The human prostate cancer cell line LNCaP C4-2 (cat. no. CRL-3314) was purchased from the American Type Culture Collection. Human mesenchymal stem cells (hMSCs; cat. no. C-12974) from the bone marrow were obtained from PromoCell. The C4-2 cells were routinely grown in RPMI-1640 medium (Life Technologies) supplemented with 10% fetal bovine serum (FBS; Invitrogen; Thermo Fisher Scientific, Inc.), hMSCs were cultured in Mesenchymal Stem Cell Growth Medium 2 (cat. no. C-28009; PromoCell) and maintained in humidified incubators (5% CO₂). Penicillin G (100 U•ml⁻¹) and streptomycin sulfate (0.1 mg•ml⁻¹) (cat. no. A5955; Sigma-Aldrich; Merck KGaA) were added to all conditioned media.

To quantify cell dynamics non-invasively, fluorescent proteins (GFP and RFP) were introduced into the cells. The GFP gene was introduced into C4-2 cells and the RFP gene was introduced into hMSCs using lentivirus. C4-2 cells were incubated in RPMI-1640 medium supplemented with 10% FBS in a 10-cm dish and hMSCs were incubated in Mesenchymal Stem Cell Growth Medium 2 in a 10-cm dish. C4-2 cells were transfected with 6 µg of BLIV 2.0 Reporter: MSCV-Luciferase-EF1α-copGFP-T2A-Puro Lentivector Plasmid (cat. no. BLIV713PA-1; System Biosciences, LLC) and hMSCs were transfected with 6 µg of BLIV 2.0 Reporter: pCDH-CMV-MCS-EF1-RFP-T2A-Puro (cat. no. CD516B-2, System Biosciences, LLC) and 4 µg of pPACKH1 HIV Lentivector Packaging Kit (cat. no. LV500A-1; System Biosciences, LLC) using X-tremeGENE HP DNA Transfection Reagent (cat. no. 6366244001; Roche). After overnight incubation, the medium was replaced with fresh medium and the cells were incubated for 3 days. The medium was transferred to a 15 ml tube, followed by centrifugation at 3,000 rpm for 5 min, and the supernatant containing lentiviral particles was passed through a syringe filter (cat. no. SLPES2545S; Hawach Scientific). C4-2 cells were incubated for 3 days in RPMI-1640 medium supplemented with 10% FBS containing lentiviral particles and polybrene (8 µg/ml; cat. no. H9268; Sigma-Aldrich; Merck KGaA). hMSCs were incubated for 3 days in Mesenchymal Stem Cell Growth Medium 2 containing lentiviral particles and polybrene. Since the downstream region of the plasmid contains a puromycin resistance gene, GFP-expressing or RFP-expressing cells were selected using puromycin (1 mg/ml; cat. no. A1113802; Thermo Fisher Scientific, Inc.).

To promote osteogenic differentiation, hMSCs were cultured in Mesenchymal Stem Cell Osteogenic Differentiation medium (cat. no. C-28013; PromoCell). The medium was changed every 3–4 days for 2 weeks. After differentiation, the cells were fixed with 10% formalin solution, stained with 1% Alizarin Red S (cat. no. 5533-25G, Sigma-Aldrich) to confirm extracellular calcium deposits, and stained with BCIP/NBT (cat. no. B5655; Sigma-Aldrich; Merck KGaA) to confirm alkaline phosphatase activity.

RFP-transfected hMSCs were plated (5×10⁴ cells/cm²) onto a chitosan nanofiber-coated culture plate (Hokkaido Soda Co., Ltd.) and incubated with Mesenchymal Stem Cell Growth Medium 2. When cells reached 100% confluency, the medium was changed to Mesenchymal Stem Cell Osteogenic Differentiation medium (Day 0), followed by 14 days of incubation to induce human osteoblasts. On day 14, GFP-transfected C4-2 cells (1.5×10³ cells/cm²) were added and co-cultured with human osteoblasts. On day 15, the medium was changed to androgen-free, phenol red-free RPMI-1640 medium (Life Technologies) supplemented with 5% charcoal/dextran-treated fetal bovine serum (HyClone). Penicillin G (100 U•ml⁻¹) and streptomycin sulfate (0.1 mg•ml⁻¹) were added to all conditioned media. The medium was changed every 3–4 days.

