TRIM24 as a therapeutic target in endocrine treatment–resistant breast cancer

Significance

Estrogen receptor alpha (ERα) is a transcription factor that drives breast cancer development, serving as the primary therapeutic target in this disease. Endocrine therapies are designed to inhibit ERα action, but resistance to treatment is common, resulting in incurable metastatic disease. In this study, we found TRIM24 to be a key ERα cofactor that facilitates ERα chromatin interactions and maintains active histone marks, supporting ERα-driven transcription and tumor growth. Using a recently developed TRIM24 degrader, we effectively block ERα transcriptional output and tumor growth in both endocrine-responsive and -resistant models, as well as patient-derived organoids. Thus, therapeutic interventions that deplete TRIM24 may serve as an interesting emerging strategy to treat metastatic therapy–resistant breast cancer patients.

Abstract

While Estrogen receptor alpha (ERα)+ breast cancer treatment is considered effective, resistance to endocrine therapy is common. Since ERα is still the main driver in most therapy-resistant tumors, alternative therapeutic strategies are needed to disrupt ERα transcriptional activity. In this work, we position TRIM24 as a therapeutic target in endocrine resistance, given its role as a key component of the ERα transcriptional complex. TRIM24 interacts with ERα and other well-known ERα cofactors to facilitate ERα chromatin interactions and allows for maintenance of active histone marks including H3K23ac and H3K27ac. Consequently, genetic perturbation of TRIM24 abrogates ERα-driven transcriptional programs and reduces tumor cell proliferation capacity. Using a recently developed degrader targeting TRIM24, ERα-driven transcriptional output and growth were blocked, effectively treating not only endocrine-responsive cell lines but also drug-resistant derivatives thereof as well as cell line models bearing activating ESR1 point mutations. Finally, using human tumor-derived organoid models, we could show the efficacy of TRIM24 degrader in the endocrine-responsive and -resistant setting. Overall, our study positions TRIM24 as a central component for the integrity and activity of the ERα transcriptional complex, with degradation-mediated perturbation of TRIM24 as a promising therapeutic avenue in the treatment of primary and endocrine resistance breast cancer.

Estrogen receptor alpha (ERα) is a hormone-driven transcription factor, expressed in over 75% of breast cancer (BC) patients (1). In these tumors, ERα is considered the key driver of tumor development and progression and represents the main target for treatment in the adjuvant and metastatic setting (2). ERα is activated through its natural ligand estradiol (E2), inducing receptor dimerization and subsequent association with distal regulatory DNA elements—enhancers—where it recruits a large multi-megadalton protein complex, to orchestrate the expression of specific ERα-responsive genes through long-range chromatin interactions and drive tumor cell proliferation (3).

Endocrine therapies aim to prevent ligand-mediated activation of the receptor [aromatase inhibitors (AI)], block coregulator recruitment (tamoxifen), or induce receptor degradation (fulvestrant), all of which inhibit tumor cell proliferation. However, around 30% of BC patients relapse despite having received adjuvant treatment with endocrine therapeutics, resulting in metastatic disease for which there is no cure (4). Importantly, for most endocrine-resistant metastatic BCs, ERα remains the critical driver. Point mutations in the ESR1 gene (encoding for ERα) are found in around 20 to 40% of ERα+ metastatic cases, mostly after long exposure to AIs (5, 6). Mechanistically, the most impactful somatic mutations fall in the ESR1 ligand-binding domain (Y537 and D538), allowing for hormone-independent transcriptional activity of ERα (7, 8).

As a substantial fraction of BC patients present with acquired resistance to endocrine drugs, further research efforts are being made in the identification of novel therapeutic strategies for these patients. Since most coregulators represented in the ERα-transcription complex are conserved between endocrine-sensitive (9) and -resistant BC cells (10), these proteins may represent an interesting novel therapeutic avenue to explore, as was reported for FEN1 (11), SRC-3 (12), or CREBBP/EP300 (13).

TRIM24 is an E3-ubiquitin ligase that, together with TRIM28, TRIM33, and TRIM66, constitutes the transcriptional intermediary factor 1 (TIF1) family of chromatin-binding proteins (14). Increased expression levels of this subfamily of the large tripartite motif (TRIM) family of E3 ligases have been linked to BC development (15–17), serving as mediators of chromatin-associated DNA damage responses (18), epigenetic modulation (15), and metastatic potential (17, 19). TRIM24 overexpression in mice has been shown to induce the formation of mammary tumors that lack ERα, progesterone receptor, and HER2 expression, closely resembling human metaplastic breast cancers (MpBC), a rare and aggressive triple-negative BC subtype (20). However, the role of TRIM24 in hormone receptor-positive BC, particularly in the context of acquired therapy resistance, is incompletely understood.

TRIM24 has previously been reported to physically interact with ERα in BC (21, 22), serving as ERα coregulator (15). These findings position TRIM24 as a putative therapeutic target in endocrine resistance, but this has to date been unexplored. Moreover, a heterobifunctional protein degrader targeting TRIM24 was recently developed (23), showing promising results in fully degrading TRIM24 in acute leukemia cells and capable of diminishing TRIM24 protein levels in MCF7 cells.

In this study, we performed comprehensive immunoprecipitation–mass spectrometry analyses that support previous findings and confirm TRIM24 as a genuine interactor of ERα. Genetic perturbation of TRIM24 in MCF7 cells effectively impaired ERα-dependent gene transcription, blocked ERα-driven gene transcriptional networks, and diminished tumor cell proliferation. Importantly, treatment of endocrine therapy–resistant cell line models and patient-derived organoids with a TRIM24 degrader fully recapitulated the blocking in proliferation as observed in endocrine-sensitive cells.

Collectively, our work positions TRIM24 as a critical regulator of ERα activity in therapy-resistant BC cells. Furthermore, degradation-mediated disruption of TRIM24 may provide a viable therapeutic strategy for endocrine-resistant BC treatment.

Results

TRIM24 Interacts with ERα in BC Cells.

