Patients were included in this study by our inclusion standard as follows: i) All patients met the EAU guidelines for confirming CRPC (29). Serum testosterone levels at castration levels (<1.7 nmol/l) plus either: a) Three consecutive increases in serum prostate-specific antigen (PSA) levels, 1 week apart, leading to two 50% increases over the nadir with PSA levels >2.0 ng/ml; b) the appearance of new lesions and the progression of the primary lesion: New bone lesions and a soft tissue lesion (including prostate, bladder neck, seminal vesicle and other viscera) using TRUS or/and MRI. ii) All patients had available matched PPC and CRPC specimens. iii) All patients had complete clinical data, including PPC and CRPC data. If patients met the inclusion standard 'i', the tissues obtained from the prostate lesions were regarded as CRPC specimens (30). According to the inclusion standard, 45 CRPC and 41 matched PPC samples (4 cases with clinical data of PPC, but without PPC tissue specimens) were collected at the Department of Urology at the First Affiliated Hospital of Chongqing Medical University, Chongqing, China between April, 2008 and September, 2016. CRPC specimens of prostate lesions were obtained from the patients by transurethral resection of the prostate (TURP, 30 cases) or needle biopsy (15 cases). All samples were reviewed by a pathologist for the confirmation of PCa. Gleason's score was evaluated not only in the PPC tissues, but also in the CRPC tissues with the help of a pathologist who was blinded to the clinical data and assessed Gleason's scores in the tissue samples. This study was approved by the Ethics Committee of Chongqing Medical University. Informed consent was obtained from the patients or their family members who agreed to the use of their samples in this study.

All the embedded samples, including the 41 PPC specimens and 45 matched CRPC specimens (30 cases from TURP and 15 cases from needle biopsy), were cut into 5-μm-thick sections. The immunoreactivities of HepaCAM, Notch1 and Hes1 were detected using a standard immunoperoxidase staining procedure (anti-HepaCAM, 1:200; cat. no. 18177-1-AP; ProteinTech, Wuhan, China; anti-Notch1, 1:200, cat. no. ab52627; anti-Hes1, 1:200, cat. no. ab108937; both from Abcam, Cambridge, UK). Staining scoring was semi-quantitatively assessed using staining intensity and was defined as 0, no staining; 1, weak staining; 2, light staining; 4, moderate staining; and 6 and 8, strong staining. Staining scores of ≤1 were regarded as negative expression, while staining scores of ≥2 were regarded as positive expression.

Total RNA was extracted from all cell lines using TRIzol reagent, and reversed transcribed into cDNA using the Prime Script™ RT reagent kit (both from Takara, Dalian, China). SYBR PremixEx Taq™ II kit (Takara) was used for RT-qPCR with the CFX96™ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The sequences of the primers were as follows: HepaCAM sense, 5′-TACTGTAGATGTGCCCATTTCG-3′ and antisense, 5′-CTTCTGGTTTCAGGCGGTC-3′; Notch1 sense, 5′-GAAC GGGGCUAACAAAGAUTT-3′ and antisense, 5′-AUCUUU GUUAGCCCCGUUCTT-3′; Hes1 sense, 5′-GGACTAGTATGCCAGCTGATATAATGGAG-3′ and antisense, 5′-GAAGATCTAGGTGGGCTAGGGACTTTAC-3′; Jagged1 sense, 5′-GTGCCGCCATAGGTAGAGT-3′ and antisense, 5′-CCAGCCAACCACAGAAAC-3′; and β-actin sense, 5′-TGACGTGGACATCCGCAAAG-3′ and antisense, 5′-CTGGAAGGTGGACAGCGAGG-3′. The thermocycling conditions of RT-qPCR were as follows: Initial denaturation, 95°C for 3 min; 95°C for 10 sec, 60°C for 20 sec, 72°C for 20 sec, 40 cycles; final extension: 72 °C for 5 min. The mRNA expression levels were calculated using the comparative 2^−ΔΔCq method (31) and β-actin served as a calibrator. All gene expression experiments were repeated at least 3 times.

