Mutation of the androgen receptor causes oncogenic transformation of the prostate

Abstract

Recent evidence demonstrates that the androgen receptor (AR) continues to influence prostate cancer growth despite medical therapies that reduce circulating androgen ligands to castrate levels and/or block ligand binding. Whereas the mutation, amplification, overexpression of AR, or cross-talk between AR and other growth factor pathways may explain the failure of androgen ablation therapies in some cases, there is little evidence supporting a causal role between AR and prostate cancer. In this study, we functionally and directly address the role whereby AR contributes to spontaneous cancer progression by generating transgenic mice expressing (i) AR-WT to recapitulate increased AR levels and ligand sensitivity, (ii) AR-T857A to represent a promiscuous AR ligand response, and (iii) AR-E231G to model altered AR function. Whereas transgenes encoding either AR-WT or AR-T857A did not cause prostate cancer when expressed at equivalent levels, expression of AR-E231G, which carries a mutation in the most highly conserved signature motif of the NH2-terminal domain that also influences interactions with cellular coregulators, caused rapid development of prostatic intraepithelial neoplasia that progressed to invasive and metastatic disease in 100% of mice examined. Taken together, our data now demonstrate the oncogenic potential of steroid receptors and implicate altered AR function and receptor coregulator interaction as critical determinants of prostate cancer initiation, invasion, and metastasis.

Fig. 1.

Generation and identification of transgenic mice. (A) The AR-N-terminal signature (ANTS) sequence was identified from multiple AR sequence alignments by using a clustalw algorithm (42) with human AR as the profile sequence. Alignment revealed a conserved region of amino acids corresponding to mouse residues 229–242. Blue, identical; yellow, conserved; white, nonhomologous. (B) The AR-E231G transgene. The E231G mutation (GAG > GGG) was introduced into a construct carrying an HA epitope-tagged WT mouse AR cDNA (AR-WT) between the rPB with a rabbit β-globin fragment with a small intron and a bovine growth hormone poly(A) signal sequence. Arrows indicate the primer pairs. The DNA sequences were confirmed in germ-line DNA. (C) The AR-T857A transgene. The T857A mutation (ACT > GCT) was created as described in B. (D) Expression analysis of AR-E231G and AR-WT transgenic mice. Tissue RNA at 12 weeks of age was analyzed by RT-PCR with P2 primers. Primers for L-19 were included as internal controls. Nontransgenic littermates were the controls (NT). Transgene plasmid DNA was the positive control (+). Transgenes were expressed predominantly in ventral prostate in independent lines. Expression of AR-E231G was primarily detected in the ventral prostate, but some expression was detected in the dorsolateral prostate. TE, testis; SV, seminal vesicle; SP, spleen; LU, lung; LV, liver; KD, kidney; BL, bladder; H, heart; MU, muscle; TH, thymus; BR, brain; M, DNA molecular weight markers. (E) RNase protection analysis. Probe was hybridized with RNA from VP lobes of AR-T857A, AR-E231G, AR-WT, and NT mice. Protected probe was separated by acrylamide gel. (F) Expression of AR-WT, AR-E231G, and AR-T857A protein. Extracts prepared from the DLP, VP, and AP lobes of transgenic and NT littermates were fractionated by SDS/PAGE and probed with anti-HA (Upper) or anti-GAPDH (Lower) antibodies. Transgene protein was detected only in the VP lobes of transgenic mice.

Fig. 2.

Pathobiology of transgenic mice. Paraffin sections (5 μm) were prepared from NT (A–D), AR-WT (E–H), AR-E231G (I–L), and AR-T857A (M–P) mice at 12 weeks of age. Representative sections stained with H&E are shown for DP lobe (A, E, I, and M), LP lobe (B, F, J, and N), VP lobe (C, G, K, and O), and AP lobe (D, H, L, and P). Histological features consistent with prostatic intraepithelial neoplasia were observed in the VP lobes of AR-E231G mice (K, Q–T). The anti-HA antibody was used to probe serial sections of VP (T) and NT (U), AR-WT (V) and AR-T857A mice (W). Brown nuclei indicate immunoreactivity. Sections were counterstained with methyl green. All ×20 except S and T, which were ×40 original magnification. (X) Increased proliferation in the VP lobes of AR-E231G transgenic mice. An anti-Ki67 antibody was used on sections of VP lobes from NT, AR-WT, and AR-E231G mice at 12 weeks of age. Quantitation of Ki-67 positive cells was from at least three mice for each group. *, Significant difference between AR-E231G and NT and AR-WT, P < 0.05 by Student's t test.

Fig. 3.

Pathobiology of prostate cancer in AR-E231G mice. (A–D) Adjacent sections (5 μm) were prepared from AR-E231G mice at 50 weeks of age. (A) Representative section of VP lobes stained with H&E shows adenocarcinoma with lymphocytic infiltration. (B) Analysis with anti-E-cadherin antibodies demonstrates E231G tumors to be of epithelial origin. (C) Analysis with anti-AR-specific antibodies demonstrates collocation of AR expression in adenocarcinoma. (D) Analysis with anti-HA antibody demonstrates expression of transgene collocates with tumor and AR expression (compare with C). (E–H) Sections of lungs of AR-E231G mice at 50 weeks stained with H&E demonstrating minimal residual glandular structure of a metastatic deposit (E) and a poorly differentiated lung deposit from an independent AR-E231G mouse (F). (G) Stained section adjacent to that in E with anti-HA specific antibodies demonstrates expression of the AR-E231G transgene in the metastatic lesion. (H) Stained section adjacent to that in G with anti-NKX2.1-specific antibody demonstrates the metastatic lesion is not of lung origin.

Fig. 4.

Structural analysis of AR mutations. (A–C) Three-dimensional molecular model structures representing WT or mutant AR-NTD peptide sequences constructed by using chemsite pro. Nonpolar hydrogen atoms were added, and the structure was solvated in a box of single-point charge water molecules and subjected to energy minimization (20 steps of steepest descent with a cutoff for nonbonded interactions of 10 Å). Ten independent molecular dynamic simulations, performed at constant temperature (300 K) over a period of 100 ps, are shown for each peptide. rmsd values for all atoms were calculated by using the relevant minimized starting structures as templates, and solutions are colored according to displacement from the starting minimized structure (blue, minimal displacement; red, maximum displacement). For each mutant, the WT peptide is shown in pink. Average rmsd (+SD) of the 10 solutions for each peptide is indicated. Solvent is not shown. (D) Existing structures for the human AR (Protein Data Bank ID code 1I37) and its LNCaP variant (Protein Data Bank ID code 1I38) were analyzed by using spdbv (Version 3.7). The environment within 6 Å of bound ligand is shown in magenta or aqua, with bound ligand shown in blue and hydrogen bonds represented by dashed green lines.