Exploiting AR-Regulated Drug Transport to Induce Sensitivity to the Survivin Inhibitor YM155 Free

Prostate cancer is the most frequently diagnosed cancer in men and the second leading cause of male cancer mortality (1). The androgen receptor (AR) is a master regulator of prostate development and the maintenance of prostate epithelial cell viability and secretory activity. The AR-directed transcriptional program also serves to maintain the survival of prostate cancer and is the focal point for therapeutic strategies in localized disease where it is combined with radiation treatment, and for advanced disease where AR pathway suppression remains first-line therapy. The physiologic role of the AR changes as prostate cells undergo tumorigenesis and progression (2). Notably, as AR signaling promotes cellular differentiation and can suppress proliferation, early oncogenic events, including MYC gain, loss of tumor suppressors such as PTEN and TP53, and ERG overexpression, allow transforming cells to uncouple the suppressive functions of AR signaling while benefiting from growth and metabolic advantages (3–5).

Although suppressing AR activity through ligand reduction in the form of androgen deprivation therapy (ADT) is initially effective in metastatic prostate cancer, disease progression, termed castration-resistant prostate cancer (CRPC), inevitably manifests after 2 to 3 years. This occurs due to the reactivation of AR through AR amplification, AR mutations, intratumoral synthesis of androgens, and the production of AR splice variants (6). The relationship between AR and prostate cancer further evolves upon progression to CRPC through genomic (7), cistromic (8), and transcriptional (9) alterations. Because of the maintenance of AR activity, survival benefits are achieved by retargeting AR signaling with next-generation AR-directed therapeutics, although responses are generally measured in months rather than years (10, 11). Although resistance is again often accompanied by persistent AR activity, prolonged and effective AR suppression can induce epithelial-to-mesenchymal transition (12–14), acquisition of stem-like cell characteristics (13, 15–17), and transdifferentiation into an AR-null neuroendocrine phenotype. There are also significant complications and quality-of-life issues that arise with long-term ADT (18, 19). For these reasons, there has been a longstanding interest in the discovery of therapeutic modalities that may acutely synergize with AR-directed therapy to improve long-term outcomes in men with advanced prostate cancer.

AR activity can be dichotomous in action by promoting prostate cancer growth under normal circumstances and retarding prostate cancer growth when overstimulated with excessive androgens. AR signaling induces cell-cycle arrest, in part, by upregulating negative regulators of the cell cycle (20, 21). AR is also a licensing factor for DNA replication, and excessive androgens prevent AR cycling and relicensing, resulting in cell-cycle block (22, 23). In addition, AR can recruit TOP2B to sites of active transcription, resulting in double-strand DNA breaks (24).

The observation that prostate cancer cells can adjust to too much or too little AR signaling over time has led to the development of an approach termed bipolar androgen therapy that rapidly cycles androgen levels to maximize the suppressive benefits of ADT while attempting to prevent prostate cancers from adapting to a static low androgen environment (25, 26). This approach has been evaluated in phase I and II clinical trials (27, 28). Furthermore, the use of sustained high-dose androgens or bipolar therapy has also been evaluated to exploit potential synergies with other therapeutics such as ionizing radiation (29). In this study, we sought to determine whether the growth-suppressive effects of high-dose androgen, or other cellular activities regulated through AR, would synergize with anticancer drugs to further inhibit proliferation or induce prostate cancer cell death.

