Background: Enhancers and promoters are cis-acting regulatory elements associated with lineage-specific gene expression. Previous studies showed that different categories of active regulatory elements are in regions of open chromatin, and each category is associated with a specific subset of post-translationally marked histones. These regulatory elements are systematically activated and repressed to promote commitment of hematopoietic stem cells along separate differentiation paths, including the closely related erythrocyte (ERY ) and megakaryocyte (MK) lineages. However, the order in which these decisions are made remains unclear. Results: To characterize the order of cell fate decisions during hematopoiesis, we collected primary cells from mouse bone marrow and isolated 10 hematopoietic populations to generate transcriptomes and genome-wide maps of chromatin accessibility and histone H3 acetylated at lysine 27 binding (H3K27ac). Principle component analysis of transcriptional and open chromatin profiles demonstrated that cells of the megakaryocyte lineage group closely with multipotent progenitor populations, whereas erythroid cells form a separate group distinct from other populations. Using H3K27ac and open chromatin profiles, we showed that 89% of immature MK (iMK)-specific active regulatory regions are present in the most primitive hematopoietic cells, 46% of which contain active enhancer marks. These candidate active enhancers are enriched for transcription factor binding site motifs for megakaryopoiesis-essential proteins, including ERG and ETS1. In comparison, only 64% of ERY-specific active regulatory regions are present in the most primitive hematopoietic cells, 20% of which containing active enhancer marks. These regions were not enriched for any transcription factor consensus sequences. Incorporation of genome-wide DNA methylation identified signifi- cant levels of de novo methylation in iMK, but not ERY. Conclusions: Our results demonstrate that megakaryopoietic profiles are established early in hematopoiesis and are present in the majority of the hematopoietic progenitor population. However, megakaryopoiesis does not constitute a “default” differentiation pathway, as extensive de novo DNA methylation accompanies megakaryopoietic commit - ment. In contrast, erythropoietic profiles are not established until a later stage of hematopoiesis, and require more dramatic changes to the transcriptional and epigenetic programs. These data provide important insights into lineage commitment and can contribute to ongoing studies related to diseases associated with differentiation defects. Keywords: Hematopoiesis, Megakaryopoiesis, Erythropoiesis, ATAC-Seq, RNA-Seq, ChIP-Seq, Lineage commitment, Epigenetics *Correspondence: firstname.lastname@example.org NHGRI Hematopoiesis Section, GMBB, Bethesda, MD, USA Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 2 of 18 hematopoiesis. As part of the ValIdated Systematic Inte- Background gratiON of hematopoietic epigenomes (VISION) project, Hematopoiesis is the process by which proliferating we are creating comprehensive catalogs of genomic regu- hematopoietic stem cells (HSC) undergo continued tran- latory elements in primary mouse cells to compare and scriptional and epigenetic changes associated with line- contrast epigenetic regulation in mice with human data age-restriction and cell-specific function, giving rise to from the Encyclopedia of DNA Elements (ENCODE) all of the cell types in the hematopoietic system. Classi- project and the NIH Roadmap Epigenomics Project cal models of hematopoiesis describe a series of progres- [21–24]. VISION also emphasizes bioinformatics mod- sively restricted cell fate decisions in which HSC give rise eling and machine learning to predict regulatory interac- to multipotent progenitors (MPP), which in turn give tions which is enabled by the ability to manipulate mouse rise to common myeloid progenitors (CMP) and com- hematopoiesis at the genetic level. mon lymphoid progenitors (CLP). In myelopoiesis, CMP To characterize the relationships between hemat- differentiate into granulocyte–macrophage progenitors opoietic populations and the changes that accompany (GMP), from which mature granulocytes and monocytes erythro-megakaryopoiesis, we enriched 10 primary are generated, and megakaryocytic erythroid progenitors hematopoietic cell populations from C57BL6 mice for (MEP), which have classically been described as the com- RNA-Seq, ATAC-Seq, ChIP-Seq, and MBD-Seq analy- mon erythrocyte and megakaryocyte progenitor [1–4]. sis. These data have been integrated into transcriptional Erythro-megakaryopoiesis describes the subset of cell and epigenetic maps to identify candidate regulatory fate decisions associated with the production of eryth- elements in the ERY and MK lineages and to study the roblasts (ERY; erythropoiesis) and megakaryocytes (MK; establishment and maintenance of these regulatory ele- megakaryopoiesis). The process of erythroid and mega - ments during erythro-megakaryopoiesis. Our results karyocyte commitment is accompanied by the substitu- demonstrate that committed megakaryocytes have simi- tion of GATA1 for GATA2 at key chromatin regulatory lar transcriptional and epigenetic profiles to hematopoi - sites to increase expression of linage-specific transcrip - etic stem and progenitor cells, but do not represent a tion factors including FLI1 and ETS in the MK lineage, “default” developmental program. By contrast, the eryth- and KLF1 and GATA1 in the ERY lineage [5–8]. Tradi- roid population is the most dissimilar from hematopoi- tional hierarchical models predict that ERY and MK etic stem and progenitor populations and requires more lineages are derived from a homogeneous population of extensive changes to the transcriptional and epigenetic bipotential MEP (Fig. 1a) . However, there is mount- programs to permit erythropoiesis. ing evidence that the MK program is established prior to the emergence of erythroid cells. Several studies have Results noted transcriptional and immunophenotypic similari- Committed megakaryocyte precursors have similar ties between HSC and MK populations [10–13]. Single- transcriptional profiles to hematopoietic progenitor cells cell transcriptional studies of human MEP demonstrated RNA-Seq was performed on two biological replicates that traditionally defined MEP included erythroid- and of total RNA isolated from 10 populations of primary megakaryocyte-committed cells mixed with a small com- mouse bone marrow separated on the basis of cell surface partment of bipotential intermediates . These studies marker expression (Fig. 1a, b). We defined “expressed” have led to revised models where many MK are derived transcripts as having a transcripts per million count from a multipotent stem cell, while others can be gener- (TPM) ≥ 1 (Table 1). Because our primary megakaryo- ated from CMP and MEP intermediates . While MK cytes were not treated with thrombopoietin, these cells appears to have multiple origins, in these revised models do not express many of the genes involved in platelet the erythroid lineage appears to remain a single branch function; we therefore refer to this population as imma- downstream of the CMP and formed through the MEP. ture MK (iMK). Determining when lineage specification occurs is Unsupervised hierarchical clustering of the individual important for understanding mechanisms behind dis- RNA-Seq replicates showed that lineage-restricted mega- eases with maturation defects. These diseases include karyocytic populations (CFU-MK and iMK) clustered anemias, thrombocytopenias, neutropenias, myelod- with progenitors (LSK, CMP, GMP, and