Defective DNA Damage Response Is a Targetable Therapeutic Vulnerability in ESR1-Mutant Breast Cancer.

Introduction
Nearly 4 million women currently live with breast cancer in the United States, and up to 13% of women will be diagnosed with breast cancer in their lifetime.1,2 While mortality rates have decreased in recent years, up to 20% of breast cancer will recur or metastasize.1,2 For patients with breast cancer expressing the estrogen receptor (ER), one of the leading causes of metastatic progression is the emergence and enrichment of mutations in the ER gene (ESR1m) conferring ligand independent activity and resistance to ER-targeting endocrine therapy (ET).3,4 Genomic surveillance of metastatic breast cancer (MBC) has revealed emergent ESR1m in up to 50% of ER-positive (ER+) MBC patients.5 Acquisition of these mutations contributes to both decreased therapeutic benefit and reduced overall survival.5 Though emergence and enrichment of ESR1 mutations is a significant clinical issue in MBC, to date only one therapeutic option, the selective estrogen receptor degrader (SERD) elacestrant, has been approved specifically for ESR1m MBC.6 Therefore, identification of effective therapeutic options for the treatment of ESR1m MBC is a top clinical priority.
While only one therapeutic specifically targeting mutant ERs has been FDA-approved6, extensive pre-clinical investigations have identified a variety of potentially targetable phenotypes associated with ESR1m acquisition. Acquired phenotypes include enhanced stem-cell activity, enrichment of the epithelial to mesenchymal transition (EMT), and increases in growth factor receptor (GFR) expression.7–10 Unfortunately, while evaluation of GFR targeting drugs is underway in ER+ metastatic breast cancer7,11, the other phenotypes lack effective targeted therapeutics to date. We therefore sought to identify a unique “druggable phenotype” with a variety of therapeutic options. In this study, we identified an enrichment of the replication stress response (RSR) and DNA damage response (DDR) pathways enabled by poly ADP ribose polymerase 1 (PARP1) in multiple ESR1m models. The RSR induces cell cycle checkpoints and recruits DDR proteins to ensure effective DNA replication and facilitate DNA repair through activation of ataxia telangiectasia and Rad3 related (ATR) and checkpoint kinase 1 (Chk1) proteins.12 PARP1 is a critical protein for the effective repair of DNA strand breaks, and fidelity of DNA replication.13 PARP inhibitors are synthetic lethal in breast cancer with defective DNA repair such as breast cancer gene (BRCA) mutations.14 The clinical efficacy of disrupting the DDR with PARP inhibitors has led to significant efforts in the clinical development of cell cycle checkpoint inhibitors of ATR and Chk1 to target the RSR.15,16 Cancer is highly proliferative which can lead to genomic instability and accumulation of DNA damage17. Herein, we propose that the constitutively active and highly proliferative ESR1m breast cancers are a unique subset of breast cancer that exhibits a novel sensitivity to both PARP and cell cycle checkpoint inhibitors due to defective RSR and enriched DDR.
We identified significant enrichment of defective RSR in ESR1m models resulting in elevated intrinsic DNA damage. Inhibition of the RSR and DDR utilizing Chk1 and PARP inhibitors was synergistic in cell line models and significantly reduced the progression of ESR1m metastatic lesions. PARP inhibition in combination with the SERD fulvestrant (Fulv) reduced primary tumor growth in ESR1m models and was synergistic in ESR1m cell lines. Thus, defective RSR in ESR1m breast cancer drives metastatic tumor growth, and inhibition of the RSR and downstream DDR utilizing Chk1 and PARP inhibitors is an effective therapeutic strategy.

Materials and Methods

Cell Culture and Maintenance
MCF-7 cells (RRID: CVCL_0031), CRISPR-engineered MCF-7 Y537S ESR1m (Y537S), and long-term estrogen deprived cells with emergent Y537C (Y537C)7 were cultured in MEM (Corning MT10–010-CV) media supplemented with 10% fetal bovine serum (GeminiBio 100–500-500), 100 U/mL penicillin and 100 μg/mL streptomycin (GeminiBio 400–109-100), 1X MEM non-essential amino acids (Corning MT-25–025-CI), 50 μg/mL Gentamicin (ThermoFisher, 15750060), and 5 μg/mL Plasmocin Prophylactic (InvivoGen, NC9886956). ZR75–1 (RRID:CVCL_0588) cells were cultured in RPMI 1640 (Corning 10104CV) media supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin. Both cell lines were originally obtained from the Bill McGuire laboratory at UT San Antonio Health Science Center and did not undergo STR authentication. Cells were screened for ER and ER-regulated protein expression and morphology changes upon start of each culture. For studies with estrogen reduced conditions, cells were cultured in phenol-free MEM media (Corning, MT 17–305-CV) supplemented with 5% charcoal stripped fetal bovine serum (GeminiBio 100–119-500) instead of 10% fetal bovine serum, the additions above, and 200 μM L-glutamine (Corning, MT-25–005-Cl). Cells were cultured at 5% CO2 in a 37°C incubator. Cells with long-term treatment exposure to Fulv (100 nM) or Abemaciclib (Abema) (500 nM) were treated twice per week. All cells were passaged weekly for up to 10 passages and tested for mycoplasma every 30–60 days utilizing the MycoAlert Detection Kit (Lonza, LT07–318). Detailed information for drug sources can be found in Supplemental Table 1.

Immunoblot Assays
Cells were harvested utilizing cell scraping on ice and washed with PBS. Cells were lysed for 30 minutes utilizing RIPA buffer (Boston BioProducts, BP-115) supplemented with 1:100 phosphatase and protease inhibitor cocktail sets I, II, and III (EMD Millipore, 539134, 524625, 524624) followed by centrifugation at 14000 rpm for 15 minutes at 4°C. Protein concentrations were determined utilizing the Pierce BCA Protein Assay kit (ThermoFisher 23225) per manufacturer’s instructions. Cell extracts were resolved by electrophoresis in 4–15% polyacrylamide gels (BioRad 5671085) in 1X Tris/Glycine/SDS buffer (BioRad 1610772) at 80V for 20 minutes followed by 120V until appropriate sample separation was reached. Proteins were transferred utilizing Trans-Blot Turbo Nitrocellulose Transfer Packs (BioRad 1704159) in 1X Tris/Glycine Buffer (BioRad 1610771) at 1A, 25V for 30 minutes. Membranes were stained with Ponceau solution (Sigma P7170) to confirm protein transfer. Blots were blocked overnight at 4°C in 5% milk in 1X TBST solution (Fisher Scientific AAJ77500K8). Antibodies were incubated with membranes overnight in 5% BSA (Sigma A7030) in 1X TBST for phosphorylation antibodies or 5% milk in 1X TBST for other antibodies. Membranes were washed with 1X TBST and secondary antibodies were incubated for 1 hour at room temperature in 5% milk in 1X TBST. Membranes were imaged utilizing ChemiGlow reagents (BioTechne PN 60–12596-00) on the BioRad ChemiDoc instrument. Gels re-probed with multiple antibodies (such as phosphorylated and total protein) were stripped after imaging with stripping buffer (ThermoFisher 21063) followed by 3×10-minute washes with 1X TBST followed by membrane blocking overnight in 5% milk in 1X TBST. Primary and secondary antibody incubation for re-probes and imaging were repeated as described above. Antibody details and concentrations can be found in Supplemental Table 2. Blots provided are representative.

