The Selective Estrogen Receptor Degrader ZN-c5 Has Broad Antitumor Activity in Wild-Type and Mutant ER-Positive Breast Cancer Models.

Introduction
Breast cancer is the second leading cause of cancer death in women after lung cancer (1) and the incidence rates continue to increase. An estimate of 313,510 new cases of breast cancer and 42,780 cancer-related deaths in 2024 according to The American Cancer Society has been predicted (1). A majority of breast cancers diagnosed are estrogen receptor–positive (ER+) and rely on the activity of this receptor for tumor growth and survival. Therefore, ER has been an attractive target and many therapeutic strategies have been developed to antagonize ER signaling for treatment of patients with ER-dependent breast cancers (2).
Endocrine therapies, including aromatase inhibitors (AI) that block estrogen production and selective ER modulators (SERM; e.g., tamoxifen) that directly inhibit receptor activity, have been the standard of care for patients with ER+ breast cancers. Although these strategies can be effective, many patients develop resistance, which ultimately leads to disease progression that continues to rely on ER signaling (3). The acquired resistance remains a critical clinical problem, and effective therapeutic strategies are greatly needed (4). The mechanisms of resistance include somatic alterations, epigenetic changes, activation of bypass pathways, and changes in the tumor microenvironment (5). Mutations of ESR1, the gene encoding ERα, have been reported in 20% to 40% of patients who progress on AI therapy and have been associated with endocrine therapy resistance. The most frequent mutations are D538G, Y537S, and Y537N (6–8). Circulating tumor DNA analysis from patients with ER+ breast cancer in the BOLERO-2 trial suggested that the Y537S and D538G mutations are associated with shorter overall survival (OS) and more aggressive diseases (9). The Y537S and D538G mutations occur in the ERα ligand-binding domain (LBD) and compromise affinities for not only its natural ligand estradiol but also the antagonists, such as 4-hydroxytamoxifen (4-OH tamoxifen). Furthermore, the ER mutants demonstrate constitutive activity that is independent of its ligands and seem to have higher transactivation activity than the wild-type (WT) receptor (7, 8, 10).
Breast cancers that are resistant to one type of endocrine therapy may still respond to other ER-targeting therapies (11). In efforts to find novel therapies that are more efficacious to treat ER+/HER2− breast cancer and to overcome endocrine resistance, selective estrogen receptor degraders (SERD) have been developed. In the clinic, patients who progressed after AI or SERM treatment typically receive fulvestrant, the first approved SERD. Fulvestrant competitively binds to ERα, antagonizes ERα function, and induces ERα degradation, which further limits the availability of ER for downstream signaling (12, 13). Fulvestrant has demonstrated superiority over AIs as monotherapy in hormone treatment–naïve patients in the FIRST and FALCON trials (14, 15). In patients who previously received hormone therapies, fulvestrant combined with CDK4/6 inhibitors further improved progression-free survival (PFS) compared with fulvestrant alone in several clinical trials (16–18). Although it is clinically beneficial, fulvestrant has poor pharmacokinetic (PK) properties, which limit its exposure and oral bioavailability. This led to the research and development of oral SERDs with improved drug profiles and other oral ER antagonists, such as ER proteolysis-targeting chimeras (PROTAC; ref. 19), which bind the intracellular E3 ligase cereblon and the ligand-binding domain of ER to direct the degradation of ER.
Many early-generational novel SERDs have been developed and advanced to clinical trials, including GDC-0810 and GDC-0927 from Genentech, AZD9496 from AstraZeneca, LSZ102 from Novartis, and G1T48 from G1 Therapeutics. Although these SERDs showed decent anti-ER activity in preclinical studies, the effect did not translate well to clinical efficacy (20–23). Safety and efficacy of a new generation of SERDs with improved potency, such as giredestrant (GDC-9545) from Genentech, camizestrant (AZD9833) from AstraZeneca, and palazestrant (OP-1250) from Olema, are currently being evaluated in the clinic (24–29). Although demonstrating clinical benefit, adverse events (AE), such as bradycardia, visual disturbance, and neutropenia, have been observed in some studies and this needs to be properly managed, especially for combination therapy and long-term treatment (30–32). Vepdegestrant (ARV-471), from Arvinas, an oral PROTAC targeting ER, has entered clinical trial and showed evidence of activity based on clinical benefit rate. However, the benefits of vepdegestrant were only seen in patients with ESR1-mutant tumors as disclosed recently (NCT05654623; ref. 33). Recently, two oral SERDs, elacestrant from Menarini and imlunestrant from Loxo/Eli Lilly and Company, were granted FDA approval for patients with ER+/HER2−, ESR1-mutated advanced or metastatic breast cancer who had progressed on prior lines of endocrine therapy.
Here, we describe the identification and characterization of ZN-c5, a novel small molecule with potent antagonism and degradative properties against the ER both in vitro and in vivo. ZN-c5 showed a high oral bioavailability across several preclinical species as compared with other SERDs. ZN-c5 induced robust tumor growth inhibition (TGI) as a single agent in ER+ breast cancer xenograft models. The combination of ZN-c5 with cell cycle inhibitors such as CDK4/6 or PI3K inhibitors resulted in enhanced antitumor activity. More importantly, ZN-c5 shows improved antitumor activity over fulvestrant or elacestrant in human tumor xenograft models. We believe that the high exposure of ZN-c5 coupled with its potency and degradative properties could therapeutically benefit patients with ER+ breast cancer. ZN-c5 has been evaluated in clinical trials as a single agent and in combination studies, and initial clinical data demonstrated excellent safety and tolerability and preliminary signs of activity. The PK profile of ZN-c5 in patients with breast cancer indicates that ZN-c5 has greater than fivefold exposure than fulvestrant. Early evidence from clinical trials has confirmed the exceptional exposure and target engagement of ZN-c5. These data suggest that ZN-c5 has the potential to benefit patients with ER+ breast cancer in the clinic.