The LNCaP cell line (ATCC, CRL-1740) was cultured in RPMI1640, no phenol red (GIBCO, cat. no. 11835030) with 10% FBS (GIBCO, cat. no. 10437-02). VCaPs (ATCC, CRL-2876) were cultured in DMEM/F-12, no phenol red (GIBCO, cat. no. 21041025) with 10% FBS. Cell lines were lineage and mycoplasma validated by DDC (DNA Diagnostic Center) Medical. Cells were cultured for no more than 20 passages from the validated stocks. Drug screening was carried out at Quellos High Throughput Screening Core (University of Washington, Seattle, WA). The epigenetics, apoptosis, and stem cell modifier screening libraries were obtained from SelleckChem. LNCaPs were plated in 384-well plates at 2,500 cells per well in 50 μL of complete media (10% FBS) with either 10 μmol/L enzalutamide, 10 nmol/L R1881, or DMSO control using a PerkinElmer Wellmate Dispenser. They were incubated at 37°C in 5% CO₂ overnight. The next day, the compound libraries were added using a CyBio CyBi-Well Vario outfitted with tip washing stations and a 384-head equipped with a pin tool using 50-nL slotted pins (V&P Scientific). Fifty nanoliters of compounds dissolved in 100% DMSO (1,000× concentrated) were adsorbed onto the pins and washed off into cell assay wells containing 50 μL of complete media (0.1% DMSO final). Plates were then incubated for 72 to 75 hours at 37°C, 5% CO₂. Cells were harvested with CellTiter-Glo (Promega, cat. no. G7572) according to the manufacturer's protocol and read with PerkinElmer Envision Multi-label Plate Reader outfitted with a plate stacker and ultrasensitive luminescence detection. Viability signal was blank subtracted and normalized to DMSO control and plotted using Microsoft Excel and Tibco Spotfire.

LNCaP or VCaP cells were plated into 96-well plates at 5,000 cells/well in complete growth media (10% FBS). The next day, a one-fourth dose dilution series of YM155 was added in 25 μL complete media following a one-fifth R1881 dose dilution series or a one-third enzalutamide dose dilution series also in 25 μL in complete growth media. Cells were incubated for 5 days at 37°C, 5% CO₂, then harvested with 30 μL/well CellTiter-Glo (Promega, cat. no. G7572). LNCAP and VCAP cells were plated as described for isobolographic analysis except either 10 nmol/L R1881 or 2.5 nmol/L testosterone was added and cultured for 5 days unless otherwise noted. IC₅₀ values were generated with GraphPad Prism 6 and evaluated with the extra sum of squares F-test applying a P value of 0.05 (n = 4).

Western blots were run on NuPAGE 4% to 12% Bis-tris gels (Thermo Fisher Scientific, cat. no. NP0321) with MOPS SDS buffer (Thermo Fisher Scientific, cat. no. NP0001) then transferred to PVDF membranes (Thermo Fisher Scientific, cat. no. LC2002) for SLC35F2 or nitrocellulose membranes (Thermo Fisher Scientific, cat. no. LC2000) for other proteins with NuPAGE transfer buffer (Thermo Fisher Scientific, cat. no. NP0006). Membranes were blocked in TBS with 0.1% Tween-20, 5% milk, and 2.5% BSA. The antibodies used were SLC35F2 (Proteintech, cat. no. 25526-1-AP), GAPDH (Cell Signaling Technology, cat. no. 2118), ATG5 (Cell Signaling Technology, cat. no. 8540), Beclin (Cell Signaling Technology, cat. no. 3495), γH2AX (Cell Signaling Technology, cat. no. 9718), PARP (Cell Signaling Technology, cat. no. 9542), BIRC5 (Abcam, cat. no. ab76424), and caspase 3 (Cell Signaling Technology, cat. no. 9662).

Gene expression microarray data were analyzed as described previously (30, 31). Expression across primary prostate tumors (n = 11; ref. 32) and CRPC tumor samples (n = 171; ref. 30) was plotted in GraphPad Prism version 7.00 (GraphPad Software) using normalized, log2-transformed microarray signal intensities. The AR activity score was determined by the expression of a 20-gene signature (33) and calculated as described previously (34). Briefly, the activity score is defined as the sum of the expression Z-scores converted to a percent. Pearson correlation coefficient of AR activity and SLC35F2 expression (normalized, log2 signal intensity) in patient CRPC tumors (n = 171) was assessed using the cor.test function in R. qRT-PCR was performed Power Sybr Green (Applied Biosystems, cat. no. 4367659) and run on a Bio-Rad CFX384 real-time system according to the manufacturer's recommendation. Primers used were RPL13A_F-CCTGGAGGAGAAGAGGAAAGAGA, Hs_RPL13A_R-TTGAGGACCTCTGTGTATTTGTCAA, Arv567es_F TGCTGGACACGACAACAA, Arv567es_R GCAGCTCTCTCGCAATCA SLC35F2_F-AGGCAAACTCTTCACCTGGAAT, and SLC35F2_R-TCTGAAGCATGGGGGTGTTC.

