Introduction

The cell cycle is precisely controlled in benign cells by cyclin-dependent kinases (CDKs), regulatory cyclins, and checkpoint regulators. However, in cancer, the cell cycle is frequently altered because of somatic mutations in genes encoding cell-cycle regulators and defects in checkpoint proteins (1, 2). CDK4/6 inhibitors used clinically include palbociclib, abemaciclib, and ribociclib, and these are administered in combination with hormone therapy for the treatment of breast cancer. When the RB pathway is intact, these CDK 4/6 inhibitors block the transition of cells from the G1- to the S-phase, which leads to accumulation of cells in G1 (3).

Androgen receptor (AR) is a transcription factor that regulates the expression of thousands of genes including those involved in cell growth and survival. Altered expression and transcriptional activity of AR contribute to the development and progression of castration-resistant prostate cancer (CRPC) and some triple-negative breast cancers (TNBCs). AR associates with regulators of the cell cycle (4). Overexpression of CDK6 occurs in approximately 50% of prostate cancer and enhances AR transcriptional activity to promote disease progression (5). Increased AR transcriptional activity is associated with disruption of cyclin D1 repression in prostate cancer (6–8). In LNCaP human prostate cancer cells, AR interacts with the pre-replication complex during mid-G1-phase and associates with DNA synthesis, indicating its essential role during the G1–S transition (9). Luminal AR is a subtype of TNBC that is particularly sensitive to CDK4/6 inhibitors, and this sensitivity is associated with AR expression and low cyclin E1 levels (10).

Full-length AR consists of N-terminal domain (NTD), DNA-binding domain, a hinge region, and a ligand-binding domain (LBD) to which androgens and antiandrogens bind. Deletion of LBD results in constitutively active AR variants (AR-Vs), of which AR-V7 is most clinically relevant. AR-V7 is associated with the failure of CRPC to therapies targeting the full-length AR such as antiandrogens and androgen deprivation therapies (11, 12). AR-V7 is also detected in human breast cancer cell lines as well as in clinical samples of breast cancer (13–16). AR-V7 mediates a growth response and regulates expression of a subset of genes that is unique from those regulated by the full-length AR (15, 17).

Ralaniten (aka EPI-002) and its analogs specifically bind to transactivation unit-5 (Tau-5) within activation function 1 (AF-1) of the intrinsically disordered AR NTD (18–22). Uniquely, ralaniten compounds inhibit the transcriptional activities of full-length AR, AR-Vs, as well as gain-of-function mutations in AR LBD that are found in patients with CRPC who are resistant to antiandrogens and abiraterone (18, 20, 22). Antiandrogens and AR-NTD inhibitors cause prostate cancer cells that express full-length AR to arrest in the G1 phase and not progress to the S-phase of the cell cycle (9, 20, 23). Thus, a functional AR is required for G1–S progression. Cyclin D1-CDK4/6 and late G1 cyclin E-CDK2 kinases phosphorylate the RB protein, which leads to E2F regulation of expression of cyclins that are required for the transition to the S-phase (24). Based on the importance of CDK4/6 in controlling G1–S progression, we propose a therapy of inhibiting CDK4/6 combined with an AR-NTD antagonist that blocks transcription activities of all AR species (full-length AR and AR-Vs).

Monotherapies for metastatic cancers rarely lead to durable responses or cures primarily due to tumor plasticity and heterogeneity with the survival and outgrowth of resistant clones. The goals of combination therapies include deeper and more durable responses and can include approaches such as targeting pathway reactivation and/or targeting multiple components of a signaling pathway. In this study, we examined the effects of targeting both AR-NTD and CDK4/6 in CRPC and AR-expressing breast cancer with a combination of a second-generation ralaniten analog and palbociclib. EPI-7170 is a more potent analog of ralaniten that specifically blocks AR transcriptional activity (23, 25, 26). Here, we show that EPI-7170 disrupted the cell cycle in CRPC and AR-expressing breast cancer cell lines, and when combined with palbociclib had improved antitumor activity. As expected, palbociclib caused G1 accumulation, whereas EPI-7170 caused both G1 accumulation and also delayed the cell cycle. Sequential versus concomitant combinations resulted in differential effects on the cell cycle.

