BET Bromodomain inhibition reverses CDK4/6 inhibitor resistance in estrogen receptor-positive breast cancer via induction of miR-34a-5p

Abstract

Purpose:

CDK6 overexpression is one critical determinant of acquired CDK4/6 inhibitor resistance. Because BRD4 is recruited to the CDK6 promoter, we investigated the potential of bromodomain and extra-terminal domain (BET) inhibition to reverse CDK4/6 inhibitor resistance.

Experimental Design:

Cell viability and survival assays and cell line xenografts were utilized to evaluate BET inhibition in palbociclib-resistant breast cancer cells. Vehicle- and BET inhibitor-treated cells were subjected to RNA sequencing. CDK6 promoter activity was assessed with luciferase assays, and the miRPathDB V2.0 database was used to identify potential miRNAs mediating the effects of BET inhibition. Experiments were conducted to determine whether continued palbociclib treatment is essential for BET inhibitor efficacy and to explore associated mechanisms.

Results:

In CDK4/6 inhibitor-resistant models overexpressing CDK6, a cell cycle gene signature was differentially downregulated following BET inhibition. The BET inhibitors JQ1 and ZEN-3694 reduced the expression of CDK6 and cyclin D1, reinstated CDK4/6 inhibitor-induced cell cycle arrest, and triggered apoptosis in vitro, as well as tumor regression in vivo. Mechanistically, BET inhibition downregulated CDK6 expression through the induction of miR-34a-5p, rather than by directly repressing the CDK6 promoter. Introduction of a miR-34a-5p inhibitor abrogated BET inhibitor-mediated molecular changes, whereas a miR-34a-5p mimic replicated the effects of BET inhibition. Lastly, resistant cells exhibited downregulation of BCL-2 in the presence of continued palbociclib, associated with reduced ER⍺ expression, facilitating sensitivity to BET inhibition.

Conclusions:

Our findings highlight BET inhibition or application of miR-34a-5p mimics as promising strategies to reverse CDK4/6 inhibitor resistance in a subset of ER+ breast cancers.

Translational Relevance

Resistance to CDK4/6 inhibitors is a significant clinical challenge for patients with ER+ breast cancer. Our previous work identified CDK6 overexpression as a driver of acquired resistance. The bromodomain and extra-terminal domain (BET) protein BRD4 has been shown to bind to the CDK6 promoter, and BET inhibition reduces both CDK6 mRNA and protein levels. Here, we explored the potential of BET inhibition to overcome CDK4/6 inhibitor resistance. Our findings demonstrate that BET inhibition effectively reverses CDK4/6 inhibitor resistance primarily by suppressing CDK6 expression through induction of miR-34a-5p. These results suggest BET inhibitors or miR-34a-5p mimics as therapeutic candidates for a subset of CDK4/6 inhibitor-resistant ER+ breast cancers. Furthermore, since miR-34a-5p targets components of multiple oncogenic pathways, including MYC and BCL2, these strategies may have broader applications in ER+ breast cancers with diverse mechanisms of acquired CDK4/6 inhibitor resistance, other breast cancer subtypes, or in cancers of other tissue origins.

Introduction

The advent of selective inhibitors of CDK4/6, critical regulators of cell cycle progression, has transformed the treatment of estrogen receptor-positive (ER+) breast cancer. The addition of these agents, including palbociclib, ribociclib, or abemaciclib, to standard hormonal treatment has significantly improved progression-free survival, and in some cases, overall survival (1). Nonetheless, intrinsic resistance occurs, and acquired resistance is universal. Mechanisms of resistance are diverse and include alterations in upstream oncogenic pathways such as FGFR amplification, ERBB2 mutation, and MAP kinase or AKT activation (2,3). Alternatively, other alterations that induce CDK4/6-independent cell cycle progression may occur, including RB1 loss (2,4), CDK2 activation via cyclin E overexpression or p27^Kip1 loss, and AURKA amplification (2,5,6). Identifying novel strategies to overcome these resistance mechanisms addresses a critical unmet need to maximize the clinical benefit afforded by CDK4/6 inhibition in ER+ breast cancer.

Overexpression of the target kinase CDK6 has also been shown to confer either intrinsic or acquired resistance in both model systems and in primary tumors, which may occur through multiple mechanisms. For example, Loss of FAT1, PTEN and ARID1A have been linked to elevated YAP/TAZ transcription factor activity and overexpression of CDK6 as determinants of intrinsic resistance (7,8), whereas CDK6 amplification has been described in cell line models with acquired CDK4/6 inhibitor resistance (9). We recently demonstrated that overexpression of CDK6 in CDK4/6 inhibitor-resistant ER+ breast cancer cells is dependent on increased levels of the microRNA (miRNA) miR-432-5p that targets SMAD4 expression to suppress transforming growth factor β (TGF-β) pathway signaling (10). Furthermore, analysis of biopsy samples displaying intrinsic or acquired resistance from ER+ breast cancer patients treated with a CDK4/6 inhibitor, showed higher levels of miR-432-5p and CDK6 mRNA and lower levels of SMAD4 mRNA than biopsies from patients with tumors considered sensitive to CDK4/6 inhibition (10). Importantly, a prolonged drug holiday, with removal of palbociclib from these miR-432-5p-driven, CDK4/6 inhibitor-resistant cells, resulted in reduced CDK6 expression and reversal of resistance, suggesting that strategies reducing CDK6 expression may restore sensitivity in a subset of CDK4/6 inhibitor-resistant breast cancer.

Recently, the bromodomain and extra-terminal domain (BET) family member BRD4 was shown to be recruited to the CDK6 gene promoter (11) and treatment with the BET inhibitor JQ1 (12) reduced CDK6 expression in small cell lung cancer cell lines, where the degree of reduction of CDK6 correlated with JQ1 sensitivity (13). Additionally, studies of BET inhibitor resistance have pointed to synergism between BET and CDK4/6 inhibition. In NUT carcinoma, a rare malignancy driven by a BRD4-NUT or similar fusion oncoprotein, upregulation of the D-cyclin-RB1 axis protected cells from BET inhibitor-induced cell cycle arrest, with tumor proliferation controlled both in vitro and in vivo following addition of a CDK4/6 inhibitor (14). Similarly, resistance to BET inhibition in triple-negative breast cancer cell lines has been shown to emerge rapidly, but with synergistic apoptosis induced by the addition of CDK4/6 inhibition, particularly in BET inhibitor-resistant derivatives (15,16).

Here, we report that BET inhibition reverses CDK4/6 inhibitor resistance in CDK6-overexpressing breast cancer cells in vitro and in vivo by reducing CDK6 protein levels. Surprisingly, BET inhibition suppresses expression of CDK6 through induction of a second miRNA, miR-34a-5p, rather than by reducing CDK6 promoter activity. In CDK4/6 inhibitor-resistant cells, maximal effects of BET inhibition require continued CDK4/6 inhibitor exposure, contributing to suppressed expression of pro-survival proteins. Our results support the vulnerability of CDK4/6 inhibitor-resistant cells to BET inhibition or miR-34a-5p mimics and provide evidence of synergism between BET and CDK4/6 inhibitors in CDK4/6 inhibitor-resistant cancer models.

