INTEGRATIVE GENOMIC APPROACH IDENTIFIES DEREGULATED MICRORNAS IN HUMAN MYELOMA CELL LINES

The role of microRNAs (miRNAs) in multiple myeloma (MM) remainsto be fully elucidated. To identify miRNAs potentially deregulated inMM, we performed an integrative analysis of genome-wide DNA copynumber (CN), gene expression profiling and miRNAs expression profilingin a panel of 16 human myeloma cell lines (HMCLs). Global miRNAand mRNA expression data were generated on Agilent miRNA microarrays(representing 470 human mature miRNAs) and AffymetrixGeneChip HG-U133A arrays, respectively; genome-wide profiling datawere generated on Affymetrix GeneChip Human Mapping 250K Nsparrays. CN values were inferred using the DNAcopy R Bioconductorpackage. To measure the correlation between the expression levels ofeach miRNA and the corresponding CN values, non-parametric analyseswere performed (Kendall's tau correlation). We identified 22 miRNAswith a good correlation (p<5×10–2) between expression levels and localCN variations, suggesting that a dosage effect could be a main mechanismunderlying their expression. The identified miRNAs are mappedwithin different genomic regions, including chromosome 22q11.21,8q24.22, 17, 7 and 16q22, and for some of them a role in other types ofcancer as well as in MM has already been reported or suggested. Thesame approach was applied in order to investigate the associationbetween intronic miRNA and corresponding host-gene expressions. Apositive correlation (p<5×10–2) was found for 22 intronic miRNAs mappedwithin 20 different deregulated host genes, strongly supporting the suggestionthat intronic miRNAs and their host genes share the same regulatorysequences and can be co-transcribed. Among the most correlatedmiRNA/host-gene pairs we identified miR-342-3p/EVL and miR-335/MEST, for which the miRNA expression levels were validated bymeans of Q-RT-PCR. The presence of a correlation was further confirmedin a fraction of primary tumors. Some miRNA/host-genes expressionalso shared a significant correlation with CN variations. In conclusion, wedemonstrated that the presence of genomic lesions and the expression ofhost-genes can affect miRNA expression in HMCLs. We are nowattempting to investigate these findings in primary tumors in order to betterdefine the role of miRNAs in the pathogenesis of multiple myeloma.