Materials and MethodsCell lines, cell culture, and inhibitor treatment

LNCaP cells were obtained from Leland Chung (Cedars-Sinai Medical Center, Los Angeles, CA) and maintained in RPMI1640 supplemented with 5% [volume for volume (v/v)] FBS. LNCaP95 cells were obtained from Stephen Plymate (University of Washington, Seattle, WA) and maintained in RPMI1640 supplemented with 10% (v/v) charcoal-stripped serum (CSS). SUM159PT cells were obtained from Sharon Gorski (Canada's Michael Smith Genome Sciences Centre at BC Cancer, Vancouver, British Columbia, Canada) and maintained in Ham's F-12 Nutrient Mix supplemented with 5% (v/v) FBS, l-Glutamine, HEPES, 1 μg/mL hydrocortisone, and 0.2 U/mL bovine insulin. MFM-223 cells were purchased from DSMZ (DSMZ No.ACC422) and maintained in MEM supplemented with 10% (v/v) FBS, 1% (v/v) L-Gln and insulin-transferrin-selenium. MDA-MB-453 cells were purchased from ATCC (ATCC HTB-131) and maintained in RPMI1640 supplemented with 10% (v/v) FBS. MDA-MB-468 cells were purchased from ATCC (ATCC HTB-132) and maintained in MEM supplemented with 10% (v/v) FBS and 1% (v/v) L-Gln. All cell lines were incubated at 37°C with 5.0% CO2. All cell lines have been authenticated using short tandem repeat profiling and were routinely tested every 6 months to ensure Mycoplasma free using the Venor GeM Mycoplasma Detection kit (Sigma-Aldrich). The last test for Mycoplasma was done July 23, 2021. All treatments with inhibitors or vehicle controls were carried out in culture maintaining media supplemented with FBS for each of the cell lines except LNCaP95 cells which were in 5% CSS. All repeated experiments were conducted using cells within five to 10 passages. EPI-7170 was synthesized by Raymond Andersen (University of British Columbia, Vancouver, British Columbia, Canada; refs. 23, 25). Enzalutamide (ENZA) was purchased from Omega Chem. Palbociclib (PD-0332991) HCl was purchased from Selleckchem (catalog No. S1116). These three inhibitors were dissolved in DMSO. Vehicle controls were DMSO (final = 0.02–0.04% v/v). For combination treatments, concomitant regimens (two inhibitors were mixed in media and added to the cultures at the same time) were performed unless a sequential regimen was otherwise stated.

Cell viability assay

LNCaP (3,000 cells/well), LNCaP95 (3,000 cells/well), and SUM159PT (2,000 cells/well) were plated in 96-well culture plate (Corning Primaria). The next day (as day 0), cells were treated with inhibitors at the indicated concentrations or with vehicle control. On days 2 and 3 of treatment, alamarBlue was added to the treated cells and incubated at 37°C for 2 to 4 hours before fluorescence measurement (560-nm excitation and 590-nm emission) by Infinite M1000 (Tecan). EPI or enzalutamide in medium without cells did not change alamarBlue fluorescence measurement as compared to medium only. GraphPad Prism 8 was used for plotting data and statistical analyses.

Clonogenic assay

LNCaP (400–600 cells/well), LNCaP95 (500 cells/well), and SUM159PT (400 cells/well) were plated in 24-well culture plates. The next day (as day 0), cells were treated with inhibitors at indicated concentrations or with vehicle control without medium change. For consistency, the same cell numbers were plated before treatments, instead of plating the same cell number after treatments which would require adjusting cell numbers after collecting cells from each of the treatments. For experiments with a one-time medium change: on day 3 (LNCaP95) or day 4 (SUM159PT) of treatment, the media from one set of treated cells was changed to fresh media without inhibitors. On day 15 (LNCaP and LNCaP95) or day 7 (SUM159PT) of treatments, cells were fixed with 4% paraformaldehyde, washed with PBS, and stained with 0.01% (w/v) crystal violet. Images of colonies were captured by ChemiDoc XRS+ Imaging System with a trans-white conversion screen and Image Lab 5.1 software (Bio-Rad).

Cell-cycle analysis

For cell cycle with bromodeoxyuridine (BrdU) incorporation, non-synchronized cells were plated (200,000 or 300,000 cells per dish for LNCaP; 300,000 cells per dish for LNCaP95 and MDA-MB-453; 150,000 cells per dish for SUM159PT; 500,000 for MFM-223; 250,000 for MDA-MB-468) in 60-mm culture dishes. The next day (as day 0), cells were treated with inhibitors at the indicated concentrations or with vehicle control. At the indicated timepoints, cells were labeled with 10 μmol/L BrdU for 2 hours. After BrdU incorporation, culture media containing both nonattached cells and attached cells that were trypsinized by TrypLE were collected. Cells were then passed through a 50-μm filter to obtain single cells and fixed by cold 70% (v/v) ethanol. Fixed cells were stained with anti–BrdU-FITC antibody (clone B44, BD Biosciences, catalog No. 347583) and 7-aminoactinomycin D (7-AAD from Sigma, catalog No. A9400), analyzed by flow cytometry using FACSCalibur and data were acquired by CellQuest Pro software (BD Biosciences). For cell-cycle tracing with BrdU incorporation, experimental procedures were the same as described above except cells were labeled with BrdU for 2 hours prior to inhibitor treatments. Cells were harvested and collected at the indicated timepoints. All cell-cycle data were analyzed with FlowJo V.10. Single-cell gating was performed on FSC-H versus FL3-W profiles. Data calculations for bar or line graphs used GraphPad Prism 8.