To comprehensively analyze the ERα transcription complex in BC, we performed Rapid-immunoprecipitation mass spectrometry of endogenous proteins (RIME) for ERα in MCF7 cells (Fig. 1A). TRIM24, a previously reported interactor of ERα in BC (15, 21) and interesting potential therapeutic target due to the recently developed degrader (23), was one of the top interactors in our analyses (Fig. 1A). Importantly, ERα RIME upon 6 h of β-estradiol (E2) induction, after hormone-deprivation, enhanced TRIM24 interactions in MCF7 cells compared to DMSO-treated control cells (SI Appendix, Fig. S1 A and B). These results suggest a hormone-dependent recruitment of TRIM24 to ERα transcriptional complex. Coimmunoprecipitation of TRIM24 in MCF7 and T47D cells further confirmed its interaction with ERα (Fig. 1B and SI Appendix, Fig. S1C). To further solidify our findings, we performed the reciprocal experiments and performed a TRIM24 RIME, confirming interactions with ERα along with numerous of its classical coregulators, including GREB1 (9), NCOA3 (24), GATA3 (25), and TLE3 (21) in MCF7 cells (Fig. 1C, SI Appendix, Fig. S1 D and E, and Datasets S1 and S2) which were independently confirmed in TRIM24 RIME experiments in T47D cells (SI Appendix, Fig. S1F). Previously, TRIM24 was reported to associate with poor outcome in ERα-positive BC (15). In line with these observations, TRIM24 expression levels significantly correlated with higher BC tumor grade [Fig. 1D, (26)] and with Ki-67 expression (Fig. 1E); a well-known marker of tumor proliferation and poor prognosis (27). Moreover, TRIM24 transcript levels were progressively increased from healthy mammary tissue to primary tumors and eventually metastases, suggesting a potential role of TRIM24 in disease progression (Fig. 1F). In summary, TRIM24 serves as an interactor for ERα and is associated with unfavorable clinical features in BC patients.

Fig. 1.

TRIM24 is a member of the ERα transcriptional complex, and its expression is correlated with poor BC prognosis. (A) Volcano plot summarizing the quantitative results of ERα RIME in MCF7. TRIM24 is depicted in red. ERα (ESR1 gene) and several known ERα interactors are labeled. (B) Coimmunoprecipitation of TRIM24, followed by western blot for TRIM24 and ERα in MCF7 cells. (C) Volcano plot summarizing the quantitative results of TRIM24 RIME. ERα (ESR1 gene), several known ERα interactors, and members of the TIF1 family (TRIM28 and TRIM33) are labeled. (D) Box plot showing TRIM24 expression in different ERα+ BC grade tumors. Gene expression data originated from the METABRIC dataset (26), filtered for ERα+/HER2- BC patients (n = 997) that have received hormone therapy. Gene expression category per patient was defined by the median gene expression for the group (above = HIGH, below = LOW). The t test was performed for association with Grade (1,2,3) and gene expression categories. (E) Correlation plot of normalized TPM expression (RNA-seq) between TRIM24 and Ki-67 in 879 ERα+ female BC patient samples from TCGA-BRCA cohort (www.cancer.gov/tcga). Pearson’s correlation coefficient is displayed. (F) Violin plot showing the RNA-seq TRIM24 gene expression from Breast Invasive Carcinoma TNMplot data including tumor, metastatic, and normal samples as control. Average of 113 healthy, 1,097 tumors, and 7 metastatic samples is shown. Global P-value of the Kruskal–Wallis test, as well as post hoc Wilcoxon’s rank-sum test, is indicated.

TRIM24 Is Recruited to the Chromatin at ERα-Binding Sites after Stimulation.

TRIM24 binds the chromatin (28, 29) and regulates activity of multiple transcription factors including RXRα, RARα, and AR (30, 31). To determine genome-wide action of TRIM24, we performed chromatin immunoprecipitation followed by sequencing (ChIP-seq) in MCF7 cells (Fig. 2 A–E, SI Appendix, Fig. S2 A and B, and Dataset S3). As exemplified for classical ERα-target loci (GREB1, TFF1, and IGFBP4; Fig. 2A), and comprehensively shown on a genome-wide scale (Fig. 2 B and C and SI Appendix, Fig. S2B), TRIM24 shows strong overlap with ERα-occupied chromatin regions. Even though the raw ChIP-seq signal of TRIM24 widely overlaps with ERα (Fig. 2B and SI Appendix, Fig. S2B), the number of called peaks for TRIM24 is relatively low (Dataset S3). This discrepancy may have either technical (e.g., antibody quality, epitope availability) or biological (e.g., transient interactions) causes that are yet to be addressed in future studies.

Fig. 2.

TRIM24 binds together with ERα at the same genomic locations, and its binding is hormone dependent. (A) Snapshots of ERα and TRIM24 ChIP-seq signal at GREB1, TFF1, and IGFBP4 loci. The genomic coordinates are annotated. (B) Heatmap disclosing ChIP-seq signal for Input, ERα, and TRIM24 in full medium. Regions were sorted according to decreasing ERα signal. Data are centered at each factor peak, depicting a ±1 kb window around the peak center. (C) Correlation plot between ERα and TRIM24 ChIP-seq signal at ERα peaks. Pearson’s correlation coefficient (R) is indicated. Red line depicts the linear regression (glm, y ~ x) ± SE. Color scale indicates the dot density. (D) Motif enrichment analysis at TRIM24 binding sites. Font size represents log₁₀(P-value) × 10³. (E) GIGGLE enrichment analysis for the top transcription factors binding at the enriched TRIM24 sites. (F) Snapshots of TRIM24 ChIP-seq signal at GREB1, TFF1, and IGFBP4 loci upon treatment with DMSO or E2 for 3 h. The genomic coordinates are annotated.

Motif enrichment analyses on TRIM24 bound sites reveal enrichment for motifs of ESR1, as well as its classical pioneer factor FOXA1 (32) (Fig. 2D). To identify, in an unbiased fashion, which transcription factors co-occupy TRIM24-bound sites, we overlaid the genomic coordinates of TRIM24 sites with publicly available ChIP-seq datasets (n = 13,976) from the Cistrome Database (33, 34). In addition to ESR1, enrichment of classical ERα-interactors is found, including EP300 (35), GREB1 (9), RARα (36), and GATA3 (25) (Fig. 2E). TRIM24 itself was not found enriched in this analysis, which can likely be attributed to the absence of sufficient quality TRIM24 ChIP-seq data in this repository. ChIP-seq analyses in MCF7 cells illustrated that TRIM24 chromatin binding is increased upon E2 treatment, as visualized genome-wide (SI Appendix, Fig. S2 D–F and Dataset S3), and exemplified for well-annotated ERα-driven enhancers for GREB1, TFF1, and IGFBP4 loci (Fig. 2F). Cumulatively, these findings illustrate, in a genome-wide manner, a hormone-enhanced interaction of TRIM24 with ERα sites at the chromatin level.

TRIM24 Is Required for Transcriptional Activity of ERα.

