Blocking the Feedback Loop between Neuroendocrine Differentiation and Macrophages Improves the Therapeutic Effects of Enzalutamide (MDV3100) on Prostate Cancer

Prostate cancer is the second-most common cancer in men and one of the most common causes of cancer-related death in men worldwide (1). Androgen deprivation therapy (ADT), the mainstay treatment for advanced prostate cancer, is beneficial during the initial phase of treatment (2). However, progression to the lethal form of the disease, known as castration-resistant prostate cancer (CRPC), is inevitable (3). Enzalutamide (MDV3100), an FDA-approved agent, is a second-generation androgen receptor (AR) inhibitor designed to overcome resistance to first-generation ADT agents (4). In addition, phase III studies of enzalutamide in patients with metastatic CRPC have provided evidence of survival advantages, confirming the relevance of targeting AR in late-stage prostate cancer (5). However, prostate cancer can become resistant to enzalutamide and develop more aggressive phenotypes (6), but the underlying mechanisms are unknown.

Prostate cancer represents a clinically and biologically heterogeneous tumor type. Neuroendocrine differentiation (NED) in prostate cancer has been increasingly observed as a treatment-emergent adaptive response to ADTs, including enzalutamide (7–9). Several biomarkers, including neuron-specific enolase (NSE), chromogranin A (CHGA), and CD56, have been used to detect NED in prostate cancer (10, 11). Although repressor element (RE)-1 silencing transcription factor (REST) inactivation and serine/arginine repetitive matrix 4 (SRRM4; refs. 12, 13), an RNA splicing factor, mediate NED in prostate adenocarcinoma in the context of ADTs such as enzalutamide, further studies are needed to elucidate the mechanisms underlying enzalutamide-induced NED associated with therapeutic resistance in prostate cancer.

In addition to the internal characteristics of tumors, the tumor microenvironment, which provides support to tumors, should be simultaneously considered (14). Among the immune cells recruited to the tumor site, tumor-associated macrophages (TAM), derived from monocytes/macrophages, are present during all stages of tumor progression (15, 16). Furthermore, disseminated TAMs are useful biomarkers of advanced disease, indicating the vital role of TAMs in tumor cell migration (17). In prostate cancer, CC chemokine ligand 2 (CCL2) increases tumor growth and bone metastasis by signaling through TAMs (18). In addition, the administration of a C-C motif chemokine receptor 2 (CCR2) antagonist reverses enzalutamide-induced TAM migration and inhibits prostate cancer invasion (19). Moreover, strategies targeting colony-stimulating factor 1 receptor (CSF1R) in prostate cancer can reverse TAM-mediated resistance to ADT (20). However, researchers have not determined whether enzalutamide-induced NED in prostate cancer is closely associated with TAMs, and the underlying mechanism has not been well examined.

In this study, we investigated the association between enzalutamide-induced NED in prostate cancer and the recruitment and polarization of TAMs to explore the clinical significance and underlying mechanisms. First, a positive correlation between NED and TAMs was observed in ADT-treated prostate cancer or CRPC tissues with high NSE and CHGA expression and in enzalutamide-resistant xenografts in mice. Second, we observed a positive feedback loop between NED and TAMs following treatment with enzalutamide; this feedback loop was based mainly on the IL6/STAT3 pathway and high mobility group box 1 (HMGB1). Third, the concomitant increase in the expression of NSE (or CHGA), CD163, and HMGB1 was a useful indicator of clinicopathologic features, biochemical recurrence (BCR), and disease-free survival (DFS) in prostate cancer patients. Finally, blockade of the association between NED and TAMs by targeting IL6R and HMGB1 inhibited in vivo enzalutamide resistance in an established orthotopic prostate cancer mouse model. Our study thus provides a new potential target that can be used during the development of ADT resistance in patients with advanced or metastatic prostate cancer.

This study included patients who were pathologically diagnosed prostate cancers (including naïve, ADT-treated prostate cancer and CRPC, n = 5/group) in Changhai Hospital (Shanghai, China), and 140 consecutive patients with prostate cancer pathologically diagnosed in Changhai Hospital between 2012 and 2013. Tumor stage and Gleason Scores (GS) were assessed according to the American Joint Committee on Cancer (AJCC) 2002 and the World Health Organization (WHO) classification guidelines. The time to BCR (cutoff: PSA = 0.2 ng/mL) and disease progression identified by MRI, CT, or ECT were selected as the clinical endpoint of biochemical relapse-free survival and disease-free survival, respectively.

The samples were obtained after writing informed consent from patients according to an established protocol approved by the Ethics Committee of Changhai Hospital (Shanghai, China).

The IHC was done as we reported previously (21), primary antibodies as follows were used, mouse anti-NSE (M0873, Dako), mouse anti-CHGA (sc-13090, Santa Cruz Biotechnology) and rat anti-F4/80 (ab6640), mouse anti-AR (ab9474), rabbit anti-CD163 (ab182422), rabbit anti-HMGB1 (ab18256), mouse anti-β-catenin (ab22656), rabbit anti-p-STAT3 (Y705) (ab76315), rabbit anti-IL6R (ab128008), rabbit anti-STAT3 (ab68153), mouse anti-β-catenin (ab22656) from Abcam, respectively. The protein expression was scored by staining intensity and percentage of positively stained cells as we previously reported (21).