MEP), while lin- ysplastic and myeloproliferative disorders, and hema- eage-restricted erythroid (CFU-E and ERY) and granu- tologic malignancies [16–20]. Fluorescence-activated locyte populations grouped in different clusters (Fig. 2a). cell sorting and next-generation sequencing technolo- We used principal component analysis (PCA) to display gies make it possible to generate transcriptome profiles relationships among the transcriptomes on a plane repre- (RNA-Seq), genome-wide profiles of chromatin accessi - senting the largest variance in the datasets. As expected, bility (ATAC-Seq), histone modifications (ChIP-Seq), and replicates for each cell population fell closer to each other DNA methylation (MBD-Seq) of cells at specific stages of Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 3 of 18 Fig. 1 Primary mouse bone marrow cells isolated via flow cytometry. a Hematopoietic populations assayed in this study. Branches represent lineage-commitment points. Cell surface markers used for flow cytometry enrichment are shown. b Sort logic is shown for flow cytometry enrichment Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 4 of 18 Table 1 Numbers of features ascertained in each cell population Population Number of transcripts Number of transcripts Number of ATAC peaks Number (TPM ≥ 1) (TPM ≥ 10) of methylation peaks LSK 13,066 7515 64,043 33,754 CMP 12,348 7065 107,281 10,288 CFU-E 9631 3834 41,805 16,439 ERY 10,025 3525 42,260 5218 CFU-MK 11,889 7059 122,757 69,405 iMK 11,751 6848 97,463 52,754 MEP 11,574 6834 76,334 N/A GMP 11,749 6972 112,947 N/A NEU 12,250 6330 44,239 N/A MONO 11,857 6239 53,531 N/A than to other cell types. Principal component 1 (PC1), clustered separately (Fig. 3a). These groups of commit - which accounted for 30% of the variance, separated ted erythroid cells and committed granulocytes were erythroid, monocyte, and neutrophil cells from pro- also observed in the PCA, while CFU-MK and iMK again genitor cells. PC2, which accounted for 23% of the total grouped with the multilineage progenitor cells (Fig. 3b). variance, separated erythroid cells from all other cells Unlike the transcriptional profiles, profiles of MEP chro - (Fig. 2b). Strikingly, the transcriptome of the committed matin accessibility clustered with erythroid cells rather MK progenitor, CFU-MK, grouped with progenitor cells, than with the multilineage progenitors. while the iMK grouped closest to this population. Of ~ 64,000 ATAC-Seq peaks in LSK, more than 95% Of the 13,000 transcripts expressed in our multipotent of these were present in the set of ~ 100,000 CMP peaks progenitor population (LSK), 81% were also expressed (Fig. 3c, Table 1). Similarly, 76% (~ 48,695) of LSK ATAC- in the iMK population (Fig. 2c, solid circle); this frac- Seq peaks were present in the set of ~ 97,000 iMK peaks tion represents 89% of all iMK transcripts (Fig. 2c, dot- (representing 72% of all iMK peaks) (Fig. 3c, Table 1). In ted circle). In contrast, 63% of transcripts expressed in contrast, only 33% (~ 21,000) of LSK ATAC-Seq peaks LSK were expressed in ERY, representing 82% of all ERY were present in the set of ~ 42,000 ERY peaks (repre- transcripts. This fraction is significantly smaller than senting 50% of all ERY peaks). While less overlap was that observed for the iMK population (proportion test observed among CMP, ERY, and iMK ATAC-Seq pro- p value < 0.001) (Fig. 2c, Table 2). Of the almost 12,000 files compared to the overlap observed in RNA-Seq pro - expressed CMP transcripts, 1930 (16%) were expressed files, a substantially larger number of peaks were shared in iMK but not in ERY, whereas only 294 (2%) were between iMK and CMP than between ERY and CMP expressed in ERY but not in iMK (proportion test p (Fig. 3d). We conclude that iMK maintains much of the value < 0.001) (Fig. 2d, Table 2). These results show that transposase-accessible chromatin that is established in cellular transcriptomes of the megakaryocytic lineage are LSK, while erythropoiesis involves significant loss of similar to those of multilineage progenitor cells, whereas accessibility in chromatin that was open in progenitors, erythroid cells repress the multilineage transcriptome. implying substantial compaction of chromatin during erythropoiesis. Committed megakaryocytes have a similar chromatin accessibility profile to hematopoietic progenitor cells Megakaryocytic regulatory elements are established We generated maps of accessible chromatin regions using in multipotent progenitor cells, while many erythroid the Assay for Transposase-Accessible Chromatin (ATAC) regulatory elements are established in more differentiated on the same populations of primary mouse bone mar- cells row cells used for transcriptional profiling. Unsupervised Active enhancers and promoters are marked by the pres- hierarchical clustering of the average ATAC-Seq signal ence of histone H3 acetylated at lysine 27 (H3K27ac) . in each peak region showed that, like the transcriptional We generated genome-wide maps of candidate enhancers profiles, the ATAC profiles of iMK and CFU-MK clus - and promoters in ERY and iMK by performing ChIP-Seq tered with LSK, CMP, and GMP, while the ATAC profiles of H3K27ac. Using four replicates for each population, of the erythroid, neutrophil, and monocyte populations we identified 8566 ERY and 12,594 iMK H3K27ac peaks Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 5 of 18 Fig. 2 Transcriptional profiles of hematopoietic populations. a Unsupervised hierarchical clustering analysis of RNA-Seq-derived transcriptomes. b Principle component analysis plotting PC1 versus PC2. c Expressed transcripts in LSK and maintained throughout erythro-megakaryopoiesis. Solid circle size is proportional to the percentage of expressed transcripts maintained from the LSK, dotted circle is proportional to the percentage of all transcripts in the cell relative to the number maintained from LSK, and line length is inversely proportional to percentage of expressed transcripts maintained from the progenitor. The number and percentage of maintained LSK transcripts is shown. d Proportional Venn diagram indicating overlap of expressed genes between CMP, ERY, and iMK populations (Fig. 4a, b). Since active enhancers and promoters are AREs to actively expressed candidate target genes: 82% of almost always in regions of open chromatin, we gener- ERY-specific AREs, 89% of iMK-specific AREs, and 93% ated a set of high-confidence active regulatory elements of shared AREs (Table 3). To infer the stage at which ERY (AREs) by overlapping the H3K27ac peaks with ATAC- or iMK AREs appear (inferred AREs), we intersected the Seq peaks. We found that over 95% of H3K27ac peaks sets of ERY and iMK AREs with the ATAC-Seq peaks of overlapped with ATAC regions. We identified 2098 progenitor cells (Fig. 4c). Between 80 and 90% of these ERY-specific and 6386 iMK-specific AREs, and 5989 inferred AREs were associated with an actively tran- common AREs shared by both cell types (Fig. 4b). We scribed gene (Table 3). Over 98% of shared ERY and iMK assigned AREs to candidate target genes using the clos- AREs were present in LSK. 89% of iMK-specific AREs est transcriptional start site (TSS), regardless of distance were present in LSK, while 97% were present in CMP. To or activation state. This method assigned the majority of calculate the significance of the overlap, we performed Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 6 of 18 Table 2 Number of transcripts with maintained expression at subsequent stages of hematopoiesis Expressed in LSK and CMP Expressed in LSK, CMP, Expressed in LSK, CMP, and committed progenitor committed progenitor, and mature cell ERY # Expressed since LSK 11,765 8760 8239 % Expressed since LSK 90 67 63 % Expressed from immediate progenitor 90 74 94 iMK # Expressed since LSK 11,765 11,161 10,522 % Expressed since LSK 90 85 81 % Expressed from immediate progenitor 90 95 94 permutation tests with the R package “regioneR” . 