific transcriptional program promotes a metastatic phenotype

(A) Correlation of the Y537S-mutant specific cistrome and differential gene expression. The red line represents the genes (2598 genes) that have Y537S-mutant unique peaks (log2FC > 1; 3,491 peaks). The blue line represents the genes (966 genes) that have peaks associated with them that were unique to WT-ER (log2FC < -1; 2180 peaks). The black line represents the genes (5250 genes) associated with peaks common to mutant and WT-ER (-1<|log2FC| < 1). The X-axis is the log2FC of the genes in mutant condition vs WT condition. The Y-axis is the average normalized BETA score per gene bin. (B) Overlap of the genes with a ranked product <0.001 determined by BETA using the mutant (Y537S, Y537N and D538G) unique binding sites and gene sets significantly enriched in the 344 overlapping genes using ranked gene set enrichment analysis (GSEA). The red line in the graph represents the q value and the blue bars represent the normalized enrichment score (NES). (C) Tumor growth of orthotopic xenografts of MCF7 cells expressing the indicated ER mutants or WT MCF7 cells with estradiol (E2) pellets and without E2 supplements prior to the survival surgery. The Y-axis represents the tumor volume measured by calipers. Comparison of Y537S to D538G mutant cells was statistically significant * p value <0.05 for this comparison. ** p value <0.005 in the comparison of the WT xenografts without E2 to all other conditions. (D) Tumor growth of the orthotopic xenogratfs of MCF7 cells expressing the Y537S or D538G mutation from the time of the intra-mammary injections of the cells through the survival surgery and monitoring of local recurrence and distant metastases (mets). (E) Representative pictures of the mice with the Y537S xenografts. Survival surgery was performed on day 35 and the DOX diet was discontinued on day 98. (F) Representative pictures of the mice with the D538G xenografts. The survival surgery was performed on day 70. (G) Tumor growth of the orthotopic xenogratfs of WT-MCF7 cells with E2 (WT+E2) or without E2 (WT) supplements from the time of the intra-mammary injections of the cells through the survival surgery and monitoring of local recurrence and distant metastases. (H) Representative pictures of the mice with WT MCF7 cells +E2 pellets. The survival surgery performed on day 63. (I) Representative pictures of the mice with WT MCF7 cells without E2 pellets. Survival surgery was not performed in these mice. Error bars in (C-E) represent SEM, 6-8 mice were included in each arm. See also Figure S4 and Table S5.

Figure 5. CRISPR screen identifies ER-mutant essential genes

(A) Essentiality scores from the CRISPR screen in T47D-Y537S mutant cells grown in HD conditions. (B) Comparison of beta scores of CRISPR-CAS9 library screens in Y537S-ER mutant cells in HD conditions versus library screen in ER-WT cells in FM conditions. Dark gray dots=genes that are essential for both WT-ER and Mutant-ER, green= uniquely essential for mutant ER, orange= uniquely positively selected in mutant ER and light gray= not significant in the mutant cells (FDR<0.05). (C) –log rank product values integrating the beta essentiality scores and the geometric mean of the gene rankings from at least 3 of 4 BETA analysis results from the Y537S, D538G and Y537N MCF7 cells and T47D Y537S cells. (D) ChIP-seq tracks showing ER, HA and H3K27 acetylation binding at the TFAP2C promoter region in WT cells with E2 treatment and mutant cells in HD conditions. (E) Immunoblotting for AP2γ in control (CON) cells and cells with CRISPR/Cas9 suppression of TFAP2C (KO) (F) Cell proliferation studies in HD conditions of WT and Y537S or D538G mutant (DOX treated) control cells and after suppression of TFAP2C using CRISPR-Cas9. Error bars represent ±SEM, n=3. See also Figure S5 and Table S6.

Figure 6. CDK7 is essential for mutant ER constitutive activity

(A) Immunoblotting for CDK7 in AAV1-control (AAV1-con) cells, and cells with CDK7 silencing with 3 different CRISPR/cas9 gRNAs. Cell proliferation studies of T47D ER-WT cells with AAV1-con and CDK7 silencing in FM and Y537S mutant cells in HD conditions. (B) Immunoblotting for ER and phospho-ER at serine 118 (S118) in WT cells (-DOX) in HD and E2 stimulated conditions and Y537 or D538G mutant (+DOX) ER in HD conditions. (C) ER and pS118 levels by immunoblotting after vehicle (VEH) and THZ1 treatment at 4 hr and 24 hr with increasing doses. (D) 2D and 3D synergy maps of THZ1 and fulvestrant (FUL) treatment combination in MCF7 (left) and T47D (right) cells expressing the Y537S mutation. (E) Growth curves of orthotopic xenografts of the MCF7 –Y537S cells after treatment with vehicle (VEH), FUL, THZ1 or the combination of FUL and THZ1. Error bars represent ±SEM, n=8. See also Figure S6.