Previously published chromatin immunoprecipitation sequencing (ChIP-seq) data for AR were obtained from the following Gene-Expression Omnibus and Sequence Read Archive: LNCaP-1F5 vehicle reps 1-3 (GSM973815-GSM973817), LNCaP-1F5 DHT reps 1-3 (GSM973818-GSM973820; ref. 35); LNCaP ethanol (GSM353643), LNCaP R1881 (GSM353644; ref. 36); LNCaP vehicle (GSM696839), LNCaP R1881-stimulated (GSM696840; ref. 37). Raw reads were aligned to hg19 with bowtie (38). Reads that had more than a single alignment were suppressed. Data were visualized with IGV (39).

AR-directed ChIP-qPCR was performed as described previously (40). Briefly, LNCaP cells in 5% CSS medium were pretreated with enzalutamide for 2 hours and followed by 10 nmol/L of DHT for 4 hours. ChIP was performed using anti-AR antibody (Santa Cruz Biotechnology, 816×). The qPCR analysis was carried out using the SYBR Green method on the QuantStudio 3 Real-Time PCR system (Applied Biosystems). The primers used were SLC35F2_AR1_F:5′-AGAGAATCGTCCTTCAGAACC, SLC35F2_AR1_R:5′-GGACTGAGCACAAACAAACC, SLC35F2_AR2_F:5′-GGTCACTACCAAATGAACTGATCATG, SLC35F2_AR2_R:5′-AGTAGATAAGAAGGCTGACACCTG, SLC35F2_AR3_F:5′-GTTGAACTAACAGAGGTTTCAG, SLC35F2_AR3_R:5′-GATATGAATCAATACGGGCTGGCAC.

The SLC35F2 ORF was obtained from a previously published ORF library (41) and expressed using the pLX304 (Addgene, plasmid no. 25890) lentivirus backbone. The ABCB1 expression vector was constructed from pHAMDRwt (Addgene, plasmid no. 10957) by PCR amplifying out ABCB1 using the primers ABCB1_F-gccaccATGGATCTTGAAGGGGACCGCAATGG and ABCB1_R-TCACTGGCGCTTTGTTCCAGCCTGGAC and cloning it into PCR8-GW-TOPO (Thermo Fisher Scientific, cat. no. K250020); then, a gateway reaction was used to clone into the pL6.3/V5 lentivirus (Thermo Fisher Scientific, V53306). The SLC35F2 shRNAs were the GIPZ Open Biosystems Human shRNAmirs V3LHS_377258 and V2LHS_154477.

Animal studies were carried out in strict accordance with NIH guidelines and with protocols approved by the Fred Hutchinson Center and the University of Washington Institutional Animal Care and Use Committees. All surgeries were performed under isoflurane anesthesia, and all efforts were made to minimize suffering. Five different LuCaP patient-derived xenograft (PDX) models established as part of the University of Washington tissue bank (42, 43) were used (LuCaP 23, LuCaP 35, LuCaP 96, LuCaP 86.2, and LuCaP 136). All lines express the wild-type AR and secrete PSA. Intact 6- to 8-week-old male C.B-17 SCID mice (Charles River Laboratories) were implanted subcutaneously with 30-mm³ tumor pieces. When tumors reached an average of 100 mm³, a subset of mice were castrated (Cx). Tumor volume was determined by the following formula (long and short axis lengths in mm): long × (short²)/2. Tumors from a subset of mice in each cohort were harvested at days 7 and 21 after castration, and the rest of the animals were followed and sacrificed until tumors exceeded 1,000 mm³ (end of study) or sacrificed if animals became compromised. Xenografts were harvested and flash frozen for the determination of tissue androgens and extraction of total RNA.