15 and 79% of LSK-accessible shared AREs were classi- This analysis demonstrated that the observed number of fied as poised and active enhancers, respectively. 21% of overlaps was statistically higher than one would expect by LSK-established ERY-specific and 9% of LSK-established random chance (LSK p value 0.002, CMP p value 0.002) iMK-specific AREs did not overlap with either H3K4me1 (Fig. 4d). In contrast, while 64% of ERY-specific AREs had or H3K27ac peaks, and were therefore classified as open already been established in LSK, this portion is signifi - AREs. As validation, 23 of our ERY AREs overlapped cantly lower compared to 89% of iMK-specific AREs (p with a set of 39 enhancers validated by luciferase assays value 0.002). More ERY-specific AREs (84%) were present , while none were found in the set of open AREs. Per- in CMP, but this was significantly lower than the 97% of mutations tests using 500 iterations of randomized peak iMK-specific AREs (p value 0.002). Approximately 14% of overlaps demonstrated that these intersections were sta- ERY-specific AREs were established de novo in CFU-E (p tistically higher than expected by chance alone (Fig. 5b). value 0.002), whereas only 1% of iMK-specific AREs are To visualize the integration of RNA-Seq, ATAC-Seq, established de novo in CFU-MK (p value 0.002) (Fig. 4d). and ChIP-Seq profiles, we examined several genes sig - We conclude that the regulatory element profiles showed nificant for ERY and iMK maturation (Fig. 6, Additional a greater amount of lineage-specific activation in ERY file 2: Fig. S2). The gene Slc4a1, encoding the anion trans - than in iMK, similar to what was observed for the tran- porter BAND3, is strongly induced during erythroid scriptional profiles. maturation. This induction was accompanied by the ERY- Poised enhancers are defined as regions of accessi - specific induction of transposase-accessible chroma - ble chromatin that contain histone H3 monomethylated tin around the promoter and 3′ end of the gene, as well at lysine 4 (H3K4me1), but they can be distinguished as in the upstream noncoding gene Bloodlinc (Fig. 6a). from active enhancers by the absence of H3K27 acety- Chromatin in these regions was also was modified with lation . To determine whether the inferred AREs in H3K27ac and H3K4me1, indicative of active elements progenitor cells were poised (H3K4me1 only) or active (Fig. 6a). The closely linked gene Slc25a39, encoding a (H3K4me1 and H3K27ac) in LSK, we compared our data constitutively expressed mitochondrial carrier protein, with the indexing-first (iChIP) H3K4me1 and H3K27ac was in accessible chromatin in all the examined cell types regions identified by Lara-Astiaso et al. . We accessed (Fig. 6a). In contrast, the Itgb3 gene (encoding the surface iChIP sequencing reads from Gene Omnibus (GSE60103) marker glycoprotein IIIa or CD61) was expressed and in and performed genome alignment and peak calling using regions of open chromatin in LSK, CMP, and MK-com- our pipeline (see Materials and Methods). This analysis mitted populations, but was repressed and in non-acces- identified 62,849 H3K4me1 and 16,098 H3K27ac peaks in sible chromatin regions in erythroid cells (Fig. 6b). the iChIP LSK set. Approximately 57% of the LSK-estab- lished ERY-specific and 41% iMK-specific AREs over - Differing properties of ERY‑ and iMK‑specific AREs lapped with H3K4me1 peaks but not H3K27ac peaks, We plotted the proximity of AREs established during classifying them as poised enhancers (Fig. 5a, Additional hematopoiesis to the closest TSS. AREs within 1 Kb of file 1: Fig. S1). Approximately 20% of LSK-established the TSS were defined as candidate promoter elements ERY-specific and 46% of LSK-established iMK-specific (cPE), and AREs outside of this region were defined as AREs overlapped with both H3K4me1 and H3K27ac, candidate enhancer elements (cEE). Based on these cri- classifying them as active enhancers. In comparison, teria, approximately 85% (1800) of ERY-specific and 55% Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 7 of 18 Fig. 3 ATAC-Seq profiles of hematopoietic populations. a Unsupervised hierarchical clustering analysis of ATAC-Seq-derived peak profiles. b Principle component analysis plotting PC1 versus. c ATAC-Seq peaks in LSK and maintained throughout erythro-megakaryopoiesis. Circle size is proportional to the percentage of peaks maintained from the LSK, dotted circle is proportional to the percentage of all peaks in the cell relative to the number maintained from LSK, and line length is inversely proportional to percentage of peaks maintained from the progenitor. The number and percentage of maintained LSK peaks is shown. d Proportional Venn diagram indicating overlap of ATAC-Seq peaks between CMP, ERY, and iMK populations (3500) of iMK-specific AREs established during differ - In addition to AREs established at different stages of entiation were categorized as cPE (Fig. 7a), with primi- hematopoiesis, we plotted the proximity of active, poised, tive cells having more cPE than committed cells. We also and inactive AREs to the nearest TSS. As with ERY- observed that cEE established de novo during differen - specific AREs established during differentiation, ERY- tiation tended to form closer to the TSS in both ERY and specific active, poised, and inactive AREs were almost iMK populations. Together these data demonstrate that exclusively categorized as cPE (Fig. 7b). While iMK- ERY-specific AREs established early in hematopoiesis are specific inactive AREs are almost exclusively defined as more likely to be cPE, whereas a substantially larger frac- cPE, approximately 20% of active and poised AREs were tion of iMK-specific AREs are comparatively more likely classified as cEE. We conclude that LSK-established ERY- to be cEE. specific AREs are more likely to be candidate promoters, Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 8 of 18 Fig. 4 Establishment of ERY and iMK AREs throughout hematopoiesis. a Heatmap comparing H3K27ac immunoprecipitation peaks in ERY and iMK samples. Calculations were performed using the R package DiffBind (v2.2.6). b Active regions are defined as the intersection of ATAC and H3K27ac peaks in ERY and iMK. c Establishment of open chromatin was defined as intersecting AREs from ERR or iMK with ATAC-Seq peaks in sequentially more primitive cell populations. d Significance of overlap was calculated by randomizing peak positions and calculating random versus expected overlap (500 iterations). Z-score of overlap between AREs and ATAC peaks. Hashed bars indicate p value > 0.05 Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 9 of 18 Table 3 Number of AREs assigned to the closest TSS ARE establishment stage Category ERY‑specific Common MEG‑specific LSK AREs 1112 5926 5385 No. genes assigned as ARE targets 1118 7963 4462 No. genes expressed 940 84% 7459 93% 3965 89% CMP AREs 1727 5967 6068 No. genes assigned as ARE targets 1606 7982 4767 No. genes expressed 1294 81% 7415 94% 4083 86% Committed progenitor AREs 2054 5975 6286 No. genes assigned as ARE targets 1829 7987 4848 No. genes expressed 1443 79% 7266 90% 4238 87% Terminally differentiated cell AREs 2098 5989 6386 No. genes assigned as ARE targets 1853 7998 4890 No. genes expressed 1527 82% 7464 93% 4341 89% Fig. 5 Establishment of active and poised enhancers in LSK. a LSK-accessible