Total protein was extracted from the cell lines (please see cell lines below) and tissue samples using RIPA buffer containing the phosphatase inhibitors, NaF and Na₃VO₄, and the protease inhibitor, PMSF (Beyotime Institute of Biotechnology, Beijing, China). The protein concentration was determined using the BCA Protein Assay kit (Beyotime Institute of Biotechnology). Protein samples (50 μg), stacked by 5% SDS-PAGE and separated by 10 or 12% SDS-PAGE, were transferred to PVDF membranes (EMD Millipore, Billerica, MA, USA). After blocking with 5% non-fat milk for 2 h at room temperature, the membranes were incubated with the following primary antibodies overnight at 4°C: Anti-E-Cadherin (1:1,000; cat. no. 3195), anti-N-cadherin (1:1,000; cat. no. 4061), anti-Snail (1:1,000; cat. no. 3895) were obtained from Cell Signaling Technology (Danvers, MA, USA). Anti-Jagged1 (1:1,000; cat. no. ab109536), anti-Notch1 (1:2,000; cat. no. ab52627), anti-NICD (1:500; cat. no. ab83232), anti-Hes1 (1:1,000; cat. no. ab108937) were from Abcam. Anti-HepaCAM (1:500; cat. no. 18177-1-AP) was purchased from ProteinTech. Anti-GAPDH (1:1,000; cat. no. 5174; Cell Signaling Technology) was used as loading control. The membranes were then incubated with the following secondary antibodies for 2 h at room temperature: Goat anti-mouse IgG (1:3,000; cat. no. SA00001-1), goat anti-rabbit IgG (1:3,000; cat. no. SA00001-2) (obtained from ProteinTech). The enhanced chemiluminescent (ECL) kit was purchased from Merck Millipore (Billerica, MA, USA). The intensity level of the protein expression bands was evaluated using Image-Pro plus 6.0.

Human prostate cell lines (RWPE-1, LNCaP and DU145) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The 293A cell line was a gift from Professor Wenli Luo, Key Laboratory of Laboratory Medical Diagnostics, Ministry of Education, Department of Laboratory Medicine, Chongqing Medical University, Chongqing, China. All the cell lines were cultured in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) (both from Gibco-Life Technologies, Carlsbad, CA, USA) and 1% penicillin/streptomycin (Beyotime Institute of Biotechnology). To generate bicalutamide-resistant cells and enzalutamide-resistant cells, the LNCaP cells, one of the androgen-dependent prostate cancer cell strains, were treated with enzalutamide (10 μM) (32) and bicalutamide (10 μM) (Selleck Chemicals, Houston, TX, USA), respectively for at least 6 months. For the generation of bicalutamide-resistant (Bica-R) cells, the cells were first cultured with 1 μM (33), or 5, 10 or 25 μM (34) bicalutamide, respectively. We found that the concentration of 1 μM bicalutamide had almost no effect on the LNCaP cells, and the concentration of 25 μM bicalutamide killed too many cells to induce the cells continually (data not shown). Moreover, similar to treatment with 10 μM enzalutamide, treatment with 10 μM bicalutamide inhibited cell growth by 60 to 70% (data not shown). After screening, we selected the concentration of 10 μM of bicalutamide by ourselves to generate Bica-R cells. The HepaCAM plasmid was transfected into 293A cells using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) according to the manufacturer's instructions. Adenoviruses carrying HepaCAM (Ad-HepaCAM) were stored at −80°C and amplified in 293A cells. The viral fluid was obtained after freezing and thawing the 293A cells repeatedly. The prostate cancer cell strains were transfected with Ad-HepaCAM or Ad-GFP, respectively. After 72 h of incubation, follow-up experiments were performed. The cells were treated with the concentration of 5 μM PF-3084014 for 48 h (Med Chem Express, Monmouth Junction, NJ, USA).

We also constructed docetaxel-resistant cells based on the LNCaP cell line and sequential dual-resistant cells to enzalutamide and docetaxel based on the Enza-R cells. The LNCaP and Enza-R cells were respectively incubated with various concentrations of docetaxel (0.1, 0.5, 1, 2 and 5 nM; Med Chem Express) and the growth of the cells was observed. We found that the concentrations of 0.1 and 0.5 nM docetaxel were not able to inhibit cell growth effectively, and the concentrations of 2 and 5 nM docetaxel killed too many cells to culture continuously (data not shown). Moreover, the concentration of 1 nM docetaxel inhibited the growth of both the LNCaP and Enza-R cells by 60 to 70% (data not shown). Therefore, we selected the concentration of 1 nM of docetaxel as the initial concentration of administration. The LNCaP and Enza-R cells were treated with 1 nM docetaxel every 24 h for 3 weeks. Moreover, when the morphology of the cells exhibited alterations, such as cell membrane shrinkage and even disruption, and acquired a thin and small, polygonal shape, or the cells stopped growing, treatment was halted until the cells recovered. The drug concentration was increased when the cells were able to tolerate the current concentration. Each time the drug concentration was increased, some aliquots of cells were stored. When the cells were killed or contaminated, we resuscitated the aliquots. The frozen cells were removed from liquid nitrogen and placed in a 37°C water bath for 20–30 sec. The thawed cells were added to RPMI-1640. Following centrifugation (1,000 rpm; 5 min), the cells were incubated with a lower concentration of docetaxel. By the stepwise exposure method, the cells were cultured until they were able to tolerate 10 nM docetaxel (35) in 2 months. The cells were maintained in 10 nM docetaxel for at least 4 months. The docetaxel-resistant cells were termed Doce-R cells and sequential dual-resistant cells (resistant to enzalutamide and docetaxel) were termed E+D-R cells.

