Antibodies to HER2, HER3, AKT1, phospho-AKT1 (S473) and PARP1 were purchased from Cell Signaling (Danvers, MA, United States). Antibodies to cytokeratin-18, phosphor HER3, Ki-67 and GADPH were purchased from Abcam (San Francisco, CA, United States). Anti-pAKT (S473) for IHC staining was from GeneTex (Hsinchu, Taiwan). AR antibody was from Santa Cruz Biotechnology (Dallas, TX, United States). Antibodies to phoso-HER2 and cleaved caspase 3 came from Merck Millipore (Taipei, Taiwan).

PC-3, DU145, 22Rv1 and LNCaP cell lines were obtained from American Type Culture Collection (ATCC, Manassas, VA, United States). All cell lines were authenticated by comparing them with the ATCC database of short tandem repeat DNA profiles. For in vivo measurement of tumor growth, the cell lines were stably transfected with firefly luciferase luc2 of pGL4 (Promega, Madison, WI, United States) driven by a hybrid EF1α/eIF4g promoter through lentivirus infection (Hsu et al., 2012). All cell lines were cultured in RPMI-1640 (Thermo Fisher Scientific, MA, United States) supplemented with 2 mM glutamine, 1 mM sodium pyruvate and 10% fetal bovine serum (Thermo Fisher Scientific). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂.

The plant materials are cultivated in a greenhouse in Annan District, Tainan City, Taiwan and authentication of the W. chinensis plant was performed based on sequencing results of the internal transcribed spacer DNA sequence (ITS) analysis compared with the reference in NCBI databank (Accession: AY947415.1). The whole fresh plants were air-dried, ground and extracted by immersion with ethanol. After condensing, the ethanolic extract was acid-hydrolyzed with HCl at pH 2.0, at 80°C for 30 min to increase aglycone flavonoid content, then neutralized with NaOH and applied to flash LC with a C18 column (SNAP 400 KP-C18-HS Column, Biotage, Uppsala, Sweden) to separate the extract into fractions. To assure consistent WCE quality, the chemical profiles of all WCE lots were analyzed by HPLC-CAD, quantified by triple quadruple LC-MS for contents of wedelolactone, apigenin and luteolin, and analyzed by PSA reporter assay for AR-inhibitory activity.

Wedelia chinensis was dissolved in DMSO for in vitro studies. For in vitro mimic of the castration effect, cells were incubated in medium containing 10% CD-FBS with indicated treatments of vehicle control, 5α-dihydrotestosterone (DHT), or WCE for 2 days. After treatment, cell lysates were prepared by lysis with RIPA buffer supplemented with protease and phosphatase inhibitor cocktails (Thermo Fisher Scientific) for immunoblotting. A quantitative polymerase chain reaction (qPCR) analysis was conducted using a Thermo Fisher scientific SYBR Master Mix and an ABI 7500 real-time PCR system. Raw counts were normalized to GAPDH (ΔCt) and values are presented as fold change relative to vehicle-treated control (2^-ΔΔCt). For cell growth assay, 5000 cells were seeded in 96-well plates overnight and treated with different concentrations of WCE for 4 days. The readout of cell number was measured by CyQUANT Direct Cell Proliferation Assay (Thermo Fisher Scientific).

For luciferase activity, 22Rv1/103E cells (2 × 10⁴) that were stable clone transfected with PSA promoter-Luciferase gene were grown on 96-well plates with RPMI-1640 medium containing 10% CD-FBS for 18 h. The culture was refreshed with the same medium containing treatments as indicated in the figures and grown for another 18 h. Cells were lysed by passive lysis buffer (Promega) and luciferase assay was performed by luciferase assay system (Promega). Luciferase activity was measured by VICTOR³ counter (PerkinElmer, Taipei, Taiwan) and data were normalized to lysate protein as measured by Coomassie (Bradford) Protein Assay Kit (Thermo Fisher Scientific). Inhibition of AR by each treatment was calculated by relative luciferase activity induced by 10 nM DHT as 0% inhibition and vehicle as 100% inhibition.

