Attenuation of SRC Kinase Activity Augments PARP Inhibitor–mediated Synthetic Lethality in BRCA2-altered Prostate Tumors

Metastatic castration-resistant prostate cancer (mCRPC) is lethal and incurable (1). Treatment options for patients with mCRPC have expanded significantly, but de novo and acquired resistance occur frequently. Large-scale genomics studies (2, 3) have revealed a high degree of genomic instability (4) and frequent alterations in DNA damage repair (DDR) genes (2, 5) in patients with mCRPC. Alterations in BRCA2, a hallmark DDR and cancer susceptibility gene, are also prevalent in men with advanced prostate cancer (6, 7), including homozygous and heterozygous BRCA2 deletions (8). BRCA2 alterations, especially loss-of-function mutations, have been observed in a higher proportion of men with mCRPC and are associated with a worse prognosis (5, 6). In the PROREPAIR-B mCRPC cohort, BRCA2 germline mutations were reported to negatively affect outcome (9). However, the molecular mechanisms through which BRCA2 loss accelerates prostate cancer progression are poorly understood.

A significant fraction of localized prostate cancer and mCRPC harbors BRCA2 deletions (10). BRCA2 is frequently codeleted with RB1, and codeletion was significantly enriched in mCRPC and associated with higher genomic instability. BRCA2/RB1 codeletion impaired DNA repair, induced castration resistance, and augmented an epithelial-to-mesenchymal transition–like aggressive phenotype (10). Beltran and colleagues reported BRCA2 deletions in 30%–50% of mCRPCs and in TP53/RB1-deficient neuroendocrine prostate cancer (8). They also showed that alterations in DDR genes (primarily BRCA2) were significantly associated with poor overall survival (P = 0.001; ref. 8). These data suggest that BRCA2 deletion leads to aggressive prostate cancer and indicate that the underlying signaling pathways need to be investigated.

PARP inhibitors (PARPis) cause single-strand break accumulation (11) and synthetic lethality in DDR-impaired tumors because of mutated or deleted BRCA1/2 (12). PARPis are approved for the treatment of ovarian cancer, breast cancer, and prostate cancer (13). In a phase II trial in 49 patients with mCRPC, 16 (33%) showed a significant response to olaparib. Importantly, 88% of those who responded to olaparib harbored homologous recombination repair defects due in large part to BRCA2 and ATM alterations (14). The landmark TRITON2, TOPARP-B, and PROfound 3 trials showed that germline and somatic BRCA2 alterations were associated with increased response to olaparib and rucaparib (15–18). Recent data have also shown that patients with loss of BRCA1/2 in multiple cancers, including prostate cancer, experienced clinical benefit with PARPis (19). In TRITON2, patients with BRCA1/2 alterations treated with rucaparib had a 52% PSA response rate and a 43.9% response rate in patients with measurable disease. The duration of response was ≥6 months in 15 (56%) of the 27 patients, with confirmed objective responses (20). However, resistance is common, resistance mechanisms are poorly understood (21), and treatment modalities for patients resistant to PARPis are limited. Combination strategies that are well tolerated and have a synergistic effect are needed.

Increasing evidence has connected oncogenic nonreceptor tyrosine kinase SRC activity with aggressive prostate cancer (22). Phosphoproteomic analysis of mCRPC showed that SRC is one of the most activated kinases (23). Src knockout reduced primary tumor growth and metastasis (24). Although SRC activation is known to be required for checkpoint recovery termination and SRC inhibition delays G₂-phase DNA damage checkpoint recovery following DNA double-strand break (DSB) repair (25, 26), the role of SRC in DDR is incompletely understood.

We analyzed the transcriptomic profile of BRCA2-deleted mCRPC, and, for the first time, demonstrated that SRC pathway activation is directly associated with BRCA2 loss. We observed strong SRC activation and increased dasatinib sensitivity in BRCA2-knockout prostate cancer cells. We demonstrated that the combined use of PARPis and SRC inhibitors in BRCA2-deleted prostate cancer cell lines has a synergistic effect on cell viability. We found that SRC knockdown increases PARPi sensitivity, indicating that SRC activation may be important in PARPi resistance. Our study suggests that there may be significant promise in exploring combined inhibition of PARP and SRC in BRCA2-altered mCRPC, as well as in TP53/RB1-altered neuroendocrine prostate cancer.

The association between BRCA2 genomic deletion (heterozygous and homozygous) and disease progression in various prostate cancer clinical cohorts was analyzed in cBioPortal (27, 28). The transcriptomic profile of tumors with BRCA2 alteration was generated in cBioPortal and pathway analysis for oncogenic and hallmark signature was performed using gene set enrichment analysis (GSEA). Ten prostate cancer cohorts were used in this study (Supplementary Table S1). Graphs and Kaplan–Meier survival curves were plotted using GraphPad prism (version 7).

