BMI1 Drives Metastasis of Prostate Cancer in Caucasian and African-American Men and Is A Potential Therapeutic Target: Hypothesis Tested in Race-specific Models

Metastasis is the major cause of mortality in prostate cancer patients (1, 2). The post-therapy recurrence of metastatic prostate cancer is a major impediment in the management of this lethal disease (3). Although often initially sensitive to androgen depravation, prostate cancer cells eventually become androgen insensitive and continue to thrive in the absence of androgenic steroids (4, 5). Metastatic prostate cancer cells usually have a relatively low mitotic rate, aberrant apoptotic machinery, and high genetic diversity, which render these cells resistant to chemotherapies, such as taxanes, which generally target highly proliferative tumor cells (6). Factors such as cancer cell stemness, tumor heterogeneity, genetic makeup, diverse immunobiology and race play critical roles in the clinical outcome of patients with prostate cancer to therapy (7). African-American prostate cancer patients reportedly exhibit greater rates of metastasis development, recurrence of disease, and metastasis-associated mortality than their Caucasian counterparts (8, 9). Studies affirm racial disparity as African-American prostate cancer patients exhibit worse clinical outcomes in response to taxane, hormone, or androgen deprivation therapy (ADT) relative to Caucasian patients (10, 11). Therefore, a need exists to develop prostate cancer therapies that are more effective in African-American patients, and ideally, among all races. Critical to this need is understanding of the molecular mechanisms that contribute to disease recurrence, chemoresistance, metastasis, and growth of metastatic prostate cancer tumors.

B-cell–specific Moloney murine leukemia virus integration site 1 (BMI1) stem cell factor is a component of the polycomb group complex 1 (PRC1), which functions through chromatin remodeling and is involved in embryonic development (12). We previously reported that BMI1 is a soluble protein that is secreted by metastatic prostate cancer cells and is detectable in the blood of patients with prostate cancer (13). We showed that BMI1 drives the proliferation of docetaxel-resistant metastatic prostate cancer cells by regulating several molecular pathways and suggested that BMI1 is a druggable target to treat metastatic prostate cancer in humans (13, 14). Efforts are being made by several investigators to develop chemical inhibitors targeting BMI1. Our laboratory is also pursuing this endeavor. Kresso and colleagues showed that PTC-209, a small molecule, has the potential to inhibit the expression of BMI1 (15). In this study, we tested the relevance of BMI1 molecule in the metastasis of prostate cancer in African-American and Caucasian patients. We tested the therapeutic efficacy of small-molecule inhibitor of BMI1 against localized and metastasis models of African-American and Caucasian prostate cancer disease in vitro and in vivo using xenograft mouse and zebrafish models.

Cell lines were purchased from ATCC (MDA-PCa2b, PC3, WBNB26) and MD Anderson Cancer Center (Houston, TX; PC3M). E006AA and E006AA-hT cell lines were obtained from Dr. Shahriar Koochekpour (Roswell Park Cancer Center, Buffalo, NY). RC-77N/E and RC-77T/E were provided by Johng S. Rhim (Uniformed Services University, Bethesda, MD). Cell lines were authenticated (for short tandem repeat analysis) and the report is attached as Supplementary Data file. PC3, PC3M-luc were cultured in RPMI1640 medium, whereas MDA-PCa2b cells were cultured in HPC1 + 20% FBS medium. WBNBA26 cells (purchased from ATCC) were grown in basic KFC medium.

Frozen and paraffin-embedded prostatic tissues were procured from NCI-sponsored Cooperative Human Tissue Network (CHTN, Columbus, OH). The metastatic tissues and tumor-RNA of patients with prostate cancer were obtained from (i) BioNet-University of Minnesota, (ii) and the University of Washington (Seattle, WA) and the Prostate Cancer Biorepository Network (Johns Hopkins University, Baltimore, MD). The metastatic tissues (Mets) were obtained from patients who died of mCRPC and who signed written informed consent for a rapid autopsy to be performed within hours of death, under the aegis of the prostate cancer Donor Program at the University of Washington.

Docetaxel and PTC-209 were purchased from Selleckchem (Houston, TX). The ChIP-grade anti-BMI1 antibody was obtained from Millipore. We used following antibodies: anti-BCL2, anti-BMI1, anti-mitochondrial protein, anti-PCNA, anti-Cyclin D1 (Cell Signaling Technology), anti-P16, anti-c-MYC (Abcam) and anti-VEGF (Santa Cruz Biotechnology). We used following plasmids pTK-TCF-Luc (Upstate Laboratories); pGL3-MMP2, pGL3-VEGF, pGL3-MYC (Addgene) and pGL3-BMI1 (gifted by Goberdhan Dimri, George Washington University, Washington, D.C.).