TRIM24 occupies ERα-bound regulatory elements, but is TRIM24 required for ERα transcriptional activity? To address this question, we generated TRIM24 CRISPR-Cas9 Knockout (KO) models (Fig. 3A) in MCF7 cells, which resulted in reduced tumor cell proliferation (Fig. 3B). Importantly, depletion of TRIM24 did not affect ERα protein levels (Fig. 3A). Using monoclonal TRIM24-KO A cell populations (SI Appendix, Fig. S2G), we performed ChIP-seq for ERα (Fig. 3 C and D, SI Appendix, Fig. S2 H–K, and Dataset S4) and observed reduced chromatin binding of this transcription factor in the absence of TRIM24 (Fig. 3D and SI Appendix, Fig. S2K), as exemplified for classical ERα-bound loci such as GREB1, TFF1, and IGFBP4 (Fig. 3C). This observation was confirmed using the second monoclonal population of TRIM24-KO (TRIM24-KO B; SI Appendix, Figs. S2G and S3A). Moreover, TRIM24 is known to interact with acetyl lysine residues of histones, in particular H3K23ac, an understudied epigenetic mark associated with cancer development (15, 31, 37), but also H3K27ac, a marker of active enhancer and promoter regions (15, 38, 39). Previous work showed reduced H3K23ac and H3K27ac levels at a specific ERα-responsive locus upon TRIM24 depletion (15). Therefore, we performed ChIP-seq of both H3K23ac and H3K27ac to chart the profiles of these two epigenetic marks on a genome-wide scale (Fig. 3 C and D and SI Appendix, Figs. S2H and S3 B and C) and further confirmed by ChIP-qPCR for the second monoclonal TRIM24-KO model (TRIM24-KO B; SI Appendix, Figs. S2G and S3A). Although the overall H3K23ac and H3K27ac occupancy remains unchanged in the absence of TRIM24 (SI Appendix, Fig. S3 B and C), we observed a decreased signal for both histone marks specifically at ERα-bound sites following TRIM24 perturbation (Fig. 3 C and D). Altogether, these results indicate a potential impairment of ERα transcriptional activity in the absence of TRIM24.

Fig. 3.

Depletion of TRIM24 leads to reduced ERα activity. (A) Western blot for TRIM24-KO in MCF7 cells. NT: nontarget. Actin is used as a loading control. (B) Cell viability assay on TRIM24-KO MCF7 cells and NT MCF7 cells, by CTG (n = 7). P-value represents the two-tailed paired t test. (C) Snapshots of ERα, H3K23ac, and H3K27ac ChIP-seq peaks at GREB1, TFF1, and IGFBP4 loci. The genomic coordinates are annotated. (D) Heatmap disclosing ChIP-seq signal for ERα, H3K23ac, and H3K27ac in NT (clone 1) or TRIM24-KO (Clone A) MCF7 cells at ERα binding regions in NT MCF7 cells. Regions were sorted according to decreasing ERα average signal. Data are centered at each factor peak, depicting a ±2.5 kb window around the peak center. (E) Mean signal intensity for p-Ser RNA Polymerase 2 at genes found near an ERα binding site in NT1. Below is a zoom shot of the starting site ±2.5 kb distance. (F) Enrichment plot of the hallmark estrogen response late pathway performed on differential transcriptomic data between TRIM24-KO A and NT1. Genes are ranked by differential expression upon TRIM24-KO. (G) Heatmap depicting the differential expression of genes included in the Hallmark Estrogen Response Late gene set between both TRIM24-KO MCF7 clones and both NT. Color scale represents the gene expression (z-score).

To confirm a direct functional impact of TRIM24 on ERα action, we performed phospho-Ser2 (p-Ser2) RNA Polymerase II CTD ChIP-seq, showing reduced occupancy at putative ERα target genes upon TRIM24-KO (Fig. 3E and SI Appendix, Fig. S3D). Importantly, ERα sites that are not shared with TRIM24 were not affected on p-Ser2 RNA Polymerase II ChIP-seq signal, upon TRIM24-KO (SI Appendix, Fig. S3D). In line with this, transcriptomic data in both TRIM24-KO monoclonal populations (Fig. 3 F and G and SI Appendix, Fig. S3 E and F) show robust gene expression alterations, with a significant decrease of the late estrogen-response hallmark signature (Fig. 3 F and G). In agreement with this, we find a positive correlation between TRIM24 expression and ERα responsive genes (Estrogen Receptor Early and Late responses) on GSEA from the TCGA-BRCA dataset (www.cancer.gov/tcga) (SI Appendix, Fig. S3 G and H), clinically confirming our cell line–based observations.

TRIM24 Degrader Phenocopies Knockout Effects, Inhibits ERα Activity, and Impairs BC Cell Growth.

Heterobifunctional protein degraders provide new possibilities to selectively target a protein-of-interest for degradation, using the ubiquitin proteasome system (40). Recently, a molecule inducing selective degradation of TRIM24 in leukemia cells was developed (23). Due to the crucial role of TRIM24 on ERα function and therefore on ERα-positive BCs (Figs. 2 and 3), we decided to explore the biological impact of this TRIM24 degrader (dTRIM24) in ERα-positive BC cells.

dTRIM24 activity is dependent on the binding of TRIM24 to VHL, the E3 ubiquitin ligase responsible for ubiquitinating TRIM24 to be degraded by the proteasome (23). Thus, to identify any putative off-target activity of the degrader, we treated MCF7 WT with a previously established TRIM24 inhibitor—IACS-9571 (41) that does not possess any VHL targeting moiety. Both ERα DNA-binding and consequent transcriptional activity were unaltered by treatment with IACS-9571 (SI Appendix, Fig. S4 A–C), contrarily to our observations in the TRIM24-KO cells. These data illustrate that TRIM24’s role in the ERα transcriptional complex is independent of its E3 ligase activity.

Moreover, when we deplete VHL in MCF7, the degradation of TRIM24 by dTRIM24 treatment is impaired (SI Appendix, Fig. S4D), and we observe a phenotypic rescue of VHL-KD MCF7 cell proliferation from dTRIM24 treatment (SI Appendix, Fig. S4E). Additionally, MCF7 TRIM24-KO cells exposed to dTRIM24 show a reduced sensitivity on the level of cell proliferation impairment, in contrast of TRIM24-WT cells (SI Appendix, Fig. S4 F and G). Cumulatively, these results confirm target-specificity of dTRIM24 and exclude off-targets effects of the degrader driving the phenotype in MCF7 cells.

We observe successful degradation of TRIM24 upon dTRIM24 treatment [as reported previously by others (23)], with no impact on ERα protein levels (Fig. 4A and SI Appendix, Fig. S4H). Interestingly, and in contrast to previous results (23), we observed a significant reduction on cell proliferation of ERα-positive BC cells upon dTRIM24 exposure (Fig. 4 B and C and SI Appendix, Fig. S4 I and J). These observations are in full concordance with our TRIM24 knockout data, impairing MCF7 cell proliferation (Fig. 3B). Moreover, treatment of MCF7 cells with a second recently developed TRIM24 degrader (dTRIM24_2; Fig. 4D and SI Appendix, Fig. S4K) confirmed our original observations, with a significant growth impairment. Specifically, dTRIM24 treatment prevented E2-driven cell proliferation of MCF7 cells (Fig. 4E), stressing the impact of TRIM24 depletion on ERα activity and BC tumor growth.

Fig. 4.