Materials and Methods

Cell culture and reagents
The MCF-7 cell line was purchased from ATCC (HTB-22, RRID: CVCL_0031; sex: female) in 2018 and maintained in DMEM/F12 (Gibco, 11330-032), supplemented with nonessential amino acids (NEAA; Gibco, 11140-050), Na-pyruvate (Gibco, 11360-070), and FBS (HyClone, SH30084). CAMA-1 (HTB-21, RRID: CVCL_1115; sex: female) was purchased in 2020, T47D (HTB-133, RRID: CVCL_0553; sex: female) was purchased in 2017, ZR-75-1 (CRL-1500, RRID: CVCL_0588; sex: female) and HCC1428 (CRL-2327, RRID: CVCL_125; sex: female) were purchased in 2018, and HCC1500 (CRL-2329, RRID: CVCL_1254; sex: female) was purchased in 2020 from ATCC and cultured according to ATCC recommendation in RPMI-1640 complete medium. MCF-7/luciferase cells (AKR-234, RRID: CVCL_5J36; sex: female) were purchased from Cell Biolabs, Inc. in 2019. CRISPR–knocked-in ERY537S mutants were generated with Synthego in 2020. Cells were cultured in DMEM (high glucose) supplemented with 10% FBS, 0.1 mmol/L minimum essential medium NEAA, 2 mmol/L L-glutamine, and 1% penicillin–streptomycin. Cell culture medium and supplements, unless otherwise indicated, were purchased from either ATCC or Gibco. The chemistry design strategies and synthesis of ZN-c5 have been described previously (34). 4-OH tamoxifen, elacestrant, palbociclib, ribociclib, abemaciclib, and alpelisib were purchased from Selleckchem. Fulvestrant and 17β-estradiol were purchased from Sigma-Aldrich. Antibodies for ERα were purchased from Abcam (ab16660, RRID: AB_443420). Progesterone receptor (PR; #8757, RRID: AB_2797144) and GAPDH (#2118, RRID: AB_561053) were purchased from Cell Signaling Technology, and anti–β-actin (A2228, RRID: AB_476697) was obtained from Sigma-Aldrich. All cell lines were used within 20 passages since collection (date of purchase). All cell lines were confirmed to be Mycoplasma negative by PCR and cell identity was authenticated by the vendor by short tandem repeat profiling at the time of purchase. No further tests were performed except MCF-7 cells which were tested for Mycoplasma by PCR and cell identity was confirmed by short tandem repeat profiling at IDEXX, with the most recent test run on April 26, 2024.

Cell proliferation assay
For SERD assay, MCF-7 cells, MCF-7/luciferase, or CRISPR–knocked-in ERY537S and CAMA-1 cells were seeded at a density of 3,000 cells per well into flat clear-bottom tissue cultured–treated 96-well plates (Corning) in assay medium. Assay medium components included phenol red–free DMEM/F12 (HyClone, SH30272.01) 500 mL for MCF-7 cells and phenol red–free DMEM for MCF-7/luciferase or CRISPR–knocked-in ERY537S cells, NEAA (Gibco, 11140-050) 5 mL, Na-pyruvate (Gibco, 11360-070) 5 mL, restripped charcoal-stripped FBS (Gemini, 100-119) 45 mL, and (8%; charcoal stripped) FBS. The charcoal-stripped FBS was re-stripped as follows: 10 mg/mL activated charcoal Norit SA 2 and dextran 70 (1 mg/mL) were added to FBS and incubated at 56°C for 45 minutes with shaking. The solution was then centrifuged at 4°C at ∼3,500 rpm for 30 minutes. The supernatant was filtered through a 0.22-μm filter. T47D, ZR-75-1, HCC1428, and HCC1500 cells were seeded at 10,000 cells per well in 96-well plates in assay medium consisting of phenol red–free RPMI-1640 medium (Thermo Fisher Scientific, cat. #11835030) plus 8% restripped charcoal-stripped FBS (Gemini, 100-119).
Cell viability was assessed after 6-day incubation with different compounds as indicated in the presence of 0.1 nmol/L 17β-estradiol (Sigma) using CellTiter-Glo (Promega) according to the manufacturer's protocol, and relative luminescence units (RLU) were measured using an Envision Multilabel Reader (PerkinElmer). The RLUs of the treated samples were normalized to those of the untreated samples, and cell viability was expressed as a percentage of the value of the untreated cells.

Western blot assay
For SERD treatment, cells were seeded in phenol red–free media supplemented with charcoal-stripped FBS. Forty-eight hours later, cells were treated with SERDs at the indicated concentrations. For other drug treatment, cells were seeded in regular culture medium. Cells were lysed in RIPA buffer protease and phosphoprotease inhibitors (Thermo Fisher Scientific) after 24 or 48 hours of treatment, and total proteins were separated by SDS-PAGE and immunoblotted using antibodies as indicated. For pharmacodynamic (PD) studies using tumor samples from xenograft studies, flash-frozen tumors were pulverized into powder, and then were lysed in RIPA buffer. Total protein was analyzed by Western blotting.

Rat uterine wet weight assay
From Vital River Laboratory Animal Technology Co., Ltd, 21-day-old female Sprague–Dawley CD-IGS rats with body weight 30 to 50 g were purchased. Animals were quarantined for 5 days before being randomized into groups (n = 5) and treated once a day for 3 days with compounds of interest at indicated doses. Body weights of all animals were recorded before dosing during the study. Animals were euthanized 24 hours after the last dose and uteri were dissected, weighed, and fixed in 10% neutral buffered formalin for histologic examination. Formalin-fixed uteri samples were processed for paraffin embedding, sectioned at approximately 5 μm, and stained with hematoxylin and eosin. Endometrial layer thickness was measured using whole-slide images obtained from digital scans at 20× magnification. Uterine wet weight (UWW) index was calculated as follows: (UWW/BWD4)/vehicle mean UWW/BWD4, in which BWD4 is body weight at day 4. For each mouse, digital measurements were taken from three separate sections. Results are displayed as the mean endometrial layer thickness ±SD.

RNA isolation and qPCR
RNA was extracted from cells treated with DMSO or indicated compounds using the RNeasy Kit (Qiagen) per the manufacturer’s instructions, quantified using NanoDrop 8000 (Thermo Fisher Scientific), and reverse-transcribed with the cDNA Archive Kit (Applied Biosystems). PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) was used to quantify PGR, GREB, TIFF, and the housekeeping gene GAPDH. The relative quantities were determined using ΔΔ threshold cycle (ΔΔCt).

RNA sequencing
MCF-7 cells were seeded in phenol red–free medium supplemented with 10% charcoal-stripped FBS. Forty-eight hours later, the cells were treated with ZN-c5 or 4-OH tamoxifen at 1 µmol/L for 24 hours. RNA was extracted using the methods described above. The Poly (A) tail RNA selection strand-specific cDNA library was prepared by LC Sciences, and sequencing was performed on the Illumina NovaSeq 6000 sequencer, providing paired-end 50 base pair reads and around 20 million reads per sample. The aggregated gene counts were used for differential gene expression analyses using StringTie by calculating fragments per kilobase per million reads.