Isolation of chromatin bound protein fraction
The isolation of the chromatin bound protein fractions were performed in two ways: Cells were treated using single agent or a combination of Olaparib (5 μM), Veliparib (5 μM) and Fulvestrant (10–100 nM) for 48–72 hours, then harvested. The chromatin bound fraction was isolated using the subcellular protein fractionation kit for cultured cells according to the manufacturer protocol (ThermoFisher #78840) or were isolated according to a published protocol with small adjustments. 18 The chromatin bound nuclear extract protein concentration was determined with the Pierce BCA Protein Assay kit (ThermoFisher 23225) per manufacturer’s instructions and 15 ug of protein per sample were resolved by electrophoresis as described above for the immunoblot assays. Chromatin-bound protein immunoprecipitation was performed as described previously.18 Cells were lysed in cold cytoskeleton (CSK) buffer, followed by extraction using IP lysis buffer supplemented with protease inhibitors (Roche) and 20 mM N-ethylmaleimide (Sigma-Aldrich) to enrich the chromatin-bound protein fraction.18 Proteins were eluted with 2X Laemmli sample buffer, boiled, and separated by SDS–PAGE. Immunoblotting was performed using antibodies against PARP, ER (BioRad) and Histone H3 as a loading control. Chromatin-bound input lysates were run in parallel to assess enrichment of PARP and ER.

Microarray Analysis
Microarray analyses were performed as described in Gu et al.8 and Dustin et al.7

Gene Set Enrichment Analysis
Gene set enrichment analysis (GSEA) was performed utilizing the GSEA software from the Broad Institute (RRID:SCR_003199) as described in Dustin et al. with a significance cutoff of FDR<0.25.7 Genes in microarray (Y537S and Y537C mutant cell lines) or RNA-seq (F+A mutant and Abema control cell lines) analyses were sorted by high to low fold change as ranked lists. These ranked lists were analyzed for enrichment of the defective RSR signatures and of HALLMARK MSigDB pathways. For defective RSR signatures we utilized differentially expressed genes reported from the Ataxia telangiectasia and Rad3 related (ATR), and checkpoint kinase 1 (Chk1) shRNA knockdown models developed by McGrail et al.19

Survival Analysis
McGrail et al. utilized shRNA to individually knockdown key replication checkpoint proteins ATR or Chk1 in a hyperplastic cell line to develop ATR and Chk1 defective RSR signatures.19 We overlapped shared differentially expressed genes reported from the shATR and shChk1 models (named the ATR/Chk1 defective RSR signature) with differentially expressed genes in our Y537S or Y537C cell lines versus parental WT from microarray analyses (FC>1, step-up p-value<0.05). 313 genes exhibited shared differential gene up-regulation in the Y537S and Y537C ESR1m models and shChk1 and shATR McGrail et al.19 knockdown cell lines. We named this 313 gene signature the ESR1m defective RSR signature. These 313 genes were converted into gene signature scores in patient expression data utilizing the sum of z-scores method as described in Gu et al.8 High vs low patient signatures were evaluated utilizing cut points as indicated in figure legends. We assessed disease specific survival and overall survival using the METABRIC dataset20 (1508 ER+ samples) and overall survival using the SCAN-B cohort (1747 ER+ samples).21 Log-rank p-values were calculated utilizing the “survdiff” command in the “survival” package of R; significance was achieved at p<0.05. Kaplan-Meier curves were drawn with the “survfit” command in the same R package. Plots were generated utilizing the ggplot2 package (RRID:SCR_014601). Studies were conducted in accordance with U.S. Common Rule.

Comet Assays
Comet assays (R&D Systems 4250–050) were performed with minor modifications to the manufacturer’s protocol. Cells were cultured for 48 hours in MEM charcoal stripped serum media. Cells were resuspended at 1×105 cells/mL in charcoal stripped serum media. Cell suspensions were mixed with 1% low melt agarose at a 1:10 ratio and plated on 2-well comet slides. Cells were lysed overnight at 4°C in lysis solution. For the alkaline comet assay DNA was treated with alkaline unwinding solution (200 mM NaOH, 1 mM EDTA in distilled water) for 20 minutes at 25°C. Slides were electrophoresed in alkali unwinding solution at 20V for 30 minutes. For the neutral comet assays, DNA was unwound in neutral unwinding solution for 30 minutes at 25°C and electrophoresed in the neutral unwinding solution at 21V for 45 minutes. After electrophoresis the slides were washed in distilled water and 70% ethanol before drying at 37°C for 30 minutes. DNA was stained with SYBR gold solution according to manufacturer’s directions (ThermoFisher Scientific, S11494). Fluorescence microscopy to quantitate comet tails was performed at 10X magnification using the Keyence BZ-X800 microscope and 4X using the BioTek Cytation 5 epi-fluorescence microscope. Analysis of tail moments was performed using the Comet Assay IV software from Instem with quantification of at least 150 comet tails per sample. Statistical significance was determined utilizing one-way ANOVA in GraphPad PRISM (RRID:SCR_002789) with a significance cutoff of p-value<0.05.

