Castration resistance in human prostate cancer is conferred by a frequently occurring androgen receptor splice variant

Abstract

Progression of prostate cancer following castration is associated with increased androgen receptor (AR) expression and signaling despite AR blockade. Recent studies suggest that these activities are due to the generation of constitutively active AR splice variants, but the mechanisms by which these splice variants could mediate such effects are not fully understood. Here we have identified what we believe to be a novel human AR splice variant in which exons 5, 6, and 7 are deleted (ARv567es) and demonstrated that this variant can contribute to cancer progression in human prostate cancer xenograft models in mice following castration. We determined that, in human prostate cancer cell lines, ARv567es functioned as a constitutively active receptor, increased expression of full-length AR (ARfl), and enhanced the transcriptional activity of AR. In human xenografts, human prostate cancer cells transfected with ARv567es cDNA formed tumors that were resistant to castration. Furthermore, the ratio of ARv567es to ARfl expression within the xenografts positively correlated with resistance to castration. Importantly, we also detected ARv567es frequently in human prostate cancer metastases. In summary, these data indicate that constitutively active AR splice variants can contribute to the development of castration-resistant prostate cancers and may serve as biomarkers for patients who are likely to suffer from early recurrence and are candidates for therapies directly targeting the AR rather than ligand.

Figure 1. Identification of a novel AR splice variant.

(A) Agarose gels showing PCR amplification of AR from human prostate LuCaP xenografts. Three sets of primers were used for PCR amplification of AR. One set is specific for exons 1–3 (amplicon 1), and another is specific for exons 2–8 (amplicon 2). The final set is specific for the 4–8 junction present in AR^v567es (amplicon 3). Note that xenografts 86.2 and 136 have a smaller PCR product with the amplicon 2 primers, while several xenografts are positive for the deletion using the amplicon 3 primers. Inverted agarose images are shown. NTC, no template control. (B) A graph of relative amounts of AR^v567es (amplicon 3) using qt-RT-PCR (mean ± 1 SD). Note that when xenografts occur as castrate-sensitive and castrate-resistant pairs (e.g., 35 and 35AI), the castrate-resistant sample (labeled AI) shows increased levels of AR^v567es. The differences were significant at P < 0.05 for 35 versus 35AI and 96 versus 96AI.

Figure 2. AR^v567es enhances AR^fl activity in benign prostate epithelial cells.

(A) P69 SV40T immortalized, nontransformed human prostate epithelial cells transfected with AR^v567es demonstrated a marked increase in AR^fl protein compared with P69 pcDNA empty vector control cells (P69 pc). (B) ARR3-Luc reporter assay of P69 pcDNA control cells versus P69 AR^v567es cells, showing increased transcriptional activity basally and in response to DHT, consistent with increased AR^fl expression and AR^v567es expression (mean ± SEM). ^#P < 0.05, P69 AR^v567es compared with P69 pc control cells in the absence of androgen; *P < 0.01, P69 AR^v567es compared with P69 pc cells with and without DHT added.

Figure 3. Constitutive activation of the AR^v567es in M12 prostate cancer cells.

(A) The AR-null M12 human prostate cancer cells were transiently transfected with AR^fl or the splice variant AR^v567es. The Western immunoblot of transfected cells shows expression of either AR^fl or AR^v567es. AR was detected with AR sc441 antibody, which detects both full-length and variant AR. GAPDH was used as a loading control. (B) ARE luciferase assay with the ARR3-Luc reporter. The M12 pcDNA empty vector control cells show that no AR activity is detected for any of the treatments. The M12 AR^fl cells had very low reporter activity in the absence of androgen as well as in the presence of the AR antagonist, flutamide, but had a clear increase in luciferase activity when 10^–9 M DHT was added. In contrast, M12 AR^v567es cells showed maximal reporter activity regardless of treatment. Values are mean ± SEM. *P < 0.01, DHT vs. no added DHT or DHT plus flutamide compared with DHT alone for AR^fl. There were no differences among treatments for pcDNA or AR^v567es cells. (C) Immunofluorescence staining of AR^fl and AR^v567es. In the absence of ligand, AR^fl is in the cytoplasm and translocates to the nucleus after addition of DHT. However, the constitutively active AR variant is primarily intranuclear in the absence of DHT and no change is seen when DHT is added. Nuclei are shown with DAPI staining. Arrows indicate examples of cells that are positive for nuclear translocation of AR. Scale bars: 10 μm.

Figure 4. Increased AR^fl activity in LNCaP cells transfected with AR^v567es.