ER activity score
ER activity score was determined as described before (35). Specifically, the gene list was first generated by including genes that were significantly induced or suppressed by E2 relative to DMSO across cell lines (MCF7, HCC1500, CAMA-1, MDA-MB-330, BT-474, EFM-19, T-47D, and MDA-MB-134-VI) by RNA sequencing (RNA-seq). This list was further narrowed down by leveraging public RNA-seq data from 939 breast tumors (726 ER+ and 213 ER− by IHC) from The Cancer Genome Atlas (TCGA). Genes that (i) exhibited detectable expression in ER+ TCGA tumors and (ii) that the E2-induced genes be more highly expressed in ER+ compared with ER− TCGA tumors (one-sided t test P < 0.05) were used as the core sets, including 21 E2-induced and 17 E2-repressed genes. The ER activity score was applied by obtaining ligand-induced fold change in expression of each gene, followed by averaging these fold changes across the genes. Once the E2-induced score and E2-repressed score had been independently determined, the E2-repressed score was subtracted from the E2-induced score to generate a single score for ER pathway activity.

Osteoporosis mouse model
This study was approved and conducted in compliance with the Institutional Animal Care and Use Committee (IACUC) of WuXi AppTec. Female C57BL/6 mice, aged 10 weeks, were ovariectomized (OVX) bilaterally to induce osteoporosis. Progesterone in plasma was tested at the third week after surgery to confirm the successful induction of osteoporosis. After confirmation, mice were randomized to six groups and treated with vehicle or SERD compounds at indicated doses. The bone mineral density (BMD) was measured by the dual-emission X-ray absorptiometry (DXA) method weekly from the sixth week of drug administration. At the end of the study (12 weeks), micro-CT of femur and tibia of animals in vitro was performed. Micro-CT is an analysis method with higher resolution than ordinary CT (DXA) and its cone beam can obtain three-dimensional (3D) bone tissue scanning image. Different from bone densitometer in vivo, BMD here refers to the weight of bone tissue per cubic centimeter (g/cm3). Micro-CT also conducts nondestructive 3D imaging of bone trabecula to display the microstructure of bone trabecula.

In vivo efficacy studies
All in vivo efficacy studies were approved by the IACUC of each vendor, including Pharmaron, ChemPartner, Envigo, and WuXi AppTec. For MCF-7 and ZR-75-1 models, mice were inoculated subcutaneously on the second right mammary fat pad with the single-cell suspension of 95% viable tumor cells (1.5 × 107) in 200 µL medium Matrigel mixture (1:1 ratio) without serum for the tumor development. The treatments were started when mean tumor size reached ∼200 mm3. Mice were randomized into 10 groups. In addition, estradiol benzoate injection was delivered subcutaneously (40 µg/20 µL, twice weekly). For MCF-7 ERY537S model, 3 days prior to the tumor cell implantation, estrogen pellet (17β-estradiol pellet 90 day release, 0.18 mg/pellet) was subcutaneously implanted in the lateral side of the neck by microsurgery under ketamine/xylazine anesthetization. Each animal was inoculated into the right mammary fat pad with a single-cell suspension of >95% viable 1 × 105 tumor cells in 10 µL of serum-free DMEM. For T47D and HCC1428 models, each mouse was inoculated subcutaneously at the right flank with tumor cells (1 × 107) in 0.2 mL mixture of PBS and Matrigel (PBS:Matrigel = 1:1) for tumor development, which has been subcutaneously implanted with 17β-estradiol tablets (0.18 mg, 90-day release) 2 days before cell implantation. For WHIM20 patient-derived xenograft (PDX) study, female athymic nude mice were implanted with a single-cell suspension of 1.5 × 106 WHIM20 cells, passage 10. The cells were mixed 1:1 with PBS:Matrigel (Corning, ref 354234, lot #0090009) in a total volume of 100 μL/mouse. Once tumors reached a volume of approximately 100 to 300 mm3, mice were randomized by tumor volume into 1 of 10 treatment groups (eight mice/group) using Biopticon TumorManager software.
Mice were sacrificed after the final dose and tumors were excised, cut into approximately 30 mg fragments, and flash-frozen for RNA and protein PD analysis. Additional tumor fragments for IHC were placed in 10% neutral buffered formalin for 24 hours and transferred to 70% ethanol until processing. Sections (5 μm) were labeled for ERα (clone SP1, 790–4325, Ventana Roche) and stained with the Horseradish Peroxidase/3,3′-Diaminobenzidine Detection Kit (Abcam, #64261).

IHC staining
For patient sample IHC staining, the formalin-fixed samples were paraffin-embedded, and ERα and PR IHC assays were conducted at Discovery Life Sciences. ERα mAb 6F11 (ACA093C) and PR mAb clone 16 (ACA424C) were purchased from Biocare Medical. ER and PR expression were evaluated in the nucleus of tumor cells using a semiquantitative approach that includes percentages at differential intensities to determine H-scores. The percentage of tumor cells with biomarker staining was recorded at a corresponding differential intensity on a four-point semiquantitative scale (0, 1+, 2+, and 3+). On this scale: 0 or <1+ = null, negative, or nonspecific staining; 1+ = low or weak staining; 2+ = medium or moderate staining; and 3+ = high or strong staining. H-score was calculated by using a formula H-score = (1 × percent of cells at 1+) + (2 × percent of cells at 2+) + (3 × percent of cells at 3+), resulting in a score range from 0 to 300. For tumor samples from the WHIM20 xenograft, tumors were collected and fixed in formalin for 24 hours and transferred to 70% ethanol until use. Formalin-fixed, paraffin-embedded tumor tissues were sectioned at 5 µm. ER mAb 6F11 was used. H-score was calculated similarly as described above.

Clinical trials
ZN-c5-001 (NCT03560531) and ZN-c5-002 clinical trials (NCT04176757) were conducted in accordance with U.S. ethical guidelines (i.e., U.S. Common Rule) and the Declaration of Helsinki, Good Clinical Practice, and all federal, state regulatory guidelines and approved by Western Institutional Review Board (IRB) and NYSDOH IRB. In the ZN-c5-001 study, ZN-c5 was administered once daily or twice daily with dosage ranging from 25 to 300 mg per day in 28-day cycles to patients enrolled under the protocol. Plasma samples were collected at cycle 1, day 1 (single dose) or cycle 1, day 15 (steady state) for PK analysis (MicroConstant). In the ZN-c5-002 study, ZN-c5 was given once daily or twice daily with dosage ranging from 25 to 300 mg per day for 28 days. Pretreatment and posttreatment tumor tissues were collected from patients and were used for ER and PR IHC assays at Discovery Life Science or for PK analysis at MicroConstant. All patients included in the clinical trial provided written informed consent prior to study enrollment. Study representativeness information is provided as Supplementary Table S1.

