Mutant p53 binds and controls estrogen receptor activity to drive endocrine resistance in ovarian cancer.

Results

Mutant p53 interacts with ERα in ovarian cancer
We previously identified a gene signature that can delineate B lymphocytes from carriers of cancer-prone germline variants of TP53 compared with cells from siblings with wild-type (WT) p53. A significant component of the gene signature was decreased expression of target genes for estrogen receptor (Leung et al. 2023). There is considerable evidence in the literature that WT p53 and estrogen receptor interact and that estrogen receptor α (ERα) inhibits WT p53 activity (Liu et al. 2006, 2009; Sayeed et al. 2007; Konduri et al. 2010; Bailey et al. 2012; Das et al. 2025), but the impact of mutant p53 on ERα function has not been demonstrated. To begin to explore this research area, we first assessed the interaction between mutant p53 and ERα in two TP53 mutated ERα-positive breast cancer cells (T47D, p53 L194F mutation and MDA-MB-134IV, p53 E285K mutation). For this analysis, we used proximity ligation assay (PLA) and immunoprecipitation-Western blot (IP-Western). Our PLA and IP-Western results revealed that mutant p53 interacts with ERα in these breast cancer lines and that silencing mutant p53 eliminated the PLA signals (Supplemental Fig. S1A,B).
High-grade serous ovarian cancer (HGSOC) has the highest frequency of TP53 mutation; ∼96% of HGSOCs have mutations in TP53, and ∼60% of these are missense mutations that lead to abnormal accumulation of mutant p53 protein (Ahmed et al. 2010; The Cancer Genome Atlas Research Network 2011; Cole et al. 2016). PLA revealed a positive signal between mutant p53 and ERα in two ERα-positive HGSOC cell lines: PEO1 (p53 G244D mutation) and OV2085 (p53 K132R mutation) (Fig. 1A,B). We also analyzed SKOV3 ovarian cancer cells that were stably transfected with the R175H or Y220C mutants of p53; together, these two mutants account for up to 10% of HGSOCs with mutant p53. We readily observed a mutant p53–ERα interaction in these cells (Fig. 1C; Supplemental Fig. S1C). The mutant p53–ERα interaction was confirmed using IP-Western blots in two cell lines with IP in both directions (Fig. 1D; Supplemental Fig. S1D), and we confirmed that this interaction is direct using purified Y220C mutant p53 (Fig. 1E). We next analyzed paraffin sections of human serous tubal intraepithelial carcinoma (STIC) samples. STIC is regarded as a precursor of HGSOC (Kroeger and Drapkin 2017), and 92% of STIC samples contain a mutation in TP53 (Kuhn et al. 2012). Immunohistochemistry of a STIC sample revealed high levels of p53 protein (suggestive of mutant p53) in the areas of STIC (cellular atypia and high Ki-67 stain) but not in the normal adjacent regions, as well as high levels of ERα (Fig. 1F). As in the HGSOC lines, PLA revealed an interaction between p53 and ERα in three independent STIC samples with high p53 staining, but only in the STIC and not in noncancerous adjacent tissue (Fig. 1G; Supplemental Fig. S1E); quantification of PLA signals from all three combined STIC samples is shown in Figure 1G.