CCK-8 assay for cell viability, the cells were plated in 96-well plates (2,000 cells/well), and incubated for 12 h. The cells were then cultured with the various treatment agents in each 3 replicate wells. Each well was supplemented with 10 μl CCK-8 reagent (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China). Following incubation for 1 h. Optical density was detected at absorbance of 450 nm using a microplate reader (Bio-Rad Laboratories, Inc., Hercules, CA, USA). CCK-8 assay for the half maximal inhibitory concentration (IC₅₀) of enzalutamide or docetaxel to the cells, the resistant cells (4,000 cells/well) pretreated with various reagents, such as Ad-HepaCAM or Ad-GFP, were seeded into 96-well plates and incubated for 12 h. The cells were then treated with various concentrations of enzalutamide or docetaxel in each 3 replicate wells for 24 h. DMSO (Sigma-Aldrich; Thermo Fisher Scientific, Inc.) was used as the control. For CCK-8 assay for the viability of the Enza-R, Doce-R, E+D-R cells treated with PF-3084014, the cells (4,000 cells/well) were treated with increasing concentrations of PF-3084014 for 48 h (5, 10, 20, 30, 40, 60, 80 and 100 μM) and DMSO for 48 h was used as the control.

The cells (400 cells/well) were plated in 6-well plates, and were consecutively cultured until the numbers of each clone reached 50 cells under a microscope (Nikon, Tokyo, Japan). The clones were stained with 0.05% crystal violet solution (Beyotime Institute of Biotechnology) for 20 min at room temperature. The colony formation experiments were performed at least 3 times.

For Transwell assay, 1.0×10⁴ cells were plated in the upper chamber of the insert with Matrigel (BD Biosciences, San Jose, CA, USA). The cells were incubated with serum-free medium for 48 h. The cells were then stained with 0.1% crystal violet and 4% formaldehyde (Beyotime Institute of Biotechnology). The number of cells, fixed on the bottom membrane of the inserts was counted under a microscope (Nikon). For wound healing assay, 5×10⁴ cells/well were seeded into a 6-well plate. Following 24 h of incubation, the cells were wounded with a yellow pipette tip. The cells were then cultured for 24 h and the wound healing was observed under a microscope (Nikon) at indicated time-points.

A total of 1.0×10⁵ cells/well were plated into a 12-well plate inserted with glass coverslips. Following incubation for 24 h, the cells were fixed in 4% paraformaldehyde for 20 min, and incubated with primary antibody (anti-Notch1 1:50; anti-Hes1, 1:100) (both from Abcam) overnight at 4°C. The cells were then incubated with secondary antibody (Zhongshan Golden Bridge Biotechnology, Beijing, China) for 50 min in a dark room at room temperature. The cell nuclei were stained with DAPI (Zhongshan Golden Bridge Biotechnology) for 10 min. Immunofluorescence images were obtained using a fluorescence microscope (Nikon).

Statistical analyses were performed using SPSS 19.0 software (IBM SPSS Corp., Armonk, NY, USA). All the numerical data are expressed as the means ± SD. Data were analyzed using Kaplan-Meier survival analysis, one-way ANOVA, two-way ANOVA, the Student's t-test, Pearson's correlation analysis, Spearman's correlation analysis, the Mann-Whitney test, McNemer test, the Chi-square test for trend and Pearson's Chi-square test where appropriate. Values of P<0.05 were considered to indicate statistically significant differences.

We collected 45 CRPC samples and 41 matched PPC samples (PPC specimens of 4 patients were unavailable) (Table I). The expression of HepaCAM was detected in the matched PPC and CRPC tissues by immunohistochemistry assay. In total, 71% (32/45) of the CRPC samples exhibited HepaCAM negative staining (staining scores ≤1), whereas HepaCAM expression was negative in 58% (24/41) of the matched PPC samples (Table I and Fig. 1A).

