MED1 mediates androgen receptor splice variant induced gene expression in the absence of ligand

Abstract

The appearance of constitutively active androgen receptor splice variants (AR-Vs) has been proposed as one of the causes of castration-resistant prostate cancer (CRPC). However, the underlying mechanism of AR-Vs in CRPC transcriptional regulation has not been defined. A distinct transcriptome enriched with cell cycle genes, e.g. UBE2C, has been associated with AR-Vs, which indicates the possibility of an altered transcriptional mechanism when compared to full-length wild-type AR (ARfl). Importantly, a recent study reported the critical role of p-MED1 in enhancing UBE2C expression through a locus looping pattern, which only occurs in CRPC but not in androgen-dependent prostate cancer (ADPC). To investigate the potential correlation between AR-V and MED1, in the present study we performed protein co-immunoprecipitation, chromatin immunoprecipitation, and cell proliferation assays and found that MED1 is necessary for ARv567es induced UBE2C up-regulation and subsequent prostate cancer cell growth. Furthermore, p-MED1 is bound to ARv567es independent of full-length AR; p-MED1 has higher recruitment to UBE2C promoter and enhancer regions in the presence of ARv567es. Our data indicate that p-MED1 serves as a key mediator in ARv567es induced gene expression and suggests a mechanism by which AR-Vs promote the development and progression of CRPC.

Figure 1. AR^v567es increased UBE2C expression and improved prostate cancer cell proliferation, which was blocked by MED1 silencing

(A) Western blot and qRT-PCR showed significantly increased UBE2C expression in transiently transfected LNCaP cells with 3Flag-CMV-AR^v567es expression vector. Up-regulation of UBE2C by AR^v567es could be attenuated by MED1 siRNA at mRNA and protein level. (B) AR^v567es stable expressing cell line, LNCaP-AR^v567es showed significant higher proliferation rate (**p < 0.01) compared with Lenti virus empty vector control cell line in the absence, but not presence of DHT. (C) AR^v567es induced LNCaP cell proliferation could be blocked by MED1 silencing (*p < 0.05) when DHT was not present, but not significant with DHT.

Figure 2. AR^v567es could bind p-MED1 independent of full length AR

(A) In androgen-dependent LNCaP cells transiently transfected with Flag-tagged AR^v567es, pulling-down of Flag could co-precipitate both AR^fl and p-MED1. With the same input volumes of 100ug, AR^v567es showed more abundant p-MED1 pull-down than that of full-length AR (AR^fl). (B) Reversed pull-down with p-MED1 antibody could co-precipitate Flag-tagged AR^v567es in the cumate-inducible M12-ARv567es prostate cancer cell line.

Figure 3. AR^v567es recruited p-MED1 to UBE2C enhancer and promoter regions independent of androgen

(A) p-MED1 binding to UBE2C promoter and enhancers increased in LNCaP-AR^v567es cells compared with LNCaP-Lenti cells, in the conditions of androgen deprivation (T+S) and full-length AR inhibition (MDV, MDV+DHT); AR^fl activation by DHT abated the extra recruitment induced by AR^v567es. (B) The reversed effect of DHT and MDV on the recruitment of p-MED1 to UBE2C promoter and enhancers in LNCaP-AR^v567es cells and LNCaP-Lenti cells. (C) AR^v567es also showed increased binding of p-MED1 to PSA enhancers but not in the promoter region especially when androgen receptor was inhibited (MDV3100). *p < 0.05 and #p < 0.01.

Figure 4. AR^v567es Increased UBE2C enhancer transcriptional activities

(A) The diagrammatic view of the UBE2C Enhancer-1 region (20 kb upstream of the UBE2C transcriptional start site) used in the Luciferase reporter assay. (B) Even though lower than the E4TATA control, the luciferase activities of UBE2C E1-2 and E1-3 region were significantly higher ( p < 0.01) in M12-AR^v567es cell compared with LNCaP-Lenti cells. (C) MED1 silencing attenuated the transcriptional activation of AR^v567es on UBE2C enhancer specially showing significance on E1-2 ( p < 0.01).

Figure 5. AR^v567es/P-MED1 complex cross talks with PI3K-AKT pathway, but not with MAPK pathway

Inhibition of PI3K kinase by LY294002 led to decreased MED1 phosphorylation and UBE2C expression both in LNCaP-AR^v567es cells (A) and M12-AR^v567es cells (B). MAPK inhibitor PD98059 had no effect either on MED1 phosphorylation or UBE2C expression in M12-AR^v567es cells (C).

Figure 6. FoxA1 was involved in the AR^v567es/p-MED1 transcriptional regulation

(A, C) Co-IP showed the binding of FoxA1 to p-MED1 and AR^v567es in M12-AR^v567es and LNCaP–AR^v567es cell lines. (B, D) Silencing of FoxA1 decreased UBE2C expression in both cell lines.

Figure 7. “The p-MED1 switch hypothesis”: Model of AR^fl/AR^v567es/p-MED1 transcriptional regulation on UBE2C in prostate cancer

(a) In ADPC cells with the presence of DHT, p-MED1 goes to AR^fl but no chromatin looping forms and there is a low level of activation on UBE2C transcription; (b) In CRPC cells with DHT depleted, AR splice variants appear which have higher affinity to p-MED1, and recruit more p-MED1 to the UBE2C promoter and enhancers with the assistance of FoxA1, resulting in looping formation and vigorously activating UBE2C expression and leading to cell survival and increased cell proliferation.