Silencing mutant p53 leads to increased half-life of ERα protein
We next sought to determine the impact of mutant p53 on ERα level and function. We found that silencing mutant p53 using siRNA, but not the siNC (nontargeting control), led to an approximately twofold- to fourfold increase in ERα protein level in PEO1 and OV2085 cells (Fig. 2A,B). There was a modest increase in the RNA level for ESR1 in sip53 cells (Fig. 2A, right panel). Ectopic expression of the p53 hotspot mutants R175H or R273H in SKOV3 cells led to a modest but consistent decrease in ERα levels (Fig. 2B). We used a cycloheximide chase experiment to assess ERα half-life and found that silencing mutant p53 led to an ∼2.5-fold increase in ERα half-life (Fig. 2C). In PEO1 and OV2085 cells transfected with HA-tagged ubiquitin, silencing of mutant p53 led to a consistent decrease in ERα ubiquitylation (Fig. 2D–E). Conversely, in SKOV3 cells transfected with increasing amounts of the R273H hotspot mutant, there is clear evidence for increased ubiquitylation of ERα (Fig. 2F). The decrease in ERα ubiquitylation upon silencing mutant p53 was also evident when HA-ubiquitin was immunoprecipitated (Fig. 2G). The combined data suggest that mutant p53 can negatively impact the half-life of ERα. These data were consistent with our IHC findings in STIC. Here we observed reduced ERα staining in sections of STIC from three patients with high levels of (mutant) p53 compared with adjacent normal tissue. Conversely, there were no changes in ERα immunostaining in three independent p53-null STIC samples (Fig. 2H,I).
To assess the influence of mutant p53 on ERα activity, we used 17β-estradiol (E2) to activate the transcriptional function of ERα. When ERα is activated, Ser118 is phosphorylated and ERα translocates to the nucleus to regulate the expression of canonical estrogen-responsive genes like GREB1 (Joel et al. 1998). Western blot analysis revealed that silencing mutant p53 led to a modest increase in p-ERα (Ser118), which was evident despite the proteasome-mediated degradation of ERα that occurs upon binding to E2 (Supplemental Fig. S2A; Nawaz et al. 1999). There was also a modest but significant increase in the induction of the E2/ERα transcriptional target GREB1 (Supplemental Fig. S2A). Conversely, stable expression of mutant p53 (R175H) in SKOV3 cells led to a modest decrease in p-ERα (Ser118) upon E2 treatment, along with a decrease in GREB1 induction (Supplemental Fig. S2B). Western blot analysis of isolated cytoplasmic and nuclear protein revealed increased ERα translocation to the nucleus when E2 was added, and this was increased by sip53 (Supplemental Fig. S2C). These results were in contrast to MCF7 cells with WT p53, where silencing WT p53 led to a modest decrease in the E2-mediated induction of GREB1 and PDZK1 (Supplemental Fig. S2D).

Mutant p53 negatively regulates the estrogen-mediated induction of a subset ERα target genes
We next performed RNA sequencing (RNA-seq) on PEO1 cells transfected with nontargeting control (siNC) or siRNA for p53 (sip53) and treated with E2 for 0, 2, and 8 h. Consistent with our results above, the majority of known ERα target genes were modestly or not obviously affected by silencing mutant p53 (Fig. 3A). Approximately four dozen genes showed significantly increased E2-mediated upregulation when mutant p53 was silenced (Fig. 3B,C). Estrogen response pathways were the major pathways activated in E2-treated cells when mutant p53 was silenced (Fig. 3D). We used qPCR on independent samples to confirm these findings for the genes DEPTOR, LOXL4, CADM1, and ACP6 (Fig. 3E). We then focused on DEPTOR, which is a known ERα target gene (Cuesta et al. 2019). Western blot analysis confirmed that the expression of DEPTOR is increased after E2 treatment and that silencing mutant p53 leads to markedly increased DEPTOR protein induction after E2 treatment in both PEO1 and OV2085 cells (Fig. 3F,G). In SKOV3 cells, the stable expression of the R175H mutant of p53 nearly ablated the ability of E2 to induce DEPTOR (Fig. 3H). Because DEPTOR inhibits mTOR function, we also assessed mTOR activity by analyzing the phosphorylation of AKT (p-Ser473 and p-Ser308) after E2 treatment in PEO1 cells; in this experiment, sip53 cells showed increased DEPTOR and decreased phospho-AKT after E2 treatment, suggesting that the downstream mTOR pathway was inhibited (Fig. 3I). Analysis of a HGSOC data set in cBioPortal revealed that DEPTOR levels are reduced in patients with advanced stages of HGSOC (Fig. 3J), suggestive of a tumor-suppressive role for DEPTOR in HGSOC, as supported by the literature (Caron et al. 2018; Rogers-Broadway et al. 2019).
We next sought to probe the potential significance of the subset of E2-induced genes that showed enhanced induction in sip53 cells. Toward this goal, we analyzed the ER-positive METABRIC database of breast cancer for the influence of this subset of ERα target genes on survival. Patients with high expression of this subset of ERα target genes showed improved overall and relapse-free survival (Supplemental Fig. S3A). To probe more deeply, each gene was analyzed for its correlation with survival (overall, relapse-free, or disease-free); this analysis revealed that most of these individual genes correlated with improved survival of ER-positive breast cancer patients—more so in the METABRIC database than in TCGA (Supplemental Fig. S3B). We found that four of these “ER-induced/mutant p53-repressed” target genes (KRT5, KRT15, KRT16, and DIRAS3) had strong predictive value for improved outcomes in both the TCGA and METABRIC cohorts (Supplemental Fig. S3C). We then analyzed this four gene signature on a TCGA data set of HGSOC that is quantified for p53 immunostaining. We found a statistically significant correlation between high p53 immunostaining and high p53 expression with a decreased level of the four gene signature; however, there was no association with the “all-gene” signature (Supplemental Fig. S3D–F). A HGSOC data set stratified for the ER+ immunostaining level would enable a better assessment of correlations between mutant p53 and this gene signature.