MTT Growth Assays and Synergy
Cells were plated at 1000 cells per well in 96-well plates with all treatments in minimum quadruplicate. Cells were treated for 7 days with media and drug replenished every 48 hours. At day 7, cells were treated with MTT solution (1:2 dilution of 1 part 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (Thermo Fisher M6494) in PBS to 2 parts cell culture media) for 2–4 hours followed by DMSO for 20 minutes. Absorbances were read at 570 nM and background subtracted from 655 nM. Viability was determined relative to no treatment control. Statistical significance was determined utilizing GraphPad PRISM software to perform multiple t-tests with a significance cutoff of p<0.05. Synergy values and IC50s were calculated utilizing the CompuSyn software (RRID:SCR_022931) for non-constant or constant ratios.

Cell Cycle Analysis
Cells were starved for 48 hours before treatment with 10 μM Lovastatin for 36 hours. Media was replaced after Lovastatin treatment and cells were treated with additional drugs for 48 hours (PF00477736 (500 nM Chk1i) or Olaparib (Olap) (1 μM or 5 μM). Cells were harvested via trypsinization and washed twice with 1X PBS. Cells were manually single cell suspended by repeat pipetting in 70% ethanol overnight for fixation. Approximately 1×106 cells per condition were pelleted and washed with 1X PBS to remove ethanol after fixation. Cells were resuspended in propidium iodide staining solution (0.1% Triton X-100 (Sigma 93443), 0.2 mg/mL DNA-free RNase A (Invitrogen 12091021), 0.02 mg/mL propidium iodide (Molecular Probes P4864) in 1X PBS) for 15 minutes protected from light. Propidium iodide positive cells were analyzed utilizing the BD FACSCanto II flow cytometer with 10,000 events per condition. Cell cycle analysis was performed utilizing FlowJo software (BD BioSciences, RRID:SCR_008520).

Digital Droplet Polymerase Chain Reaction
ddPCR was performed utilizing the BioRad platform as described in Dustin et al.7 Probe IDs were as follows, with 3/1000 droplets defined as positive detection of ESR1 mutations: Y537C dHsaMDS732897750, Y537N dHsaMDS296069817, Y537S dHsaMDS975379796, D538G dHsaMDS460485301.

In Vivo Tumor Growth and Metastasis Experiments
Xenograft tumor methods for the WHIM20 and MCF-7 CRISPR Y537S in vivo experiments were performed as described in detail in Gu et al.8 Removal of estrogen supplementation and treatment randomization was performed when tumors reached approximate volumes of 200 mm3 for WHIM20 experiments and 350 mm3 for CRISPR Y537S experiments. WHIM20 experimental treatment specifications: Olap 50 mg/kg 5x/week via oral gavage, Fulvestrant 200 mg/kg 1x/week via subcutaneous injection. MCF-7 CRISPR Y537S xenograft treatment specifications: Olap 50 mg/kg 5x/week via oral gavage, Chk1 inhibitor (Chk1i) PF00477736 7.5 mg/kg BID 2x/week via intraperitoneal injection. Tumor resections were performed at 800 mm3 and mice were harvested when exhibiting signs of moribund behavior or 6 months post-resection. Statistical analysis of metastatic frequency was conducted by comparing lung micrometastatic foci/mouse between treatment groups utilizing two-way ANOVA. Primary tumor growth of all xenograft experiments was compared utilizing Kaplan-Meier analysis in the PRISM software (GraphPad, RRID:SCR_002798) with growth determined relative to volume at time of randomization. All animal studies were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee (IACUC) and were conducted in accordance with institutional guidelines.

RNA-Sequencing
Total RNA was isolated using the RNeasy Mini Kit (QIAGEN 74104) according to the manufacturer’s instructions. Library preparation and RNA-sequencing analysis was conducted by the Baylor College of Medicine Genomic and RNA Profiling Core (GARP) using the Illumina NovaSeq 6000 platform. Paired-end reads were trimmed using TrimGalore and mapped to the UCSC hg38 genome build using HISAT2.22 Aligned reads were counted against the Gencode gene model annotation23 to obtain expression values by using FeatureCounts.24 Differential gene expression was evaluated using the R package edgeR25, with TMM (trimmed means of m values) normalization. Significance was achieved with an adjusted p-value of <0.05.

Overrepresentation Analysis
Overrepresentation analysis was performed to detect enrichment of gene sets corresponding to pathways and biological processes based on differential expressed genes (DEGs) in the ESR1m defective RSR signature or long-term ESR1m models compared to parental cells (fold change>1.5, FDR<0.05) using the Hallmark and KEGG compendia from MSigDB (v7.5.1) and the Molecular Signature Database methodology (MSigDB).26 A hypergeometric test was used to assess the enrichment with significance achieved at an adjusted p-value<0.05.

BrdU Assays
Cells were treated with indicated drugs (0.5 mM Chki, 5 μM Olap) for 48 hours, with 0.5 mM hydroxyurea treatment to induce replication stress (RS) introduced 24 hours into relevant drug conditions. After treatment, cells were washed 1x with PBS before incubation with 20 μM BrdU (BD Pharmigen 550891) for 2 hours at 37°C. Cells were washed with PBS, trypsinized, and fixed in 70% ethanol in PBS and stored at −20°C until analysis. Approximately 1.5×106 cells were collected per sample and resuspended in 2 M HCl (Sigma 1.09063) denaturing solution for 20 minutes at room temperature. Cells were washed with 0.5% bovine serum albumin solution in PBS before incubation in 0.1 M sodium tetraborate (pH 8.5) (ThermoFisher J62902.AP) for 2 minutes at room temperature. Cells were again washed in 0.5% bovine serum albumin solution in PBS. Cells were incubated with FITC anti-BrdU antibody (BioLegend 364104) for 1 hour protected from light. Cells were again washed then labeled with propidium iodide as described above. Cells were resuspended in 1x PBS and analyzed utilizing the BD FACSCanto II flow cytometer with 10,000 events per condition. Cells were gated for BrdU and propidium iodide staining and counted utilizing the FlowJo software (BD BioSciences, RRID:SCR_008520).