Materials and Methods

Cell lines

ER+ breast cancer cells, MCF7 (RRID: CVCL_0031), T47D (RRID:CVCL_0553), and ZR-75-1 (RRID:CVCL_0588) were purchased from American Type Culture Collection (ATCC). Cells were routinely confirmed negative for mycoplasma contamination using a PCR detection Kit (ABM, Cat. No. G238). All cells were maintained in a humidified incubator with 95% air plus 5% CO2 at 37°C and were cultured in Dulbecco’s modified Eagles medium (MCF7, T47D) or RPMI 1640 medium (ZR-75-1) supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 1% penicillin:streptomycin solution (10,000 units/mL of penicillin and 10,000 μg/mL of streptomycin, Gibco) and 0.01 mg/ml human recombinant insulin.

Palbociclib-resistant cells (MCF7-R100, T47D-R100, and ZR-75-1-R100) were generated by continuous exposure to 100 nM palbociclib as previously described (10). The MCF7-CDK6 stable cell line was established from MCF7 cells depleted of CDK6 by CRISPR-knockout and re-introduced exogenous V5-tagged-CDK6 as previously described (10). MCF7-CDK6-R100 cells are palbociclib-resistant MCF7-CDK6 cells maintained in media with 100 nM palbociclib.

Human bone marrow CD34 + stem/progenitor cells (HSPCs) from two individual donors (#1 and #2) were purchased from StemCell Technologies (Cat# 70002.2) and were cultured in StemSpan-XF medium (StemCell Technologies, Cat# 100-0073) supplemented with StemSpan CD34+ Expansion Supplement (10X) (StemCell Technologies, Cat# 02691). HSPCs were maintained in a humidified incubator with 95% air plus 5% CO2 at 37°C.

Reagents

Compounds, oligonucleotides, recombinant DNA constructs, and transfection and RNA extraction reagents are described in the Supplementary Materials.

Western blot analysis

Cells were lysed with RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS, Boston BioProducts) supplemented with protease and phosphatase inhibitor cocktails (Calbiochem). Lysates were incubated on ice for 15 mins, vortexed every 5 mins and then centrifuged at 15,000 g for 15 min. Protein concentration was measured using a Pierce-BCA assay kit. Proteins were separated by gel electrophoresis and transferred to PVDF membranes. Membranes were blocked with 5% milk, immunoblotted with protein-specific primary antibodies overnight at 4°C on a shaker, and then washed with TBST and probed with horseradish peroxidase (HRP) conjugated anti-mouse or anti-rabbit secondary antibodies (GE Healthcare) for 1 hour at room temperature. Signals were detected by chemiluminescence, intensity of which was analyzed to reveal relative expression levels of specific proteins. Antibodies used are listed in the Supplementary Materials.

RNA-Seq

RNA was extracted from vehicle or JQ1-treated cells using the RNeasy Mini Kit (Qiagen) and treated with DNase using RNase-Free DNase Set (Qiagen). rRNA depletion was performed from 100ng of purified RNA using QIAseq FastSelect rRNA HMR reagents according to manufacturer’s protocol. Libraries were prepared using Roche Kapa Biosystems RNA HyperPrep sample preparation reagents on a Beckman Coulter Biomek i7. Finished dsDNA libraries were quantified by Qubit fluorometer and Agilent TapeStation 4200. Uniquely dual indexed libraries were pooled in an equimolar ratio and shallowly sequenced on an Illumina MiSeq to further evaluate library quality and pool balance. The final pool was sequenced with paired-end 150bp reads on an Illumina NovaSeq 6000at the Dana-Farber Cancer Institute Molecular Biology Core Facility. Sequenced reads were aligned to the UCSC hg38 reference genome assembly and gene counts were quantified using STAR (v2.7.3a; RRID: SCR_004463 (17). Differential gene expression testing was performed by DESeq2 (v1.22.1; RRID:SCR_000154) (18). RNAseq analysis was performed using the VIPER snakemake pipeline (19).

qPCR

mRNA was extracted from cultured cells using the PureLink RNA Mini Kit (Life Technologies of Thermo Fisher Scientific). cDNA was generated by using the iScript cDNA synthesis kit from BIO-RAD. miRNA was extracted from cultured cells using MIRVANA(TM) MIRNA Isolation kit (Thermo Fisher Scientific). For miRNA-based qPCR, reverse transcription was performed using the miRCURY LNA RT Kit (Qiagen). SYBR Green super-mix for running qPCR was from Applied Biosystems. qPCR was performed using primers for human CDK4, CDK6, CCND1, MYC, BIRC5, ACTB, miR-34a-5p, miR-34a-3p, miR-432-5p and U6. Primer sequences are listed in the Supplementary Materials. SYBR green signal was detected using the QuantStudio^™ 3 Real-Time PCR System from Applied Biosystems. Relative expression of each gene was analyzed using the 2-ΔΔCT method (20). Data are reported from 3 biological replicates, each in duplicate, unless otherwise indicated.

Flow cytometry

For cell cycle analysis, cells were harvested by trypsinization and washed with PBS. Cell pellets were resuspended and fixed with 80% ethanol (−20°C) for 30 minutes on ice, washed with PBS, and stained with propidium iodide/RNase solution (BD Biosciences) for 15-30 min. Cells were filtered and cell cycle analysis was performed using a BD Fortessa cytometer with FACS Diva software (BD Biosciences). For Edu Labeling, cells were pulsed with 10 μM EdU for 30 min prior to harvest. Trypsinized cells were washed with PBS, fixed in 4% formaldehyde in PBS for 15 min at room temperature, washed with 1% BSA in PBS and permeabilized in 70% ethanol at −20°C for 30 min to overnight. After additional washes with 1% BSA in PBS, incorporated EdU was labeled with a CLICK reaction (2 mM CuSO4, 100 μM THPTA, 100 mM sodium ascorbate, and 2 μM Calfluor 647 Azide in PBS) by rotating for 30 min at room temperature. Samples were then washed with 1% BSA in PBS and stained with a 1 μg/ml DAPI/100 ng/ml RNase A staining solution. Samples were analyzed on Fortessa analyzers (BD Biosciences) and 50,000 total events were captured for each sample. Annexin V apoptosis assays were performed following the manufacturer’s instructions (BD Biosciences).

Growth inhibition assays

For short-term growth inhibition assays, MCF7-R100 or T47D-R100 cells (1,000 cells/well) were seeded in 96-well plates and incubated for one day prior to DMSO, JQ1, or ZEN-3694 treatment for 5 days. Cell viability was quantified using CellTiter-Glo reagent from Promega following the manufacturer’s protocol. Relative cell viability percentage was calculated by first subtracting background (CellTiter-Glo reagent in medium) from the drug-treated wells and then normalizing to the DMSO-treated control. All data were derived from 3 biological replicates, each in triplicate. BLISS Synergy/antagonism analysis was performed using Combenefit software (21).