Mutant p53 specifically disrupts the induction of SP1/ERα target genes
To identify the mechanism whereby mutant p53 inhibits the E2-dependent induction of this subset of ERα target genes, we performed an analysis to identify transcription factor(s) that regulate this group of genes. The top enriched transcription factor for this subset of target genes was SP1 (Fig. 4A). SP1 is known to use ERα as a coactivator for SP1 target genes in response to E2 (Safe 2001). We first confirmed that ERα and SP1 interact in HGSOC cells using proximity ligation analyses (PLAs) (Fig. 4B) and IP-Western (Supplemental Fig. S4A). Next, we found that silencing mutant p53 in PEO1 cells led to a statistically significant increase in ERα–SP1 complexes in PLA assays (P = 0.034) (Fig. 4B); this was recapitulated in IP-Westerns (Supplemental Fig. S4A). We then tested whether three of these target genes (DEPTOR, LOXL4, and CADM1) were SP1 target genes by assessing their E2-mediated induction when SP1 is silenced. Our qPCR and Western blot results confirmed that DEPTOR, LOXL4, and CADM1 were upregulated by E2 treatment and showed significantly reduced E2-mediated induction when SP1 was silenced (Fig. 4C,D). To gain a broader understanding of the impact of mutant p53 on SP1 and ERα chromatin binding, we performed chromatin immunoprecipitation sequencing (ChIP-seq) for ERα and SP1 in E2-treated PEO1 cells that were transfected with siNC or sip53. ChIP-seq analysis revealed an overall increase in ERα binding to target loci in sip53 cells compared with siNC cells in response to E2 (Fig. 4E–G). Increased joint SP1–ERα peaks were evident for LOXL4, DEPTOR, and CADM1, while DEPTOR also showed nonoverlapping peaks that were increased (Fig. 4H). Increased ERα binding when mutant p53 is silenced was confirmed by ChIP-qPCR (Fig. 4I). ChIP-seq revealed little to no difference in SP1 binding when mutant p53 was silenced (Supplemental Fig. S4B–D). Overall, there was an over twofold increase in joint SP1–ERα peaks upon E2 treatment when mutant p53 was silenced (Supplemental Fig. S4E). Importantly, the majority of genes identified by RNA-seq to show increased expression by E2 when mutant p53 was silenced showed joint binding by both ERα and SP1 in ChIP-seq (44 out of 57 genes) (Supplemental Fig. S4F). The combined data best fit a model whereby mutant p53 inhibits the E2-mediated transactivation of joint SP1–ERα target genes, likely by inhibiting SP1–ERα interactions (Supplemental Fig. S4G).

