Emergence of Enzalutamide Resistance in Prostate Cancer is Associated with BCL-2 and IKKB Dependencies

Other than skin cancer, prostate cancer is the most common cancer in North American males and the second leading cause of cancer-related deaths (1). Androgen deprivation therapy, as well as anti-androgen therapies [such as enzalutamide (ENZ) and abiraterone acetate (AA)], are now well-established treatments for both de novo and castration-resistant metastatic disease (2, 3). Unfortunately, despite significant efficacy, these therapies are not curative, and resistance inevitably develops (4, 5). For example, in the castrate-resistant setting with ENZ, the median time to PSA progression is only approximately 11 months (6, 7). Previous studies have suggested several possible mechanisms of ENZ resistance including androgen receptor (AR) gene amplifications, mutations, and/or splice variants; activation of other prosurvival or proliferation pathways including PI3K/AKT (8), MEK/ERK (9), and glucocorticoid receptor (10); or transformation into a neuroendocrine phenotype (11). There remains a need for additional therapeutic strategies to delay the emergence of resistance, and to identify putative biomarkers and targets that may indicate a higher probability of de novo or acquired resistance. In this study, we carried out a chemical screen and identified two drugs, venetoclax (ABT199) and IMD0354, targeting BCL-2 and IKKB, respectively, as potential therapies to target ENZ-resistant prostate cancer in vitro and in vivo. We subsequently validated both the drugs and their genomic targets in several preclinical assays and prostate cancer samples, to provide a basis for future clinical development.

Human ENZ-sensitive, castration-resistant prostate cancer (CRPC) cell line 16D^CRPC and ENZ-resistant, CRPC cell lines including 34B^ENZR, 40C^ENZR, and 30C^ENZR were acquired from the Vancouver Prostate Centre (Vancouver, Canada) and derived as described previously (12). They were maintained in RPMI1640 supplemented with 10% FBS at 37°C in 5% CO₂ atmosphere. Regular cell-line authentication was done by short-tandem repeat profiling. Mycoplasma testing was routinely done with the MycoAlert Mycoplasma Detection Kit (LT07-118, Lonza). ENZ-resistant cells were routinely tested for their resistance by growing them in 20 μmol/L ENZ for 2-week periods and quantifying their growth rates. All ENZ-resistant and ENZ-sensitive lines were also routinely tested for their ability to grow in phenol red–free RPMI1640 supplemented with 10% charcoal-stripped serum and quantifying their growth rates. Unless otherwise indicated, cells used for growth assays and drug screens were plated in media supplemented with 10% charcoal-stripped serum. Where appropriate in assays, ENZ concentration was 20 μmol/L. ENZ, ABT199, and IMD0354 were purchased from Selleck Chemicals. For experiments, cells were plated for 24 hours before media was replaced with fresh drugs. Cells were incubated for 5–6 days without further media changes.

Four prostate cancer cell lines (16D^CRPC, 34B^ENZR, 40C^ENZR, and 30C^ENZR) were dissociated into single cells and seeded at 1,000–1,500 cells per well in 50 μL RPMI1640 medium (supplemented with 10% charcoal-stripped serum) in 384-well microplates. Compounds were dissolved in DMSO as 10 mmol/L stocks. Compound addition was performed at Ontario Institute for Cancer Research (OICR; Toronto, Ontario, Canada) using HP Tecan D300 digital dispenser. For each compound, 12-point, 2-fold serial dilutions were performed. Drug effects were compared with cells optimally proliferating in 0.1% DMSO alone, while wells filled with media served as background. CellTiter-Glo (5 μL, Promega) was added after 5 days, and fluorescence intensity measured after 20 minutes on a PHERAstar microplate reader, equipped with a λ540 excitation/λ590 emission filter.

Cells were seeded into black, clear-bottom 96-well assay plates or 6-well plates [for Annexin-V/propidium iodide (PI) assays] and incubated for 24 hours prior to treatment with serially increasing concentrations of ENZ, ABT199, or IMD0354. After 5 days of exposure to drugs, Alamar Blue (Life Technologies) reagent was added and incubated for 4 hours. Fluorescence intensity was measured, and inhibition of proliferation was calculated by normalizing to an untreated control. Apoptotic cells were analyzed by FACS using the BioVision Annexin-V/PI kit as per the manufacturer' instructions.

Wells in 96-well assay plates were coated by 0.6% agar. Cells (100/well) were seeded in 0.4% agar with DMSO, ABT199, or IMD0354 in RPMI1640 supplemented with 10% FBS (complete media). 0.5% agar in complete media with DMSO or inhibitors were covered on top of the cells. After 14-day incubation at 37°C in 5% CO₂ atmosphere, bright-field pictures were captured under microscope. Alamar Blue (Life Technologies) reagent was added and incubated for 4 hours. Fluorescence intensity was measured, and fold changes were calculated by normalizing to the DMSO control of 16D^CRPC cells.

The plasmids of BCL-2-GFP and IKKB-mCherry were purchased from GeneCopoeia. Plasmid transfection was performed with Lipofectamine 3000 (Life Technologies) in accordance with the manufacturer's guidelines. The siRNAs targeting BCL-2 and IKKB were purchased from Dharmacon (Horizon Discovery). Transfection of siRNAs at multiple concentrations was performed with Lipofectamine RNAiMAX (Life Technologies) in accordance with the manufacturer's guidelines.

