Key functions of non-coding RNAs in human cancer have recently been described and several classes of non-coding RNAs (e.g., miRs, ucRNAs, lincRNAs, snoRNAs) are now known whose expression is dysregulated in the disease because of existing oncogenic stimuli, genome amplifications and deletions, mutations, and epigenetic silencing [18, 19]. Here, we examined the expression of transcripts encoded by 481 UCRs in prostate cancer, defined as ucRNAs, and found that ucRNAs can be detected in the cancerous human prostate and show a disease-specific expression pattern. When we compared the expression of ucRNAs in prostate cancer with other human cancer types, e.g., leukemia, colon cancer, liver cancer, and neuroblastoma [6, 10, 20, 21], we did not find a common ucRNA expression signature among them. Thus, ucRNA expression is rather tissue-specific and cancer type-specific, which is reminiscent of miR expression patterns in solid human tumors [22, 23]. However, it appears that even fewer ucRNAs than miRs are commonly dysregulated across tumor types, as we could not identify any commonly dysregulated ucRNAs between our study and other ucRNA studies. We also observed that only very few UCRs encoded for transcripts that were associated with both disease onset and progression. Among them were UCR106 and UCR346 that were found to encode for sense and antisense transcripts (uc.106+, uc.106 + A, uc.346+, uc.346 + A) and whose expression was dysregulated in primary tumors and with disease stage and Gleason grade. Previous studies identified robust miR signatures only for the tumor versus non-cancerous tissue contrast, but did not find those signatures when studying tumor grade and disease stage differences in prostate cancer [24–26]. Our study of ucRNAs concurs with these findings.
The underlying mechanisms that cause cancer-specific ucRNA expression are largely unknown but may include mutational events and epigenetic regulation. Some UCRs are regulated by epigenetic silencing, as was shown recently , which is consistent with the well established regulation of other non-coding RNAs, like miRs, by this mechanism . Our study identified 6 ucRNAs whose expression significantly increased after treatment of LNCaP cells with the two epigenetic drugs, 5-AzaC and TSA. Of those, uc.283 + A was up-regulated by these drugs in both our study and the study by Lujambio and coworkers, who also showed that the uc.283 + A locus is silenced by CpG hypermethylation in a human colon cancer cell line . We did not examine the uc.283 + A locus in more detail but the consistent finding in the two studies (present study and ) identifies uc.283 + A as an epigenetically regulated transcript. UCR283 is located in an intergenic genomic region and does not overlap with the coding region for any mRNA. While epigenetic regulation of UCR283 could be observed in the LNCaP cells, we did not find that transcription from UCR283 was significantly altered in human prostate tumors but noticed that two of the six epigenetically regulated ucRNAs, uc.241 + A and uc.285+, were down-regulated in these tumors when compared with adjacent non-cancerous tissue. Thus, future investigations are needed to define the importance of ucRNA silencing in human prostate cancer biology.
Another important mechanism of gene regulation in prostate biology and prostate cancer progression is the activation of the androgen receptor signaling pathway . It has been shown that non-coding RNAs like miRs are regulated by androgen signaling [25, 29–31]. We examined the effect of the androgen receptor agonist, R1881, in androgen-sensitive LNCaP cells and surveyed global expression changes by ucRNAs. Our experiment was restricted to one dose of R1881 and future experiments would have to investigate R1881 effects on UCR transcription using a broader dose range and a larger panel of cell lines. Nevertheless, this exploratory analysis identified several ucRNAs that were induced by R1881, though only the up-regulation of uc.287+ was robust. In contrast, multiple ucRNAs were found to be significantly down-regulated in response to R1881. Among them was uc.283+, which was the most significantly repressed transcript. We did not find that any of these androgen-responsive ucRNAs are differently expressed in primary human prostate tumors, when compared with non-cancerous prostate tissue, or were associated with Gleason grade or disease stage, nor did we find that these ucRNAs were, with the exception of uc.135+, described in other ucRNA studies . Thus, they do not appear to be cancer-associated ucRNAs in the primary disease. Nevertheless, it remains a possibility that these transcripts are differently expressed between castration-resistant and castration-sensitive prostate tumors, which we could not investigate.
Like with most non-coding RNAs other than miRs, our current knowledge with respect to the function of ucRNAs is very limited. Only a few full length ucRNA transcripts have been described. Because of these limitations, we used siRNA to knockdown uc.106+ expression and also decided to apply computational analyses to predict RNA-RNA interactions between ucRNAs and mRNAs based on the available coding sequence from the 481 described UCR loci, and to link these interactions to possible functions of the ucRNAs. In addition, we applied correlation analyses of ucRNA, mRNA, and miR expression data to discover functional associations between selected ucRNAs and both miRs and mRNAs. These are exploratory tools that can yield new insight into ucRNAs in the absence of other supporting information in this largely unexplored research field. Down-regulation of uc.106+ generated a robust gene expression profile in LNCaP cells, indicating that uc.106+ is a functional transcript and suggesting that UCR106 encoded transcripts may have a function in prostate cancer. Future research is needed to clone candidate non-coding transcripts encoded by UCR106 to allow functional analysis of them in phenotypic assays. Other analyses also revealed a significant enrichment for predicted ucRNA:mRNA interactions in processes related to ion binding and calcium signaling, suggesting that ucRNAs may target calcium signaling processes. Our approach also showed that expression of uc.106 + A (antisense transcript of UCR106) in prostate tumors may influence the interferon signaling pathway by either a direct or indirect mechanism in prostate tumors, as was indicated from the inverse relationship between uc.106 + A expression and expression of multiple interferon pathway genes. Notable, UCR106 also encodes a DNA damage-regulated gene and ATPase, termed OLA1 or Obg-like ATPase, which is a putative GTP-binding protein involved in mitochondrial function and regulation of the oxidative stress response [32, 33]. UCR106 is located in an intronic region of this gene and we did not find a correlation between OLA1 expression and the expression of either uc.106+ or uc.106 + A. While these are only few examples of candidate functions for ucRNAs in prostate cancer, our exploratory work shows that these approaches can be useful in potentially uncovering the biology of ucRNAs in cancer biology.