Colony formation assays

For long-term colony formation assays, 3,000 cells per well were seeded in 6-well plates and treated with DMSO, JQ1, ZEN-3694 and/or palbociclib for 14 days. Colonies were fixed using fixation solution (75% methanol + 25% Acetic acid) for 20 minutes, stained with 1% Crystal Violet in water for 15 minutes, washed in tap water, and quantified using ImageJ software (RRID:SCR_003070). All colony formation assays were performed on 3 biological replicates, with representative images of the plates shown.

Luciferase assays

For luciferase assays, the CDK6 promoter (1,000 bp upstream of the transcription start site) driven luciferase reporter construct (CDK6-p1000-Luc) (22) was introduced into MCF-R100 cells via transfection. Cells were then split into 96-well plates and incubated overnight before adding DMSO or JQ1 of 100nM or 300nM for 24 hours. Assays were performed using Luciferase Assay System with Reporter Lysis Buffer following the manufacturer’s instructions (Promega).

Animal studies

Female nude mice (NCRNU-F) (NCr-Foxn1^NU, Taconic Laboratories), approximately 10 weeks old, were maintained and handled in accordance with local guidelines; therapeutic interventions were approved by the Animal Care and Use Committee of Dana-Farber Cancer Institute. The generation of palbociclib-resistant breast tumors from injected palbociclib-resistant MCF7 cells was as described previously (10). Briefly, 3 days prior to cell implantation, mice were implanted with 17β-estradiol pellets (0.36 mg, 90-day release, Innovative Research of America). 7 x 10⁶ MCF7-R100 cells in 100 μl DMEM medium + 50% matrigel (BD Biosciences) were subcutaneously implanted into the mammary fat pads of mice. Palbociclib treatment was initiated 5 days after cell implantation. Palbociclib exposure (dissolved in 10mM sodium lactate buffer, pH 4) was maintained at 100 mg/kg via daily oral gavage throughout the study. JQ1 was dissolved in 10% DMSO plus 90% HP-β-CD solution (10% hydroxypropyl-β-cyclodextrin in saline) and was dosed at 40 mg/kg daily i.p. Mice were randomized to receive treatment with vehicle or JQ1 when tumors became palpable (115-135 mm³).

Statistical analyses

For comparison of two sets of measurements, unpaired two-tail Student’s t tests were performed. At least three observations were recorded in each group, unless otherwise indicated. Two-way ANOVA was performed to determine statistical significance of tumor volume changes in response to drug treatment over time in in vivo experiments. A P value < 0.05 was considered significant (*P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001). Statistical analyses were performed using the software GraphPad Prism 7 (RRID:SCR_002798).

Data Availability

The RNA-seq data generated in this study have been deposited to NCBI’s Gene Expression Omnibus (GEO) repository (RRID:SCR_005012) and are publicly available through series accession number GSE273133. Additional raw data are available on request from the corresponding author.

Results

BET inhibition reverses palbociclib resistance

To explore combined BET and CDK4/6 inhibition in ER+ breast cancer models with acquired CDK4/6 inhibitor resistance, we employed a widely-used first-generation BET inhibitor, JQ1 (12), to treat ER+ MCF7 breast cancer cells resistant to 100nM palbociclib (MCF7-R100) expressing miRNAs targeting the TGF-β pathway, with consequent CDK6 overexpression (10). JQ1 suppressed the viability of MCF7-R100 cells cultured in the presence of 100nM palbociclib with an IC₅₀ of 168.3nM (Fig. 1A). In clonogenic assays, JQ1 treatment suppressed colony forming ability at nanomolar potency, i.e., 50nM (Fig. 1B). Treatment with another pan-BET inhibitor, ZEN-3694, a second-generation agent currently in clinical development (23), also demonstrated nanomolar sensitivity and impaired colony forming ability in MCF7-R100 cells maintained in palbociclib (Supplementary Fig. S1A and S1B). These cells had relative cross resistance to abemaciclib (10) but were highly sensitive to ZEN-3694 as monotherapy or when combined with abemaciclib (Supplementary Fig. S1C). A second cell line model of palbociclib-resistant T47D-R100 cells recapitulated these results, with JQ1 and ZEN-3694 treatment resulting in comparable effects on viability (Fig. 1A; Supplementary Fig. S1A) and colony formation (Fig. 1B; Supplementary Fig. S1B). Notably, although parental MCF7 cells are also sensitive to JQ1 or ZEN-3694 alone, results in viability assays indicated increased sensitivity of palbociclib-resistant derivatives. Differences in sensitivity to JQ1 and ZEN-3694 in resistant compared to parental cells were even more marked in clonogenic assays (Supplementary Fig. S1D and S1E). These data suggest emergence of an acquired vulnerability to BET inhibition with CDK4/6 inhibitor resistance.

Figure 1.

BET inhibition reverses palbociclib resistance. A, CellTiter-Glo(CTG) luminescent cell viability assay evaluating viability of palbociclib-resistant ER+ breast cancer cells (MCF7-R100 and T47D-R100) in response to JQ1. B, Clonogenic survival assay demonstrating cell survival upon JQ1 treatment. C, The top up- and down-regulated KEGG pathways calculated based on all the differentially expressed genes detected by RNA-Seq 24 hours post-treatment with JQ1 or vehicle control. D, Heatmap displaying the top 65 differentially expressed genes from the cell cycle KEGG pathway. E, Real-time qPCR analysis of CDK6 mRNA and Western blot analysis of CDK6 protein levels in MCF7 and MCF7-R100 cells treated with JQ1 for 24 hours. Data are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

To transcriptionally interrogate differential gene changes in response to JQ1 treatment, RNA-sequencing was performed to analyze mRNA profiles of vehicle and JQ1-treated MCF7-R100 cells. Principal component analysis demonstrated a large variation in the transcriptional landscape in response to JQ1 treatment (Supplementary Fig. S2A–C). Gene set enrichment analysis (GSEA) using the KEGG pathways identified signatures for the cell cycle and DNA replication pathway as the most significantly downregulated following JQ1 treatment (Fig. 1C), consistent with negative regulatory effects of BET inhibition on cell growth and cell cycle progression. Particularly relevant to these palbociclib-resistant cells, CDK6 gene expression was significantly downregulated by JQ1, as well as expression of CCND1, CDK2, CCNE1, E2F1–3 and numerous other cell cycle-related genes (Fig. 1D). The reduction in CDK6 expression was validated by qPCR in both parental MCF7 and MCF7-R100 cells (Fig. 1E). ZEN-3694 similarly reduced CDK6 expression (Supplementary Fig. S2D). Additionally, expression of numerous genes involved in DNA replication such as MCM2, MCM5 and PCNA was downregulated by JQ1 (Supplementary Fig. S2E).