Methods for the determination of steroids in tissue samples by mass spectrometry were as previously reported (44). Briefly, frozen tissues were weighed, added to 60°C water containing deuterated internal standards, heated to 60°C for 10 minutes, and homogenized using a tissue homogenizer (Precellys; Bertin); supernatant was extracted twice with hexane [ethyl acetate (80:20, v/v)], and the organic layer was dried (SpeedVac; Thermo Fisher Scientific), derivatized with 0.025 mol/L hydroxylamine hydrochloride for 24 hours at room temperature to form oximes, and quantified using liquid chromatography electrospray-ionization tandem mass spectrometry. The lower limit of detection for steroids in tissue was 0.49 pg/sample (0.02 pg/mg) for DHT and testosterone.

The LNCaP prostate cancer cell line grows maximally in 10% FBS containing media, and growth is repressed when androgens are added (Supplementary Fig. S1). To identify drugs capable of enhancing the activity of AR inhibition or AR activation, we screened a library of 145 well-characterized pharmacologic agents that impair cancer cell proliferation or survival using the androgen-sensitive prostate cancer cell line LNCaP (Fig. 1A). Briefly, we plated LNCaP cells in standard growth media containing 10% FBS with either the AR antagonist enzalutamide, the synthetic androgen R1881, or DMSO vehicle. After 24 hours, a 14-point range of concentrations of each library compound was applied to the plates and incubated for an additional 72 hours, at which point, cell viability was quantitated. Strikingly, in control cultures with no other added drugs, R1881 suppressed cell proliferation as potently as the enzalutamide treatment (Fig. 1B). No compounds synergized with enzalutamide under these conditions. However, YM155, which suppresses transcription of the antiapoptotic protein survivin/BIRC5, displayed a supra-additive effect with R1881.

To validate the synergy between R1881 and YM155, we measured YM155 IC₅₀ values in response to varying R1881 concentrations. These data demonstrated a potent synergistic interaction between R1881 and YM155 in LNCaP cells with R1881 increasing LNCaP sensitivity to YM155 from IC₅₀ 24 nmol/L without R1881 to 3.85 nmol/L with an R1881 concentration of 160 pmol/L (Fig. 1C). Synergy with YM155 was not achieved with enzalutamide (Fig. 1D). Surprisingly, the VCaP prostate cancer cell line, which has amplification of AR, was initially more sensitive to YM155 but not further sensitized to YM155 by the addition of androgen (Fig. 1C).

The mechanism by which YM155 is reported to elicit cell death is controversial and varies based on cell line, time, and YM155 concentration (45). However, YM155 is reported to suppress BIRC5 expression, increase DNA damage, and induce autophagy (45). Given that BIRC5 is involved in DNA repair (46), and DNA damage can induce autophagy (47), we sought to determine the action of YM155-mediated cellular effects before significant cell loss and secondary effects occurred. To this end, we performed Western blots on LNCaP cells preincubated for 48 hours with 10 nmol/L R1881 or vehicle control and subsequently dosed with 0.5 or 1 μmol/L of YM155 for 24 hours (in contrast to the proliferation studies carried out at 96 hours). R1881 alone suppressed BIRC5 levels, consistent with androgen-induced G₁ cell-cycle arrest (Fig. 2A; ref. 21). YM155 and R1881 coordinately suppressed BIRC5 levels (Fig. 2A). Modest DNA damage (γH2AX) was observed at the higher dose of YM155 with R1881 (Fig. 2A). Reductions in γH2AX and cleaved PARP1 were observed in AR-suppressed cells in the control and low-dose YM155 groups, consistent with reduced replicative stress (48) and DNA repair supportive effects of AR signaling (Fig. 2A; ref. 49). In addition, there was no change in the levels of autophagy (as measured by Beclin and AT5 levels) and negligible changes in apoptosis (as measured by cleaved caspase-3 and PARP1) between the YM155 and control groups (Fig. 2A).