Athymic nude mice (6 weeks old) were obtained from the National Laboratory Animal Center (Taiwan) and all animal work was conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee, Academia Sinica. For orthotopic implantation, male mice were anesthetized by 2.5% isoflurane and 2 × 10⁵ cancer cells in 20 μL of DPBS were injected into the anterior prostate by using a 27 gauge needle fitted on a 50 μL Hamilton syringe. After 1 week of tumor establishment, WCE was administered through gavage at a dosage of 10 mg/kg. For the in vivo bioluminescence imaging (BLI) of tumors, mice received D-luciferin at 150 mg/kg by intraperitoneal injection at 10 min prior to imaging and tumor growth was monitored once per week. For ex vivo BLI of lung tissues, mice were sacrificed by CO₂ overdose and lung tissues were removed, immersed in DPBS and imaged for 10 s. The images were acquired with a Xenogen IVIS 50 Imaging System and quantitatively analyzed with Living Image 2.50 software (PerkinElmer).

The formalin-fixed, paraffin-embedded tissue sections were deparaffinized and hydrated in a series of graded alcohol to water. After antigen retrieval, tissue sections were incubated with specific antibodies for 1 h at room temperature. PromARK micro-polymer detection systems (Biocare Medical, Concord, CA, United States) were used to detect the primary antibodies. All images were captured with a Zeiss AxioCam HRc camera attached to a Zeiss AxioImager.Z1 microscope (Munich, Germany). For determination of positive cells, the random five fields per tumor sections (n = 6) were examined following staining.

The median inhibition concentration (IC₅₀) values of PSA reporter assay were calculated using GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA, United States) using non-linear regression. The significant differences between the groups were compared with Student’s t-test (two-tailed) by the same software. A P-value of less than 0.05 was considered statistically significant. All values presented in the study were expressed as means with standard error (SEM).

To evaluate the response of PCa to WCE treatment, we first measured the cell cycle distribution following WCE treatment in three different PCa cell lines: AR-positive, androgen-dependent LNCaP, AR-positive but castration-resistant 22Rv1, and AR-negative, hormone-refractory PC-3. After 24 h incubation with WCE, low-dose WCE (25 μg/mL) induced apoptosis in LNCaP and 22Rv1 cells with 12% and 13% increases, respectively, in the sub-G1 population, while causing cell cycle arrest in the G2/M phase in PC-3 cells (Figure 1A). Furthermore, high-dose of WCE (50 μg/mL) induced a 64 and 39% increase, respectively, in the sub-G1 fraction in LNCaP and 22Rv1cells. In apoptotic cells, caspases cleave PARP1 (110-kDa) into 86- and 24-kDa fragments. Immunoblotting of PARP1 showed that the WCE treatment caused a dose-dependent rise and fall of the cleaved 86-kDa fragment and intact PARP1 in 22Rv1 and LNCaP, not PC-3 or DU145 cells (Figure 1B). These results suggest that WCE induced caspase-dependent apoptosis in AR-positive PCa cells, not in AR-negative PCa cells. Due to the promising suppression in AR-positive PCa cells, we next determined the WCE effect on AR expression in LNCaP and 22Rv1 cells and found that WCE significantly inhibited the AR expression in PCa cells (Figure 1B). As determined by qRT-PCR analysis, the WCE effect on AR was at least in part due to a decrease in AR expression at the transcript and protein levels in PCa cells (Figure 1C). Our analysis of WCE responsive genes in 22Rv1 cells using microarray and Gene Set Enrichment Analysis (GSEA) identified a group of androgen responsive genes (Figures 2A,B and Supplementary Table S1). Genes of FKBP5, STEAP1 and TMPRSS2 are AR target genes with functional androgen-responsive elements in their promoters. As analyzed by qRT-PCR, these AR-responsive genes were dose-dependently downregulated by WCE in both 22Rv1 and LNCaP cells (Figure 2C). EZH2 and monoamine oxidase A (MAOA) were correlated with worse clinical outcomes in PCa patients (Xu et al., 2012; Wu et al., 2014). Here, we found that WCE treatment downregulated the mRNA expression of EZH2 and MAOA in AR-positive PCa cells (Figures 2A,C). Interestingly, the WCE regulated genes are also downregulated in AR-negative PCa cells except MAOA and TMPRSS2 did not reach statistical significance. It is worth noting that the truncated AR arising from alternative splicing in 22Rv1 was also downregulated by WCE treatment (Figure 1B). The C-terminally truncated AR protein lacking the ligand-binding domain (ΔLBD) was constitutively nuclear and actively bound to DNA independent of androgens, resulting in androgen-independent expression of AR target genes and PCa growth (Tepper et al., 2002; Ceraline et al., 2004). In analyses of benign and cancerous prostate tissue, the AR-ΔLBD isoforms were found frequently expressed in clinical PCa (Libertini et al., 2007). WCE suppressed the expression of the AR transcripts and protein and thus suffocated the AR signaling in castration-resistant 22Rv1 cells and induced apoptosis, suggesting the potential use of WCE in patients with hormone-naïve and CRPC.

