22 recurring tumors and develop in females and in males castrated or treated with bicalutamide. Although C4-2 cells divide in androgendepleted medium, androgens still efficiently induced the transcription of a 4xARE-luciferase reporter construct, confirming that AR is still functional in these cells. However, bicalutamide failed to antagonize the androgen-induced transcription. 14985929 In contrast, AR silencing in C4-2 cells, abolished the induction of transcription by androgens. We then transfected AR-siRNA into C4-2 and 22RV1 cells and compared the kinetics of the AR protein level, the cells’ proliferation, quantified by the MTT activity, and apoptosis, measured with a caspases assay were affected. In C4-2 cells, the AR protein level was markedly reduced 2 days after transfection. Cells’ proliferation diminished starting from day 3 and apoptosis increased significantly in C42 cells on day 4 and 14500812 5. In 22RV1 cells, the AR-siRNA effect appeared less pronounced: the AR protein level was unaffected until day 3 and the proliferation was reduced only after day 4. Moreover, AR silencing did not induce any significant increase in caspases-dependent apoptosis during the 5 first days following transfection. Therefore, 5 days after transfection, the number of viable cells was higher in 22RV1 than in C4-2 cells. The Vascular Endothelial Growth Factor, which is a key regulator of angiogenesis in the prostate, is upregulated by androgens in the normal prostate and in androgen-dependent cells, through indirect transcriptional or post-transcriptional mechanisms. Although CRCaP tumors are highly angiogenic, the role of AR in the control of VEGF production in these tumors was not studied so far. Importantly, we observed that AR silencing reduced the VEGF synthesis in LNCaP as well as in C4-2 and 22RV1 cells, demonstrating that VEGF is still regulated by AR in advanced prostate cancers. Our in vivo studies using the LNCaP model demonstrated the efficiency and specificity of the antitumoral effects produced by AR silencing. We then studied the effects of the two different ARsiRNAs on the in vivo growth of castration-resistant tumors. We first grafted C4-2 cells to nude mice and, after a month, once vascularized tumors were exponentially growing, and reached a mean tumor volume of 129.9629.1 mm3, mice were randomized to receive cont-, or panAR-, or 
hAR-siRNA. In contrast with castration or bicalutamide, which do not affect the development of C4-2 tumors, both the panAR- and the Odanacatib site hAR-siRNA efficiently inhibited the C4-2 tumor growth. In nonnecrotic regions, mainly at the periphery of the tumor, a strong reduction in the level of AR expression and in the proportion of KI67-positive proliferating cells was observed. Similarly, treatment of mice bearing 22RV1 tumors with AR-siRNA markedly repressed the tumor growth. Despite the presence of large necrotic regions in C4-2 and 22RV1 tumors treated with AR-siRNA, we only seldom observed shrinkage of tumors. This may be due to the strong fibrotic reaction also observed in these tumors. The external volume of the tumors therefore is likely not proportional to the number of viable tumor cells. To further establish that 22RV1 were resistant to castration levels of testosterone and that the ARsiRNA was not dependent on indirect effects on prostate or testes, 22RV1 cells were grafted into female nude mice. Mice bearing exponentially growing tumors with a mean volume of 300 mm3 were treated with hAR-siRNA. This treatment inhibite