TRIM24 degrader impairs BC cell proliferation through inhibition of ERα transcriptional activity. (A) Representative western blot for TRIM24 and ERα protein levels upon dTRIM24 treatment for 24 h in MCF7 (TRIM24+ and ERα+ BC cell line) and MDA-MB-231 (TRIM24− and ERα− BC cell line). Actin is used as a loading control. (B) Relative growth of MCF7 or MDA-MB-231 cells treated with 5 μM of dTRIM24 compared to DMSO control (n = 7). P-value is determined by two-way ANOVA with Bonferroni’s post hoc test ****P < 0.0001; n.s.; nonsignificant. (C) Cell viability analysis of MCF7 and MDA-MB-231 cells treated with 5 μM of dTRIM24. CTG was performed after 8 d of treatment with dTRIM24 (n = 4). P-value is determined by the two-tailed paired t test. **P < 0.01, n.s.; nonsignificant. (D) Cell viability assay of MCF7 treated with 5 μM of dTRIM24_2. CTG was performed after 8 d of treatment (n = 6). P-value is determined by the two-tailed paired t test: ***P < 0.001. (E) Cell viability assay by CTG for MCF7 cells deprived for 3 d of hormones and incubated for 8 d with the depicted treatments (n = 3). P-value is determined by two-way ANOVA with Sidak’s correction: **P < 0.01, ***P < 0.001. (F) GSEA for hallmark gene sets. Shown are the top differentially enriched pathways upon 24 h treatment with dTRIM24 (5 µM) or vehicle (DMSO). The X-axis indicates the normalized enrichment score. Only significantly enriched pathways are indicated (P < 0.05). (G) Volcano plot summarizing the quantitative results of ERα RIME in MCF7 treated for 24 h with 5 μM dTRIM24 or vehicle (DMSO).

Importantly, dTRIM24 had no impact on cell fitness of ERα-negative cell line MDA-MB-231 (Fig. 4 A–C and SI Appendix, Fig. S4I), which did not express any detectable TRIM24 on protein level (Fig. 4A). To comprehensively assess the impact of dTRIM24 on the transcriptional program in ERα-positive BC, we performed RNA Sequencing (RNA-seq) analysis after 24 h treatment with 5 μM dTRIM24 (Fig. 4F). GSEA revealed decrease of cell cycle progression and MYC target gene signatures -indicative of diminished cell proliferation capacity- and reduction of both early and late estrogen response gene sets upon dTRIM24 treatment (Fig. 4F). To identify where TRIM24 is positioned in the ERα complex, and whether TRIM24 is required for association of other ERα interactors to the complex, we performed ERα RIME experiments in MCF7 cells treated with 5 μM of dTRIM24 (Fig. 4G). In addition to the expected loss of TRIM24, we observed a reduced interaction of the Bromodomain Adjacent To Zinc Finger Domain 1B (BAZ1B) protein with ERα upon TRIM24-KO. BAZ1B is a critical component of the Imitation switch (ISWI) family of chromatin remodelers, involved in nuclear organization (42). These data suggest that TRIM24 may be required to retain proteins involved in chromatin structure maintenance, allowing for successful chromatin binding of ERα. Cumulatively, these data show that targeting TRIM24 through a degrader phenocopies a genetic knockout for TRIM24, perturbing ERα-driven transcriptional output and cell proliferation.

TRIM24 Degrader Impairs Cell Growth of Treatment-Resistant BC Cells and Patient-Derived Organoids.

To further explore the potential use of dTRIM24 as an alternative treatment in endocrine therapy–resistant BC, we made use of MCF7 cells that were rendered resistant to tamoxifen treatment through long-term exposure [MCF7 TAMR (43)], or long-term deprived from estrogen to model resistance to aromatase inhibitors [MCF7 LTED (44)] and MCF7 variants that have been CRISPR engineered to contain point-mutations in the ligand-binding pocket [MCF7 Y537S Clone 1 and 2; MCF7 D538G Clone 3 (10)] that have been found to render ERα constitutively active in metastatic relapse upon prolonged endocrine exposure (7). We then incubated all the abovementioned cell lines with 5 μM of dTRIM24 (Fig. 5A). In agreement with the persistent dependency of resistant cells on ERα signaling, both TAMR and LTED MCF7 were significantly perturbed in their cell proliferation capacity following degrader-mediated depletion of TRIM24 (Fig. 5 B and C and SI Appendix, Fig. S5A). MCF7 TAMR growth capacity could also be significantly decreased by a second TRIM24 degrader (dTRIM24_2) (Fig. 5D and SI Appendix, Fig. S5 B and C), independently confirming our observations and reinforcing the potential applicability of targeting TRIM24 in therapy-resistant BC. Somatic mutations in the ESR1 locus render the receptor’s activity ligand-independent, reducing response to endocrine therapy (7). Proliferation capacity of cell lines expressing constitutively active ESR1 (Y537S [clone C1 and C2] and D538G) was significantly reduced upon dTRIM24 exposure (Fig. 5E). Thus, targeted degradation of TRIM24 successfully impairs growth of several endocrine therapy–resistant BC cells, including those with clinically relevant ESR1 activating point mutations.

Fig. 5.

TRIM24 degrader blocks therapy-resistant BC cell growth and impedes the formation of ERα+ patient-derived BC organoids. (A) TRIM24 protein levels upon treatment with 5 μM of dTRIM24 in several endocrine-resistant cell lines. Actin is used as a loading control. (B) Cell proliferation percentage relative to DMSO of MCF7 TAMR and LTED treated with 5 μM of dTRIM24 for 12 d (n = 3). Results shown are normalized to the vehicle control. P-values are determined by two-way ANOVA with Bonferroni post hoc test: ****P < 0.0001. (C) Cell viability assay by CTG in MCF7 TAMR (n = 5) and LTED (n = 6) treated with 5 μM of dTRIM24 for 12 d. P-value is determined by the two-tailed paired t test: *P < 0.05, ***P < 0.001. (D) Colony Assay quantification of MCF7 TAMR treated with increased concentrations of dTRIM24_2 (n = 3). P-values represent two-way ANOVA with Tukey’s multiple test correction. ***P < 0.001, ****P < 0.0001. (E) Cell proliferation analysis of MCF7 ESR1-mutant Y537S C1, C2, and MCF7 ESR1-mutant D538G C3, treated with 5 μM of dTRIM24 for 8 d (n = 5). Results shown are normalized to the vehicle control. P-values are determined by two-way ANOVA with Bonferroni post hoc test: ***P < 0.001; ****P < 0.0001. (F) TRIM24 and ERα protein levels in the depicted organoid models after treatment with 10 μM of dTRIM24. Actin is used as a loading control. (G) 3D patient-derived organoid culture cell viability assay. Each patient-derived organoid was incubated either with dTRIM24 (5 or 10 μM) and vehicle control for 7 d before cell-viability was measured (n = 6). TNBC: triple negative breast cancer. Two-way ANOVA test: *P < 0.05; **P < 0.01; ***P < 0.001. (H) GREB1 protein level after treatment with 10 μM of dTRIM24 in the two depicted organoid models. Actin is used as a loading control. (I) mRNA level of GREB1, XBP1, and RARα upon treatment with 10 μM of dTRIM24 in T347 (Left) and T4-1 (Right) organoid models. Results are representative of n = 3 biological replicates. Two-tailed t test: *P < 0.05; **P < 0.01; ***P < 0.001.