Based on GSEA signatures, we next analyzed the effects of JQ1 on cell cycle distribution by flow cytometry evaluating EdU-DAPI staining in MCF7-R100 and T47D-R100 cells. JQ1treatment induced G1 arrest in both cell lines 24 hours post-treatment (Fig. 2A). Further analysis of cell cycle proteins in JQ1-treated palbociclib-resistant cell lines demonstrated a dose-dependent reduction in CDK6 (Fig. 1E; Fig. 2B and 2C) indicating reduced expression of the primary determinant of resistance in MCF7-R100 and T47D-R100 cells, whereas the effect on CDK4 expression was modest (Fig. 2C). The reduction in CDK6 was both concentration and time-dependent, occurring at doses as low as 50 nM and as early as 12 hours post-treatment (Supplementary Fig. S3A and S3B). In addition, there was marked reduction of cyclin D1 and cyclin E2, with reduced Rb phosphorylation (Fig. 2B and 2C), as well as reduced expression of estrogen receptor alpha (ER⍺) (Fig. 2C), which may also contribute to the JQ1-induced antiproliferative effects in these cells. ZEN-3694 treatment also resulted in similar patterns of protein expression in a dose-dependent manner (Supplementary Fig. S3C).

Figure 2.

BET inhibition suppresses cell cycle progression and induces apoptosis. A, EdU and DAPI staining and flow cytometry analysis of cell-cycle profiles of cells treated with JQ1 at different doses or for different time courses. Upper panel: Horseshoe plots depicting G1, S, and G2/M phases; Lower panel: quantification and analyses of cell populations in each cell cycle phase. B, Western blot analysis of Rb phosphorylation in palbociclib-resistant ER+ breast cancer cells (MCF7-R100 and T47D-R100) treated with JQ1 for 24 hours. GAPDH serves as the loading control. C, Western blot analysis of cell cycle-related proteins in palbociclib-resistant ER+ breast cancer cells (MCF7-R100 and T47D-R100) treated with JQ1 for 24 hours. Vinculin serves as the loading control. D, Propidium iodide (PI) staining and flow cytometry analysis of the cell-cycle profiles of cells treated with JQ1 at 300 nM for different time courses. Sub-2N represents sub-G1 DNA content. E, Annexin V staining and flow cytometry to assess apoptosis induced by JQ1 at 300 nM. Early and late apoptosis contribute to the total apoptotic population. F, Western blot analysis of CDK6 and PARP cleavage in palbociclib-resistant ER+ breast cancer cells treated with JQ1 for 48 hours. Vinculin serves as the loading control. G, Relative mRNA levels of BCL2 and BIRC5 as demonstrated by TPM values from RNA Seq analyses of MCF7-R100 cells treated with JQ1 or vehicle control. Data are reported as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Because c-MYC is a known target of BET inhibition, we also studied its expression in our resistant cells. Compared to parental cells, c-MYC RNA expression is not increased (10) and is even slightly decreased at the protein level (Supplementary Fig. S3D). JQ1 and ZEN-3694 both reduced c-MYC protein expression (Fig. 2B and Supplementary Fig. S3C). However, c-MYC is comparably reduced in parental and resistant cells in response to JQ1 (Fig. 2B and Supplementary Fig. S3E) and siRNA-mediated depletion affects viability of parental and resistant cells equally (Supplementary Fig. S3F). Therefore, changes in c-MYC expression do not account for the differences in sensitivity to BET inhibition between parental and resistant cells.

Following cell cycle arrest at 24 hours, treatment with JQ1 for extended time periods (i.e., 2, 3, or 5 days) showed a time-dependent increase in the sub-G1 cell population (Fig. 2D) as apoptosis induction occurred, evidenced by Annexin V/Propidium iodide staining (Fig. 2E; Supplementary Fig. S4A), as well as by induction of cleaved PARP in a dose-dependent manner (Fig. 2F). Consistent with the apoptotic biological outcome, JQ1 reduced expression of BCL-2 and BIRC5 mRNAs encoding anti-apoptotic BCL-2 and survivin in MCF7-R100 cells maintained in palbociclib (Fig. 2G). Treatment with ZEN-3694 similarly induced G1 arrest over the first 24 hours of exposure, with evidence of apoptosis after longer exposures (Supplementary Fig. S4B and S4C).

Because senescence is another biological outcome of CDK4/6 inhibition, we asked whether the re-sensitization of palbociclib-resistant cells by JQ1 was accompanied by senescence. Following prolonged JQ1 treatment, there was evidence of senescence, demonstrated by positive staining for senescence-associated β-galactosidase (SA-β-gal) in T47D-R100 cells harboring loss of function TP53 mutation, accompanied by a p53-independent increase in p21^Waf1/Cip1. However, these events did not occur in MCF7-R100 cells, where p21^Waf1/Cip1 expression was reduced by JQ1 (Supplementary Fig. S4D and E). Therefore, after prolonged BET inhibition, palbociclib-resistant MCF7-R100 cells are more prone to apoptosis, whereas the effects on T47D-R100 cells are heterogeneous.

Palbociclib-resistant tumors are sensitive to JQ1 in vivo

To confirm the sensitivity of palbociclib-resistant cells to JQ1 in vivo, we established MCF7-R100 orthotopic xenografts in the mammary fat pads of immunodeficient mice receiving daily palbociclib administration to maintain the resistant phenotype. In addition to palbociclib, mice were treated with vehicle or JQ1 (n=7/group). Tumor growth suppression was noted as early as one week after JQ1 treatment commenced (Fig. 3A), with all mice demonstrating evidence of significant tumor regression (Fig. 3B), translating to significantly improved animal survival (Fig. 3C). Moreover, tumors completely regressed in 2 animals, with these complete responses sustained through the study endpoint of 130 days (Supplementary Fig. S5A).

Figure 3.

Palbociclib-resistant tumors are sensitive to JQ1. A, Overall tumor growth of orthotopic tumors in two groups of mice treated with palbociclib or palbociclib+JQ1 (n=7 per group). Xenograft tumors were established in the presence of palbociclib using MCF7-R100 palbociclib-resistant cells. Palbociclib treatment at 100 mg/kg/day was maintained throughout the study. ANOVA was performed for the analysis and data are reported as the mean ± SD. B, Post-treatment maximal relative tumor volume changes of individual tumors. C, Kaplan-Meier plot of survival of mice treated with vehicle or JQ1 (n=7/group). D, Body weight changes of mice treated with vehicle or JQ1. Data are reported as the mean ± SD. E, Tumor growth curves of orthotopic tumors in a new group of mice treated with vehicle or JQ1 for 9 days. Data are reported as the mean ± SD. F, Western blot analysis of proteins in xenograft tumors harvested after 9 days of vehicle or JQ1 treatment. G, Real-time qPCR analysis of mRNA levels in xenograft tumors documenting the in vivo effects of vehicle or JQ1 treatment. Data are reported as the mean ± SEM. *P < 0.05, **P < 0.01, **** P < 0.0001.