Transfections, luciferase reporter activity, immunostaining, immunoblot analysis, 3[H]-thymidine incorporation, cell growth, migration, and invasion assays were performed as described previously (13, 14, 16).

Total cDNA prepared from control and treated cells were used for qRT-PCR–based microarray analysis as per vendor's protocol (microarray# PPAHS-135Z; Qiagen). The data analysis was performed by using array analyzer software. The array data have been deposited to NIH-Gene bank database (accession number- GSE113309).

Cells were cultured in cancer stem cell medium using ultra-low binding culture plates (ScienCell Research, Carlsbad, CA). At fifth day of postseeding, cell spheroids were treated with therapies (PTC-209 and docetaxel) in fresh culture medium. Therapeutic agents were added to the cells at every 48th hour in fresh media, and the treatments continued for 8–12 days. At the termination, images were captured under microscope fitted with a CCD camera and the spheroid volumes and size were measured using software.

BMI1 occupancy on promoter region of PTEN and INK4a/p16 gene was determined by using the method as described by Song and colleagues (17). Briefly, cells were cross-linked with formaldehyde and trypsinized in lysis buffer followed by the nuclear digestion to obtain 300–800 bp DNA fragments. Equal quantity of chromatin supernatants was subjected to immunoprecipitation with anti-BMI1 and anti-IgG antibodies. Purified DNA was analyzed by real-time PCR (ABI Prism 7500) using PTEN and INK4A/p16 promoter. All ChIP assays were performed three times. The sequences of the qPCR primers are listed in Supplementary Table S1.

Cancer cells were trypsinized, counted, and labeled with Cell Tracker Orange CMTMR (Invitrogen) according to the manufacturer's instructions. Cells were resuspended in PBS containing DNase I and heparin, and 50–200 cells were microinjected in anesthetized embryos into the injection site, which varied according to the specific metastasis model used. Two different zebrafish metastasis models were used: (i) PC3 cells were microinjected in the yolk sac of 48 hours postfertilization (hpf) of wild-type zebrafish as described previously (18). Following injection, the zebrafish were put in 34°C embryo water (water supplemented with salt for maintaining zebrafish embryos). Twelve hours postinjection, zebrafish were examined for the presence of cells in the vasculature and these fish were excluded from the analysis. The zebrafish were imaged 4 days postinjection on a Zeiss Axio Observer ApoTome fluorescent microscope using standard dsRed filter set. (ii) MDA-PCa2b cells were microinjected into the pericardium of 72 hpf Tg(fli:GFP) zebrafish as described previously (19, 20). Embryos showing cancer cells in the bloodstream after injection were selected, put in 34°C embryo water for 24 hours, and then imaged on a Zeiss Axio Observer ApoTome fluorescent microscope using standard FITC and dsRed filter sets. All zebrafish studies were conducted in accordance with institutionally approved Institutional Animal Care and Use Committee protocols.

These were performed under two protocols (i) tail-vein metastasis protocol, and (ii) subcutaneous tumor protocol. These are described as follows:

• (A) Tail-vein metastasis protocol: Under this protocol the aim was to measure micrometastasis in lung of mice. For this purpose, we employed an intravenous cell implantation technique. The PC3M-luc cells (5 × 10⁶) were injected through the tail vain using a 30-gauge needle under aseptic conditions. At 0 hours postimplantation, all mice received an intraperitoneal injection of 100 μL of luciferin (150 mg/kg weight of mice) in PBS solution. The mice detected with positive signal (for luciferase) were randomly distributed into the experimental groups: (i) Control (n = 6) receiving the vehicle alone [100 μL solution DMSO (0.001) %/PBS], and (ii) BMI1-inhibitor (PTC-209)-treated group (n = 6; 60 mg/kg; twice/week; i.p. route; for 6 weeks).

• (B) Subcutaneous tumor protocol: This was performed as described previously (21).

Student's t test for independent analysis was applied to evaluate differences between the treated and untreated cells. Statistical analyses were carried out by using PRISM statistical software. A P value of < 0.05 was considered to be statistically significant.