Cells were washed twice with cold PBS and scraped in 50 mmol/L Tris HCl pH 8.0, 150 mmol/L NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS supplemented with protease and phosphatase inhibitors (Roche). Lysates were clarified by centrifugation and protein concentration determined via BCA. A total of 20–60 μg reduced protein was resolved by NuPAGE 12% gels (Life Technologies) in MES gel running buffer. Proteins were transferred onto polyvinylidene difluoride membrane, blocked with 5% skim milk and probed overnight at 4°C with antibodies listed in Supplementary Table S1. Bound protein was visualized using horseradish peroxidase–conjugated secondary antibodies (GE Healthcare) and chemiluminescence. Western blots analyzed using the mean grayscale value of inverted scanned Western blots in Adobe Photoshop, and ANOVA performed with the use of PRISM7 for MacOSX.

RNA was isolated using TRI reagent (Sigma) and samples were reverse transcribed using qScript cDNA SuperMix (Quantas). qRT-PCR was performed on an Eppendorf Realplex mastercycler using SYBR green (Quantas). Specific primers used are listed in Supplementary Table S2.

Male NOD/SCID mice (4–6 weeks old) were anesthetized using 2% isoflurane (inhalation) and 1 × 10⁶ 16D^CRPC or 30C^ENZR prostate cancer cells suspended in 100 μL of PBS with 50% Matrigel (BD Biosciences) were implanted subcutaneously into the dorsal flank on the right side of the mice. Once the tumors reached a palpable stage (∼200 mm³), the animals were randomized and treated with vehicle (DMSO), 10 mg/kg ENZ (dissolved in DMSO), 100 mg/kg ABT199 (dissolved in 50% PEG300 + 5% Tween 80 + ddH₂O), or 10 mg/kg IMD0354 (dissolved in 5% Tween 80 + 0.5% carboxymethyl cellulose + ddH₂O) intraperitoneally, every other day. Growth in tumor volume was recorded using digital calipers, and tumor volumes were estimated using the formula (π/6) (L × W²), where L is the length of tumor and W is the width. Animal survival was determined on the basis of the tumor sizes reaching maximal volumes allowable (1,500 mm³) under the University Health Network Institutional Animal Care and Use Committee (IACUC). At the endpoint, mice were sacrificed, and tumors were extracted. All cells planned for inoculation into mice require the authentication of being free of all Mycoplasma contamination. Typically, qRT-PCR profiling of Mycoplasma contamination was performed prior to inoculation. All experimental procedures were approved by the University Health Network IACUC under the protocol AUP 5643.

Patients were identified from a population of men with CRPC treated at the Royal Marsden NHS Foundation Trust (RMH, London, UK) in accordance with the Declaration of Helsinki. All patients had given written informed consent and were enrolled in institutional protocols approved by the RMH (London, UK) ethics review committee. Twenty patients with sufficient formalin-fixed, paraffin embedded (FFPE) diagnostic (archival) castration-sensitive prostate cancer (CSPC) biopsies, and matched FFPE CRPC biopsies were identified. All CSPC biopsies demonstrated adenocarcinoma and were from prostate needle biopsies (15 patients), transurethral resection on the prostate (TURP; 1 patient), prostatectomy (2 patients), transurethral resection on the bladder (1 patient) and a neck mass (1 patient). CRPC tissue was obtained from metastatic biopsies of bone (9 patients), lymph node (8 patients), TURP (1 patient), and pelvis (soft tissue; 2 patients). All tissue blocks were freshly sectioned and only considered for IHC analyses if adequate material was present (≥50 tumor cells; reviewed by B. Gurel). Demographic and clinical data for each patient were retrospectively collected independently from the hospital electronic patient record system.

IHC for BCL-2 and IKKB was performed on patient samples described above. BCL-2 IHC was performed using the mouse anti-BCL-2 mAb (Dako; M0887). Briefly, antigen retrieval was performed for 10 minutes with Bond ER2 solution (Leica Biosystems) and anti-BCL-2 antibody (1:750 dilution) incubated with tissue for 15 minutes and the reaction visualized using Bond Polymer Refine (Leica Biosystems) system. Appendix tissue was used as a positive control. Cell pellets from LNCaP95 cells treated with control and BCL-2 siRNA were used to confirm specificity of the antibody for BCL-2 (Supplementary Fig. S1). Mouse IgGs were used as negative controls. IKKB IHC was performed using the rabbit anti-IKKB mAb (Cell Signaling Technology; D30C6). Briefly, antigen retrieval was performed for 10 minutes with Bond ER2 solution (Leica Biosystems) and anti-IKKB antibody (1:500 dilution) incubated with tissue for 15 minutes and the reaction visualized using Bond Polymer Refine (Leica Biosystems) system. Appendix tissue was used as a positive control. Cell pellets from DU145 cells treated with control and IKKB siRNA were used to confirm specificity of the antibody for IKKB (Supplementary Fig. S2). Rabbit IgGs were used as negative controls.