Mutant p53 depletion can overcome endocrine resistance in ovarian cancer
For breast cancer, ERα is an ideal target for endocrine therapy, and agents like fulvestrant and tamoxifen are commonly used and efficacious in the clinic (Hanker et al. 2020). Endocrine therapy for HGSOC has shown somewhat promising results, but the presence of ERα is not predictive of response, suggesting that HGSOC demonstrates endocrine resistance through unknown mechanism (Papadimitriou et al. 2004; Simpkins et al. 2012, 2013). These findings prompted us to test the hypothesis that mutant p53 might confer resistance to endocrine therapy. Cell viability analysis in endocrine-resistant PEO1 cells revealed that silencing mutant p53 increased the sensitivity to fulvestrant by sevenfold (Fig. 5A). No effect of sip53 was evident on the viability of p53-null SKOV3 cells (Supplemental Fig. S5A), while a short hairpin to p53 (shp53) also increased the sensitivity of PEO1 cells to fulvestrant; these data indicate that our findings are not likely to be an off-target effect of sip53 (Supplemental Fig. S5B). Colony suppression assays showed that silencing mutant p53 with sip53 or shp53 led to a dramatically enhanced and dose-dependent suppression by fulvestrant (P < 0.0001) (Fig. 5B; Supplemental Fig. S5C). Conversely, stable expression of mutant p53 (R175H or R273H) in SKOV3 cells conferred resistance to fulvestrant (Fig. 5C). Silencing mutant p53 in cells treated with fulvestrant led to a significant increase in cell cycle arrest (Fig. 5D) and a modest decrease in spheroid size (Supplemental Fig. S5D) but no increase in apoptosis (Supplemental Fig. S5E,F). We next analyzed HGSOC cell lines grown as spheroids. For this analysis, we used PEO1 cells transfected with siRNA and grown as spheroids on low-attachment plates; here we measured spheroid size after fulvestrant treatment. In addition, we analyzed PEO1 cells stably infected with a short hairpin to p53 (shp53) and grown as organoids (see the Materials and Methods). In both cases, sip53 and shp53 led to enhanced inhibition by fulvestrant and elacestrant, as assessed by spheroid size and calcein staining of live cells (Fig. 5E,F). To follow up on this further, we silenced one of the SP1–ERα target genes, DEPTOR, and found that this led to increased survival of PEO1 cells following fulvestrant treatment (Supplemental Fig. S6A,B). Consistent with these findings, analysis of TCGA data revealed that increased DEPTOR expression is associated with increased survival in HGSOC (Supplemental Fig. S6C). These data preliminarily link our findings on SP1–ERα target genes to endocrine resistance, but this premise awaits further analysis.
To further explore the impact of mutant p53 on the sensitivity to endocrine therapy, we performed RNA-seq analysis on PEO1 cells stably infected with vector alone or shp53 and treated with vehicle or fulvestrant. This analysis revealed a large cluster of genes with altered expression (induced or repressed) following fulvestrant treatment and with enhanced differences in cells when mutant p53 was stably silenced (Fig. 6A). Ingenuity pathway analysis (IPA) and gene set enrichment analysis (GSEA) revealed that the majority of these differentially regulated genes played roles in the cell cycle (Fig. 6B,C). Consistent with this, flow cytometric analyses revealed that silencing of mutant p53 led to markedly increased G1 arrest in response to either fulvestrant or elacestrant (P < 0.001 for both agents) (Fig. 6D,E).

Mutant p53 confers endocrine resistance in part through p27Kip1 (CDKN1B)
An important component of the anticancer activity of fulvestrant is the induction of CDKN1B, encoding the cyclin-dependent kinase inhibitor p27Kip1 (Cariou et al. 2000; Simpkins et al. 2012). Consistent with this, low p27Kip1 expression is associated with poorer outcomes from endocrine therapy (Filipits et al. 2009). Western blot analysis revealed that treatment with fulvestrant or elacestrant led to a modest increase in p27Kip1 protein levels in PEO1-siNC cells; this upregulation was markedly enhanced in cells in which mutant p53 was silenced (Fig. 7A). qPCR analysis revealed that the induction of p27Kip1 protein following fulvestrant or elacestrant was likely post-transcriptional (Fig. 7B). One gene with significantly decreased expression following fulvestrant treatment in our RNA-seq data was SKP2, which encodes a component of the ubiquitin ligase that targets p27Kip1 for degradation. Both fulvestrant and elacestrant led to decreased SKP2 expression in HGSOC cells, but this effect was more pronounced in shp53 cells at both the RNA (Fig. 7C) and protein (Fig. 7D) levels. To probe the impact of p27Kip1 on the cell cycle arrest in PEO1-shp53 cells, we treated these cells with fulvestrant or elacestrant in the presence of siNC or siCDKN1B and performed cell cycle analysis. Cells expressing shp53 showed dramatic G1 arrest by fulvestrant and elacestrant, but this was completely abrogated by silencing CDKN1B (p27Kip1) (Fig. 7E,F). The combined data best fit a model whereby mutant p53 contributes to the resistance to selective estrogen receptor degraders like fulvestrant and elacestrant, and part of this effect is mediated by the upregulation of p27Kip1, potentially through repression of SKP2 (Fig. 7G).