Statistical analysis
Unless stated otherwise, all data were statistically analyzed using GraphPad Prism software (version 10). Multiple comparisons were assessed by one-way ANOVA and grouped data were analyzed by two-way ANOVA. Data were presented as mean ± SEM or violin plots. All data were analyzed using GraphPad Prism 10.0. Results were considered statistically significant if P ≤ 0.05 (* or #), P ≤ 0.01 (** or ##), P ≤ 0.001 (*** or ###).

Results

ZN-c5 selectively targets ER to inhibit breast cancer cell proliferation in vitro and in vivo
To identify a potent nonsteroid SERD with improved bioavailability and efficacy, we used in vitro and in vivo screening, employing the MCF-7 breast cancer cell model to evaluate antitumor activity, ER degradation, and PK. ZN-c5 was identified as the best compound for further development, with chemistry structure as shown in Fig. 1A (34). ZN-c5 specifically binds to ERα with high affinity (IC50 of 0.4 nmol/L) and showed no agonist and antagonist activity against other nuclear steroid receptors, such as androgen receptor (AR), PR, mineralocorticoid receptor, and glucocorticoid receptor (Supplementary Table S2). To determine whether ZN-c5 induces ERα degradation, ERα protein level was evaluated in MCF-7 cells treated with ZN-c5. Treatment with increasing concentrations of ZN-c5 resulted in 96% degradation of ER protein level, generating DC50 (half maximal degradation concentration) of 0.3 nmol/L, comparable with fulvestrant (Fig. 1B; Supplementary Fig. S1A and S1B). In addition, ZN-c5, like fulvestrant, degraded ER protein regardless of the presence of 17b-estradiol (E2). Moreover, ZN-c5 treatment was able to decrease the protein level of PR, an ER target protein that is usually measured as a marker of ER activation, in a dose-dependent manner (Fig. 1C). As a result of decreased ER signaling, ZN-c5 treatment inhibited E2-mediated proliferation in MCF-7 cells with an IC50 of 0.4 nmol/L, which is comparable with fulvestrant and elacestrant with IC50 of 0.2 and 0.3 nmol/L, respectively (Fig. 1D). These data demonstrate that ZN-c5 can selectively target ER to induce its degradation and inhibit ER-dependent cancer cell proliferation.

In vivo antitumor activity of ZN-c5 was evaluated in the MCF-7 orthotopic xenograft model established from this human breast cancer cell line. The treatment of MCF-7 xenografts with ZN-c5 led to tumor regression, i.e., TGI >100%, at 10 mg/kg (Fig. 1E; Supplementary Fig. S1C). Importantly, ZN-c5 showed better efficacy than fulvestrant at 3 mg/week dose (200 mg/kg/week in mice), which is about eightfold higher dose than the clinically achievable, which is reported to be 25 mg/kg/week (36). ZN-c5 was well tolerated on this schedule as indicated by minimum body weight loss (Supplementary Fig. S1D and S1E). To confirm that the TGI correlated with ER signaling downregulation, PK and PD relationship was analyzed in MCF-7 xenograft tumors treated with ZN-c5 at 10 mg/kg for 7 days. ZN-c5 concentrations in tumors were measured (Fig. 1F) to confirm tumor penetration. As expected, there was a significant reduction of PR expression that is maintained up to 48 hours following the last ZN-5 treatment (Fig. 1G). These results demonstrate that ZN-c5 could induce sustained inhibition of ER signaling and thus lead to TGI effectively. PK analysis at the end of efficacy (steady state) showed that ZN-c5 demonstrated high plasma and tumor exposure, with an AUC from time zero to last measurable concentration (AUClast) of 9.9 and 8.3 µg × hour/mL, respectively, when dosed at 5 mg/kg, and 21.3 and 19.5 µg × hour/mL, respectively, when dosed at 10 mg/kg (Table 1), suggesting that ZN-c5 efficiently penetrated tumor and inhibited tumor growth through antagonizing the ER pathway.

ZN-c5 transcriptionally inhibits ER pathway and has minor agonist activity in endometrial tissue
To further investigate the mechanism of ZN-c5 effect on the ER pathway, transcription of ER-regulated genes was analyzed after ZN-c5 treatment in MCF-7 cells. As shown in Fig. 2A, the most common ER target genes, PGR, TFF1, and GREB1 were examined by qPCR. As expected, E2 stimulated the transcription of the above genes whereas treatment with ZN-c5, fulvestrant, and tamoxifen downregulated the expression of the genes. We further analyzed a core set of ER-regulated genes (35) by RNA-seq. A subset of these genes is presented in Fig. 2B, including the ones upregulated by E2 (35) but are reversed by 4-OH tamoxifen and those that are similarly upregulated by E2 and 4-OH tamoxifen. Unlike 4-OH tamoxifen which has an agonistic effect on the expression of some genes listed here, ZN-c5 treatment largely resulted in downregulation of gene expression, opposite to those observed for E2 treatment (Fig. 2B; Supplementary Tables S3, S4). The ER activity score was further analyzed as previously described (35) and confirmed that ZN-c5 has no agonist activity (Fig. 2C).
We then tested whether ZN-c5 could act as a partial agonist in other tissues using a well-established rat UWW assay. Immature rats were treated with ZN-c5 once daily at indicated dose for 3 days, and uterine weight was measured and compared with fulvestrant, AZD9496, tamoxifen, and E2. Consistent with published data, both E2 and tamoxifen significantly increase UWW. In contrast, ZN-c5 and AZD9496 did not affect uterine weight significantly whereas fulvestrant seemed to decrease uterine weight. In addition, all three SERDs antagonized E2-mediated uterine stimulation (Fig. 2D). Histologic staining of uterine tissues indicated that epithelia height was slightly increased by ZN-c5 and AZD9496, but not by fulvestrant, compared with the vehicle control group (Fig. 2E). Tamoxifen and E2, however, dramatically increased epithelia height, consistent with previous observation that tamoxifen acts as a partial agonist in certain tissues (Fig. 2E; ref. 37). No dose-dependent agonist effect was observed for ZN-c5 (Supplementary Fig. S2A and S2B). These data revealed that ZN-c5 had slight agonism in endometrial tissue but to a significantly lower extent than tamoxifen.