C4-2 cells, which was kindly provided and authenticated by Dr. Leland Chung (Cedars-Sinai Medical Center, Los Angeles, CA), were cultured in phenol red–free RPMI1640 medium (11835030, Gibco) containing 10% charcoal-stripped FBS (CS-FBS, 100-119, Gemini) supplemented with 1% penicillin/streptomycin (15140122, Gibco). C4-2-ENZ-R cells were established by long-term culture in medium containing 30 μmol/L enzalutamide (S1250, Selleck Chemicals), then maintained in medium containing 20 μmol/L of enzalutamide. THP-1 was obtained from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China) in June 2016. THP-1 cells were maintained in RPMI1640 medium (C11875500CP, Gibco) supplemented with 10% FBS (10099-141, Gibco) and 1% penicillin/streptomycin with 0.05 mmol/L β-mercaptoethanol (07604, Sigma). All cell lines were cultured at 37°C under 5% CO₂. Under a microscope, C4-2 cells with two or more branches were considered to have an NE phenotype, and cells with rounded shapes were considered to have an epithelial phenotype. Then, the average percentage of cells with an NE or epithelial phenotype was calculated (n = 5 microscope view/group).

All cell lines were authenticated by STR profiling and tested for mycoplasma contamination with a Mycoplasma Detection Kit (B39032, Selleck Chemicals). The most recent tests were performed in August 2017. The cell lines used in this study were within 40 passages. Permanent stocks of the cell lines were prepared and stored in liquid nitrogen until use.

Human monocytes were isolated from human peripheral blood mononuclear cells (PBMC) from healthy donors, and subsequently were sorted via magnetic-activated cell sorting (MACS) with human CD14 microbeads (130-050-201, Miltenyi Biotec) according to the manufacturer's protocol. The purity of monocytes was analyzed using recombinant anti-CD14-PE (130-110-577, Miltenyi Biotec) via flow cytometry. C4-2 cells were seeded into culture dishes at 30% confluencyand had grown to an approximately 70% confluent monolayer 18 hours later. THP-1 cells or peripheral blood mononuclear cells were pelleted via centrifugation at 300 × g. Thereafter, the pelleted cells were resuspended the medium as described above and laid on top of the monolayer of C4-2 cells (number of monocytes: C4-2 monolayer cells = 3:2). After 5-day coculture in the 37°C, 5% CO₂ condition, both the supernatant and cells were collected for subsequent analysis.

The proliferation of cells was evaluated using a CCK-8 (CK-04, Dojindo) according to the manufacturer's instructions. Prior to the assay, the medium was replaced with fresh medium, 10% v/v CCK-8 was added to each well and the samples were incubated at 37°C for 2 hours. Then, the OD values were measured by a microplate reader (EXL800, BioTek Instruments) at an absorbance of 450 nm. The proliferation rates were presented as a proportion of the control value which was detected at the first time point.

A total of 1 × 10⁵ THP-1/monocytes or mononuclear cells were seeded in RPMI1640 medium without FBS into the top chamber of each uncoated transwell, RPMI1640 medium with 20% FBS and conditioned medium (CM) was placed in the bottom chamber. Forty-eight hours after seeding, noninvasive cells in the top chamber were removed with a cotton swab, and the cells on the bottom surface of the membrane were fixed with 4% paraformaldehyde fix solution (E672002, Sangon Biotech) and stained with crystal violet (E607309, Sangon Biotech), then photographed at 400 × magnification. Data represent the means ± SD of three independent experiments.

Short hairpin RNA (shRNA) interference vector pLKO.1-GFP containing an U6 promoter upstream of the shRNA, lentivirus packaging vector pVSVG-I and pCMV-GAG-POL were obtained from Shanghai Integrated Biotech Solutions Co.,Ltd., (Shanghai, China). GFP was used as an internal control with an independent promoter. The C4-2 cell line was cultured in 6-well plates, inoculated at a density of 5 × 10⁴ cells/mL, and transfected with the shRNA-expressing lentivirus (sh-HMGB1, sh-STAT3, or sh-CTNNB1) or control lentivirus at a multiplicity of infection (MOI) of 45. After 72-hour transfection, they were observed and photographed under microscope. Stable C4-2 knockdown of HMGB1, STAT3, and CTNNB1 was also generated using lentiviral constructs. The sequences for shRNA are presented in Supplementary Table S8.

Total RNAs were extracted by RNAiso Plus (9109, Takara) and cDNAs were synthesized using PrimeScript One Step RT reagent Kit (RR037B, Takara). Real-time PCR was performed using SYBR Green Realtime PCR Master Mix (QPK201, Toyobo) with ABI PRISM 7300HT Sequence Detection System (Applied Biosystems). Each measurement was performed in triplicate and the results were normalized by the expression of the β-actin gene. Fold change relative to mean value was determined by 2^−ΔΔC_t. The primer sequences were presented in Supplementary Table S8.

The Western blot analysis was done as reported previously (22), and the following primary antibodies were used: mouse anti-NSE (MAB324, Millipore) and rabbit anti-CHGA (ab68271), rabbit anti-HMGB1 (ab18256), rabbit anti-STAT3(ab68153), rabbit anti-p-STAT3 (Y705) (ab76315), mouse anti-β-catenin (ab22656), rabbit anti-β-TrCP (ab71753) from Abcam, and rabbit anti-GAPDH (#2118S), rabbit anti-β-actin (#4970), rabbit anti-Histone H3 (#4499) from Cell Signaling Technology, respectively. To analyze HMGB1 and β-catenin protein interactions, coimmunoprecipitation experiments were performed using C4-2 cells according to previously published protocols (22) and the following primary antibodies were used: rabbit anti-HMGB1 (ab18256, Abcam) and mouse anti-β-catenin (ab22656, Abcam).