To discount the possibility that the effects we observed are specific to R1881, a potent synthetic androgen, we performed a dose–response curve of LNCaP viability to YM155 with a physiologic dose of testosterone (2.5 nmol/L). The addition of testosterone also increased LNCaP sensitivity to YM155 (Fig. 2B). To determine the time dependence of androgen and YM155 synergy, we preincubated LNCaPs with 10 nmol/L R1881 or vehicle for 24 hours and then determined YM155 IC₅₀ values after 24, 48, and 72 hours (Fig. 2C). Both the vehicle and R1881 groups increased sensitivity to YM155 with longer incubation times, consistent with reported observations that efficacy of YM155 increases with time (50). Given that R1881 potently suppresses BIRC5 without inducing apoptosis and does not induce DNA damage under these experimental conditions (Fig. 2A), these data suggest the mechanism by which androgen therapy sensitizes cells to YM155 is mediated through AR transcriptional activity rather than through genotoxicity or loss of BIRC5.

A recent study demonstrated that YM155 is transported by SLC35F2, a member of the solute carrier group of membrane transport proteins (51). Therefore, one possible mechanism of the R1881-YM155 synergy is through transcriptional upregulation of SLC35F2 by AR. To establish a transcriptional relationship, we performed qRT-PCR to measure SLC35F2 transcript levels in LNCaP and VCaP cells cultured with 10 nmol/L R1881 or ethanol vehicle for 48 hours. In cells treated with 10 nmol/L R1881, transcript levels were approximately 5-fold higher than the vehicle control in LNCaP and approximately 4-fold higher in VCaP (Fig. 3A). Importantly, a similar increase was not seen with 10 nmol/L enzalutamide, suggesting transcriptional upregulation of SLC35F2 is not simply associated with cell-cycle arrest, which occurs in response to either R1881 or enzalutamide (Fig. 3A). To compare expression levels across multiple prostate cancer cell lines, we plotted the RNA-seq counts per million (CPM) of SLC35F2 after exposure to 1 nmol/L R1881 or vehicle control for 24 hours. The prostate cancer cell lines used were androgen-sensitive lines LNCaP and VCaP, castration-resistant lines C42, C42B, LNCAP-ABL, and VCaP-AI, and androgen-independent lines 22Rv1 and E006. As expected, SLC35F2 transcription levels were regulated by androgens with the exception of the androgen-independent lines (Fig. 3B).

The androgen-induced increase in SLC35F2 transcript levels was accompanied by increases in SLC35F2 protein levels in LNCaP cells (Fig. 3C). We then compared levels of SLC35F2 protein across a dose range of R1881 in LNCaP cells. In concordance with the 160 pmol/L of R1881 at which the maximum response to YM155 was achieved (Fig. 1C), SLC35F2 levels were dramatically increased at 100 pmol/L R1881 (Fig. 3D). Of note, SLC35F2 upregulation corresponded with the downregulation of BIRC5 (Fig. 3D).

Despite the androgen-mediated increase of SLC35F2 transcript levels in VCaP, SLC35F2 protein levels were barely detectable in VCaP cells cultured for 48 hours with 10 nmol/L R1881 (Fig. 3E). A band close to the correct size of 45 kD for SLC35F2 was observed by Western blot analysis in VCaP but was not masked by a SLC35F2-blocking peptide, in contrast with AR-regulated product observed in LNCaP cells, which was masked by the blocking peptide (Fig. 3E). However, a faint band corresponding to SLC35F2 was detected in VCaP cells. Notably, SLC35F2 did not appreciably increase or decrease in VCaP cells incubated with the translational inhibitor cycloheximide or the proteasome inhibitor MG132, respectively (Fig. 3F). Taken together, we hypothesize that the lack of synergy between R1881 and YM155 in VCaP is due to the inefficient translation of additional SLC35F2 transcripts.

