MicroRNA Expression Profiling of High and Low Risk MDS

Abstract

MicroRNA's (miR) are small noncoding RNA's of 18–25 nucleotides that have a critical impact on gene regulation affecting cell growth & differentiation. Importantly, miR expression profiles can help distinguish normal cells from cancerous cells. This is particularly relevant to myelodysplastic syndromes both because of their heterogeneity as well as the difficulties associated with the early diagnosis of these disorders. We investigated the miR profiles of CD34+ and total nucleated cells (TNC) from different subtypes of MDS (n=135) and compared these to AML (n=20), normal CD34+ (n=5) and normal TNC's (n=6). We used the Illumina universal array matrix to interrogate 470 validated miR's from the Sanger mir-Base database and a further 265 putative miR's from the literature. Array intensity data was analysed using the Partek Genomics Suite v6.3. Data was normalised using quantile normalisation and unchanging miR's removed. A 1-way ANOVA was used to identify differentially expressed miR's and a FDR correction applied to control for Type I errors. Our initial analysis compared TNC (n=6) with CD34 cells (n=5) from normal controls. As expected both groups clustered separately with mir-199a (p<10−9), mir-30a-3p (p<10−7), mir-507 (p<10−7) showing the most significant differences out of a total of 250 differentially expressed miR's. Our results clearly identified an expression signature for low risk (LR), high risk (HR) and AML using data from both TNC's and CD34+ cells compared to normal TNC and CD34+ cells. Cluster analysis of miR's of CD34+ cells from normal controls (n=5) and monosomy 7 (−7) patients (n=6) identified a panel of 18 miR's that differentiated both groups. Similarly, analysis of CD34+ cells from HR MDS (n=7) and normal CD34+ controls identified a unique panel of 34 miR's differentiating both groups. Analysis of CD34+ cells from both del(5q) (n=4) and LR MDS cases (n=8) identified a panel of 154 and 23 miR's respectively that differentiated these from the normal controls. A 4-way intersect analysis of miR expression from CD34+ cells from LR, HR, del(5q) and −7 groups did not identify any common miR's when compared to normal CD34+ cells. However, a 3-way analysis using LR, HR and −7 groups identified mir-34a as a single common miR. The mir-34 family (a, b, c) are transcriptional targets for p53 and have been shown to be induced by DNA damage and oncogenic stress (He et al, 2007). A direct comparison of miR expression in CD34+ cells from LR and HR showed only 4 miR's that are differentially expressed with significant fold changes; mir-656 (53x), mir-498 (32x), mir-181b (−17x) and mir-130a (−4x). Such differences where not observed in TNC from LR and HR samples. A 4-way intersect analysis on TNC's from LR (n=47), del(5q) (n=5), HR (n=50) and −7 (n=8) cases identified 8 common miR's; mir-34b, mir-139, mir-429, mir-519a, mir-548b, mir-561, mir-580 and mir-619. Comparison of the miR expression in TNC from AML samples against normal TNC identified 49 differentially expressed miR's including let-7e, mir-181a, mir- 181b, mir-199a and mir-221 in accordance with previous published miR profiles in AML (Dixon-McIver et al, 2008). We also observed a 17 fold down regulation of mir-127 which is often silenced in cancer cells due to the methylation and acetylation status of the mir-127 promoter. The proto-oncogene BCL6 is a verified target of mir-127 (Saito et al, 2006). To our knowledge this is the first study in MDS that shows unique miR profiles in TNC for each subgroup when compared to normal TNC. Furthermore, miR profiles from CD34+ cells distinguish between the different MDS subtypes. In conclusion our analysis has shown that profiling miR's in MDS results in unique profiles which may be used to discriminate between HR and LR MDS subgroups.