Upregulation of Scavenger Receptor B1 Is Required for Steroidogenic and Nonsteroidogenic Cholesterol Metabolism in Prostate Cancer

Cholesterol is essential for rapid cancer growth (1), and has been specifically linked to prostate cancer progression to castration-resistant disease (CRPC; refs. 2, 3). Its levels are elevated in patient serum and bone metastasis after androgen deprivation therapy (ADT), and hypercholesterolemia correlates with increased prostate cancer–specific mortality (4–6). In addition, association of elevated squalene monooxygenase (SQLE) expression with higher Gleason grade and disease-specific mortality indicates a role for de novo intratumoral cholesterol synthesis in lethal prostate cancer (7). The increased appreciation that statin use is correlated with decreased prostate cancer occurrence and improved disease prognosis (8–10), together with evidence linking statin use to improved PSA declines and overall survival in abiraterone-treated patients (11, 12), highlight the benefit of reducing de novo cholesterol and androgen synthesis to achieve maximal suppression of androgen receptor (AR) pathway activation, and management of advanced prostate cancer (13–16).

Cholesterol needs can also be met by elevating systemic uptake via the actions of low-density lipoprotein receptor (LDLr) and scavenger receptors (SR), particularly the Class B1 allele, SR-B1 (SCARB1; ref. 17). LDLr transcript levels are lower in more aggressive tumors (7, 18). Although elevated SCARB1 transcript levels have been suggested to correlate with decreased disease-free survival (18), analyses of the well-annotated Health Professional Follow-up, Physicians' Health Study, and Swedish Watchful Waiting cohorts demonstrated unchanged expression relative to tumor Grade or disease outcome (7). Whether SR-B1 expression persists in CRPC, and how it might promote mechanisms of malignant transformation, remain to be determined.

SR-B1 internalizes high-density lipoprotein (HDL) cholesterol, and acetylated or oxidized LDL, and has allelic variants linked to increased risk of atherosclerosis and an impaired innate immune response (19). It is also critical for cholesterol uptake as a precursor for androgen synthesis in steroidogenic tissues (20). Experimentally, linkage of SR-B1 expression to prostate cancer aggressiveness includes elevated expression in de novo androgenic CRPC derivatives of LNCaP (13, 16), and increased tumor growth in TRAMP (21). SR-B1 also signals growth and survival of nonsteroidogenic endothelial (22), and breast cancer cells (23), and association of elevated expression with aggressive characteristics and poor prognosis of breast, and clear cell renal carcinomas, indicates roles for SR-B1 in multiple malignancies (24–26).

Hypothesizing that SR-B1 expression may help facilitate malignant transformation by increasing levels of metabolically available cholesterol, we demonstrate increased SR-B1 expression in the transition from normal prostatic tissue to cancerous tissue, and persistent high expression in metastases. We go on to show sensitivity of androgenic prostate cancer cell lines to SR-B1 antagonism, and how targeting SR-B1 suppresses cancer growth through induction of endoplasmic reticulum (ER) stress and autophagy via both steroid and nonsteroid-based mechanisms. These results implicate systemic cholesterol uptake mechanisms, particularly SR-B1, as potentially actionable targets for managing CRPC.

IHC staining of the prostate cancer Donor Rapid Autopsy Program at the University of Washington (UWRA, Seattle, WA) metastatic CRPC tissue microarray was performed using SR-B1 primary antibody: AB52629 (Abcam; ref. 27). Metastatic specimens were obtained from patients who died of metastatic CRPC, who signed written informed consent for a rapid autopsy performed within 6 hours of death under the aegis of the prostate cancer Donor Program at the University of Washington with Institutional Review Board approval. SR-B1 staining was scored by experienced independent pathologists (0 = no staining, 1 = low staining, 2 = moderate staining, 3 = high staining). Expression data for cholesterol metabolism gene transcripts was obtained from 27 patients with paired normal prostatic, and local cancerous, tissue from the Shanghai Changhai Hospital (Shanghai Shi, China) and Fudan University Shanghai Cancer Center (Shanghai Cohort, SC; ref. 28) and from 83 patients with CRPC from the UWRA; an expansion of the 63 CRPC patient data previously reported (27).

