Nrdp1-Mediated Regulation of ErbB3 Expression by the Androgen Receptor in Androgen-Dependent but not Castrate-Resistant Prostate Cancer Cells

Because prostate cancer (PCa) cells are initially dependent on androgens for growth, the standard therapy for recurrent PCa is the pharmacologic removal of circulating androgens (androgen withdrawal, AW). Although initially effective, this therapy ultimately fails, indicating the development of castration-resistant PCa (CRPC). The treatment options for patients who fail AW therapy are limited; hence, there is an urgent need for the elucidation of molecular pathways leading to CRPC. Previous studies concluded that AW resulted in cell cycle arrest, whereas CRPC is associated with a release from that arrest (1). In this study, we show that AW results in an increase in the receptor tyrosine kinase ErbB3, which induces an increase in androgen receptor (AR) transcriptional activity and cell cycle progression.

The ErbB receptor tyrosine kinase family regulates proliferation and survival in PCa (2). It consists of the type 1 tyrosine kinases ErbB1/human epidermal growth factor receptor (HER)1/epidermal growth factor receptor (EGFR), ErbB2/HER2/neu, ErbB3/HER3, and ErbB4/HER4 (3). PCa cells express EGFR, HER2, and ErbB3, but not ErbB4 (2, 3), which are activated by ligand binding, dimerization, and phosphorylation. ErbB receptors, except HER2, have specific ligands (4); however, ErbB3 is unique in that its tyrosine kinase domain is functionally defective so it must heterodimerize with other ErbB receptors for signaling activity (4). Despite this, multiple studies suggested that ErbB3 plays a role in promoting PCa. Treatment of androgen-dependent LNCaP cells with the cytokine interleukin-6, known to promote PCa progression, stimulates HER2 and ErbB3 (5). Phosphatidylinositol 3-kinase, which regulates cell survival by activating Akt, associates with ErbB3 (6), whereas microarray analysis showed increased ErbB3 expression in PCa compared with normal prostate (7), and immunohistochemical analysis showed that ∼90% of PCa tissues displayed significant ErbB3 staining (8–10). Alternate splicing caused multiple soluble, truncated forms of ErbB3 (11, 12), and this receptor tyrosine kinase was also shown to localize to the nucleus (8, 13). These studies point to the importance of ErbB3 in PCa signaling.

Substantial evidence also underscores a link between ErbB3 activation and AR activity. The AR is known to remain active in CRPC, and activation of ErbB3 in a mouse model of PCa was associated with AR phosphorylation (10), whereas activation of HER2/ErbB3 heterodimers modulated AR transcriptional activity (14) and promoted AR transactivation of reporter genes (15). The effects of ErbB3 on the AR are likely mediated by suppression of the ErbB3-binding protein Ebp1 (16, 17), which inhibits both E2F1 and AR activity (17).

In this study, we make the novel observation that ErbB3 levels are increased during AW, and that the effect of AR on ErbB3 is mediated by the RING finger E3 ubiquitin ligase neuregulin receptor degradation protein-1 (Nrdp1), which was discovered as an ErbB3-interacting protein by yeast two-hybrid analyses (18), but was not, until now, investigated in PCa. Nrdp1 associates with ErbB3 and mediates its ubiquitination and rapid degradation in a ligand-independent manner (19), thus regulating steady-state levels in breast cancer cells (20). We show that although the AR regulates Nrdp1 transcription in androgen-dependent cells, this regulation is lost in AR-positive CRPC cells due to the suppression of Nrdp1 by elevated ErbB3. Our data explain how ErbB3 levels are increased in CRPC: during AW, AR levels are decreased, and the suppressive effect of the AR on ErbB3 is relieved, resulting in an increase in ErbB3 levels. However, in AR-positive CRPC cells, ErbB3 levels remain high because elevated ErbB3 suppresses Nrdp1 expression and the AR loses control of Nrdp1 transcription; therefore, it cannot suppress ErbB3 levels even when AR levels rebound.