For analysis of LNCaP95 cell cycle with EdU incorporation and marker expression, Click-iT EdU Alexa Fluor 647 Flow Cytometry assay kit (Invitrogen) was used, and manufacturers' protocol was followed with modifications on fixation and intracellular marker staining. Cells (1,200,000 per dish) were plated on a 100-mm culture dish. The next day (as day 0), cells were treated with inhibitors or with vehicle control. On day 3 of treatments, cells were labeled with 10 μmol/L EdU for 2 hours prior to collecting nonattached and attached cells. Collected cells were fixed with 70% cold ethanol. Incorporated EdU was detected using Click chemistry. Antibody against phospho-histone H2A.X (Ser 139; clone: 20E3; Cell Signaling Technology No. 9718) was used for intracellular marker staining with staining carried out using TBS/FBS (4% v/v). Secondary antibody conjugated with Alexa Fluor 488 was used (Invitrogen A11008 for anti-rabbit IgG). After intracellular marker staining, cells were stained with 7-AAD, analyzed by LSRFortessa, and data were acquired by FACSDiva software (BD Biosciences). All cell-cycle data were analyzed using FlowJo V.10. Single-cell gating was performed on SSC-A versus SSC-W profiles. Data calculations for bar graphs were done using GraphPad Prism 8.

Cell number counting was carried out using an automatic cell counter (EVE Automated Cell Counter, NanoEnTek Inc.). The same number of cells from different treatment conditions at different timepoints were used for analyses of the cell cycle (BrdU) and EdU/γH2AX except for cell-cycle tracing experiments where all collected cells were used for BrdU staining to show cell number changes using 7AAD staining.

Tumor xenograft

Animal studies were approved by the University of British Columbia Animal Care Committee. Mice were maintained in Animal Resource Centre at BC Cancer Research. Six-week-old male NSG mice were castrated. Two weeks after castration, 2 × 106 cells of LNCaP95-D3 cells prepared in 1:1 ratio of RPMI and Matrigel (Corning, Product 354234) were inoculated subcutaneously on the back of mice. When tumor volume reached 50 to 90 mm3, mice were randomly separated into four treatment groups: vehicle control, EPI-7170 (30 mg/kg body weight), palbociclib (75 mg/kg body weight), and combination (EPI-7170 at 30 mg/kg body weight and palbociclib at 75 mg/kg body weight). Treatments were given daily by oral gavage for 21 days. Treatment solutions were freshly prepared each day. A working solution of palbociclib (isethionate salt purchased from LC Laboratories P-7766) was prepared daily from powder by dissolving in sterile Milli-Q water. To be completely soluble, each treatment solution was prepared in the following order of DMSO or EPI-7170 (dissolved in DMSO), Tween-80, methyl cellulose solution (Sigma M0512; 0.5% w/v prepared in sterile Milli-Q water), and sterile Milli-Q water or palbociclib solution to have final treatment concentrations in 5% DMSO/1.5% Tween-80/0.5% methyl cellulose. Body weight was measured every day. Tumor volume was measured every 3 days and calculated using formula: length × width × height × 0.5236.

Protein expression analysis

Whole-cell lysates were extracted from LNCaP95 cells treated with inhibitors or vehicle control after 3 days by lysing cells in lysis buffer RIPA buffer containing protease inhibitors (cOmplete EDAT-free protease inhibitor cocktail and PhosStop from Roche). Xenograft tumors were homogenized in lysis buffer in Lysing Matrix M with FastPrep-24 homogenizer (MP Biomedicals). Protein concentrations in cell lysates were quantified by the bicinchoninic acid assay (Pierce BCA Protein assay kit). An equal amount of protein (40 μg) of each sample was loaded and resolved on 10% or 12.5% SDS-PAGE and transferred to nitrocellulose membrane for immunoblotting analyses. Antibodies were diluted in TBS (pH 8.0)/Tween20 (0.05%) containing 5% milk or 5% BSA (w/v). Washes were done using TBS (pH8.0)/Tween20 (0.05%). Antibodies included AR [clone: ER179(2); abcam ab108341], RB (clone: 4H1; Cell Signaling Technology No. 9309), phospho-RB at Ser780 (clone: D59B7; Cell Signaling Technology No. 8180), CDK4 (clone: D9G3E; Cell Signaling Technology No. 12790), CDK6 (clone: DCS83; Cell Signaling Technology No. 3136), CDK2 (clone: D-12; Santa Cruz Biotechnology sc-6248), cyclin D1 (clone: 92G2; Cell Signaling Technology No. 2978), cyclin A2 (clone: BF683; Cell Signaling Technology No. 4656), p27 Kip1 (clone: D69C12; Cell Signaling Technology No. 3686), Skp2 (clone: D3G5; Cell Signaling Technology No. 2652), RAD51 (clone: D4B10; Cell Signaling Technology No. 8875), and β-actin (clone: AC-15; Sigma-Aldrich A5441). Secondary antibodies conjugated with horseradish peroxidase were used (Cell Signaling Technology No. 7074 for anti-rabbit IgG and Cell Signaling Technology No. 7076 for anti-mouse IgG) and enhanced ECL (GE Healthcare Lifescience Amersham ECL Prime Western Blotting Detection Reagent) was applied for chemiluminescence detection by ChemiDoc XRS+ Imaging System and Image Lab 5.1 software (Bio-Rad). Relative protein levels were quantified by Image Lab 5.1 software. Data and bar graphs were calculated and generated by GraphPad Prism 8.