To provide further evidence that SLC35F2 transcription is directly regulated by the AR, we evaluated AR ChIP-seq data from three published studies of LNCaP cells and one study of VCaP cells evaluating AR-binding sites throughout the genome. Three peaks, termed AR1-3, were observed in intron 1 of SLC35F2 under androgen exposure but not the androgen-depleted control condition (Fig. 4A). We designed PCR primers recognizing these regions and performed qPCR on LNCaP lysates following AR crosslinking and immunoprecipitation. Each peak demonstrated a substantial and significant increase in product following androgen treatment relative to either no androgen control or exposure to enzalutamide (Fig. 4B).

To further examine SLC35F2-mediated YM155 sensitivity, we generated SCL35F2 knockdown and overexpression cell lines and measured viability in response to a range of YM155 doses (Fig. 5A). IC₅₀ values of YM155 shifted from 46.5 nmol/L in GFP-control LNCaP cells treated with vehicle to 679 pmol/L in LNCaP SCL35F2-overexpressing cells treated with vehicle (∼68-fold change; Fig. 5B). The YM155 IC₅₀ only shifted from 679 to 363 pmol/L (∼2-fold) when 10 nmol/L R1881 was added to SLC35F2-overexpressing cells, whereas the IC₅₀ values shifted from 46.5 to 4.35 nmol/L when 10 nmol/L R1881 was added to GFP control cells. In agreement, VCaP sensitivity to YM155 increased from 14.8 nmol/L in the GFP cells with vehicle to 1.49 nmol/L when SLC35F2 was overexpressed. Consistent with the lack of androgen-induced changes in SLC35F2 protein levels in VCaP cells, IC₅₀ values did not shift in VCaP cells dosed with 10 nmol/L R1881 or vehicle (Fig. 5C; Supplementary Fig. S2A).

Conversely, the reduction of SLC35F2 by shRNAs decreased the sensitivity of LNCaP cells to YM155, with IC₅₀ values going from 35.8 nmol/L with the nontargeting shRNA to 267 nmol/L for SLC35F2-directed shRNA#7 and 100 nmol/L for shRNA#8 (Fig. 5D; Supplementary Fig. S2B and S2C). VCAPs displayed a similar shift in IC₅₀ values following SLC35F2 suppression (Fig. 5E; Supplementary Fig. S2D). YM155 has also been reported to be effluxed by the ABCB1 transporter with a consequent resistance to drug treatment (52, 53). Strikingly, ABCB1 overexpression rendered LNCaP cells completely resistant to YM155 regardless of androgen levels, suggesting the ratio of SLC35F2 to ABCB1 expression in tumor cells is an important determinant of sensitivity to YM155 (Fig. 5F).

Identifying the patient population most likely to respond to YM155 is crucial for studies designed to establish YM155 as a targeted therapy for prostate cancer. To this end, we compared the expression of SLC35F2 with a metric of in vivo AR activity, which is based on a panel of AR-regulated genes not including SLC35F2 (30, 31). We analyzed SLC35F2 transcript levels and calculated an AR activity score for 171 CRPC metastases by gene expression microarrays (30). SLC35F2 expression was significantly correlated with AR activity score (r = 0.62, P < 0.001; Fig. 6A). To examine the distribution of SLC35F2 expression across various states of prostate cancer progression, we compared microarray signal intensities for SLC35F2 in primary prostate tumors (n = 11; ref. 32) and CRPC tumors (n = 171; Fig. 6B; ref. 30). Notably, SLC35F2 was broadly expressed in all progression states and spanned a particularly large range of expression in CPRC. A subset of tumors had very high levels of SLC35F2 expression, suggesting that a subset of patients harbor tumors that may be highly responsive to YM155 treatment.