Previous study has indicated that suppression of AR signaling leads to upregulation of HER2/3 signaling and AKT reactivation, which can stabilize the AR and contribute to the eventual failure of ADT (Gao et al., 2016). To determine the role of WCE in the feedback regulation, we examined the expression of HER2 and HER3 upon WCE treatment. Immunoblotting of HER2 and HER3 showed that increased HER2 but decreased HER3 occurred in 22Rv1 cells following WCE treatment, while HER2, phosphorylated HER2, and HER3 proteins were all decreased in LNCaP, PC-3, and DU145 cells (Figure 1B). Downstream of HER2/3, AKT phosphorylation at S473 was suppressed after WCE treatment in all four cell lines (Figure 1B). Next, we determined whether the effect is regulated at the transcriptional level. In 22Rv1 cells, HER2 mRNA was dose-dependently upregulated by WCE treatment (Figure 1C). This is consistent with previous reports that inhibition of AR transcriptional activity in 22Rv1 cells induced HER2 expression (Pignon et al., 2009). However, the mRNA level of HER2 was lowered in LNCaP, PC-3, DU145 cells following WCE treatment compared to vehicle-treated control cells, suggesting that WCE-mediated HER2 regulation involves mechanisms dependent on cell context but probably independent of AR (Figure 1C). WCE-mediated transcription of HER3 in PCa cells is not fully constant with respective HER3 protein level, especially at low dose of WCE treatment, suggesting that posttranslational regulation may also be involved in WCE-mediated HER3 degradation (Figures 1B,C). Although WCE treatment raised HER2 expression in 22Rv1 cells, WCE also reduced the AKT expression to impair HER2/3 signaling (Figure 1B). The mRNA level of AKT was unaffected by WCE, suggesting translational or posttranslational regulation of WCE-mediated AKT expression was involved (Figure 1C). WCE decreased the expression of HER2 and HER3 at the protein and mRNA levels in PCa cell lines except HER2 expression in 22Rv1 cells. Consistently, AR-expressing PCa cell lines were more sensitive to WCE than AR-negative cell lines. Of note, WCE effectively disrupted the crosstalk between the HER2/3-AKT pathway and AR signaling in LNCaP and 22Rv1 cells.

To determine the effect of WCE combined with ADT, cells were grown in a medium stripped by charcoal to mimic ADT, and treated with WCE in the presence or absence of androgen (1 × 10^-9 mol/L DHT). In androgen-dependent LNCaP cells, lack of androgen significantly decreased the expression of AR protein but increased the expression of HER2, HER3 and downstream AKT phosphorylation without induction of apoptosis, (Figure 3A, lane 1 vs. lane 3). Unlike ADT, WCE in the presence of androgen concurrently abolished both AR and HER2/3 signaling and induced apoptosis (Figure 3A, 1ane 2 vs. lane 1); WCE combined with ADT further augmented PARP1 cleavage over WCE or ADT alone (Figure 3A, lane 4 vs. lanes 2, 3). In castration-resistant 22Rv1 cells, ADT slightly decreased AR expression and resulted in activated HER2-AKT signaling (Figure 3A, lane 7 vs. lane 5). Despite the HER2 upregulation by WCE in 22Rv1, WCE decreased the expression of AKT and AR proteins, still leading to cell apoptosis (Figure 3A, lane 6 vs. lane 5). Moreover, WCE combined with ADT decreased the expression of AR, HER2 and HER3, enhancing the cell apoptosis (Figures 3A, lane 8 vs. lane 6, 7). These effects were next examined in vivo using a 22Rv1 orthotopic xenograft model. The anti-androgen drug Casodex moderately delayed the tumor growth leading to 30% lower tumor weight compared to the vehicle control; however, WCE treatment significantly inhibited tumor growth and caused 60% lower tumor weight at end-point (Figures 3B,C).