Previously, we showed a progressive increase in the expression of BMI1 in primary tumor tissues that correlated with the disease stage of Caucasian prostate cancer patients (13, 14). In this study, we first determined the expression of BMI1 in primary tumor tissues of African-American and Caucasian patients by employing IHC (Fig. 1Ai). On the basis of IHC staining score, we compared BMI1 protein levels between Caucasian and African-American patients exhibiting stage III/IV disease. It is to be noted that majority of patients with stage IV disease develop metastasis. The expression of BMI1 was observed to be greater in African-American primary tumors than in Caucasian counterparts (Fig. 1A, ii). We determined the level of BMI1 mRNA expression in the matched normal and tumor tissues of African-American patients using qPCR (Fig. 1B, i). BMI1 expression was greater in prostate cancer patient tumor tissue relative to corresponding matched normal tissue (Fig. 1B, i; P < 0.05). Next, we determined the level of BMI1 protein in the metastatic tissues (Mets) derived from African-American and Caucasian prostate cancer patients by employing IHC. Elevated BMI1 levels were observed in the lymph node Mets of prostate cancer patients (Fig. 1Bii). We performed bioinformatics analysis of The Cancer Genome Atlas (TCGA), Oncomine (22), and NIH/Geo (GEO# GSE64331; GEO# GSE71016; refs. 23, 24) databases, and we found that metastatic prostate cancers exhibit higher BMI1 levels than primary tumors (Supplementary Fig. S1A). The data mining of the NIH-GEO database showed enrichment of BMI1 in prostatic tumors derived from African-Americans (Fig. 1B, iii), thus validating our observations. Next, using a cohort of Mets (lymph node, lung, bone) retrieved from African-American (n = 25) and Caucasian (n = 35) patients, we measured mRNA expression of BMI1 by employing qRT-PCR. Increased mRNA expression of BMI1 in the Mets of Caucasian and African-American patients was noted, thus strengthening the relevance of BMI1 in the metastasis of prostate cancer (Fig. 1C, i and ii). Subsequent biochemical studies (to establish the relevance of BMI1 in the metastasis) were performed in relevant race-based prostate cancer models.

Several genes associated with chemoresistance also play a key role in the metastasis (25). The prominent genes whose overexpression or downregulation has been reported to contribute to the metastasis in castration-resistant prostate cancer (CRPC) patients are AR, MMP2, cMYC, VEGF, BCL2, Cyclin D1, INK4a/p16, and PTEN (25). To determine the association of BMI1 to these molecular events, we measured the expression level of these genes in Mets of African-American and Caucasian patients by employing qRT-PCR. BMI1-rich Mets exhibit elevated mRNA expression of c-MYC, VEGF, Cyclin D1, and MMP2 (Fig. 1C, i). When compared between the races, African-American Mets exhibited the highest enrichment of BMI1, c-MYC, and Cyclin D1 (Fig. 1C, ii). Notably, AR expression was found to be low in both Caucasian and African-Americans Mets (Fig. 1C, i and ii). Two important suppressor genes PTEN and INK4a/p16 were observed to be significantly low in expression in the Mets of patients with prostate cancer (Fig. 1C, i). When compared, African-American Mets exhibited significantly reduced expression of INK4A/p16, whereas PTEN expression status remains similar in both races (Fig. 1C, ii). BMI1 is a transcriptional repressor and INK4a/p16 is its direct target (26). Inactivation of PTEN, an AR-independent event, has been implicated in the CRPC development in patients (27). We asked if INK4A/p16 and PTEN suppression during prostate cancer progression could be due to transcriptional repressor activity of BMI1 protein. We therefore investigated whether BMI1 is recruited to the promoter region of PTEN and INK4A/p16 genes in prostate cancer cells. We selected PTEN-positive and INK4A-positive prostate cancer models from both races as such: Caucasian (22Rv1 primary and DU145 brain metastasis) and African-Americans prostate cancer (RC-77T/E primary and MDA-PCa2b bone metastasis) and performed the ChIP analysis. The occupancy of BMI1 on the PTEN promoter was observed in RC-77T/E cells (Fig. 1D, i). BMI1 protein recruitment was also observed at the promoter regions of INK4a/p16 and PTEN genes in metastatic prostate cancer models (Fig. 1D, ii).

To establish the relevance of BMI1 as a regulator of metastatic genes in prostate cancer cells, we used siRNA to genetically knockdown BMI1 expression in metastatic prostate cancer cells. BMI1-suppressed MDA-PCa2b cells cotransfected with reporter plasmids were measured for MMP2, VEGF, c-MYC promoter activity and NFκB transcriptional factor activity using a dual luciferase assay. We found that silencing of BMI1 significantly decreases the activation of MMP2, VEGF, c-MYC, and NFκB in metastatic cells suggesting a direct association between BMI1 and metastatic genes (Fig. 1E, i–v). These data also suggest that BMI1 is a druggable target for treating metastatic prostate cancer.