Cells were purchased from the American Type Culture Collection (ATCC), unless otherwise specified in Supplementary Table S2. LNCaP, DU145, 22RV1, and TRAMP-C2 were purchased from the ATCC. PC3M cells were provided by Raymond C. Bergan (Knight Cancer Institute, Oregon Health & Science University, Portland, OR). Cells were cultured in RPMI1640 medium (LNCaP, 22RV1, and PC3M) or DMEM (DU145 and TRAMP-C2) supplemented with 10% FBS, 2 mmol/L l-glutamine, and 1 × antibiotic/antimycotic (Gemini Bio-Products) at 37°C in 5% CO₂. The LNCaP-abl cell line was provided by Zoran Culig (Innsbruck Medical University, Innsbruck, Austria) and was maintained in phenol red–free RPMI1640 media supplemented with 10% CCS, 2 mmol/L l-glutamine, and 1 × antibiotic/antimycotic. RWPE1 cells were obtained from the ATCC and cultured in Keratinocyte Serum-free Medium (Thermo Fisher Scientific) at 37°C in 5% CO₂. Prostate organoids derived from patients with mCRPC were provided by Dr. Yu Chen [Memorial Sloan Kettering Cancer Center (MSK), New York, NY] and cultured as described previously (29). Cells were acquired between 2017 and 2019; in general, there were between two and six passages between collection and thawing. Cells were authenticated by human short tandem repeat profiling at the MSK Integrated Genomics Operation Core Facility (New York, NY) in December 2017 and June 2019. Mycoplasma testing was performed at the MSK Antibody & Bioresource Core Facility (New York, NY) using the MycoAlert PLUS assay in January 2019. CRISPR/Cas9-mediated BRCA2 knockout was performed in LNCaP as described previously (10). Guide RNA (gRNA) sequences are shown in Supplementary Table S2. Single-cell–derived BRCA2-knockout LNCaP clones were analyzed by Sanger sequencing at the MSK Gene Editing & Screening Core Facility (New York, NY).

Cells were washed with Hank's Balanced Salt Solution and lysed in RIPA (50 mmol/L TRiS-HCl, pH 7.4, 150 mmol/L NaCl, 1 mmol/L EDTA, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% SDS) supplemented with Protease and Phosphatase Inhibitors (Thermo Fisher Scientific). Protein concentrations were measured using the Bradford protein assay. Western blot analysis was performed using specific antibodies (Supplementary Table S2). Freshly prepared cell lysates were used for BRCA2 Western blot analysis, as described previously (10).

RNA isolation was performed using the Direct-zol RNA Kit (Zymo Research) and reverse transcribed with qScript cDNA SuperMix (Quantabio). cDNA corresponding to approximately 10 ng of starting RNA was used for one reaction. qPCR was performed with TaqMan Gene Expression Assay (Applied Biosystems). GAPDH was used as an internal control. TaqMan probe sequences are shown in Supplementary Table S2.

Cell viability was measured by MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide; Invitrogen) assay. Cells were plated in 96-well plates in complete media (2.5 × 10³ cells for LNCaP, 22RV1, LNCaP-abl, and RWPE1; 1 × 10³ cells for PC3M; and 5 × 10³ cells for MSKPCa1 and MSKPCa3). 96-well plates were precoated with 50 μg/mL Collagen I (rat tail; Thermo Fisher Scientific) for MSKPCa1 and MSKPCa3 cells. Cells were treated with DMSO or indicated inhibitors at days 0 and 3, unless otherwise mentioned. After indicated times, cells were incubated in 0.5 mg/mL MTT for 2 hours at 37°C. MTT crystals were dissolved in isopropanol and absorbance was measured in a plate reader at 570 nm.

PC3M and LNCaP-abl cells were treated with PARPis (olaparib and talazoparib) and dasatinib alone or in combination for 5 days. Drugs were added to the cells at days 0 and 3. The combination index–isobologram equation was used to quantitatively calculate the drug interaction. The Chou-Talalay method (CompuSyn) is a standardized definition for synergy in drug combinations and is based on the median-effect equation. The different identifications are based on the combination index. The dose–response and combination index–effect plots show the combinational effect of indicated drugs on cell proliferation, where combination index < 1 indicates synergism, combination index = 1 is additive, and combination index > 1 indicates antagonism (30).

Cells were treated with dasatinib (0.3 μmol/L) for 24 hours. DNA damage was quantified using MUSE Multi-color DNA Damage Kit (Luminex; MCH200107) in Guava MUSE cell analyzer system according to the manufacturer's instructions. Experiments were performed in triplicate.

1105× LNCaP scrambled and BRCA2-knockout (pooled population and single-cell–derived clones) cells were incubated in media supplemented with dasatinib (0.3 μmol/L) for 24 hours. Comet assay was performed using the Comet Assay Kit (Abcam; ab238544) according to the manufacturer's instructions. Photographs were captured under fluorescence microscopy.

Male wild-type (WT) C57BL/6J mice and C57BL/6-Tg(TRAMP) mice were purchased from The Jackson Laboratory. Normal prostates (from WT C57BL/6J mice) and localized prostate tumors [from C57BL/6-Tg(TRAMP) mice] were isolated from 30-week-old mice and fixed in 4% paraformaldehyde, followed by paraffin embedding, sectioning, and staining using indicated antibodies. All animal experiments were performed at the MSK Research Animal Resource Center Core Facility (New York, NY) according to the Institutional Animal Care and Use Committee protocol.

Organoid assays were performed as described previously (31). Briefly, PC3M, LNCaP-abl, MSKPCa1, and MSKPCa3 cells were detached using Accutase (Innovative Cell Technologies), collected using 70-μm cell strainers, counted (1 × 10³ cells/well for PC3M cells and 5 × 10³ cells/well for LNCaP-abl, MSKPCa1, and MSKPCa3 cells), and resuspended in prostate organoid media (29) and mixed with Matrigel Membrane Matrix (Thermo Fisher Scientific, CB-40234C) in a 1:1 ratio. The cell and Matrigel mixtures were plated on ultra-low attachment plates and allowed to grow for the indicated times. Detailed representations of the organoid experiments are shown in Supplementary Fig. S6A. Organoids were counted and photographed using GelCount Colony Counter (Oxford Optronix).

Results are reported as mean ± SD, unless otherwise noted. Comparisons between groups were performed using an unpaired two-sided Student t test, unless otherwise noted. Graphs were generated with GraphPad Prism (version 7.0 GraphPad Software, Inc).