Tissue microarrays from the University of Michigan Prostate Cancer Specialized Program of Research Excellence were constructed from benign prostate (n = 36 cores in triplicate, i.e., total 108), high-grade prostatic intraepithelial neoplasia [HGPIN; n = 21 cores in triplicate (63 total)], prostate tumor tissue [n = 65 cores in triplicate (195 total)] obtained by prostatectomy from men with localized PCa, and CRPC metastatic lesions [n = 68 cores in triplicate (204 total)] from “warm” autopsies (mean 3 h lapsed from death to commencement of autopsy) of patients succumbing to castrate-resistant disease (21, 22). Tissue details, immunohistochemical techniques, and statistical analyses were described earlier (23) and in Supplementary Materials.

LNCaP, PC-3, DU-145, CWR22Rv1, and RWPE-1 cells were purchased from the American Type Culture Collection, whereas C4-2 cells were from UroCor. pRNS-1-1 cells were from Dr. Johng Rhim, University of the Health Sciences, Bethesda, MD (24), whereas PC-346C cells were provided by Dr. W.M. van Weerden, Josephine Nefkens Institute, Erasmus MC, Rotterdam, the Netherlands. All cells were cultivated in RPMI 1640 with 5% serum unless otherwise noted. CRPC sublines of LNCaP cells (LNCaP-AI cells) were developed by prolonged culture of LNCaP cells in phenol red–free RPMI 1640 with 5% charcoal-stripped fetal bovine serum (CSS; ref. 25). The cell lines were not tested or authenticated specifically for this study; however, they have been authenticated elsewhere (24–27). RPMI 1640 and fetal bovine serum (FBS) was from Invitrogen, whereas CSS was from Gemini Bio-products. 4,5α-Dihydrotestosterone (DHT), cycloheximide, and concanamycin were obtained from Sigma-Aldrich. Casodex (bicalutamide) was from AstraZeneca. Rabbit polyclonal EGFR, HER2, ErbB3, β-actin, AR, and α-tubulin antibodies were from Santa Cruz Biotechnology. Rabbit polyclonal anti–phospho-Akt (Ser 473), anti–phospho-EGFR (Y1068), anti–phospho-HER2 (Y1248), and phospho-ErbB3 (Y1289) were from Cell Signaling Technology. Affinity-purified rabbit antibodies to Nrdp1 were previously described (18).

Plasmids expressing wtAR or AR(T877A) resulted from cloning, into pCEP4, the full-length wild-type AR cDNA or AR cDNA whose protein product contains a Thr→Ala mutation at the 877 residual position isolated from LNCaP cells (24). pCDNA-HER2 and pCDNA3-ErbB3 plasmids were from Dr. John Koland, University of Iowa, Iowa City, IA (28). Wild-type AR (pAR0) plasmid was from Dr. Albert Brinkman, Erasmus University, the Netherlands (29). Plasmids encoding human Nrdp1 COOH terminally tagged with a FLAG epitope and a pSuper-Nrdp1 RNAI were described earlier (18, 20, 30). Another small interfering RNA (siRNA) pool against human Nrdp1 was from Santa Cruz Biotechnology (against 5′-CAA-GCA-GUA-UCC-CUG-UUC-ATT-3′, 5″-CUG-UGA-CUG-UAG-UUA-GUU-ATT-3′, and 5′-CUU-CCU-CUC-UUC-CUG-UGA-ATT-3′). siRNA against human AR (Invitrogen): hAR1, 5′-GAC-UCC-UUU-GCA-GCC-UUG-CUC-UCU-A-3′ and 5′-UAG-AGA-GCA-AGG-CUG-CAA-AGG-AGU-C-3′; hAR2, 5′-GCC-UUG-CUC-UCU-AGC-CUC-AAU-GAA-C-3′ and 5′-GUU-CAU-UGA-GGC-UAG-AGA-GCA-AGG-C-3′. siRNA against ErbB3 (Santa Cruz Biotechnology): 5′-CCA-AUA-CCA-GAC-ACU-GUA-CTT-3′. Control siRNA: pool of four scrambled nonspecific siRNA (siCONTROL, Santa Cruz Biotechnology.). Reverse transcription-PCR (RT-PCR) was carried out using the following primers: c-erbB3-F, 5′-GGA-GTA-CAA-ATT-GCC-AAG-GG-TA-3′; c-ErbB3-R, 5′-CAG-GTC-TGG-CAA-GTA-TGG-AT-3′; EGFR-F, 5′-GAG-AGG-AGA-ACT-GCC-AGA-A-3′; EGFR-R, 5′-GTA-GCA-TTT-ATG-GAG-AGT-G-3′; β-actin-F, 5′-ACT-CTT-CCA-GCC-TTC-GTT-C-3′; β-actin-R, 5′-ATC-TCC-TTC-TGC-ATC-CTG-TC-3′; Nrdp1-F, 5′-GCA-GTG-GAG-TCT-TGG-AGG-AG-3′; Nrdp1-R, 5′-GCC-TTT-AGC-AGC-TGG-ATG-TC-3′.