The p53 refolder rezatapopt enhances the efficacy of endocrine therapy in a Y220C cell line
Our data indicating that mutant p53 enhances the resistance of HGSOC to endocrine therapy brought up the possibility that refolding compounds for mutant p53 might enhance the efficacy of endocrine therapy for HGSOC. The Y220C mutant of p53 is a “hotspot” mutant for HGSOC that occurs in ∼4% of HGSOCs with mutant p53 and for which a refolding compound (rezatapopt) has been discovered that stabilizes Y220C in a wild-type conformation (Puzio-Kuter et al. 2025). We confirmed that rezatapopt decreases the colony-forming ability of SKOV3 cells stably transfected with the Y220C mutant but not parental cells transfected with vector alone (Supplemental Fig. S7A,B). Treatment with either 3 µM or 10 µM rezatapopt led to robust induction of the p53 target genes p21/CDKN1A and MDM2 (Supplemental Fig. S7C,D), but importantly, this lower dose (3 µM) shows little inhibition in colony-forming assays (Supplemental Fig. S7F). Notably, in both IC50 and colony-forming assays, this dose of rezatapopt led to significantly increased sensitivity to fulvestrant in SKOV3 cells harboring the Y220C mutant (Supplemental Fig. S7E,F), thus offering a rationale for a potential combination of rezatapopt with endocrine therapy in HGSOC harboring the Y220 mutant.

Discussion
p53 and the estrogen receptor are two critical proteins in female cancers. In this study, we show that mutant p53 binds and negatively regulates the level and activity of ERα. There is abundant evidence in the literature that wild-type p53 directly interacts with ERα (Liu et al. 2006, 2009; Berger et al. 2012, 2013); therefore, our finding that mutant p53 interacts with ERα is not unexpected. The available evidence indicates that ERα inhibits WT p53 function (Liu et al. 2006, 2009; Sayeed et al. 2007; Konduri et al. 2010; Bailey et al. 2012) and that breaking the interaction between ERα and WT p53 can restore p53 function (Konduri et al. 2010); this led to a window of opportunity trial for tamoxifen use in early stage breast tumors with wild-type p53 (Oturkar et al. 2024). Here we show in three different cell lines (PEO1, OV2085, and SKOV3) and in three independent STIC samples that mutant p53 binds and regulates the level of ERα. Our RNA sequencing data indicate that mutant p53 modestly inhibits the E2-mediated induction of canonical ERα target genes but markedly inhibits jointly regulated SP1/ERα target genes. Some of these jointly regulated SP1/ERα target genes, like LOXL4 and CADM1, have overlapping ERα and SP1 ChIP-seq peaks, suggesting that ERα may be acting as an estrogen-regulated coactivator for SP1. However, DEPTOR appears to contain distinct ERα and SP1 binding sites, raising the possibility that ERα and SP1 may interact on these promoters through looping of chromatin. Notably, some studies suggest that the SP1/ERα complex plays a particularly critical role in the regulation of proliferation by estrogen (Deroo and Korach 2006; Fuentes and Silveyra 2019; Chen et al. 2022).
Our study shows that silencing mutant p53 increases ERα stability and modestly enhances ERα activity on canonical targets. In contrast, we found that silencing mutant p53 dramatically enhances the E2-mediated induction of joint SP1/ERα target genes. Somewhat surprisingly, this also restores the sensitivity of HGSOC lines to the selective estrogen degraders (SERDs) fulvestrant and elacestrant. We note that whereas restoring the SP1/ERα arm of the ERα pathway may be critical for restoring sensitivity to endocrine therapy, our data indicate that silencing mutant p53 also increases ERα nuclear localization, and this may explain part of the increased sensitivity to fulvestrant. Along these lines, it is also known that SERDs not only target ERα for degradation but also inhibit ERα mobility on chromatin, and this may contribute to their anticancer activity (Hanker et al. 2020). Overall, our data are consistent with the findings of others that restoring ERα expression in ER-negative breast cancer renders these cells sensitive to tamoxifen (Sharma et al. 2006) and that mutations in ERα and p53 are mutually exclusive toward conferring endocrine resistance in breast cancer (Li et al. 2022). The combined data suggest that ER+ tumors with mutant p53 may respond less well to endocrine therapy compared with tumors that are WT or null for p53; however, we note that there are many other proteins that regulate ERα function and potentially many other mechanisms for endocrine resistance. Therefore, such a conclusion awaits extensive clinical validation.
There are some caveats to this study. We have not determined whether the growth arrest induced by fulvestrant when mutant p53 is silenced is transient or more prolonged (e.g., senescence). Also, we focused predominantly on the impact of mutant p53 on ERα but not on ERβ; the latter can interact with mutant p53 (Mukhopadhyay et al. 2019; Oturkar et al. 2022; Scarpetti et al. 2023). We found that PEO1 cells do not express ERβ, consistent with published reports (Banerjee et al. 2022). We note that the relationship between p53 and ERα is more complex than outlined here and involves joint transcriptional regulation: ERα is a transcriptional target of p53 (Shirley et al. 2009), and p53 is a transcriptional target of ERα (Berger et al. 2012). We were able to analyze only a subset of mutant forms of p53, and whether the abundance of the ERα mutant p53 complex is different in different tumors and different tumor types is unclear at present. While we showed that the ERα–p53 complex is a direct interaction, we did not map the domain of interaction. The difference in survival in tumors with high DEPTOR expression could be influenced by its proximity to the MYC oncogene, which can be amplified in some ovarian tumors. A final caveat is that, in order for this pathway to be leveraged in the clinic, we must find ways to silence or refold mutant p53. Unfortunately, most mutant p53 refolding compounds are not approved in the clinic, and some that are approved do not require mutant p53 for efficacy, so they clearly have a different mechanism of action (Wang et al. 2023; Tuval et al. 2024). One interesting exception is the recent discovery of rezatapopt, and our data suggest that combining rezatapopt with endocrine therapy could have promising therapeutic potential.