Treatment of combined JQ1 and palbociclib in mice was well tolerated (Fig. 3D). To further address potential tolerability of combined BET and CDK4/6 inhibition, we treated hematopoietic stem and progenitor cells (HSPCs) with monotherapies and the combination in vitro. Compared to HSPCs, palbociclib-resistant cells were significantly more sensitive to combination treatment (Supplementary Figure S5B).

Additional mice harboring MCF7-R100 tumors with daily palbociclib administration were treated with vehicle or JQ1 for 9 days, after which tumors were harvested for pharmacodynamic assessments. JQ1 treatment again demonstrated tumor growth inhibition by day 9 (Fig. 3E). At this time point, analysis of JQ1-treated tumors showed reduced levels of CDK6 expression compared to vehicle control (Fig. 3F) as well as a consistent transcriptional reduction in CDK6 mRNA (Fig. 3G).

BET inhibition regulates the expression of CDK6 via miR-34a-5p

Previous studies from our group demonstrated that CDK6 overexpression in MCF7-R100 and T47D-R100 cells was associated with induction of miR-432-5p, which targeted SMAD4, resulting in TGF-β pathway suppression (10). We therefore examined the effects of JQ1 treatment on expression of miR-432-5p and SMAD4. Surprisingly, treatment with JQ1 increased expression of miR-432-5p in MCF7-R100 cells (Fig. 4A), as well as in T47D-R100 cells (Supplementary Fig. S6A). This increase was associated with a further reduction in SMAD4 expression in MCF7-R100 cells, where expression is already reduced compared to parental cells (Fig. 4B). JQ1 treatment also suppresses SMAD4 expression in parental MCF7 cells, where effects are enhanced by palbociclib exposure (Fig. 4C). Consistent with these findings, treatment with JQ1 strongly enhanced CDK6 promoter activity, demonstrated in a luciferase reporter assay (Fig. 4D) (22).

Figure 4.

BET inhibition regulates the expression of CDK6 via miR-34a-5p. A, Real-time qPCR analysis of miR-432-5p in MCF7-R100 cells treated with JQ1 for 24 hours. B, Western blot analysis of SMAD4 protein levels in MCF7-R100 cells treated with JQ1 or vehicle for 48 hours. C, Western blot analysis of SMAD4 protein levels in MCF7 and MCF7-R100 cells treated with JQ1 or vehicle in the presence or absence of 100 nM palbociclib for 48 hours. JQ1, 300 nM. V, vehicle; J, JQ1; P, palbociclib. D, Luciferase-reporter assay assessing CDK6 promoter activity in MCF7-R100 cells treated with JQ1 for 24 hours. E, Venn diagram demonstrating the top 10 miRNAs involved in each of the top three pathways downregulated by JQ1 treatment and identifying one common miRNA as miR-34a-5p. F, Real-time qPCR analysis of miR-34a-5p and miR-34a-3p in MCF7-R100 cells treated with JQ1 for 24 hours. G, Real-time qPCR analysis of CDK6 mRNA levels in MCF7-R100 cells treated with JQ1 for 24 hours in the presence or absence of a miR-34a-5p inhibitor. The miR-34a-5p inhibitor or control were introduced via transfection. H, Western blot analysis of MCF7-R100 cells treated with JQ1 for 24 hours in the presence or absence of miR-34a-5p inhibitor. I, Real-time qPCR analysis of miRNA levels in xenograft tumors documenting the in vivo effects of vehicle or JQ1 treatment. J, CTG luminescent cell viability assay evaluating viability of palbociclib-resistant ER+ breast cancer cells (MCF7-R100 and T47D-R100) in response to the introduction of miR-34a-5p mimic via transfection. K, Clonogenic survival assays demonstrating reduced colony formation of palbociclib-resistant cells in response to miR-34a-5p mimic. L, Western blot analysis of cell cycle-related protein expression and PARP cleavage in MCF7-R100 cells upon the introduction of different doses of miR-34a-5p mimic. M, Western blot demonstrating reduced MYC, ERα, and CDK4 levels along with low CDK6 expression in response to miR-34a-5p mimic. Quantitative data are reported as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001.

Because these effects were not in accordance with the reduced CDK6 mRNA levels observed upon JQ1 treatment (Fig. 1E), we asked whether suppression of CDK6 expression could be through an alternative miRNA-mediated mechanism. We analyzed the public miRPathDB 2.0 database to identify potential miRNAs associated with the top three differential KEGG pathway signatures identified by RNA-seq in response to short-term JQ1 treatment (Supplementary Table S1). To identify common microRNAs that regulate pathways involved in cell cycle, DNA replication, and pyrimidine metabolism, we merged the top 10 significant hits for each pathway to discover a single common microRNA, miR-34a-5p (Fig. 4E). miR-34a is a known tumor suppressor that targets the 3’UTR of CDK6 (24,25), destabilizing the mRNA and reducing the translation of the corresponding protein (26). Additionally, it has been demonstrated to target mRNAs encoding cyclin D1, CDK4, cyclin E2, and SMAD4 (27–29). This suggests that miR-34a-5p may be a potential mediator of the mechanistic reduction in CDK6 expression induced by BET inhibitor treatment.

To validate this finding, we transcriptionally assessed both end products of the miR-34a gene, miR-34a-5p and miR-34a-3p, in response to JQ1 treatment. Both miR-34a-5p and miR-34a-3p were significantly up-regulated by JQ1 in MCF7-R100 cells (Fig. 4F), as well as in T47D-R100 cells (Supplementary Fig. S6B). Similarly, miR-34a transcriptional upregulation also occurred following treatment with ZEN-3694 in MCF7-R100 cells (Supplementary Fig. S6C). Transfection of MCF7-R100 cells with a miR-34a-5p inhibitor prior to JQ1 treatment significantly attenuated the downregulation of CDK6 transcription (Fig. 4G), results reflected with CDK6 protein expression (Fig. 4H). In contrast, an inhibitor of miR-34a-3p did not prevent downregulation of CDK6 protein (Supplementary Fig. S6D), suggesting that effects of JQ1 treatment are predominantly mediated through miR-34a-5p, consistent with the miRNA database analysis (Fig. 4E). Moreover, miR-34a-5p and miR-34a-3p were both transcriptionally increased by JQ1 treatment in tumors from the in vivo study (Fig. 4I), recapitulating effects observed in vitro.

In addition, miR-34a-5p also targets expression of the c-MYC oncogene, as well as the BIRC5 gene, encoding survivin (27). Consistent with our previous data (Fig. 2B), in MCF7-R100 cells, JQ1 treatment reduced mRNA levels (Supplementary Fig. S6E) and protein expression of c-MYC, as well as survivin (Fig. 4H), the latter effect possibly contributing to JQ1-mediated induction of apoptosis (Fig. 2E). As with CDK6, effects of JQ1 treatment on expression of c-MYC and survivin were reversed by an inhibitor of miR-34a-5p (Fig. 4H), but not by an inhibitor of miR-34a-3p (Supplementary Fig. S6D).