The growth of C4-2 and osteoblasts in the bone microenvironment model was quantified using an imaging system, Cell³ iMager duos (SCREEN Holdings Co., Ltd.) every 3-days. The sum of the green or red intensity area of each well was determined using the Cell³ iMager duos software version 1.4 rev 2.1, (SCREEN Holdings Co., Ltd.).

A total of four ARATs (enzalutamide, apalutamide, darolutamide, and abiraterone), D4A (abiraterone metabolite with AR antagonist) and dutasteride (5α-reductase inhibitor) were selected for drug sensitivity testing using a microenvironment model. All drugs were purchased from Selleckchem. In the drug sensitivity test over time, each drug was added to the culture medium at a concentration of 5 µM dissolved in ethanol from day 15. The final concentration of ethanol in all the drugs and controls was 0.1%. Investigational agents were added every time the medium was changed. In the dose-response curve, each drug was added to the culture medium at each concentration dissolved in ethanol on day 15. The drug exposure time was 48 h, and the antiproliferative effect of the drug on C4-2 cells was determined by comparison with the concentration at 0.01 µM of each drug. CompuSyn software was used to calculate the combination index (CI) at several effective doses (CI=1; additive effect, CI<1; synergy effect, CI>1; antagonistic effect) (14).

Total RNA was extracted using the RNeasy Mini Kit (Qiagen), according to the manufacturer's protocol. RNA quality and quantity were determined using a NanoDrop Lite spectrophotometer (Thermo Fisher Scientific, Inc.). First-strand cDNA was synthesized from 1 µg of total RNA using the RT² First Strand Kit (Qiagen). RNA expression of EMT-related genes was analyzed using the RT² Profiler PCR Array Human Epithelial to Mesenchymal Transition (EMT) (cat. no. PAHS-090ZA; Qiagen), and RNA expression of osteogenesis-related genes was analyzed using the RT² Profiler PCR Array Human Osteogenesis (cat. no. PAHS-026Z, Qiagen). The expression of AR and PSA was analyzed using TaqMan Gene Expression Assays (Applied Biosystems Inc.). The TaqMan MGB probes used in this study were as follows: AR (Hs00171172_m1), PSA (Hs02576345_m1, GAPDH (Hs02758991_g1). Quantitative real-time PCR was performed in triplicate using Applied Biosystems StepOnePlus (Applied Biosystems) according to the manufacturer's protocol. RNA expression levels were determined using StepOnePlus software (version 2.2.2; Applied Biosystems) and normalized to GAPDH expression levels. The following thermocycling conditions were used: initial denaturation at 95°C for 20 sec, followed by 40 cycles at 95°C for 1 sec and 60°C for 20 sec. RNA expression levels were determined using the 2^−ΔΔCT method (15).

Cell samples were collected in RIPA lysis buffer (Santa Cruz Biotechnology Inc.), and protein concentrations were determined using a BCA Protein Assay Kit (Takara Bio Inc.). Each lysate sample (20 µg) was separated by SDS-PAGE and electro-transferred to a polyvinylidene fluoride (PVDF) membrane. Following blocking with 5% non-fat milk in TBS with 0.05% Tween-20 (TBST), the membranes were incubated with each primary antibody overnight at 4°C. The primary antibodies used were as follows: N-cadherin (1/500 dilution, cat. no. 13116), E-cadherin (1/1,000 dilution, cat. no. 3195), Snail (1/1,000 dilution, cat. no. 3879), TGF-β (1/1,000 dilution, cat. no. 3709), and GAPDH (1/2,000 dilution, cat. no. 5174), from Cell Signaling Technology. AR (1/1,000 dilution; cat. no. ab133273) from Abcam. After washing with TBST, the membranes were incubated with HRP-conjugated anti-rabbit IgG secondary antibody (1/5,000 dilution, cat. no. ab6721; Abcam) for 1 h at room temperature. After washing with TBST, membrane signals were detected using an ECL detection system (Amersham Imager 600; GE Healthcare Life Sciences). GAPDH was used for normalization of the protein bands.