The immortalized human prostate epithelial cell line, BPH-1, was generously provided by Dr. S. Hayward (NorthShore Research Institute, Evanston, IL). Prostate cancer cell lines: C4-2, VCaP, 22Rv1, and PC3, were obtained from ATCC. BPH-1, 22Rv1, and PC3 were maintained in DMEM (Invitrogen) supplemented with 10% FBS (Invitrogen). C4-2, was maintained in RPMI1640 (Invitrogen) supplemented with 10% FBS. VCaP was maintained in low-bicarbonate DMEM (ATCC) supplemented with 10% FBS. Unless otherwise noted, all other reagents were from VWR or Thermo Fisher Scientific.

Block lipid transport-1 (BLT-1; ChemBridge), a selective inhibitor of cholesteryl ester transfer through SR-B1 (29), or dimethyl sulfoxide (DMSO, vehicle), was added to cells cultured in phenol red-free media supplemented with 5% charcoal-dextran–stripped FBS (CSS, Invitrogen) at the indicated final concentrations. Unless otherwise specified, all assays were conducted 3 days posttreatment initiation.

One day after transfection with either Stealth RNAi duplexes targeting SR-B1 (SRB1-KD: Oligo ID HSS101571: AUAAUCCGAACUUGUCCUUGAAGGG, catalog. no. 1299001) or Lo GC Negative Control duplexes (NC: catalog. no. 12935-110; Invitrogen), cells were cultured in phenol red-free RPMI1640 with 5% CSS for C4-2 cells, or DMEM with 10% FBS for PC3 cells (30). Unless otherwise specified, all assays were conducted 4 days posttransfection.

Metribolone (R1881, Perkin Elmer), dehydroepiandrosterone (DHEA, Steraloids), and progesterone (Sigma-Aldrich) were added to culture medium simultaneously with BLT-1. DHEA was added to culture media 1 day post-SRB1-KD/NC transfection.

Cellular protein levels were determined by immunoblot analysis as described in Supplementary Methods. Samples were normalized using primary antibodies targeting GAPDH (sc-32233) from Santa Cruz Biotechnology, and β-actin (A2228), or vinculin (V4505) from Sigma-Aldrich. Antibodies targeting the AR (sc-7305) was from Santa Cruz Biotechnology, SR-B1 (NB400-104) was from Novus Biologicals, and clusterin (CLU: 4214S and sc-6419) were from Cell Signaling Technology and Santa Cruz Biotechnology in Figs. 4 and 5, respectively. Antibodies targeting phospho-mTOR 923 (9234), mTOR (2983), BiP (3177), IRE1α (3294), p21 (2947), phospho-RB 807/811 (9308), phospho-RB 780 (9307), and LC3B (2775) were from Cell Signaling Technology, and TP53 (OP03) was from EMD Millipore.

Growth rates of BLT-1–treated or interfering RNA–transfected cells were determined by phase contrast image analysis using an Incucyte Zoom System (Essen Bioscience) with confluency measured from sequential images used to determine cell growth kinetics using the system software. Propidium iodide (PI)- and Annexin V–positive fractions were determined by automated image analysis at 72 hours posttreatment initiation. The Live/Dead Cytotoxicity Assay (Invitrogen) was performed following the manufacturer's instructions. Cell-cycle analysis was performed using the previously described PI-based flow cytometry method (31).

HDL-derived cholesterol uptake was approximated using the fluorescent lipid, 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI), from labelled HDL particles (DiI-HDL, Alfa Aesar) as modified from established techniques described in Supplementary Methods (29).

Quantitative mRNA expression analysis was performed as described in Supplementary Methods using Qiagen SYBR Green probes targeting SR-B1 (QT00033488), HMGCR (QT00004081), PSA (QT00027713), and NKX3.1 (QT00202650) normalized to GAPDH (QT00079247).

Cellular androgen levels were quantitatively assessed from approximately 100-mg cell pellets by LC/MS as described previously (15) and detailed in Supplementary Methods.

PSA secreted into media was quantified using an electrochemilluminescent immunoassay on a Cobas e 411 Analyzer (Roche) and analyzed as described previously (15).

Formalin-fixed C4-2 cells were stained with wheat germ agglutinin (WGA) conjugated to Alexa Fluor 647 (WGA-647, Life Technologies) and 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories) and imaged by confocal microscopy.

Senescence-activated β-galactosidase (SA-β-gal) activity was detected in cells treated with 100 μmol/L chloroquine for 2 hours, then with 33 μmol/L 5-dodecanoylaminofluorescein di-β-D-galactopyranoside (C₁₂FDG, Invitrogen) for 1 hour prior to flow cytometry analysis as adapted from previously published methods (32).