cell-specific and shared AREs were compared against the indexing-first H3K4me1 and H3K27ac chromatin immunoprecipitation profiles (Lara-Astiaso et al, Science, 2014). b Z-score of overlap between ARE and iChIP peaks Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 10 of 18 Fig. 6 Transcriptional and epigenetic features illustrating different modes of regulation. a Induction of expression and AREs at Slc4a1, Bloodlinc, and Slc25a39. RNA-Seq data (left panel) are shown for each pair of replicates only for transcription of the minus strand (and thus RNA from Rundc3a is not shown). ATAC-Seq patterns (central panel) are shown for each pair of replicates. Histone modifications (right panel) are shown as single determinations, although some are available as replicates. b Retention of expression and AREs and from LSKs to MKs with loss in ERY for Itgb3. Displays are arranged as in (a) except RNA-Seq is shown for the plus strand whereas a larger proportion of LSK-established iMK-spe- suggest that megakaryocytes maintain epigenetic and cific AREs qualify as candidate enhancers. transcriptional profiles present in the progenitor cell AREs often contain consensus sequences for tran- populations, while erythroblasts acquire their distinct scription factor binding sites that regulate cellular profiles later in hematopoiesis. development and lineage specificity. We identified instances of consensus transcription factor binding site The megakaryocytic lineage is associated with de novo motifs (TFBS motifs) that were significantly enriched DNA methylation (q-value ≤ 0.05) in the set of AREs (Table 4, Additional We generated genome-wide DNA methylation maps file 3: Table S1). Poised iMK-specific AREs included in LSK, CMP, CFU-E and ERY, and CFU-MK and iMK motifs for ETS-family proteins FLI1 (q-value 0.029), populations using recombinant Methyl Binding Domain ETS1 (q-value 0.004), and PU.1 (q-value 0.004). In 2 (MBD2) protein to enrich methylated DNA fragments, addition to those found in poised iMK-specific AREs, followed by next-generation sequencing (MBD-Seq) . active iMK-specific AREs also had matches to motifs We identified ~ 33,754 LSK and ~ 10,288 CMP meth- for RUNX1 (q-value 0.046) and RUNX2 (q-value 0.010). ylation peaks; 86% of CMP peaks were present in LSK ERY-specific AREs established in CMP were enriched (Table 1). In differentiating erythroid cells, we observed for TFBS motifs of the master regulators GATA1, a further loss of the DNA methylation peaks present GATA2, and NF-E2 (q-value ≤ 0.03) (Table 4). Note in CMP accompanied by an additional ~ 9000 de novo that while many TFBS motifs are enriched in AREs pre- methylation peaks in CFU-E. Mature erythroblasts sent at all stages of hematopoiesis, Table 4 only shows had a global loss of methylation peaks accompanied by those motifs whose enrichment passes a stringent additional de novo methylation (Fig. 8a). Megakary- FDR threshold of q-value ≤ 0.05. In summary, our data opoiesis was associated with an increase of ~ 60,000 de novo methylation peaks in CFU-MK, followed by a loss Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 11 of 18 Fig. 7 Distance of ARE to closest TSS. a Distance of LSK-established ARE to the transcriptional start site. b Distance of active, poised, and inactive ARE to the transcriptional start site of ~ 20,000 of these de novo methylation peaks between methylation in LSK (Table 6). We conclude that, unlike CFU-MK and iMK (Fig. 8b). These patterns are illus - the ATAC and transcriptional profiles, the methylation trated in the transferrin receptor 2 (Tfr2) gene (Fig. 8c). profile of the megakaryocytic lineage is acquired during Because changes in DNA methylation are associated differentiation while erythropoiesis is associated with with differences in expression, we compared the overlap global DNA demethylation. of DNA methylation peaks with cell-specific ARE pro - files. We found that 43 (2.1%) of 2098 ERY-specific AREs Discussion established in ERY overlapped with ERY DNA meth- It is well known that megakaryocytes and stem cells ylation peaks (Table 5). 236 (11.5%) of 2054 ERY-specific share a number of molecular features [10, 11]. Both HSC AREs established in CFU-E overlapped with CFU-E DNA and megakaryocytes have similar dependencies on the methylation peaks. In comparison, 789 (12.4%) of 6368 thrombopoietin receptor MPL and express CD41 and iMK-specific AREs established in iMK overlapped with CD117 (cKit) [10, 11, 30]. HSC and megakaryocytes iMK DNA methylation peaks, while 1121 (17.8%) of 6286 also have similar dependencies on transcription fac- iMK-specific AREs established in CFU-MK overlapped tors, including expression of RUNX1, GATA2, and EVI1 with CFU-MK DNA methylation (Table 5). . Our data suggest that there is a large population We compared the overlap of DNA methylation peaks of megakaryocyte-primed cells within the multipotent with active, poised, and inactive regions in LSK. We progenitor population (MPP) that shares the chromatin, found that nearly 20% of active and poised ERY-specific enhancer/promoter, and transcriptional profiles of the regulatory elements overlapped with regions of DNA MPP. It is also possible that only a subpopulation of HSC methylation in LSK, whereas ~ 8% of active and poised are megakaryocyte-primed, but our data suggest that the iMK-specific regulatory elements overlapped with DNA majority of MPP share transcriptional and regulatory Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 12 of 18 Table 4 Partial list of TFBS motifs enriched in AREs at different stages of hematopoiesis Transcription ERY‑specific AREs MEG‑specific AREs factor LSK (active) LSK (poised) CMP CFU‑E LSK (active) LSK (poised) CMP CFU‑MK AP-1 0.005 < 0.0001 < 0.0001 BACH1 0.001 BATF 0.005 0.0001 0.002 CEBP 0.001 0.001 CHOP 0.0001 0.001 NF-E2 0.02 0.0002 EHF 0.0001 0.005 < 0.0001 < 0.0001 ELF1 0.0001 < 0.0001 < 0.0001 ERG 0.0009 < 0.0001 < 0.0001 ETS 0.0001 0.0003 < 0.0001 < 0.0001 ETS1 0.0008 < 0.0001 < 0.0001 PU.1 0.0148 0.0039 < 0.0001 < 0.0001 SpiB 0.0004 0.0017 < 0.0001 < 0.0001 FLI1 < 0.0001 < 0.0001 IRF1 0.0138 0.0017 < 0.0001 < 0.0001 IRF2 0.0052 0.0001 0.0005 IRF8 0.0459 RUNX1 0.0004 0.002 RUNX2 0.002 0.01 GATA1 0.03 0.01 GATA2 0.03 0.01 KLF3 0.0281 SP1 0.04 Benjamini-adjusted q-value for the enrichment score is reported characteristics with those of the megakaryocyte popula- profiles are first detected at the level of the MEP, where tion [4, 15, 31–34]. the MEP ATAC-Seq profile clusters with that of mature The clearest distinction between MPP and mega - erythroid cells. These data are consistent with the greater karyocytes is in their DNA methylation profiles, where frequency of erythroid committed cells in the human de novo methylation increases in CFU-MK but subse- MEP population . Based on the transcriptional pro- quently decreases in iMK . Similar patterns have been files, we propose that many cells in the mouse MEP observed in muscle and neuronal lineages (reviewed in population are in transition from multipotent progenitor [35–37]). Our data suggest that de novo methylation is cells (as evidenced by the similar transcriptional profiles) an important step in megakaryopoiesis , while the to erythroid committed cells. Our observations support chromatin accessibility, enhancer/promoter profiles, this model and previous publications: ERY-specific AREs and transcriptional programs remain highly consist- in CMP, but not LSK, contain TFBS associated with ent. Changes in DNA methylation have been linked to GATA-switching. However, in-depth single-cell analyses induction of several megakaryocytic processes, including will be needed to test this hypothesis [43, 