ZN-c5 treatment protects bone loss in OVX mice
One of the most common AEs of fulvestrant is bone pain, which is caused by the degradation of ERs (38). On the other hand, SERMs such as tamoxifen have been shown to prevent bone loss because of their estrogen agonist–like effect in bone tissue (39), highlighting the importance of estrogen signaling in bone homeostasis regulation. We wanted to assess whether ZN-c5 has any bone-specific effect by monitoring BMD change after ZN-c5 treatment in OVX young mice. The success of ovariectomy was confirmed by decreased progesterone level compared with the sham control (Supplementary Figure S3A). Following ovariectomy, mice were treated with vehicle, ZN-c5, fulvestrant, or tamoxifen for 12 weeks and BMD of femur and tibia was monitored weekly. As expected, ovariectomy led to lower BMD compared with sham control, indicating bone loss. Treatment with tamoxifen significantly increased BMD overtime, reaching a level that exceeded the sham control, consistent with previous studies of SERMs in animals and humans (39, 40). In contrast, fulvestrant, as a pure estrogen antagonist, further decreased BMD compared with the OVX-vehicle group. ZN-c5 treatment at both dose levels prevented bone loss caused by ovariectomy as indicated by increased BMD over the whole treatment period (Supplementary Fig. S3B and S3C). At the end of the treatment, ZN-c5 groups showed a higher level of BMD than the sham group but not as dramatic as that in the tamoxifen group (Fig. 3A). Micro-CT analysis of the distal femur and tibia indicated that ovariectomy induced significant changes in bone microarchitecture (Fig. 3B; Supplementary Fig. S3D) and the corresponding parameters, including decreased bone density/volume and bone cortex density. Consistently, ZN-c5 treatment prevented such changes whereas fulvestrant further amplified the effect of ovariectomy (Fig. 3B–D; Supplementary Fig. S3E and S3F). UWW was also examined at the end of the study. Similar to the earlier observations, ZN-c5 showed no agonist effect and tamoxifen significantly increased uterine weight index (Supplementary Fig. S3G).

ZN-c5 demonstrates potent antitumor activity as a single agent and in combination with CDK4/6 and PI3K inhibitors in multiple breast cancer models
The studies described above established ZN-c5 as a potent inhibitor in MCF-7 cells in vitro and in vivo. To further assess the antitumor activity of ZN-c5, we evaluated the activity of ZN-c5 in five additional ER+ breast cancer cell lines, CAMA-1, T47D, ZR-75-1, HCC1428, and HC1500. ER and PR expression were detectable in these cells (Fig. 4A). We first assessed the ability of ZN-c5 to modulate the levels of ER, as shown in Fig. 4B, and ZN-c5 treatment led to more ER degradation than elacestrant in MCF-7 cells. The ER degradation profile of ZN-c5 was comparable with that of AZD9496 and elacestrant in the other five cell lines (Fig. 4B; Supplementary Fig. S4A).
It is well established in clinical practice that the combination of endocrine therapies and CDK4/6 inhibitors is an effective treatment strategy for patients with ER+/HER2− metastatic breast cancers as this combination substantially improves PFS and OS compared with monotherapy (41–43). Mechanistically, ER signaling regulates cyclin D1 expression and thus the cyclin D1/CDK4/6 function, and synergy between CDK4/6 inhibitors and anti-estrogen therapies is expected and has been observed in the clinic. We showed that treatment with palbociclib, a CDK4/6 inhibitor, increased cyclin D1 levels but had no effect on ER expression (Supplementary Fig. S4B). Both ZN-c5 and fulvestrant treatment decreased cyclin D1 expression and phosphorylation of RB1, a downstream target of CDK4/6 kinases (Supplementary Fig. S4B). The combination of ZN-c5 or fulvestrant with palbociclib eliminated the elevated cyclin D1 by palbociclib and further decreased RB1 phosphorylation. The signaling change eventually led to more profound cell-cycle arrest at G1 phase and decreased S and G2–M phases (Supplementary Fig. S4C and S4D). This synergistic effect was further evaluated in the MCF-7 tumor model. As previously observed, ZN-c5 alone demonstrated dose-dependent TGI, and combination with palbociclib resulted in enhanced tumor regression at both doses tested (Fig. 4C). Similarly, ZN-c5 combined with ribociclib and abemaciclib, two other approved CDK4/6 inhibitors, also induced significant tumor regression (Fig. 4D). We then extended the combination treatment to other breast cancer models, including T47D, ZR-75-1, and HCC1428. Combination treatment of ZN-c5 with palbociclib resulted in tumor regression in all models (Fig. 4E–G). Despite its superior activity as a SERD in multiple cell lines in vitro, fulvestrant only generated partial tumor inhibition in xenograft models, likely due to its poor exposure. Increased efficacy was also observed in the combination of palbociclib and fulvestrant but was much less pronounced than that of palbociclib and ZN-c5 (Fig. 4C and E).
Resistance to endocrine therapy can be driven by activation of growth factor receptors pathways, such as PI3K/AKT signaling. Moreover, about 40% of ER+ tumors carry a PIK3CA (PI3K subunit p110α) mutation, constitutively activating the PI3K pathway. Based on these observations, alpelisib, a PIK3α inhibitor, was tested in the clinic and approved in combination with fulvestrant in patients with ER+/HER2−, PIK3CA-mutated, advanced breast cancer. Here, we examined the impact of combination of ZN-c5 with alpelisib on TGI. To this end, the MCF-7 tumor model, which harbors E545K mutation in the PI3K p110 subunit α (44), was used for this study. Combination treatment of ZN-c5 and alpelisib produced a greater TGI compared with each single agent and led to tumor regression (Fig. 4H). This result, in line with the other efficacy data generated, indicates that ZN-c5 is efficacious in multiple ER+ breast cancers and supports the application of ZN-c5 in a combination regimen with currently approved therapeutics in the clinic.