The precipitates obtained in the coimmunoprecipitation (co-IP) experiments for protein quantification were boiled with 50 μL of 50 mmol/L NH₄HCO₃. For in-solution digestion, samples of the protein solutions were first reduced using DTT and all cysteine residues were alkylated with iodoacetamide and cleaned using desalting columns or ethanol precipitation. The samples were then digested with sequencing grade–modified trypsin (V5111, Promega) in digestion buffer (ammonium bicarbonate 100 mmol/L, pH 8.5). The dissolved peptide samples were subsequently analyzed with a NanoLC-ESI-MS/MS system. For the identified proteins reported here, the certainty should be >98% if the identification is based on LC/MS-MS sequencing of one peptide and >99.9% if based on the sequencing of two or more peptides.

Cells were fixed in 4% paraformaldehyde fix solution (E672002, Sangon Biotech) for 15 minutes at room temperature. Then, the cells were permeabilized with 4% Triton X-100 (A110694, Sangon Biotech) for 10 minutes. After a 30-minute incubation with Tris-buffered saline solution containing 5% bovine serum albumin (BSA), samples were immunostained with rabbit anti-HMGB1 (ab18256), rabbit anti-STAT3 (ab68153), rabbit anti-β-catenin (ab32572) from Abcam at 4°C overnight. Subsequently, samples were incubated with Alexa Fluor 488–conjugated goat anti-rabbit (#4412S, Cell Signaling Technology). Nuclei were stained by DAPI (E607303, Sangon Biotech). Fluorescence images were observed and collected under a Leica DM5000B fluorescent microscope (Leica).

This technique was performed using the Duolink II Red Starter Kit (DUO92101, Sigma). A video summarizing the steps of this technique can be found online (www.olink.com/products-services/duolink/how-useduolink). Briefly, cells were prepared as above for immunofluorescence before being incubated overnight at 4°C with primary antibodies against HMGB1 (1:100; 6893, Cell Signaling Technology) and β-catenin (1:50; ab22656, Abcam). Slides were then incubated for 1 hour at 37°C with a mix of the MINUS and PLUS PLA probes. Hybridized probes were ligated using the Ligation-Ligase solution for 30 minutes at 37°C and then amplified utilizing the Amplification-Polymerase solution for 100 minutes at 37°C. Slides were finally mounted using Duolink II Mounting Medium with DAPI.

The IL6, HMGB1, and PSA levels in serum or cell culture medium were measured using ELISA Kit for IL6 (SEA079Hu, Cloud-Clone Corp), ELISA Kit for High Mobility Group Protein 1 (SEA399Hu, Cloud-Clone Corp), and PSA Quantikine ELISA Kit (DKK300, R&D Systems), respectively, according to the manufacturer's instructions.

Chromatin immunoprecipitation (ChIP) assays were performed using the EZ-Magna ChIPA/G Chromatin Immunoprecipitation Kit (17-10086, Millipore), strictly according to the manufacturer's specifications. Briefly, cells were fixed in 1% formaldehyde for 10 minutes at room temperature, and then were lysed and proceed to sonication using 10-second pulses with 50-second rest in between pulses with power setting at 20% (Digital Sonifier 450, Branson) for 10 cycles. Immunoprecipitation was carried out overnight (8 hours) using rabbit anti-STAT3 antibody (ab32500, Abcam). Anti-RNA polymerase II antibody (kit supplied) and normal rabbit anti-IgG (#2729, Cell Signaling Technology) were used as positive control and negative control. Free DNA was obtained after reverse cross-linking and DNA purification of which the product was analyzed by real-time quantitative PCR. Primers complementary to the promoter region of HMGB1 (forward: 5′-GAAGCCCTGCCTTTCCAGAGAC-3′; reverse: 5′-CGCCCAAAGCCGTTGAGATAAG-3′) were used to detect HMGB1 genomic DNA and primers specific to human GAPDH promoter was used as control (kit supplied). Enrichment of the targets was calculated as follows: fold enrichment = 2^{(C_t[HMGB1ChIP] −C_t[IgG])}.

The STAT3-binding sites of the HMGB1 promoter (NM_002128: −1000 to +180 relative to the HMGB1 transcription site) or its mutant sequence (−869 to −859, TCACAATTCCG) was cloned into the pGL3-basic luciferase reporter vector (Promega). C4-2 cells were cotransfected with 10 ng pTK-RL reporter control plasmid and 200 ng pGL3-basic-HMGB1-WT or pGL3-basic-HMGB1-Mut using Lipofectamine 3000 reagents (L3000015, Invitrogen) according to the manufacturer's protocols. Cells were collected 48 hours after transfection and HMGB1 transcription activity was evaluated by measuring luminescence with the Dual-Luciferase Assay Kit (E1910, Promega). Fold induction was derived relative to normalized reporter activity.

Cytokine profiles were measured by Quantibody Human Inflammatory Array 3 (QAH-INF-G3, RayBiotech) that permitted detection of 40 inflammation-associated cytokines. The 40 antibodies against chemokines and cytokines printed in triplicates by the company. Total 9 samples (cell culture medium) were processed: (i) centrifuged at 4°C, 10,000 rpm for 10 minutes, the supernatant was transferred to a new centrifuge tube and centrifuged twice; (ii) proteins concentration was determined; (iii) add 100-μL sample into each well and incubate at room temperature for 30 minutes to block slides; (iv) incubation with Cy3 Equivalent dye-streptavidin and Wash and then perform fluorescence detection; (v) the signals were detected and visualized through use of a laser scanner equipped with a Cy3 wavelength (green channel) by GenePix 4000B. Data extraction can be done using the GAL file that is specific for this array along with the microarray analysis software GenePix.