44]. endomitosis, transcription factor expression and DNA The classic model of hematopoiesis is displayed as a affinity, and enhancer activity [39–42]. The requirement hierarchy in which multipotent progenitors traverse a for de novo DNA methylation in megakaryopoiesis sug- series of oligopotent and bipotent intermediates that gests that despite similar transcriptional and regulatory progressively restrict lineage potential [1, 45, 46]. Recent profiles with MPP, megakaryopoiesis does not constitute studies, including single-cell transcriptional analyses and a “default” developmental program. clonogenic assays, have identified a subset of megakary - In contrast to megakaryopoiesis, erythropoiesis is ocyte-primed progenitor cells in the MPP compartment associated with significant changes in the chromatin, that is proposed to give rise directly to megakaryocytes enhancer/promoter, transcriptional, and methylation [4, 12, 15, 47, 48]. Our data are consistent with this model. profiles. Our data suggest that, for most elements, these Further down the hierarchy, multipotent progenitors give Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 13 of 18 Fig. 8 Overlap of DNA methylation and chromatin-accessible AREs. a Establishment of DNA methylation in the erythroid lineage. b Establishment of DNA methylation in the megakaryocytic lineage. c The Tfr2 locus showing example regions of maintained, lost, and de novo DNA methylation during erythropoiesis and megakaryopoiesis Table 5 Comparison of DNA methylation and AREs established during hematopoiesis ARE DNA methylation Mature cell Committed progenitor CMP LSK ERY-specific 43/2098 (2.1%) [ERY ] 236/2054 (11.5%) [CFU-E] 64/1727 (3.7%) 207/1112 (18.6%) iMK-specific 789/6386 (12.4%) [iMK] 1121/6286 (17.8%) [CFU-MK] 138/6068 (2.3%) 422/5385 (7.8%) lineages [9, 49]. The existence of many of the cell types Table 6 Comparison of DNA methylation and cell-specific regulatory elements in LSK in this hierarchy is supported by the ability to grow colo- nies comprised of single or multiple lineages from single ARE DNA methylation in LSK cells in semisolid medium in vitro. However, there are no Active Poised Inactive reliable assays to culture both erythroid and megakaryo- cytic cells under the same conditions. Our data support ERY-specific 28/144 (19.4%) 122/605 (20.2%) 56/335 (16.7%) models in which many cells within the classically defined iMK-specific 126/1651 (7.6%) 189/2489 (7.6%) 86/978 (8.8%) MEP compartment have already become committed to one or the other lineage. The data and analysis in this paper were generated rise to common myeloid progenitor (CMP) cells, then to as part of ValIdated Systematic IntegratiON of hemat- the megakaryocyte-erythroid lineage at the MEP stage, opoietic epigenomes (VISION) project, which aims to and finally to either the megakaryocyte or erythroid Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 14 of 18 generate comprehensive catalogs of genomic regulatory Table 7 List of antibodies elements in mouse and human hematopoietic cells and Antibody Clone Catalog number to conduct integrative statistical modeling and machine Rat α-CD4 clone GK1.5 14-0041-86 learning to predict regulatory interactions that are then Rat α-IL7Ra clone A7R34 14-1271-85 validated by manipulating hematopoiesis at the genetic Rat α-CD8 clone 53-6.7 14-0081-86 level. The data come both from efforts in the project lab - Rat α-Mac-1 clone M1/70 14-0112-86 oratories, such as those generated for this paper, and also Rat α-Gr-1 clone RB6-8C5 14-5931-86 from other laboratories and consortia such as the Ency- Rat α-B220 clone RA3-6B2 14-0452-86 clopedia of DNA Elements (ENCODE) project and the Rat α-Ter119 clone TER-119 14-5921-85 NIH Roadmap Epigenomics Project [21–24]. These data Rat α-Sca-1 (PE) clone D7 12-5981-83 are available at our website, usevision.org, for use by the Rat α-CD34 (FITC) clone RAM34 11-0341-85 larger community. Rat α-CD16/32 (PeCy7) clone 93 25-0161-82 Rat IgG2b clone eB149/10H5 14-4031-85 Conclusion Rat IgG1 eBRG1 14-4301-85 Our studies confirm that the majority of the epigenetic Rat IgG2a serum eBR2a 14-4321-85 and transcriptional profiles found in the hematopoietic Rat α-cKit (APC) clone 2B8 17-1171-83 stem cell population are present in megakaryocytes, Rat α-CD41 (PE) clone eBioMWReg30 12-0411-83 while the erythropoietic epigenetic and transcriptional Rat α-CD61 (FITC) clone 2C9.G3 11-0611-82 programs are not established until erythroid-lineage Rat α-Sca-1 (PerCP-Cy5.5) clone D7 45-5981-82 commitment. By performing “bulk” analyses, we achieve Rat α-Ter119 (APC-780) clone TER-119 47-5921-82 greater depth of coverage that emphasizes the dominant Rat α-CD44 (eFluor450) clone IM7 48-0441-82 characteristics of cell populations. In contrast, single-cell Rabbit a-H3K27ac polyclonal ab4729 analyses can resolve finer distinctions between popula - Rabbit a-H3K4me1 polyclonal ab8895 tions, but at a lower depth of coverage. Rat α-Mac-1 (eFluor450) clone M1/70 48-0112-82 Rat α-Gr-1 clone RB6-8C6 45-5931-80 Methods All antibodies were purchased from eBiosciences (San Diego, CA, USA) Cell isolation All primary hematopoietic cell populations were enriched from 5-to-8-week-old C57BL6 male mice. LSK, CMP, using the Ribo-Zero removal reagents and fragmented. MEP, GMP, CFU-E, ERY, CFU-MK, and iMK populations First-strand cDNA was then synthesized using a 5’ tagged were collected from bone marrow. Following collection, random hexamer and reverse transcription, followed by samples were lineage-depleted using antibodies (Fig. 1) annealing of a 5’ tagged, 3’-end blocked terminal-tagged and conjugated to BioMag Goat anti-Rat IgG magnetic oligo and second-strand synthesis. The Di-tagged cDNA beads (cat # 310007, Qiagen, Venlo, Netherlands). Lin- fragments were purified, barcoded, and PCR-amplified eage-depleted cells were then stained with fluorescently for 15 cycles. conjugated antibodies and sorted by flow cytometry on The size and quality of each library were then evalu - an FACSAria II (BD Biosciences). Neutrophils (NEU) ated by Bioanalyzer 2100 (Agilent Technologies, Santa and monocytes (MONO) were isolated directly from Clara, CA), and quantified using qPCR. Libraries were peripheral blood (PB). After Ficoll separation NEU were sequenced in paired-end mode on the NextSeq500 to collected from the cell pellet following red cell lysis and generate 2 × 75 bp reads using Illumina-supplied kits as MONO (Gr-1+Mac-1+ were sorted from the mononu- appropriate. The sequence reads were processed using clear fraction. All antibodies are listed in Table 7. the ENCODE3 long RNA-Seq pipeline (https ://www. encod eproj ect.org/pipel ines/ENCPL 002LP E/). In brief, RNA‑Seq reads were mapped to the mouse genome (mm10 assem- Cells were sorted into media and total RNA was bly) using STAR v2.5.4 , followed by RSEM v1.3.0 extracted using TRIzol and the Ambion PureLink RNA  for gene quantifications. RNA-Seq was repeated to Mini Kit (Life Technologies, cat# 12183018), treated generate at least two biological replicates. Analyzed data with DNase to remove genomic DNA using the DNA- are included as Additional file 4. free Kit (Life Technologies, cat# AM1906). Sequencing libraries were then constructed from 100 ng of treated, ATAC‑Seq total RNA using the ScriptSeq Complete Kit (Illumina Approximately 50,000 cells were collected by centrifu- cat# BHMR1224) according to manufacturer’s specifica - gation at 600 × g for 10 min at 4 °C. Cells were washed tions. In brief, the RNA was subjected to rRNA depletion Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 15 of 18 Cells were washed in 2X PIC (Roche mini-tabs, 1 tab in once with cold 1× PBS and centrifuged