The Cistrome And Gene Signature Of Androgen Receptor Splice Variants In Castration-resistant Prostate Cancer

Androgen deprived therapy (ADT) is still the first line therapy for prostatecancer which is not suitable for surgical therapy or late-stage prostate cancer.Androgen could take function through binding to androgen receptor (AR), leads to theinitiation and progression of prostate cancer. ADT targets the AR signaling and treatprostate cancer. However, although prostate cancer could regress after initial ADTtreatment, most tuomor would relapse and progress from androgen-dependent intoandrogen-independent prostate cancer (AIPC), also known as castration resistantprostate cancer (CRPC), and resist ADT.AR is a110kDa ligand-dependent transcription factor belonging to the nuclear receptor (NR)superfamily. The full-length androgen receptor (AR-FL) contains four functional domains,including an N-terminal domain (NTD) which has transactivation activity, a centralDNA binding domain (DBD), a short hinge region, and a ligand-binding domain (LBD).In the absence of androgens, the AR is located in the cytoplasm where it associates withheat shock proteins (HSPs). Androgens enter the cell by passive diffusion and bind tothe LBD of AR. The conformational rearrangements induced by the ligand-AR complexresult in dissociation of HSPs. The AR form homodimers and expose its nuclearlocalization signal (NLS) which could mediate the translocation of AR into nucleus.The activated nuclear AR binds to androgen response elements (AREs) as a homodimerand recruits other proteins, such as AR coregulators, general transcription factors andRNA polymerase II. The formation of these diverse multiprotein complexes results inspecific transcriptional activation or repression of target genes.Androgen receptor splice variants (AR-Vs) are truncated isoforms of AR-FL.Compared to the AR-FL, these truncated AR-Vs proteins still have NTD and DBD domain, but lack LBD domain. AR-Vs instead contain cryptic or novel exons, whichresult in lack of androgen binding. This distinct feature leads to resist ADT targetingcanonical AR signaling. Because AR-Vs still maintain DBD, they can bind to DNAand modulate downstream target genes expression. Thus, AR-Vs could be one ofmechanisms that result in ADT resistance and CRPC progression.It has been reported that AR-Vs have androgen-independent constitutive activity, butwhether this characteristic is dependent of AR-FL is still controversial. The differentresults are also shown on that if AR-Vs has their unique regulated genes. It has not beenreported that if genes regulated by AR-Vs are correlated with progression of prostatecancer, especially CRPC. Thus, the main objectives of this thesis are:(1) to identify thecistrome of AR-Vs in the absence of androgen, and whether or not the binding events aredependent on AR-FL;(2) to identify the gene expression profile of AR-Vs in CRPC cellmodel, and determine the unique gene signature;(3) to determin the clinical significanceof the AR gene signature in a cohort of prostate cancer specimens.We assess the level of AR-Vs by absolute quanitification real-time RT-PCR in apanel of androgen-dependent and androgen-independent human PCa cell lines. Theresults suggest besides AR-FL, four AR-Vs mRNA are detected inandrogen-independent22Rv1cell line. The AR-Vs in22Rv1have androgen-independent consititutive transactivity. Because of the similar heterogeneity in AR-Vsto clinical samples,22Rv1is chosen as perfect CRPC cell model.Then the cistrome of AR-Vs are identified in22Rv1cells. Three gourps,22Rv1-ARFL+/ARVs+,22Rv1-ARFL-/ARVs+and22Rv1-ARFL-/ARVs-, weregenerated through RNA interference. ChIP-seq are conducted with antibody targetingN-terminal of AR. Quaitve analysis shows about75%of AR binding identifiedbetween22Rv1-ARFL+/ARVs+and22Rv1-ARFL-/ARVs+are overlapping.Quantitative analysis shows the peak densities of AR binding in these two groups arecomparable. These results suggest AR-Vs could bind to DNA independently ofandrogen and AR-FL.In addition, gene expression profiles are analyzed in22Rv1cells. The experimental design is the same as above. The genes modulated by all AR isoformsand AR-FL are obtained in the absence of androgen through microarray. Thencomparing these two groups, a subgroup of genes modulated uniquely by AR-Vs isidentified. The genes with fold-change>1.3are selected as AR-Vs gene signature.The integrated analysis of AR-Vs cistrome and AR-Vs regulated genes indicates thatAR-Vs bind to DNA and modulate some AR-Vs regulated genes directly.Finally we evaluate the correlation between AR-Vs gene signature and clinicalprostate cancer specimens. Two prostate cancer clinical database are used.Unsupervised clustering analysis shows, AR-Vs gene signature not only separatelocalized prostate cancer and metastatic CRPC tissue significantly, but alsodistinguish benign and malignant prostate tissue. Further analysis suggests someprobes in AR-Vs gene signature are correlated with Gleason pattern and lymph nodemetastasis positively and negatively. It is also suggested some probes are correlatedwith PSA failure. These results indicate that AR-Vs gene signature correlate withCRPC and prostate cancer progression.In conclusion,22Rv1cells are chosen as CRPC cell model based on theexpression of AR-Vs in a panel of prostate cancer cell lines. The cistrome of AR-Vsare first analyzed, demonstrating AR-Vs could bind to DNA independently ofandrogen and AR-FL. The AR-Vs gene expression profile suggest that AR-Vs couldmodulate a distinct set of genes independently of AR-FL, some of which may bemodulated by AR-Vs binding directly. The identified AR-Vs gene signature correlatewith CRPC and prostate cancer progression, showing the potential roles of AR-Vs inCRPC. This finding could help to identify novel prognostic biomarkers and noveltherapeutic targets for prostate cancer.