For tumor inductions, the C4-2 cells incorporated with the luciferase reporter gene (C4-2-Luc) were established by transfection and stable clone selection procedures. A total of 1 × 10⁶ cells mixed with Matrigel (1:1, v/v) were orthotopically injected into both anterior prostates of NOD-SCID mice (Shanghai Laboratory Animal Center, SLAC, China) through an abdominal incision, three weeks after tumor inoculation, using live-animal bioluminescence optical imaging with the IVIS Lumina II imaging system (PerkinElmer) after an intraperitoneal injection of d-luciferin (150 mg/kg; Gold Biotech) in 100 μL of DPBS. Three weeks after the cell injections, the animals were grouped so that the average bioluminescence index was similar in the indicated groups, tumor growth (bioluminescence optical imaging) and serum PSA (ELISA assay of 50-μL blood collected by saphenous vein puncture) measured once a week for the entire duration of the experiment (3–12 weeks). All mice were castrated by bilateral orchiectomy at 4 weeks while the treatment was initiated. These mice in treatment group were treated with enzalutamide (30 mg/kg twice a week) by intraperitoneal injection, combined with or without tocilizumab (5 μg/mL) and shHMGB1, while the control group was treated with vehicle. At the end of week 12, all mice were euthanized 5–15 minutes after the final in vivo bioluminescence measurement.

All experimental animal procedures were approved by the Animal Care and Use Committee of the Second Military Medical University, Shanghai, China. NOD-SCID mice were housed under specific pathogen-free conditions.

Continuous data were presented as median (range) and categorical data were presented as number (%). Numerical data were expressed as the mean ± SD. Statistical differences between variables were analyzed by two-tailed Student t test, χ² test, or Fisher exact test for categorical/binary measures and ANOVA for continuous measures. Survival curve was plotted by the Kaplan–Meier method and compared using the log-rank analysis. Difference was considered significant at P < 0.05. All experiments for cell cultures were performed independently at least three times and in triplicate each time. Data analysis was performed by the GraphPad Prism 5 (GraphPad Software, Inc.) and SPSS 22.0 (IBM Corporation).

To elucidate the association between ADT-induced NED and TAMs, we used IHC to investigate the expression of TAM marker CD163 in prostate cancer samples with differentially expressed neuroendocrine (NE) markers (NSE and CHGA). First, CD163 was expressed at higher levels in ADT-treated prostate cancer tissues (also with high expression of NE markers) than in naïve prostate cancer tissues (also with negative or low expression of NE markers; Fig. 1A and B; Supplementary Fig. S1A and S1B). Second, the expression level of CD163 was higher in the tissues from patients with metastatic prostate cancer after ADT treatment (also with high expression of NE markers) than in the corresponding tissues collected before ADT treatment (Fig. 1C and D; Supplementary Fig. S1C and S1D; Supplementary Table S1). In addition, we compared the expression of CD163 in CRPC tissues with negative or high expression of NE markers, and observed a higher expression of CD163 in the latter group of tissues (Fig. 1E and F; Supplementary Fig. S1E and S1F). Thus, there was a positive correlation between ADT-induced NED and TAMs in prostate cancer. In addition, we examined whether the combined expression of CD163 and NSE (or CHGA) was associated with clinicopathologic features and prognosis in prostate cancer patients. On the basis of the expression levels of CD163 and NSE (or CHGA) in tumors, a separate cohort of prostate cancer patients was classified into 4 groups (n = 140; Supplementary Fig. S1G; Supplementary Tables S2 and S3). The concomitant increase in the expression of the two markers was a helpful indicator of the Gleason score (GS), T stage, capsule penetration, surgical margin, and seminal vesicle invasion (Supplementary Tables S2 and S3). Furthermore, prostate cancer patients who expressed high levels of both NSE (or CHGA) and CD163 displayed poorest BCR and disease-free survival (DFS; Fig. 1G and H; Supplementary Fig. S1H and S1I; Supplementary Tables S2 and S3), indicating that elevated levels of NE and TAM markers predict prostate cancer progression and recurrence.

Furthermore, an orthotopic tumor transplantation model was employed in which C4-2 cells (a prostate cancer cell line) stably expressing luciferase were injected into the prostates of male mice. On the fourth week after inoculation, the mice were castrated and simultaneously treated with enzalutamide (Supplementary Fig. S1J). Initially, the serum prostate-specific antigen (PSA) levels decreased, and tumor growth plateaued; however, at late stages, both PSA levels and tumor growth recovered (Fig. 1I and J), indicating the development of enzalutamide resistance in orthotopic prostate cancer treated with ADT. Moreover, F4/80, NSE, and CHGA were expressed at higher levels in the enzalutamide-resistant xenografts than in the control xenografts (Fig. 1K and L; Supplementary Fig. S1K–S1M). Thus, ADT-induced NED is closely related to TAM infiltration, which accounts for ADT resistance in prostate cancer.