Our previous and current data show that genetic targeting BMI1 (by siRNA) causes a significant downregulation of metastatic gene activities in prostate cancer cells (Fig. 1D, i–v). We asked whether a similar outcome could be achieved through pharmacologic inhibition of BMI1 activity. Kreso and colleagues identified a small molecule (PTC-209) as potential inhibitor of BMI1 expression in colon cancer cells (15). For the subsequent in vitro and in vivo studies in African-American and Caucasian models, we used PTC-209. We selected a spectrum of African-American and Caucasian models representing localized, primary invasive and metastatic phenotypes. By employing an MTT assay, we determined the efficacy of BMI1 inhibitor (PTC-209) on the growth of African-American prostate cancer models, and measured IC₅₀. The IC₅₀ of PTC-209 for RC-77T/E, E006AA, E006AA-hT, and MDA-Pca2b was estimated to be 1.4, 1.6, 1.8, and 1.75 μmol/L, respectively (Fig. 2A). African-American normal prostate cell model RC-77N/E did not show any response to PTC-209 therapy up to 5 μmol/L at 24 hours of treatment (Fig. 2A).

Next, we determined the impact of PTC-209 treatment (low dose of 1.0 μmol/L for 24 hours) on BMI1 mRNA and protein expression levels in African-American (E006AA, E006AA-hT, and MDA-PCa2b) and Caucasian (WB-NB26 and PC3) models using qRT-PCR and immunoblot analysis. PTC-209–treated cells exhibited reduced BMI1 transcript (>50% decrease; P < 0.05) and protein levels in all cell models (Fig. 2B and C). We asked whether downregulation of BMI1 mRNA expression in treated cells was due to an impact on the transcription activity of BMI1 gene. To address this question, WPNB26, PC3, E006AA, E006AA-hT, and MDA-PCa2b cell lines transfected with BMI1 promoter reporter (pGL3-BMI1-luc) were treated with PTC-209 (1.0 μmol/L) and evaluated for reporter activity after 24 hours of treatment. Notably, PTC-209 treatment was observed to significantly inhibit (<2-fold) the promoter activity of BMI1 gene in cells [Fig. 2D (i) and E (ii)]. The mechanism underlying the effect of PTC-209 on the promoter activity of BMI1 gene is of great interest; however, it is beyond the scope of the current study.

We investigated the effect of pharmacologic inhibition of BMI1 on the mRNA expression of well-characterized genes (known for their role in prostate cancer) using MDA-PCa2b cells treated with PCT-209 or control (DMSO) for 24 hours and subjected to the qRT-PCR–based microarray analysis (microarray# PPAHS-135Z; Qiagen). The prostate cancer–specific gene expression array data generated from control and treated cells were deposited to the NIH-Gene bank database (accession number: GSE113309). Genes showing modulation above or below 2.0 fold were considered and the data presented as a heatmap (Fig. 3A, i) is summarized in Fig. 3A, ii. PTC-209 treatment decreased the expression of VEGF (32-fold) and NFκB1 (160-fold); and PTC-209 increased the expression of PTEN and CDKN2A (which encodes p16/INK4A) in MDA-PCa2b cells (Fig. 3A, i and ii; P < 0.05). We next asked whether targeting BMI1 by PTC-209 impacts the transcription activity of such genes. As assessed by luciferase-based reporter assays, the reporter-transfected cells treated with PTC-209 (1.0 μmol/L for 24 hours) exhibited a significant reduction (P < 0.05) in the transcriptional activation of MMP9 (>1.5-fold), BCL2 (>3.5-fold), VEGF (>2.0-fold), and Cyclin D1 genes (Fig. 2B, i–viii). In addition, the activity of NFκB (>1.5-fold) and TCF transcriptional factors were decreased in PTC-209–treated E006AA-hT and MDA-PCa2b cell lines (Fig. 2B, i–viii). The inhibitory effect of PTC-209 was more prominent in metastatic models than in primary models (Fig. 3B, iii–viii). The inhibitory effect of PTC-209 on metastatic gene activity was higher (>3-fold) in African-American derived prostate cancer cells than the Caucasian counterparts (Fig. 2B, vi vs. C, i) and (Fig. 2B, v vs. C, iii). The data using PTC-209 to pharmacologically inhibit BMI1 correspond to genetic siRNA–mediated silencing of BMI1 findings and suggest that pharmacologic targeting of BMI1 carries high significance for futuristic drugs to treat metastatic prostate cancer.

