Abstract
Abstract 780
Gene expression analyses of mammalian erythroid precursors have been limited to time series generated from in vitro maturation model systems, one or two time point analyses from in vivo-derived samples, or pairwise comparisons of grouped precursors compared with a mutant phenotype. Despite the fact that erythroid cells comprise >25% of the cells of the mammalian fetus and adult, there have been no analyses of gene expression 1) of multiple stages of primary erythroid precursors, or 2) of similar maturational stages derived from primitive, fetal definitive and adult definitive erythroid lineages. Erythroid precursors have classically been defined using morphological characteristics following Wright-Giemsa staining, including cell size, nuclear condensation, nuclear to cytoplasmic ratio, and loss of cytoplasmic basophilia due to decreased ribosomes and increased hemoglobin. Recently, progressive stages of erythroid precursors have been defined by cell surface expression of glycophorin A/Ter-119, CD71 and CD44. It has been difficult to compare and interpret data derived from these two different approaches. We devised a cell sorting strategy utilizing a combination of cell surface expression and scatter related to size with stains for RNA and DNA to purify progressive stages of erythroid precursors (proerythroblast, ProE; basophilic erythroblast, BasoE; polychromatophilic/orthochromatic erythroblast, Poly/OrthoE; reticulocyte, Retic) that correlate well with the morphological series identified by Wright-Giemsa staining. RNA was isolated from four maturational stages (ProE, BasoE, Poly/OrthoE, and Retic) derived from three erythroid lineages: 1) “primitive” erythroid, from yolk sac and embryonic bloodstream, 2) “fetal definitive” erythroid, from E14.5 liver, and 3) “adult definitive” erythroid, from the bone marrow. Gene expression data from these samples were obtained using Affymetrix Genechip arrays. Initial analysis of the dataset indicates robust, reproducible clustering of samples within replicates of each stage/lineage. Hierarchical clustering analysis reveals both stage- and lineage-specific gene sets. A large number of genes are differentially expressed in the reticulocyte stage, regardless of lineage. Intriguingly, initial analysis also indicates that of the 12 stage/lineage data sets, the adult ProE and primitive Poly/OrthoE had the most divergent gene expression patterns distinguishing them from the other samples. Genes representing different expression patterns predicted by abundance data were confirmed using qPCR analysis. Cluster analysis as well as gene ontology mapping indicate a diverse set of expression patterns and molecular functions are present during erythroid maturation. Lineage-specific gene-interaction networks have been constructed and we are analyzing their topology to determine those most essential to erythroid maturation. Gene interactions were determined based on ranked co-expression of genes across our cell stages. These interactions are annotated by known and computationally predicted transcription factor targets, pathways (e.g., metabolic, cellular process, cell-signaling), and known erythroid-specific interactions and can be filtered according to cell-stage specific gene expression and gene function. We are developing a public access website that will aid in the analyses of these data through a searchable database of predicted and known gene-interactions. The site will facilitate comparison of gene-expression and function among the erythropoietic lineages by allowing the visualization and annotation of lineage-specific local-gene interaction networks. These studies provide the first gene expression data from defined stages of normal, primary erythroid precursors that constitute a significant portion of the embryonic, fetal and adult erythron.
Disclosures:
No relevant conflicts of interest to declare.
Multi-Omics Data Analysis Uncovers Molecular Networks and Gene Regulators for Metabolic Biomarkers