Materials and methods

Cell culture, reagents, and plasmids
All cell lines used in this study were purchased from ATCC and used within 6 months of purchase or thawing. The exceptions are PEO1 and OV2085 cells, which were generously provided by Andrew Godwin (University of Kansas Medical Center) and A.-M.M.-M. (University of Montreal), respectively. PEO1, T47D, and MDA-MB-134 cells were cultured in RPMI 1640 (Corning) supplemented with 10% FBS (HyClone) and 1% penicillin/streptomycin (Gibco). OV2085 cells were cultured in medium OSE (Wisent) supplemented with 10% FBS and 1% penicillin/streptomycin. SKOV3, MCF7, and HEK293 cells were cultured in DMEM (Corning) supplemented with 10% FBS and 1% penicillin/streptomycin. For hormone deletion, cells were cultured in phenol red-free RPMI 1640, medium OSE, or DMEM supplemented with 10% charcoal-stripped FBS (Gibco) at least 48 h before treatment. For fulvestrant or elacestrant treatment, 100 pM 17β-estradiol (E2) was added, and cells were treated in phenol red-free media supplemented with 10% charcoal-stripped FBS. All cells were grown at 37°C in a 5% CO2 humidified incubator. The following reagents were used: 17β-estradiol (E2; Sigma), MG132 (Sigma), cycloheximide (CHX; Sigma), puromycin (Gibco), etoposide (MedChemExpress), rezatapopt (MedChemExpress), fulvestrant (Cayman), and elacestrant (MedChemExpress). pCMV-ERα, pCMV-Bam p53 R175H, and pCMV-Bam p53 R273H plasmids were purchased from Addgene. RNA interference, immunoblotting, immunoprecipitation, qPCR, IHC, IF, cell viability, and RNA sequencing were all performed as described previously (Indeglia et al. 2023). Detailed methods using these procedures are included in the Supplemental Material.

Spheroid formation and organoid analysis
For the organoid analysis, the PEO1-pLKO.1 and PEO1-shp53 cells were grown in pretreatment media (phenol red-free RPMI 1640 supplemented with 10% charcoal-stripped FBS) for 24 h prior to plating cells in dome matrices. A 20 µL cell suspension mixed with 40 µL of Cultrex UltiMatrix (R&D Systems BME001-05) was made to a final cell concentration of 1000 cells/5 µL. The cell and matrix suspensions were carefully mixed to avoid bubbles, and 5 µL was plated into each well of a 96 well plate. The plate was flipped upside down for 30 min to allow the dome to polymerize before adding 100 µL of media to each well. Elacestrant (3 µM) was added on the same day the cells were plated in the matrix domes, and fresh drug was added every 4 days up to day 10. On day 10, the organoid-containing domes were incubated in Calcein-AM for live-cell imaging and quantification. The quantification was done by reading the green fluorescence on a Tecan Infinite plate reader at the appropriate wavelength for at least 15 different organoid-containing domes for both untreated and treatment conditions. Images of the entire domes were taken by the Tecan Spark live-cell imager using a 2× objective.