Immunofluorescence staining

LNCaP95 cells were plated on an 8-well chamberslide (15,000 cells/well). On the day, cells were treated with vehicle control, EPI-7170, palbociclib, or EPI-7170 combined with palbociclib as indicated. After 3 days of treatments, cells were fixed with 4% paraformaldehyde (v/v) and washed with TBS, pH8.0. Fixed cells were penetrated with TBS/0.1% TritonX-100 (v/v), blocked with TBS/0.05% Tween20 (v/v)/5% BSA (w/v) and stained in the same blocking buffer with antibodies against phospho-histone H2AX (Ser 139; clone: D7T2V; Cell Signaling Technology No.80312) and 53BP1 (Cell Signaling Technology No.4937S). Secondary antibody conjugated with Alexa Fluor 488 (Invitrogen A11008 for anti-rabbit IgG) and Alexa Fluor 568 (Invitrogen A11031 for anti-mouse IgG) was used. Secondary antibodies only were used as negative controls. After staining, cells were washed and mounted with mounting media containing DAPI. Stained cells were imaged by Axio Imager.M2 (ZEISS) and images of two-dimensional planes were acquired and processed by ZEN blue edition 2.6.

Foci numbers were quantified by ImageJ 1.45s and cell nuclei identified by ImageJ were manually confirmed. Cell clusters without clear separation or boundary were excluded. Thresholds (or noise tolerance in ImageJ) to identify foci of both 53BP1 and γH2AX were set to have a 1:1 ratio of both proteins in control within each independent experiment to allow comparison among experiments. At least three different fields from each condition and total number of cells (control: 1753, EPI-7170: 854, palbociclib: 1207, and E+P: 617) were analyzed from independent experiments (n = 4).

Statistical analysis

Statistics analyses were facilitated using GraphPad Prism 8. Two-way ANOVA with Tukey, Dunnett, or Sidak multiple comparisons test and One-way ANOVA with Tukey multiple comparisons test were used to calculate statistical significance among treatments. t test was used to determine statistical significance for tumor weight, Ki67, and TUNEL analyses in xenograft study. For estimation of cell doubling times, the exponential growth model (Malthusian model) was used to estimate the best-fit values.

Data availability

All data are available within the article and its Supplementary Data.

ResultsCombination of palbociclib and EPI-7170 affects cell viability, cell number, and colony formation

A combination of palbociclib with EPI-7170 was tested in CRPC and AR-expressing breast cancer cells to determine their effective concentrations to reduce tumor growth, colony formation, and cell viability. Enzalutamide was included as a control because it targets the full-length AR by binding to the AR-LBD and has no effect on AR-Vs. Antiandrogens including more potent second-generation compounds such as enzalutamide are routinely used in cell culture at a concentration of 10 μmol/L (27) presumably to saturate the AR (28).

LNCaP human prostate cancer cells express full-length AR, are negative for AR-Vs, and are sensitive to androgen and enzalutamide. These cells exhibited a dose-dependent decrease in cell viability in response to palbociclib when used as a monotherapy and when combined with EPI-7170 or enzalutamide (Fig. 1A). As expected, enzalutamide caused an approximately 25% to 40% decrease in the viability of LNCaP cells that plateaued at 2.5 μmol/L (Fig. 1B, orange line). The addition of 0.25 μmol/L palbociclib resulted in a synergistic effect with enzalutamide (Fig. 1B, light orange line; Supplementary Fig. S1) in LNCaP cells.

Figure 1.

EPI-7170 and palbociclib inhibit cell growth and viability. A–C, One day after plating, LNCaP or LNCaP95 cells were treated with various concentrations of Palbo (dark blue), Palbo + 5 μmol/L EPI-7170 (light blue), or Palbo + 5 μmol/L ENZA (orange; A); various concentrations of EPI-7170 (dark blue), EPI-7170 + 0.25 μmol/L Palbo (light blue), various concentrations of ENZA (orange), or ENZA + 0.25 μmol/L Palbo (light orange; B). One day after plating, SUM159PT cells (C) were treated with various concentrations of EPI-7170 (dark blue), EPI-7170 + 0.125 μmol/L Palbo (green), or EPI-7170 + 0.25 μmol/L Palbo (light blue). Cell viability was measured by alamarBlue after day 2 or day 3 of treatments. Measurements were normalized to vehicle controls in each of the independent experiments. Data are shown as mean ± SEM (n ≥ 3 independent experiments). Single cells of LNCaP (D), LNCaP95 (E), or SUM159PT (F) were plated in low density to allow colony formation. The next day, cells were treated with vehicle control (DMSO), various concentrations of EPI-7170, ENZA or Palbo with a fixed concentration for one of the inhibitors. On day 3 (LNCaP95) or day 4 (SUM159PT) of treatments, media were changed to fresh medium without inhibitors in one set of treated cells. The media in the other set of treated cells was not changed. On day 15 (LNCaP and LNCaP95) or day 7 (SUM159PT) of treatments, cells were fixed for crystal violet staining. Representative data are shown (n = 3 independent experiments). ENZA, enzalutamide; Palbo, palbociclib.