To further validate miR-34a-5p as a primary effector of BET inhibition in CDK4/6 inhibitor-resistant breast cancer cells, we introduced an miRNA mimic of miR-34a-5p via transfection into MCF7-R100 and T47D-R100 cells. Introduction of the mimic recapitulated the results observed with JQ1 treatment demonstrating reduced cell viability (Fig. 4J) and colony formation (Fig. 4K) in both resistant cell lines, as well as cell cycle arrest and induction of apoptosis (Supplementary Fig. S6F and S6G), along with reduction in expression of CDK6, CDK4, survivin, MYC, and ERα and increase of PARP cleavage (Fig. 4L and 4M).

BET inhibition reverses palbociclib resistance via downregulation of CDK6 and survivin

Our previous data indicated that CDK6 overexpression was a primary determinant of palbociclib resistance in MCF7-R100 and T47D-R100 cells. We therefore sought to demonstrate that growth suppression by BET inhibition in these cells was related to CDK6 downregulation. To this end, we expressed exogenous CDK6 or CDK4 in MCF7-R100 cells and assessed the antiproliferative effects of JQ1 treatment in clonogenic survival assays. Similar experiments were done with exogenous expression of survivin, also significantly down regulated by JQ1 (Fig. 2G; Fig. 5A–C). Colony formation during JQ1 exposure was significantly increased by exogenous CDK6 expression and to a lesser extent by exogenous survivin expression but was not meaningfully affected by increased levels of CDK4. These data suggest downregulation of CDK6 and survivin both contribute to the ability of JQ1 treatment to overcome palbociclib-mediated resistance.

Figure 5.

BET inhibition reverses palbociclib resistance via downregulation of CDK6. A, B, &C, Clonogenic survival assays demonstrating cell survival after JQ1 treatment of MCF7-R100 cells transfected with plasmid vector control or vectors expressing CDK6, survivin, or CDK4. Western blots demonstrate levels of endogenous protein and the increases afforded by the respective vectors. D, Clonogenic survival assay demonstrating cell survival upon JQ1 treatment of MCF7-R100 or MCF7-CDK6-R100 cells. MCF7-CDK6-R100: cells depleted of CDK6 by CRISPR-knockout and re-introduced exogenous CDK6 and maintained in media containing 100 nM palbociclib. E, Western blot analysis of exogenous CDK6 protein levels in MCF7-CDK6 cells in response to JQ1 treatment. BSJ-03-123 (BSJ), is a specific degrader of CDK6,³⁹ confirming identity of the protein in these cells. BSJ-bump, the non-targeting control, was added to JQ1 treated cells. MCF7-CDK6: cells depleted of CDK6 by CRISPR-knockout with re-introduced exogenous V5-tagged CDK6. F, Western blot analysis to compare PARP cleavage in MCF7-R100 and MCF7-CDK6-R100 cells treated with JQ1. MCF7-CDK6-R100: MCF7-CDK6 cells resistant to and maintained with 100nM of palbociclib in culture medium. G, Annexin V staining and flow cytometry to assess apoptosis induced by JQ1 treatment in MCF7-R100 and MCF7-CDK6-R100 cells. Early and late apoptosis contribute to the total apoptotic population. Quantitative data are reported as the means ± SEM. *P < 0.05, ***P< 0.001.

In a further experiment, we utilized parental MCF7 cells harboring CRISPR/Cas9 CDK6 knockout (KO), in which WT exogenous CDK6 was introduced and expressed at high level (MCF7-CDK6 cells), conferring palbociclib resistance (10). Like MCF7-R100 cells, these cells were maintained in palbociclib. Whereas colony formation was markedly suppressed in MCF7-R100 cells treated with JQ1, the sensitivity of MCF7-CDK6 was significantly diminished (Fig. 5D). Notably, in MCF7-CDK6 cells, the level of exogenous viral vector-mediated expression of CDK6 was not reduced by JQ1 but surprisingly increased (Fig. 5E). Importantly, exogenous CDK6 expression markedly reduced PARP cleavage and apoptosis induced by JQ1 treatment (Fig. 5F and 5G). We performed similar engineering of T47D-R100 cells to demonstrate that CDK6 overexpression also compromises the effects of BET inhibition (Supplementary Fig. S6H).

Palbociclib downregulates BCL-2 and sensitizes cells to JQ1

We next tested the importance of continued palbociclib exposure during JQ1 treatment of CDK4/6 inhibitor-resistant cells. We treated MCF7-R100 cells with JQ1 in the presence or absence of palboclicb, i.e., after removal of palbociclib for 14 days, at a time when CDK4/6 inhibitor resistance was not yet reversed. Notably, MCF7-R100 cells were significantly more sensitive to JQ1 treatment in the continued presence compared to the absence of 100nM palbociclib (Fig. 6A). Across a range of drug concentrations, JQ1 and palbociclib co-treatment demonstrated anti-proliferative synergy in MCF7-R100 cells, despite acquired resistance to palbociclib monotherapy (Supplementary Fig. S7A). Similarly, MCF7-R100 cells also recapitulated this phenotype with significantly more sensitivity to ZEN-3694 in the presence compared to the absence of 100nM palbociclib (Supplementary Fig. S7B).

Figure 6.

Palbociclib downregulates BCL-2 and sensitizes cells to JQ1. A, Clonogenic survival assay demonstrating cell survival with JQ1 treatment of MCF7-R100 cells in the absence or presence of 100 nM palbociclib. B, Western blot analysis of BCL-2 protein levels in MCF7 cells treated with palbociclib for a series of time courses (upper panel) and comparison of BCL-2 levels in MCF7 and MCF7-R100 cells (lower panel). C, Real-time qPCR analysis of BCL2 mRNA levels in MCF7-R100 cells treated with JQ1 for 24 hours in the presence or absence of a miR-34a-5p inhibitor. The miR-34a-5p inhibitor or control were introduced via transfection. Quantitative data are normalized to vehicle controls (JQ1=0 nM) and reported as the mean ± SEM. D, Comparison of PARP cleavage and BCL-2 levels in MCF7-R100 cells treated with JQ1 in the absence or presence of palbociclib. E, Clonogenic survival assay demonstrating cell survival with JQ1 treatment of MCF7-R100 cells transfected with plasmid vector control or the vector expressing BCL-2. Western blot displays protein expression of each vector. Quantitative data are normalized to vehicle controls (JQ1=0 nM) and reported as the mean ± SEM. F, Western blot analysis of ERα protein levels in MCF7 cells treated with palbociclib for the indicated time points. G, Western blot analysis of BCL-2 in vehicle-treated MCF7 cells and MCF7-R100 cells treated with vehicle or 17b-estradiol for 48 hours. GAPDH serves as the loading control. H, A schematic diagram depicting the mechanism for the synergistic effects of CDK4/6i and BETi on CDK4/6i-resistant cells, with the associated miRNA network. Created in BioRender. Liu, R. (2025) https://BioRender.com/mhj8zzu. *P < 0.05, ***P < 0.001.