To perform these studies, we used two preclinical in vivo metastasis models as following:

• (i) Zebrafish metastasis model: During metastasis, tumor cells migrate through the bloodstream, adhere to distant capillaries, and extravasate from the bloodstream to invade and colonize a distant metastatic site. We used two zebrafish metastasis models to test the efficacy of BMI1 pharmacologic inhibition in counteracting extravasation (19, 20) and subsequent metastasis of tumor cells (18). The clinical significance of zebrafish models to study drug responsiveness is emerging and complements in vitro and mouse models. Moreover, the transparency and experimental accessibility of zebrafish embryos enables the visualization of the metastatic process in real time at single-cell resolution (28). First, we determined the effect of pharmacologic inhibition of BMI1 on MDA-PCa2b cell extravasation using Tg(fli:GFP) transgenic zebrafish, which express GFP under the control of a vasculature-specific promoter, to facilitate in vivo visualization of the vascular system (19, 20). Either control or PTC-209-treated (1.0 μmol/L for 24 hours) MDA-PCa2b cells were transiently labeled with a fluorescent tracker dye and microinjected into the bloodstream of 72 hpf Tg(fli:GFP) zebrafish embryos. The following day at 24 hours postinjection of cells, the embryos were imaged using a fluorescence microscope. In the control group, we observed MDA-PCa2b cells in the extravascular space, whereas PTC-209 pretreated MDA-PCa2b cells remained in the vasculature (Fig. 3D and E). These results demonstrate that pharmacologic inhibition of BMI1 reduces prostate cancer cell extravasation in a vertebrate model. Second, we determined the efficacy of PTC-209 on the metastatic potential of prostate cancer cells in zebrafish. Either control or PTC-209-treated fluorescently labeled PC3 cells were microinjected in the yolk sac of 48 hpf zebrafish embryos. The extent of metastasis was evaluated by quantitating detectable fluorescent cancer cells that had migrated to the trunk/tail region of the embryo. Compared with control, the pretreatment of xenotransplanted PC3 cells with PTC-209 resulted in a significant reduction in metastasis (Fig. 3F). These data show that PTC-209–mediated inhibition of BMI1 reduces two important metastasis processes, extravasation, and cell migration, in vertebrate animals.

• (ii) Mouse metastasis model: Prostate cancer cells often remain undetected for a long periods of time in patients until they appear as well-established metastatic tumors in lymph nodes and skeletal tissue (29). We next determined the effect of BMI1 inhibition on the development of micrometastases in vivo. For this purpose, metastasis was induced in athymic mice via intravenous injection of PC3M-luc reporter cells. Animals were imaged to confirm the presence of tumor cells at the time of administration and distributed blindly into two groups: (i) control (n = 6) and (ii) PTC-209–treated (n = 6). The treatment group received PTC-209 (60 mg/kg; i.p., 2×/wk) and control received the vehicle [DMSO (0.001%)/saline, 100 μL]. All animals were sacrificed at 5 weeks postimplantation, and soft tissues (lungs, liver, kidney, brain) were tested for micrometastasis by using an India ink method. As expected, 5 of 6 mice in the control group exhibited lung micrometastases. Notably, in the treatment group, 4 of 6 mice did not exhibit micrometastases and rest of the treatment group (2 of 6 mice) exhibited a significantly reduced number of micrometastasis nodules (Fig. 3G, i and ii). We next validated the presence of prostate cancer -originated tumor cells by performing IHC for detection of a human mitochondrial protein in the mouse lung tissue. The IHC analysis showed that the metastatic nodules are positive for human mitochondrial markers (Fig. 3H, i–iv), indicating the presence of xenografted human prostate cancer cells. These data suggest a clinical potential of BMI1-inhibitors as antimetastatic agents.

Previously, we reported that genetic targeting of BMI1 (by shRNA) could sensitize resistant tumors to docetaxel therapy (13, 14). We investigated the efficacy of BMI1 inhibitor–based combination therapy in vitro and in vivo as following:

• (i) In vitro studies: We tested the efficacy of combination using Caucasian (WBNB26, PC3) and African-American (E006AA, E006HAA-ht, MDA-PCa2b) prostate cancer models on cell growth, proliferation, 3D organoid formation, invasiveness, and apoptosis by employing MTT, thymidine uptake, spheroid formation, chemoinvasion and FACS assays, respectively. The effect of PTC-209 and docetaxel monotherapies were noted as moderate in all Caucasian models. Docetaxel monotherapy generated only a marginal response in African-American models (Fig. 4). Notably, the combination therapy exhibited a significant (P < 0.01) impact on African-American models when measured for various functional indices (viability, rate of proliferation, invasion). The combination therapy resulted in 60%–75% reduction in growth and 70%–80% reduction in the rate of proliferation of African-American and Caucasian prostate cancer models (Fig. 4A and B). We observed an inhibitory effect of combination therapies on the clonogenic potential and induction of apoptosis in cell models (E006AA-hT, MDA-PCa2b, PC3) exposed to therapies (Supplementary Figs. S1C and S2). We compared the efficacy of mono- and combination therapies using 3D prostatosphere models. In this assay, therapeutic agents were added to the culture on alternate days in fresh medium and the treatments were continued for 2 weeks. When compared for size (diameter in μm units), the control-treated models exhibited significantly increased formation of spheroids with an average size of 600 ± 50 μm, 1,100 ± 150 μm, and 700 ± 90 μm for the E006hAAt, MDA-PCa2b, and PC3 models, respectively (Fig. 4B, i and ii). The docetaxel-treated models exhibited an average of 300 ± 50 μm, 800 ± 65 μm, and 210 ± 40 μm for E006AA-hT, MDA-PCa2b, and PC3, respectively (Fig. 4B, i and ii). PTC-209 monotherapy caused almost a 3- to 5-fold reduction in the spheroid formation in all models (Fig. 4B, i and ii). Whereas, docetaxel monotherapy caused a 2× reduction in spheroid formation, the PTC-209–based combination therapy caused a reduction of 45–50 times (P < 0.001) in the spheroid formation of African-American models (Fig. 4B, i and ii). We observed that combination therapy performs better than PTC-209 or docetaxel monotherapies when E006AA-hT and MDA-PCa2b cells were tested for proliferation (Fig. 4C, i–iv) migration and invasiveness potential (Fig. 4D, i–vii). We next determined the efficacy of mono- and combination therapies on the expression of BMI1, BMI1-target proteins (p16, BCL2, Cyclin D1) and a surrogate marker of proliferation (PCNA) by employing immunoblotting analysis. Docetaxel therapy did not cause a change in the expression of INK4A/p16 and BCL2; however, docetaxel increased the expression of Cyclin D1 in African-American invasive models (E006AA-hT, MDA-PCa2b; Fig. 4E ii–iii). An increase in Cyclin D1 due to docetaxel therapy suggests the possible association between the poor outcomes of docetaxel therapy in patients of this ethnic group. It is noticeable that PTC-209 in combination with docetaxel significantly reduced the expression of Cyclin D1 and BCL2 in E006AA-hT and MDA-PCa2b cells (Fig. 4E ii–iii). Similar data was observed for Caucasian models (Fig. 4E, iv and v).

• (ii) In vivo studies: We tested the efficacy of combination therapy on the growth of invasive tumors in mouse models. For this purpose, we selected E006AA-hT and PC3M-luc preclinical models. The E006AA-hT tumor growth is slow until 8 weeks of age at which measurable (200 mm³) globular shaped tumors become visible to naked eye in athymic mice. On the basis of this experience, we randomly distributed the mice in 4 different groups (n = 6) at 6 weeks postimplantation when average tumors measured 100 mm³. Mice received therapies for 6 weeks (2 times/week), and all mice were withdrawn from the protocol when control mice reached an endpoint tumor volume of 1,000 mm³. Tumors were harvested and measured for volume and weight. As shown in boxplots, the weight of tumors ranged from 850 to 1,100 mg in the control group (Fig. 5A, i). By comparing the final tumor weight among the groups, the docetaxel therapy caused a 20%–25% reduction; PTC-209 monotherapy caused a 65%–70% reduction and the combination (PTC-209 + docetaxel) caused an 80% reduction (Fig. 5A, i).

We next used PC3M-luc reporter cells for testing the efficacy of mono and combination therapy in athymic nude mice. Bioluminescence was detected in mice at 1 week postimplantation (Fig. 5B, ii). A follow-up for tumor growth of PC3M-luc mice with imaging continued for 6 weeks and all mice were sacrificed at this time point. The imaging data of growing PC3M-luc tumors obtained on weekly basis show that control tumors continued to grow in 6 of 6 mice, whereas the PC3M-luc tumors continued to grow in majority of (4/6) mice receiving docetaxel (Fig. 5B, i and ii). However, PTC-209–treated mice exhibited a significant inhibition in tumor growth. The differences in tumor growth are apparent from the data showing average photo flux indices of all groups at a given time (Fig. 5B, iii). The average normalized photon flux of control, PTC-209, docetaxel, and combination treated groups were as 5.3 × 10⁸, 3.2 × 10⁷, 3.3 × 10⁸, and 1.1 × 10⁷, respectively (Fig. 5B, iii). In combination therapy, a 90% reduction in photo flux was recorded as compared with control mice. When compared for final tumor weight and volume (at sacrifice), the docetaxel therapy was observed to cause a 10%–15% reduction; PTC-209 monotherapy caused a reduction of 60%–65% and the combination (PTC-209 + docetaxel) therapy caused 80%–90% reduction (Fig. 5B, iv and v). The IHC analysis of harvested tumors showed that as compared with control and docetaxel-treated groups, a significant decrease in the levels of marker of proliferation (Ki67), BMI1, c-Myc, and VEGF were recorded in PTC-209 mono- and combination therapy groups (Fig. 5C). Animals were monitored for health conditions (movement, drinking, fur condition) on daily basis and the weights were recorded on weekly basis (Supplementary Fig. S3). We did not find any potential toxicity with the therapies.