Highly purified C4-2 cells were isolated from co-cultured cells using MACS^® (Miltenyi Biotec), using positive selection. Co-cultured cells (1×10⁷) were centrifuged (200 × g for 4 min) and then resuspended in 100 µl of MACS buffer (PBS containing 2 mM EDTA and 0.5% FBS). Biotin anti-human PSMA (FOLH1) antibody (10 µl, cat. no. 342510; BioLegend) was added and incubated for 5 min at 4°C. Cells were washed in 2 ml of MACS buffer and centrifuged again at 200 × g for 4 min. The cellular pellet was resuspended in 200 µl of MACS buffer and 20 µl of anti-biotin microbeads (Miltenyi Biotec). After incubation for 15 min at 4°C, the cells were washed in 2 ml of MACS buffer and centrifuged at 200 × g for 4 min. The cellular pellet was resuspended in 500 µl of MACS buffer loaded onto an LS MACS column (Miltenyi Biotec) in a magnetic field. Pass-through (unlabeled) cells were collected, and the column was washed three times with 500 µl of MACS buffer. The MACS column was removed from the magnetic stand, and PSMA-positive (labeled) cells were eluted with 5 ml of MACS buffer. The PSMA-positive cells and pass-through cells were analyzed by flow cytometry.

An allophycocyanin (APC)-conjugated anti-human PSMA (FOLH1) antibody (cat. no. 342508, BioLegend) was used in this study. Flow cytometric data were acquired using a CytoFLEX S System cell (Beckman Coulter) and analyzed using FlowJo software (FlowJo, LLC).

C4-2 cells and osteoblasts were co-cultured for 24 h, and abiraterone alone or a combination of abiraterone and dutasteride were added to the culture medium at a concentration of 5 µM dissolved in ethanol on day 15. Cell suspensions were collected on days 17 and 28 (2 and 13 days after drug addition), and C4-2 cells were isolated by MACS^® using an anti-PSMA antibody. C4-2 cells (1×10⁵ cells) were washed with PBS by centrifugation at 200 × g for 5 min. They were then washed twice under the same conditions. The cell pellets were lysed with 50 µl of 70% acetonitrile and then centrifuged at 17,4000 × g for 5 min at 4°C. The resulting supernatants were collected, diluted to 1/20 with 80% methanol in water, and analyzed by liquid chromatography (LC)-electrospray ionization (ESI)-time-of-flight/mass spectrometery (TOF-MS). The LC system used was a Nexera SFC/SFE-HPLC system (Shimadzu). The ESI-TOF/MS system used was Impact II (Bruker Daltonics). MS was operated in the positive ion mode. The concentrations of abiraterone and abiraterone metabolites, including D4A, 3-keto-5α-abiraterone, 3α-OH-5α-abiraterone, and 3β-OH-5α-abiraterone in the supernatant were measured using the exact mass value of the protonated molecular ion of each compound.

Details of this method have been reported elsewhere (16). Briefly, abiraterone and its metabolite (D4A, 3-keto-5α-abiraterone) were measured before and after combination treatment with abiraterone and dutasteride in a phase II clinical trial in patients with CRPC (UMIN Clinical Trial Registry: UMIN000027795). Pharmacokinetic data in patients who were judged as effective in combination therapy were compared with the data from our in vitro model.