Previous studies have shown that up-regulation of anti-apoptotic protein expression contributes to JQ1 resistance (30) and that there is synergy between JQ1 treatment and inhibitors of BCL-2 and/or BCL-xL (16,30). We therefore assessed BCL-2 levels in palbociclib-treated MCF7 cells or MCF7-R100 cells maintained in palbociclib (Fig. 6B). Interestingly, BCL-2 levels were reduced by palbociclib in parental MCF7 cells after 2 days of exposure, with an even greater reduction in expression by days 4-5. In MCF7-R100 cells maintained in palbociclib, BCL-2 levels are markedly lower than in untreated parental MCF7 cells (Fig. 6B).

BCL-2 is a well-established target of miR-34a-5p (27,31). Indeed, JQ1 reduced BCL-2 mRNA levels in MCF7-R100 cells (Fig. 2G and 6C) and the effect was significantly reversed by an inhibitor of miR-34a-5p (Fig. 6C). Treatment with JQ1 further suppressed the already low BCL-2 protein expression in MCF7-R100 cells (Fig. 6D). Importantly, removal of palbociclib for 5 days from the culture of MCF7-R100 cells resulted in an increase in BCL-2 expression in both vehicle and JQ1-treated cells with reduced PARP cleavage induced by JQ1 (Fig. 6D). These results suggest that alteration in BCL-2 levels may modulate the sensitivity of MCF7-R100 cells to JQ1 and that suppression of BCL-2 expression is maximized by combined exposure to palbociclib and a BET inhibitor. As expected, overexpression of ectopic BCL-2 in MCF7-R100 cells significantly enhanced cell survival (Fig. 6E). This result was again recapitulated with ZEN-3694 treatment displaying similar effects on BCL-2 expression and PARP cleavage (Supplementary Fig. S7C).

We next investigated the mechanism for the downregulation of BCL-2 in MCF7-R100 cells. A previous study identified positive BCL-2 expression by immunohistochemistry in 73% of breast cancer cases, 86% of which were ER+ (32), implying a positive correlation between BCL-2 expression and active ER signaling. Indeed, by analyzing ESR1 and BCL-2 mRNA levels in 981 samples from patients with breast invasive carcinoma (TCGA, PanCancer Atlas) we identified a positive correlation between BCL-2 and ESR1 expression (Supplementary Fig. S8A). 17β-estradiol has been shown to induce BCL-2 gene transcription in MCF-7 cells via two estrogen-responsive elements in the coding region of the gene (33). Additionally, acquired resistance to abemaciclib has been correlated with reduced ERα expression (9).

In MCF7 cells, we found that ERα levels were reduced early after exposure to palbociclib, particularly by day 5 of treatment (Fig. 6F). Furthermore, in MCF7-R100 cells, both ERα and BCL-2 expression levels were lower compared with those in parental MCF7 cells (Fig. 6G). Stimulation of MCF7-R100 cells with 17β-estradiol rescued the expression of BCL-2 (Fig. 6G), suggesting that its reduced expression in MCF7-R100 cells is at least partially the consequence of diminished ER signaling. Notably, reduction of ERα alone by the SERD fulvestrant is not sufficient to induce apoptosis in resistant cells, indicating that the consequences of combined JQ1 and palbociclib cannot solely attributed to reduced ERα expression (Supplementary Fig. S8B).

Because T47D cells do not express readily detectable BCL-2 (Supplementary Fig. S3D), we utilized ZR-75-1 ER+ breast cancer cells to confirm our results in an additional cell line. We previously reported on ZR-75-1-R100 cells that also express high levels of CDK6 compared to their parental counterparts (10). JQ1 induced similar molecular changes and PARP cleavage in ZR-75-1-R100 cells (Supplementary Fig. S8C) to those observed in MCF7-R100 and T47D-R100 cells (Fig. 2B and C). Additionally, JQ1 induced miR-34a expression in these cells (Supplementary Fig. S8D). As in MCF7-R100 cells, ZR-75-1-R100 cells expressed lower levels of BCL-2 compared to parental cells, associated with reduced expression of ERα that was reversed with estradiol treatment (Supplementary Fig. S8E). Therefore, the mechanism of BCL-2 suppression is similar in both ZR-75-1-R100 and MCF7-R100 cells.

Taken together, our results indicate that treatment of JQ1 is more effective in MCF7-R100 cells in the continued presence of palbociclib. Mechanistically, JQ1 treatment re-sensitizes cells to palbociclib by reducing expression of CDK6 and survivin, whereas palbociclib exposure sensitizes cells to JQ1 treatment via suppression of BCL-2 expression (Fig. 6D). To this end, since palbociclib treatment reduces BCL-2 expression even before the emergence of frank acquired resistance, the combination of JQ1 and palbociclib treatment also demonstrates synergy in parental, palbociclib-sensitive MCF7 cells (Supplementary Fig. S7A).

Discussion

Proliferation and maintenance of ER+ breast cancer is tightly linked to cyclin D1-CDK4 activity (34), accounting for the impact of CDK4/6 inhibitors in improving patient outcomes. Nonetheless, both intrinsic and acquired resistance complicate use of these agents and mandate development of novel approaches. Mechanisms of resistance are heterogeneous, although several may converge on elevated CDK2 activity, prompting substantial investment in the evaluation of selective CDK2 inhibitors (35). There has been less focus on the role of the target kinases in mediating CDK4/6 inhibitor resistance, although several mechanisms resulting in elevated CDK6 expression have now emerged and appear relevant beyond preclinical models, with documentation in primary tumor samples (7,10). Overexpression of CDK6, also an NF-κB co-factor (36,37), may offer a survival advantage to ER+ breast cancer cells as well. The potential for CDK6 to contribute to cellular survival was supported by our prior work demonstrating that the knockdown of the low levels of CDK6 in parental cells resulted in compromised survival during palbociclib exposure, whereas knockdown of cyclin D1 or CDK4 did not produce similar effects (10). Notably, to reduce hematologic toxicity related to cyclin D3-CDK6 inhibition, highly selective CDK4 inhibitors are also under active development (38); although resistance mechanisms specific to these agents have not yet been extensively documented, breast cancer cells may adapt to these agents via elevated CDK6 expression (39).

In previous studies, CDK6-specific degraders have been proposed as a strategy to address expression of CDK6 and its overactivity in CDK4/6 inhibitor-resistant cells (8). Here, we demonstrate that BET inhibition may also be useful in CDK6-overexpressing palbociclib-resistant cells, with suppression of cell growth, cell cycle arrest, and induction of apoptosis in vitro, as well as tumor regression in vivo. Our results support those previously reported for the anti-proliferative effects of combined ZEN-3694 and abemaciclib in ER+ breast cancer cells, including those with CDK4/6 inhibitor resistance, although this study did not find similar effects with palbociclib and did not examine miRNA induction by BET inhibition (40).

