A Non-canonical Role of YAP/TEAD Is Required for Activation of Estrogen-Regulated Enhancers in Breast Cancer

Abstract

YAP/TEAD are nuclear effectors of the Hippo pathway, regulating organ size and tumorigenesis largely through promoter-associated function. However, their function as enhancer regulators remains poorly understood. Through an in vivo proximity-dependent labeling (BioID) technique, we identified YAP1 and TEAD4 protein as co-regulators of ERα on enhancers. The binding of YAP1/TEAD4 to ERα-bound enhancers is augmented upon E₂ stimulation and is required for the induction of E₂/ERα target genes and E₂-induced oncogenic cell growth. Furthermore, their enhancer binding is a prerequisite for enhancer activation marked by eRNA transcription and for the recruitment of the enhancer activation machinery component MED1. The binding of TEAD4 on active ERE-containing enhancers is independent of its DNA-binding behavior, and instead, occurs through protein-tethering trans-binding. Our data reveal a non-canonical function of YAP1 and TEAD4 as ERα cofactors in regulating cancer growth, highlighting the potential of YAP/TEAD as possible actionable drug targets for ERα⁺ breast cancer.

Keywords: ERα; Hippo signaling; YAP/TEAD; breast cancer; enhancer; estrogen signaling; transcriptional regulation.

Figure 1.. Identification of YAP/TEAD as previously unknown ERα-interacting cofactors by proximity BioID proteomics

(A) Schematic diagram demonstrating BioID (proximity-dependent biotin identification) approach for identification of ERα-interacting nuclear proteins. (B) Western blots confirming the inducible expression and in vivo biotinylation in the established ERα-BioID tet-on stable cell line. The fractionation of cytoplasmic (Cyto) and nuclear (Nuc) fractions of MCF7 cells was confirmed with Western blots for GAPDH (cytoplasm-specific marker) and Histone H3 (nucleus-specific marker). The doxycycline-inducible ERα-BirA*-HA fusion protein expression was detected by antibodies recognizing HA and ERα (the endogenous ERα was labeled with * and the tagged exogenous ERα was labeled with #). Biotinylated proteins by ERα-BirA* were detected by streptavidin-HRP blot. (C) Identification of ERα-interacting cofactors by mass spectrometry analyses on the protein complex pulled down from the nuclear fractions of ERα-BioID tet-on stable cell line under the indicated treatments. Biotinylated proteins in nuclear fraction were enriched and purified using streptavidin beads before subjected to mass spectrometry. Besides many listed known cofactors, YAP1 and TEAD4 are two previously unknown ERα-interacting cofactors identified from our studies. Peptide numbers detected from mass spectrometry analyses are listed in the table for each protein. (D) Co-IP assays in MCF7 cells with the indicated treatments confirming protein-protein interactions of endogenous ERα, YAP1 and TEAD4. Nuclear fractions treated with vehicle or E₂ treatment were used for immunoprecipitation with antibodies against ERα, TEAD4 and YAP1 respectively. (E) Western blots confirming the inducible expression and in vivo biotinylation in the established FOXA1-BioID tet-on stable cell line. The fractionation of cytoplasmic (Cyto) and nuclear (Nuc) fractions of MCF7 cells was confirmed with Western blots for GAPDH and Histone H3 respectively. The doxycycline-inducible Myc-BirA*-FOXA1 fusion protein expression was detected by antibodies recognizing Myc and FOXA1 (the endogenous FOXA1 was labeled with * and the tagged exogenous FOXA1 was labeled with #). Biotinylated proteins by BirA*-FOXA1 were detected by streptavidin-HRP blot. (F) Identification of FOXA1-interacting factors by mass spectrometry analyses on streptavidin bead pulldowns from the nuclear fractions of FOXA1-BioID tet-on stable cell line with the indicated treatments. Peptide numbers detected from mass spectrometry analyses are listed in the table. See also Figure S1.