LNCaP95 cells express full-length AR and AR-Vs, but their proliferation is driven by AR-Vs, and hence they are resistant to enzalutamide (22). The addition of 5 μmol/L enzalutamide had no effect on its own nor did it improve the response achieved with palbociclib in LNCaP95 cells, whereas 5 μmol/L EPI-7170 had a profound inhibitory effect on its own in the absence of palbociclib, which could not be improved with the addition of higher doses of palbociclib (plateaued; Fig 1A). Maintaining the concentration of palbociclib at 0.25 μmol/L and adding increasing amounts of EPI-7170 yielded dose-dependent decreases in the viability of LNCaP and LNCaP95 cells (Fig. 1B, light blue line) that were synergistic at lower concentrations of EPI-7170 in LNCaP cells but at higher concentrations of EPI in LNCaP95 cells (Supplementary Fig. S1). No increase in cell density or in cell number was detected with increased concentrations of EPI-7170 with or without palbociclib (Supplementary Figs. S2 and S3).

The TNBC cell line, SUM159PT, exhibited a dose-dependent decrease in cell viability in response to EPI-7170 monotherapy (Fig. 1C, dark blue line), and its combination with palbociclib resulted in synergistic decreases at higher concentrations of EPI-7170 (Fig. 1C, green and light blue lines; Supplementary Fig. S1).

Interestingly, LNCaP cells were unable to form colonies in the presence of 0.25 μmol/L palbociclib (Fig. 1D) even though cell viability and cell number were not reduced to zero on day 3 (Fig. 1A; Supplementary Fig. S2). On the contrary, LNCaP95 cells were able to form colonies that were reduced in a dose-dependent manner by EPI-7170 and by palbociclib, whereas enzalutamide had no effect (Fig. 1E, left: no medium change; Supplementary Fig. S4A). Holding the concentrations of EPI-7170 and enzalutamide constant at 5 μmol/L, palbociclib inhibited colony formation in a dose-dependent manner regardless of inhibitor in LNCaP95 cells. A combination of EPI-7170 with palbociclib resulted in a synergistic effect at higher concentrations of EPI-7170 (Supplementary Fig. S4B and S4C). In addition, the inhibitory effect was reversible as indicated by increased colonies after removal of both inhibitors after 3 days of treatments compared with conditions without removal of the inhibitors (Fig. 1E, right: medium change on day 3). The reversible effect was not 100% complete, as colony numbers were less at the higher concentrations tested (10 or 15 μmol/L EPI-7170; 500 nmol/L palbociclib) than vehicle control, which may suggest that some of the cells were either dead or permanently affected in their ability to form colonies or proliferate. Similar results were observed in SUM159PT cells (Fig. 1F).

Combination of EPI-7170 and palbociclib alters the cell cycles of CRPC and breast cancer cells

Attenuation of AR transcriptional activity and inhibition of CDK4/6 activity both independently block cell-cycle progression from G1- to the S-phase (4, 9, 29–31). Hence, we next examined whether EPI-7170 that targets AR NTD would affect the cell cycle in both prostate cancer and breast cancer cells and whether it would deepen any response mediated by palbociclib. Non-synchronized cells were used to reflect what would be encountered in tumor tissues with all phases of the cell cycle present. In addition, one particular method of synchronizing cells may result in different phases in differing types of cells. Single cells were selected for cell-cycle analyses as shown in Supplementary Fig. S5. On the basis of the above results from colony formation assays and cell viability, concentrations of 5 and 10 μmol/L of EPI-7170 were chosen with 10 μmol/L being approximately 10-fold its IC50 for blocking AR transcriptional activity in LNCaP cells (23). Consistent with previous reports (23, 25), EPI-7170 significantly inhibited the transcriptional activities of androgen-induced full-length AR as well as androgen-independent AR-V7 transcriptional activity as demonstrated by luciferase reporter gene assays (Supplementary Fig. S6A and S6B) and measurement of expression of endogenous genes regulated by AR and AR-V7 with quantitative RT-PCR (Supplementary Fig. S6C and S6D). Palbociclib had no effect on the transcriptional activities of androgen-induced full-length AR or AR-V7 as measured using two reporter gene constructs (Supplementary Fig. S6A and S6B). As expected palbociclib had no effect on androgen-regulated genes (KLK3 for PSA and KLK2) while impacting the cell-cycle genes UBE2C, CDC20, and CCNA2 that are also known AR-V7 target genes (ref. 17; Supplementary Fig. S6C and S6D). These data support that the synergistic effects observed on cell viability and colony formation measured with the combination therapy of palbociclib with EPI-7170 are not due to palbociclib impacting AR transcriptional activity.