Next, we aimed to determine the efficacy of TRIM24 depletion in patient-derived models. For this, we exposed treatment naïve and endocrine therapy–resistant patient-derived organoids to 5 or 10 μM of dTRIM24 and evaluated their cell viability (Fig. 5 F and G). Strikingly, two out of three treatment naïve and one out of three therapy-resistant patient-derived BC organoids showed significantly reduced cell viability in preformed three-dimensional (3D) cultures after TRIM24 depletion (Fig. 5G). Next, the impact of dTRIM24 treatment was tested in two organoid models (T4-1 and T347), revealing protein and transcriptomic levels of canonical ERα-responsive genes (Fig. 5 H and I) being significantly reduced upon dTRIM24 treatment. These results confirm in patient-derived model systems the therapeutic potential of TRIM24 treatment to eradicate therapy-resistant BC.

In agreement with our observations in the ERα-negative cell line MDA-MB-231 (Fig. 4 A–C and SI Appendix, Fig. S4I), TRIM24 depletion did not impact cell viability capacity of preformed triple-negative BC (TNBC) 3D cultures (Fig. 5G). Altogether, we provide preclinical proof of concept for targeted treatment of treatment naïve and endocrine therapy–resistant BC by degradation-mediated depletion of the ERα coregulator TRIM24.

Discussion

Endocrine therapy resistance represents a major clinical unmet need. The treatment-induced representation of ESR1 mutations, ESR1 gene fusions (5), and elevated expression of ERα coactivators, such as NCOA1 (45), are observed in BC patients after long-term exposure to endocrine therapies. Novel therapeutic strategies that indirectly target ERα activity may still be effective in these settings, in which conventional endocrine therapies have lost their efficacy.

In our study, we confirm TRIM24 as a key player on the ERα transcriptional complex and identify TRIM24 as an attractive therapeutic target in the treatment of endocrine-resistant BC. The functional impact of TRIM24 on ERα action appeared to act globally, as TRIM24 binds almost all ERα DNA-binding sites, and its depletion reduces ERα transcriptional activity.

ERα binding to chromatin is a highly orchestrated process involving many players in a temporal and functional order (3). Importantly, ERα requires a permissive chromatin structure to correctly recognize and bind its specific binding motifs (32). In this work, we show that TRIM24 links ERα activity to the local chromatin state, possibly through BAZ1B. BAZ1B is a protein that, together with SMARCA5, belongs to the ISWI subfamily of chromatin remodelers, that also encompass SWI/SNF protein complexes (46). BAZ1B is overexpressed in BC, and its higher expression was associated with worse overall and relapse-free survival in ERα+ BC patients (47). We show that TRIM24 depletion reduces BAZ1B interaction to ERα and that may lead to a more compact chromatin state at these loci. However, deeper exploration is needed to formally test this hypothesis.

The role of TRIM24 in cancer has been extensively studied in the past years, being critically involved in the development and progression of several cancer types (48), rendering TRIM24 an interesting therapeutic target. Recently, a TRIM24 degrader (23) has been developed, showing reduction of TRIM24 protein levels in acute leukemia cells but also in metastatic PDX models of ERα-negative BCs (20). In our study, and in contrast to previous work (23), TRIM24 degrader successfully impairs MCF7 cell growth. Since the previous report (23) only followed MCF7 proliferation for 5 d in the presence of TRIM24 degrader, while MCF7 doubling time is around 29 h (ATCC), the experimental design of the prior work may have prevented the detection of reduced proliferation potential of these cells. Transcriptomic analysis of dTRIM24-exposed MCF7 cells reveals strongly impaired cell proliferation and perturbation of estrogen-related gene expression (Fig. 4F). Interestingly, dTRIM24 treatment was associated with a significantly increased expression of HIF1A-related genes (Fig. 4F). HIF1A has previously been reported to be overexpressed in many cancer types and responsible for tumor survival and progression (49). One possible explanation for this phenomenon is that dTRIM24 hijacks VHL to promote TRIM24 degradation. VHL is known to function as an E3 ubiquitin ligase that targets HIF1A for degradation (50). Consequently, the pool of VHL available in the cell may be insufficient to efficiently degrade HIF1A, leading to increased expression of HIF1A target genes upon dTRIM24 treatment. In line with this hypothesis, the developers of the dTRIM24 molecule have previously reported an increase in HIF1A protein levels as a result of VHL depletion (23).

Current therapeutic strategies for ERα-positive BC patients are centered around blocking ERα activity, either by inhibiting estradiol production or by directly targeting the ERα protein itself (2). Yet, resistance is commonly observed (4). Therefore, there is a growing need to provide alternative druggable targets that can indirectly interfere with ERα activity, which remains the main driver of BC in this disease stage. We now show that targeting TRIM24 reduces ERα-positive BC cell growth, not only in the endocrine-sensitive space but also in therapy-resistant models, including drug-exposed cell lines (TAMRs, LTEDs), CRISPR engineered models (mutant-ESR1), and patient-derived organoids. Interestingly, only a subset of organoids showed clear sensitivity to dTRIM24. This intersample heterogeneity likely reflects differing dependency on TRIM24/ERα signaling between tumors and will require dedicated mechanistic follow-up not covered in this study. Moreover, mechanistic assays for apoptosis and senescence induction were not performed in the TRIM24 degrader-treated organoids, as such analyses remain technically challenging in 3D culture systems. Cumulatively, this work lays the foundation for future in vivo and clinical studies exploring the potential of TRIM24 degraders as therapeutic agents in ERα+ BC, particularly in overcoming endocrine resistance.

Materials and Methods

Cell Lines and Cell Culture.