Cytoplasmic BCL-2 and IKKB protein expression was determined for each case by a pathologist blinded to clinical data using the modified H score (HS) method, a semiquantitative assessment of staining intensity that reflects antigen concentration. HS was determined according to the formula: [(% of weak staining) × 1] + [(% of moderate staining) × 2] + [(% of strong staining) × 3], yielding a range from 0 to 300.

HS were reported as median values with interquartile range (IQR). For paired, same patient, CSPC and CRPC expression studies, the Wilcoxon matched-pair signed-rank test (nonparametric; BCL-2) or paired Student t test (parametric; IKKB) was used to compare differences in cytoplasmic protein expression levels. All analyses were conducted, and graphs were generated, using GraphPad Prism v6.

ENZ-sensitive CRPC cell line 16D^CRPC and ENZ-resistant cell lines 34B^ENZR, 40C^ENZR, and 30C^ENZR were established from LNCaP xenografts as described previously (ref. 12; Supplementary Fig. S3A). Initially, we characterized the three ENZ-resistant cell lines 34B^ENZR, 40C^ENZR, and 30C^ENZR, along with the ENZ-sensitive line 16D^CRPC. While all four lines had similar growth kinetics (Supplementary Fig. S3B), the ENZ-resistant lines consistently showed greater tolerance to higher ENZ concentrations (mean IC₅₀ ranged from 49.7 to 50.8 μmol/L) when compared with the ENZ-sensitive 16D^CRPC line (mean IC₅₀ of 12.3 μmol/L) or the hormone-sensitive LNCaP parental line (mean IC₅₀ 9.6 μmol/L; Supplementary Fig. S3C and S3D). All tested cell lines were AR and PSA positive (Supplementary Fig. S3E), and neither the ENZ-resistant 40C^ENZR/30C^ENZR lines nor the ENZ-sensitive 16D^CRPC line expressed AR-V7 (Supplementary Fig. S3F and S3G) or neuroendocrine biomarkers including synaptophysin and neuron-specific enolase (Supplementary Fig. S3H). However, we identified the known AR point mutation T878A (formerly T877A) in both LNCaP and all the derivative cell lines. Furthermore, all the ENZ-resistant cell lines (34B^ENZR, 40C^ENZR, and 30C^ENZR) harbored a heterozygous mutation F877L (formerly F876L; Supplementary Fig. S3I), which has been reported to be a possible inducer of ENZ resistance (13, 14).

To identify small-molecule inhibitors that may be used to target ENZ resistance, we carried out pharmacologic screens using the CellTiter-Glo assay on the ENZ-sensitive line and the three ENZ-resistant lines. Our compound pool consisted of two libraries, “OICR Kinase Inhibitor Library” (480 compounds) and “OICR Tool Compound Library” (300 compounds; Supplementary Table S3). The initial screen with the two libraries was performed using 40C^ENZR and 30C^ENZR cell lines, in a randomized-plating manner (Supplementary Fig. S4A). The drug effects were classified into four subsets: “cytotoxic,” “inactive,” “partial response,” and “cytostatic” (Supplementary Fig. S4B). For all the 184 compounds that were classified as “cytotoxic” in both 40C^ENZR and 30C^ENZR (Supplementary Fig. S4C), we performed a secondary screen using all the four cell lines including 16D^CRPC, 34B^ENZR, 40C^ENZR, and 30C^ENZR. To identify inhibitors that specifically target ENZ-resistant cells, we compared the IC₅₀ values for 16D^CRPC and the mean IC₅₀ values of the three ENZ-resistant cell lines (IC₅₀ Mean_ENZR), and excluded those compounds that had lower IC₅₀ values to 16D^CRPC cell line compared with any of the ENZ-resistant cell lines (138 compounds left; Supplementary Fig. S4D). To meet the criteria that the inhibitors should have at least 2-fold greater effect to the ENZ-resistant lines compared with the ENZ-sensitive line, that is, IC₅₀ Mean_ENZR/IC₅₀ (16D^CRPC) < 0.5, the secondary pharmacologic screen identified only two hits: venetoclax (ABT199), a BCL-2 inhibitor, and IMD0354, an IKKB inhibitor (Supplementary Fig. S4E). BCL-2 is an antiapoptosis canonical member of the BCL-2 family of regulator proteins while IKKB (IKK-β, inhibitor of NFκB kinase subunit beta) is an enzyme that serves as a protein subunit of IκB kinase.

We used the Alamar Blue–based proliferation assays to validate the in vitro effect of ABT199 and IMD0354. For ABT199, the mean IC₅₀ ranged from 0.67 to 0.76 μmol/L for the ENZ-resistant lines, much lower than the ENZ-sensitive line 16D^CRPC (mean IC₅₀ of 35.3 μmol/L; Fig. 1A). For IMD0354, the mean IC₅₀ ranged from 0.24 to 0.25 μmol/L for the ENZ-resistant lines, compared with the mean IC₅₀ of 0.64 μmol/L for the ENZ-sensitive line 16D^CRPC (Fig. 1B). The two inhibitors have similar effects on ENZ-resistant cell lines in the presence or absence of ENZ (Fig. 1C and D). In contrast, both hormone-sensitive (LNCaP) and castrate-resistant (16D^CRPC) cells were not as sensitive as the ENZ-resistant lines to ABT199 or IMD0354 (Fig. 1E and F). We validated this finding using the Incucyte Live-Cell Analysis Systems to measure the cell confluency. We observed that with the treatment of ABT199 or IMD0354, the cell growth of ENZ-resistant lines including 40C^ENZR and 30C^ENZR was significantly slowed down compared with ENZ-sensitive line 16D^CRPC (Supplementary Fig. S5A and S5B). These data suggest that ABT199 and IMD0354 decrease the cell proliferation of ENZ-resistant cell lines in vitro.