On the basis of these data, we examined whether NE-like prostate cancer cells regulate the recruitment and polarization of TAMs after enzalutamide treatment. Specifically, we continuously treated C4-2 cells (derived from the androgen-sensitive LNCaP parental line under androgen-depleted conditions), which were cultured in medium containing CS-FBS, with enzalutamide and established a stable enzalutamide-resistant cell line (C4-2-ENZ-R) that displayed multiple neurite extensions and robust expression of NSE and CHGA (Fig. 2A and B; Supplementary Fig. S2A and S2B). During enzalutamide treatment, C4-2 cells exhibited increased expression of NSE and CHGA (Fig. 2A and B; Supplementary Fig. S2A). In addition, a larger number of migratory THP-1 cells (a monocyte cell line) or isolated human CD14⁺ monocytes from healthy donors (Supplementary Fig. S2C) were observed in the presence of conditioned medium (CM) from enzalutamide-treated C4-2 cells and C4-2-ENZ-R cells than in the presence of CM from naïve C4-2 cells (Fig. 2C). Macrophages that originate from blood monocytes differentiate into cells of heterogeneous phenotypes, including the classically activated M1 type and the alternatively activated M2 protumor type (23). As expected, THP-1 cells or sorted monocytes cultured in CM from enzalutamide-treated C4-2 cells or C4-2-ENZ-R cells, expressed higher levels of M2 markers and lower levels of M1 markers than did cells cultured in CM from naïve C4-2 cells (Fig. 2D; Supplementary Fig. S2D). Therefore, enzalutamide-induced NE-like prostate cancer cells enhance the recruitment of monocytes to tumors and educate the monocytes toward TAMs.

Next, we employed a coculture system, shown in Supplementary Fig. S2E, to investigate whether NE-like prostate cancer cell–activated TAMs enhance the enzalutamide-induced NED in prostate cancer. First, the cocultured enzalutamide-treated C4-2 cells displayed more neurite extensions than did the enzalutamide-treated C4-2 cells that had not been cocultured with THP-1 cells or sorted monocytes from healthy donors (Fig. 2E; Supplementary Fig. S2F). Second, NSE and CHGA were expressed at higher levels when enzalutamide-treated C4-2 cells were cocultured with NE-like prostate cancer–activated THP-1 cells and sorted monocytes (Fig. 2F). In addition, a Cell Counting Kit-8 (CCK-8) was used to investigate whether the TAMs counteracted the effects of enzalutamide on C4-2 cells. As expected, a higher survival rate was observed for enzalutamide-treated C4-2 cells that were cocultured with TAMs than for cells exposed to enzalutamide alone (Supplementary Fig. S2G). Furthermore, to test whether the NE-like prostate cancer cell–TAM interactions were important for tumor growth and enzalutamide-induced NED in vivo, C4-2 cells cultured with or without an admixture of monocytes were injected into the prostates of NOD/SCID mice, which were then subjected to castration/enzalutamide treatment. The addition of THP-1 or monocytes from healthy donors resulted in a larger tumor size, faster tumor growth, and a more enhanced NED phenotype during enzalutamide treatment in C4-2–derived tumors than in C4-2–derived tumors without exposure to monocytes (Fig. 2G–I; Supplementary Fig. S2H and S2I). Thus, enzalutamide-induced NE-like prostate cancer-activated TAMs reciprocally enhance enzalutamide-induced NED in prostate cancer.

On the basis of the association between enzalutamide-induced NED and TAMs, we further investigated the underlying signaling networks. HMGB1 plays important roles in treatment-induced resistance, tumor development, leukocyte recruitment, and macrophage infiltration (24–26). In addition, HMGB1 is overexpressed in patients with prostate cancer and plays an important role in prostate cancer progression (27). Therefore, we next evaluated whether enzalutamide increased HMGB1 expression and secretion in prostate cancer. Indeed, we observed an increase in HMGB1 mRNA levels in C4-2 cells with enzalutamide treatment in a time-dependent manner (Fig. 3A). To address whether HMGB1 was required for enzalutamide-induced NED in prostate cancer, enzalutamide-treated C4-2 cells were transfected with shRNAs targeting HMGB1 (Supplementary Fig. S3A). As expected, HMGB1 knockdown in C4-2 cells weakened enzalutamide-induced NED (Fig. 3B and C; Supplementary Fig. S3B), which prompted us to further investigate the mechanism by which HMGB1 promotes NED in prostate cancer. Therefore, we immunoprecipitated the HMGB1 protein with an anti-HMGB1 antibody and analyzed the results by Nano LC-ESI-MS/MS. Among the HMGB1-interacting proteins identified by Nano LC-ESI-MS/MS (Fig. 3D; Supplementary Table S4), β-catenin had been reported to drive NED in prostate cancer (28, 29). Co-IP demonstrated that the direct interaction between endogenous HMGB1 and β-catenin mainly occurred in the cytoplasm rather than in the nucleus (Fig. 3E). Furthermore, the interaction between HMGB1 and β-catenin could be enhanced by enzalutamide treatment, which was confirmed by Duolink proximity ligation assays (Fig. 3F). Notably, β-catenin expression was upregulated in enzalutamide-treated C4-2 cells, and HMGB1 knockdown alleviated these effects (Fig. 3C). In addition, β-catenin knockdown ameliorated the enzalutamide-induced NED phenotype and increased the levels of associated marker (Fig. 3B and C; Supplementary Fig. S3B and S3C), suggesting that HMGB1 might trigger NED in prostate cancer by enhancing β-catenin activity. β-Catenin has been reported to be phosphorylated by a cytoplasmic destruction complex, subsequently ubiquitinated by beta-transducin repeat–containing E3 ubiquitin protein ligase (β-TrCP) and then degraded by proteasomes (30). Stabilized β-catenin enters the nucleus to exert its biological function (31). Compared with that in naïve C4-2 cells, increased interaction between HMGB1 and β-catenin and decreased interaction between β-catenin and β-TrCP were observed in the cytoplasm of enzalutamide-treated C4-2 cells, and HMGB1 knockdown reversed these effects (Fig. 3G). Furthermore, enzalutamide increased the abundance of β-catenin in the nucleus of C4-2 cells, but HMGB1 knockdown alleviated this effect (Supplementary Fig. S3D). On the basis of these results, HMGB1 is required for enzalutamide-induced NED in prostate cancer, particularly in the stabilization of β-catenin.

