Integrative Genomic Approach Identifies Deregulated MicroRNAs in Human Myeloma Cell Lines.

Abstract

The recent discovery of microRNAs (miRNAs), small noncoding RNAs involved in the regulation of cell cycle, survival, and differentiation programmes, has added a further level of complexity to normal and cancer cell biology. Loss or amplification of miRNA genes by broad cytogenetic abnormalities or minute molecular aberrations has been observed in a variety of human malignancies, with the consequent altered expression of these regulatory genes. Additionally, approximately one third of miRNAs are located within the intronic regions of coding transcription units, and recent evidence indicates that the expression of these miRNAs largely coincides with the transcription of the corresponding host genes. To date, no evidence of deregulated miRNA expression has been reported in multiple myeloma (MM). To provide insights into miRNA biology in MM, we performed an integrative analysis of genome-wide, gene expression and miRNA expression profilings in a panel of 16 human myeloma cell lines (HMCLs). Global miRNA and mRNA expression data were generated on Agilent miRNA microarrays (representing 470 human mature miRNAs) and GeneChip® HG-U133A arrays, respectively, and both quantile-normalized. Genome-wide profiling data were generated on GeneChip® Human Mapping 250K NspI arrays and copy number (CN) values were inferred using the circulary binary segmentation (DNAcopy R Bioconductor package). To measure the correlation between the expression levels of each miRNA and the corresponding CN value or host gene expression, conventional non-parametric analyses were performed (Kendall's tau and Wilcoxon rank-sum tests). As regards miRNA gene CN, the most frequent alteration identified was represented by gain/amplification (for all miRNA genes investigated, an increased CN was present in at least 3 HMCLs, with an average frequency of 58%), followed by loss (5%) and biallelic deletion (0.3%). Our analysis revealed that 14 different miRNA transcripts (miR-15a, miR-19a, miR-21, miR-22, miR-30d, miR-99b, miR-130b, miR-132, miR-140, miR-185, miR-339, miR-491, miR-503, miR-768-3p) had concordant levels with the inferred CN value of the corresponding miRNA gene. Notably, the identified miRNAs mapped to different genomic regions, some of which are involved in recurrent CN alterations in MM, such as 8q24, 19q13.33, or chromosome arms 13q, 16q, 17p, 17q, 22q, and for some of the miRNAs a role in other types of cancer has already been suggested. As regards intragenic miRNAs, 187 miRNA/host gene pairs were obtained after localizing miRNAs within the absolute 5′ and 3′ regions of genes represented on the HG-U133A arrays; 25 of these showed a significant correlation between miRNA and mRNA levels. Among the most correlated miRNA/hostgene pairs we identified miR-152/COPZ2, miR-342-3p/EVL, miR-335/MEST, miR-25 and miR-106b/MCM7. For some of the identified pairs, miRNA expression levels were validated by means of Q-RT-PCR. In conclusion, we showed that miRNA expression in HMCLs could be affected by the presence of genomic lesions or may correlate with host-gene modulation, suggesting a possible role in the molecular pathogenesis of MM. Our integrative approach represents the basis for further investigations, also in primary tumors, aimed at functionally characterizing specific miRNAs in MM.