Abstract
Background: Prior studies using microarray platforms have shown alterations of gene expression profiles (GEPs) in MDS CD34+ marrow cells related to clinical outcomes (Sridhar et al, Blood 2009, Pellagatti et al, JCO 2013). Given the increased sensitivity and accuracy of high-throughput RNA sequencing (RNA-Seq) (Mortazavi et al, Nat Meth 2008, Soon et al, Mol Syst Bio 2012) for detecting and quantifying mRNA transcripts, we applied this methodology for evaluating differential gene expression between MDS and normal CD34+ marrow cells.
Methods:RNA was isolated from magnetic bead affinity-enriched CD34+ (>90%) marrow aspirate cells (Miltenyi Biotec, Auburn, CA) and amplified using the Smarter Kit (Clontech, Mt View, CA). The amplified product (ds DNA) was fragmented to a size distribution of ~200-300bp using the E220 Focused Ultrasonicator (Covaris Inc, Woburn, MA). End repair, adapter ligation and PCR amplification were performed using the NEBNext Ultra RNA library prep kit for Illumina (New England Biolabs, Ipswich, MA). The indexed cDNA libraries were sequenced (paired end, 100bp) on an Illumina HiSeq2000 platform with median read counts of 69 million. The sequences were aligned to Human Reference sequence hg19 using DNAnexus mapper with gene detection focused on known annotated genes. The differential expression was analyzed using edgeR. DAVID and Ingenuity IPA programs were used for pathway analyses. Gene Set Enrichment Analysis (GSEA) was used to identify biologic processes in our dataset present across phenotypes.
Results: Correlations of RNA-Seq data from unamplified to amplified transcripts demonstrated high fidelity of transcripts obtained (Pearson and Spearman R2 = 0.80). After filtering samples for adequate read counts, 12,323 genes were evaluated. Differential expression analysis yielded 719 differentially expressed genes (DEGs) in MDS (n=30) vs normal (n=21) with FDR <.05. Among the DEGs, 548 and 171 were over- and under-expressed ≥2 fold in MDS vs Normal, respectively: 20% of the overexpressed genes were present in >50% of the patients. Hierarchical cluster analysis using these DEGs confirmed clear separation of MDS patients from normals, with 2 differential expression clusters—one region overexpressed and one underexpressed. A distinctive trend toward clustering of the patients was seen which related to their IPSS categories and marrow blast %. In functional pathway analysis of the 2 distinctive gene clusters which distinguished MDS from normal, the underexpressed MDS DEGs demonstrated enrichment of inflammatory cytokines, oxidative stress and interleukin signaling pathways, plus mitochondrial calcium transport; whereas the MDS overexpressed DEG cluster showed enrichment of adherens junction/cytokeletal remodeling, cell cycle control of chromosome replication and DNA damage response pathways. Using GSEA analysis, significantly increased numbers of genes in MDS vs normal, common to those in gene sets present within curated public databases, were involved with TP53 targets and mTOR signaling pathways.
Conclusions: Our study demonstrated that RNA-Seq methodology, a high-throughput and more comprehensive technique than most gene expression microarrays, was capable of showing significant and distinctive differences in gene expression between MDS and normal marrow CD34+ cells. Specific clustering of the DEGs was demonstrated to distinguish patient subsets associated with their major clinical features. Further, the stringently identified DEGs shown to be engaged in functional pathways and biologic processes highly relevant for MDS were extant within the patients’ CD34+ cells. These transcriptomic data provide information complementary to exomic mutational findings contributing to improved understanding of biologic mechanisms underlying MDS.
Disclosures
No relevant conflicts of interest to declare.
RankerGUI: A Computational Framework to Compare Differential Gene Expression Profiles Using Rank Based Statistics