Proximity Ligation Assays
Cells were plated in 10 cm cell culture dishes and treated with 5% charcoal stripped serum media for 5 days. Cells were trypsinized and plated in 8-well glass imaging slides (Ibidi 80826) at 20K cells per well in 5% charcoal stripped serum media and treated with indicated drugs for 48 hours. Cells were washed with PBS and fixed in 4% paraformaldehyde solution (Fisher Scientific 043368.9M, diluted in ultrapure H2O) for 10 minutes at room temperature. Fixed cells were washed with PBS 3x followed by 10 minutes in permeabilization solution (0.2% triton-100x, 0.2% bovine serum albumin in PBS). Permeabilized cells were washed with PBS and blocked with 1% bovine serum albumin in PBS for 30 minutes. Primary antibodies (ER6F11 1:100, PARP1 ActiveMotif,1:100) were incubated overnight at 4°C diluted in 1% bovine serum albumin in PBS. Cells were washed 3x with PBS and incubated with 1:5 diluted PLA probes (Duolink In Situ PLA Probes anti-rabbit DUO92002 and anti-mouse DUO92004) in 1% bovine serum PBS for 1 hour at 37°C. Ligation and amplification steps were performed per manufacturer instructions (Wash buffers: DUO82049, Duolink In Situ Detection Reagents Green: DUO 92014). Slides were rinsed with PBS and cells stained with 300 nM DAPI (ThermoFisher D1306) in PBS for 10 minutes. Slides were washed in PBS 2x and covered with Vectashield Mounting Medium (Vector Laboratories H-1000–10) for long-term storage.
PLA were imaged utilizing the Olympus IX83 epifluorescence deconvolution microscope with the 40X dry objective for quantitation located in the Baylor College of Medicine Integrated Microscopy Core.. Quantitation was performed utilizing CellProfiler27(RRID:SCR_007358) in consultation with the Baylor College of Medicine Integrated Microscopy Core. Graphs are depicted as foci/nucleus from representative images. Statistical significance was determined via one-way ANOVA relative to 5% charcoal stripped serum media control.

Immunocytochemistry
Cells were seeded on chamber slides (ThermoFisher Lab-Tek II 154526) at 20K cells per chamber, then grown in 5% charcoal stripped media for 4 days and fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton-X100, and blocked with 5% BSA in PBSt buffer, and probed with a diluted primary antibody (1:100–1:500) in 1% BSA in PBSt at 4 °C for 12 h on a shaker. After washing, the cells were incubated in the secondary antibody (1:200, Thermo Fisher AlexaFluor 488 A11008) diluted in 1% BSA in PBSt in the dark, for 1 h at 25 °C. Slides were rinsed with PBS and cells stained with 300 nM DAPI (ThermoFisher D1306) in PBS for 10 minutes. Slides were washed in PBS 2x and covered with Vectashield Mounting Medium (Vector Laboratories H-1000–10) for long-term storage. Slides were imaged with the Olympus IX83 epifluorescence deconvolution microscope with the 40X dry objective for quantitation. Quantitation was performed utilizing CellProfiler27 (RRID:SCR_007358).

Organoid Growth Assays
Organoids were plated, treated, and imaged as described in Dustin et al.7

Statistical Analysis
Statistical tests are noted in the above relevant methods sections. Significance is set at p<0.05 unless otherwise noted. Replicates of samples are indicated in relevant figure legends.

Supplemental Information
Supplemental information includes 20 figures and 8 tables and can be found with this article.

Data Availability
mRNA expression (z-scores) and clinical outcome data for the METABRIC breast cancer cohort20 were obtained from CBioPortal. SCAN-B21 breast cancer expression dataset was obtained from the Gene Expression Omnibus (RRID:SCR_005012) under accession GSE81540 and associated sub-series. Survival analyses using SCAN-B21 were performed using the outcome variables available for this dataset. RNA sequencing data generated by the authors was deposited in the Gene Expression Omnibus and is publicly available under accession GSE314294. Additional data generated by the authors is available upon request.

Results

ESR1 mutant cells exhibit defective RSR and DDR.
We previously developed two ESR1m cell lines, a CRISPR/Cas9 MCF-7 cell line expressing the homozygous ligand binding domain activating Y537S ESR1m (clone YS1), and a long-term estrogen deprived (LTED) MCF-7 cell line with spontaneously acquired ESR1 mutation Y537C.7,8 Extended culture of the LTED cell line resulted in an outgrowth of the Y537C ESR1m at an allele frequency of 50% (Supplemental Figure 1 A,B). The CRISPR Y537S and LTED Y537C cell lines exhibited both enhanced metastatic potential and up-regulation of epithelial-mesenchymal transition (EMT), growth factor and proliferative cell cycle-driven pathways.7,8 These cell lines were also estrogen-independent, and exhibited constitutive ER-regulated activity with hyperproliferation compared to WT ER cells.7,8
Due to their hyperproliferative phenotype we investigated RSR activation in ESR1m models. RSR activation is a critical pathway for maintaining DNA integrity and preventing lethal genomic instability in response to enhanced proliferation.17,28 To investigate the clinical significance of defective RSR in ESR1m breast cancer, we developed a prognostic gene signature from up-regulated differential gene expression shared between four models: the Chk1 and ATR knockdown cell lines developed by McGrail et al19 and our Y537S and Y537C ESR1m models (Figure 1A). The 313 gene signature representing up-regulated gene expression shared between these four models was termed the ESR1m defective RSR signature (Supplemental Table 3). We next utilized the METABRIC ER+ breast cancer cohort20 to evaluate survival outcomes of the ESR1m defective RSR signature in primary breast cancer patient tumors. Primary tumor enrichment of the ESR1m defective RSR signature was associated with both worse overall survival and disease specific survival in the METABRIC cohort (Figure 1B, p<0.0001 and Supplemental Figure 2, p=0.00569, respectively). We validated the signature using the SCAN-B ER+ primary breast cancer cohort21 and signature enrichment in this cohort also significantly predicted reduced overall survival (Figure 1C). The ESR1m defective RSR signature was not prognostic for DSS or OS in the METABRIC ER-negative cohort (Supplemental Figure 3 A,B), though therapeutics targeting the RSR and DDR are currently utilized in ER-negative tumors with homologous recombination deficiency, especially BRCA-mutant patients.29 Thus, the ESR1m defective RSR signature was prognostic and identified defective RSR activation as an early pathway driven by ER-signaling significantly associated with worse prognosis in patients with primary ER+ breast cancer.
We next identified the specific DNA repair and replication pathways enriched in the ESR1m defective RSR signature using overrepresentation analysis. The top enriched KEGG pathways were cell cycle followed by DNA replication, mismatch repair, nucleotide excision repair, non-homologous end joining, homologous recombination, and base excision repair pathways (Figure 1D, Supplemental Table 4). Overrepresentation analysis of MSigDB HALLMARK pathways revealed enrichment of DNA repair, G2M checkpoint, and E2F targets (Figure 1E, Supplemental Table 4). Thus, ESR1m defective RSR signature demonstrated significantly up-regulated numerous DNA repair and replication pathways.
Key RSR proteins ATR and Chk1 initiate cell cycle checkpoints to facilitate critical DNA repair, particularly at regions of single-stranded DNA.12 Enhanced phosphorylated and total protein levels of ATR and Chk1 were observed in the Y537S and Y537C ESR1m cell lines (Figure 1F). While activation of the RSR induces cell cycle arrest, which could lead to cellular senescence or apoptosis, defective RSR can enhance tumor proliferation and promote tumorigenesis in cancer models.19 Thus, we next evaluated our ESR1m cell lines for enrichment of a defective RSR phenotype. We utilized the defective RSR signature driven by ATR or Chk1 knockdown as a comparator, as described in Methods section.19 ATR and Chk1 defective RSR signatures were enriched in our ESR1m models using GSEA (Supplemental Table 5). These results suggest that in response to hyperproliferation, ESR1m cells activate a defective RSR to tolerate intrinsic RS.
Disruption of the RSR can significantly impact DNA repair and lead to downstream accumulation of DNA damage eventually resulting in cell death in multiple cancers.12,17 Therefore, we next investigated activation of key DNA repair proteins, replication protein A2 (RPA2) and RAD51 recombinase (RAD51) which are both essential for DNA repair pathways addressing ss30 and ds31 breaks. Since the protein level of these DNA repair proteins can be increased during DNA replication, we also evaluated the cell cycle protein Cyclin A2 which is enriched in S- and G2-M-phase,.32,33 Protein levels of RAD51 were increased 2-fold in Y537S and Y537C cell lines, but RPA2 was only moderately elevated in ESR1m cell lines (Figure 1G). These results suggest that ESR1m cells may exhibit enhanced DNA repair activation at the protein level. Cyclin A2 protein was also elevated in the mutant cells which is consistent with the hyperproliferation phenotype previously reported for ESR1m and could contribute to elevated RAD51 protein levels.7,8,33
We further assessed DNA damage and repair activation in the ESR1m cells using quantification of pH2AX foci and found significantly elevated pH2AX foci in nuclei of both Y537S and Y537C cells. Elevated pH2AX foci are a known marker for both induction of DNA ds-breaks and activation of DNA repair (Figure 1H, Supplemental Figure 4).34 We found significantly higher levels of endogenous DNA damage in mutant cells using alkaline comet assays for the detection of both ss- and ds-DNA breaks (Figure 1I–J, p<0.0001), and neutral comet assays that reliably identify ds-DNA breaks and are insensitive to replication intermediates (Figure 1K–L, p>0.0001).35 Treatment with the DNA intercalating chemotherapeutic agent cisplatin was used as positive controls. These results together demonstrate significant activation of DNA repair pathways driven by defective RSR in mutant cells. Activation of RSR and DDR pathways may be essential to repair the high levels of intrinsic DNA damage present in ESR1m cells to promote tumor survival.