As expected in a cell line that depends on a transcriptionally active full-length AR for proliferation, the percentages of cells in the S-phase were reduced when LNCaP cells were treated with enzalutamide and EPI-7170 (Fig. 2A; Supplementary Fig. S7A). At 10 μmol/L EPI-7170, LNCaP cells in the S-phase decreased to 1.71 ± 0.52% (mean ± SEM) on day 2 and 3.75 ± 0.59% on day 3, and importantly decreased to almost 0% when EPI-7170 was combined with palbociclib. Unique to treatment with 10 μmol/L EPI-7170, either alone or with palbociclib, was the increase in the percentage of cells in G2–M (∼twofold compared with controls) that was not observed with enzalutamide, thereby differentiating these two inhibitors of AR in a cell line that expresses solely full-length AR. A combination of 5 μmol/L EPI-7170 and palbociclib resulted in a similar effect on the S-phase, but not on G1 or G2–M, as measured with 10 μmol/L EPI-7170 monotherapy (i.e., reduced S percentages compared with controls). Palbociclib monotherapy increased the percentage of LNCaP cells in G1 on day 2 as expected, with little to no effect on day 3 compared with controls. However, the combination on day 3 with either enzalutamide or EPI-7170 had a significant effect compared with the individual treatments. Similar cell distribution patterns (i.e., ratios among different phases of cell cycle) were detected between day 2 and day 3 of treatments when comparing among the different treatment conditions. However, S percentage increased on day 3 (compared with day 2) under EPI-7170 treatment with or without palbociclib, whereas S percentage decreased on day 3 (compared with day 2) for enzalutamide treatment with or without palbociclib.

Figure 2.

EPI-7170 and palbociclib alter the cell cycle in AR-expressing cancer cells. A–F, Cells were treated with different treatments as indicated and collected at two timepoints prior to 2-hour BrdU labeling for cell-cycle analyses by flow cytometry. Distribution of each cell-cycle phase for different cell lines are shown: LNCaP-expressing AR (A), LNCaP95-expressing AR and AR-V (B), SUM-159PT–expressing AR (C), MFM-223–expressing AR and AR-V (D), MDA-MB-453–expressing AR and AR-V (E), and AR-negative/RB-negative MDA-MB-468 (F). Data are presented as mean ± SEM (n ≥ 3 independent experiments). *, P < 0.05 for monotherapy treatment compared with control or combination treatment compared with palbociclib alone; #, P < 0.05 for combination treatment compared with respective ENZA alone or EPI-7170 alone; +P < 0.05 for comparison between ENZA and EPI-7170 at the same concentrations in each cell-cycle phase (two-way ANOVA); symbol color corresponds to bar color for each cell-cycle phase. ENZA, enzalutamide; Palbo, palbociclib.

Another important finding was the unique increase of percentages of cells in the S-phase without BrdU incorporation [S BrdU-negative (S BrdU−)] seen in cells treated with 10 μmol/L EPI-7170 and with a combination of EPI-7170 (5 and 10 μmol/L) with palbociclib that was not observed with enzalutamide monotherapy or combination. This S BrdU− population was more profoundly increased in cells treated with a combination of 10 μmol/L EPI-7170 plus palbociclib (day 2: 3.72 ± 0.69% in combination and 0.24 ± 0.03% in control; day 3: 3.38 ± 0.61% in combination and 0.34 ± 0.06% in control).

Proliferation of LNCaP95 cells is driven by AR-Vs and in these cells EPI-7170 had similar impact on the cell-cycle profile as seen with LNCaP cells, but this was not the case for enzalutamide that had no effect when compared with controls (Fig. 2B; Supplementary Fig. S7B). Interestingly, even though higher S percentages were detected in LNCaP95 controls (day 2: 38.95 ± 0.68%; day 3: 29.70 ± 1.10%) than in LNCaP controls (day 2: 18.66 ± 0.62%; day 3: 14.52 ± 0.59%), the combination treatment of EPI-7170 and palbociclib caused reduction in the S population of LNCaP95 cells to almost 0% as observed in LNCaP cells. The G1 accumulation induced by palbociclib in LNCaP95 was consistent with a previous report using other AR-V–driven CRPC cells treated with this compound (29). In addition, the percentages of S BrdU− cells increased to approximately 13% with 10 μmol/L EPI-7170 plus palbociclib compared with controls.