A total of 2 × 10⁶ PC3 cells were inoculated subcutaneously on the hind flank of athymic nude mice (Crl:NU-Foxn1^nu; Harlan) under the auspices of UBC animal ethics protocol UBC ACC A16-0072. Tumor volumes (TV) were measured using calipers and the equation TV = length × width × height × 0.5326. Once tumors exceeded 100 mm³, mice were randomized into vehicle (propylene glycol) or 25 mg/kg BLT-1 treatment (by oral gavage) cohorts administered daily for 4 weeks, or until tumor burden exceeded 10%, or weight loss exceeding 20%, in accordance with UBC Committee on Animal Care standards.

Statistical analyses were performed using GraphPad Prism (GraphPad). Student t tests, χ² tests, and ANOVA with Tukey or Sidak multiple comparisons test were used to determine differences between treatment groups. Means (±SEM) of the datasets were considered to be significantly different if P < 0.05.

We assessed SR-B1 expression in localized and metastatic prostate cancer by comparing IHC staining of rapid autopsy specimens from the UWRA (27) to levels in clinical mRNA expression datasets. SR-B1 staining intensity in the UWRA samples of cancerous and patient-matched adjacent normal prostatic tissue, and of bone, lymph node, liver, and lung metastasis rapid autopsy specimens, was scored as moderate to high in 24% of normal prostate samples (56/236 cores), in 71% of local prostate cancer samples (167/236 cores), and in 57% of bone (118/207 cores), 77% of liver (42/54 cores), 84% of lymph node (56/67 cores), and 84% of lung (22/26 cores) metastasis samples (Fig. 1A and B). Furthermore, statistical comparisons can be found in Supplementary Table S1. Overall, these results indicated that SR-B1 expression is increased in both local and metastatic samples compared with normal prostatic tissue, with bone exhibiting lower expression when compared with other metastatic sites.

We used transcriptional profiling data of matched UWRA specimens to analyze expression of the SR-B1 transcript, SCARB1, and of related cholesterol metabolism regulators, LDLr and HMGCR, in metastatic prostate cancer tissue (Fig. 1C). Consistent with our IHC analysis, we observed higher SCARB1 expression compared with LDLr and HMGCR, with the highest expression in liver metastasis, and the lowest expression in bone metastasis. Analysis of TCGA-aggregated transcript level data also demonstrated elevated SCARB1 and decreased LDLr levels in prostate cancer and indistinguishable levels of HMGCR between benign and cancerous samples (Supplementary Fig. S1). We assessed the consistency of these observations in a uniformly collected, independent prostate cancer cohort: the SC radical prostatectomy series (28). Comparing transcript levels for the cholesterol influx proteins SCARB1 and LDLr, multiple mevalonate pathway enzymes, and the cholesterol efflux proteins ABCA1 and ABCG1, between treatment-naïve prostate cancer and matched normal tissues (Fig. 1D), we determined that the prostate cancer group exhibited increased expression of SCARB1, decreased expression of LDLr, and no difference in expression of the other factors. The consistent upregulation of SR-B1 expression in the assessed cancerous specimens provides validation that SR-B1 expression is upregulated in prostate cancer, and the first demonstration that this elevated expression pattern persists in metastatic CPRC lesions. Furthermore, the observed decreased LDLr expression, suggests a potential shift in the mechanism prostate cancer cells use to obtain exogenous cholesterol necessary to meet the metabolic demands of a rapidly proliferating and metastatic disease (33).

Because cholesterol is the essential metabolic precursor for steroid synthesis and is essential for de novo steroidogenesis by Leydig cells, we assessed relative expression of SR-B1, and full-length and splice variant AR in a benign prostatic hyperplasia cell line, BPH-1, and three known AR-driven CRPC cell lines, C4-2, VCaP, and 22Rv1. SR-B1 expression was detected at generally equivalent levels in the four cell lines (Fig. 2A). Consistent with previous reports, full-length AR was robustly expressed by VCaP and C4-2 cells, and detected at much lower levels in BPH-1 and 22Rv1 cells, while AR splice variant levels were highest in 22Rv1 and VCaP cells.

We previously demonstrated that SR-B1 antagonism suppressed growth of LNCaP cells during CRPC progression (16). We next tested whether there was any differential sensitivity to SR-B1 antagonism on AR-driven growth of these cell lines. Cell viability was indistinguishable in BPH-1 cells treated with the small-molecule SR-B1 inhibitor, BLT-1 (Fig. 2B and C). Treatment of the cancer lines with BLT-1 dose-dependently induced cell death of C4-2 (≥10-fold) and VCaP (∼4-fold) as measured by increased PI uptake and Annexin V staining, while 22Rv1 cell viability was indistinguishable (Fig. 2B and C). The insensitivity of the nonmalignant BPH-1 line, and the relative sensitivity of prostate cancer lines proportional to AR full-length and splice variant levels, suggested an increased reliance on SR-B1 in C4-2 cells as compared with cells with higher full-length and/or splice variant AR levels.

