Novel Immediate Direct Targets of EPO Signalling in Human Erythropoiesis

Introduction: Erythropoietin (EPO) regulates expression of genes that drive proliferation, survival and differentiation of erythroid progenitor cells into mature erythrocytes. The EPO receptor signals via JAK2 and STAT, PI3K and MAPK pathways. Only a few direct target genes of these pathways have been identified to date. Aim: To determine the repertoire of EPO-induced target genes in human erythroid cells Methods: We employed a well characterized conditionally immortalized EPO-dependent human erythroid progenitor model (HUDEP-2 cells). The cells require stem cell factor and EPO for growth, but can be starved of both and then re-challenged with either growth factor. A multi-omics approach was employed to find direct target genes of EPO stimulation. We performed ChIP-seq for pSTAT5, ATAC-seq, and a new method of primary RNA-seq based on SLAM-seq but using full length cDNA libraries. Results: Three hours of starvation followed EPO (5U/ml) stimulation for 1 hour resulted in rapid phosphorylation of STAT5, but not STAT1 or STAT3. ChIP-seq or pSTAT5 identified 3128 EPO-induced binding sites. The majority of peaks contain a palindromic 'GAS' motif (TTCYXRGAA), and are located at intronic (50%), distal (29%) and intergenic (15%) enhancers; only 3% are located at promoters. De novo motif discovery identified significant enrichment of DNA-binding motifs for GATA and KLF transcription factors (TFs), suggesting co-operativity between EPO signalling and the essential basal erythroid TFs, GATA1 and KLF1. We show GATA1 and KLF1 bind most of these by ChIP-seq. We found 14,535 new ATAC-seq peaks after EPO stimulation, at both enhancers and promoters. Only ~8% overlap with pSTAT5 ChIP-seq peaks, suggesting EPO-mediated phosphorylation and DNA-binding of undiscovered TFs. By searching for enriched motifs within these regions, we identified binding sites for TFs of the NFY, EGR, and NRF1 families and others, suggesting a hidden complexity of transcription factors that mediate responses to EPO. ChIP-seq for EGR1 is in progress and will be reported. We developed a novel metabolic labelling technique and bio-informatic pipeline, called BodySLAM-seq, to determine the immediate transcriptional targets of EPO. We used this to find ~100 differentially transcribed genes (DTGs) as well as differentially expressed genes (DEGs) in immediate response to EPO (45 min labelling period). Some are direct pSTAT5 target genes such as BCL2L1, PIM1, and CISH; others are novel targets involved in transcription regulation, erythropoiesis and signalling. Some are known immediate early genes such as EGR1. The RNAs for these are rapidly degraded in the absence of EPO and rapidly induced. Other genes are not degraded in the absence of EPO but are nevertheless rapidly induced. These were only detectable by BodySLAM-seq. Conclusions: We have discovered new pSTAT5 direct target genes, and also new DNA motifs that mediate responses to EPO, in human erythropoiesis. We have discovered a set of immediate early genes (IEGs) that are rapidly degraded in the absence of an EPO stimulus and rapidly induced again by EPO. These are involved in negative regulation of JAK-STAT signalling, cell proliferation, cell survival, and transcription. They are not involved in erythroid cell differentiation specifically. This suggests the main function of EPO is to induce cell proliferation and survival rather than differentiation.