To identify the frequency of BRCA2 alterations in prostate cancer, we analyzed data from cBioPortal for Cancer Genomics (27, 28). We observed that BRCA2 is frequently deleted in localized prostate cancer, mCRPC, and neuroendocrine prostate cancer, with heterozygous/shallow deletions more frequent than homozygous/deep deletions (Supplementary Fig. S1A). Biochemical recurrence was more common in patients with BRCA2 deletions than in patients with WT BRCA2 (P = 9.35e-7 and Q = 1.964e-5; Supplementary Fig. S1B). The Gleason scores of patients with BRCA2 deletions were significantly higher than the Gleason scores of patients with WT BRCA2 (P = 1.902e-5 and Q = 3.551e-4; Supplementary Fig. S1C). The proportion of metastatic cases was significantly higher in patients with homozygous or heterozygous BRCA2 deletions than in patients with WT BRCA2 (P < 10e-10 and Q < 10e-10; Supplementary Fig. S1D). These data suggest that BRCA2 deletion may be associated with aggressive prostate cancer.

To investigate the signaling pathways associated with BRCA2-deleted mCRPC, we used publicly available data from the Kumar and colleagues' cohort (32). BRCA2 deletions were found in 56% of mCRPC cases (Fig. 1A). We compared gene expression between patients with homozygous and heterozygous deletions of BRCA2 and patients with WT BRCA2 (Fig. 1B; Supplementary Table S3). We performed GSEA and found increased expression of several important oncogenic signaling pathways (e.g., E2F3, KRAS, and EZH2) and significant enrichment of the oncogenic tyrosine kinase SRC signaling pathway [normalized enrichment score (NES) = 1.8; P = 0.003; FDR = −0.05; Fig. 1C and D; Supplementary Table S3]. We also performed GSEA comparing cases with BRCA2 homozygous deletion with cases with WT BRCA2, cases with BRCA2 heterozygous deletion with cases with WT BRCA2, and cases with BRCA2 homozygous deletion with cases with BRCA2 heterozygous deletion (Supplementary Fig. S1E–S1G). We observed SRC signaling pathway enrichment in cases with homozygous BRCA2 deletion compared with cases with WT BRCA2 (NES = 1.3; P = 0.046; FDR = 0.073; Supplementary Table S3) and in cases with heterozygous BRCA2 deletion compared with cases with WT BRCA2 (NES = 1.55; P = 0.003; FDR = 0.02; Supplementary Table S3). The SRC pathway was not enriched when cases with homozygous and heterozygous BRCA2 deletions were compared (Supplementary Table S3).

We also performed the hallmark pathway analysis using GSEA and found that attenuation of androgen signaling was associated with BRCA2-deleted mCRPC (Fig. 1E and F; Supplementary Table S3). To identify drug targets associated with BRCA2-deleted mCRPC, we performed ToppGene suite analysis (33) of the upregulated transcriptome. The potent SRC inhibitor, dasatinib, was one of the top drug targets associated with the BRCA2-deleted mCRPC transcriptome (Fig. 1G). Our data suggest that SRC pathway activation may be associated with BRCA2-deleted mCRPC.

To investigate the direct effect of BRCA2 loss on SRC pathway activation, we used CRISPR/Cas9 to generate BRCA2-knockout LNCaP cells, a hormone-dependent human prostate cancer cell line with WT BRCA2 (10). LNCaP cells transfected with nontargeting/scrambled gRNA served as a control, as described previously (10). We confirmed BRCA2 loss in BRCA2 gRNA–transduced cells and observed increased SRC phosphorylation at tyrosine-416 (Y416) in cells transduced with all gRNAs targeting BRCA2 compared with cells transduced with the control (scrambled) gRNA (Fig. 2A; Supplementary Fig. S2A). SRC phosphorylation at Y416, which is in the activation loop of the kinase domain, is a key step leading to high SRC activity (34). We also assessed whether BRCA2 loss induced activation of other signaling pathways by Western blotting, particularly those pathways that we found to be enriched in BRCA2-null mCRPC (Fig. 1C and D). We did not observe increased phosphorylation of ERK (KRAS pathway) and JNK or induction of EZH2 in BRCA2-knockout cells compared with control cells (Fig. 2B). We did observe activation of the ATR-ATF2 pathway, indicating the induction of DNA damage in BRCA2-null LNCaP cells. Our data also showed increased phosphorylation of FAK, a downstream target of SRC (35), in BRCA2-knockout cells compared with control cells, further indicating that the SRC signaling pathway is activated in BRCA2-null prostate cancer cells (Fig. 2B). We also observed induction of AKT phosphorylation (PTEN pathway) in BRCA2-knockout cells relative to control cells.

To demonstrate that BRCA2 loss induces SRC phosphorylation in the absence of any exogenous factors, we assessed SRC phosphorylation in cells grown in serum-free media and then treated with EGF. Levels of SRC phosphorylation were higher in the serum-starved BRCA2-knockout cells than in the control cells, even without EGF treatment (Supplementary Fig. S2B). Increased SRC phosphorylation was observed in BRCA2-null LNCaP-abl, DU145, and PC3M cells, which harbor BRCA2 heterozygous deletions (10) and low BRCA2 mRNA levels (Supplementary Fig. S2C), compared with LNCaP cells (Fig. 2C). RNAi-mediated knockdown of SRC inhibited FAK phosphorylation in BRCA2-null LNCaP-abl cells, indicative of SRC-mediated FAK activation in these cells (Supplementary Fig. S2D). Our data indicate that loss of BRCA2 is sufficient to induce SRC activation in prostate cancer cells.