Mice (4–5-week-old, nu/nu athymic male) were obtained from Harlan Sprague-Dawley, Inc. and implanted s.c. with sustained release testosterone pellets (12.5 mg, 90-d release; Innovative Research of America). Suspensions of CWR22 cells were made in 50% Matrigel solubilized basement membrane (BD Biosciences), and xenografts were established by s.c. injections of 2.5 × 10⁶ cells/site into both flanks. When palpable tumors were observed, animals were treated with (a) vehicle or (b) bicalutamide, delivered by oral gavage at a dose of 50 mg/Kg, 100 μL per dose, five times per week, dissolved in ethanol, and delivered as a suspension in peanut oil. After 2 weeks, the mice were euthanized, tumors were collected and divided into sections for paraffin embedding, and snap frozen in liquid nitrogen.

For additional methods, see previous publications (31, 32) and Supplementary Text.

We stained tissue microarrays representing benign prostates, HGPIN, and localized prostate tumors obtained by prostatectomy, as well as metastatic lesions from warm autopsies of men who died with hormone refractory PCa, with an anti-ErbB3 antibody that recognizes cytoplasmic ErbB3. Only the tumor tissue stained strongly for ErbB3, which was seen exclusively in the epithelial cells (Fig. 1, top). The extent of ErbB3 staining increased from benign prostate (mean staining score, 1.65 ± 0.80) to PIN (1.9 ± 0.61) to localized PCa (2.37 ± 0.63) to metastatic CRPC specimen (2.51 ± 0.74; benign versus localized tumors: P = 0.0001; Benign versus CRPC: P = 0.0001; PIN versus loc. tumors: P = 0.0039; PIN versus CRPC: P = 0.0005; Supplementary Table S1; Fig. 1, bottom). Similarly, comparison of ErbB3 levels in various prostate-derived cell lines revealed that “normal-like” RWPE-1 cells expressed very little ErbB3, whereas its expression was significantly higher in androgen-dependent LNCaP PCa cells and further increased in its castration-resistant subline LNCaP-AI (see Supplementary Fig. S1A and B; Fig. 1A, top). Similarly, CRPC cell lines PC-3 and DU-145 cells expressed very high levels of ErbB3 compared with pRNS-1-1 cells derived from a normal prostate (Fig. 1A, bottom), supporting the assertion that ErbB3 increased with PCa progression.

To determine the functional significance of ErbB3 increase, LNCaP cells were stably transfected with plasmids encoding ErbB3 (pCDNA3-ErbB3) or HER2 (pCDNA3-HER2; Fig. 2A). The overexpressors used in subsequent experiments were chosen from a screen of multiple clones to select the ones that expressed HER2 and ErbB3 at levels comparable with that in LNCaP-AI cells (two ErbB3 clones shown in Fig. 1B). Overexpression of ErbB3 in LNCaP cells increased AR expression and AR transcriptional activity as determined by reporter gene assay on a prostate-specific antigen promoter construct (Fig. 1B) in an EGFR- and HER2-dependent manner (Supplementary Fig. S1C).