Defective RSR and DDR pathways reveal new therapeutic vulnerabilities in ESR1m breast cancer.
Chk1 and PARP inhibitors demonstrate synergy in cancer models harboring defective RSR and DDR phenotypes.36,37 We identified a significant increase in PARP RNA expression in the Y537S and Y537C cell lines (Supplemental Table 6). PARP1 facilitates DNA repair through multiple pathways13 making it an ideal target with the activated RSR and DNA repair phenotype in ESR1m cells. Therefore, we next investigated the therapeutic potential of Chk1 and PARP inhibitors in combination, and found that Chk1i treatment was synergistic when used with the PARP inhibitor Olap, especially at the higher clinically relevant doses in ESR1m breast cancer,38 resulting in substantially decreased cell viability at Chk1i and Olap concentrations below single agent IC50s in ESR1m cells (Combination treatment shown in Figure 2A–B, single agents in Supplemental Figure 5 A,B). ESR1m cells were significantly less sensitive to single agent Chk1i but demonstrated increased sensitivity to Olap treatment compared to WT cells (Supplemental Figure 5 A, B). Chk1 inhibition was evaluated at the concentrations used by demonstrating a decrease in Chk1 protein levels, and an increase in p-Chk1, both of which are used as pharmacodynamic markers of Chk1 inhibition (Supplemental Figure 6). 39 Inhibition of PARP1 enhances cellular reliance on the RSR pathway to promote DNA damage repair, particularly for ss breaks.28,37 Chk1i have been shown to enhance therapeutic efficacy of PARP inhibitors due to synthetic lethality.37 Our results suggest that inhibition of both Chk1 and PARP in combination blocks compensatory RSR DNA repair activity in mutant cells resulting in enhanced toxicity.
We next evaluated cell cycle and DNA replication dynamics to interrogate cell cycle effects of Chk1 and PARP inhibitors on ESR1m breast cancer. ESR1m cells accumulated in the G2/M phase of the cell cycle after Chk1i treatment both as a single agent and in combination with Olap, indicating potential premature mitotic entry28,40 (Figure 2C). Mutant cells exhibited reduced BrdU uptake after Chk1i treatment alone and in combination (Figure 2D, Supplemental Figure 7, Supplemental Figure 8, Supplemental Table 7). Mutant cells also tolerated higher levels of RS induced with hydroxyurea (HU) treatment resulting in increased DNA replication and increase in DDR protein RAD51 (Figure 2D, Supplemental Figure 9). DNA replication in ESR1m was greatly reduced with Chk1i alone and in combination with Olap, but not in WT cells (Figure 2D), suggesting that both Chk1 and PARP activities are necessary to maintain the high levels of DNA replication induced by RS in ESR1m cells. Disruption of DNA replication and cell cycle progression significantly reduces DNA repair efficacy and induces DNA damage in cells41, identifying a potential mechanism for the synergistic cytotoxicity of Chk1i and Olap in mutant models, and thus a new therapeutic vulnerability.
Since Chk1i and Olap treatments reduced cell viability, cell cycle and DNA replication in ESR1m models, we next evaluated these agents on in vivo tumor growth and metastatic progression using the Y537S tumor xenograft model. Olap treatment reduced levels of multiple ER-regulated proteins including progesterone receptor (PR), trefoil factor 1 (TFF1), and the proto-oncogene c-Myc in addition to PARP42 and RSR proteins ATR and Chk1 in Y537S primary tumor xenografts (Figure 2E). Olap plus Chk1i numerically increased time to primary tumor doubling (Ctrl 5.5 weeks vs. Olap plus Chk1i 8 weeks median time to tumor doubling), but it did not reach statistical significance (Supplemental Figures 10 and 11 A–C). However, Olap treatment as a single agent or in combination with Chk1i significantly reduced the number of distant lung metastases (Figure 2F, p<0.05). These results agree with earlier data demonstrating that the antiestrogen tamoxifen significantly reduced distant metastatic frequency, but did not impact primary tumor growth.8 These results demonstrate that PARP inhibition impacts ER-regulated gene expression with significant effects on metastatic progression to the lung. ESR1m are dominant drivers of distant metastasis.8 Thus, PARP function is a key regulator of both ESR1m gene expression and metastatic dissemination.