We next examined the effects of SRC inhibitor treatment on BRCA2-null prostate cancer cells. As expected, SRC-Y416 phosphorylation was reduced in BRCA2-null PC3M and LNCaP-abl cells treated with dasatinib (Supplementary Fig. S2E and S2F). However, treatment with dasatinib did not inhibit AKT phosphorylation in LNCaP-abl cells (Supplementary Fig. S2F). We observed that a pooled population of BRCA2-null LNCaP cells (LNCaP BRCA2-gRNA 2) exhibited enhanced sensitivity to dasatinib compared with cells transduced with a scrambled/nontargeting gRNA (Fig. 2D, top). We also generated single-cell–derived clones from BRCA2-knockout cells (LNCaP BRCA2-gRNA-cl7) and confirmed BRCA2 loss by CRISPR sequencing (WT BRCA2 elimination >99%; Supplementary Table S4). Dasatinib reduced the viability of LNCaP BRCA2-gRNA2-cl7 cells (Fig. 2D, top). There was no significant difference in cell viability with dasatinib treatment between the pooled cells and the single-cell–derived clones. Importantly, we observed that dasatinib also suppressed the elongated morphology of control cells, LNCaP BRCA2-gRNA 2 cells, and LNCaP BRCA2-gRNA-cl7 cells (Fig. 2D, bottom). Similarly, other SRC-specific inhibitors, bosutinib (SKI606) and saracatinib (AZD-0530), strongly reduced the viability of BRCA2-null LNCaP cells (LNCaP BRCA2-gRNA 2) compared with control cells (Fig. 2E). However, inhibition of AKT by ipatasertib did not have a significant inhibitory effect on cell viability in BRCA2-null LNCaP cells compared with scrambled LNCaP cells (Fig. 2F). These data indicate that inhibition of the SRC pathway reduces the viability and invasiveness of these prostate cancer cell lines.

We assessed whether increased levels of SRC phosphorylation were associated with poor outcomes in patients with localized prostate cancer. Increased SRC-Y416 phosphorylation was modestly, but not statistically significantly, associated with decreased disease-free survival in The Cancer Genome Atlas (TCGA) cohort of primary prostate cancer [P_trend = 0.0546; 95% confidence interval (CI), 0.9319–3.619; Fig. 2G]. We were unable to detect an association between levels of SRC-Y527 phosphorylation or levels of SRC and disease-free survival in the same cohort (Supplementary Fig. S2G and S2H). We did not detect a significant association between SRC-Y416 phosphorylation and Gleason grade (Supplementary Fig. S2I, top) or between BRCA2 deletion and SRC-Y416 phosphorylation levels (Supplementary Fig. S2I, bottom). To further understand the role of SRC activation in localized prostate cancer, we compared gene expression between patients with high SRC activation (top quartile; Fig. 2G) and those with low SRC activation (bottom quartile; Fig. 2G) and performed GSEA. We found enrichment of several oncogenic signaling pathways, including induction of the PIGF pathway (Fig. 2H). Previous studies have shown that placental growth factor/PIGF, a secreted stromal factor that can induce SRC activation (36), plays an important role in cancer progression (37). Taken together, these data suggest that SRC-Y416 phosphorylation may be associated with primary prostate cancer progression; however, additional oncogenic alterations may lead to SRC activation in primary prostate cancer.

Recently, PARPis have shown promise in patients with mCRPC harboring DDR pathway defects, in particular, but not exclusively, BRCA2-altered tumors, leading to FDA approval (14–18). Despite responses, resistance is common, and treatment modalities for patients resistant to PARPis are limited. Given these data, we hypothesized that combined inhibition of SRC and PARP could be a therapeutic option for patients with BRCA2 alterations.

To study the effects of dasatinib in combination with PARPis, we used CompuSyn synergism/antagonism analysis (30). Briefly, cells were treated with increasing concentrations of dasatinib and either olaparib or a PARPi that exhibits higher PARP1 trapping, talazoparib. We observed that dasatinib acted synergistically (combination index < 1) with both talazoparib and olaparib in BRCA2-null LNCaP-abl and PC3M cells (Fig. 3A and B; Supplementary Fig. S3A and S3B). Although the dasatinib–talazoparib combination was strongly synergistic in both LNCaP-abl and PC3M cells, the synergistic effect of the dasatinib–olaparib combination was stronger in PC3M cells than in LNCaP-abl cells (Fig. 3A and B; Supplementary Fig. S3A and S3B).

We also examined the effect of combination treatment on the viability of LNCaP-abl (0.3 μmol/L dasatinib and 0.03 μmol/L talazoparib alone or in combination) and PC3M cells (0.3 μmol/L dasatinib and 3 μmol/L olaparib alone or in combination). The combination of dasatinib with PARPis significantly reduced cell viability over treatment with either drug alone (Supplementary Fig. S3C and S3E). In particular, treatment of LNCaP-abl cells with dasatinib alone had no effect on cell viability, but the combination of talazoparib and dasatinib significantly reduced cell viability (Supplementary Fig. S3C).

We next assessed the effect of combination treatment on the viability of LNCaP BRCA2-gRNA 2 cells over time. We treated cells with 0.3 μmol/L dasatinib and 3 μmol/L olaparib, the concentrations that produced a synergistic effect on cell viability. Combination treatment of control/scrambled LNCaP cells reduced cell viability, but the effect was modest compared with olaparib treatment alone (Fig. 3C). We observed a significant reduction in the viability of LNCaP BRCA2-gRNA 2 cells treated with the combination of dasatinib and olaparib. Treatment with either drug alone only partially inhibited viability (Fig. 3C and D). Olaparib alone did not alter the invasive elongated morphology of LNCaP BRCA2-gRNA 2 cells (Fig. 3E). These data demonstrate that combination treatment inhibits cell viability and suppresses the invasive elongated morphology of BRCA2-knockout prostate cancer cells.