We then determine whether there were any differences in HepaCAM, Notch1 and Hes1 expression levels between the matched PPC and CRPC tissues. In comparison to the matched PPC tissues, the expression of HepaCAM was lost more frequently, (P=0.036; Fig. 1A, panels 1 and 2, and D), and the expression of Hes1 was upregulated (P=0.0237; Fig. 1C, panels 1 and 2, and F) in the CRPC tissues. We failed to observe any differences in the expression of Notch1 between the matched PPC and CRPC tissues (P= 0.063; Fig. 1B, panels 1 and 2, and E). We also evaluated whether the loss of HepaCAM correlated with increased Notch1 and Hes1 expression levels in the CRPC samples using Pearson's linear correlation. As shown in Fig. 1G and H, the loss of HepaCAM negatively correlated with an increase in Notch1 expression (r=−0.652, P<0.01), as well as an increase in Hes1 expression (r=−0.442, P=0.02). The results of western blot analysis revealed a similar result in 14 CRPC samples obtained by needle biopsy and TURP (part of the 45 CRPC samples) (Fig. 2A–C).

To examine the association between the protein expression of HepaCAM and dynamic alterations in gland morphology in the matched tissues, Gleason's score, a system for assessing gland morphology, was evaluated not only in the PPC tissues, but also in CRPC tissues with the help of a pathologist. Our data revealed that, compared to HepaCAM positivity in the CRPC tissues, HepaCAM negativity was associated with higher Gleason scores (P=0.011) (Table I), suggesting that HepaCAM plays an important role in maintaining normal gland morphology. Moreover, the loss of HepaCAM in the CRPC samples, but not in the matched PPC samples was found to be associated with bone metastases (P=0.001) (Table I) suggesting that patients with CRPC with HepaCAM negativity are prone to bone metastases. Kaplan-Meier survival analysis revealed that the median PFS was 39 months (95% CI, 26–52 months) in the patients with CRPC with HepaCAM positivity, while the median PFS was 27 months (95% CI, 20–34 months) in the HepaCAM-negative patients. HepaCAM negativity in the CRPC tissues was associated with a shorter PFS in the patients with CRPC (P=0.039) (Fig. 2D).

In China, the cost of the use of enzalutamide is high, and thus the majority of patients cannot afford treatment with this agent. Patients with PCa are willing to be treated with bicalutamide. In this study, 78% of the patients (35/45) were treated with bicalutamide (Table I). Enzalutamide is widely used in the treatment of PCa in developed countries. Therefore, in this study, we constructed both Bica-R cells and Enza-R cells, as described in the Materials and methods. Western blot analysis was performed to detect the expression levels of HepaCAM in the RWPE-1, LNCaP, Bica-R cells and Enza-R cells. As shown in Fig. 3A and B, HepaCAM was highly expressed in the RWPE1 cells, whereas it was almost undetectable in the other cell lines. To determine the role of HepaCAM in the proliferative capacity of the resistant cells, adenoviral vectors, carryuing the HepaCAM gene, were transfected into the LNCap, Bica-R and Enza-R cells, respectively (Fig. 3A and B). The results of CCK-8 assay revealed that the overexpression of HepaCAM suppressed the proliferation of the DU145, LNCaP, Bica-R and Enza-R cells (Fig. 3C–F). The results of colony formation assay revealed similar results (Fig. 4A). To explore the role of HepaCAM in the invasion and migration of the resistant cells, Transwell assay and wound healing assay were performed using the Bica-R and Enza-R cells. The data revealed that the overexpression of HepaCAM inhibited the invasion and migration of the resistant cells (Fig. 4B and C).

It is well known that AR amplification is responsible for CRPC. Our previous study revealed that HepaCAM decreased the expression of AR (17). Thus, we hypothesized that HepaCAM may reverse the resistance of Enza-R cells to enzalutamide via the downregulation of AR. The IC₅₀ value of enzalutamide for the Enza-R cells was determined by CCK-8 assay. Unexpectedly, however, we failed to observe any significant changes in the resistance of the Enza-R cells to enzlutamide when HepaCAM was overexpressed (Fig. 4D).

As mentioned above, the expression of HepaCAM negatively correlated with Notch signaling in the CRPC samples (Figs. 1G–H and 2A–C). Thus, we hypothesized that HepaCAM may inhibit the proliferation and invasion of the resistant cells through the downregulation of the Notch signaling pathway.