To determine whether the sensitivity of the ENZ-resistant lines to ABT199 or IMD0354 is due to increased cell death, we performed trypan blue counts after exposure to either drug for 5 days. We found that there was a significantly higher percentage of trypan blue positive cells in the inhibitor-treated ENZ-resistant cell lines 40C^ENZR and 30C^ENZR when compared with the ENZ-sensitive line 16D^CRPC (Supplementary Fig. S5C). Also, the treatment of ABT199 or IMD0354 significantly repressed the soft agar colony formation of 40C^ENZR and 30C^ENZR cells, compared and 16D^CRPC (Supplementary Fig. S5D and S5E). To assess the contribution of apoptosis to the cell death effect with ABT199 or IMD0354, we measured the protein cleavage of PARP1. Intriguingly, we observed that ABT199 triggered apoptosis in both 40C^ENZR and 30C^ENZR cells while IMD0354 did not (Supplementary Fig. S5F), despite evidence of cell death with trypan blue assays (Supplementary Fig. S5C). We used Annexin-V staining to further validate whether the cells would undergo apoptosis with the treatment of ABT199 at increasing concentrations (2.5–10 μmol/L). All ENZ-resistant cell lines underwent apoptosis in response to increasing concentrations of ABT199, and the 30C^ENZR cells showed a significant increase in sensitivity to the ABT199 at a concentration of 10 μmol/L (P < 0.0001; Supplementary Fig. S5G).

We interrogated the effect of ABT199 and IMD0354 on 16D^CRPC or 30C^ENZR subcutaneous xenografts in vivo. Consistent with the in vitro results, the growth of 16D^CRPC-derived xenografts responded to the treatment of ENZ, but neither ABT199 nor IMD0354 alone (Fig. 1G), suggesting that the 16D^CRPC cells were not sensitive to the inhibitors in vivo. Surprisingly, the treatment of ABT199 or IMD0354 alone appeared to stimulate the growth of 16D^CRPC xenografts (Fig. 1G). For 30C^ENZR-derived xenografts, in the presence of pharmacologic castration with ENZ, we found that the tumors were resistant to ENZ as expected (Fig. 1H; ENZ + vehicle control). In contrast, the growth of 30C^ENZR-derived xenografts was significantly diminished by the two inhibitors (ENZ + vehicle vs. ENZ + ABT199, P < 0.001; ENZ + vehicle vs. ENZ + IMD0354, P < 0.001), strongly implying the sensitivity of 30C^ENZR cells to ABT199 and IMD0354. The body weight of all the animals at the endpoint was equivalent, indicating that neither ABT199 nor IMD0354 adversely affected animals' well-being (Supplementary Fig. S5H). Taken together, the inhibition of BCL-2 and IKKB, by ABT199 and IMD0354, respectively, showed selective targeting of ENZ-resistant cell lines and decreased cell growth both in vitro and in vivo.

To further validate the BCL-2 or IKKB inhibitor effects, we performed genetic manipulation for both BCL-2 and IKKB. We measured cell growth on ENZ-sensitive (16D^CRPC) and ENZ-resistant lines (40C^ENZR) when transfected with increasing concentrations of two independent siRNAs targeting BCL-2 or IKKB. The negative control siRNA did not affect the cell growth of either 16D^CRPC or 40C^ENZR (Fig. 2A). As we increased the concentration of siBCL-2 (Fig. 2B and C) or siIKKB (Fig. 2D and E), the ENZ-resistant cell line 40C^ENZR showed a greater repression of cell proliferation in a dose-dependent manner, compared with the ENZ-sensitive cell line 16D^CRPC. We confirmed that the repression of cell growth with increased siRNA concentrations correlated with decreased levels of BCL-2 or IKKB proteins (Fig. 2F and G). These data suggest that the proliferation of ENZ-resistant cell lines is BCL-2 and IKKB dependent.