The AR-positive LNCaP cells were transfected with AR^v567es or empty vector (pcDNA). (A) Western blot using AR sc441 antibody, which detects both AR^fl and AR^v567es. In cells transfected with AR^v567es, the AR^fl protein is markedly increased over that of control cells. Lanes were run on the same gel but were noncontiguous. (B) LNCaP cells were grown in vitro in CS serum. Cells containing the constituently active AR^v567es had the cuboidal appearance of the androgen-treated LNCaP cells. Original magnification, ×100. (C) Luciferase ARR3-Luc reporter assays performed on LNCaP cells showed a significant increase in AR signaling in the absence of DHT for cells expressing the AR^v567es protein. When DHT was added, there was a further increase in AR transactivation in the AR^v567es cells (133.2 RLU) compared with LNCaP pc control cells (54.6 RLU). *P < 0.01, **P < 0.001. (D) mRNA from several AR-regulated genes was examined by qt-RT-PCR (RQ). Interestingly, IGF-IR, which is regulated by nongenomic activities of the AR, was suppressed by the presence of AR^v567es, suggesting a loss of nongenomic AR activity by the variant. *P < 0.01, **P < 0.001, ***P < 0.0001, compared with LNCaP pc cells with same treatment. (E) MTS assay of LNCaP pc and LNCaP AR^v567es cells treated with decreasing concentrations of DHT. ^#P < 0.05, compared with LNCaP control cells. Values are mean ± SEM.

Figure 5. AR^v567es increases endogenous AR^fl expression and activity to similar levels as overexpression of AR^fl in LNCaP cells.

(A) Western blot of LNCaP cells transfected with empty vector, AR^fl, or AR^v567es. Immunoblot with AR C-19 antibody, which only detects AR^fl. Note the increase in AR^fl following transfection with either AR^fl or AR^v567es compared with empty vector control. The panel on the right shows immunoblot using AR sc441 antibody to demonstrate presence of AR^v567es in LNCaP AR^v567es cells. The graph depicts relative amounts of AR^fl present, using LNCaP pc plus DHT as the baseline. Values are mean ± SEM. (B) ARR3-Luc reporter assay on cell lines from A, grown with and without DHT (10^–9 M). Note that for the LNCaP pc cells, DHT resulted in more than a 10-fold increase (P < 0.01) in reporter activity compared with baseline. This increase in activity is difficult to discern in this figure due to the scale used to show changes with the AR^v567es construct. Values are mean ± SEM. *P < 0.01, LNCaP AR^fl cells compared with LNCaP control cells with same treatment; **P < 0.001, LNCaP AR^v567es cells compared with LNCaP AR^fl cells with same treatment.

Figure 6. The splice variant AR^v567es forms a complex with AR^fl.

(A) Immunoprecipitate with an HA antibody in M12 cells double transfected with AR^fl and HA-AR^v567es, followed by immunoblotting with AR C-19, which only recognizes AR^fl, and AR sc441, which recognizes both AR^v567es and AR^fl. As a positive control, Flag-tagged AR^fl was transfected into M12 cells and brought down with a Flag antibody. Lanes were run on same gel but were noncontiguous. (B) M12 cells transfected with the AR^fl construct alone or in combination with the HA-AR^v567es construct. Cells were grown in serum-free media and then treated with DHT 10^-9 M or vehicle (EtOH). AR^fl was immunolabeled with the AR C-19 antibody (red) and nuclei were immunolabeled with DAPI (blue). In cells containing both AR^fl and AR^v567es, AR^fl translocates to the nucleus in the absence of ligand. Scale bar: 10 μm. (C) Relative quantitative nuclear translocation of AR^fl. A minimum of 100 AR-positive cells were included for each construct. For comparisons, the population with the lowest percentage of translocation was considered 0 (AR^fl with no DHT), and the population with the highest percentage of translocation was considered 1 (AR^v567es with DHT). Values are mean ± SEM. **P < 0.001, ***P < 0.0001, compared with AR^fl with same treatment. (D) Tumor lysates were made from LuCaP 35 and 136 xenografts taken from castrated SCID mice, immunoprecipitated with AR C-19, and immunoblotted with AR sc441. AR^v567es was brought down with AR^fl in the LuCaP 136 xenograft.

Figure 7. Stability of AR^fl in presence of AR^v567es.

(A) LNCaP cells were transfected with either the empty vector (pcDNA) or the AR^v567es construct. After transfection, RNA was extracted from cells at the times noted, and qt-RT-PCR was performed for AR^fl. The relative levels of AR^fl at each time point were compared with levels at time 0 hours. Following transfection with the AR^v567es construct, there was an increase in AR^fl mRNA at 3 hours that rapidly returned to control levels by 6 hours after transfection. ^#P < 0.05 AR^v567es vs. pcDNA. Values are mean ± SEM. (B) PCR for AR^fl in actinomycin D–treated LNCaP pc and LNCaP AR^v567es cells. Note that AR^v567es did not significantly affect AR^fl mRNA stability. (C) The Western blot of cycloheximide-treated LNCaP pc and LNCaP AR^v567es cells demonstrates that AR^v567es increases AR^fl protein stability in the presence of DHT. (D) Graph depicting relative protein levels of AR^fl following treatment with cycloheximide in LNCaP pc versus LNCaP AR^v567es cells. *P < 0.01 versus 2 hour time point. Values are mean ± SEM.