The MCF7, T47D, ZR-75-1, and MDA-MB-231 cell lines were both purchased from the American Type Culture Collection (ATCC). MCF7 Parental, MCF7 Y537S Clone 1 and 2 (referred to in this study as MCF7 Y537S C1, C2) and MCF7 D538G Clone 3 (referred to in this study as MCF7 D538G C3) mutant cell lines were generated using CRISPR-Cas9, as previously described (7) and kindly provided by Simak Ali (Division of Cancer, CRUK Labs, University of London Imperial College, London). MCF7 LTED (44) and MCF7 TAMR (43) were previously described and generously provided by Lesley-Ann Martin (Breast Cancer Now Toby Robins Research Centre, The Institute of Cancer Research, London) and Robert I. Nicholson (Tenovus Centre for Cancer Research, Welsh School of Pharmacy, Cardiff University, United Kingdom), respectively. All cell lines, except MCF7 LTED, were kept in DMEM, high glucose, pyruvate (Gibco) and supplemented with 10% fetal bovine serum (FBS, Capricorn Scientific) and 1% penicillin–streptomycin (5,000 U/mL, Life Technologies) (full medium). MCF7 TAMR cell line was further supplemented with 100 nM 4-Hydroxytamoxifen (HY-16950, MedChemExpress). MCF7 LTED cell line was maintained in phenol-red free DMEM (Gibco) supplemented with 5% dextran/charcoal-stripped FBS (DCC medium) and 1% penicillin–streptomycin (5,000 U/mL, Life Technologies). For ligand treatment, 10 nM estradiol (E2) (HY-B0141; MedChemExpress) was used. TRIM24 degrader (dTRIM24; HY-111519 MedChemExpress) was reconstituted to a 10 mM stock solution, in DMSO and used in the described concentrations and time points. dTRIM24_2 was obtained by fee-for-service). The synthesis of dTRIM24_2 is detailed in Patent No. US 10,702,504 B2: Degradation of TRIM24 by conjugation of TRIM24 inhibitors with E3 ligase ligand and methods of use, applied by the Dana-Farber Cancer Institute in Boston. Assayed compounds were tested as TFA salts, and purities of assayed compounds were in all cases greater than 95%, as determined by reverse-phase UPLC analysis [Waters Acquity UPLC/MS system (Waters PDA eλ Detector, QDa Detector, Sample manager—FL, Binary Solvent Manager) using Acquity UPLC® BEH C18 column (2.1 × 50 mm, 1.7 μm particle size): solvent gradient = 85% A at 0 min, 1% A at 1.7 min; solvent A = 0.1% formic acid in water; solvent B = 0.1% formic acid in Acetonitrile; flow rate: 0.6 mL/min]. NMR spectra were acquired on a 500 MHz Bruker Avance III spectrometer, operating at the denoted spectrometer frequency in MHz for the specified nucleus. Unless otherwise noted, all experiments were acquired at 298.0 K with a calibrated Bruker Variable Temperature Controller. The chemical shifts are reported in parts per million (ppm), and coupling constants (J) are given in Hertz (Hz). 1H NMR spectra are reported with the solvent resonance as the reference unless noted otherwise (CDCl3 at 7.26 ppm, CD3OD at 3.31 ppm, DMSO-d6 at 2.50 ppm). Peaks are reported as (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet or unresolved, br = broad signal, coupling constant(s) in Hz, integration). All cell lines were cultured at 5% CO₂ at 37 °C, were subjected to regular Mycoplasma testing, and underwent authentication by short tandem repeat profiling (Eurofins Genomics).

Transient Cell Line Transfections.

Transient MCF7 cell transfections were performed according to the manufacturer’s instructions using Lipofectamine RNAiMAX (Invitrogen) with 50 nM of siRNA knockdown of VHL (ON-TARGETplus Human VHL siRNA, L-003936-00-0005, Dharmacon), and nontargeting control (siGENOME Non-Targeting Control siRNA Pool #2, D-001206-14-20, Dharmacon). dTRIM24 (5 μM) was added after 24 h, and siVHL transfection and dTRIM24 treatment were refreshed every 72 h.

IACS-9571 Treatment.

IACS-9571 was purchased from MedChemExpress (HY-102000B) and added to MCF7 cells at a final concentration of 5 μM. After 24 h, cells were fixed for ChIP-qPCR according to the method below described. For ERα-responsive gene levels, MCF7 were treated with IACS-9571 for 24, 48, and 72 h in FBS conditions (SI Appendix, Fig. S4B) or for 24 h together with E2 (10 nM) or DMSO, after MCF7 cells were hormone-deprived for 3 d (SI Appendix, Fig. S4C). Total RNA was isolated using Invitrogen™TRIzol™Reagent (15596026, Thermo Fisher Scientific) according to the manufacturer’s instructions. First-strand cDNA was synthesized from 2 μg of isolated RNA using SuperScript^© III First-Strand Synthesis System for Reverse Transcriptase-PCR (18080-051, Life Technologies). SensiMix™ SYBR^© No-ROX kit (QT650-05, Bioline) was used to perform RT-qPCR in a QuantStudio™ 5 System (Thermo Fisher Scientific), and results were analyzed using the QuantStudio Software. Primers used: GREB1 (forward: CCACATGGACTACGGCAAC; reverse: CTGAGGAACTGCAGGTAGGC), XBP1 (forward: GGGAAGGGCATTTGAAGAAC; reverse: ATGGATTCTGGCGGTATTGA), RARα (forward: GACCAGATCACCCTCCTCAA; reverse: GTCCGAGAAGGTCATGGTGT), and TFF1 (forward: ATCGACGTCCCTCCAGAAGA; reverse: TGGGACTAATCACCGTGCTG). β-Actin was used as a housekeeping gene (forward: CCTGGCACCCAGCACAAT; reverse: GGGCCGGACTCGTCATACT).

RIME and Coimmunoprecipitation.

RIME experiments were performed in full medium. For the ERα RIME experiment comparing between vehicle and E2 stimulation, cells were cultured in hormone-deprived medium (DCC medium) for 3 d prior to incubation with 10 nM of E2 or DMSO for 3 h.

Cells were fixed, lysed, and sonicated accordingly to the protocol previously described (51), with the difference of having performed 15 cycles of 30 s ON, 30 s OFF in the sonication step, using the Bioruptor^® Pico (B01060001; Diagenode). Obtained nuclear lysates were incubated overnight with 50 μL of magnetic protein beads A (10008D, Invitrogen) conjugated with 10 μg of ERα antibody (06-935, Merck Millipore), 10 μg of TRIM24 antibody (NB100-2596, NovusBio), or 10 μg rabbit IgG control (12-370, Merck Millipore). Peptide mixtures were prepared and analyzed by mass spectrometry as previously described (51), with the following modifications.

Peptide mixtures (10% of total digest) were loaded directly onto the analytical column and analyzed by nanoLC-MS/MS on a mass spectrometer (Dataset S2) equipped with either a Proxeon nLC system (Thermo Scientific). Solvent A was 0.1% formic acid/water, and solvent B was 0.1% formic acid/80% acetonitrile. Peptides were eluted from the analytical column at a constant flow of 250 nL/min in a 90-min gradient (Dataset S2) or equipped with an Evosep One LC system. The peptides were reconstituted in 0.1% formic acid and loaded on the Evotip Pure™ (Evosep). Peptides were separated using the preprogrammed gradient (Dataset S2, 88 min gradient 15SPD) on an EV1137 (Evosep) column with an EV1086 (Evosep) emitter. Raw data were analyzed by MaxQuant (Dataset S2) using standard settings for label-free quantitation (LFQ). MS/MS data were searched against the SwissProt Human database (Dataset S2) complemented with a list of common contaminants and concatenated with the reversed version of all sequences. The maximum allowed mass tolerance was 4.5 ppm in the main search and 0.5 Da for fragment ion masses. False discovery rates for peptide and protein identification were set to 1%. Trypsin/P was chosen as cleavage specificity allowing two missed cleavages. Carbamidomethylation was set as a fixed modification, while oxidation and deamidation were used as variable modifications. LFQ intensities were Log2-transformed in Perseus (Dataset S2), after which proteins were filtered for at least three out of four valid values in at least one sample group. Missing values were replaced by imputation based on a normal distribution (width: 0.3 and downshift: 1.8). Differentially expressed proteins were determined using a Student’s t test.