ZN-c5 effectively antagonized ESR1 mutants in vitro and in vivo

ESR1 gene point mutations are highly prevalent in metastatic breast cancers after AI therapy but relatively rare in primary disease (45). Although cancer cells that have ER mutants are resistant to endocrine treatment, they continue to rely on ERα signaling and thus respond to ER-targeting therapies. In order to test ZN-c5 activity against ER mutants, we generated ERY537S CRISPR–knocked-in MCF-7/luciferase cells because the Y537S mutation has been known to confer more resistance to therapies than other mutations (12). Both heterozygous and homozygous ERY537S mutation cells were obtained and validated by Sanger sequencing. ERα and PR expression were examined, and the results showed that although ERα level was lower in ERY537S cells, PR expression remained high at both protein and mRNA levels, especially in the heterozygous cells compared with WT cells (Fig. 5A; Supplementary Fig. S5A). ERY537S cells grew faster than WT cells in hormone-deprived medium although it seemed to be the opposite in regular medium, suggesting less dependence on estradiol in the mutant cells (Fig. 5B). Greater increase of ER downstream targets relative to DMSO control in WT cells than mutant cells in response to E2 treatment (log2 fold change 4 vs. 0.5) was also observed (Supplementary Fig. S5B), indicating that estradiol had more effect in WT cells than the ESR1-mutant cells. The absence of estradiol did not affect the proliferation rate of ERY537S cells, whereas WT cells stopped growing without estradiol (Fig. 5B). These results agree with reported data that ERα mutants lost the dependence on ligands and showed higher transcriptional activity than WT ERα (7, 46). We wanted to determine whether there was a sustained reliance of ER or PR for ERY537S cells. Knockdown of ERα inhibited the proliferation of both WT and heterozygous cells but failed to deliver the same effect on homozygous cells. PR knockdown, on the other hand, was not able to affect any cells significantly (Fig. 5C). Similarly, both ZN-c5 and fulvestrant treatment reduced ERα levels and ER downstream target GREB1 expression in WT and heterozygous ERY537S cells but spared homozygous cells (Fig. 5D; Supplementary Fig. S5B). Note that ZN-c5 treatment resulted in more efficient ERα degradation in parental cells than that in heterozygous ERY537S cells (73% vs. 45%; Fig. 5D). The ERα reduction could be blocked by a proteasome inhibitor MG132 treatment, indicating that this was proteasome-dependent degradation [Fig. 5D (bottom)].
Because ESR1 mutations reported in tumors are typically heterozygous (47), we focused on heterozygous ERY537S cells in the following studies. ZN-c5 was able to effectively antagonize the proliferation of heterozygous cells with a slightly higher IC50 than observed in WT cells (10.7 vs. 1.8 nmol/L). The heterozygous cells seemed less responsive to elacestrant and more resistant to tamoxifen treatment while retaining sensitivity to fulvestrant, CDK4/6 inhibitors, and alpelisib (Supplementary Table S5). In addition, ZN-c5 at low concentrations dramatically decreased PR levels while showing limited effect on ER degradation in heterozygous cells, highlighting its function as an antagonist in addition to an ER degrader (Supplementary Fig. S5C).

In vivo activity of ZN-c5 in the ERY537S context was first evaluated in immune-deficient mice bearing the ERY537S heterozygous cells. ZN-c5 treatment resulted in a dose-dependent inhibition on tumor growth, with a TGI of 44% and 77% at 5 mg/kg and 20 mg/kg, respectively. Strikingly, a higher dose of ZN-c5 (30 mg/kg) led to tumor regression (23%) in this model, whereas fulvestrant at its clinically relevant dose did not show any efficacy (Fig. 5E). Next, ZN-c5 efficacy was tested in a PDX model which has been shown to recapitulate the characteristics of human disease more closely. The WHIM20 model that is heterozygous for ERY537S mutation was chosen. ZN-c5 at 30 mg/kg, as a single agent, significantly suppressed tumor growth, and combination with palbociclib resulted in tumor regression of 28% (Fig. 5F). Although ZN-c5 at the low dose of 10 mg/kg generated moderate efficacy as a single agent, combination with palbociclib completely inhibited tumor growth, with a TGI close to 100%. Fulvestrant as a single agent showed minor antitumor activity, and combination with palbociclib did not improve the efficacy (Fig. 5F). PD analysis showed that ZN-c5 treatment substantially decreased protein expression (Supplementary Fig. S5D). Moreover, in a separate WHIM20 efficacy study, ERα degradation in tumors was evaluated after ZN-c5 treatment at higher doses of 40 and 70 mg/kg. No further reduction of ERα in the high-dose group was observed (Supplementary Fig. S5E), indicating that ZN-c5 may have reached the target engagement and efficacy plateau around 40 mg/kg. Taken together, these data further demonstrate that ZN-c5, as an oral SERD, can inhibit ER signaling in vivo resulting in strong antitumor efficacy in the ESR1-mutant context. Additional benefit can be achieved in combination with CDK4/6 inhibitors as shown by producing regressions in the PDX model.

ZN-c5 demonstrated exceptional exposure and target engagement in patients with ER+/HER2− breast cancer
To further understand the properties of ZN-c5 as a novel SERD, we first analyzed the PK data obtained from phase I/II clinical trial of ZN-c5 monotherapy in patients with advanced ER+/HER2− breast cancer (NCT03560531). Following ZN-c5 once daily or twice daily oral doses of 50 mg up to 300 mg, single dose (cycle 1, day 1) or steady state (cycle 1, day 15) plasma PK data of ZN-c5 are shown in Fig. 6A and B and Table 2. ZN-c5 was rapidly absorbed and reached maximum concentration (Cmax) within 1 to 6 hours. There was no accumulation as similar exposure was observed on day 1 and day 15. In general, both Cmax and area under the ZN-c5 concentration–time curve from time 0 to 24 hours after dose (AUC0-24) were dose dependent but less than dose proportional. ZN-c5 exposure at 50 mg reached an average AUC of 65.7 µg × hour/mL, which is higher than the efficacious exposure (>90% TGI) in the MCF-7 mouse model (ZN-c5 at 10 mg/kg; Table 1). Tumor concentration around 2 hours after treatment (Table 2), however, was lower than tumor Cmax (2.07 µg/g) observed in MCF-7 at the efficacious dose of 10 mg/kg (Table 1). To reach the equivalent Cmax seen at the efficacious dose, ZN-c5 needed to be administered at 150 mg or 300 mg daily.
In a window-of-opportunity study assessing the PD activity of ZN-c5 (NCT04176757), ZN-c5 was administered at doses ranging from 25 mg up to 300 mg, as monotherapy for 21 days in subjects with ER+/HER2− breast cancer. A total of 33 paired biopsies before and after 21 days of ZN-c5 treatment were obtained, and ER degradation was evaluated by IHC. ER H-scores decreased after treatment in 26 of 33 (78.8%) samples and were observed across all doses tested (Fig. 6C), with the average relative change of −16% and absolute change of −46 (Supplementary Table S6). PR H-scores also decreased after treatment in PR-positive tumors (excluding PR cutoff ≤10%; ref. 48), with average relative change and absolute change of −5% and −109, respectively. Notably, PD modulation, especially PR levels, was variable and dose independent, with absolute changes ranging from −245 to 110, indicating that certain tumors may respond better than others. Overall, these data, together with the preclinical findings, substantiate ZN-c5 as a novel SERD with exceptional exposure and decent target engagement in certain patients with ER+/HER2− breast cancers who may benefit from it.