We also investigated members of Notch signaling by western blot analysis and RT-qPCR in the LNCaP and CRPC cells. We found that the levels of Jagged1, Notch1, NICD (protein level) and Hes1 were enhanced in the resistant cells (Fig. 5A–C). To determine the role of Notch signaling in the resistant cells, we treated the Enza-R cells with 5 μM PF-3084014 (a γ-secretase inhibitor) for 48 h, which maintains Notch signaling in inactivation. As shown in Fig. 5D, PF-3084014 induced a decrease in Enza-R cell viability, suggesting that Notch signaling plays an important role in PCa cells which are resistant to treatment. Moreover, when used in combination with Ad-HepaCAM and 5 μM PF-3084014, the viability of the Enza-R cells was inhibited more significantly than the use of either reagent alone (Fig. 5D). Furthermore, we found that the combination of Ad-HepaCAM and PF-3084014 exerted a more potent promoting effect on the protein expression of E-cadherin, and a more potent suppressive effect on the protein expression of N-cadherin and Snail (Fig. 5E and F). Taken together, these findings indicated a synergistic suppressive effect of HepaCAM and PF-3084014 on the proliferation and migration of PCa cells which are resistant to treatment.

We then determined the possible mechanisms responsible for the suppressive effects of HepaCAM overexpression on the survival of PCa cells which are resistant to treatment. The results of immunofluorescence assay revealed that the overexpression of HepaCAM decreased Notch1 and Hes1 expression in the Enza-R cells, which was similar to the effects of PF-3084014. Importantly, when the cells were treated with a combination of Ad-HepaCAM and PF-3084014, the expression levels of Notch1 and Hes1 were downregulated more significantly. Note that in order for Fig. 6 to be more concise, the panels for DAPI staining alone are not shown (Fig. 6A). Western blot analysis and RT-qPCR were performed to further determine the mechanisms responsible for the suppressive effects of HepaCAM overexpression on the viability of the resistant cells. As shown in Fig. B–E, both the mRNA and protein levels of Jagged1, Notch1, NICD (only the protein level) and Hes1 were downregulated when the resistant cells were treated with Ad-HepaCAM or/and PF3084014. Taken together, our data indicated that HepaCAM inhibited the biological behavior of the PCa cells which are resistant to treatment via the downregulation of Notch signaling.

A recent study indicated that overactivated Notch signaling plays an important role in the resistant of PCa to docetaxel, and that the downregulation of Notch reverses docetaxel resistance (22). The results of this study revealed that the overexpression of HepaCAM downregulated Notch signaling in the resistant cells. Thus, we hypothesized that the downregulation of Notch signaling induced by the overexpression of HepaCAM possibly re-sensitizes the docetaxel-resistant cells to docataxel, instead of re-sensitizing the Enza-R cells to enzalutamide.

To confirm our hypothesis, we constructed Doce-R cells and sequential dual-resistant cells (resistant to enzalutamide and docetaxel) (E+D-R cells) as described in the Materials and methods. The IC₅₀ values of docetaxel for the Doce-R, E+D-R and their parental cells were evaluated by CCK-8 assay. As shown in Fig. 7A and B, compared to their parental cells, the LNCaP, Doce-R cells exhibited an 81-fold increase in their resistance to docetaxel, whereas the E+D-R cells displayed a 56-fold increase in their resistance to docetaxel, compared to the Enza-R cells. Moreover, our data indicated that docetaxel inhibited the viability of both the LNCaP and Enza-R cells within the same range of concentrations, indicating that there was no cross-resistance between enzalutamide and docetaxel (Fig. 7A and B).

Our data indicated that the overexpression of HepaCAM suppressed the proliferation of both the Doce-R and E+D-R cells (Fig. 7C–E). When the cells were treated with PF-3084014 followed by Ad-HepaCAM, the inhibitory effects were enhanced in the E+D-R cells (Fig. 7E). However, we failed to observe that the overexpression of HepaCAM restores the sensitivity of the Doce-R and E+D-R cells to docetaxel (Figs. 7A and B, and 8B and C). After the Enza-R and E+D-R cells were treated with 5 μM PF-3084014 for 48 h, the IC₅₀ values of enzalutamide and docetaxel were determined by CCK-8 assay, respectively. Surprisingly, we found that PF-3084014 restored the sensitivity of the Enza-R cells to enzalutamide by 4-fold, and that of the E+D-R cells to docetaxel by 7-fold (Fig. 8A and C) indicating that PF-3084014 may be regarded as a sensitizer of docetaxel and enzalutamide in the treatment of refractory PCa.

To determine whether PF-3084014 exerts antitumor effect on refractory PCa, CCK-8 assay was performed to evaluate the viability of the Enza-R, Doce-R and E+D-R cells following treatment with various concentrations of PF-3084014 (5, 10, 20, 30, 40, 60, 80 and 100 μM) for 48 h. The results revealed that, at an increasing concentration, PF-3084014 exerted a gradual but potent antitumor effect on the Enza-R, Doce-R and E+D-R cells (Fig. 8D–F), indicating that PF-3084014 may be considered as a novel therapy for refractory PCa.