Effect of PARP inhibitors on ER-PARP genomic interactions.
The primary function of PARP is PARylation to facilitate protein recruitment for effective DNA repair and replication.13 We next investigated endogenous levels of PARylation in Y537S cells and identified a more than 5-fold increase confirming significantly increased PARP activity in mutant cells (Figure 3A). Olap treatment for 6 hours reduced PARylation levels of PARP1 and PR protein in mutant cells. Reduction in c-Myc levels were seen at later time points (24 hours). In contrast, reduction of these proteins was not sustained in WT ER-expressing cells with longer treatment times. Thus, PARP inhibition exhibited a prolonged and greater reduction of ER-regulated protein expression in mutant cells. While treatment with Olap inhibited PARylation at 300 nM (Supplemental Figure 12), reduction of estrogen-regulated proteins was observed only at higher concentrations (Figure 3A, 5 μM), that have been used to inhibit growth of various cancer models.43
We next evaluated the efficacy of Olap in combination with the SERD Fulv as a novel and clinically relevant therapeutic strategy for ESR1m metastatic breast cancer. Fulv plus Olap treatment significantly reduced PR and ER levels in both mutant models (Figure 3B). Olap treatment as a single agent also decreased PR levels in Y537S but not Y537C cells which co-express WT ER. Olap treatment induced higher levels of phosphorylated and total Chk1 protein in all models (Figure 3B).28,37 We conclude that defective RSR may function as a compensatory pathway in ESR1m cells to promote continuous DNA repair and replication. Fulv blocked Chk1 activation (Figure 3B). Fulv is used as ET for ESR1m metastatic breast cancer in combination with targeted therapeutics such as CDK4/6 and mTOR inhibitors.1,2,29 Thus, reduction of ER-regulated protein levels and Chk1 activation with Fulv plus Olap treatment effectively targets two metastatic growth pathways.8,44 To confirm that combined treatments induced increased DNA damage and also activated DNA repair, we quantified pH2AX foci that are a hallmark of DNA double strand (ds) breaks (Figure 3C, Supplemental Figure 13). Olap increased the number of pH2AX foci/nucleus for both the WT and mutant cells, but only Y537S cells demonstrated a significant increase in pH2AX foci compared to WT with combination treatment. Y537C cells with 50% WT allele frequency showed only a moderate increase. These results are consistent with DDR activation in mutant cells to promote DNA repair with Olap treatment.28,37
We next investigated genomic interactions between PARP1 and mutant ER to elucidate PARP1’s role in ER regulation in mutant cells.45 We quantitated PARP1 and mutant ER co-localization in the nucleus in response to Olap and Fulv as single agents and combination therapy. WT cells exhibited no change in nuclear co-localization with Olap treatment. A reduction in ER-PARP1 foci were observed with Fulv as a single agent (p<0.0001) or in combination with Olap (p=0.0011). This result is likely due to the known effects of Fulv treatment on ER degradation.46 In contrast we observed significant enrichment of ER-PARP1 nuclear co-localization after all treatments in Y537S mutant cells, with the highest increase seen with combination treatment (p<0.0001) (Figures 3D–E with quantitation in Figures 3F–G). Olap as a single agent also enhanced the number of ER-PARP1 foci in Y573C cells (Figure 3H, Supplemental Figure 14), although Fulv significantly reduced co-localization foci likely due to the presence of WT ER in these cells. PARP inhibitors can trap PARP1 on DNA which enhances cytotoxicity in addition to blocking enzymatic PARylation activity.47 Thus, combination treatment increased DNA damage in mutant cells.