Figure 2.. Identification of non-canonical binding of YAP/TEAD on ERα enhancers

(A) Venn diagram showing the overlap of ERα binding peaks with those of YAP1 and TEAD4. Two different YAP/TEAD binding groups were identified: the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ group and the 4,552 YAP1⁺/TEAD4⁺/ERα⁻ group. The binding peaks in MCF7 cells (+E₂) were mapped based on ChIP-seq with antibodies recognizing these three proteins. (B) Aggregate plots showing the normalized tag density of ERα, YAP1 and TEAD4 ChIP-seq data (±E₂) for the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites or the 4,552 YAP1⁺/TEAD4⁺/ERα⁻ sites. The binding of YAP1, TEAD4 and ERα on the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites, but not on the 4,552 YAP1⁺/TEAD4⁺/ERα⁻ sites, was significantly enhanced upon E₂ stimulation (see fold changes (FC) and P-values (p) between E₂ and Veh control). (C) Heatmaps of ERα, FOXA1, YAP1, TEAD4, and H3K27Ac ChIP-seq data (±E₂), supporting the distinct binding patterns for YAP1/TEAD4 in the two groups shown in (B). The top 10 enriched motifs for each group (± 300 bp from the center of ChIP-seq binding sites) are listed. (D) Genome browser view of ChIP-seq peaks on a well-known ERα target gene GREB1 locus showing the E₂-enhanced binding of ERα, FOXA1, YAP1, and TEAD4 at several ERα enhancers. (E) Genome browser view of ChIP-seq peaks on a well-known Hippo pathway target gene CTGF locus showing that the canonical binding of YAP1 and TEAD4 at CTGF promoter is independent of ERα or E₂ stimulation. See also Figures S2 and S3.

Figure 3.. The non-canonical binding of YAP/TEAD to ERα enhancers is required for transcriptional activation of ERα target genes and E₂-induced breast cancer growth

(A) GREAT analysis against MSigDB Pathway identified the potential pathways enriched in genes associated with 2,863 sites and 4,553 sites. (B) Box plots of RNA-seq showing the effects of YAP1 or TEAD4 knockdown on E₂-induced changes in gene expression. Genes associated with the 2,863 sites were divided into 3 groups based on expression changes in response to E₂ treatment. (C) Box plots of RNA-seq showing the E₂ effects on expression of the coding genes targeted by 3 groups of ERα enhancers with high, median or low activity (see STAR METHODS for enhancer group definition) in cells transduced with shControl, shYAP1 or shTEAD4. YAP1 or TEAD4 knockdown significantly attenuated the E₂-induction in the group of genes associated with “High Activity” enhancers, but not the groups with “Median” or “Low Activity” enhancers. (D) Cell proliferation assay using CCK-8 showing that YAP1 and TEAD4 are required for E₂-induced breast cancer growth. MCF7 cells were knocked down with control shRNA or shYAP1 or shTEAD4 and cultured in medium with or without E₂ after stripping. (E) Colony formation assay confirming that YAP1 and TEAD4 are required for E₂-induced breast cancer cell growth. MCF7 cells were treated with control shRNA or shYAP1 or shTEAD4 and cultured in stripping medium with or without E₂ followed by fixation and staining with crystal violet. Numbers of colonies were measured using ImageJ. (F) YAP1 or TEAD4 knockdown inhibited cancer xenograft growth. MCF7 cells treated with shControl, shYAP1 or shTEAD4 were injected into the mammary fat pads of female nude mice with 17β-Estradiol pellets implanted on back underneath skin. Tumors were measured 5 weeks post cell implantation. Statistics for (D-F): ***P < 0.001, **P < 0.01, *P < 0.05, One-way ANOVA. See also Figure S4.

Figure 4.. YAP1 and TEAD4 are required for ERα enhancer activation

(A) Aggregate plots for GRO-seq tag density showing both sense (plus) and anti-sense (minus) eRNA expression levels for the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites in MCF7 cells transduced with control shRNA, shYAP1 or shTEAD4 (±E₂). E₂ treatment in control cells significantly increased eRNA transcription, and this E₂-induced eRNA production was greatly abolished in cells with YAP1 or TEAD4 knockdown. (B-C) Genome browser views of GRO-seq signals at representative ERα active enhancers and their target genes GREB1 and SMAD7. Transcriptional activities at enhancers and at target gene bodies were positively correlated (see changes in eRNA level and changes in gene body expression upon E₂ stimulation and YAP1/TEAD4 knockdown), suggesting that YAP1 and TEAD4 are required for E₂-induced transcription from both ERα enhancers (eRNA) and the adjacent target coding genes (mRNA). See also Figure S5.