All numerical data are presented as the mean ± standard deviation. Unpaired numerical data were compared using an unpaired Student's t-test (two groups) or analysis of variance (ANOVA; more than two groups). We used Tukey test as the post-hoc test following ANOVA. The synergistic effect of the combination of drugs was calculated using the CompuSyn software (Combosyn Inc.). Statistical analysis was performed using JMP software (Pro.13; SAS Institute, Inc.). P-values were two-sided, and statistical significance was defined as P<0.05, in all tests.

To demonstrate the bone microenvironment model of prostate cancer, prostate cancer cells and osteoblasts were co-cultured on a microfiber scaffold to examine the characteristics of the co-culture system. We used a microfiber scaffold composed of a three-dimensional chitosan fiber matrix. The scaffold was composed of random fibers with an average diameter of ~200 nm. Using a three-dimensional microfiber scaffold, cells may proliferate in a three-dimensional morphology. To recapitulate the bone metastasis microenvironment of castration-resistant prostate cancer, we used C4-2, a cell line of castration-resistant prostate cancer. C4-2 maintains AR activation and signaling through de novo intratumoral steroidogenesis (17,18). Osteoblasts were differentiated from human mesenchymal stem cells (hMSCs), which were collected from the bone marrow using a special medium. Differentiation of hMSCs into osteoblasts was confirmed by Alizarin red S staining and alkaline phosphatase staining (Fig. S1). To facilitate separate quantification of C4-2 and osteoblast cells after co-culture, we stably transfected C4-2 cells with GFP and hMSCs with RFP using lentiviral vectors. We used flow cytometry to determine whether GFP was successfully introduced into C4-2 cells or RFP was introduced into hMSCs (Fig. S2). Fig. 1-A shows the schema of the bone microenvironment model. GFP-transferred C4-2 cells and RFP-transferred osteoblasts differentiated from hMSCs were co-cultured with chitosan fiber matrix 3D culture. The medium was changed to androgen-free medium 24 h after starting the co-culture to reflect the bone metastasis environment of castration-resistant prostate cancer. The growth of C4-2 cells and osteoblasts was quantified using a live-cell imaging and analysis system, Cell3 iMager duos (SCREEN). We could non-invasively quantify the static survival of osteoblasts and could maintain a continuous count of C4-2 cells for a maximum of 30 days (Fig. 1B). Fluorescence images at days 15, 32, and 46 showed that C4-2 grew to form colonies (Fig. 1C). The C4-2 colonies were physically in contact with the osteoblasts. We also examined the effects of co-culturing with osteoblasts. Compared with monoculture, co-culturing with human osteoblasts demonstrated a significant growth enhancement on C4-2 cells (t-test, P<0.01, day 52), but there was no obvious difference in morphology of the C4-2 cells (Fig. S3).

We used this model to evaluate the drug sensitivity of androgen receptor-axis-targeted agents (ARATs) and D4A (abiraterone metabolite with AR antagonist). First, we compared the growth curves of the GFP-transfected C4-2 cells (Fig. 2A). Significant difference in growth inhibition was observed from day 18 to day 25 (ANOVA, P<0.01), and significant growth inhibition was observed with the addition of enzalutamide (t-test, P<0.01), apalutamide (P<0.01), darolutamide (P<0.01), abiraterone (P<0.01), or D4A (P<0.01) compared to the control on day 28. Significant differences in growth inhibition were observed among all the investigational agents except between enzalutamide and apalutamide (P=0.195) and between darolutamide and abiraterone (P=0.118). Second, we compared the dose-response curves of each investigational agent against C4-2 cells (Fig. 2B). Significant difference in growth inhibition between ARATs was observed at concentrations of 10–100 µM (ANOVA, P<0.01). The 50% inhibition concentration (IC₅₀) of each drug calculated from the dose-response curve is shown in Table I. The IC₅₀ values of each investigational agent were compared, and significant differences were found among the investigational agents (t-test, P<0.05), except between enzalutamide and apalutamide (P=0.982) and between darolutamide and abiraterone (P=0.106). The results were similar to the drug effects of the growth curves (Fig. 2A). Next, we compared the drug sensitivity of C4-2 cells with and without co-culture with human osteoblasts in chitosan nanofiber-coated 3D culture plates. Co-culture with human osteoblasts had a inhibitory effect on the growth of all investigational agents on day 28 (Fig. 2C).