The prototypical BET inhibitor, JQ1, has been shown to suppress CDK6 gene transcription (11,13). Consistently, in both palbociclib-resistant MCF7-R100 and T47D-R100 cells, we observed significant suppression of CDK6 expression by JQ1 in a time and dose-dependent manner. Whole transcriptome sequencing with GSEA using KEGG pathways demonstrated CDK6 mRNA expression as significantly reduced in the top downregulated pathway, i.e., the cell cycle pathway, validated by qPCR. However, promoter activity of CDK6 was enhanced upon JQ1 treatment. Similar to palbociclib monotherapy, treatment with JQ1 increases miR-432-5p, resulting in reduced SMAD4 expression and suppression of the TGF-β pathway, events that have been associated with increased CDK6 expression. Instead, CDK6 mRNA and protein downregulation in response to treatment with JQ1 or ZEN-3694 in palbociclib-resistant breast cancer cells occur via induction of a second miRNA, miR-34a-5p. The finding was further supported by the observation that expression of exogenously introduced CDK6 lacking the 3’UTR of the endogenous CDK6 mRNA was not reduced by JQ1.

miR-34a-5p (of the miR-34 family) is a well-established versatile tumor-suppressive miRNA and its dysregulation is involved in a large variety of cancers including both solid tumors and hematological malignancies (27,29). Therefore, in addition to the blockade of oncogene transcription machinery, the induction of the tumor-suppressive miR-34a-5p by BET inhibition may underlie its efficacy in multiple types of cancers. Further work will aim to delineate the mechanism of miR-34a-5p induction by BET inhibition, and to generalize our findings by assessing the contribution of miR-34a-5p to BET inhibitor-mediated antitumor activity in a variety of models including triple-negative and ER+ breast cancers.

Moreover, it is well-established that BET inhibition downregulates MYC transcription (41). Here, we demonstrated a novel mechanism by which BET inhibition reduces c-MYC expression via the induction of miR-34a-5p. Future studies will elucidate whether this is also the major mechanism underlying the anti-tumor efficacy of BET inhibition in MYC-dependent tumors. Notably, early exposure of parental breast cancer cells to palbociclib and cells fully CDK4/6 inhibitor-resistant are characterized by reduced expression of ER⍺, which is linked to low expression of BCL-2 (33). The palbociclib-mediated reduction in BCL-2 expression likely sensitizes cells to BET inhibition. The effects of palbociclib on BCL-2 expression are reversible, so that in the absence of palbociclib, the ability of BET inhibitor treatment to initiate apoptosis or suppress colony formation was markedly compromised. When used as a single agent in the absence of palbociclib, BET inhibitor treatment only partially reduces BCL-2 expression (even when CDK6 expression is ablated) and this residual BCL-2 likely limits JQ1-mediated antiproliferative and apoptotic effects. In summary, reduced BCL-2 expression mediated by palbociclib sensitizes cells to JQ1, just as reduced CDK6 and survivin expression mediated by JQ1 sensitizes cells to palbociclib, explaining the strong synergistic effects of combined CDK4/6 and BET inhibition in ER+ breast cancer.

In models of T-cell acute lymphoblastic leukemia (T-cell ALL), CDK4/6 inhibition has been shown to induce apoptotic responses (42), in contrast to the cell cycle quiescence and/or senescence that has been described in solid tumor models (43). It will be of interest to determine whether CDK4/6 inhibition mediates reduced BCL-2 expression in these models, contributing to the apoptotic biological outcome. Based on the critical roles of kinase-dependent and independent CDK6 activities (44) in T-cell ALL cell proliferation and survival (45), inhibition of CDK4/6 (or selective CDK6 inhibition/degradation) combined with BET inhibition may be of substantial interest in this leukemia subset. Similar considerations may exist for acute myelogenous leukemia (AML), where CDK6 inhibition/degradation and BET inhibition have individually shown preclinical promise (46–49). Reduced BCL-2 expression mediated by CDK inhibition may also address mechanisms of venetoclax resistance, including those related to BCL-2 mutation that preclude venetoclax binding (50).

One limitation of our experiments is the primary focus on CDK4/6 inhibitor-resistant cells, including work in vivo, where we did not study combined JQ1 and palbociclib in parental MCF7 xenografts. Nonetheless, our data suggest that acquired palbociclib resistance renders MCF7 and T47D more sensitive to JQ1 or ZEN-3694 treatment, particularly in colony formation assays, likely related to the acquired CDK6 dependence of resistant cells. Therefore, although parental cells have some degree of sensitivity to BET inhibition, this sensitivity is maintained and even heightened in palbociclib-resistant cells, so that our strategy is highly relevant to the CDK4/6 inhibitor-resistant state. However, our data on the synergism between palbociclib and JQ1 emphasize the importance of evaluating such combinations in ER+ breast cancer models that have not yet developed acquired drug resistance. Because palbociclib suppresses BCL-2 expression even before the emergence of frank resistance, and because JQ1 is expected to suppress the emergence of CDK6 overexpression, palbociclib and JQ1 were indeed synergistic in parental cells, as well as in resistant cells. Therefore, comprehensive studies investigating the efficacy of BET inhibition as monotherapy or in combination with CDK4/6 inhibition + hormonal therapy to induce regression and to delay or prevent resistance in ER+ parental tumors will be important for fully realizing the utility of BET inhibition in ER+ breast cancer.

Our study focused on palbociclib-resistant cells that were previously well characterized. After acquired resistance to initial therapy with palbociclib and an aromatase inhibitor develops, use of a different CDK4/6 inhibitor combined with appropriate endocrine therapy may be superior to endocrine therapy alone (51,52) so that combined ZEN-3694 and abemaciclib or ribociclib may be a compelling option after resistance to first-line treatment. Indeed, our palbociclib-resistant cells, with relative cross-resistance to abemaciclib, were sensitive to combined ZEN-3694 and abemaciclib. Additionally, as abemaciclib and ribociclib are now more likely to be used in the adjuvant and first-line metastatic settings based on emerging survival data (1), it will be important to document the activity of BET inhibition in breast cancer cells rendered resistant to these agents either by elevated CDK6 expression or other mechanisms that may be addressed by miR-34a-5p induction.

In summary, our findings demonstrate the effectiveness of combined BET and CDK4/6 inhibition in ER+ breast cancer, particularly in CDK6 overexpressing CDK4/6 inhibitor-resistant tumors and provide a novel treatment strategy. It will be important to determine if synergy extends beyond models utilizing elevated CDK6 expression as a mechanism of acquired CDK4/6 inhibitor resistance. Furthermore, since miR-34a-5p mediates the major effects of BET inhibition, miRNA-based therapeutics, including a fully modified version of miR-34a that overcomes delivery and pharmacokinetic barriers, may provide a new approach for the treatment of a subset of CDK4/6 inhibitor-resistant cancers (53). Currently, we are conducting a Phase 1 study to test the safety and preliminary efficacy of the BET inhibitor ZEN-3694 combined with abemaciclib primarily in patients with NUT Carcinoma (NCT05372640). Once the recommended Phase 2 dose is established, the trial will be extended to patients with endocrine- and CDK4/6 inhibitor-resistant ER+ breast cancer, to translate our findings to the breast cancer population.