Next, we investigated how prostate cancer that had undergone enzalutamide-induced NED recruited and activated TAMs. It is well known that HMGB1 is secreted by tumor cells in response to therapy (32). As expected, HMGB1 levels were gradually upregulated in both the supernatant (Fig. 3H) and the cytoplasm (Supplementary Fig. S3E) of cells after enzalutamide administration, indicative of enhanced HMGB1 secretion. Furthermore, compared with naïve prostate cancer tissues, ADT-induced CRPC tissues (with high expression of NSE and CD163) displayed increased levels of cytoplasmic HMGB1 (Fig. 3I; Supplementary Fig. S3F–S3H). When a neutralizing antibody against HMGB1 was applied to a coculture of sorted monocytes (or THP-1 cells) and enzalutamide-induced C4-2 cells, monocyte migration and polarization were suppressed (Fig. 3J and K; Supplementary Fig. S3I–S3J). We also assessed whether the concomitant expression of HMGB1 and NSE (or CD163) was indicative of malignant clinicopathologic features in patients with prostate cancer. To this end, all patients in the cohort (n = 140) were classified into 4 groups based on the expression levels of these markers in their tumors (Supplementary Tables S5 and S6). As expected, the concomitant increase in expression of the two markers HMGB1 and NSE (or CD163) was a significant indicator of the Gleason score (GS), T stage, capsule penetration, surgical margin, and seminal vesicle invasion (Supplementary Tables S5 and S6). Furthermore, prostate cancer patients who highly expressed both HMGB1 and NSE (or CD163) exhibited unfavorable BCR and DFS (Fig. 3L and M; Supplementary Fig. S3K–S3L; Supplementary Tables S5 and S6). In addition, the concomitant increase in expression of the three parameters indicated poorest BCR and DFS (Supplementary Fig. S3M and S3N). Thus, HMGB1 is required for enzalutamide-induced NED in prostate cancer and TAM activation.

To identify the inflammatory factors secreted by HMGB1-activated TAMs, the cytokine profiles in the CM from monocytes cultured alone or cocultured with enzalutamide-treated C4-2 cells in the absence or presence of an HMGB1-neutralizing antibody were analyzed using a RayBio Human Cytokine Antibody Array (Supplementary Fig. S4A and S4B; Supplementary Table S7). Compared with the CM from monocytes cultured alone, the CM from monocytes cocultured with enzalutamide-treated C4-2 cells displayed significantly upregulated levels of eight cytokines. In addition, compared with the CM from cocultured monocytes in the absence of the HMGB1-neutralizing antibody, the CM from cocultured monocytes in the presence of HMGB1-neutralizing antibody displayed six differentially downregulated cytokines (Fig. 4A and B). We then compared the significantly differentially expressed cytokines using a Venn plot and identified IL6 and its signaling pathway effectors to be the crucial factors that were altered (Fig. 4C; Supplementary Fig. S4C–S4F). On the basis of the ELISA results, IL6 secretion was elevated in the CM from monocytes or THP-1 cells that were cocultured with enzalutamide-treated C4-2 cells, but the HMGB1-neutralizing antibody attenuated this effect (Fig. 4D). Furthermore, the addition of a neutralizing antibody against IL6 to the CM from cocultured enzalutamide-treated C4-2 cells and monocytes suppressed enzalutamide-induced NED (Fig. 4E–G). Thus, IL6 is the cytokine responsible for enzalutamide-induced NED in prostate cancer cells that are stimulated by HMGB1-activated TAMs.

On the basis of the crucial role of IL6, we examined the mechanism by which IL6 facilitates enzalutamide-induced NED in response to HMGB1-activated TAMs. IL6 exerts its biological effects by binding to IL-6R, which triggers the activation of downstream signaling molecules, including STAT3 (33). In addition, STAT3 is activated by tyrosine phosphorylation (34). The results of Western blot analysis revealed that compared with enzalutamide-treated C4-2 cells cultured alone, enzalutamide-treated C4-2 cells cocultured with monocytes displayed elevated levels of p-STAT3 (Y705; Fig. 4H). However, when the FDA-approved drug tocilizumab (an IL6R inhibitor) was added to cells during the coculture, p-STAT3 (Y705) levels were reduced (Fig. 4H). Accordingly, coculture with monocytes increased the number of neurite extensions and the expression of NSE and CHGA in enzalutamide-treated C4-2 cells, but the inhibition of IL6R with tocilizumab attenuated these effects (Fig. 4E–G). To determine whether STAT3 is required for TAM-mediated enzalutamide-induced NED, STAT3 expression was stably knocked down in C4-2 cells using lentiviruses expressing shRNAs targeting STAT3 (Supplementary Fig. S4G). As expected, although coculture with monocytes increased the number of neurite extensions and the expression levels of NE markers in enzalutamide-treated cells, these changes were not observed in the STAT3 knockdown enzalutamide-treated cells (Fig. 4E–G). Thus, IL6/STAT3 is required for TAM-mediated enzalutamide-induced NED in prostate cancer.