Chromatin immunoprecipitation, qPCR, and sequencing
ChIP was performed as described previously (Indeglia et al. 2023) using the primers listed in Supplemental Table S1. The following antibodies were used: ERα (Santa Cruz Biotechnology sc-8002X), SP1 (Abcam ab231778), and H3K27ac (Abcam ab4729). ChIP-seq libraries were made using the NEBNext Ultra II DNA library preparation kit according to the manufacturer's instructions. Sequencing was performed on a NextSeq 2000. Reads were quality-checked and aligned to the hg19 genome. Peak calling was performed using MACS2 against their respective input samples. deepTools was used to generate ChIP-seq signal tracks. Signal tracks were visualized by Integrative Genomics Viewer (IGV). Motif enrichment analysis was performed by HOMER and MEME suite with default settings.

Correlation of ER-induced genes with breast cancer survival
TCGA breast cancer gene expression data (RSEM RNA-seq Z-scored, with diploid samples used as reference) (The Cancer Genome Atlas Network 2012) and clinical data, as well as METABRIC microarray (Z-scored with diploid samples used as reference) (Curtis et al. 2012) and clinical data, were obtained from cBioPortal (Cerami et al. 2012). The list of ER-induced genes was cross-referenced with both the TCGA-BRCA and METABRIC data sets to identify associations with patient outcomes. ER-positive patients were extracted from both data sets using positive ER IHC status, and these samples were used for all further analyses. First, the average Z-score across all ER-induced genes was correlated with patient prognosis, showing a predictive capacity in the METABRIC cohort (predicting both OS and RFS), thus supporting the premise that the ER-induced genes are correlated with improved survival of ER-positive breast cancer patients. Second, to derive a predictive signature across cohorts, each ER-induced gene was correlated with overall survival (OS) and either recurrence-free survival (RFS; METABRIC data set) or disease-free survival (DFS; TCGA-BRCA) through log rank test, separating groups by median expression (using the lifelines package in Python). The log rank P-values and hazard ratios were obtained for each gene, and then P-values were adjusted for false discovery rate (FDR-corrected P < 0.1). We found that most genes correlated with improved survival of ER+ breast cancer patients (75% of all genes with significant associations and 100% of those that validated in both data cohorts). Genes that significantly correlated with survival in the TCGA-BRCA cohort and validated in the METABRIC cohort were extracted, yielding four genes as predictors of OS and none as predictors for DFS/RFS. The average Z-score across the four genes validated across both cohorts (namely, KRT5, DIRAS3, KRT15, and KRT16) was used to predict prognoses, demonstrating a strong predictive value of improved outcomes in both the TCGA and METABRIC cohorts. Kaplan–Meier survival curves for each gene or signature were plotted using the mean expression of each gene or signature. TP53 mutation status was not a confounder of the analysis, as demonstrated by using stratified Cox regression on the METABRIC data set, where the results remained significant when stratifying by TP53 mutations. Pathway enrichment analysis was conducted on the set of all ER-induced genes that were significantly predictive of ER+ breast cancer survival in either the training or validation cohort using GSEA gene sets and a hypergeometric enrichment P-value (using FDR correction, with corrected P < 0.1). For correlations in HGSOC, box plots comparing the average Z-score across the four genes (KRT5, DIRAS3, KRT15, and KRT16) between invasive ovarian cancer samples with high versus low p53 marker staining and between samples with high versus low TP53 gene expression were determined using the data set presented by Kader et al. (2025).

Statistical analysis of data
The ovarian cancer survival curves for DEPTOR were analyzed using the Kaplan–Meier plotter online database (https://www.kmplot.com/analysis/index.php?p=service&cancer=ovar). For all survival analyses, patients with serous histology, advanced stage (3+4), and high grade (3) were selected. A two-tailed unpaired Student's t-test was used to determine P-value (P < 0.05 [*], P < 0.01 [**], and P < 0.001 [***]), and the standard deviation of at least three biological replicates is shown in the bar graphs. The number of replicates is denoted in the figure legends.

Illustrations
All working models were created using BioRender, for which the authors have a license.

Data availability
RNA-seq and ChIP-seq data were deposited to GEO under the following accession numbers: GSE292625, GSE292583, and GSE292582.

Supplemental Material
Supplement 1