Androgen responsive intronic non-coding RNAs

BACKGROUND: Transcription of large numbers of non-coding RNAs originating from intronic regions of human genes has been recently reported, but mechanisms governing their biosynthesis and biological functions are largely unknown. In this work, we evaluated the existence of a common mechanism of transcription regulation shared by protein-coding mRNAs and intronic RNAs by measuring the effect of androgen on the transcriptional profile of a prostate cancer cell line. RESULTS: Using a custom-built cDNA microarray enriched in intronic transcribed sequences, we found 39 intronic non-coding RNAs for which levels were significantly regulated by androgen exposure. Orientation-specific reverse transcription-PCR indicated that 10 of the 13 were transcribed in the antisense direction. These transcripts are long (0.5-5 kb), unspliced and apparently do not code for proteins. Interestingly, we found that the relative levels of androgen-regulated intronic transcripts could be correlated with the levels of the corresponding protein-coding gene (asGAS6 and asDNAJC3) or with the alternative usage of exons (asKDELR2 and asITGA6) in the corresponding protein-coding transcripts. Binding of the androgen receptor to a putative regulatory region upstream from asMYO5A, an androgen-regulated antisense intronic transcript, was confirmed by chromatin immunoprecipitation. CONCLUSIONS: Altogether, these results indicate that at least a fraction of naturally transcribed intronic non-coding RNAs may be regulated by common physiological signals such as hormones, and further corroborate the notion that the intronic complement of the transcriptome play functional roles in the human gene-expression program. Keywords: Time course study – effect of androgen on gene expression Prostate carcinoma cell line LNCaP were treated with synthetic androgen or ethanol vehicle (control) and harvested for RNA isolation at several time points: 0, 6, 9, 12, 18, 24, and 48 hours. Each time point was hybridized twice, giving a total of 26 Samples, with at least 4 expression data of each spotted cDNA (2 internal replicates in each slide). In addition, hybridizations included a reference pool of RNA from 3 prostate carcinoma cell lines: LNCaP, DU145 and PC3 (cultured following ATCC suggested media), labeled using Cy3-dCTP. We decided to use the single-color approach to calculate the treatment to control sample ratios using only Cy5-labelled sample measurements. The mean intensity signal (40% trimmed) of each dataset was used for normalization between experiments (best normalization procedure tested). The Cy3-labeled reference RNA pool was used only as a general quality control of experiments. Spots with statistically significant expression changes in at least 3 consecutive time points in response to androgen stimulation were identified using the Significance Analysis of Microarray data (SAM) approach, using as parameters: two-class response (paired data), 1000 permutations, K-Nearest Neighbors Imputer, fold-change ≥ 1.5 and false discovery rate (FDR) < 5%.