The impact of these inhibitors on the cell cycle of several breast cancer cell lines was next examined. SUM159PT TNBC cells exhibited two distinct the S-phase populations (early S and late S) on days 3 and 4 (Supplementary Fig. S8A). EPI-7170 treatment with or without palbociclib in these cells did not result in substantial changes in total percentages of cells in the S-phase (Fig. 2C) as compared with that measured in LNCaP or LNCaP95. However, the percentages of SUM159PT cells in early the S-phase were reduced by EPI-7170 on day 4 (Supplementary Fig. S8A and red bars in Supplementary Fig. S8B). Enzalutamide (10 μmol/L) reduced the percentages of cells in early the S-phase but to a lesser extent than EPI-7170. As expected, palbociclib increased the percentage of cells in the G1-phase (Fig. 2C; Supplementary Fig. S8A). However, palbociclib reduced the S BrdU− population (gray bars in Fig. 2C; Supplementary Fig. S8A) and reduced the number of tetraploid cells which are the cells with doubled DNA content (i.e., non-gated cells, higher 7AAD intensity, and excluded from calculation in bar graphs).

In MFM-223 TNBC cells that express both full-length AR and AR-V7 (15), the percentages of cells in the S-phase were reduced by EPI-7170, but there were no distinct the S-phase populations (Fig. 2D; Supplementary Fig. S8C). The combination of 10 μmol/L EPI-7170 with palbociclib reduced the percentage of MFM-223 cells in the S-phase to almost zero, which was consistent to that observed in LNCaP and LNCaP95 cells as shown above.

MDA-MB-453 breast cancer cells are negative for both estrogen and progesterone receptors but importantly express AR-V7 (15). These cells responded to EPI-7170 more robustly compared to enzalutamide especially with respect to accumulation in G1 and reduction of cells in the S-phase (Fig. 2E; Supplementary Fig. S8D).

MDA-MB-468 TNBC cells were included as a negative control because they do not express either AR or RB (Supplementary Fig. S9A). As predicted, there was no impact on the cell-cycle phases of MDA-MB-468 cells in response to EPI-7170, enzalutamide, or palbociclib (Fig. 2F; Supplementary Fig. S9B), thereby supporting the specificity of these inhibitors.

Taken together, these results indicate EPI-7170 disrupted the cell cycle in prostate cancer and breast cancer cells that express AR and that its effects were greater in combination with palbociclib. Enzalutamide did not disrupt cell cycle as profoundly as EPI-7170 for cell lines that express AR-Vs such as in LNCaP95, MFM-223, and MDA-MB-453.

Combination of EPI-7170 and palbociclib improves in vivo antitumor activity

To investigate the effectiveness of combination treatments in AR-V7–driven tumors, the LNCaP95-D3 human CRPC xenograft model was employed. This enzalutamide-resistant cell line expresses AR-Vs and is less heterogeneous than parental LNCaP95 cells (32). Reduced tumor growth was observed in animals receiving EPI-7170 monotherapy and palbociclib monotherapy (Fig. 3A–C). The combination treatment significantly reduced the tumor growth compared with the control group and EPI-7170 monotherapy (Fig. 3A and B). Body weight was not significantly altered among the different treatment groups (Fig. 3D). The xenograft tumors from three individual best responders under combination treatment exhibited decreased percentage of proliferating cells (i.e., Ki67-positive cells) and increased percentage of apoptotic cell death (TUNEL; Supplementary Fig. S10).

Figure 3.

Combination treatment blocks the growth of CRPC xenografts. A, Mice bearing LNCaP95-D3 xenografts were randomly separated into four treatment groups: vehicle control, EPI-7170 (30 mg/kg body weight), palbociclib (75 mg/kg body weight), and combination (EPI-7170 at 30 mg/kg body weight and palbociclib at 75 mg/kg body weight). Treatments were given daily by oral gavage for 21 days. Tumor size was measured every 3 days. Data are presented as mean ± SEM (n = 9 tumors in each treatment group except n = 8 for palbociclib group). ****, P < 0.0001 compared with control; ###, P < 0.001 compared with EPI-7170 monotherapy (two-way ANOVA); symbol color corresponds to treatment group. B, Weights of collected tumors were measured. Data are presented as mean ± SEM (n = 9 tumors in each treatment group except n = 8 for palbociclib group). ns, not significant; *, P < 0.05; **, P < 0.01 compared between any two treatments (t test). C, Images of collected xenograft tumors on day 22 were shown. D, Body weight was measured every day over the course of treatments. Data are presented as mean ± SEM (n = 9 tumors in each treatment group except n = 8 for palbociclib group). E, Levels of AR and cell cycle–related proteins were analyzed in whole-cell lysates from xenograft tumors. Relative levels of proteins were quantified by normalization to β-actin levels in each sample and then normalization to the average of control samples on the same immunoblot membrane. Each data point represents one individual tumor. Three individual tumors from the best responders in each treatment groups were labeled with the color orange. Blue line indicates the mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 compared with control (two-way ANOVA). F, Protein levels in LNCaP95 cells treated with vehicle control DMSO, 10 μmol/L EPI-7170, 0.25 μmol/L palbociclib, or 10 μmol/L EPI-7170 + 0.25 μmol/L palbociclib for 3 days. Representative data are shown from n = 3 independent treatments. Protein levels were quantified by normalization to β-actin levels in each of the treatments and to the average of control samples. Bar graphs are presented as mean ± SEM (n = 3). *, P < 0.05; **, P < 0.01; ***, P < 0.001 compared with control (two-way ANOVA); asterisk colors correspond to types of treatments indicated on the graph.