As C4-2 appeared to be the most susceptible to SR-B1 inhibition, it was selected for further studies using the SR-B1–targeted RNA interference to silence expression (Fig. 2D). siSR-B1-C4-2 cells displayed complete growth arrest by 48 hours posttransfection, while scr-C4-2 cells displayed progressive growth over the time course such that the growth rate of siSR-B1-C4-2 cells was 84% less than that of scr-C4-2 cells (Fig. 2E). Similarly, BLT-1 treatment dose-dependently suppressed C4-2 cell growth by 17%, 81%, and 95% at 1, 10, and 20 μmol/L compared with the vehicle control, respectively, and replicated the growth arrest observed in siSR-B1-C4-2 cells at ≥ 10 μmol/L. (Fig. 2F). These observations demonstrate SR-B1 expression is critical for growth of C4-2 cells under androgen-deprived conditions.

To compare how varying doses of BLT-1 or siRNA suppression affected proliferation, or death rates, calcein AM ester hydrolysis, vital dye exclusion, and cellular DNA content were assessed. Despite the observed profound suppression of proliferation, siSR-B1-C4-2 cell viability was indistinguishable, albeit slightly lower, on average from that of scr-C4-2 cells (Fig. 2G). By cellular DNA content analysis, siSR-B1-C4-2 cells exhibited a 25% increase in G₀/G₁ phase cells compared with scr-C4-2 cells and a 2-fold increase in sub-G₀ cells, suggesting that suppressed SR-B1 expression did result in a modest cytotoxicity (Fig. 2H). At concentrations that replicated the growth arrest observed in siSR-B1-C4-2, BLT-1 treatment resulted in a dose-dependent decrease in viability: 20% at 10 μmol/L, and 70% at 20 μmol/L (Fig. 2G). Similarly, by DNA content analysis, 5 and 10 μmol/L BLT-1 treatment increased the sub-G₀ phase population 5- and 6-fold, respectively, to approximately 20% of the population at 10 μmol/L (Fig. 2H). These results indicate that SR-B1 antagonism induced both a G₀–G₁ growth arrest, and suggest a modest increase in cytotoxicity, particularly at high concentrations of BLT-1.

The ability of SR-B1 antagonism to reduce HDL-derived cholesterol uptake was assessed using RNA interference, and small-molecule antagonism. Both siSR-B1 and BLT-1 significantly reduced DiI uptake in C4-2 cells (siSR-B1-C4-2 vs. scr-C4-2: −39%, BLT-1 vs. vehicle: −62%, Fig. 3A). SR-B1 transcript levels were reduced 65% in siSR-B1 cells relative to scr-C4-2 cells, while SR-B1 protein was nearly absent (Fig. 3B). We previously reported that HMGCR inhibition induced SR-B1 expression in LNCaP-derived castrated xenografts (15), so we assessed whether SR-B1 antagonism had the converse effect on HMGCR expression. Finding that HMGCR expression was more than doubled in siSR-B1-C4-2 cells relative to scr-C4-2 cells (Fig. 3B) suggests that under androgen-deprived conditions, there is a compensatory response for increased de novo cholesterol synthesis to SR-B1 knockdown in this hormone-responsive prostate cancer model.

SR-B1 is critical for providing HDL cholesterol to steroidogenic tissues (20), and we previously demonstrated reduced PSA secretion from C4-2 cells following SR-B1 knockdown (13, 16). We therefore examined how cellular androgen levels and AR activation were impacted by SR-B1 antagonism. C4-2 cells are steroidogenic under androgen-deprived conditions (30); however, intracellular testosterone levels were decreased by nearly 60% in siSR-B1-C4-2 cells relative to scr-C4-2 (Fig. 3C) and dose-dependently by BLT-1 treatment (by approximately 50% and 65% at 5 and 10 μmol/L relative to vehicle, respectively, Fig. 3D). In addition, DHT levels were decreased by 80% in siSR-B1-C4-2 compared with scr-C4-2 cells, and 70% and 80% in 5 μmol/L and 10 μmol/L BLT-1–treated C4-2 cells, respectively, as compared with vehicle alone (Fig. 3C and D). These findings demonstrate the ability of SR-B1 antagonism to impede the accumulation of AR-activating androgens. Concurrently, transcript levels of the AR target genes: PSA and NKX3.1, were found to be suppressed by 86% and 43%, respectively, in siSR-B1-C4-2 compared with scr-C4-2 cells (Fig. 3B). Furthermore, PSA secretion was reduced 5-fold in siSR-B1 cells, and dose-dependently up to 5-fold in BLT-1–treated cells (Fig. 3E). The ability of both interfering RNA and small-molecule approaches to reduce accumulation of intracellular androgens and AR-mediated signaling in androgen-deprived C4-2 cells emphasize the ability of SR-B1 antagonism to impede de novo steroidogenesis, and suggests that the observed decreased AR activation is likely due to reduced presence of AR-activating androgens.

Although autophagy can be regulated by several independent mechanisms, mTOR is generally considered to be the primary regulator (34, 35). By Western blot analysis, we observed that mTOR phosphorylation was noticeably decreased in siSR-B1-C4-2 cells, and dose-dependently decreased in response to BLT-1 treatment (Fig. 4A). Decreased mTOR phosphorylation was correlated with activation of autophagy pathways, through increased expression of precursor, and robust expression of mature, CLU in both siSR-B1-C4-2 cells and BLT-1–treated cells, and increased LC3I/II conversion observed in siSR-B1-C4-2 cells (Fig. 4A). Induction of an autophagic phenotype was further supported by accumulation of perinuclear vacuoles visualized by fluorescent microscopy in siSR-B1-C4-2 cells (Fig. 4B). These observations suggest that decreased cholesterol import through SR-B1 antagonism promotes autophagic flux as a mechanism to promote survival in response to cell stress in a manner similar to that described previously (36).

Altered cholesterol metabolism can promote ER stress (37) that, in turn, promotes induction of autophagy (38). Consistent with this, both siSR-B1 knockdown, and BLT-1 treatment, of C4-2 cells strongly induced expression of the essential ER stress chaperone, BiP, and more modestly induced expression of the inducer of ER stress chaperones, IREα (Fig. 4A). Because ER stress and autophagy are appreciated to arrest growth (35), we interrogated the impact of SR-B1 antagonism on expression of cell-cycle check point markers. siSR-B1-C4-2 cells exhibited hypophosphorylation of RB serine 780 and serine 807/811, and increased expression of p53 and p21 (Fig. 4A). We additionally observed that cell-cycle and growth arrest in siSR-B1-C4-2 cells correlated with increased activity of SA-β-gal (Fig. 4C). Although typically used as a senescence marker, upregulated SA-β-gal activity is also known to be associated with autophagy (39). These results suggest that SR-B1 antagonism induces a strong autophagic phenotype, that is, at least in part, through the activation of ER stress pathways that inhibit mTOR.

To determine whether the stress responses observed in androgen-deprived, and SR-B1–antagonized, C4-2 cells resulted from decreased de novo steroidogenesis and AR activation, cells were costimulated with testosterone precursors, progesterone and DHEA, and the testosterone mimetic, R1881, to bypass the requirement for uptake/conversion of cholesterol as an androgen source. None of these factors reversed the growth arrest effects seen with 10 μmol/L BLT-1 treatment of androgen-deprived C4-2 cells (Fig. 5A). Furthermore, cytotoxicity, as measured through PI and Annexin V staining, were not reversed by these factors (Fig. 5B and C). As DHEA was the most clinically relevant steroid assessed (40), further studies were performed combining DHEA treatment with NC or SR-B1–targeted RNA interference. In scr-C4-2 cells, DHEA stimulated a near 10-fold increase in AR activity as measured by PSA secretion (Fig. 5D). Consistent with observations in Fig. 3, basal AR activity in siSR-B1 C4-2 cells was approximately 20% of that observed in unstimulated scr-C4-2 cells. Although DHEA treatment did stimulate a near 10-fold increase in AR activity in siSR-B1 C4-2 cells, the repressed basal AR activity of siSR-B1 C4-2 cells meant the resulting DHEA-induced maximal activity remained indistinguishable from the basal activity of scr-C4-2 cells, some 75% less than that of the DHEA-stimulated scr-C4-2 cells.

AR activation is a driver of C4-2 proliferation (41), but because the threshold of signaling required to maintain optimal growth or survival is not precisely defined, we also assessed the impact of DHEA stimulation on cell-cycle distribution of siSR-B1 C4-2 cells (Fig. 5E). The increased G₀–G₁ and sub-G₀ population of siSR-B1 C4-2 cells relative to scr-C4-2 cells were indistinguishable in the presence of DHEA. Lastly, DHEA stimulation did not affect the robust expression of CLU observed in siSR-B1-C4-2 cells relative to scr-C4-2 cells (Fig. 5F). We conclude that the arrested phenotype observed with SR-B1 antagonism correlates with reduced AR activation that cannot be restored solely by replenishing steroid levels.

Because restoring intracellular androgen levels appeared to be insufficient to reverse the antiproliferative effects of SR-B1 antagonism, we assessed how SR-B1 antagonism impacted proliferation and viability of the AR-null, androgen-independent prostate cancer cell line, PC3 (42). SR-B1–targeted interfering RNA–silenced PC3 cells (siSR-B1-PC3) exhibited little SR-B1 protein expression, and suppressed HDL-cholesterol uptake by 81% relative to scr-transfected cells (scr-PC3), while HDL-cholesterol uptake was suppressed 38% in BLT-1–treated PC3 cells relative to vehicle-treated cells (Fig. 6A). These results were consistent with the impact of SR-B1 antagonism in C4-2 cells and presented the opportunity to assess the impact of SR-B1 antagonism on a nonsteroidogenic, AR-independent prostate cancer model.

Using a growth kinetics assay, scr-PC3 cells confluency increased from 15% to nearly 80%, while siSR-B1-PC3 cells showed essentially no increase in cell density over the time course, never surpassing 20% confluency (Fig. 6B). Similarly, vehicle-treated PC3 cells grew from an initial density of 8% to approximately 50% confluency after 120 hours, while BLT-1 treatment profoundly suppressed PC3 growth, with the lowest dose (1 μmol/L) decreasing growth by 80% relative to vehicle, and the highest dose (20 μmol/L) resulting in a nearly complete growth arrest (Fig. 6C). Using proliferative and death indexes to determine how SR-B1 antagonism impacts growth kinetics of PC3 cells, we observed siSR-B1 transfection to be strongly cytotoxic, with 80% of the population reporting as dead, compared with 15% of the scr-PC3 cell population (Fig. 6D). Using DNA content analysis, this response was linked to a profound induction of cell death in siSR-B1-PC3 cells as compared with scr-PC3 cells (Fig. 6E). In contrast, while dose-dependently sensitive, BLT-1–treated cells only displayed a significant induction of cell death at 20 μmol/L (30%) when compared with the vehicle-treated cells (Fig. 6D). BLT-1–treated cells, instead, underwent up to a 15% increase in G₀–G₁ arrest relative to vehicle-treated cells, with little change in sub-G₀ levels (Fig. 6E). Despite the cytotoxic differences between interfering RNA–transfected and small molecule–treated cells, the robust response of PC3 cells indicates the importance of non–AR-mediated effects to SR-B1 antagonism bringing further credence to the impact of nutrient starvation and induction of cellular stresses.

To date, there are no reports assessing BLT-1 as a pharmacologic agent in mice. To determine whether potentially efficacious levels of circulating BLT-1 could be achieved, serum samples were obtained from mice dosed by oral gavage to develop a pharmacokinetic profile. Mice dosed with 25 mg/kg BLT-1 had a C_max of 552.5 ng/mL (2.28 μmol/L) at 0.5 hours postdose, the first measured time point (Supplementary Fig. S2A). The elimination half-life was 10.4 hours, with a final concentration, measured 24 hours post dose, of 0.783 ng/mL (0.189 μmol/L) for a calculated AUC_0-∞ of 3971.2 ng*hr/mL. Furthermore, BLT-1 was rapidly metabolized by liver microsomes having a half-life of 8 minutes (Supplementary Fig. S2B). Mice dosed with 50 mg/kg BLT-1 were observed to suffer evidence of liver and kidney toxicity (Supplementary Table S2).

Because PC3 cell growth was more sensitive to BLT-1 than C4-2 cells, with a nearly complete cessation of growth observed with 1 μmol/L BLT-1, we concluded that 25 mg/kg would be a sustainable daily dose that could achieve the in vitro therapeutic dose for PC3 cells to assess whether it might impact xenograft growth. Over the treatment course, neither 25 mg/kg BLT-1, nor vehicle dosing impacted body weight or behavior (Supplementary Fig. S3). Although all mice displayed progressive tumor growth over the experiment course, the 5.4-fold increase in tumor growth in the BLT-1 cohort was significantly less than the 7.5-fold increase in the vehicle-treated cohort (Fig. 6F; Supplementary Table S3). Using a linear regression model as described previously (15), tumor growth rate of the BLT-1 cohort (110.3 ± 9.28 mm³/week) was approximately 30% less than the vehicle cohort (156.6 ± 10.74 mm³/week). These results indicate that, despite a narrow therapeutic window, BLT-1 is capable of slowing PC3 tumor growth as a single agent.

Increased SR-B1 expression has been suggested to be related to aggressive characteristics in several cancer types (24, 25); however, its role in prostate cancer remains enigmatic. Here, we validate increased SR-B1 expression in localized prostate cancer by comparison with adjacent normal prostatic tissue, and present the first evidence for persistent elevated expression in metastatic lesions where, anecdotally, specimens displaying high SR-B1 expression exhibited staining predominantly along the plasma membrane of carcinoma cells adjacent to stroma. While lower SR-B1 expression in bone metastasis could be attributed to effects of decalcification on antigenicity (43), decreased SCARB1 mRNA levels in matched specimens are consistent with these IHC observations. These corroborative findings, combined with the increasing appreciation of a role for cholesterol accumulation in disease aggressiveness, implicate SR-B1 as a factor in prostate cancer progression.

SR-B1 antagonism using interfering RNA and small-molecule approaches lead to robust reduction of prostate cancer cell growth in vitro, while small-molecule treatment resulted in a moderate reduction in xenograft growth. Here, we describe how SR-B1 inhibition reduces androgen accumulation and AR activation in steroidogenic prostate cancer cells; however, the differential responses of prostate cancer models underscore the heterogeneous nature of the disease. The cell lines used represent distinct recurrent CRPC phenotypes. C4-2 cells maintain AR activation and signaling through de novo steroidogenesis (44, 45). Similarly, VCaP harbor steroidogenic potential, but also express higher levels of full-length and splice variant AR isoforms (46), while 22Rv1 are predominantly AR splice variant–driven (47), and PC3 are fully AR-independent (48). In CRPC, AR-driven lipogenesis is associated with poor prognosis and linked to AR splice variant expression (49). The decreased sensitivity of VCaP, and insensitivity of 22Rv1, to BLT-1 treatment, are consistent with the possibility that AR splice variant expression could be sufficient to bypass the need for de novo cholesterol synthesis, or to drive lipogenic pathways under ARPI conditions (49).

Management of metastatic CRPC with second-line ARPIs (50, 51), and indication that HMGCR inhibition can restore castration sensitivity of CRPC models (49, 52), implicate a role for de novo steroidogenesis in ARPI resistance. If inhibition of HDL-derived cholesterol uptake through SR-B1 also impacts androgen accumulation by impeding de novo steroidogenesis, SR-B1 antagonism offers the potential to overcome several proposed mechanisms of ARPI resistance, including CYP17A1 amplification, and intratumoral accumulation of higher-order steroids, and AR mutations that allow for responsiveness to steroid precursors (45, 53, 54). However, our previous observation that statin treatment increased SR-B1 expression in LNCaP xenografts (15), and, here, that SR-B1 antagonism increased HMGCR expression in C4-2 cells, suggests that these mutually compensatory mechanisms should be considered to take best advantage of targeting cholesterol availability in CRPC. Furthermore, these results suggest that effectiveness of targeting SR-B1 to suppress intratumoral steroidogenesis might be limited to full-length AR-expressing CPRCs.

Cancer cells enduring nutritional, or other external, stresses employ survival mechanisms, including autophagy, in which cells degrade and recycle cellular constituents to meet metabolic demands (39). Autophagic responses to prostate cancer treatments are common, and include responses to ADT and ARPIs, taxanes, and kinase inhibitors (35). mTOR is an essential regulator of autophagy (35), and here is linked to induction of perinuclear vacuoles, LC3 lipidation, and CLU expression. Inhibiting de novo lipogenesis in CPRC models reduces growth, and suppresses mTOR activity, and HMGCR and AR expression (49, 52). In addition, perturbing lipid and cholesterol homeostasis induce activation of ER stress and the unfolded protein response (UPR; ref. 55, 56). During the UPR, IRE1α activation leads to activation of key genes responsible for preventing hypocholesterolemia, and may therefore drive compensatory alterations in HMGCR expression (57). Such adaptations may underlie the enhanced efficacy of combining ARPIs with biguanides to disrupt mTOR nutrient sensor pathways, and statins to suppress de novo cholesterol synthesis (58).

The ability of AR pathway activation to negatively regulate autophagic activity under suboptimal environmental conditions, such as culturing in charcoal-stripped serum, was initially considered (54); however, R1881, DHEA, and progesterone were unable to reverse the effects of SR-B1 antagonism. Although DHEA is a weak AR activator, it serves as precursor to more potent AR-activating androgens (40), and potently induced AR activity in these studies. The inability of DHEA to return AR activity to non–SR-B1–antagonized levels indicates that reduced de novo androgen synthesis, due to reduction of precursor, is, but, part of other extrasteroidal effects of SR-B1 antagonism that result in induction of an ER stress response program. Although SR-B1 knockdown induced an autophagic response, in C4-2 cells, it resulted in a strong cytotoxicity in PC3 cells, even though PC3 cells are capable of becoming autophagic following treatment with 26S proteosome or mTOR inhibitors (36, 48). The lack of any AR axis–mediated signaling may impact their ability to initiate antiproliferative, but prosurvival, stress responses, resulting in a nutrient-depleted induction of cellular death. Therefore, the loss of AR functionality in an increasing fraction of patients failing second-line ARPIs (49), may help identify patients particularly sensitive to SR-B1 targeting.

BLT-1 is an established SR-B1–selective small-molecule inhibitor, found to enhance HDL binding to SR-B1 but prevent intracellular transfer of cholesterol or cholesteryl ester (29). Although not ideal for further development due to high hydrophobicity (59), rapid metabolism, and toxicity at high concentrations, sufficiently efficacious circulating BLT-1 levels were achieved to slow PC3 xenograft growth. In light of the hepato- and nephrotoxicity of the 50 mg/kg BLT-1 dosing, it is possible that xenograft growth was impacted, at least in part, because of subclinically impaired general health of the 25 mg/kg–treated mice. Despite this caveat, the combined results of these proof-of-principle findings indicate that SR-B1 antagonism can impact CRPC growth. These findings identify SR-B1 as an important contributing factor in the sustained proliferation of malignant prostatic disease, and highlight the potential for development of a novel SR-B1 inhibitor designed with intention for in vivo use. The ability of SR-B1 antagonism to arrest growth independent of AR activity, while also reducing AR activity in steroid-responsive prostate cancer, provides a promising therapeutic prospect across the CRPC spectrum.

E.S. Tomlinson Guns is a consultant/advisory board member for Prostate Cancer Canada and Prostate Cancer Foundation BC. P.S. Nelson is a consultant/advisory board member for Astellas, Janssen, and Genentech and has provided expert testimony for Venable. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.A. Gordon, A. Midha, F. Derakhshan, E.S. Tomlinson Guns, K.M. Wasan, M.E. Cox

Development of methodology: J.A. Gordon, J.W. Noble, H.H. Adomat, E.S. Tomlinson Guns, M.E. Cox

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.A. Gordon, J.W. Noble, G. Wang, H.H. Adomat, C.C. Collins, P.S. Nelson, C. Morrissey

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.A. Gordon, F. Derakhshan, G. Wang, H.H. Adomat, Y.-Y. Lin, S. Ren, P.S. Nelson, M.E. Cox

Writing, review, and/or revision of the manuscript: J.A. Gordon, F. Derakhshan, G. Wang, H.H. Adomat, E.S.T. Guns, P.S. Nelson, C. Morrissey, K.M. Wasan, M.E. Cox

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.A. Gordon, H.H. Adomat, E.S. Tomlinson Guns, S. Ren, M.E. Cox

Study supervision: J.A. Gordon, E.S.Tomlinson Guns, K.M. Wasan, M.E. Cox

Others (carried out in vitro work): A. Midha

We thank Mary Bowden, Lisha Brown, and Mei Yieng Chin for animal husbandry support, and Jonathan Frew for confocal microscopy assistance. We thank the patients and their families, Celestia Higano, Evan Yu, Heather Cheng, Bruce Montgomery, Elahe Mostaghel, Mike Schweizer, Andrew Hsieh, Dan Lin, Funda Vakar-Lopez, Lawrence True, and the rapid autopsy teams for their contributions to the UWRA program. This work is supported by the Prostate Cancer Foundation BC and Prostate Cancer Canada (2012-917). The UWRA tissue acquisition was supported by the Department of Defense Prostate Cancer Biorepository Network (PCBN; W81XWH-14-2-0183), the Pacific Northwest Prostate Cancer SPORE (P50CA97186), the PO1 NIH grant (PO1 CA163227), and the Institute for Prostate Cancer Research (IPCR).

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.