Taken together, our data provides strong evidence that targeting BMI1 (by pharmacologic agents) in a monotherapy or combination setting has the potential to inhibit the growth of invasive prostatic tumor cells at primary as well as secondary sites, such as lungs. This preclinical data suggests that pharmacologic inhibitors of BMI1 possess strong translational potential for clinical use, pending detailed investigations in relevant patient models (such as PDX) and genetically engineered mouse models of prostate cancer, which represents a current pursuit.

As per the GLOBOCAN program of the International Agency for Research on Cancer (IARC, Lyon, France), prostate cancer is the leading cancer in terms of incidence and mortality in men of African ethnic background (30). According to American Cancer Society data, 4,450 deaths and 29,530 new cases of prostate cancer were reported to be in African-American men in the United States (31). It is estimated that 1 in 6 (18.2%) of African-Americans have a lifetime probability of developing prostate cancer as compared to 1 in 8 (13.3%) of Caucasians and 44% to 75% relative higher risk of progressing to metastatic disease (32–34). Recent clinical trials which included African-American patients demonstrated to have minimal outcome in the advanced/metastatic castration-resistant prostate cancer (CRPC) stage of the disease (35, 36). Lowrance and colleagues showed that nonmetastatic CRPC is a common early clinical manifestation in Caucasian patients; however, 100% of the African-Americans exhibit metastasis at CRPC diagnosis (37). The taxanes are approved for the treatment of metastatic CRPC (38). An Eastern Cooperative Oncology Group–led phase III randomized trial conducted in patients with prostate cancer (89% Caucasian) suggested that docetaxel performs better under combination settings (39). Bernard and colleagues showed that African-Americans exhibit poor prognosis (OS) with ADT monotherapy compared with Caucasians; however, the ADT+ docetaxel combination therapy produced comparable outcome in patients of both races (39). Studies now have established that difference in ethnic/or genetic background could contribute to the disparity in the therapeutic outcome of prostate cancer patients (40). This study is focused on understanding the mechanism which underlies the metastasis and therapy outcome in patients with prostate cancer, particularly with race as a determinant factor. In this study, we performed a careful analysis of human tumors and employed tumor models representing African-Americans and Caucasian prostate cancer patients.

Metastasis of prostate cancer involves multiple steps, including angiogenesis, local migration, invasion, intravasation, extravasation of tumor cells, angiogenesis, and homing at the new site to form secondary tumors (41). Each step in the metastasis process requires the involvement of different set of genes or cooperation of multiple molecular pathways (41) The complete understanding of molecular events at each disease-stage would present an opportunity to identify druggable targets for intervention and treatment. BMI1 regulates signaling pathways critical for the emergence of chemoresistance in tumors as reviewed by us previously (42). This study identifies BMI1 as a master molecule that controls the activation of key molecules involved in the different stages of metastasis in humans. These genes include PTEN, p16, Cyclin D1, TCF4, NFκB, VEGF, MMP2, and c-MYC. Second, using relevant race-specific models and tools to measure metastasis, this study shows that BMI1-targeted therapy has the potential to effectively treat metastasis in patients with prostate cancer cutting across the ethnic backgrounds.

The increased nuclear localization of β-catenin is reported to increase nuclear activity of AR and TCF transcriptional factors, which in turn induces the metastasis of prostate cancer cells (7). We previously reported that BMI1 drives the nuclear activity of TCF4, causing docetaxel resistance in metastatic prostate cancer cells (14). In this study a notable disparity was observed in β-catenin activity in race-specific models; restricted to metastatic cells in Caucasians, whereas highly active in both primary and metastatic cells of African-Americans. We post that this could be one of the possible mechanisms underlying the aggressive nature of disease in patients with African-American prostate cancer. The progression of prostate cancer disease has been well-correlated to the loss of PTEN tumor suppressor in patients (43). Although loss of heterozygosity is considered a possible reason for loss of expression, it is not the lone factor causing the PTEN loss. On the basis of our data showing the occupancy of BMI1 as a repressor protein on promoter region of PTEN gene in African-American and Caucasian prostate cancer models, the current study adds another dimension to the mechanisms that regulate PTEN expression in neoplastic prostate cells.