The essential role of AR on cell-cycle progression has been shown by experiments examining androgen depletion, antiandrogens, and knockdown of AR (9, 33–36). Knockdown of AR or inhibition of AR transcriptional activity by antiandrogens or androgen depletion blocks progression from G1 to S by mechanisms involving decreased levels of CDK2, and cyclin A and increased levels of p27Kip1 (34–36). Here we examined in vivo and in vitro levels of AR and cell cycle–associated proteins in response to individual and combination treatments. Levels of these proteins detected from the harvested xenografts were not statistically significantly different between treatment groups and control group due to a wide range of variability (Fig. 3E; Supplementary Table S1). However, individual tumors from the best responders in each of the treatment groups appeared to be distinct from the average values as shown for AR, AR-V, pS780-RB, CDK6, cyclin A2, p27Kip1, and Skp2 (labeled orange in Fig. 3E). The original immunoblot images are provided in Supplementary Figs. S11 and S12.

Levels of these cell cycle–associated proteins were also measured from in vitro studies using LNCaP95 cells that were treated with EPI-7170 and palbociclib. Consistent with previous studies in enzalutamide-resistant prostate cancer cell lines (25), EPI-7170 treatment was associated with decreased protein levels of AR-V7, whereas not altering levels of full-length AR (Fig. 3F). EPI-7170 altered the protein levels of pS780-RB, CDK2, cyclin A2, p27Kip1, Skp2, and RAD51, which was consistent with its on-target effects on AR and mechanism of action in blocking cell-cycle progression from G1 to S (34–36). Palbociclib significantly altered the protein levels of CDK4, cyclin D1, and CDK2 (Fig. 3F; original immunoblot images are provided in Supplementary Fig. S13). The increased levels of CDK4 and cyclin D1 proteins induced by palbociclib are consistent with stabilization of CDK4–cyclin D1 complex as demonstrated by others with this compound (37). Interestingly, when compared with the control, the combination treatment significantly decreased the protein levels of total RB, pS780-RB, CDK2, cyclin A2, Skp2, and RAD51, but significantly increased the levels of p27Kip1 protein (Fig. 3F). This increased level of p27Kip1 protein was also detected in LNCaP95-D3 xenografts from the best responders treated with EPI-7170 monotherapy and the combination (Fig. 3E); whereas palbociclib did not alter the level of p27Kip1. The differential effects of EPI-7170 and palbociclib on the levels of these cell cycle–associated molecules supports their distinct inhibitory mechanisms on the cell cycle. Importantly, EPI-7170, palbociclib and the combination reduced the levels of CDK2 and cyclin A2 proteins, which have critical roles in DNA replication (38). These reduced protein levels were consistent with decreased percentages of cells in the S-phase with EPI-7170, palbociclib and combination treatments. EPI-7170, palbociclib, and the combination significantly reduced the protein levels of RAD51, which is important for DNA damage repair.

EPI-7170 affects the DNA damage response

Antiandrogens and abiraterone have been reported to increase DNA damage in LNCaP, LNCaP-AR, and VCaP cells (39, 40). Decreased levels of RAD51 protein as measured here with EPI-7170 are also consistent with studies using antiandrogens and androgen ablation (41, 42). Hence, we examined whether EPI-7170 behaved similarly to other AR antagonists and would increase DNA damage. Phosphorylated H2AX (γH2AX) was detected in LNCaP95 cells treated with EPI-7170, and these levels could be further increased by combination treatment (Fig. 4A and B). Interestingly, with EPI-7170 or the combination treatment, the majority of the S-phase EdU− (same as S BrdU−) cells expressed γH2AX (∼60%–70% in Fig. 4A, bottom and 4C). On the contrary, only a small portion (∼20%) of the S-phase EdU− cells expressed γH2AX in the control or palbociclib-treated cells. Cells in G2–M also expressed γH2AX at a significantly higher level with the combination than cells treated with control or solely palbociclib (Fig. 4C). The expression trend of γH2AX in cells in G1 was similar to the trend in G2–M cells but to a lesser extent (Fig. 4C). Immunofluorescence staining demonstrated that the foci of γH2AX and 53BP1 were increased in cells treated with EPI-7170 and the combination as compared with control and indicated by the number of foci per cell and the percentage of cells with more than five foci in the nucleus (Fig. 4D–G). Even though the number of foci for both proteins was correlated in a linear manner in each cell regardless of treatment (Fig. 4G), they did not always colocalize (Fig. 4D). The increase of DNA double-strand breaks was also observed in LNCaP cells treated with EPI-7170 or combination (Supplementary Fig. S14). Taken together, these data suggest that EPI-7170 may induce DNA double-strand breaks or impair the DNA-damage response similar to other inhibitors of the androgen axis.