To further confirm the effect of dual inhibition of SRC and PARP on the viability of BRCA2-null prostate cancer cells, we treated LNCaP-abl cells with bosutinib (0.3 μmol/L) and saracatinib (0.3 μmol/L) alone and in combination with talazoparib (0.03 μmol/L) for 7 days (Fig. 3F; Supplementary Fig. S3D). We found that the combined inhibition of SRC and PARP strongly reduced the viability of LNCaP-abl cells over the inhibition of SRC or PARP alone (Fig. 3F; Supplementary Fig. S3D).

Recent clinical data have shown that patients with mCRPC with DDR mutations (particularly BRCA2) respond to platinum-based chemotherapy (38). To investigate whether inhibition of SRC increases the efficacy of platinum-based therapy, we treated cells (LNCaP-scrambled and LNCaP BRCA2-gRNA 2) with 0.3 μmol/L dasatinib and 0.1 μmol/L cisplatin for 5 days (Fig. 3G and H). We observed a strong reduction in the viability of LNCaP BRCA2-gRNA 2 cells treated with the combination of dasatinib and cisplatin, while treatment with either drug alone only partially inhibited viability (Fig. 3G and H). Taken together, our data suggest that SRC inhibitors can increase the growth inhibitory effect of both PARP and platinum-based therapy in BRCA2-null prostate cancer cells.

To test our hypothesis that dasatinib induces DNA damage, we treated PC3M cells with dasatinib for 24 hours and assessed DNA damage by phospho-γH2Ax and phospho-ATM as measured by flow cytometry. Dasatinib-treated PC3M cells had increased γH2AX phosphorylation relative to DMSO-treated cells, indicative of defective DSB repair (Fig. 4A). There was a modest increase in ATM phosphorylation in dasatinib-treated cells, indicating that SRC phosphorylation inhibition induces DNA damage (Fig. 4A).

To determine whether BRCA2 loss affected dasatinib-induced DNA damage, we treated LNCaP BRCA2-gRNA2-cl7 and control cells with dasatinib (0.3 μmol/L) for 24 hours. Dasatinib induced DNA damage in BRCA2-knockout cells, but not in control cells (Fig. 4B and C). We also examined DNA damage in dasatinib-treated cells using the comet assay, which directly measures DNA DSBs. We observed comet tails in both LNCaP BRCA2-gRNA2-cl7 and LNCaP BRCA2-gRNA 2 cells treated with dasatinib, whereas DMSO-treated cells did not have any significant comet tails (Supplementary Fig. S4A).

Dasatinib also induced DNA damage in RWPE1 cells, an immortalized benign human prostate cell line (Supplementary Fig. S4B). RWPE1 cells are BRCA2 WT (10) and express low levels of TP53 and RB1 proteins due to the presence of a single copy of human papillomavirus 18 (HPV18; ref. 39). We also examined the effect of dasatinib on RWPE1 cell viability. Only the combination of olaparib and dasatinib significantly reduced cell viability; there was minimal decrease with dasatinib alone (Supplementary Fig. S4C). These data suggest that the combination of dasatinib and PARPis may inhibit the growth of prostate cancer cells without alterations in canonical DDR genes (e.g., BRCA2), but with alterations in tumor suppressors, such as TP53 or RB1. Taken together, our data show that inhibition of SRC signaling by dasatinib induces DNA damage in prostate cancer cells, thereby increasing sensitivity to PARPis.

We extend our study and investigated the effects of combined inhibition of SRC and PARP on the growth of TP53/RB1-null prostate cancer cells. Loss of TP53 and RB1 is observed in advanced prostate cancer, and frequently observed in very aggressive neuroendocrine/neuroendocrine-like prostate cancer and is associated with poor outcome (8). To examine SRC activation in TP53/RB1-null neuroendocrine prostate cancer, we used localized prostate tumors from C57BL/6-Tg(TRAMP) mice; this genetically engineered mouse model expresses simian virus 40 early genes (large and small tumor antigens, Tag) in the prostate, resulting in abrogation of Trp53 and Rb1 and the rapid development of androgen receptor–negative aggressive and metastatic neuroendocrine prostate cancer (40). We observed strong SRC phosphorylation and increased DNA damage (phosphorylation of γH2AX) in the Tg(TRAMP) tumors relative to the normal prostate of C57BL/6J mice (Fig. 4D). We treated TRAMP-C2 cells [tumorigenic cell lines developed from 32-week-old localized prostate of a Tg(TRAMP) mouse; ref. 41] with dasatinib (0.3 μmol/L) or bosutinib (0.3 μmol/L) alone and in combination with talazoparib (0.03 μmol/L) for 5 days. There was a significant reduction of cell viability when treatment included SRC inhibitors, dasatinib or bosutinib, in combination with talazoparib compared with treatment with either dasatinib, bosutinib, or talazoparib alone (Fig. 4E and F). We also treated TRAMP-C2 cells with 0.3 μmol/L dasatinib and 0.1 μmol/L cisplatin, and observed a strong reduction in the viability of cells treated with the combination of dasatinib and cisplatin compared with treatment with either drug alone (Fig. 4G; Supplementary Fig. S4D). These data suggest that the combinatorial treatment of SRC inhibitors with PARPis or platinum-based chemotherapy strongly reduces viability over treatment with either SRC inhibitors or PARPis alone in Trp53/Rb1-null neuroendocrine prostate cancer.

We were unable to find any significant overlap between published SRC signaling genes and DDR genes (Supplementary Fig. S4E), indicating that further study to uncover the connection is warranted.