Studies by Matusik and colleagues have established the fundamental role of NFκB in the development of CRPC disease (44). On the basis of our data, we speculate BMI1/NFκB nexus as a possible mechanism for (i) ADT and chemotherapy resistance and (ii) disease recurrence in patients with metastatic prostate cancer, and suggest that pharmacologic targeting of this molecular nexus as a futuristic treatment strategy for treating drug-resistant prostate cancer. Dissociation of primary tumor cells and homing of metastatic cells require the degradation of extracellular matrix of host tissues caused by matrix metalloproteases (MMP; ref. 45). We reported that MMP 2/9 are activated in metastatic prostate cancer (21). Morgia and colleagues showed that patients with metastatic prostate cancer exhibit high concentrations of MMP-2 and MMP-9 in plasma (46). A study in a transgenic mouse model showed that MMP2 knockdown reduced lung metastasis (47). MMP activity inhibitors such as batimastat, marimastat, prinomastat were largely unsuccessful in clinical trials due to several shortcomings, yet MMPs remain highly desirable therapeutic targets because of promising results in preclinical studies (45). Our study identifies an alternative route of MMP regulation at transcription level by BMI1. Therefore, BMI1-targeting could be envisaged as a strategy to improve the clinical outcome of MMP inhibitor therapies for metastatic disease. In this study we observed a disparity in responsiveness to the docetaxel in Caucasian vs. African-American models. However, it is noticeable that the pharmacologic inhibition of BMI1 rendered both African-American and Caucasian prostate cancer cells responsive to the docetaxel therapy. To summarize, we suggest that BMI1 a major driver of metastasis in prostate cancer is amenable to drug targeting. Our laboratory is currently pursuing this endeavor. We suggest that BMI1-based docetaxel combination would be an ideal arsenal for treating advanced prostate cancer because this combination consists of agents having the potential to eliminate highly proliferating tumor cells as well as slow-growing cancer stem cell cells (which are responsible for disease recurrence).

No potential conflicts of interest were disclosed.

Conception and design: B.R. Konety, M. Saleem

Development of methodology: A. Ganaie, F.H. Beigh, M. Astone, R. Maqbool, S. Umbreen, A.S. Parray, H.R. Siddique, M. Saleem

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Ganaie, F.H. Beigh, M. Astone, M.G. Ferrari, H.R. Siddique, P. Murugan, L.H. Hoeppner

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Ganaie, F.H. Beigh, M. Astone, T. Hussain, Y. Deng, B.R. Konety, L.H. Hoeppner

Writing, review, and/or revision of the manuscript: A. Ganaie, M. Astone, H.R. Siddique, P. Murugan, C. Morrissey, S. Koochekpour, Y. Deng, B.R. Konety, L.H. Hoeppner, M. Saleem

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F.H. Beigh, R. Maqbool, S. Umbreen, C. Morrissey, M. Saleem

Study supervision: L.H. Hoeppner, M. Saleem

Other (performed the laboratory experiments): F.H. Beigh

Other (performed animal experiments): T. Hussain

This work was partially supported by the bridge funding from the Department of Urology, Masonic Cancer Center, University of Minnesota. The author (M. Saleem) is supported by the US PHS grants (CA193739, CA184685, CA184685–02S1; pilot award U54); the author (L.H. Hoeppner) is supported by the US PHS grant (CA187035), the author (YD) is supported by the US PHS grant (CA155522; CA160333), and author (BRK) is supported by US PHS grant U54MD008620-06 and DOD grant W81XWH-17-1-0462. We thank the Prostate Cancer Biorepository network team (Dr. Bruce J. Trock, Kathy Willey Johns Hopkins University, Baltimore, MD) for providing race-based matched RNA/DNA of patients. This work is supported by the Department of Defense Prostate Cancer Research Program, Award No W81XWH-14-2-0182, W81XWH-14-2-0183, W81XWH-14-2-0185, W81XWH-14-2-0186, and W81XWH-15-2-0062 Prostate Cancer Biorepository Network (PCBN). We thank the patients and their families, Celestia Higano, Pete Nelson, Elahe Mostaghel, Bruce Montgomery, Lawrence True, Paul Lange, Robert Vessella and the rapid autopsy teams for their contributions to the University of Washington Medical Center Prostate Cancer Donor Rapid Autopsy Program. This work was supported by the Department of the Pacific Northwest Prostate Cancer SPORE (P50CA97186), the PO1 NIH grant (PO1 CA163227) and the Institute for Prostate Cancer Research (IPCR). We also acknowledge the support of NCI-sponsored Cooperative Tissue Network (CHTN) and the BioNet-biorepository of the Masonic Cancer Center, University of Minnesota for providing tumor tissues. We thank Kim Klukas, Terry S. Jones and Neelofar Jan for providing help in animal studies.

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.