AW in LNCaP cells was induced by culture in CSS-containing medium, causing growth arrest, which was relieved by ErbB3 overexpression, (Supplementary Fig. S2B; Fig. 2A) similar to HER2 (Supplementary Fig. S2A), whereas inhibition of ErbB3 using a human ErbB3 siRNA pool in LNCaP-AI cells significantly decreased their growth rates (Fig. 2C). Overexpression of ErbB3 in normal-like RWPE-1 cells also resulted in increased proliferation (data not shown). These results indicate that ErbB3 regulates cell proliferation in PCa cells.

We next investigated the cause for ErbB3 increase in PCa cells with a history of AW exposure. Prolonged culture in CSS-containing medium decreased AR expression, but increased ErbB3 levels (Fig. 3A, left). To investigate whether this effect was caused by AR downregulation, AR levels were suppressed in LNCaP cells by two different RNAi, both stimulated ErbB3 levels, although to different extent (Fig. 3A, right). These results indicated that the AR suppresses ErbB3 levels in LNCaP cells, likely to prevent AR-independent cell signaling, whereas AW relieves this suppression.

Similarly, transfection with increasing amounts of wild-type AR (pAR0) in pRNS-1-1, a cell line derived from benign prostate tissue that had lost the expression of its endogenous AR, decreased ErbB3 levels, but not EGFR (Fig. 3B). Stable transfection with AR(T877A) (a mutant AR found in LNCaP cells, which remains ligand dependent), but not the vector alone, also downregulated ErbB3 levels in pRNS-1-1 cells (Fig. 3C, left). DHT, a strong AR ligand, suppressed ErbB3 in pRNS-1-1/AR(T877A), whereas prolonged culture in CSS-containing medium upregulated ErbB3 (Fig. 3C, right). These results show that the AR is a negative regulator of ErbB3 expression in normal prostate also.

Our data indicated that the AR regulated ErbB3 levels by a posttranscriptional, rather than a transcriptional mechanism (Supplementary Fig. S3). To determine whether AR affected ErbB3 degradation, pRNS-1-1 cells transfected with vector alone or with wtAR (Fig. 3D) were treated with the protein synthesis inhibitor cycloheximide, and the rate of ErbB3 degradation in the absence of de novo synthesis was determined over time. In cells transfected with vector alone, inhibition of protein synthesis with cyclohexmide caused a 36% decline in ErbB3 after 24 hours; however, transfection with wtAR greatly reduced ErbB3 half-life (<4 h; Fig. 3D). These results show that the increase in ErbB3 expression in the absence of AR is due to a decrease in protein degradation rates.

Previous studies identified the RING finger E3 ubiquitin ligase Nrdp1 as a promoter of ErbB3 degradation in breast cancer (18, 20). Nrdp1 overexpression in LNCaP cells decreased ErbB3 levels (Fig. 4A, left) and decreased cell proliferation (Supplementary Fig. S4A), whereas Nrdp1 downregulation (19, 20) increased ErbB3 (Fig. 4A, left), indicating an inverse relationship between ErbB3 and Nrdp1 in PCa as well. Hence, we hypothesized that the effect of AR on ErbB3 may be mediated by Nrdp1.

Culture in CSS medium decreased Nrdp1 expression, whereas transfection of wtAR into AR-null pRNS-1-1 cells increased Nrdp1 expression (Fig. 4A). We also tested this effect in androgen-dependent PC-346C cells derived from a nontreated human prostate tumor extracted by transurethral resection of the prostate (26, 27). Similar to LNCaP, culture of PC-346C cells in CSS medium decreased AR and Nrdp1, whereas ErbB3 increased (Fig. 4B). These results confirmed that the AR positively regulated Nrdp1 expression in androgen-dependent cells. Hence, we investigated whether the effect of AR on ErbB3 half-life is mediated by Nrdp1. ErbB3 half-life in pRNS-1-1 cells stably expressing wtAR was ∼3.5 hours, whereas downregulation of Nrdp1 increased ErbB3 half-life to >24 hours (Fig. 4C). Taken together, these results show that AR-regulated decrease in ErbB3 half-life is mediated by AR-induced Nrdp1 transcription. Thus, during AW, AR levels sharply decline, decreasing Nrdp1 levels, which in turn increased ErbB3 levels.