Because HMGB1 was necessary for enzalutamide-induced NED and TAM activation (as shown in Fig. 3; Supplementary Fig. S3), we determined whether the IL6/STAT3 pathway is responsible for HMGB1 expression and transcription. First, the expression of HMGB1 mRNA was upregulated in enzalutamide-treated C4-2 cells cocultured with sorted monocytes from healthy donors or THP-1 cells; upregulation in HMGB1 expression was inhibited when IL6R or STAT3 was blocked (Fig. 5A and B). We then examined whether STAT3 is required to regulate HMGB1 transcription. To this end, the online JASPAR software (http://jaspar.genereg.net) was employed to predict putative transcription factor binding sites on the HMGB1 promoter (Supplementary Fig. S5A). As expected, STAT3 was found bound to the HMGB1 promoter by ChIP assay in C4-2 cells using antibodies against STAT3 (Fig. 5C), indicating that STAT3 directly regulates HMGB1 transcription. In addition, we blocked STAT3 binding on the HMGB1 promoter using reporter constructs harboring mutated variants of the binding sites. We did not observe an increase in HMGB1 transcriptional activity in C4-2 cells in response to enzalutamide treatment in these cells relative to that in naïve C4-2 cells, and these findings were also validated by luciferase assays (Fig. 5D). Furthermore, as shown in Fig. 5E, coculture of enzalutamide-treated C4-2 cells with monocytes augmented HMGB1 promoter activity. However, blocking the IL6/STAT3 pathway with tocilizumab or STAT3 knockdown decreased HMGB1 transcription (Fig. 5E). We also investigated whether TAM-mediated HMGB1 secretion occurs through the IL6/STAT3 pathway. Coculture of enzalutamide-treated C4-2 cells with monocytes increased the concentration of HMGB1 measured using ELISA, whereas the inhibition of the IL6/STAT3 pathway attenuated the increase in HMGB1 secretion (Fig. 5F). Thus, the IL6/STAT3 pathway is required for HMGB1 expression and transcription in enzalutamide-treated prostate cancer cells.

On the basis of the crucial role of the IL6/STAT3 pathway and HMGB1 in enzalutamide-induced NED and TAMs, we examined whether their expression and related pathways are related to ADT resistance in prostate cancer patients. According to IHC analysis, the expression of IL6R, p-STAT3, STAT3, HMGB1, and β-catenin was higher in tissues from patients with metastatic prostate cancer complicated with ADT failure (also with a high expression of NSE and CD163) than in corresponding tissues collected before ADT (Fig. 5G; Supplementary Fig. S5B–S5F), indicating that strategies targeting IL6/STAT3 and HMGB1 might reverse ADT resistance in prostate cancer.

On the basis of the specific associations between enzalutamide-induced NED in prostate cancer and TAMs discussed above, we examined whether a strategy targeting IL6/STAT3 signaling can improve the efficacy of enzalutamide in prostate cancer in vivo. C4-2 cells were injected into the prostates of mice, and after three weeks, all mice were randomly divided into six groups as shown in Fig. 6A. As shown in Fig. 6B and C, the orthotopic tumors of enzalutamide-treated mice became resistant to treatment. However, when monocytes were coinjected with C4-2 cells, the serum PSA levels and tumor growth recovered 1–2 weeks earlier than expected (Fig. 6B and C), indicating that the time to enzalutamide resistance had decreased. Conversely, tocilizumab treatment, HMGB1 knockdown, and, particularly, the combination of the two treatments prolonged the efficacy of enzalutamide on prostate cancer (Fig. 6B and C).

Furthermore, levels of NSE, F4/80, HMGB1, p-STAT3, and STAT3 were elevated in the enzalutamide-resistant xenografts, especially in the presence of sorted monocytes (Fig. 6D; Supplementary S6A–S6E). However, tocilizumab treatment and HMGB1 knockdown attenuated the enzalutamide-induced expression of NSE and F4/80, and the combination of the two treatments exerted even better effects (Fig. 6D; Supplementary S6A and S6B), indicating that drug resistance provoked by enzalutamide-induced NED in the prostate cancer is decreased when the IL6/STAT3 pathway and HMGB1 are inhibited.

Despite the initial effectiveness of ADT in advanced and metastatic prostate cancer, drug resistance is inevitable (35). Enzalutamide, an FDA-approved drug, is an effective treatment for metastatic CRPC. Nevertheless, prostate cancer acquires drug resistance after enzalutamide administration (36). Therefore, an effective remedy is required for patients with ADT-resistant prostate cancer. In this study, targeting IL6/STAT3 signaling with tocilizumab and HMGB1 silencing, which disrupted the link between enzalutamide-induced NED and TAMs, reversed enzalutamide-induced drug resistance in an orthotopic prostate cancer transplantation mouse model (Fig. 6E). IL6 signaling and HMGB1 may be potential targets for overcoming enzalutamide-resistance in prostate cancer, which is also mediated by TAMs.