To further explore the BCL-2 and IKKB dependencies in ENZ resistance, we initially determined the protein levels of BCL-2 and IKKB in the ENZ-sensitive cell line 16D^CRPC and two of the ENZ-resistant cell lines including 40C^ENZR and 30C^ENZR, and observed that the protein level of BCL-2 and IKKB in 40C^ENZR and 30C^ENZR cells was much higher compared with 16D^CRPC (Fig. 3A). We generated 16D^CRPC variants overexpressing BCL-2, IKKB, or both (Fig. 3B) which had similar growth kinetics as parental 16D^CRPC cells (Supplementary Fig. S6A), and showed greater tolerance to higher ENZ concentrations (mean IC₅₀ ranges from 27.5 to 52.7 μmol/L) compared with the parental 16D^CRPC line (mean IC₅₀ of 10.5 μmol/L; Fig. 3C and D). When treated with 20 μmol/L ENZ, the proliferation of 16D^CRPC was dramatically repressed (Fig. 3E), whereas the BCL-2– or IKKB-overexpressing lines were partially resistant to ENZ treatment (Fig. 3F and G). The dual-overexpressing variant showed complete ENZ resistance, comparable with the ENZ-resistant lines used previously (Fig. 3H and I; Supplementary Fig. S6B). The direct comparison for ENZ fold inhibition of cell viability between 16D^CRPC parental control and BCL-2– and/or IKKB-overexpressing cells is also significant (Supplementary Fig. S6C).

To further determine the role of IKKB in regulating ENZ resistance, we performed IKKB knockdown assays using the two previously used siIKKBs, and tested the cell growth in the presence of ENZ. When transfected with 20 nmol/L siIKKBs, the ENZ-resistant 40C^ENZR cell line exhibited a greater sensitivity to ENZ (IC₅₀: 17.58–19.01 μmol/L vs. 58.27 μmol/L of negative control), whereas the knockdown of IKKB did not change the ENZ sensitivity in ENZ-sensitive 16D^CRPC cell line (IC₅₀: 10.91–12.81 μmol/L vs. 11.93 μmol/L of negative control; Supplementary Fig. S6D).

As IKKB presumably activates NFκB pathway, we assessed the NFκB activity by measuring the protein level of nuclear p65 and cytoplasmic IκBα. We observed no nuclear expression of p65 and no change of IκBα level between 16D^CRPC parental control and its variants overexpressing BCL-2 and/or IKKB (Supplementary Fig. S6E), indicating that the NFκB pathway was not activated in this setting.

We also tested the expression of multiple AR downstream targets including KLK3, NKX3.1, and TMPRSS2 in the presence of ENZ in the 16D^CRPC cell variants, and found that the overexpression of BCL-2 and/or IKKB did not affect the decrease of all the three AR targets with the treatment of ENZ (Supplementary Fig. S6F), suggesting that the ENZ resistance induced by BCL-2 and/or IKKB overexpression may not be relevant to canonical AR signaling. These data of overexpression and knockdown assays, especially the dual overexpression of BCL-2 and IKKB, suggest that the emergence of ENZ resistance in prostate cancer is driven by BCL-2 and IKKB.

To explore the effect of the inhibitors to prevent the emergence of de novo resistance, we treated the 16D^CRPC tumor-bearing mice with ENZ in combination with vehicle control, BCL-2 inhibitor, or IKKB inhibitor in two cohorts. For the first cohort, the drugs including ENZ, ABT199, and IMD0354 were administered to the animals when the tumor size reached 200 mm², the time before ENZ resistance developed. We found that the treatment of ABT199 or IMD0354 could significantly repress the tumor growth when the lesion started to relapse from day 28 (ENZ + vehicle vs. ENZ + ABT199, P < 0.01; ENZ + vehicle vs. ENZ + IMD0354, P < 0.001), suggesting that the inhibitor of BCL-2 or IKKB suppresses the development of ENZ resistance in 16D^CRPC tumor-bearing mice (Fig. 4A; Supplementary Fig. S7A). For the second cohort, we explored the effect of the drugs on acquired ENZ resistance. We treated the 16D^CRPC tumor-bearing mice with vehicle control or ENZ alone when the tumor size reached 200 mm². After the ENZ-treated tumors reached 300 mm² (meaning the emerged ENZ resistance), we randomly grouped the animals and administered them the BCL-2 or IKKB inhibitor or vehicle control in addition. Consistently, we found the growth of the relapsed tumors was significantly slowed by the treatment with ABT199 or IMD0354 (ENZ + vehicle vs. ENZ + ABT199, P < 0.01; ENZ + vehicle vs. ENZ + IMD0354, P < 0.05; Fig. 4B; Supplementary Fig. S7B). Setting 15 mm of tumor's length as endpoint, the Kaplan–Meier analysis showed that ENZ in combination with ABT199 or IMD0354 increased the endpoint-free survival of 16D^CRPC relapsed tumor-bearing mice, compared with the treatment of ENZ plus vehicle control (P = 0.017 and P = 0.067, respectively; Fig. 4C and D). Taken together, these results suggest that the BCL-2 or IKKB inhibitor not only prevents the emergence of ENZ resistance, but also targets the ENZ-resistant disease in vivo.