If AR always negatively regulated ErbB3 levels, then as AR increased during CRPC development, ErbB3 levels should decrease. However, we see that both LNCaP-AI and C4-2, another androgen-independent subline of LNCaP cells (33), expressed higher AR, as well as ErbB3, compared with LNCaP cells (Fig. 5A). Therefore, we compared the effect of AR stimulation on ErbB3 in LNCaP and LNCaP-AI cells. Increasing doses of DHT in LNCaP stimulated AR, as well as Nrdp1 protein and mRNA levels, but suppressed ErbB3 (Fig. 5B, top left). However, in LNCaP-AI cells, DHT did not affect Nrdp1 or ErbB3 levels (Fig. 5B, top right), although AR activity increased (Fig. 5B, bottom). Our results showed that the AR regulates the levels of Nrdp1, and, as a consequence that of ErbB3 expression, in LNCaP cells, but not in LNCaP-AI.

Both LNCaP-AI and C4-2 expressed higher ErbB3 and lower levels of Nrdp1 compared with LNCaP cells. In addition, overexpression of ErbB3 in LNCaP cells also decreased Nrdp1 (Fig. 5C, top). We conclude that ErbB3 overexpression suppressed Nrdp1 by an AR-independent mechanism, which prevented AR-mediated Nrdp1 transcription, underscored by the observation that AR-independent Nrdp1 decrease by short hairpin RNA prevented stimulation by DHT (Fig. 5C, bottom). Thus, despite increased AR, inability of AR to regulate Nrdp1 prolonged ErbB3 half-life in LNCaP-AI (Supplementary Fig. S4B).

We also tested the effect of AR on Nrdp1 and ErbB3 in androgen-dependent CWR22 tumors in nude mice (34) and in CRPC CWR22Rv1 cells, derived from a relapsed CWR22 tumor (35). CWR22 xenograft tumors were cultivated in the flanks of athymic nu/nu mice treated with vehicle or 50 mg/kg bicalutamide (an AR antagonist). Bicalutamide did not affect AR expression but severely decreased AR transcriptional activity in androgen-sensitive cells (32, 36). After 2 weeks, the mice were euthanized and the tumors were extracted, part was paraffin-embedded, and the rest were frozen. ErbB3 levels determined by Western blot (Fig. 6A) reflected only the epithelial cells, as immunohistochemical studies revealed ErbB3 staining only in epithelial cells (Fig. 6B) and showed higher ErbB3 in bicalutamide-treated animals both by immunoblotting and immunohistochemistry, whereas Nrdp1 levels were significantly suppressed. These results indicate that ErbB3 and Nrdp1 levels were regulated by AR in CWR22 tumors as well. However, in CRPC CWR22Rv1 cells, which express very low levels of Nrdp1 (Supplementary Fig. S3C), neither bicalutamide nor culture in CSS medium affected either Nrdp1 or ErbB3 (Fig. 6C). These results show that, similar to LNCaP and PC-346C cells, AR regulates Nrdp1 and ErbB3 in androgen-dependent CWR22 PCa cells but not in its CRPC subline CWR22Rv1.

Men who undergo AW therapy are known to become refractory to this therapy, although investigators continue to debate the causes leading to the development of castration resistance. The present study identifies ErbB3 as a likely mediator of increased proliferation during CRPC development, which releases cells from cell cycle arrest imposed by AW therapy, because increased ErbB3 resulted in enhanced proliferation both in the presence and the absence of androgens. We show that ErbB3 is increased during AW, and that this is caused by a novel mechanism involving the negative regulation of ErbB3 by the AR. In androgen-dependent cells, AR suppresses ErbB3, likely because although the AR promotes androgen-dependent cell cycle progression, it also resists the activation of androgen-independent pathways that promote cell growth. AW therapy relieves this suppression, thereby elevating ErbB3 levels, which stimulate proliferation and likely promote resistance to this therapy. Our data identifies the inability of the AR to suppress ErbB3 expression as one cause of CRPC.

