Abstract
MicroRNAs (miR) are non-coding RNAs that regulate gene expression by pairing to 3UTRs of target genes inducing translational repression or mRNA cleavage. New evidence suggests that the latter mechanism markedly contributes to miRNA effects. Hence, global gene expression analyses may help elucidate the functional role of miRNA by recognizing pathways modified by their abnormal expression and identifying direct targets. MiR155, the product of the non coding gene BIC, is overexpressed in lymphomas and its role in tumorigenesis is supported by the development of B-cell malignancies in miR155 transgenic mice. However, the functional consequences of miR155 overexpression in tumor development remain unclear. To address this issue, we developed a semi-quantitative RT-PCR assay that specifically amplifies either the nuclear unspliced BIC mRNA (target of the RNase III Drosha) or the spliced BIC mRNA. We found a marked correlation between the expression levels of these two mRNAs, which in turn agreed with the levels of mature miR155 detected in northern blots. Of the 22 DLBCL cell lines studied, only 5 (DHL6, Ly3, Ly10, Farage, RCK-8) expressed significantly high levels of BIC and miR155. To isolate the effects of miR155 in DLBCL we genetically modified its expression and performed global transcription analysis on microarray. In brief, we cloned the BIC transcript in a MSCV-GFP bicistronic retrovirus and confirmed in transduced HeLa cells that the mature miR155 was expressed when this transcript was driven by an LTR promoter. Next, we used two DLBCL cell lines with low levels of miR155 (Ly8 and Ly19) to generate unique populations expressing miR155 or MSCV alone. RNA was isolated from GFP-sorted cells, hybridized to the Affymetrix U133Plus2.0 chip and the data analyzed with dChip. Remarkably, and in agreement with the role of miRNAs, supervised analysis (fold diff >1.7) revealed a vast predominance (>90%) of downregulated genes when comparing miR155 expressing cells to MSCV only. These gene groups included predicted miR155 targets and were significantly enriched for molecules involved in the immune response (p<.001), including MHC class II, chemokine receptors, TDT, NFAT and CD24, a particularly relevant target for miR155 inhibition since its activation induces apoptosis in lymphomas. To validate and extend our findings, we queried public expression datasets of primary DLBCL. First, we used our cell lines expression data to confirm that the BIC probe in the Affymetrix chip reliably reflected the expression of miR155. Since DLBCL entails at least two groups of tumors reflecting distinct normal B-cells (GC and ABC) and the expression of miRNA and its targets are highly tissue/cell specific, we compartmentalized these analyses within the groups of ABC and GC tumors. We selected the tumors within each group with the highest and lowest levels of BIC (20% percentile) and performed unsupervised hierarchical clustering analysis (filtering parameters 0.5<SD/mean<10). In agreement with our data in the GC-type miR155-expressing cell lines, we found that in GC-, but not in ABC tumors, the expression of BIC inversely correlated with that of the genes related to the immune response. Notably, the pro-apoptotic molecule CD24 was significantly downregulated (p<0.02) in BIC overexpressing primary DLBCL, underscoring the need for further characterization of the signals relayed by this surface molecule and its potential as a rational drug target. Our data start to delineate the effects of miR155 in DLBCL and show the potential of expression arrays to identify miRNA targets modified by mRNA cleavage.
Contribution of CTCF binding to transcriptional activity at the HOXA locus in NPM1-mutant AML cells