Discussion
In efforts to understand the mechanisms in endocrine-resistant breast cancers, researchers have learned that ER+ tumors continue to rely on ER for tumor growth even in the setting of endocrine resistance. This raises the possibility that an alternative strategy to target ER signaling might still be effective in these patients (11), thus driving the search for novel and more effective ER-targeting agents. This initially led to the development of fulvestrant, which works as a pure ER antagonist or SERD without any known agonist effect. Fulvestrant has shown activity in treating breast cancers that have progressed on initial endocrine treatment. However, fulvestrant’s limitations include low oral bioavailability and large-volume intramuscular injection. These limitations lead to low exposure and suboptimal ER degradation, while also reducing patient compliance. Therefore, an orally bioavailable SERD that is safe and effective against ER+ breast cancer after progression on earlier therapies is clearly an unmet medical need. Inspired by the advantage of an oral SERD approach, we developed ZN-c5, focusing on improving potency, as well as overcoming limitations of fulvestrant. We showed that ZN-c5 could directly bind to ERα and has picomolar potency in degrading ER and inhibiting ER+ cell proliferation in vitro, similar to fulvestrant and another SERD elacestrant (Fig. 1B and D). The effect of ZN-c5 on ER signaling was further confirmed by checking mRNA levels of ER downstream targets (Fig. 2A and B). Although fulvestrant demonstrated slightly better potency in all in vitro assays (Figs. 1–3), it has limited efficacy in in vivo tumor models because of its low bioavailability. In contrast, ZN-c5 demonstrated dose-dependent antitumor activity in all models, including the ESR1-mutant PDX model. PK/PD analysis revealed that ZN-c5 had high plasma and tumor exposure, inducing sustained ER signaling inhibition in tumors (Fig. 1G), resulting in TGI.
Unlike tamoxifen, ZN-c5 showed no agonist effect in RNA-seq analysis of the ER pathway in the MCF-7 cell line and generated mild agonism in endometrial tissue (Fig. 2B and D). On the other hand, fulvestrant tended to decrease the uterine weight and epithelia height but the clinical significance is yet to be defined. This low level of agonism is not considered as a concern for the development of ZN-c5 given the fact that agonism is modest relative to tamoxifen and long-term use of tamoxifen administration in the clinic only shows small absolute increases in mortality (49, 50). Furthermore, a similar small but significant change in rodent uterine weight was noted for elacestrant (51). Bone is known to be another target tissue of estrogen signaling and suppressed estrogen levels in postmenopausal women are associated with bone loss and increased risk of fracture (52, 53). Drugs that block estrogen signaling may also affect bone homeostasis, highlighted by the observation of bone pain in fulvestrant-treated patients and bone protection in tamoxifen-treated patients. The effect of ZN-c5 on bone was also evaluated and our data suggest that ZN-c5 has a tamoxifen-like effect by preventing bone loss in OVX mice. The mild agonism activity of ZN-c5 on bone may generate additional clinical benefit to patients with postmenopausal breast cancer who oftentimes have decreased bone mass. It also provides a potential advantage of ZN-c5 over fulvestrant or AIs when used in the adjuvant setting for patients with breast cancer for whom treatment tends to be long term.
Although the mechanisms underlying the bone-protective effects of certain SERDs are not fully understood, several contributing factors have been proposed. Structural features of SERDs, such as the presence of a cinnamic acid side chain, as seen in GDC-0810, AZD9496, and so in ZN-c5 may influence tissue-specific activity, including partial agonism in bone (21, 54, 55). Additionally, differential expression of coactivators and corepressors in various tissues can modulate SERD activity (56). Bone tissue likely expresses a distinct set of these cofactors compared with breast tissue, contributing to the observed agonism in bone and antagonism in tumors. Fulvestrant, in contrast, has a distinct chemical structure and functions as a pure ER antagonist. It binds to ER with high affinity, impairs dimerization, promotes receptor degradation, disrupts nuclear localization, and consistently downregulates ER expression in tumors without exhibiting agonist activity in bone or other tissues (57). A more comprehensive investigation integrating both biological and chemical approaches is warranted to fully elucidate how ZN-c5 selectively antagonizes ER in breast cancer cells while preserving ER function in bone.
Breast cancer is known as a highly heterogeneous disease which has been a key challenge for the development of therapeutics. Heterogeneity is found between different patients (intertumor heterogeneity) and even within a single tumor (intratumor heterogeneity; ref. 58). Phenotypes can change during tumor progression and heterogeneity is often observed between primary and different metastatic lesions (58, 59). This heterogeneity is a major challenge for targeted therapies which oftentimes limit therapeutic efficacy and has been related to primary or acquired resistance (60). The complexities of estrogen signaling probably add another hurdle on the way to achieve expected outcomes. In classical pathway, ligand-activated ER dimers bind to ER elements and interact with coactivators and corepressors to regulate target gene transcription (61, 62). A diverse cofactor expression profile can be found in different tissues or even among breast cancer cells; therefore, agents that downregulate the ER pathway, such as SERDs and SERMs, may not result in the same inhibitory effect in all cancer cells (63). Certain coactivator expression may also contribute to primary resistance to anti-ER treatment (64). We addressed this complexity by evaluating the activity of ZN-c5 in a variety of breast cancer cell lines and in vivo models which potentially represent different ER+ tumor subpopulations. As shown in Fig. 3, ZN-c5 was able to degrade ER across the six cell lines tested, with a comparable potency with AZD9496 and elacestrant but less than fulvestrant. As expected, ZN-c5 demonstrated strong in vivo antitumor activity across several models, with significantly improved efficacy compared with fulvestrant, likely due to the superior bioavailability. As fulvestrant has shown its advantage over AIs and SERMs in the clinic, we believe that the high exposure of ZN-c5 coupled with its potency and degradative properties could provide therapeutic benefit to patients with ER+ breast cancer. Another layer of complexity of ER signaling is the cross-talk with a variety of other signaling pathways, such as cell-cycle regulation pathways, PI3K/AKT, ERK/MAPK, and PLC/PKC. This is also known as indirect nongenomic action of estrogen (65), a common cause for acquired endocrine resistance. Treatment regimens combining SERD with other pathways have been developed for patients with breast cancer. The combination of fulvestrant or AI with three CDK4/6 inhibitors, palbociclib, ribociclib, and abemaciclib, have all been approved. PIK3CA mutations are frequently found in ER+ breast cancer and are usually associated with poor prognosis and endocrine resistance (66). Alpelisib, a PI3Kα inhibitor, in combination with fulvestrant was recently approved for patients with ER+/HER2−, PIK3CA-mutated, advanced or metastatic breast cancer (67). Our data show that the combination of ZN-c5 with all three CDK4/6 inhibitors or with alpelisib demonstrated enhance antitumor activity compared with either monotherapy (Fig. 3), which suggests that ZN-c5 potentially has better efficacy in combination therapies in the clinic.