Because SLC35F2 is regulated by androgens, and previous clinical trials of YM155 in prostate cancer were performed in men with castrate levels of testosterone (54), we evaluated the relationship between SLC35F2 expression and intratumor androgen levels in AR-positive PDX models before and after castration. Levels of testosterone and DHT in LuCaP96 and LuCap23 prostate adenocarcinoma PDX models were substantially decreased at days 7 and 21 after castration (Fig. 6C and F) but partially recovered upon progression to castration-resistant disease, consistent with previous reports of increased tumoral androgen synthesis leading to elevated levels of androgens in castration-resistant tumors (55). Concordantly, tumor SLC35F2 mRNA levels decreased upon castration and increased in castration-resistant tumors, although not necessarily to levels present in tumors from intact mice (Fig. 6D and G). Importantly, we observed strong correlations between SLC35F2 expression and tumor testosterone and/or DHT levels in both LuCaP96 (testosterone r = 0.456, P = 0.003; Fig. 6E) and LuCaP23 (DHT r = 0.459, P = 0.009; Fig. 6H), as well as in two additional PDX models (LuCaP35 DHT r = 0.483, P = 0042; LuCap136 DHT r = 0.327, P = 0.08, Supplementary Fig. S3; data for all models summarized in Supplementary Table S1), suggesting that exogenous administration of high-dose androgens has the potential for further regulating SLC35F2 expression. In contrast, SLC35F expression was not associated with tumor androgen levels in the LuCaP86.2 PDX model. This model is known to harbor a genomic AR rearrangement resulting in high-level expression of the exon-skipping androgen-independent AR splice variant ARV567 (56). Notably, we found that SLC35F2 expression was strongly correlated with ARV567 expression (r = 0.745, P = 0.0004), consistent with the ability of ligand-independent AR variants to stimulate the expression of AR target genes (57).

BIRC5/survivin is a critical mediator of cancer cell growth, survival, DNA repair, and therapy resistance (58). Although preclinical studies have repeatedly demonstrated the relevance of BIRC5 as a therapeutic target, clinical efforts to exploit these findings have largely been unsuccessful (59, 60). YM155 was originally developed to suppress the transcription of BIRC5 (50). Subsequent studies measuring the efficacy of YM155 in more than 100 human cancer cell lines and xenografts found YM155 to potently cause cell death across a broad spectrum of cancers with low systemic toxicity in xenografts (61–63). YM155 inhibits BIRC5 by perturbing transcription factor–DNA interactions of Sp-1 (64), ILF3 (65), p50 (66), and NONO (67). However, the anticancer effects of YM155 are unlikely to be due to BIRC5 suppression alone (68). Other targets related to cell survival, such as Mcl1, Bcl-2, and Bcl-xl, have also been reported (66).

A phase I pharmacokinetic study of YM155 on 41 patients, including 9 hormone- and docetaxel-refractory patients, determined YM155 was very well tolerated with very few grade 3 and 4 toxicities. Two CPRC patients had PSA responses, in addition to one complete and two partial responses in 3 patients with non–Hodgkin lymphoma. A subsequent phase II study in 35 progressing, hormone- and docetaxel-refractory, prostate cancer patients found modest single-agent activity of YM155, with 25% of patients achieving prolonged stable disease of ≥18 weeks (54). Importantly, eligibility requirements for the trial required castrate levels of serum testosterone (≤50 ng/mL). It is possible that the drug was inefficiently absorbed by tumor cells due to a loss of AR-mediated SLC35F2 expression. If so, our data suggest YM155 and related compounds may be more effective in the context of intermittent ADT or high-dose testosterone cycles. Intriguingly, our data showing a correlation between tumor androgen levels and SLC35F2 expression also suggests YM155 may be effective in a subset of patients that progress on ADT with elevated levels of intratumoral androgens caused by aberrant expression of steroidogenic enzymes (69).

SLC transporters are among the least studied protein families, and many receptors such as SLC35F2 are considered “orphan” with unknown substrates and physiologic roles (70, 71). However, there is an evolving conception of SLC transporters as mediators of intertissue communication or “remote sensing” of metabolites, signaling molecules, and morphogenic compounds (70). A growing body of literature on model–organism knockouts and human genetic diseases reveals complexity in how transporters are regulated and function in the development of an organism (72, 73). For example, both developing and mature tissues have selective SLC expression profiles, sensitizing, or insulating cells from signals in their environment (72, 74). It is tempting to speculate that SLC transport profiles change with developmental and progression states in cancer, potentially presenting targets of opportunity. In support, it has been reported that SLC35F2 is more highly expressed in non–small cell lung cancers than in surrounding tissue (75).