ESR1m cells treated with first-line metastatic therapy Fulv plus Abema exhibited defective RSR and DDR and are vulnerable to DDR inhibition.
ESR1m metastatic breast cancer patients receive Fulv plus CDK4/6 inhibitors as first-line therapy.1,2,29 The addition of CDK4/6 inhibitors Abema, ribociclib, or palbociclib to Fulv significantly improves progression-free survival in metastatic breast cancer patients.48–50 Clinical evaluation of Abema in the adjuvant setting significantly increased invasive disease free survival, however adjuvant palbociclib did not significantly improve patient outcomes.51 Thus, a significant proportion of women with high-risk primary or metastatic ER+ breast cancer will be treated with Abema in combination with ET.1,29 The PADA-1 trial identified ESR1m emergence as a significant driver of resistance to aromatase inhibitors (AIs) plus CDK4/6 inhibition.52 Switching patients with emergent ESR1m from AIs to Fulv plus CDK4/6 inhibition significantly improved progression-free survival.52 Patients switched to Fulv at the time of rising ESR1m frequency rather than at time of progression with AIs also exhibited increased efficacy with Fulv.52 These results highlight the importance of monitoring ESR1m emergence for the selection of ET for combination therapy with CDK4/6 inhibitors. The PALOMA-3 and MONARCH-2 trials identified significant acquisition and enrichment of ESR1 mutations in 20–50% of metastatic breast cancer patients after treatment with Fulv plus CDK4/6 inhibition.5,53 Therefore, investigation of the effects of Fulv plus Abema on ESR1m breast cancer is highly clinically relevant to identify effective new therapeutic approaches.
We report for the first time a novel MCF-7 cell line treated long-term with Fulv plus Abema (F+A) that demonstrated an emergent Y537S ESR1 mutation. We used long-term treatment of MCF-7 and ZR75–1 cell lines with Abema (AbemaR) or Fulv (FulvR) as single agent controls in the development of the F+A MCF-7 model (Supplemental Table 8, Supplemental Figures 15 A–F and 16 A–F). The F+A resistant MCF-7 cell line acquired the Y537S ESR1 mutation at 3% allele frequency after 8 months of treatment that was further enriched to 30% allele frequency after an additional 4 months of treatment (Supplemental Figure 17 A–K). The F+A resistant cell line models the emergence and enrichment of ESR1 mutations in metastatic breast cancer patients treated with Fulv plus Abema treatment in the MONARCH-2 trial.5 Thus, this model is most appropriate to investigate therapeutic vulnerabilities in ESR1m breast cancer after disease progression on a CDK4/6 inhibitor. It is important to note that we did not observe the emergence of ESR1m with single agent treatment controls.
We observed significant enrichment of the ATR/Chk1 defective RSR signature using GSEA in the F+A resistant model after 12 months of treatment (NES=1.56, Figure 4A). The F+A resistant and MCF-7 AbemaR cell lines exhibited transcriptional enrichment of CHEK1, CHEK2, and PARP1 using ORA, and transcriptional up-regulation of HALLMARK G2/M checkpoint, E2F targets, and DNA repair (Figures 4B–C). The F+A line exhibited significantly increased endogenous DNA damage concomitant with the emergence of ESR1m (Figures 4D–E). Thus CDK4/6 inhibitor resistance also enriches for a defective RSR phenotype in breast cancer cells.
We next evaluated the cytotoxicity of Chk1i and Olap in our long-term treated cell lines to explore the potential of these inhibitors as targeted therapeutics after resistance to Fulv and Abema. Both F+A and Y537S mutant models were sensitive to ~5 μM Olap (Supplemental Figure 18 A,B), similar with that reported for other cancer cell lines. 43,54 Combination of Chk1 and Olap inhibitors were synergistic at concentrations near the single agent and combined IC50s in F+A mutant cells (Figures 4F–G). Similarly, Fulv plus Olap inhibitor treatments were also synergistic at higher concentrations in mutant cells (Figures 4H–I). Endogenous PARylation levels were elevated and levels of ER and c-Myc proteins were reduced in F+A resistant cells (Figure 4J) similar to that seen in Y537S and Y537C models. Together our results show that tumors resistant to the CDK4/6 inhibitor Abema also exhibit a defective RSR and are sensitive to combination therapies.
We next investigated PARP1 and ER genomic interactions in the F+A cell line using PLA. Increased ER-PARP1 nuclear co-localization was observed after treatment with Olap and Fulv as single agents or in combination (Figures 4K–L, Fulv+Olap p<0.001, Olap p<0.01). Thus, both the CRISPR engineered and acquired ESR1m mutant models exhibit increased mutant ER-PARP1 genomic interactions in response to PARP inhibitor therapy.
Retention of ER on chromatin is crucial for its role as a nuclear transcription factor. To explore whether Olap treatment enhances chromatin retention of ER or PARP1, which can function as an ER co-regulator, we isolated chromatin bound protein fractions and compared PARP and ER levels after Fulv plus Olap treatment (Supplemental Figure 19A) and found higher levels of PARP1 were retained on chromatin in the mutant models. We also used Veliparib (Velip) treatment as a control because of its low trapping ability,47 and found increased levels of PARP1 protein with Olap single treatment compared to Velip. We also detected increased levels of chromatin bound ER protein for both the single agent Olap and Velip treatment in WT and mutant models (Supplemental Figure 19B). These results indicate that Olap traps both PARP1 and ER on DNA, which is associated with the reduction of ER-regulated proteins after Olap treatment in mutant cells. Additional studies examining ER protein half-life will be required to address this possibility. Previous reports identified PARP1 as a transcriptional regulator of WT ER through active enhancer formation and ER PARylation.45,55 Our identification of enhanced endogenous PARylation in the Y537S model and increased ER-PARP1-nuclear interactions after Olap treatment provides novel mechanistic data supporting PARP1’s role as an important transcriptional regulator of mutant ER.

PARP1 inhibition in combination with ET is an effective and novel therapeutic strategy for ESR1m breast cancer.
Fulv significantly improves patient outcomes in ESR1m metastatic breast cancer compared to AIs both alone and in combination with CDK4/6 inhibitors.52,56 Fulv as compared to AIs also significantly extends progression-free survival for ESR1m breast cancer patients.56 Therefore, we next investigated Olap in combination with Fulv as a clinically relevant combination for the treatment of Y537S and Y537C mutants. As single agents both Y537S (90 nM IC50) and Y537C (0.35 nM IC50) were less sensitive to single Fulv treatment compared to WT cells (0.03 nM IC50) (Figures 5A–B).7,8 However in combination, Olap was synergistic with Fulv treatment in both mutant models at physiological (10 nM)57 and supraphysiological (100 nM) Fulv concentrations (Figures 5C–D).
We next investigated the impact of combining Olap with Fulv on ex vivo tumor growth of the WHIM20 Y537S ESR1m patient derived xenograft (PDX) model, and the combination significantly reduced growth of organoids derived from WHIM20 primary tumors and lung metastases and grown ex vivo (Figures 5E–F). However, the organoids were resistant to single agent treatment with Fulv. We evaluated the effects of Fulv plus Olap on WHIM20 in vivo tumor growth and ER-regulated protein expression. The combination greatly reduced PARylation levels, as well as ER and ER-regulated proteins PR and c-Myc in the WHIM20 tumors (Figure 5G), in agreement with earlier data (Figure 3B). Fulv plus Olap treatment also reduced levels of phosphorylated ATR and total Chk1 protein levels, confirming defective RSR in WHIM20 primary tumors (Figure 5G). Both Fulv alone and in combination with Olap significantly reduced primary tumor growth, with the greatest decrease in tumor growth observed in the combination treatment group (Figure 5H, Supplemental Figure 20 A–D). Thus, the combination of Olap plus Fulv was effective in mutant models because it reduced both the DDR and ER signaling resulting in significant reduction in tumor growth. Taken together, our study results strongly support further investigation of PARP inhibitors in clinical trials in combination with Fulv or Elacestrant for the treatment of ESR1m metastatic breast cancer.