Figure 5.. YAP1 and TEAD4 regulate MED1 binding on active ERα enhancers

(A) Heatmap of ERα ChIP-seq data (+E₂) on the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites showing that ERα binding is not dependent on YAP1 or TEAD4. MCF7 cells were treated with control shRNA, shYAP1 or shTEAD4. (B) E₂-induced MED1 binding on enhancers requires YAP1. Aggregate plots showing the normalized tag density of MED1 ChIP-seq data (±E₂) on the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites. MCF7 cells treated with control shRNA or shYAP1 and maintained in stripping medium with or without E₂ were used for ChIP-seq. (C-D) Genome browser views of ChIP-seq signals for ERα/YAP1/TEAD4 (+E₂) and MED1 (±E₂) on enhancers at GREB1 and SMAD7 loci. On these YAP1/TEAD4/ERα co-bound enhancers, E₂-induced MED1 binding was significantly diminished by shRNA knockdown of YAP1. See also Figures S6A–B.

Figure 6.. Binding of YAP1 and TEAD4 on ERα enhancers requires Erα

(A) Western blots showing the downregulation effects of ICI 182,780 (ICI) on ERα protein level in the nucleus, especially under E₂ stimulation condition. ICI treatment significantly decreased ERα protein level but did not affect YAP1 or TEAD4 protein level in cells in the presence or absence of E₂. (B-D) Aggregate plots showing the normalized tag density of ERα, YAP1 and TEAD4 ChIP-seq data for the two groups of YAP1/TEAD4 binding sites. On the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ non-canonical sites, ICI treatment abolished ERα binding and reduced YAP1/TEAD4 binding at the same time. However, on the 4,552 YAP1⁺/TEAD4⁺/ERα⁻ canonical sites, their binding was not affected by ICI treatment (see fold changes (FC) and P-values (p) between DMSO and ICI). (E-F) Genome browser views of ChIP-seq signals on two well-known E₂/ERα target genes loci, GREB1 and SMAD7, confirming that binding of ERα and YAP1/TEAD4 on non-canonical binding sites was greatly reduced in response to ICI-induced ERα degradation. See also Figures S6C–E.

Figure 7.. TEAD4 is recruited to ERα enhancers through protein tethering trans-binding

(A) Heatmaps and aggregate plots of the ChIP-seq data for wildtype TEAD4 (WT) and DNA-binding domain mutant TEAD4 (Mut) to compare the effects of DNA-binding domain mutation on TEAD4 binding at three different groups of binding sites: the 4,552 YAP1⁺/TEAD4⁺/ERα⁻ sites, the 2,863 YAP1⁺/TEAD4⁺/ERα⁺ sites and the 419 the most active ERα enhancers. The fold changes (FC) and P-values (p) comparing WT and Mut binding (+Dox, +E₂) are shown. (B-C) Genome browser images of WT and Mut TEAD4 ChIP-seq signals on representative gene loci. The DNA-binding mutation completely abolished the binding of TEAD4 on a classic Hippo pathway target gene CTGF, but did not affect its binding on the enhancers associated with the ERα target gene GREB1. (D) A proposed model for the canonical and non-canonical function of YAP1 and TEAD4. In addition to their canonical role as downstream nuclear effectors of Hippo pathway, YAP1 and TEAD4 can also function as ERα co-regulators at active ERE-containing enhancers to regulate enhancer activity and target gene transcription upon E₂ stimulation. The binding of YAP1 and TEAD4 is critical for the assembly of enhancer activation machinery components, such as MED1. Different to its cis-binding to the canonical sites, the binding of TEAD4 on the most active ERα enhancers is in trans through protein-protein tethering interaction, which is independent of the DNA-binding activity of TEAD4. See also Figure S7.