as above. Cells 5 ml = 2X) and stored at − 80°C. Cells were resuspended were lysed by gently pipetting to resuspend cell pellet in in lysis buffer (10 mM Tris–HCl, pH8.0, 10 mM NaCl, cold lysis buffer (10 mM Tris–HCl pH 7.4, 10 mM NaCl, 0.2% NP40) and sonicated (QSonica) to a size range of 3 mM MgCl2, 0.1% IGEPAL CA-630) and centrifuged as 200–500 bp as determined by electrophoresis. Imme- above. For each transposition reaction, cells were sus- diately prior to immunoprecipitation, an aliquot was pended in the following mix: 25 µl 2X Tagment DNA removed to be sequenced as a matched input control. buffer (Illumina cat# FC-121-1030), 3 µl Tn5 Transposase Immunoprecipitation was performed using H3K27ac (Illumina cat#FC-121-1030), 22 µl nuclease-free H2O (Abcam ab4729) supplemented with 50 mM sodium and incubated for 30 min at 37 °C. Reactions were ter- butyrate. DNA fragments were purified with AMPure minated by adding 5 µl of 1% SDS solution and purified XP beads (cat # A63880, Agencourt, Pasadena, CA, USA) using SPRIselect beads (Beckman Coulter cat #B23318) according to the manufacturer’s protocol. Sequencing at a 1:1 ratio according to manufacturer’s instructions, was performed at the NIH Intramural Sequencing Center followed by a right side size selection using SPRIse- on an Illumina HiSeq2500. Chromatin immunoprecipita- lect beads (Beckman Coulter cat #B23318) at a 0:5 ratio tions were repeated to generate four biological replicates. according to manufacturer’s instructions. Following puri- Sequence data were aligned to the mm10 genome using fication, library fragments were amplified using 1 × NEB- Bowtie v2.3.4  and peak calling performed using next PCR master mix and 1.25 μM of custom Nextera SICER v1.1 . Peaks present in all replicates were used PCR primers 1 and 2  using the following PCR condi- for further analysis, as above. tions: 72 °C for 5 min; 98 °C for 30 s; and thermocycling We accessed indexing-first immunoprecipitation at 98 °C for 10 s, 63 °C for 30 s, and 72 °C for 1 min. To (iChIP) experimental data performed by Lara-Astiaso reduce GC and size bias, the PCR was monitored using et al.  from Gene Omnibus (GSE60103). Sequenc- qPCR in order to stop amplification before saturation. To ing reads from long- and short-term HSC (LT-HSC and this end, libraries were amplified for five cycles, and then ST-HSC, respectively) and multipotent progenitor cells a 5 µl aliquot of the PCR was removed and added 10 µl of were combined to generate “LSK” profiles from the iChIP the PCR cocktail with SYBR Green at a final concentra - data. These reads were aligned using Bowtie v2.3.4 and tion of 0.6×. This reaction was amplified for 20 cycles to peak calling performed using MACS2 . We com- determine the additional number of cycles needed for the bined LT-HSC, ST-HSC, and MPP peak sets to create the remaining 45-μL reaction. Libraries were amplified for iChIP “LSK” population for downstream comparisons. a total of 17–19 cycles and then purified using AMPure For studies involving indexing-first chromatin immu - XP beads (Beckman Coulter cat #A63881) at a 1:1 ratio noprecipitation data (iChIP), reads were accessed via according to manufacturer’s instructions. Constructed GSE60103 (Lara-Astiaso et al. ), data aligned using libraries were run on the Agilent Bioanalyzer 2100 (Agi- Bowtie v2.3.4 and peaks-calling performed using MACS2 lent Technologies) using the 7500 DNA kit (cat# 5067- . To create a set of LSK data, peaks were combined 1504) as appropriate to determine the average size and from the long-term HSCs, short-term HSCs, and multi- confirm the absence of unligated adaptors. Libraries were potential progenitor sample sets. Analyzed data are quantitated by qPCR using the Kapa SYBR FAST Uni- included as Additional file 6. versal kit (Kapa Biosystems) according to the Illumina’s Sequencing Library qPCR Quantification Guide. Librar - MBD‑Seq ies were multiplexed and sequenced on the Illumina Genomic DNA was isolated from enriched cells with the NextSeq500 using Illumina’s kits and reagents as appro- QIAGEN Puregene kit and sonicated to 200- to 400-bp priate. ATAC-Seq was repeated to generate at least two fragments (QSonica). MBD2 enrichment was performed biological replicates. Reads were mapped using Bowtie with the Active Motif Methyl Collector kit. Approxi- v1.0.0 , accessible regions identified with FSeq v1.85 mately 1 μg of sonicated genomic DNA was incubated , and peaks called using HOMER v1.0 . To iden- with MBD2-His-conjugated protein and magnetic beads tify consensus peak sets, peaks from all datasets were according to the manufacturer’s protocol. After enrich- combined and merged. Peaks present in both replicates ment, both the methylated fraction and supernatant within the same merged region were used for further fractions were purified with QIAGEN DNA purification analysis. Analyzed data are included as Additional file 5. columns. Quantitative PCR amplification of the differ - entially methylated regions regulating the imprinting ChIP‑Seq of Snrpn and Rasgrf1 and the unmethylated CpG island Approximately 20 × 10 cells were fixed in 0.4% for - promoter of Actb was performed with SYBR Green maldehyde (16% methanol-free, Thermo Scientific) for PCR master mix (Applied Biosystems) and was used to 15 min before quenching in 125 mM glycine for 5 min. Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 16 of 18 Abbreviations validate the enrichment of methylated DNA using the − + + HSC: hematopoietic stem cell; LSK: Lin Sca Kit ; MPP: multipotent progenitor; MBD2-pull-down approach. CMP: common myeloid progenitor; CLP: common lymphoid progenitor; GMP: Two biological replicates of each enriched cell popu- granulocyte/macrophage progenitor; MEP: megakaryocyte/erythroid progeni- tor; CFU-MK: colony-forming unit (megakaryocyte); CFU-E: colony forming lation and one supernatant sample per cell type were unit (erythrocyte); ERY: erythrocyte; MK: megakaryocyte; VISION: ValIdated submitted for high-throughput sequencing analysis. Systematic IntegratiON of hematopoietic epigenomes; ENCODE: Encyclopedia Between 225 and 540 ng of MBD2-enriched DNA and of DNA Elements; iMK: immature megakaryocyte; ATAC-Seq: assay for transpo- son-accessible chromatin followed by next-generation sequencing; RNA-Seq: 1 μg of supernatant for each cell type were used to con- next-generation sequencing of RNA; ChIP-Seq: chromatin immunoprecipita- struct DNA libraries according to the Illumina protocol. tion followed by next-generation sequencing; MBD-Seq: methyl-binding The libraries were sequenced on the Illumina Genome protein immunoprecipitation followed by next-generation sequencing; TPM: transcripts per million count; PCA: principle component analysis; ARE: active Analyzer platform, and 36-bp single-end reads were regulatory element; iChIP: indexing-first chromatin immunoprecipitation; TSS: used to uniquely identify the MBD2-bound fraction of transcriptional start site; cEE: candidate enhancer element; cPE: candidate the mouse genome. Sequenced reads were mapped to promoter element; TFBS: transcription factor binding site. the mouse genome (UCSC assembly mm10) using Bow- Authors’ contributions tie v2.3.4 . Peaks for each cell type were called using EFH generated H3K27ac ChIP-Seq data and analyzed, integrated, and inter- the SigSeeker peak calling ensemble . Reads from the preted next-generation sequencing experiments. CAK generated RNA-Seq and ATAC-Seq data. BG performed analyses and generated images regarding matched supernatant were used as a control for each cell RNA-Seq and ATAC-Seq data. SMA performed flow cytometry analyses. JL type. Replicate peaks that were called by