Although PARPi-based therapies are the mainstay for patients with BRCA1/2-mutated prostate cancer (14–16), responses are not durable; de novo and acquired PARPi resistance are major limiting factors of clinical therapy, although the mechanisms of resistance are not fully understood. We hypothesized that SRC activation is important in PARPi resistance in prostate cancer. We treated 22RV1 cells, which harbor a BRCA2 oncogenic mutation (T3033Nfs*11; ref. 42), with olaparib and dasatinib alone and in combination. Consistent with our previous report (10), 22RV1 cells had de novo relative resistance to olaparib (Fig. 5A). Combination treatment significantly decreased 22RV1 cell viability. Dasatinib alone had only a very modest effect.

We next performed a transient periodic experiment treating PC3M cells with dasatinib and talazoparib (Supplementary Fig. S5A). We first treated PC3M cells for 3 days with talazoparib or dasatinib alone. We then added dasatinib to the talazoparib-pretreated cells (and vice versa) and incubated the cells for 4 days. Cells treated with talazoparib or dasatinib alone or in combination for 6 continuous days were used as control. Dasatinib considerably reduced the viability of talazoparib-pretreated cells (Fig. 5B; Supplementary Fig. S5A). There was a similar reduction of cell viability in dasatinib-pretreated cells in the presence of talazoparib (Fig. 5B; Supplementary Fig. S5A). As expected, combination treatment of dasatinib and talazoparib for 6 continuous days significantly reduced cell viability relative to either drug alone (Fig. 5B; Supplementary Fig. S5A). However, there was no significant difference in viability of cells pretreated or continuously treated with the talazoparib/dasatinib combination (Fig. 5B; Supplementary Fig. S5A).

We also incubated PC3M cells in olaparib- or talazoparib-supplemented media. After 15 days, we counted the surviving cell population (PARPi resistant), replated, and treated with dasatinib with olaparib or talazoparib for 7 days (Supplementary Fig. S5B). Strikingly, the viability of PARPi-resistant cells was remarkably reduced in the presence of low concentrations of dasatinib (Fig. 5C and D). Dasatinib treatment only modestly reduced the viability of DMSO-pretreated cells (Fig. 5C and D). These data indicate that dasatinib restores PARPi sensitivity in PARPi-resistant prostate cancer cells.

We transiently knocked down SRC in PC3M cells and treated them with olaparib or talazoparib. PARPis selectively attenuated the viability of PC3M cells with decreased SRC expression (Fig. 5E). We next investigated whether SRC activation was directly involved in PARPi resistance in prostate cancer cells. We transiently overexpressed constitutively active SRC (Y527F SRC; ref. 43) or empty vector in PC3M cells. Expression of constitutively active SRC rescued the cell viability reduction in response to PARPis (Fig. 5F). Collectively, these data further demonstrate that SRC activation may be a mechanism of PARPi resistance in prostate cancer.

Recent data have shown that 3D organoids are a better model for understanding disease biology and testing drug efficacy in vitro (44). We treated 3D Matrigel organoids of PC3M cells with dasatinib or olaparib alone or in combination (Supplementary Fig. S6A). Combination treatment remarkably reduced the number and size of organoid colonies compared with treatment with either drug alone (Fig. 6A and B). Treatment of LNCaP-abl organoids with dasatinib and talazoparib produced similar results (Fig. 6C; Supplementary Fig. S6B). We also observed reduced PC3M invasion through Matrigel in dasatinib-treated organoids compared with DMSO- or olaparib-treated organoids (Fig. 6B).

We extended our study to include organoids derived from patients with mCRPC, which can be used as avatars of human cancer to study the molecular mechanisms of candidate genes and the effect of drugs (29). We have previously used FISH to assess BRCA2 status of these organoids; MSKPCa1 and MSKPCa3 have heterozygous loss of BRCA2, whereas MSKPCa2 is BRCA2 WT (10). All three patient-derived organoids exhibited high SRC activation compared with LNCaP cells (Fig. 6D). Interestingly, we observed high FAK phosphorylation only in the BRCA2-null organoids MSKPCa1 and MSKPCa3, but not in the BRCA2 WT organoid MSKPCa2 (Fig. 6D). We treated the BRCA2-null organoids with 0.3 μmol/L dasatinib and 0.03 μmol/L talazoparib alone or in combination in 2D on collagen-coated plates or in 3D Matrigel. We observed a strong reduction in the viability of organoids in 2D culture (Fig. 6E) and a statistically significant reduction in the number of organoid colonies with combination treatment relative to treatment with either drug alone (Fig. 6F and G; Supplementary Fig. S6C). Our results suggest that combined inhibition of SRC and PARP could be more effective than targeting PARP alone in patients with BRCA2-altered tumors.

Recent studies have revealed germline and somatic variants in DDR pathway components in a significant subset of patients with prostate cancer (2, 5, 6). Studies have also shown that patients with mCRPC harbor germline mutations in DDR genes, including BRCA2, more frequently than those with localized disease (6, 9). Patients with prostate cancer with germline BRCA1/2 mutations were found to have more aggressive disease and poorer survival than patients with WT BRCA1/2 (45). We have demonstrated previously that loss of even a single copy of BRCA2 results in a worse prognosis in prostate cancer (10). In addition to its function as a guardian of genomic stability during replication stress and maintaining genomic integrity, BRCA2 is a multifaceted tumor suppressor with numerous functions (46). Therefore, the discovery of the signaling pathways associated with BRCA2 loss is crucial for identifying therapeutic targets.

