From transforming growth factor-beta signaling to androgen action: identification of Smad3 as an androgen receptor coregulator in prostate cancer cells

Abstract

Although transforming growth factor-beta (TGF-beta) has been identified to mainly inhibit cell growth, the correlation of elevated TGF-beta with increasing serum prostate-specific antigen (PSA) levels in metastatic stages of prostate cancer has also been well documented. The molecular mechanism for these two contrasting effects of TGF-beta, however, remains unclear. Here we report that Smad3, a downstream mediator of the TGF-beta signaling pathway, functions as a coregulator to enhance androgen receptor (AR)-mediated transactivation. Compared with the wild-type AR, Smad3 acts as a strong coregulator in the presence of 1 nM 5alpha-dihydrotestosterone, 10 nM 17beta-estradiol, or 1 microM hydroxyflutamide for the LNCaP mutant AR (mtAR T877A), found in many prostate tumor patients. We further showed that endogenous PSA expression in LNCaP cells can be induced by 5alpha-dihydrotestosterone, and the addition of the Smad3 further induces PSA expression. Together, our findings establish Smad3 as an important coregulator for the androgen-signaling pathway and provide a possible explanation for the positive role of TGF-beta in androgen-promoted prostate cancer growth.

Figure 1

The ligand-induced transactivation of AR is enhanced by treatment with TGF-β. (A) CAT assays were performed with extracts from DU145 cells transfected with AR expression vector (pSG5-AR) (1 μg) in the presence (+) or absence (−) of DHT (10⁻⁸ M) or TGF-β1 (10 ng/ml) or specific TGF-β1 neutralizing antibody (20 mg/ml). (B) (Left) PC-3 cells were transfected with pSG5-AR (1 μg) in the presence (+) or absence (−) of DHT (10⁻⁸ M) with increasing amounts of TGF-β1. (Right) A fixed amount of TGF-β1 (10 ng/ml) was added to transfected PC-3 cells with increasing amounts of specific TGF-β1 neutralizing antibody. (C) PC-3(AR)2 cells stably transfected with AR were overexpressed with TGF-β type I (TβRI) or type II (TβRII) receptor or constitutively active TGF-β type I receptor (TβRI-T204D) as indicated. Three micrograms of MMTV-CAT or MMTV-Luc was used as a reporter plasmid in all experiments. All values represent the averages ± SD of four independent experiments.

Figure 2

The association of Smad3 with AR in a mammalian two-hybrid interaction system. (A) SW480.7 cells were cotransfected with 3 μg of Gal4-Smad3 encoding the full-length Smad3 fused to the Gal4-DBD and 4.5 μg of VP16-AR encoding the full-length AR fused to the activation domain of VP16. Interaction was estimated by determining the level of CAT activity from 3 μg of the reporter plasmid pG5-CAT in the presence of 10⁻⁸ M DHT. (B) DU145 cells were transfected with Gal4-Smad3 and VP16-AR expression vectors in the presence (+) or absence (−) of DHT and TGF-β. Each CAT activity is presented relative to the transactivation observed in the absence of DHT. All values represent the mean ± SD of four independent experiments.

Figure 3

In vivo and in vitro interaction between Smads and AR. (A and B) Coimmunoprecipitation of AR and Smad3. (A) PC-3 cells that overexpressed Flag-Smad3 and AR. (B) PC-3 and PC-3(AR)2 cells were treated with or without DHT. Cell extracts were prepared and immunoprecipitations were performed with the use of anti-FLAG antibody or anti-Smad3 antibody, followed by immunoblotting with antibody to AR. (C) The wtAR and different AR deletion mutants used in the GST pull-down assay are shown schematically. (D) Interaction domains of AR for Smad3. A series of ³⁵S-labeled mtARs incubated with GST-Smad3 or GST alone in the presence (+) or absence (−) of 1 μM DHT were tested for interaction in the GST pull-down assay.

Figure 4

The effects of Smad3 on AR-mediated transcriptional activity. (A) SW480.7 cells were cotransfected with 1 μg of pSG5-AR, 3 μg of MMTV-CAT, and 3 μg of Smad3 expression vectors in the presence (+) or absence (−) of DHT (10⁻⁸ M) or TGF-β (10 ng/ml). (B) DU145 cells were cotransfected with 3 μg of Smad3 or Smad3ΔC mutant expression vectors with 1 μg of pSG5-AR and 3 μg of MMTV-CAT, in the presence (+) or absence (−) of DHT (10⁻⁸ M) or TGF-β (10 ng/ml). Each CAT activity is presented relative to the transactivation observed in the absence of DHT, and an error bar represents the mean ± SD of four independent experiments.

Figure 5

The androgen response element is important for TGF-β/Smad3-enhanced AR transactivation. (A) DU145 cells were transiently cotransfected with AR (2 or 4 μg) and tyrosine aminotransferase–CAT, MMTV-CAT, or PSA-CAT (3 μg), in the presence (+) or absence (−) of DHT (10⁻⁸ M) or TGF-β (10 ng/ml). (B) DU145 cells were transiently cotransfected with AR (2 or 4 μg) and tyrosine aminotransferase–CAT, MMTV-CAT, or PSA-CAT (3 μg), and Smad3 expression vector (6 or 10 μg) in the presence (+) or absence (−) of DHT (10⁻⁸ M). Each CAT activity is presented relative to the transactivation observed in the absence of Smad3. All values represent the mean ± SD of three independent experiments.

Figure 6

Effect of Smad3 on the transcriptional activities of wtAR, mtAR, progesterone receptor (PR), VDR, and estrogen receptor (ER). (A) DU145 cells were transiently cotransfected with 3 μg of reporter plasmids (MMTV-CAT for AR and PR, ERE-CAT for ER, and VDRE-CAT for VDR), 1 μg of each receptor constructed in pSG5, and 4.5 μg of Smad3 expression vector in the presence of 10⁻⁸ M of each cognate ligand. Each luciferase and CAT activity is presented relative to the transactivation observed in the absence of Smad3. (B) 1.5 μg of wtAR was cotransfected with 4.5 μg of Smad3 or ARA70 in the absence or presence of DHT, E2, or HF at indicated concentrations. (C) The LNCaP mtARt877a was used to replace the wtAR to perform the same experiment as in B. All values represent the mean ± SD of three independent experiments.

Figure 7

AR-induced PSA expression is potentiated by Smad3. (A) Smad3-enhanced androgen/AR-induced PSA mRNA expression. LNCaP cells were transfected with Smad3 and parent vector as indicated for 16 h, followed by DHT treatment for another 16 h. The PSA expression level was determined by Northern blotting. The probe was obtained from exon 1 of the PSA gene and labeled with [α-³²P]dCTP. A β-actin probe was used as a control for equivalent mRNA loading. (B) A model for androgen and TGF-β pathways in AR-mediated PSA transcription.