Subsequently, we tested the AR genotype for the xenograft samples collected at the endpoint, and found that in all cases no change from the implanted cells. All the relapsed 16D^CRPC tumors in which ENZ was treated alone harbored the same AR T878A mutation as the vehicle control–treated 16D^CRPC tumors (Supplementary Fig. S7C and D), whereas the ENZ-treated 30C^ENZR xenografts exhibited both T878A and F877L mutation, similar to the 30C^ENZR cells (Supplementary Fig. S7E). We examined the expression of AR-V7 in the 16D^CRPC xenograft tissues by qRT-PCR, and found that AR-V7 expression also did not change after the development of ENZ resistance in the ENZ-treated 16D^CRPC tumors (Supplementary Fig. S7F), which is consistent with the result in the ENZ-resistant cell lines including 30C^ENZR (Supplementary Fig. S3F and S3G). To determine whether the level of BCL-2 or IKKB correlates with the emergence of ENZ resistance, we measured the expression level of BCL-2 and IKKB by both qRT-PCR and IHC assays among the xenograft tissues from the animals that develop ENZ resistance. For BCL-2, the mRNA levels in ENZ-resistant tissues were slightly upregulated, compared with the vehicle control tissues (Supplementary Fig. S7G), and the BCL-2 protein levels measured by IHC did not change (the HS values in all tested samples were defined as zero; data not shown). However, we observed an increased level of IKKB mRNA in 4 of 5 animals with ENZ resistance (Supplementary Fig. S7G), and a strong uptrend for the IKKB protein level, determined by IHC for cytoplasmic IKKB expression, in the ENZ-resistant tumors (mean HS 250; 100.0–300.0), compared with the vehicle controls (mean HS 166.7; 0.0–200.0; Supplementary Fig. S7H), consistent with the protein level of IKKB in cell lines (Fig. 3A). These in vivo data suggest that the inhibition of BCL-2 and IKKB disrupts the emergence of de novo ENZ resistance potentially driven by the upregulation of IKKB.

To further clarify the underlying mechanism of sensitivity to BCL-2 inhibitor in ENZ-resistant cell lines, we assessed the expression level of multiple BCL-2 family members in the ENZ-sensitive cell line 16D^CRPC, and the ENZ-resistant but ABT199-sensitive cell lines including 40C^ENZR and 30C^ENZR, as well as the parental cell line LNCaP. To determine which of the BCL-2 family proteins may be involved in the ABT-199–induced cell death, we measured the protein level of multiple BCL-2 family members, including BH3-only proapoptotic sensors, prosurvival proteins, and apoptotic effectors, with and without the treatment of 10 μmol/L ABT-199 (Fig. 5A). Intriguingly, the levels of BCL-2, BAK, and BAD were greater in untreated 40C^ENZR and 30C^ENZR cell lines compared with LNCaP and 16D^CRPC cells, with significantly higher levels of BCL-2 (P < 0.01), BAD (P < 0.001), and BIM (P < 0.001) observed in the 30C^ENZR cell line compared with all other lines, supporting their increased sensitivity to ABT-199 (Fig. 5B). These results suggest that BCL-2, BAD, and BIM are potentially responsible for the sensitivity to ABT199 in the ENZ-resistant cell lines.

To examine the clinical utility of these findings, we explored whether the expression level of BCL-2 and IKKB was induced by AR targeting therapy in a cohort of 20 patients with prostate cancer who developed castration-resistant disease with paired archival (castration-sensitive) and metastatic castration-resistant tissue biopsies (Supplementary Table S4). HS were determined by IHC for cytoplasmic BCL-2 and IKKB protein expression in all patient biopsies (Fig. 6A and B). Cytoplasmic BCL-2 expression did not change significantly (P = 0.11) as patients progressed from CSPC (median; IQR: HS 0.0; 0.0–23.8) to CRPC (HS 0.0; 0.0–12.5; Fig. 6C). In contrast, cytoplasmic IKKB expression increased significantly (P = 0.003) as patients progressed from CSPC (median; IQR: HS 60.0; 22.5–100.0) to CRPC (HS 110.0; 80.0–160.0), and in 15 of 20 cases the IKKB level of CRPC was increased compared with the matched CSPC ones (Fig. 6D). We further classified the CRPC samples into three subgroups based on the treatment followed by biopsy collection: post-AA, post-ENZ, and post-ENZ plus AA. For additional mechanisms known to cause ENZ resistance, we tested the nuclear expression of AR-V7, and observed that the expression pattern of BCL-2, IKKB, and AR-V7 showed no difference between these three subgroups (Fig. 6E). Furthermore, we did a correlation analysis between nuclear AR-V7 and cytoplasmic IKKB expression, but not finding any correlation (Spearman correlation coefficient r = −0.06, P = 0.82; Fig. 6F). These results suggest that, with the development of AA and/or ENZ resistance in patients, the cytoplasmic level of IKKB is enhanced, which is consistent with our in vitro data showing that the protein level of IKKB is much higher in ENZ-resistant cell lines (Fig. 3A).

There is a pressing need to identify pathways that contribute to prostate cancer disease progression during and after effective AR blockade with drugs such as AA and ENZ. In this article, we initially modeled resistance with LNCaP derivative cell lines that mimicked the majority of patients who relapse on AA and ENZ with an unchanged non-neuroendocrine phenotype that continues to produce PSA. Herein, we have shown that inhibition of either BCL-2 or IKKB is not only a strategy to selectively kill ENZ-resistant prostate cancer cell lines and xenografts, but also a potential approach to prevent the emergence of ENZ-resistance in xenografts. We found that although the increased sensitivity to IMD0354 in ENZ-resistant cell lines is more modest (IC₅₀: 0.20–0.29 μmol/L vs. 0.69 μmol/L of 16D^CRPC), compared with ABT199 (IC₅₀: 0.49–0.82 μmol/L vs. 34.86 μmol/L of 16D^CRPC), it still met the criteria to identify hits in the drug screen (IC₅₀ ratio Mean_ENZR/16D^CRPC < 0.5; Supplementary Fig. S4E). Furthermore, we have identified that while overexpression of either BCL-2 or IKKB can lead to partial ENZ resistance in cells that were previously sensitive, the overexpression of both can fully mimic ENZ resistance. Also, our results of BCL-2 profiling have suggested that several key members of BCL-2 family define the increased sensitivity to ABT-199 in ENZ-resistant cell lines.