Discussion
We discovered enrichment of defective RSR in ESR1m breast cancer. Increased expression of RSR proteins and PARP1 levels in ESR1m models demonstrated significant dysregulation of the RSR and downstream DDR. ESR1m cells exhibited significantly elevated intrinsic DNA damage, and activation of DNA repair due to defective RSR. We determined that enrichment of defective RSR is a novel therapeutic vulnerability in ESR1m breast cancer, and inhibition of Chk1 in combination with Olap reduced mutant tumor growth and significantly reduced distant lung metastasis in tumor models. Enrichment of defective RSR expression in primary patient tumors was prognostic with significant reduction in overall survival in multiple ER-positive patient cohorts. The identification of defective RSR in primary ER+ breast cancer is novel as previous literature reported this phenotype only in homologous-recombination deficient or oncogene-driven cancer models.19,28,36,44,58 Thus, we demonstrate for the first time that metastatic breast cancer patients with ESR1 mutations are an ER-positive subtype without genomic homologous recombination defects that are responsive to RSR and PARP inhibition as single or combination agents.
Aberrant RSR has been reported as a therapeutic vulnerability in cancer models enriched for EMT59 and stem-cell phenotypes.19 We previously reported that ESR1m breast cancer models exhibit enhanced activation of EMT and stem-cell phenotypes.8,10 Both stem-cell and EMT phenotypes promote tumor plasticity, which is a key hallmark of cancer metastasis.60 Herein, we demonstrated a significant reduction in the progression of ESR1m metastatic tumors by targeting the RSR and downstream DDR with a combination of Chk1 and PARP inhibitors. Thus, we provide in vivo evidence that RSR inhibition disrupts the metastatic potential of ESR1m tumors. This is important because there are no current effective therapeutics to block aberrant EMT and stem-cell phenotypes in breast cancer. Therefore, inhibition of the RSR and DDR is an attractive approach to target metastatic driver phenotypes in ESR1m breast cancer. The significance of the identified defective RSR signature in primary breast cancer patients suggests that metastatic driver phenotypes might evolve before the emergence of clinically detectable ESR1m.
We demonstrated synergistic responses to RSR and PARP inhibitors in multiple ESR1m cell line models. Our study is the first to show significant synergy in ESR1m breast cancer models utilizing these therapeutic combinations. We observed cell cycle arrest and significantly reduced levels of actively replicating DNA after treatment with Chk1 inhibition as a single or combination agent with Olap. Our data suggest that inhibition of the RSR and PARP prevents effective DNA replication in addition to disruption of DNA repair. Both Chk1 and PARP are critical players in DNA replication by facilitating fork progression. PARP1 regulates fork reversal61, and the RSR promotes fork stalling62–64 to ensure a high level of DNA fidelity. These pathways are complementary, ensuring that the cell cycle is delayed so that aberrant DNA replication can be corrected.64,65 Accumulation of ESR1m cells in the G2/M phase of the cell cycle suggests these inhibitors force premature mitotic entry which may further enhance DNA damage and promote cell death.28 Disruption of DNA replication through PARP or RSR inhibition promotes significant genomic instability and cell death in a number of cancer models.17,66,67 Thus, future investigation of the impact of Chk1 and PARP inhibitors on replication fork dynamics might provide mechanistic insights into the efficacy of RSR and DDR inhibitors in ESR1m breast cancer.
This study is the first to identify enrichment of PARP1-mutant ER DNA co-localization with PARP inhibition. PARP inhibitors block PARylation and trap PARP protein on the DNA to enhance cytotoxicity.47 Our PARP trapping experiment revealed that PARP1 and mutant ER interact at the DNA level. Reduced expression of both ER and ER-regulated proteins in ESR1m cell lines and tumors after PARP inhibition identified PARP as a critical regulator of mutant ER-driven gene transcription, consistent with an earlier report by Gadadet al.45 We previously published that ESR1m cells display significantly enhanced recruitment of PARP1 to co-activator protein-DNA complexes.68 PARP1 functions as a promoter of active enhancer formation in cells with WT ER.45 Thus, protein recruitment and regulation of chromatin structure are two potential mechanisms of PARP’s genomic actions on mutant ER. PARylation also promotes resistance to the ER antagonist tamoxifen by enhancing ER-regulated gene expression.55 Our results for the first time demonstrate ER-PARP interactions in a Fulv-resistant model. We identified enhanced endogenous PARylation in our ESR1m cell lines with intrinsic (Y537S, Y537C models) or acquired (F+A cells) Fulv resistance. Addition of Olap to Fulv was synergistic in mutant models, and significantly increased viability. We also identified a significant reduction in ESR1m primary tumor growth with the combination of Fulv plus Olap. Taken together, these results identify PARP1 as a critical co-regulatory protein for ESR1m-driven gene transcription, with PARP activity acting as a promoter of resistance to ET. Future investigation of the ESR1m genomic landscape, including chromatin accessibility and identification of gene loci with PARP1-mutant ER co-localization, will provide mechanistic insight into PARP1’s regulatory role in ESR1m transcriptional activity.
Herein, we provide compelling evidence for the inclusion of RSR or DDR inhibitors, particularly PARP inhibitors, in the treatment of ESR1m breast cancer. Addition of Olap to Fulv significantly reduced tumor growth of both primary and metastatic mutant ESR1 tumors and decreased ER-regulated protein expression in multiple mutant models. When to introduce PARP inhibitors in the treatment of ESR1m metastatic breast cancer patients remains a critical question. Fulv plus Olap treatment was synergistic in the CDK4/6 resistant Y537S mutant cell line, indicating that PARP inhibition is a viable therapeutic option for ESR1m patients with previous exposure to CDK4/6 inhibitors. The PADA-1 trial recently highlighted the importance of monitoring emergent ESR1ms for the selection of second-line therapeutics.52 Utilizing a novel “crossover” design, patients with rising baseline ESR1 mutations switched from AIs to Fulv and continued CDK4/6 inhibition. The Fulv crossover cohort exhibited significantly increased progression-free survival compared to the AI cohort. Patients switched to Fulv after detection of ESR1ms also remained on Fulv without disease progression longer than patients switched to Fulv at the time of progression with AIs. Thus, selection of ET for targeted therapy combinations is a critical challenge for patients with ESR1m. The approval of the oral SERD elacestrant for ESR1m metastatic breast cancer6 and promising preliminary clinical results of the oral SERDs camizestrant69 and imlunestrant70 provide multiple opportunities to introduce new targeted therapeutic combinations, such as PARP inhibitors, for the treatment of ER+ metastatic breast cancer. Therefore, utilizing the PADA-1 trial design, we propose that emergence of ESR1ms in MBC patients could be a biomarker to “crossover” to PARP inhibitors in combination with the new oral SERDs or Fulv.

Supplementary Material
12345678910111213141516171819202122232425262728