We examined the changes in mRNA and protein expression in C4-2 cells when co-cultured with osteoblasts in this model. Highly purified C4-2 cells were isolated from co-cultured cell suspensions by magnetic-activated cell sorting (MACS^®; Miltenyi Biotec). This was performed using positive selection using specific binding of anti-PSMA antibodies to C4-2 cells. The highly specific separation of C4-2 cells was confirmed by flow cytometry (Fig. S4). We compared the mRNA expression of C4-2 cells isolated from co-culture cell suspensions and monocultured C4-2 cells using RT-PCR (Fig. 3A). We focused on TGF-β and EMT-related genes, because accumulating evidence suggests that these molecules play an important role as promoters of tumor cell survival and development in the bone microenvironment. The mRNA expression of TGF-β (TGF-β1, TGF-β2) was significantly higher in the co-culture than in the monoculture. Similarly, the expression of EMT-related genes (TWIST1, Snail2, and N-cadherin) was also significantly higher in the co-culture than in the monoculture. The expression of AR and PSA (downstream gene of AR) was significantly higher when co-cultured with osteoblasts. The protein expression of C4-2 cells isolated from co-culture cell suspensions and monoculture was compared by western blotting (Fig. 3B). An identical expression pattern was observed between the mRNA and protein expression of each gene. We also examined the mRNA expression of osteoblast stimulatory factors in the co-culture. mRNA expression of bone morphogenetic protein 2 (BMP2) and vascular endothelial growth factor (VEGF-A, VEGF-B) were significantly higher in the co-culture than in the monoculture (Fig. S5).

We used this model to evaluate the therapeutic effects of the combination of abiraterone and dutasteride in the bone microenvironment. First, we compared the growth curves of GFP-transferred C4-2 cells among abiraterone, dutasteride, and both combinations (Fig. 4A). The combination of abiraterone and dutasteride had a greater colony inhibitory effect on tumor growth (ANOVA, P<0.01, day 46). Second, we compared the dose-response curves of each investigational agent against C4-2 cells (Fig. 4B). To examine whether this effect was additive or synergistic, dose-dependent effects with constant ratio design and combination index (CI) values were calculated according to the Chou and Talalay median effect principal (14). Dutasteride synergistically enhanced the inhibitory effect of abiraterone on the colony growth of C4-2 cells (Fig. 4C). We then examined the concentrations of abiraterone metabolites in C4-2 cells treated with abiraterone alone or abiraterone and dutasteride. C4-2 cells were isolated from cell suspensions on days 17 and 28 (2 and 13 days after drug addition). C4-2 cells were lysed with acetonitrile, and the supernatants were analyzed using LC-ESI-TOF/MS. The concentrations of each investigational agent used are listed in Table II. On day 17, the combination of dutasteride significantly decreased the concentration of 3-keto-5α-abiraterone (t-test, P<0.05) and 3β-OH-5α-abiraterone (P<0.01). On day 28, the combination of dutasteride tended to decrease the concentration of 3-keto-5α-abiraterone (P=0.06) and a significant decrease in 3β-OH-5α-abiraterone (P<0.01). These results suggest that the combination of abiraterone and dutasteride had a more potent inhibitory effect on tumor growth by reducing the androgen receptor agonist activity of 5α-abiraterone. In a phase II study of abiraterone and dutasteride combination therapy in CRPC patients (16), effective cases where PSA was decreased by the combined use of abiraterone and dutasteride showed a decrease in serum 3-keto-5α-abiraterone concentration (Fig. S6), similar to the results of the microenvironmental model using the chitosan culture substrate.