The mechanisms of ADT-associated resistance are attributed to AR-specific mechanisms (including AR splice variants and amplification; refs. 37, 38) and AR-independent mechanisms (such as glucocorticoid receptor over-expression and NED; refs. 39, 40). According to recent studies, the progression from adenocarcinoma to CRPC is induced by ADT, which is associated with NED (8, 41). In addition, transdifferentiation toward a neuroendocrine phenotype was recently reported to be a mechanism of ADT resistance in CRPC (38). Thus, an understanding of the mechanisms underlying NED will be helpful in identifying patients with prostate cancer who are likely to develop resistance to ADT. In our study, samples from patients with metastatic prostate cancer that had progressed to CRPC after ADT treatment possessed higher NSE and CHGA expression, indicating that the extent of NED in tissues of naïve prostate cancer may predict the sensitivity of prostate cancer to ADT, which should be investigated in future clinical study.

Although most previous studies have focused on the internal mechanisms of the tumor itself, the tumor microenvironment enhances the ability of a tumor to invade, metastasize, and acquire drug resistance (42). TAMs, which originate from blood cells, are driven by chemotactic signals that are released from tumor cells or nonmalignant cells that are present in the tumor microenvironment (43). Bone morphogenetic protein-6 (BMP-6) derived from prostate cancer cell lines increases IL6 expression in macrophages, inducing the NED of prostate cancer (44). ADT, through either castration or enzalutamide treatment, was recently shown to induce a protumorigenic TAM phenotype, and CSF1R inhibitors are an effective treatment for overcoming macrophage-mediated resistance to ADT (20). However, further studies are needed to determine if interactions between ADT-induced NED, such as enzalutamide-induced NED, and TAMs contribute to drug resistance in prostate cancer. According to our study, ADT-induced NED, specifically enzalutamide-induced NED, is closely associated with TAMs in patients with prostate cancer. Furthermore, the IL6/STAT3 pathway and HMGB1 are required for the interaction between enzalutamide-induced NED and TAMs.

Although previous studies have shown that autocrine or paracrine IL6 signaling induces the NED of prostate cancer (45, 46), the mechanisms by which IL6 signaling triggers NED are not completely understood. In this study, we demonstrated TAM-secreted IL6 activated STAT3 by binding to IL6R in prostate cancer cells. Subsequently, STAT3 directly bound to the HMGB1 promoter that facilitated HMGB1 transcription and drove the enzalutamide-induced NED of prostate cancer, which has not been elucidated in previous studies. In addition, some studies have indicated that HMGB1 mediates expression or stabilization of β-catenin (47–49), but the mechanisms are not well elucidated. Our study presented that HMGB1 directly interacted with β-catenin mainly in the cytoplasm, which prevented β-TrCP from binding and stabilized β-catenin. Then activated β-catenin entered into the nucleus and enhanced enzalutamide-induced NED. Thus, a crucial pathway that governs the interaction between enzalutamide–NED of prostate cancer and TAMs has been identified.

To make our findings more relevant to clinical practice, we examined the interaction between the enzalutamide-induced NED of prostate cancer and TAMs in an orthotopic prostate cancer mouse model, which enables the discovery of more accurate and reasonable therapeutic targets to overcome enzalutamide resistance. The process of enzalutamide resistance is advanced by the interaction between TAMs and the NED of prostate cancer, whereas tocilizumab treatment and HMGB1 knockdown prolongs the therapeutic effects of enzalutamide. Thus, the combined targeting of the IL6/STAT3 pathway and HMGB1 ameliorates the enzalutamide resistance of prostate cancer that underwent NED, which provides a new scientific basis for ADT-resistant prostate cancer treatments. Our future research will focus on and explore the source of enzalutamide-recruited TAMs and extend the findings of this study to patient-derived xenografts and data from a larger clinical cohort, which will provide new possibilities for and improve the feasibility of future clinical treatments for patients with prostate cancer.

No potential conflicts of interest were disclosed.

Conception and design: C. Wang

Development of methodology: C. Wang, G. Peng, F. Liu, D.-P. Kong, D. Liu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Wang, G. Peng, H. Huang, F. Liu, D.-P. Kong, K.-Q. Dong, L.-H. Dai, Z. Zhou, K.-J. Wang, J. Yang, X. Gao, M. Qu, F. Zhu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Wang, G. Peng, H. Huang, F. Liu, K.-Q. Dong, K.-J. Wang, H.-R. Wang, Q.-Q. Tian

Writing, review, and/or revision of the manuscript: C. Wang, G. Peng

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Wang, G. Peng, Y.-Q. Cheng, F. Zhu

Study supervision: L. Cao, X.-G. Cui, C.-L. Xu, D.-F. Xu, Y.-H. Sun

The authors thank Dr. Leland Chung (Cedars-Sinai Medical Center, Los Angeles, CA) for providing prostate cancer cell line C4-2. The authors acknowledge the experts (CloudGene Co., LTD, Shanghai) for valuable technical support on bioinformatics analysis. The authors also thank Dr. Jun-hui Ge and Dr. Fu-bo Wang for valuable suggestions. This work was supported by the National Natural Science Foundation of China (no. 81472397, 81773154, and 81301861), National Key Basic Research Program of China (973 Program, no. 2012CB518306), Research Program of Science and Technology Commission of Shanghai Municipality (no. 14411950100), Program for Shanghai Municipal Health and Family Planning Commission Important Diseases Joint Research Project (no. 2013ZYJB0101), Shanghai Natural Science Foundation of China (no. 13ZR1450700), and Innovation Program of Shanghai Municipal Education Commission (no. 2017-01-07-00-07-E00014).

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.