DDR is a complex, multilevel process involving several subpathways (47). Generally, mutations/deficiencies in one component can be compensated for by other components/genes (48), which may be exploited to develop novel therapeutic strategies, such as using PARPis to treat cancers with DDR alterations (49). Studies have shown the benefits of combining PARPis with other therapies in cell lines and clinical trials (50). There was a synergistic reduction of apoptosis and DNA damage in multiple prostate cancer cell lines treated with the histone deacetylase inhibitor, SAHA, and the PARPi, veliparib (51). Clinical trials combining olaparib and checkpoint inhibitors and antiandrogen therapies have demonstrated efficacy in patients with DDR defects (52). A phase I/II trial with olaparib and cediranib, an angiogenesis inhibitor, in advanced solid tumors, including mCRPC, is ongoing (NCT02484404). This indicates that combination therapies with PARPis may be beneficial for patients with mCRPC, particularly those with DDR alterations.

We prospectively investigated the signaling pathways deregulated by BRCA2 loss to identify novel therapeutic targets. We hypothesized that combining PARPis with inhibitors of activated signaling pathways would have a synergistic effect. We found that the SRC signaling pathway was activated in BRCA2-null mCRPC and increased SRC activation in BRCA2-null prostate cancer cell lines. We observed a synergistic reduction in cell viability in BRCA2-null prostate cancer cell lines, but not parental cells with WT BRCA2, treated with PARPi and dasatinib. We also showed that SRC activation may be involved in PARPis resistance in BRCA2-null prostate cancer cells. These data suggest the utility of combining dasatinib and PARPis in aggressive prostate cancer. We also found a remarkable reduction in the viability of BRCA2-null prostate cancer cells treated with dasatinib and cisplatin, indicating that dasatinib increases the efficacy of both PARPi and platinum-based chemotherapy.

Interestingly, we also observed that the combination of PARPis and SRC inhibitors resulted in a significant reduction in RWPE1 cell viability, even though this cell line has WT BRCA2 and an intact canonical DDR pathway. However, RWPE1 cells express significantly lower RB1 and P53 protein due to their expression of a single copy of HPV18 (39). We have also demonstrated that combination treatment with PARPis and SRC inhibitors reduces the growth of Trp53/Rb1-null androgen receptor–independent neuroendocrine prostate cancer. Our study is the first of its kind to combine SRC and PARP inhibition, and we hypothesize that this combination has great potential for patients harboring DDR alterations. We anticipate that this combination may also be effective in unselected patients without known DDR alterations whose tumors harbor alterations of classical tumor suppressors, RB1 and p53. However, this requires further testing to confirm the significance of the therapeutic combination beyond canonical DDR alterations.

SRC is a proto-oncogenic nonreceptor tyrosine kinase with roles in tumor cell proliferation, survival, and invasion (53). Inactivation of SRC signaling in response to DNA damage replication stress has been shown to suppress G₁–S-phase progression and maintain genomic stability (54), and induction of DNA damage has also been shown to induce SRC activation in breast cancer cells (55). We observed that inhibition of SRC by dasatinib induces defective DNA DSB repair. These data indicate that SRC activation is essential for DNA DSB repair and may be important in homologous recombination repair, which has not been reported previously. We also observed that BRCA2 loss induces SRC-Y416 phosphorylation, suggesting that increased DSBs by BRCA2 loss may activate SRC. Considering the complexity of the process, we believe that multiple mechanisms may also be involved in SRC-regulated DNA damage response.

SRC activation has been reported in several cancers, including prostate cancer (56). In a study that examined matched prostate tumor samples taken before hormone deprivation therapy and after relapse, 28% of castration-resistant tumors exhibited increased SRC activity (57). Patients with high SRC activity had significantly shorter overall survival (P < 0.0001; ref. 57). Induction of v-SRC alone is sufficient for oncogenic transformation of benign prostate cells and induces lung metastasis (58). SRC knockdown/inhibition reduces bone metastasis in a breast cancer xenograft model (59), and SRC knockdown inhibits the migration of prostate cancer cells in vitro (60). We observed that dasatinib suppressed elongated morphology of prostate cancer cells regardless of BRCA2 status and reduced invasion in 3D organoid culture, further indicating the association between SRC activation and invasive phenotype of cancer cells. Although alterations in the DDR pathway are far rarer in localized prostate cancer than in mCRPC (5), we did observe increased SRC activation (SRC-Y416 phosphorylation) in a fraction of localized prostate cancer cases from TCGA cohort, which may be associated with aggressive disease and independent from BRCA2 alteration status. Previous studies showed that activation of SRC induces androgen-independent growth of LNCaP cells and drives castration-resistant prostate cancer progression through an androgen receptor–dependent mechanism (61). These data indicate that activation of SRC may play an important role in primary prostate cancer progression and castration resistance, but further testing in a larger cohort and proteomic analyses are necessary to validate these findings.

Dasatinib (BMS-354825, Sprycel) is a commercially available, multiple tyrosine kinase inhibitor that inhibits SRC activation. Although the role of SRC in tumor progression and metastasis is well established, dasatinib monotherapy has not shown significant promise in solid tumors (62). Dasatinib has been shown to inhibit cell adhesion, migration, and invasion of prostate cancer cell lines (63). Dasatinib inhibited tumor growth and the development of lymph node metastases in both castration-sensitive and castration-resistant tumors. It was also shown to decrease proliferation and increase apoptosis in orthotopic nude mouse models (60, 62). Dasatinib suppressed disease progression in an intratibial xenograft model of PC3M cells (64). Mendiratta and colleagues reported that decreased predicted androgen receptor activity correlated with increased predicted SRC activity and sensitivity to dasatinib in androgen-sensitive LNCaP cells (65). Given these findings, dasatinib was evaluated in combination with docetaxel in the READY trial, a randomized phase III study of more than 1,500 unselected patients with mCRPC. There was no overall survival benefit in patients treated with docetaxel and dasatinib compared with docetaxel alone (66). However, dasatinib modestly prolonged time to skeletal events [P = 0.08; HR, 0.81 (64–1.02); ref. 66]. Similarly, previous preclinical observations showed that bosutinib and saracatinib also inhibited prostate cancer growth and metastasis in experimental mouse models (67, 68). These findings demonstrate that identifying patients with high SRC activation is necessary to personalize SRC inhibitor–based therapy for patients with mCRPC.

