Genome-Wide ChIP-Seq Reveals a Dramatic Shift in the EKLF Binding Profile Between Erythroid Progenitors and Erythroblasts.

Abstract 565

Erythroid Kruppel-Like Factor (EKLF; KLF1) is the founding member of the Kruppel family of C2H2 zinc finger transcription factors. First identified as an activator of the beta-globin locus, EKLF facilitates chromatin remodeling and transcriptional activation of target genes, at least in part through recognition of a 9-base consensus motif (NCNCNCCCN). By comparing the transcriptional profiles of E13.5 wild type and Eklf-/- mice, we demonstrated that the lethal failure to complete definitive erythropoiesis in the fetal liver (FL) was due in part to dysregulation of an EKLF target gene, the cell cycle control factor, E2F2 (Pilon et al. 2008). To identify further direct targets of EKLF activation that affect erythropoiesis, we are coupling chromatin immunoprecipitation with ultra high-throughput massively parallel sequencing (ChIP-seq). ChIP-seq is increasingly being used to map protein-DNA interactions in vivo, allowing simultaneous genome-wide analysis of transcription factor occupancy, defining an ‘interactome‘. Using mice whose endogenous Eklf gene was replaced with a fully functional HA-tagged form of EKLF, chromatin was isolated at E13.5 from immature erythroid progenitors and maturing erythroblasts by ChIP. Using a highly specific high-affinity anti-HA antibody, libraries of HA-EKLF-bound chromatin were subjected to fluorescent in situ sequencing on a Solexa 1G platform, providing 36-base signature tags that were mapped to the mouse genome using the Eland software package. A control library was derived from E13.5 FL chromatin that was not enriched for HA-EKLF occupancy. For both progenitors and erythroblasts, >1.1×10⁷tags were obtained. 72.5% and 78.7% of progenitor and erythroblast tags mapped to unique sites within the genome, respectively. The tags were highly enriched in the ∼10% of the genome within genes (genic; 42% of tags), sites ≤10 kb from the nearest gene (adjacent; 15%), as opposed to the ∼90% of the genome that is >10 kb from the nearest gene (intergenic; 22%) or in repetitive DNA (21%) p=2.2 ×10^-16. Using the MACS software package clustered peaks of EKLF occupancy were identified throughout the genome, defining the EKLF ‘interactome‘. The vast majority of peaks were mapped to non-repetitive regions of the genome (98% in progenitors; 95% in erythroblasts). Progenitors contained 4,383 peaks of EKLF occupancy, while erythroblasts contained 15,396 peaks. Only 100 peaks were common between populations. This >3.5-fold increase in genomic EKLF occupancy between progenitors and erythroblasts (p=1×10^-5) reflects the shift in the expression and activity of EKLF protein in erythropoiesis described previously (Bouilloux et al. 2008; Lohmann & Bieker 2008). To identify potential EKLF target genes, we partitioned the genome into 3 categories, relative to annotated RefSeq coordinates (genic) as well as adjacent and intergenic. In progenitors, the majority of EKLF binding (54%) occurred in intergenic regions, with a minority within (38%) or adjacent (7%) to genes. By contrast, the EKLF binding profile in erythroblasts was reversed, with 62% of the peaks in genic regions, and a minority at intergenic (26%) or adjacent (12%) sites.To assess the effect of this shift in EKLF binding on gene transcription, we used publicly availabel data from the inducible G1E model of erythroid maturation (GEO: GSE628) to correlate our ChIP-seq data with mRNA expression. Informatic analyses using MetaCore demonstrated that >2,200 EKLF-associated genes were differentially expressed during maturation (949 increasing expression; 1,298 decreasing expression, all p<0.05). Among progenitors, control of cell cycle S-phase entry and progression was a significantly represented network, highlighted by focal EKLF target genes like Cdk2, Cdk4, and p107, in agreement with our previous observations. Among erythroblasts, the erythropoietin (Epo) signaling pathway was most significantly represented, highlighted by focal EKLF target genes like Stat3 and Bcl-XL, reflecting the well-established importance of the Epo axis for erythroblast survival. These data indicate that shifts in EKLF occupancy during erythropoiesis correlate with distinct functional effects on gene expression. Further, these observations support a model in which transcriptional regulators (e.g., EKLF) may collect at intergenic locations when their activity is not required, but where they remain poised for rapid recruitment.