For coimmunoprecipitation analyses, the protocol follows as the previously described RIME protocol with the difference that obtained protein lysates were incubated at 95 °C for 10 min in 2× Laemmli lysis buffer (120 mM Tris, 20% glycerol, and 4% SDS) and supplied with 100 mM DTT before proceeding with the western blot protocol. The same antibodies for RIME were used for the immunoprecipitation studies.

ChIP Analyses.

Cells were fixed with 1% formaldehyde (1039991000; Merck) added directly on the cell culture medium for 10 min at room temperature and subsequently quenched with 0.125 M glycine. Cells were then lysed as previously described (51) and sonicated for 15 cycles (30 s ON, 30 s OFF) using the Bioruptor^® Pico (B01060001, Diagenode). For each ChIP sample, 50 μL of magnetic protein A beads (10008D, Thermo Fisher Scientific) were conjugated to 5 μg of antibody: ERα (06-935, Milipore), TRIM24 (NB100-2596, NovusBio), H3K23ac (39131, Active Motif), H3K27ac (39133, Active Motif), and phospho-Ser RNA polymerase II (5095, Abcam). Immunoprecipitated DNA was submitted for library preparation using a KAPA library kit (KK8234, Roche) and subsequently sequenced on the Illumina HiSeq2500 platform. Sequence of TRIM24 and ERα in MCF7 WT was performed by single-end protocol and with a read length of 65 bp, whereas TRIM24 in MCF7 treated with E2 or DMSO, ERα, H3K23ac, H3K27ac, and phospho-Ser RNA polymerase II, in FBS, were performed by paired-end protocol and with a read length of 51 bp. All samples were aligned to reference genome Hg38/GRCh38 using Burrows-Wheeler Aligner [BWA v0.5.10 (52)]. Reads were filtered based on mapping quality (MAPQ ≥ 20), and duplicate reads were marked with PicardMarkDuplicate (v2.19.0). MACS2 (v2.1.2) was used to perform peak calling over input ChIP-seq samples in both single-end and paired-end mode (respective to the type of experiment). Single-end mode MACS2 peak calling was performed with phantomQualPeaktools estimation of the fragment length (53, 54). DeepTools (v2.5.3) was used to calculate the fraction of reads in peaks [FRiP; (55)]. For visualization purposes, mapped reads of each replicate sample were merged using SAMtools [v1.10; (56)]. The DiffBind (57) R package (v2.10) was used to perform TRIM24 differential binding between E2 and DMSO-treated MCF7 cells and ERα differential binding analysis between NT1 and TRIM24-KO A, using a false discovery rate (FDR) < 0.05 and to generate consensus peaklist. The same obtained peaklist was used for subsequent H3K23ac and H3K27ac differential binding between NT1 and TRIM24-KO A. Genome browser snapshots were generated using the R v4.0.3 environment and Rseb (v0.3.1) (https://github.com/sebastian-gregoricchio/Rseb) (58), while tornado plots have been produced using deepTools (v2.5.3). Motif enrichment analysis was performed using the Galaxy Cistrome SeqPos motif tool (59).

For ChIP-qPCR, MCF7 cells TRIM24-KO B and NT 2 were seeded in normal medium (SI Appendix, Fig. S3A) or MCF7 WT cells were treated with IACS 9571, as previously described (SI Appendix, Fig. S4A). Cells were fixed and processed for ChIP as described above. The DNA samples were used for qPCR using the SensiMix SYBR kit (Bioline) according to the datasheet provided by the manufacturer. The amplification signal was detected using a QuantStudio 6 Flex system (Thermo Fisher Scientific). The primer sequences for DNA binding detection were XBP1 (forward: ATACTTGGCAGCCTGTGACC; reverse: GGTCCACAAAGCAGGAAAAA), RARα (forward: GCTGGGTCCTCTGGCTGTTC; reverse: CCGGGATAAAGCCACTCCAA), GREB1 (forward: GAAGGGCAGAGCTGATAACG; reverse: GACCCAGTTGCCACACTTTT), TFF1 (forward: CCCGTGAGCCACTGTTGTC: reverse: CCTCCCGCCAGGGTAAATAC), and IGFBP4 (forward: ACAAACCACGGTGCAGAGAA; reverse: TCCACATGTGCCTTACCCAC). As a negative control region, the last intron of CCND1 (NEG - forward: TGCCACACACCAGTGACTTT; reverse: ACAGCCAGAAGCTCCAAAAA) was used.

CRISPR/Cas9-Mediated Knockout Cell Lines Generation.

TRIM24 targeting single-guide RNA (TRIM24-KO 1: GGCAACGAATGACTCCAACT) and nontargeting control guide RNA (NT 1: AACTACAAGTAAAAGTATCG; NT 2: GTATTACTGATATTGGTGGG) were separately cloned into the lentiCRISPR v2 vector (60). Using H3K293T cells, the CRISPR vectors were cotransfected with third-generation viral vectors using polyethyleneimine (PEI, Polysciences). After lentivirus production, the medium was harvested and added to the MCF7cells. Two days after infection, cells were selected for 2 wk with 2 μg/mL puromycin (Sigma-Aldrich), and knockout efficiency was confirmed by western blot.

Monoclonal CRISPR/Cas9 TRIM24-KO MCF7 cells (TRIM-24 KO A and B) were generated by single-cell Flow cytometry-sorting (FACS) per well of the previously generated polyclonal TRIM24-KO 1 MCF7 cells line. When reaching the 6-well plate confluency, cells were incubated with 2 μg/mL puromycin and the survived populations were tested by western blot.

Western Blot.