Resistance to PARPis reduces drug efficacy and worsens patient outcomes (69). Because BRCA2 is frequently deleted in prostate cancer, the mechanisms of resistance to PARPis in prostate cancer likely involve alternative molecular mechanisms, rather than reversion mutation. We observed that the combination of dasatinib and olaparib greatly reduced the proliferation of 22RV1 cells, which harbor oncogenic mutation of BRCA2, but exhibit resistance to PARPis. Moreover, the addition of dasatinib to PARPis in BRCA2-defective PC3M cells enhanced the reduction in cell viability and may rescue secondary resistance developed during prior treatment with PARPis. We also observed that SRC knockdown increases the antiproliferative effect of PARPis, whereas overexpression of constitutively active SRC leads to relative resistance to PARPis. Further study is needed to understand the molecular mechanism. Our data suggest that SRC activation may be a possible mechanism of PARPi resistance in prostate cancer; treatment with dasatinib, bosutinib, or saracatinib may overcome this resistance.

In conclusion, we identified high SRC activation in BRCA2-altered mCRPC. We found that the combined inhibition of SRC and PARP in BRCA2-altered prostate cancer cell lines and organoids had a synergistic effect on cell viability. For the first time, we demonstrate that SRC activation may be a potential mechanism of PARPi resistance and found that SRC inhibitors (e.g., dasatinib, bosutinib, and saracatinib) may overcome this resistance. These results suggest that combination inhibition of PARP and SRC should be explored in men with BRCA2-mutated mCRPC.

W. Abida reports grants and personal fees from Clovis Oncology, grants from AstraZeneca, Zenith Epigenetics, Epizyme, and Prostate Cancer Foundation, and personal fees from Janssen during the conduct of the study, and personal fees from Daiichi Sankyo, ORIC Pharma, and Roche outside the submitted work. M.J. Morris reports personal fees from ORIC Pharmaceuticals and Curium and other from Progenics, Roche, Janssen, Bayer, Sanofi, Endocyte, and Progenics outside the submitted work. L.A. Mucci reports grants from NIH and Janssen during the conduct of the study, other from AstraZeneca, and grants and other from Bayer, Merck, and Sanofi outside the submitted work. D. Danila reports research support from U.S. Department of Defense, American Society of Clinical Oncology, Prostate Cancer Foundation, Stand Up 2 Cancer, Janssen Research & Development, Astellas, Medivation, Agensys, Genentech, and CreaTV and is a consultant for Angle LLT, Axiom LLT, Janssen Research & Development, Astellas, Medivation, Pfizer, Genzyme, and Agensys. P.W. Kantoff reports personal fees from Bavarian Nordic, GE HealthCare, Janssen, Merck, Progenity, SynDevRx, Tarveda, Genentech/Roche, and OncoCell MDX, other from Cogent Biosciences, Mirati, Placon, XLink, and personal fees and other from Context Therapeutics, DRGT, and SEER outside the submitted work, as well as spouse reports research funding from Astellas, Janssen, AstraZeneca, Bayer, and Merck and is a paid consultant for Bayer. No disclosures were reported by the other authors.

G. Chakraborty: Conceptualization, data curation, supervision, methodology, writing-original draft, project administration, writing-review and editing. N. Khan Patail: Conceptualization, data curation, methodology, writing-original draft. R. Hirani: Data curation. S. Nandakumar: Formal analysis, writing-original draft, project administration. Y.Z. Mazzu: Data curation, methodology, writing-review and editing. Y. Yoshikawa: Data curation, methodology, writing-original draft. M. Atiq: Data curation, methodology, writing-original draft. L.E. Jehane: Data curation, writing-original draft. K.H. Stopsack: Formal analysis, writing-review and editing. G.-S.M. Lee: Methodology. W. Abida: Writing-review and editing. M.J. Morris: Formal analysis, writing-review and editing. L.A. Mucci: Formal analysis, writing-review and editing. D. Danila: Formal analysis, methodology, writing-review and editing. P.W. Kantoff: Conceptualization, supervision, writing-original draft, writing-review and editing.

This work was supported by a Department of Defense Prostate Cancer Research Program award (W81XWH1910470 to P.W. Kantoff), a Department of Defense Early Investigator Research Award (W81XWH-18-1-0330 to K.H. Stopsack), and Prostate Cancer Foundation Young Investigator Awards (to G. Chakraborty, K.H. Stopsack, W. Abida, and L.A. Mucci). P.W. Kantoff and L.A. Mucci were supported by a Prostate Cancer Foundation Challenge Award and by NCI grant 5P01CA228696-02. This work was supported, in part, by a grant from the NIH to Memorial Sloan Kettering Cancer Center (P30CA008748). We thank Ralph Garippa, Hsiu Yu Liu, and staff of the MSK Gene Editing and Screening Core (formerly the RNAi Core) for CRISPR design and CRISPR sequencing, Yu Chen (MSK) for the organoids, Cindy Lee of the MSK Human Oncology and Pathogenesis Program for organoid culture media, Mesruh Turkekul of the MSK Molecular Cytology Core Facility for IHC, and Sara DiNapoli and Amy Plofker (MSK) for editing.

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.