The case for ErbB3 as a mediator of cell cycle release is strengthened by previous studies emphasizing the importance of HER2 in the development of castration resistance (37–45) and by the identification of ErbB3, rather than EGFR, as a major binding partner for HER2 in PCa (5, 14, 15). Reports described transcriptional as well as posttranslational increase of HER2 during AW (46); hence, the major binding partner of HER2 in CRPC would also be expected to increase. Our study shows that the AR regulates ErbB3 levels in PCa by promoting its degradation, whereas during AW, ErbB3 levels are increased due to decreased degradation.

We show that the AR promotes ErbB3 degradation by regulating Nrdp1 transcription. Importantly, we show that the AR transcriptionally regulates Nrdp1 expression in androgen-dependent but not in CRPC cells. Nrdp1 was originally identified as a RING finger domain–containing protein that interacts with ErbB3 (18, 47), binding to the cytoplasmic tail of ErbB3, and promoting its degradation by mediating ligand-independent ubiquitination. Nrdp1 was expressed in multiple androgen-regulated tissues including prostate, testicles (47), and skeletal muscles (18). Our data showed an inverse relationship between ErbB3 and Nrdp1 in androgen-dependent PCa cells. Nrdp1 promotes ErbB3 degradation (19, 47), and we show that it mediates AR regulation of ErbB3 half-life. These results indicate for the first time that Nrdp1 is a transcriptional target of the AR.

We also show that AR regulation of Nrdp1 and ErbB3 is lost in CRPC cells, thereby maintaining elevated ErbB3 in CRPC cells despite high AR activity. Our data show that ErbB3 overexpression suppressed Nrdp1 levels and likely prevented AR regulation of Nrdp1 transcription. This indicates androgen-dependentregulation of Nrdp1 in cells with high Nrdp1 and low ErbB3, and androgen independence of Nrdp1 transcription in cells expressing high ErbB3 and low Nrdp1. Therefore, our data explain why intermittent AW therapy, in which the effects of AW on ErbB3 may be reversed during frequent “off-cycle” phases, while Nrdp1 transcription is still AR regulated, may prolong androgen dependence (48), and indicates Nrdp1 regulation by AR as a hallmark of androgen-sensitive cell growth.

In summary, this study shows that induction of ErbB3 occurs as a result of AR inactivation during AW. However, even after AR levels rebound, ErbB3 remains elevated, eventually promoting castration-resistant proliferation of PCa cells. We show that Nrdp1 mediates the regulation of ErbB3 levels by the AR in androgen-dependent cells and is itself transcriptionally regulated by the AR, whereas this regulation is lost in AR-positive CRPC cells, likely due to high ErbB3 levels caused by AW therapy that may have suppressed Nrdp1 levels in an androgen-independent manner (Fig. 6D). These studies point to the importance of ErbB3 in PCa progression and identify it as a possible target of therapy for prevention of CRPC development.

No potential conflicts of interest were disclosed.

We thank Dr. Hsing-Jien Kung, Department of Biochemistry, UC Davis, for critical reading of the manuscript; Dr. Thomas G. Pretlow (Case Western Reserve School of Medicine, Cleveland, OH) for generously providing the CWR22 human prostate tumor xenograft and protocols for its growth; Rebekah Burich, Department of Hematology/Oncology, and Dr. Yu Wang, currently in Cleveland Clinic, Cleveland, OH, for the assistance with certain experiments; AstraZeneca, Cheshire, UK, for the gift of bicalutamide; Dr. John Koland, University of Iowa, Iowa City, IA, for the gift of the pcDNA3-HER2 and pcDNA3-ErbB3 cDNA; and Dr. Albert O. Brinkman, Erasmus University, the Netherlands, for the wild-type AR plasmid (pAR0).

Grant Support: Department of Defense, Idea Development Award PC081177 (P.M. Ghosh) and by Awards CA133209 (P.M. Ghosh), CA123541 (K.L. Carraway), and GM068994 (K.L. Carraway) from the NIH.

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.