Figure 8. The effect of the histone deacetylase inhibitor, SAHA, on AR^v567es-expressing cells.

(A) LNCaP pc or LNCaP AR^v567es cells were grown to 80 percent confluence in CS medium, with or without DHT (10^–9 M). SAHA was then added at the concentrations noted, and after 16 hours, cell lysates were collected, and Western blots were run with AR sc441 primary antibody, which detects both full-length and variant AR. (B) Two LuCaP 86.2 xenografts were removed and digested with collagenase to single cell suspensions. Cells were then plated in RPMI 5% CS medium, with or without DHT (one xenograft was grown with DHT, another without). Six hours later, cells were treated with either vehicle control (VC) or SAHA (5 mM). Six and twenty-four hours following treatment, cells were trypsinized and counted and cell lysates were collected for Western blots (blotted with AR sc441 antibody or GAPDH as a loading control). (C) Density of AR^v567es bands corrected for GAPDH. ^#P < 0.05, *P < 0.01, compared with vehicle control 6 or 24 hours. Values are mean ± SEM. (D) Cell counts corresponding to treatments in B. ^#P < 0.05, compared with 6 hour control. Values are mean ± SEM.

Figure 9. Expression profiles of LNCaP pc and LNCaP AR^v567es cells.

(A) Venn diagrams showing the gene number per comparison for expression profiles of androgen-regulated and AR^v567es-regulated genes in LNCaP cell lines. Cells were grown in CS medium with or without DHT (10^–9 M) for 24 hours, in triplicate experiments. Microarray analysis using Agilent 44K whole human genome expression oligonucleotide microarray slides was performed. Statistical analysis was conducted using 2-sample, unpaired t test with the SAM software, with a q value of less than 0.01% considered statistically significant for AR^fl cells and a q value of less than 10% considered statistically significant for AR^v567es cells. I and IV indicate genes upregulated or downregulated by DHT in LNCaP cells; II and V indicate genes upregulated or downregulated by DHT in LNCaP cells or LNCaP AR^v567es cells compared with LNCaP cells with no DHT; and III and VI indicate genes regulated by AR^v567es in LNCaP cells compared with LNCaP cells without DHT. (B) The top AR-regulated genes in the LNCaP AR^v567es cells compared with LNCaP controls grown in CS medium. Note that the right hand column shows changes in genes previously described to be regulated by DHT in LNCaP cells. Results of GO analysis of genes upregulated or downregulated by DHT in AR cells or genes uniquely upregulated or downregulated in AR^v567es cells are listed in Supplemental Tables 2 and 3.

Figure 10. Tumor growth of LNCaP AR^v567es cells and xenografts expressing various amounts of AR^v567es.

(A) 1 × 10⁶ LNCaP pc or LNCaP AR^v567es cells were mixed 1:1 with Matrigel and injected s.c. into athymic nude mice (n = 10 per line). When tumors reached a volume of 100–200 mm³, mice were castrated and animals were followed until tumors regrew and reached a volume of 1,000 mm³ or met IACUC criteria for euthanasia. There was no difference in growth rate in intact mice. Following castration, the LNCaP AR^v567es tumors, which did not decrease in volume following castration, grew to a significantly larger volume, more quickly than those of controls. *P < 0.01, LNCaP AR^v567es tumor volume versus LNCaP pc tumor volume. Values are mean ± SEM. (B–D) The response of tumor volume to castration in 3 different xenografts (intact, n = 12 per xenograft; castrate, n = 12 per xenograft). Note that LuCaP 86.2, which has the majority of its AR in the AR^v567es form, had no castration response; LuCaP 136, which has both full-length and variant AR, had a modest response to castration; and LuCaP 35, which has the majority of its AR as AR^fl, had a marked decrease in tumor volume in response to castration. Values are mean ± SEM. (E) Western blots of representative xenografts before and 6 weeks after castration using AR sc441, which recognizes AR^v567es and AR^fl. Lanes were run on different gels.

Figure 11. Expression of the splice variant AR^v567es in human prostates.

(A) AR variant PCR products from laser-captured samples of benign (B) and malignant tissue (T) from prostate tissue obtained at the time of prostatectomy from non-castrate men. Note that tumor or benign tissue may be positive in these samples. Samples 105–115 are PCR products from men, aged 35–55 years, with no evidence of prostate cancer, who were enrolled in a male contraception study. (B) Results of variant AR PCR products from metastases in a man who died from his prostate cancer. Table 1 shows results for all metastases samples. Variants include the AR^v567es described in the current report and 2 variants previously described, AR3 and AR-V7. GAPDH was used as a control for adequacy of RNA in the sample. Primers are described in the Methods section. If GAPDH could not be amplified, the sample was not included in the study. Inverted agarose images are shown for A and B. (C) Diagram of AR^v567es variant compared with previously published AR variants (AR^1/2/2b in ref. ; AR-V7 in ref. ; and AR3 in ref. 13). RRP, radical retropubic prostatectomy.