To consider the clinical feasibility of WCE, we investigated the effect of castration in combination with oral WCE treatment in an LNCaP orthotopic xenograft in a nude mouse model (Figure 4A). WCE or castration monotherapy markedly reduced tumor growth compared with the vehicle control; and a combination of both therapies further reduced tumor growth (Figures 4B,C). Consistent with the tumor growth result, immunohistochemical staining of Ki-67 and cleaved caspase 3 showed that castration or WCE alone inhibited the cell proliferation and induced apoptosis, and combination therapy further enhanced the therapeutic effect (Figures 4D–G). While either monotherapy markedly decreased lung metastasis compared with vehicle treatment, combination therapy was also more effective in inhibiting metastasis than either castration or WCE monotherapy (Figures 5A,B). Only a few micrometastatic PCa foci were detected in the lungs of the WCE or castration monotherapy groups, and none were detected in the combination group (Figure 5C).

To understand whether WCE might affect tumor growth through regulating the physiological levels of androgens, we studied the WCE function under DHT replacement using 12.5 mg of DHT pellet with 60 days release (Innovative Research of America, Sarasota, FL, United States). Despite being used alongside ectopic hormone replacement, WCE remained effective in attenuating tumor growth and metastasis, suggesting that WCE regulated the AR and downstream activities rather than the physiological levels of androgens (Figures 4B,C, 5A–C). To further understand the anti-androgenic activity of WCE in the treated mice, mouse serum containing testosterone, DHT, and WCE compounds was harvested 2 h following the last WCE dose and analyzed by PSA-reporter assay. The active compounds in sera from WCE-treated mice remained functional and inhibited both the physiological and DHT-replaced levels of androgen activity (Figure 5D).

Castration treatment for 12 weeks upregulated HER3 phosphorylation (Y1289), AKT phosphorylation (S473) and AR expression in tumors (Figures 6A–C). This alteration reflects the transition of androgen-dependent PCa to CRPC and enhancement of cancer survival during ADT by upregulating the AR and HER3 signaling pathways (Gao et al., 2016). However, WCE treatment not only suppressed the AR but also suppressed HER2-dependent phosphorylation of HER3 at Y1289, preventing ADT-induced feedback activation of HER2/HER3 (Figures 6A,B). Compared to monotherapy, combination therapy of WCE and ADT showed a more potent therapeutic outcome by suppressing HER3 activation and AR expression. Overall, in contrast with castration, which has unfavorable side effects, WCE treatment decreased the levels of AR, HER2/HER3, and AKT phosphorylation in PCa (Figures 6A–C).

Wedelia chinensis not only impaired the AR expression but also the HER2/3-AKT signaling in castration-resistant 22Rv1, suggestion a potential function in the prevention of the development of CRPC. Next, we investigated the effect of WCE on another CRPC model (Figure 7A). After LNCaP PCa relapsed from castration in mice, we performed necropsies on the mice and cultured the relapsed LNCaP tumor cells, named LNCaP-CR cells. The same procedure was repeated four times and LNCaP-CR4 cells were derived. In vitro, growth of LNCaP-CR4 was not affected by ADT in the medium, whereas growth of the parental LNCaP cells was inhibited by ADT (Figure 7B). Upon treatment with enzalutamide, a second-line anti-androgen, LNCaP-CR4 cells were ∼10-fold more resistant than the parental LNCaP cells (Figure 7C). Therefore, LNCaP-CR4 cells represented a state of CRPC. To this end, WCE elicited the same concentration-effect relationship on the cell growth of LNCaP and LNCaP-CR4 cells (Figure 7D). In addition, LNCaP-CR4 expressed higher levels of HER2, HER3, and AKT phosphorylation compared with parental LNCaP cells (Figure 7E, lane 3 vs. lane 1). Upon ADT, LNCaP-CR4 cells still expressed a high level of AR (Figure 7E, lane 3 vs. lane 5), while ADT significantly decreased AR expression in LNCaP cells, (Figure 7E, lane 2 vs. lane 1). Activation of HER2, HER3 signaling in LNCaP-CR4 might arise from upregulation of gene transcription while the increased AR expression may be influenced by HER2/3 activation (Figure 7F). Even following the failure of ADT, WCE treatment in PCa cells still effectively inhibited the expression of AR, HER2, and HER3 at the protein and mRNA levels, which also resulted in a decrease in AKT phosphorylation and promotion of apoptosis (Figure 7E, lane 6 vs. lane 5, and Figure 7F). In conclusion, the data suggest that exposure to effective doses of WCE can completely inhibit the cell growth in LNCaP and LNCaP-CR4 cells at least in part due to effective simultaneous suppression of AR and HER2/3.