Enhancer repertoires are reshaped independently of early priming and heterochromatin dynamics during B cell differentiation

Abstract

A widely accepted model posits that activation of enhancers during differentiation goes through a priming step prior to lineage commitment. To investigate the chronology of enhancer repertoire establishment during hematopoiesis, we monitored epigenome dynamics during three developmental stages representing hematopoietic stem cells, B-cell progenitors and mature B-cells. We find that only a minority of enhancers primed in stem cells or progenitors become active at later stages. Furthermore, most enhancers active in differentiated cells were not primed in earlier stages. Thus, the enhancer repertoire is reshaped dynamically during B-cell differentiation and enhancer priming in early stages does not appear to be an obligate step for enhancer activation. Furthermore, our data reveal that heterochromatin and Polycomb-mediated silencing have only a minor contribution in shaping enhancer repertoires during cell differentiation. Together, our data revisit the prevalent model about epigenetic reprogramming during hematopoiesis and give insights into the formation of gene regulatory networks.

Figure 1. Genome-wide characterization of enhancer elements in HSCs and different B cell stages.

(a) H3K4me1 versus H3K27ac enrichments (over input) for 1 kb sliding windows across the genome. Regions overlapping with annotated promoters or enriched in H3K4me3 were excluded. Green lines indicate the cut-offs used to select enriched windows for both signals (Materials and methods). Windows additionally enriched for the H3K27me3 mark are marked by red stars (*). (b) Putative enhancers (corresponding to merged windows enriched in H3K4me1) were classified according to their enrichment in H3K27ac and H3K27me3 signal. H3K4me1+/H3K27ac+ loci correspond to active enhancers; H3K4me1+/H3K27ac- loci correspond to primed enhancers and H3K4me1+/H3K27me3+ loci to poised enhancers.(c) Box plots showing expression levels of genes according to their association with distinct classes of enhancers. (d) Functional validation of enhancer states by a reporter assay. Box plots show the luciferase activity for different categories of enhancers in HSCs and Pro B cells. Group 1 consists of enhancers primed in HSCs and active in Pro B cells (n=8); group 2 consists of enhancers active in both HSCs and Pro B cells (n=11). An enhancer-less construct (see Material and methods) was used as a negative control. Raw data are presented in Supplementary Fig. 7D. (e) Gene expression levels as a function of the number of associated active enhancers. (f). Genome browser snapshots illustrating primed enhancers (Wdr25 locus), poised enhancers (Lgr6 locus), and active enhancers and promoter (Ebf1 locus) in Pro B cells. HSC, haematopoietic stem cells; Mat B, mature B cells; Pro B, progenitor B cells.

Figure 2. Enhancer repertoires are dynamically reshaped from HSCs to mature B cells.

(a) Heatmap showing the dynamic behaviour of the H3K4me1 mark in HSCs, Pro B and mature B cells. H3K4me1-positive regions are shown in blue, H3K4me1-negative regions in white. (b) Genome browser snapshot illustrating different kinds of cell type-specific enhancers: marked in HSCs and mature B but not in Pro B cells (top, Glipr1 locus), marked in Pro B and mature B cells but not in HSCs (middle, Pax5 locus) or marked in all stages (bottom, Spi1/Pu1 locus). (c) TF motif enrichments for enhancers specific to HSCs, Pro B or mature B cells. Sequence logos and P values are shown for the most highly enriched sequence motifs. Only motifs corresponding to TFs that are expressed in the respective cell type are shown (only ETS1 in HSCs and Oct4 in mature B cells were excluded).

Figure 3. Enhancers primed in HSCs rarely become active at later differentiation stages.

(a) Chromatin state of enhancers primed in HSCs was investigated in Pro B, mature B cells, spleen and thymus. H3K4me1 and H3K27ac enrichments in the indicated cell types were calculated at genomic coordinates corresponding to primed enhancers in HSCs. Lines indicate cut-offs used to select enriched regions for each signal. The proportions of the different populations are indicated in the scatter plots. (b) Heatmap displaying enhancers primed in HSCs and clustered based on their H3K4me1 and H3K27ac enrichment in the indicated cell types and tissues. (c) Similar to A, the chromatin state of enhancers primed in Pro B cells was investigated in mature B cells.

Figure 4. Most enhancers active in differentiated cells are not primed in stem cells or progenitor stages.

(a) Chromatin state of active enhancers in mature B cells was investigated in HSCs and Pro B cells. Scatter plots display H3K4me1 and H3K27ac enrichment for HSCs and Pro B cells at the coordinates of active enhancers in mature B cells. The percentage of enhancers in the respective quadrants are indicated. (b) Similar to A, the chromatin state of active enhancers in Pro B cells, spleen and thymus was investigated in HSCs. (c) Genome browser snapshot illustrating the different classes of enhancer dynamics.

Figure 5. Enhancers are used in a stage-specific manner.

(a) Heatmap showing gene expression levels dynamics in HSCs, Pro B and mature B cells. Genes were divided into two categories, expressed (blue) and not expressed (white). (b) Heatmap showing H3K27ac signal dynamics at promoters (proximal peaks, left panel) and outside of promoters (distal peaks, right panel). H3K27ac-positive regions are shown in blue and H3K27ac-negative regions in white. (c) Browser snapshot illustrating stage-specific enhancer usage for the Pax5 gene in Pro B and mature B cells. E1, E2 and E3 stand for Enhancers 1, 2 and 3. Enhancer 1 is common to Pro B and mature B cells, enhancer 2 is specific to mature B cells and enhancer 3 is solely marked by H3K4me1 in Pro B cells and acquires H3K27ac in mature B cells.

Figure 6. Heterochromatin distribution is largely stable from HSCs to mature B cells.

(a) H3K9me2 enrichments in 5-kb windows along the entire chromosome 19 (Chr19) in HSCs, Pro B and mature B cells. Identified heterochromatic domains are indicated by pink rectangles. Mb: megabases. (b) Percentage of the genome marked by H3K9me2 in HSCs, Pro B and mature B cells. Only genomic windows with at least one read in any of the input or H3K9me2 samples was considered. (c) Pairwise comparison of H3K9me2 enrichment in 5 kb sliding windows for HSCs, Pro B and B cells. Pearson correlations are indicated. (d) Similar to C, for promoters (± 1 kb around transcription start sites). (e) Venn diagrams showing the number of nucleotides (in Mb) that are common between the identified H3K9me2 domains in HSCs, Pro B and mature B cells. (f) Browser snapshot illustrating local changes in H3K9me2 at the T-cell receptor (TCR) beta locus. (g) H3K4me1 marked loci were classified according to their overlap with H3K9me2 domains. (h) Enhancers specific to HSCs and closed in Pro B cells were investigated for their overlap with H3K9me2 domains in Pro B cells; similarly, de novo enhancers generated in Pro B cells and absent in HSCs were investigated for their overlap with H3K9m2 domains in HSCs. The proportions of H3K9me2-marked and unmarked loci are indicated.