ESR1 mutations in LBD are known to drive endocrine resistance and the detection of specific ESR1 mutations in the clinic can help to guide treatment choice. Among the most common mutations, Y537S was reported to be more resistant to therapies that inhibit estrogen signaling (68). In cell proliferation assay, compared with the activity in WT-ER cells, ZN-c5 and fulvestrant maintained the ability to inhibit ERY537S cell growth, with only a minimum increase in IC50. In contrast, both elacestrant and tamoxifen were less active in this mutant cell line (Fig. 4E). The in vitro activity of ZN-c5, but not fulvestrant, translated well into in vivo efficacy as shown in Fig. 4F. In a PDX model, WHIM20, which harbors ERY537S, ZN-c5 also demonstrated significant tumor growth inhibition as a single agent and was able to induce tumor regression when co-administrated with palbociclib. There are other mutations in ER LBD that may be less resistant to endocrine therapies. We have previously tested ZN-c5 in a PDX model that harbors ERL536P mutation and found that ZN-c5 completely inhibits the tumor growth (69). This is important for antiestrogen agents that are intended to overcome the acquired resistance to AI therapy. We noticed that ZN-c5 did not degrade the mutant ER as efficiently as WT-ER. Therefore, it will be very meaningful to monitor ESR1 mutations in future clinical studies to further evaluate the activity of ZN-c5.
ZN-c5 has been evaluated as monotherapy and in combination with palbociclib in postmenopausal and premenopausal women with advanced ER+ breast cancer (Study NCT03560531). Results from a dose-escalation and -expansion study with single-agent ZN-c5 demonstrated no dose-limiting toxicities, with the most common treatment emergent AEs (TEAE) including hot flashes, nausea, and fatigue; grade 3 TEAEs included abdominal pain, hypertension, hyponatremia, pain in extremities, and γ-glutamyl transferase increases. Preliminary efficacy analysis in this population showed an overall response rate of 5% and a clinical benefit rate of 38%, which were slightly lower than observed for other SERDs that are in clinical investigation (70). This result is consistent with the modest PD modulation in patient tumor samples (Fig. 6; Supplementary Table S6). The disconnect between the robust antitumor activity observed in preclinical models and the limited clinical efficacy may be attributed to greater tumor heterogeneity in patients. It is also possible that patients with specific molecular alterations respond more favorably to ZN-c5, underscoring the need to investigate predictive biomarkers for these sensitive tumors. Another factor could be a relatively lower tumor concentration of ZN-c5 in patient tumors despite very high plasma exposure as shown in Table 2. Although increasing the dose may enhance tumor drug levels, careful evaluation is needed to balance efficacy and toxicity. Previous studies suggest that slight structural modifications can improve drug exposure and selectivity in tissues (e.g., tumor, fat pad, and bone), which further affects their success in clinical development (71, 72). Additionally, optimizing the drug formulation could enhance solubility and tumor delivery. These considerations warrant further investigation to enhance the therapeutic potential of ZN-c5.

Supplementary Material
Supplementary Figure S1Supplementary Figure S1 shows ZN-c5 is a novel SERD that inhibits growth of ER + tumors in vitro and in vivo A. ERα degradation curve after ZN-c5 treatment and DC50 (concentration at 50% degradation) determination for Figure 1B top; B. Top: western blot analysis of ERα degradation at 24 hours after fulvestrant treatment in MCF-7 cells at indicated concentration; Bottom: ERα degradation curve and DC50 determination for fulvestrant; C. Tumor growth curve of MCF-7 xenograft model. Mice bearing MCF-7 tumor cells were dosed orally once a day for 26 days. D&E, body weight changes after ZN-c5 or fulvestrant treatment.

Supplementary Figure S2Supplementary Figure S2 shows ZN-c5 demonstrates minor agonist effect on uterine tissue A. Uterine wet weight change in juvenile rats treated with different doses of ZN-c5. B. H&E staining on uterine tissues after ZN-c5 treatment. Images were digitally scanned at 20X magnification

Supplementary Figure S3Supplementary Figure S3 demonstrates ZN-c5 protects bone loss in ovariectomized mice A. Progesterone level in serum was measured after Ovariectomy (OVX) or sham surgery to confirm the success of ovariectomy. B and C. BMD in femur and tibia were monitored by InAlyzer system overtime after drug treatment. D. Micro-CT analysis of tibia microarchitecture after at the end of study (12 weeks of drug treatment). E & F. BMD measured by micro-CT in femur and tibia. G. The ratio of uterine wet weight to body weight post-mortem. Statistical significance was evaluated by 2-Way ANOVA, *P < 0.05 compared with sham vehicle. #P < 0.05, ##P < 0.01, compared with OVX-vehicle. Sham, surgery control; OVX, Bilateral ovariectomy

Supplementary Figure S4Supplementary Figure S4 shows Combination of ZN-c5 and palbociclib further arrests the cells at G1 phase A. Relative ER level to DMSO after normalized to β-actin control for Figure 4B. B. MCF-7 cells were treated with 500nM of palbociclib for indicated time, cyclin D1 and p-Rb expression were examined by western blot. C. MCF-7 cells were treated with 100nM ZN-c5 or fulvestrant, 500nM palbociclib, or combination of palbociclib with ZN-c5 or fulvestrant, for 24 hours. Western blot analysis of cyclin D1, pRb, ERα and PR were performed. D. cell cycle analysis of MCF-7 cells treated with ZN-c5 or palbociclib for 48 hours.

Supplementary Figure S5Supplementary Figure S5 shows ZN-c5 activity in ERY537S tumors. A. Key ER downstream targets expression by qPCR in WT or ERY537S heterozygous cells. B. ER downstream target GREB expression by qPCR in WT or ERY537S heterozygous cells after treatment with 0.1nM E2, 100nM ZN-c5 or 100nM fulvestrant. C. Western blot analysis of ER and PR protein after different concentrations of ZN-c5 treatment for the indicated time in MCF7 ERY537S heterozygous cells. ER or PR protein change (%) relative to non-treatment control after correction by GAPDH. D. WHIM20 tumor bearing mice were dosed with ZN-c5 or fulvestrant at PR protein level in WHIM20 tumor samples harvested at the end of an efficacy study. ZN-c5 was dosed at 40mg/kg, fulvestrant was dosed at 5mg/dose. E. ER expression in WHIM20 tumors by immunohistochemistry (IHC) at the end of efficacy study

Supplementary Table S1Supplementary Table S1 shows the Representativeness of Study Participants

Supplementary Table S2Supplementary Table S2 shows ZN-c5 human nuclear receptor profiling and estrogen receptor binding affinity

Supplementary Table S3Supplementary Table S3 shows The effect of ZN-c5 and tamoxifen on E2-induced genes

Supplementary Table S4Supplementary Table S4 shows The effect of ZN-c5 and tamoxifen on E2-suppressed genes

Supplementary Table S5Supplementary Table S5 shows Cell proliferation IC50 and maximum inhibition (Emax) in parental and ERY537S (heterozygous) after different ER ligands treatment.

Supplementary Table S6Supplementary Table S6 shows ER and PR IHC staining average H-score