Total protein lysates were obtained using Laemmli buffer complemented with 1x complete protease inhibitor cocktail (Roche) and 1× phenylmethylsulfonyl fluoride. 40 μg of protein per sample was resolved in 8% acrylamide gel (MiliQ, 40% acrylamide, 1.5 M Tris pH 6.8, 10% SDS, 10% APS, and TEMED) in SDS-PAGE 1× running buffer (25 nM Tris, 0.25 M glycine, and 0.1% SDS) and sequentially transferred to a 0.45 μm nitrocellulose membrane (Santa Cruz Biotechnology), overnight at 4 °C. Upon 2 h blocking with 3% bovine serum albumin (BSA; A8022, Sigma/Merck) diluted in PBS-T (1× PBS, 0.01% Tween), membranes were incubated overnight at 4 °C with the respective primary antibodies diluted in blocking solution: ERα (1:500 dilution; MA5-14104, Thermo Fisher Scientific; for organoids ERα protein levels: 06-935 Sigma 1:1,000), TRIM24 (1:5,000 dilution; NB100-2596, NovusBio), GREB1 (1:1,000, ab72999, Abcam; for organoids GREB1 levels: MABS62, Millipore, 1:1,000), TLE3 (1:1,000, ab94972, Abcam), GAPDH (sc-47724, Santa Cruz Biotechnology), and Actin (1:5,000 dilution MAB1501R, Merck Millipore). After three washes with PBS-T, membranes were incubated with the respective secondary antibodies: donkey-α-rabbit 800CW (926-32213, LI-COR Biosciences, 1:10,000) and donkey-α-mouse 680 RD (926-68072, LI-COR Biosciences, 1:10,000) diluted in blocking solution for 1 h. The Odyssey^® CLx Imaging system (Li-Cor Biosciences) and ImageStudio™ Lite v.5.2.5 software (LI-COR Biosciences) were used to scan and visualize the proteins.

Cell Proliferation and Viability Assays.

All cell lines were seeded at 1,000 cells/well in a 96-well plate for cell counting assay, 500 cells/well in a 364-well plate (except MCF7-mutants that were seeded 250 cell/well) for CellTiter-Glo (CTG) assay and 20,000 cells/well in a 24-well plate (except MCF7 TAMR and LTED that were seeded 10,000 cells/well) for Crystal violet assays. The next day, dTRIM24 5 μM or the indicated concentrations were added to the cells and incubated for the indicated days (with a change of medium + dTRIM24 every 72 h). For cell counting assays, cells were trypsinized on the indicated day and counted using a cell counter (CellDrop BF, DeNovix). The CTG assays were performed according to the manufacturer’s protocol, and luminescence was detected using the Tecan system (Infinite^® M Plex, Tecan). Crystal violet stainings were performed by fixing the cells with 1% formaldehyde (Merck) and, after three washes with 1× PBS, stained with 0.1% Crystal Violet solution (Sigma Aldrich, SLBW3435). For quantification, crystal violet–stained cells were incubated with 10% acetic acid solution (Honeywell), and absorbance at 590 nm was measured using the Tecan system (Infinite^® M Plex, Tecan). Obtained values were normalized to the respective DMSO conditions. Bar charts and scatter plots were plotted in GraphPad Prism 9 software.

For colony assays with dTRIM24_2, MCF7 TAMR cells were seeded in duplicate into 6-well plates at a density of 10,000 cells per well in DMEM supplemented with Tamoxifen. A range of doses from 0.1 to 5 of dTRIM24-2 were tested. The medium was refreshed twice a week. After 14 d, colony confluence was assessed using the Incucyte Live-Cell Imaging and Analysis System (Sartorius). Each cell line and condition were tested in biological triplicate.

RNA-seq.

MCF7 TRIM24-KO monoclonals (A and B) and MCF7 NT (1 and 2) cells were cultured in full medium, whereas for dTRIM24 RNA-seq, MCF7 WT cells were cultured in full medium and incubated with 5 μM of dTRIM24 or DMSO for 24 h. After two washes with cold PBS, cells were collected in RLT buffer (79216, Qiagen) prior to RNA extraction by the RNeasy mini kit (Qiagen), according to the manufacturer’s protocol. Libraries were generated with the TruSeq RNA Exome kit (Illumina) and sequenced in the Illumina NovaSeq 6000 (Illumina) platform using paired-end protocol and 51 bp reading length. After sequencing, data were aligned to the human reference genome Hg38/GRCh38 using HISAT2 [v2.1.0 (61)] and to calculate the number of reads per gene HTSeq count [v0.5.3 (62)] was used. Gene expression differences between KO/nontarget or treated/nontreated samples, after QC filter, were determined using DESeq2 [v1.22.2 (63)]. Differentially expressed genes have been ranked by log2(FoldChange expression) and used for GSEAs on the hallmark gene set (H) from msigdbr (v7.5.1) using clusterProfiler [v3.18.1 (64)], [PvalueCutoff = 0.05, pAdjustMethod = “BH”]. GSEA enrichment plots have been generated using the plot.gsea function from Rseb package [v0.3.2 (65)].

Organoids.

All organoid work was carried out at Royal College of Surgeons in Ireland University of Medicine and Health Sciences. Tumor-derived organoids were developed under the ethical approval granted by the Institutional Review Board of Royal College of Surgeons in Ireland with the exception of HCI-11 which was a kind gift from Alan Welm (Department of Oncological Sciences, University of Utah, Salt Lake City, UT; Huntsman Cancer Institute, University of Utah, Salt Lake City, UT) (66). Clinical samples for this project were sourced from the Breast Cancer Proteomics and Molecular Heterogeneity observational clinical trial (NCT01840293). Written informed consent was obtained from all patients prior to tissue acquisition for organoid culture. We utilized the standard organoid procedure to generate organoid lines from the collected tumors, supplemented with estradiol for those with ERα+ tumors (67). For the subsequent intervention experiment, these mature organoids were dissociated and cultured in organoid-specific medium enriched with 5% of Cultrex^® Reduced Growth Factor Basement Membrane Matrix (BME, Trevigen, 3533-001-02). Organoids were subjected to either vehicle or specified treatments at the denoted concentrations (n = 6; biological replicates). The viability of the cells was then evaluated 7 d posttreatment using the CellTiter-Glo^® 3D Cell Viability assay (Promega). For downstream analysis, organoids were processed as follows. After removing the BME, each pooled condition was resuspended in Buffer RLT + β-ME to lyse the organoid pellet according to the Qiagen kit protocol. RNA was isolated using the Qiagen RNeasy Kit, and cDNA was synthesized using the SuperScript III First-Strand Synthesis System (Invitrogen). Real-time PCR was performed using TaqMan probe technology (Applied Biosystems) with the following probes: Hs00940446_m1 (RARα), Hs00231936_m1 (XBP1), Hs00536409_m1 (GREB1), and β-Actin (Human ACTB 20×). The comparative CT (ΔΔCT) method was employed to analyze relative gene expression levels. Whole cell protein lysates were collected as previously described and quantified using the BCA kit (Thermo Fisher, Cat number 23227). Proteins were separated by 10% SDS–PAGE, transferred to nitrocellulose membranes, and blocked at room temperature in 3% BSA in 0.1% Triton-X 100 in TBS. Membranes were incubated with primary antibodies as detailed already in Materials and Methods section.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

The ChIP-seq and RNA-seq data have been deposited to the GEO database (GSE246796). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (68) partner repository with the dataset identifier PXD046736.