two or more generated MBD-Seq data. NISC performed sequencing for H3K27ac ChIP-Seq peak calling tools and overlapped by at least 100 bp experiments. EFH, CAK, BG, JL, RCH, and DMB analyzed data and wrote the paper. All authors read and approved the final manuscript. were considered for further analysis. Analyzed data are included as Additional file 6. Author details 1 2 NHGRI Hematopoiesis Section, GMBB, Bethesda, MD, USA. Pennsylvania State University, University Park, PA, USA. NHGRI Flow Cytometry Core, Bethesda, MD, USA. NHGRI, Rockville, MD, USA. Additional data analysis and visualization Permutation tests were performed using the R package Acknowledgements The authors would like to acknowledge Martha Kirby and Beth Psaila for their regioneR v1.10.0. Transcription factor enrichment analy- advice and assistance in preparation of this manuscript. ses were performed in HOMER v3.18 . Graphs were created in PRISM v6. Competing interests The authors declare that they have no competing interests. Additional files Availability of data and materials The datasets supporting the conclusions of this article are available in the Additional file 1: Fig. S1. Establishment of ERY and iMK enhancer/ VISION repository (http://usevi sion.org). iChIP data are available in the Gene promoter regions throughout hematopoiesis. (A–C) LSK-accessible Expression Omnibus repository, at https ://www.ncbi.nlm.nih.gov/geo (acces- cell-specific and shared AREs were compared against the indexing-first sion number GSE60103, see citation below). Additionally, datasets supporting H3K4me1 and H3K27ac chromatin immunoprecipitation profiles in (A) the conclusions of this article are included within the article and its additional LT_HSC, (B) ST_HSC, and (C) MPP (Lara-Astiaso et al., Science, 2014). (D) files. Lara-Astiaso D, Weiner A, Lorenzo-Vivas E, et al. Chromatin state dynamics Z-score of overlap between ARE and iChIP peaks. Hashed bars indicate during blood formation. 2014. doi: https ://doi.org/10.1126/scien ce.12562 71. p-value >0.05. Consent for publication Additional file 2: Fig. S2. Locus-specific example of (epi) genomic cor - Not applicable. relations. (A) Erythroid specific induction of AREs and expression (Slc4a1) and constitutive AREs and expression (Slc25a39). RNA-Seq and ATAC-Seq Ethics approval and consent to participate are shown for all 10 cell types, and histone modifications for 5 cell types Animal work followed the guidelines and policies and was approved by the of most relevance. (B) Retention of AREs and expression from LSKs in MKs Animal Care and Use Committee. The National Human Genome Research and loss in ERY. RNA-Seq and ATAC-Seq are shown for all 10 cell types, Institute is accredited by AAALAC International and is in accordance with the and histone modifications for 5 cell types of most relevance. Tracks are Public Health Service Policy for the Care and Use of Laboratory Animals. displayed on the mm10 genome. Additional file 3. Enrichment scores of transcription factor binding sites Funding in AREs. Unabridged list of enriched transcription factor binding sites in This work was supported by the National Human Genome Research Institute AREs. Scores are reported as q-values. intramural funds and the National Institute of Diabetes and Digestive and Kidney Diseases (Grant Number R24DK106766-01A1). Additional file 4. RNA-Seq data with TPM calculations. Transcripts per mil- lion counts are included for each RNA-Seq replicate. Additional file 5. Peak data from ATAC-Seq experiments. The presence Publisher’s Note of ATAC-Seq peaks in each cell type is indicated by a 1 (present) or 0 (not Springer Nature remains neutral with regard to jurisdictional claims in pub- present). lished maps and institutional affiliations. Additional file 6. Peak data from ChIP-Seq and MBD-Seq experiments. Received: 14 April 2018 Accepted: 21 May 2018 Peaks from H3K27ac ChIP-Seq and MBD-Seq are presented in BED format. Note that due to file restrictions, MBD-Seq data for CFU-MK is included across two workbooks. Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 17 of 18 References 26. Gel B, Díez-Villanueva A, Serra E, Buschbeck M, Peinado MA, Malinverni 1. Orkin SH, Zon LI. Hematopoiesis: an evolving paradigm for stem cell biol- R. regioneR: an R/Bioconductor package for the association analy- ogy. Cell. 2008;132:631–44. sis of genomic regions based on permutation tests. Bioinformatics. 2. Seita J, Weissman IL. Hematopoietic stem cell: self-renewal versus dif- 2016;32:289–91. ferentiation. Wiley Interdiscip Rev Syst Biol Med. 2010;2:640–53. 27. Lara-Astiaso D, Weiner A, Lorenzo-Vivas E, Zaretsky I, Jaitin DA, David E, 3. Krause DS. Regulation of hematopoietic stem cell fate. Oncogene. Keren-Shaul H, Mildner A, Winter D, Jung S, et al. Chromatin state dynam- 2002;21:3262. ics during blood formation. Science. 2014;345:943–9. 4. Velten L, Haas SF, Raffel S, Blaszkiewicz S, Islam S, Hennig BP, Hirche C, Lutz 28. Dogan N, Wu W, Morrissey CS, Chen K-B, Stonestrom A, Long M, Keller CA, C, Buss EC, Nowak D, et al. Human haematopoietic stem cell lineage com- Cheng Y, Jain D, Visel A, et al. Occupancy by key transcription factors is a mitment is a continuous process. Nat Cell Biol. 2017;19:271. more accurate predictor of enhancer activity than histone modifications 5. Bresnick EH, Lee HY, Fujiwara T, Johnson KD, Keles S. GATA switches as or chromatin accessibility. Epigenet Chromatin. 2015;8:16. developmental drivers. J Biol Chem. 2010;285:31087–93. 29. Lichtenberg J, Bodine DM. Evaluation of peak calling approaches for 6. Doré LC, Crispino JD. Transcription factor networks in erythroid cell and genome-wide MBD2-based methyl-seq profiling. In: 2017 IEEE 7th megakaryocyte development. Blood. 2011;118:231–9. international conference on computational advances in bio and medical 7. Ghamari A, Pregerning G, Fraenkel E, Cantor AB. GATA factor switching sciences (ICCABS), 19–21 Oct 2017; 2017. pp. 1–6. during erythroid differentiation is facilitated by FBW7 mediated clearance 30. Nishikii H, Kanazawa Y, Umemoto T, Goltsev Y, Matsuzaki Y, Matsushita K. of GATA2. Blood. 2016;128:1479. Unipotent megakaryopoietic pathway bridging hematopoietic stem cells 8. Pimkin M, Kossenkov AV, Mishra T, Morrissey CS, Wu W, Keller CA, Blobel and mature megakaryocytes. Stem Cells. 2015;33:2196–207. GA, Lee D, Beer MA, Hardison RC, Weiss MJ. Divergent functions of 31. Nishikii H, Kurita N, Chiba S. The road map for megakaryopoietic line- hematopoietic transcription factors in lineage priming and differentiation age from hematopoietic stem/progenitor cells. Stem Cells Transl Med. during erythro-megakaryopoiesis. Genome Res. 2014;24:1932–44. 2017;6:1661–5. 9. Debili N, Coulombel L, Croisille L, Katz A, Guichard J, Breton-Gorius J. 32. Macaulay IC, Svensson V, Labalette C, Ferreira L, Hamey F, Voet T. Single- Characterization of a bipotent erythro-megakaryocytic progenitor in cell RNA-sequencing reveals a continuous spectrum of differentiation in human bone marrow. Blood. 1996;88:1284–96. hematopoietic cells. Cell Rep. 2016;14:966–77. 10. Woolthuis CM, Park CY. Hematopoietic stem/progenitor cell commitment 33. Waddington CH. The strategy of the genes a discussion of some aspects to the megakaryocyte lineage. Blood. 2016;127:1242–8. of theoretical biology. London: Allen & Unwin; 1957. 11. Huang H, Cantor AB. Common features of megakaryocytes and 34. Quesenberry PJ, Goldberg L, Aliotta J, Dooner M. Marrow hematopoietic hematopoietic stem cells: What’s the connection? J Cell Biochem. stem cells revisited: they exist in a continuum and are not defined by 2009;107:857–64. standard purification approaches; then there are the microvesicles. Front 12. Sanjuan-Pla A, Macaulay IC, Jensen CT, Woll PS, Luis TC, Mead A, Moore Oncol. 