Although controversial, it is increasingly appreciated that drug transport by passive diffusion across a membrane is often negligible and instead mediated by transmembrane proteins that normally transport metabolites (76). Furthermore, drugs designed with “metabolite-likeness” (77) have a greater chance of achieving standards of efficacy, such as Lipinski's rule of five (78), suggesting an important role for SLC transporters in drug uptake. Several important cancer drugs have known transporters. SLC22A1/SLC22A2 transports daunorubicin (79), imatinib (80), cisplatin, and oxaliplatin (81, 82), and SLC22A16 transports doxorubicin (83). Further research on SLC transporter regulation and their substrates may lead to precision medicine–guided therapies with high antitumor subtype efficacies and low systemic toxicities.

In addition to the potential prostate cancer growth-suppressive effects and quality-of-life benefits of high-testosterone therapy, androgens may selectively sensitize prostate cancer to YM155. Future studies are needed to evaluate the use of YM155 with high-testosterone therapies. For example, it may be optimal to treat patients with cycles of synergistic therapeutic combinations by alternating testosterone/YM155 with ADT and docetaxel (84) or PI3K/AKT inhibitors (85). Structure–activity relationship analysis on YM155 examined the determinants of its selective lethality to transformed cells and derived several dioxonaphthoimidazolium analogues with similar potency (68). Future work is required to determine whether molecules derived from YM155 have similar efficacy and transport mechanisms (66). Finally, it is unclear what other factors regulate SLC35F2 gene expression or how protein levels are being regulated. The VCaP cell line does not upregulate SLC35F2 protein levels to the same extent as LNCaPs with androgens, despite similarly upregulating the transcript. One of the great challenges in the study of SLC transporters is their apparent functional redundancy and observed compensatory regulation (70). The mechanism by which cells and tissues coordinately regulate SLC transport profiles is largely unknown. Moreover, the regulation and functional significance of SLC transporters in cancers as well as their relevance to chemotherapeutics is increasingly appreciated yet still poorly understood (86).

In summary, we demonstrated that YM155 synergizes with endogenously achievable levels of androgens to eliminate prostate cancer cells. This is due to the upregulation of YM155 transporter SLC35F2 by AR transcriptional activity. However, YM155-mediated cell death can be counteracted by high ABCB1 levels. Furthermore, SLC35F2 expression correlates with AR activity in CRPC tumors. Interestingly, some CRPCs express high levels of SLC35F2 despite lower AR activity scores (Fig. 6A), suggesting SLC35F2 could be used as a biomarker for YM155 susceptibility. Given that YM155 is well tolerated and synergizes with various therapies (45), YM155 may be an effective and well-tolerated cotherapy for prostate cancer.

P.S. Nelson is a consultant/advisory board member for Astellas and Janssen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.D. Nyquist, P.S. Nelson, E.A. Mostaghel

Development of methodology: M.D. Nyquist, L. Riggan, E.A. Mostaghel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.D. Nyquist, J. Burns, R. Tharakan, P.S. Nelson, E.A. Mostaghel

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Corella, J. Burns, I. Coleman, S. Gao, C. Cai, E.A. Mostaghel

Writing, review, and/or revision of the manuscript: M.D. Nyquist, J. Burns, E. Corey, P.S. Nelson, E.A. Mostaghel

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Burns, E. Corey, P.S. Nelson, E.A. Mostaghel

Study supervision: P.S. Nelson, E.A. Mostaghel

The authors thank Steven Plymate, MD, for discussions.

This work was supported by the NIH: Pacific Northwest Prostate Cancer SPORE grant no. P50 CA097186 (to E.A. Mostaghel, E. Corey, and P.S. Nelson), P01 CA163227 (to E.A. Mostaghel, E. Corey, and P.S. Nelson), R21 CA194798 (to E. Corey), U.S. Department of Defense awards W81XWH-15-1-0319 (to E.A. Mostaghel), W81XWH-16-1-0206 (to M.D. Nyquist), and the Lucas Foundation.

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.