Abstract
Abstract 3330
MicroRNAs (miRNAs) are important regulators of gene expression involved in virtually all physiological and pathological processes. Transcriptional profiling with subsequent bioinformatic analysis is increasingly used to identify miRNAs critical for hematopoiesis, normal blood cell function and hematological diseases. The biological variation observed in differential expression of these RNAs has been useful for better understanding of disease mechanism, and for identifying potential biomarkers and therapeutic targets. However, conclusions about the significance of measured biological variation is only legitimate when the RNA expression levels are normalized to internal controls, and most miRNA profiling data analyses are based on assumptions that have not been validated. For example, the expression level of U6 snRNA is often used to normalize expression across comparison tissues. However, it is not known whether U6 snRNA is expressed at equivalent levels among these cells or sample groups, and no reliable reference genes have been identified in cells from different hematopoietic lineages. The goal of this study was to identify reference miRNAs with the least variation for peripheral blood T-cells, B-cells, granulocytes and platelets.
Subsets of cells were obtained from the peripheral blood of 5 healthy donors using density centrifugation followed by immunoselection. High purity (>97%) of the cells was verified by flow cytometry. Total RNA was extracted with Trizol and miRNA profiled by Nanostring technology (Nanostring Technologies, Denver, CO). Two different statistical algorithms - the NormFinder (NF) and Coefficient of variation (CV) methods - were used to identify miRNA reference genes with high stability within and among cell types. Geometric mean normalized raw data was analyzed by NF and CV, and candidate normalizer genes were selected based on three criteria: (1) stability measure of the variation in miRNA values (low values being more stable) (2) present in the top 10 candidate normalizer genes by both the NF and CV methods, and (3) a moderate expression value of 300–1000 counts (range = 50–12,000 counts for 95% of miRNAs). The last criterion is important because variation in a very low abundance reference gene would have an inappropriately large effect on normalization of the data. The reference miRNAs identified were hsa-miR-30b and hsa-miR-151-5p for B-cells, hsa-miR-484 and hsa-miR-425 for platelets, hsa-miR-301a, hsa-miR-30d and hsa-miR-424 for granulocytes, and hsa-miR-140-3p and hsa-miR-101 for T-cells. Hsa-let-7b and hsa-miR-423-3p were identified as common normalizers across T-cells, B-cells, platelets and granulocytes with stability values of 0.207 and 0.331 by NormFinder and 0.0599 and 0.0591 by CV method as shown in Fig. 1 and Fig. 2. Both hsa-let-7b and hsa-miR-423-3p were validated by RT-PCR to be stable normalizers (CV of Ct values were 9% and 12%, respectively) across T-cells, B-cells, platelets and granulocytes. Notably, hsa-let-7b showed lower variation across cell types than U6 snRNA, indicating hsa-let-7b is a more reliable reference gene for quantification of miRNA data from hematopoietic cells.
In summary, we used a rigorous and formal approach to identify miRNAs in different hematopoietic lineages that can be used as appropriate reference transcripts for genome-wide profiling studies. Most ideally, this approach can be used for any new transcriptome data set. In addition, we provide ideal reference miRNAs for comparisons within or among different hematopoietic lineages. Normalization with the reference genes identified in this report, will allow more reliable and accurate determination of the biological variation involving differential expression of hematopoietic miRNAs than currently used normalizers.
No relevant conflicts of interest to declare.