Interestingly, our in vivo data suggested that the BCL-2 and IKKB inhibitors, especially IMD0354, appeared to induce the growth of 16D^CRPC-derived xenografts when they were administered alone, compared with the vehicle control (Fig. 1G). This was not consistent with our in vitro results, such as Alamar Blue assays (Fig. 1A–D), colony formation assays in soft agar (Supplementary Fig. S5D and E), and knockdown experiments (Fig. 2B–E), although is likely not relevant to clinical care as other in vivo experiments in this study showed that ABT199 or IMD0354 combined with ENZ could treat the acquired ENZ resistance (Fig. 1H) and prevent the development of ENZ resistance (Fig. 4), suggesting that the combination of ENZ and ABT199 or IMD0354 may have therapeutic potential.

Current publications have documented multiple AR-related mechanisms of ENZ resistance, one of which is the point mutation F877L (13, 14). Our sequencing data identified the same mutation in all the ENZ-resistant cell lines including 34B^ENZR, 40C^ENZR, and 30C^ENZR. However, in our xenograft experiment which was similar to the original serial passaging process that generated the 34B^ENZR, 40C^ENZR, and 30C^ENZR cell lines, we did not find that this mutation appeared in any xenograft samples obtained from the relapsed 16D^CRPC tumors after ENZ treatment (Supplementary Fig. S7D). These results show that F877L mutation is not universally present as a mechanism of ENZ resistance, confirming the heterogeneity of underlying mechanisms of ENZ resistance. Other studies from clinical materials have also indicated that F877L is not the only driving mutation of ENZ resistance (15, 16).

IKKβ is one of two catalytic subunits that makes up the Inhibitor of Nuclear Kappa B Kinase (IKK) complex. The complex includes another catalytic subunit, IKKα, and a regulatory subunit, NEMO (IKKγ). The major role of this kinase complex is to phosphorylate and inhibit the activities of the Inhibitor of Nuclear Kappa B protein (IκB) family. The IκB family serves to inhibit the activity of Nuclear Factor Kappa B (NFκB) proteins by forming heterodimers and sequestering them in the cytoplasm. Therefore, IKKs serve as upstream activators of NFκB signaling. The role of NFκB in prostate cancer progression to a castration-resistant state has been well documented (17–19). NFκB functions as a master transcription factor that activates inflammatory cytokines/chemokines and underpins the synthesis of genes implicated in cell survival and chemoresistance, angiogenesis, and localized invasion (20–23). With ENZ resistance in particular, NFκB2/p52 has been shown to regulate AR splice variants and upregulate glucose metabolism as means to mediate resistance (24). In our study, however, NFκB signaling was not important in the resistance phenotype (Supplementary Fig. S6E). The use of various inhibitors included in our compound library, such as Ro 106-9920, pyrrolidine dithiocarbamate and dimethylamino parthenolide, did not selectively target ENZ-resistant cell lines (data not shown). Therefore, the effect of IKKβ inhibition in our study may be through an NFκB-independent manner. While IκBs have been the traditional substrates for IKKβ, there is growing evidence that IKKβ targets a variety of substrates through NFκB-independent mechanisms. Some of these substrates include β-catenin, IRS-1, Dok1, TSC1, cyclin D1, and Aurora A (25). Given the evidence that knockdown of IKKB sensitize ENZ-resistant cell line 40C^ENZR to ENZ (Supplementary Fig. S6D), it may be possible that IKKΒ expression in prostate cancer can activate one or more of these alternate pathways leading to ENZ resistance. For example, the overexpression of Aurora A signaling has been reported to drive the emergence of ENZ-resistant neuroendocrine prostate tumors (26, 27).

The expression and involvement of BCL-2 in prostate cancer and its association with the emergence of CRPC has been documented since the early 1990s (28–30). BCL-2 protein levels are often increased in tumor cells following androgen ablation therapy (31, 32), and likely act to confer an apoptosis-resistant phenotype in the malignant population (33–35). Using RNA sequencing analyses to test ENZ-resistant cells generated by a xenograft-based model, Li and colleagues also reported similarly that BCL-2 is highly upregulated in ENZ-resistant prostate cancer, and that ABT199 can repress the tumor growth of ENZ-resistant xenografts (36). Relevant to this finding, one of the strengths of our study has been the ability to validate our findings in two clinical cohorts. In the larger cohort from the Royal Marsden, surprisingly, we detected no upregulation in BCL-2 in the presence of AA/ENZ compared with the hormone-sensitive samples using standard IHC, which may imply that our preclinical findings have limited clinical relevance or limitations of our assay in small tissue samples. Current clinical trials exploring the utility of combining ENZ with venetoclax (NCT 03751436) are quantifying this benefit. The IKKB expression levels in post-AA/ENZ treatment specimens exhibited a significant upregulation compared with the pretreatment CSPC counterparts (Fig. 6D), consistent with the in vitro data in the ENZ-resistant cell lines. The reason for the dissociation between the preclinical and clinical findings with BCL-2 reinforces the importance of validated preclinical findings in human tissue prior to embarking on clinical trials with targeted therapies.