2014;4:56. S, Carella C, Matsuoka S, Jones TB, et al. Platelet-biased stem cells 35. Podobinska M, Szablowska-Gadomska I, Augustyniak J, Sandvig I, Sandvig reside at the apex of the haematopoietic stem-cell hierarchy. Nature. A, Buzanska L. Epigenetic modulation of stem cells in neurodevelopment: 2013;502:232–6. the role of methylation and acetylation. Front Cell Neurosci. 2017;11:23. 13. Pinho S, Marchand T, Yang E, Wei Q, Nerlov C, Frenette PS. Lineage-biased 36. Cheng Y, Xie N, Jin P, Wang T. DNA methylation and hydroxymethylation hematopoietic stem cells are regulated by distinct niches. Dev Cell. in stem cells. Cell Biochem Funct. 2015;33:161–73. 2018;44:634–41. 37. Tsumagari K, Baribault C, Terragni J, Varley KE, Gertz J, Pradhan S, Badoo 14. Psaila B, Barkas N, Iskander D, Roy A, Anderson S, Ashley N, Caputo M, Crain CM, Song L, Crawford GE, et al. Early de novo DNA methyla- VS, Lichtenberg J, Loaiza S, Bodine DM. Single-cell profiling of human tion and prolonged demethylation in the muscle lineage. Epigenetics. megakaryocyte-erythroid progenitors identifies distinct megakaryocyte 2013;8:317–32. and erythroid differentiation pathways. Genome Biol. 2016;17:83. 38. Lichtenberg J, Heuston EF, Anderson SM, Keller CA, Hardison RC, Bodine 15. Notta F, Zandi S, Takayama N, Dobson S, Gan OI, Wilson G, Kaufmann KB, DM, NIH Intramural Sequencing Center. De novo DNA methylation is McLeod J, Laurenti E, Dunant CF, et al. Distinct routes of lineage develop- associated with granulopoiesis and megakaryopoiesis but not erythro- ment reshape the human blood hierarchy across ontogeny. Science. poiesis. Blood. 2016;128:3868. 2016;351:aab2116. 39. Farlik M, Halbritter F, Müller F, Choudry Fizzah A, Ebert P, Klughammer J, 16. Wegman-Ostrosky T, Savage SA. The genomics of inherited bone marrow Farrow S, Santoro A, Ciaurro V, Mathur A, et al. DNA methylation dynam- failure: from mechanism to the clinic. Br J Haematol. 2017;177:526–42. ics of human hematopoietic stem cell differentiation. Cell Stem Cell. 17. Songdej N, Rao AK. Hematopoietic transcription factor mutations: impor- 2016;19:808–22. tant players in inherited platelet defects. Blood. 2017;129:2873. 40. Ji H, Ehrlich LIR, Seita J, Murakami P, Doi A, Lindau P, Lee H, Aryee MJ, 18. Duployez N, Lejeune S, Renneville A, Preudhomme C. Myelodysplastic Irizarry RA, Kim K, et al. A comprehensive methylome map of lineage syndromes and acute leukemia with genetic predispositions: a new chal- commitment from hematopoietic progenitors. Nature. 2010;467:338–42. lenge for hematologists. Expert Rev Hematol. 2016;9:1189–202. 41. Izzi B, Pistoni M, Cludts K, Akkor P, Lambrechts D, Verfaillie C, Verhamme 19. Dailey HA, Meissner PN. Erythroid heme biosynthesis and its disorders. P, Freson K, Hoylaerts MF. Allele-specific DNA methylation capacitates Cold Spring Harb Perspect Med. 2013;3:a011676. PEAR1 enhancer activity. Blood. 2016;128:1003–12. 20. Donadieu J, Beaupain B, Fenneteau O, Bellanné-Chantelot C. Congenital 42. Lichtenberg J, Heuston EF, Keller CA, Hardison RC, Bodine DM. Compari- neutropenia in the era of genomics: classification, diagnosis, and natural son of expression and epigenetic profiles in human and mouse eryth- history. Br J Haematol. 2017;179:557–74. ropoiesis and megakaryopoiesis using a systems biology model. Blood. 21. Roadmap Epigenomics C, Kundaje A, Meuleman W, Ernst J, Bilenky M, Yen 2015;126:2383. A, Heravi-Moussavi A, Kheradpour P, Zhang Z, Wang J, et al. Integrative 43. Litzenburger UM, Buenrostro JD, Wu B, Shen Y, Sheffield NC, Kathiria A, analysis of 111 reference human epigenomes. Nature. 2015;518:317–30. Greenleaf WJ, Chang HY. Single-cell epigenomic variability reveals func- 22. The ENCODE Project Consortium. An integrated encyclopedia of DNA tional cancer heterogeneity. Genome Biol. 2017;18:15. elements in the human genome. Nature. 2012;489:57–74. 44. Hogart A, Lichtenberg J, Ajay SS, Anderson S, Margulies EH, Bodine DM. 23. The ENCODE Project Consortium. The ENCODE (ENCyclopedia Of DNA Genome-wide DNA methylation profiles in hematopoietic stem and pro - Elements) Project. Science. 2004;306:636. genitor cells reveal overrepresentation of ETS transcription factor binding 24. Hong EL, Sloan CA, Chan ET, Davidson JM, Malladi VS, Strattan JS, Hitz BC, sites. Genome Res. 2012;22:1407–18. Gabdank I, Narayanan AK, Ho M, et al. Principles of metadata organization 45. Till JE, McCulloch EA. Hemopoietic stem cell differentiation. Biochim at the ENCODE data coordination center. Database J Biol Databases Cura- Biophys Acta (BBA) Rev Cancer. 1980;605:431–59. tion. 2016;2016:baw001. 46. Spangrude GJ, Heimfeld S, Weissman IL. Purification and characterization 25. Creyghton MP, Cheng AW, Welstead GG, Kooistra T, Carey BW, Steine EJ, of mouse hematopoietic stem cells. Science. 1988;241:58. Hanna J, Lodato MA, Frampton GM, Sharp PA, et al. Histone H3K27ac 47. Haas S, Hansson J, Klimmeck D, Loeffler D, Velten L, Uckelmann H. Inflam- separates active from poised enhancers and predicts developmental mation-induced emergency megakaryopoiesis driven by hematopoietic state. Proc Natl Acad Sci USA. 2010;107:21931–6. stem cell-like megakaryocyte progenitors. Cell Stem Cell. 2015;17:422–34. Heuston et al. Epigenetics & Chromatin (2018) 11:22 Page 18 of 18 48. Yamamoto R, Morita Y, Ooehara J, Hamanaka S, Onodera M, Rudolph KL. 54. Boyle AP, Guinney J, Crawford GE, Furey TS. F-Seq: a feature den- Clonal analysis unveils self-renewing lineage-restricted progenitors gen- sity estimator for high-throughput sequence tags. Bioinformatics. erated directly from hematopoietic stem cells. Cell. 2013;154:1112–26. 2008;24:2537–8. 49. Weissman IL, Anderson DJ, Gage F. Stem and progenitor cells: origins, 55. Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, Cheng JX, Murre C, phenotypes, lineage commitments, and transdifferentiations. Annu Rev Singh H, Glass CK. Simple combinations of lineage-determining transcrip- Cell Dev Biol. 2001;17:387–403. tion factors prime cis-regulatory elements required for macrophage and 50. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, B cell identities. Mol Cell. 2010;38:576–89. Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioin- 56. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat formatics. 2012;29:15–21. Methods. 2012;9:357. 51. Li B, Dewey C. RSEM: accurate transcript quantification from RNA-Seq 57. Zang C, Schones DE, Zeng C, Cui K, Zhao K, Peng W. A clustering data with or without a reference genome. BMC Bioinform. 2011;12:323. approach for identification of enriched domains from histone modifica- 52. Buenrostro JD, Giresi PG, Zaba LC, Chang HY, Greenleaf WJ. Transposition tion ChIP-Seq data. Bioinformatics. 2009;25:1952–8. of native chromatin for fast and sensitive epigenomic profiling of open 58. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, chromatin, DNA-binding proteins and nucleosome position. Nat Meth- Nusbaum C, Myers RM, Brown M, Li W, Liu XS. Model-based analysis of ods. 2013;10:1213–8. ChIP-Seq (MACS). Genome Biol. 2008;9:R137. 53. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory- 59. Lichtenberg J, Elnitski L, Bodine DM. SigSeeker: a peak-calling ensem- efficient alignment of short DNA sequences to the human genome. ble approach for constructing epigenetic signatures. Bioinformatics. Genome Biol. 2009;10:R25. 2017;33:2615–21. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your ﬁeld rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions
Epigenetics & Chromatin – Springer Journals
Published: May 28, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera