Enhancer Accessibility during Erythropoiesis and Megakaryopoiesis Correlates with Lineage-Specific Gene Expression

Enhancers are epigenetic regulatory modules critical to lineage-specific transcript expression. Understanding the development and maintenance of enhancers will help clarify lineage commitment decisions. The ENCODE Project Consortium used a variety of cell lines to define enhancer regulatory maps. Our goal is to build ENCODE-like enhancer maps for primary mouse hematopoietic cells. To accomplish this we have performed genome-wide open chromatin surveys (Assay for Transposable-Accessible Chromatin [ATACSeq]) and transcriptome analysis (RNASeq) in enriched populations of hematopoietic stem and progenitor cells (LSK), common myeloid progenitors (CMP), megakaryocyte-erythroid progenitors (MEP), and erythroid and megakaryocytic progenitors (CFUE and CFUMeg, respectively). In addition to ATACSeq and RNASeq we performed ChIPSeq in erythroblasts (EB) and megakaryocytes (MEG). Together these data provide a high-quality map of essential enhancers and correlated transcription profiles at specific stages of hematopoiesis.

Enhancers are associated with DNase I hypersensitivity and monomethylation at lysine 4 of histone H3 (H3K4me1), while active enhancer regions (AERs) are also acetylated at lysine 27 of histone H3 (H3K27ac). We identified several thousand candidate enhancer regions (cERs) in EB and MEG (Table). More than 70% of cERs also contained the H3K27ac mark and thus were candidate AERs (cAERs). In 90% of cases cAERs were closer to an active transcriptional start site (TSS) than to any other TSS. 23 of these erythroid cAERs were shown to be active in a luciferase assay.

We next evaluated the major trends in acquisition and maintenance of cERs during hematopoietic differentiation and their correlation with altered gene expression. We focused on cAERs and examined their appearance and retention in progenitor cells (Table). Almost all cell-specific (97% of EB and 93% of MEG) cAERs are in accessible chromatin as monitored by ATACSeq (HC cAERs). This high degree of correlation allows us to use overlaps with progenitor ATACSeq data to estimate the presence of the cAERs identified in the mature cells. Our initial results show that the mature cell HC cAERs overlap with progenitor cell ATACSeq data (MEP and CMP) is greater for EB cAERs than MEG, suggesting that chromatin in these cells is more similar to that of EB. However, in LSK, 50% of both EB and MEG cAERs overlap with ATACSeq peaks, suggesting that half of cell-specific cAERs are present in LSK while the rest are established during differentiation. In contrast, 82% of shared EB and MEG cAERs overlapped with LSK ATACSeq peaks and were maintained throughout differentiation.

Candidate super enhancers (cSEs) are the cAERs with the highest levels of H3K27ac as measured by ChIPSeq. Within the top 2% of AERs we identified 101 EB and 98 MEG HC cSEs (i.e., overlap ATACSeq peaks). There was no overlap between the EB and MEG cSEs, indicating that cSEs are more cell-specific than all cAERs (Χ²≤ 0.04). RNASeq data confirmed that 92% of cSEs were closest to an active TSS. These data indicate that cSEs are more cell-specific than cAERs and are associated with increased gene expression.

Approximately 96% of EB-specific cSEs correlated with ATACSeq peaks in CFUE chromatin, 94% in MEP, 82% in CMP, and 65% in LSK. In contrast, 95% of MEG-specific cSEs correlated with ATACSeq peaks in CFUMeg chromatin, 81% in MEP and CMP, and 75% in LSKs. The higher percentage of MEG cSEs in LSK is significantly different from all MEG cAERs (Χ²≤ 0.003), indicating that, unlike cAERs, MEG-specific cSEs are established early and maintained throughout differentiation.

We have shown that cAERs, especially cSEs, are highly cell-specific in primary murine erythroid and megakaryocytic cells and correlate with gene expression. Examining changes in cAERs, cSEs, and gene expression allows us to map the specific epigenetic changes in chromatin that define erythroid and megakaryocytic differentiation. These results will enable us to test hypotheses about the mechanism of erythroid and megakaryocytic lineage commitment.

 

	EB-specific		MEG-specific		Common
	No.	Percent	No.	Percent	No.	Percent
cER	7153	n/a	4196	n/a	619	n/a
cAER	6097	100%	3064	100%	468	100%
HC cAER	5934	97%	2848	93%	434	93%
HC cAER + CFUE ATACSeq	5818	98%	n/a	n/a	n/a	n/a
HC cAER + CFUMeg ATACSeq	n/a	n/a	2606	92%	n/a	n/a
HC cAER + MEP ATACSeq	5591	94%	1509	53%	394	91%
HC cAER + CMP ATACSeq	4754	80%	1473	52%	385	89%
HC cAER + LSK ATACSeq	3719	63%	1368	48%	351	81%