RNA Sequencing Of Adult Bone Marrow Erythroid Populations Reveals Significant Transcriptional Deregulation In Foxo3^-/- Mutant Cells

Transcriptional control of last stages of erythropoiesis is a complex and well orchestrated process controlled by lineage-specific transcription factors. The precise contribution of the different transcription factors to this multistep process has not been fully elucidated. Foxo3 is a transcription factor that is required for terminal erythroid maturation and Foxo3 mutant mice exhibit ineffective erythropoiesis. In order to gain further insight into the contribution of Foxo3 to the control of adult terminal erythroid maturation we analyzed the transcriptome of three adult bone marrow erythroid precursor populations: pro-, basophilic and polychromatophilic erythroblasts from wild type and Foxo3-/- mice. Populations were FACS sorted according to their TER119 and CD44 cell surface expression and FSC properties. RNA was then isolated and sequenced using the Illumina GaII platform. Genes were grouped into 3 categories according to their expression during erythroid cell maturation using the Short Time Series Expression Miner (STEM) program: no change (4577 genes), down-regulated (2868 genes) or up-regulated (2637) (Figure 1). Enrichment analysis of groups of genes using the ChEA database identified Myb, Meis1, Runx1, Fli1 and PU.1 as the main transcription factors regulating gene repression over erythroid maturation. In contrast, ChEA identified known erythroid transcription factors like Gata1, Eklf and Tal1 to drive the up-regulation of many of the erythroid-specific genes. This analysis also enabled the identification of putative novel transcription factors implicated in erythroid cell maturation. Interestingly, the difference between WT and Foxo3-/- cells increased gradually from pro- to polychromatophilic erythroblasts in correlation with increased Foxo3 expression during these steps of maturation. Strikingly, pathway enrichment analysis detected several immune-related pathways such as Toll-like receptors, TGF-β and IL-1 signaling as expressed in maturing wild type erythroid cells and significantly deregulated in Foxo3-/- cells. The expression of a number of these immune genes in erythroid cells has been validated by qRT-PCR. In addition, among others, a cluster of genes from the autophagy pathway was noted to be significantly down-regulated in Foxo3 mutant erythroid cells. In order to better dissect Foxo3 transcriptional control during erythroid maturation, STEM analysis of Foxo3-/- samples revealed an unexpected number of differences compared to WT. Most remarkably the STEM analysis identified that 90% of the 1198 genes that are continuously up-regulated during erythroid maturation from pro- to polychromatophilic are highly compromised in their level of expression during erythroid maturation in the absence of Foxo3. Interestingly, this group was also enriched for Foxo3 direct target genes as determined by ChIP-seq studies. We also identified a subset of genes whose expression increased from pro- to basophilic erythroblasts but decreased thereafter in the absence of Foxo3 in contrast to wild type cells. Interestingly, ChEA analysis on this group identified a subset of genes that are targets of Gata1, Eklf and Tal1 that may require Foxo3 for their full expression at the last stages of erythroid cell maturation. In conclusion, we present an unbiased genome-wide approach using RNA sequencing of adult bone marrow erythroid cells to study the contribution of Foxo3 to the regulation of gene expression at the last stages of erythroid cell maturation. This analysis enabled us to identify novel genes and pathways whose function in the control of red cell generation requires further investigations.