We had previously examined genome-wide expression of microRNAs and mRNAs in 60 primary human prostate tumors and 16 non-tumor prostate tissues in collaboration with Dr. Carlo Croce at Ohio State University. The analysis revealed that both key components of microRNA processing and numerous microRNAs were significantly altered in prostate tumors when compared with surrounding non-cancerous tissue. Tumor microRNAs were up- and down-regulated when compared with non-cancerous tissue and the expression profile of the tumors yielded a diagnostic microRNA signature. Notably, prostate tumors tended to express all members of the miR-106b-25 cluster at significantly higher levels than non-tumor prostate, which is consistent with the miR-106b-25 cluster having oncogenic properties in prostate tumor biology. We continued to study the specific functions of the miR-106b-25 cluster. Analyzing a large publically available microRNA expression dataset for prostate cancer revealed increased expression of the miR-106b-25 cluster in tumors that correlated with disease progression. Moreover, increased tumor miR-106b expression was associated with early disease recurrence and the combination of high miR-106b and low CASP7 (caspase-7) expression in primary tumors was an independent predictor of early disease recurrence (adjusted hazard ratio = 4.1;95% confidence interval: 1.6 to 12.3). To identify yet unknown oncogenic functions of miR-106b, we overexpressed miR-106b in LNCaP human prostate cancer cells to examine miR-106b-induced global expression changes among protein-coding genes. The approach revealed that caspase-7 is a direct target of miR-106b, which was confirmed by Western blot analysis and a 3'UTR reporter assay. Moreover, selected phenotypes induced by miR-106b knockdown in DU145 human prostate cancer cells did not develop when both miR-106b and caspase-7 expression were inhibited. Further analyses showed that caspase-7 is down-regulated in primary prostate tumors and metastatic lesions across multiple datasets and is by itself associated with disease recurrence and disease-specific survival. Using bioinformatics, we also observed that miR-106b-25 may specifically influence focal adhesion-related pathways. This observation was experimentally examined using miR-106b-25-transduced 22Rv1 human prostate cancer cells. After infection with a miR-106b-25 lentiviral expression construct, 22Rv1 cells showed increased adhesion to basement membrane- and bone matrix-related filaments and enhanced soft agar growth. In summary, miR-106b-25 was found to be associated with prostate cancer progression and may do so by altering apoptosis- and focal adhesion-related pathways. Our findings indicate that the miR-106b-25 cluster acts like an oncogene in human prostate cancer and is a predictor of early recurrence. In a second project, we characterized the expression profile of ultraconserved region-derived non-coding RNAs (ucRNAs) in prostate tumors and surrounding normal tissue using microarrays. Ultraconserved regions (UCR) are genomic segments of more than 200 base pairs that are evolutionarily conserved among mammalian species. They are thought to have functions as transcriptional enhancers and regulators of alternative splicing. Recently, it was shown that numerous RNAs are transcribed from these regions. These UCR-encoded transcripts (ucRNAs) were found to be expressed in a tissue- and disease-specific manner and may interfere with the function of other RNAs through RNA:RNA interactions. We hypothesized that ucRNAs have unidentified roles in the pathogenesis of human prostate cancer. In a pilot study, we examined ucRNA expression profiles in human prostate tumors. Using a custom microarray with 962 probesets representing sense and antisense sequences for the 481 human UCRs, we examined ucRNA expression in resected, fresh-frozen human prostate tissues (57 tumors, 7 non-cancerous prostate tissues) and in cultured prostate cancer cells treated with either epigenetic drugs (the hypomethylating agent, 5-Aza 2'deoxycytidine, and the histone deacetylase inhibitor, trichostatin A) or a synthetic androgen, R1881. Expression of selected ucRNAs was also assessed by qRT-PCR and NanoString-based assays. Because ucRNAs may function as RNAs that target protein-coding genes through direct and inhibitory RNA:RNA interactions, computational analyses were applied to identify candidate ucRNA:mRNA binding pairs.We observed altered ucRNA expression in prostate cancer (e.g., uc.106+, uc.477+, uc.363+A, uc.454+A) and found that these ucRNAs were associated with cancer development, Gleason score, and extraprostatic extension after controlling for false discovery (false discovery rate 5% for many of the transcripts). We also identified several ucRNAs that were responsive to treatment with either epigenetic drugs or androgen (R1881). For example, experiments with LNCaP human prostate cancer cells showed that uc.287+ is induced by R1881 whereas uc.283+A was up-regulated following treatment with combined 5-Aza 2'deoxycytidine and trichostatin A. Additional computational analyses predicted RNA loop-loop interactions of 302 different sense and antisense ucRNAs with 1058 different mRNAs, inferring possible functions of ucRNAs via direct interactions with mRNAs. We conclude from this first study of ucRNA expression in human prostate cancer that the disease has aberrant transcript expression of these RNAs.