Further research is required to understand the codependency of IKKB and BCL-2 in ENZ resistance. Previous reports demonstrated that BCL-2 is transcriptionally regulated by the NFκB pathway in prostate cancer (37), suggesting that BCL-2 is a downstream target gene of NFκB signaling. However, we found that IKKB-overexpressing 16D^CRPC cell line had a similar sensitivity to ABT199, compared with the parental 16D^CRPC cells, and unlike ABT199-induced apoptosis, the type of cell death triggered by IMD-0354 remained uncharacterized (data not shown). Given that IKKB may be acting in a NFκB-independent manner, we hypothesize that BCL-2 and IKKB may represent two independent mechanisms in the development of ENZ resistance.

Although multiple pathways likely converge to drive the emergence of ENZ resistance, understanding the contribution of BCL-2 and IKKB to this disease is critical for better implementation of current therapies for CRPC and may have implications across the disease spectrum, in particular in de novo metastatic disease where outcomes remain inadequate despite recent therapeutic advances. Taken together, data from our drug screen and the accompanying in vitro/vivo validation implicates BCL-2 and IKKB dependencies in ENZ-resistant prostate cancer, suggesting their potential as novel therapeutic agents for the treatment of the disease, both of which require further validation and verification in larger human cohorts.

A. Sharp reports other from Roche-Genentech and Astellas outside the submitted work. J.S. de Bono reports grants, personal fees, and nonfinancial support from AstraZeneca, Genentech/Roche, Bayer, MSD, Merck Serono, Daiichi Sankyo, Pfizer, and Sanofi Aventis; nonfinancial support from Astellas; grants and nonfinancial support from Harpoon; and personal fees from Amgen and Constellation outside the submitted work. B.G. Wouters reports grants from Canadian Institutes of Health Research and Terry Fox Research Institute during the conduct of the study; B.G. Wouters also reports personal fees from Northern Biologics and Versant Ventures, and grants from Northern Biologics outside the submitted work. A.M. Joshua reports grants from U.S. Department of Defense and Prostate Cancer Canada during the conduct of the study, and grants and other from Astellas outside the submitted work. No disclosures were reported by the other authors.

Y. Liang: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Jeganathan: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Marastoni: Conceptualization, resources, data curation, software, formal analysis, validation, methodology, writing–original draft, writing–review and editing. A. Sharp: Resources, data curation, formal analysis, validation, methodology, writing–original draft, writing–review and editing. I. Figueiredo: Resources, data curation, formal analysis, validation, methodology, writing–original draft. R. Marcellus: Conceptualization, resources, data curation, software, methodology. A. Mawson: Conceptualization, resources, data curation, software, methodology. Z. Shalev: Resources, data curation, methodology. A. Pesic: Resources, data curation. J. Sweet: Resources. H. Guo: Resources, software, formal analysis. D. Uehling: Conceptualization, resources, software, formal analysis, methodology. B. Gurel: Conceptualization, resources, investigation, methodology. A. Neeb: Resources, investigation. H.H. He: Resources, funding acquisition, investigation. B. Montgomery: Resources, funding acquisition, investigation. M. Koritzinsky: Resources, funding acquisition, writing–review and editing. S. Oakes: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. J.S. de Bono: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. M. Gleave: Conceptualization, resources, funding acquisition, writing–review and editing. A. Zoubeidi: Resources, writing–review and editing. B.G. Wouters: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. A.M. Joshua: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing.

The work in this grant was supported by “Functional Assessment and Characterization of MDV3100-Resistant Cell Lines,” Proposal Log Number PC110998, Award Number W81XWH-12-1-0239; Prostate Cancer Canada-Movember Team T2103-1; Rising Star in Prostate Cancer Research 2013 RS2013-58; Stanley Tessis Prostate Cancer Research Fund; and Hold' em for Life Prostate Cancer Research Fund (to A.M. Joshua).

The Oakes Laboratory is supported by Cancer Council NSW, Cue Clothing Co., and the Mostyn Family Foundation.

The de Bono laboratory is supported by research funding from Movember, Department of Defense, Prostate Cancer UK, Cancer Research UK, an Experimental Cancer Medicines Centres (ECMC) grant, Prostate Cancer Foundation, and the National Institute for Health Research (NIHR) Biomedical Research Centre at the Royal Marsden NHS Foundation Trust (RMH) and the Institute of Cancer Research (ICR), London. A. Sharp has been supported by the Medical Research Council (MR/M018618/1), the Academy of Medical Sciences/Prostate Cancer UK (SGCL15), and a Prostate Cancer Foundation Young Investigator Award. The views expressed are those of the authors and not necessarily those of the NIHR or Department of Health and Social Care.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.