Genetic dissection of the α-globin super-enhancer in vivo

Abstract

Many genes determining cell identity are regulated by clusters of Mediator-bound enhancer elements collectively referred to as super-enhancers. These super-enhancers have been proposed to manifest higher-order properties important in development and disease. Here we report a comprehensive functional dissection of one of the strongest putative super-enhancers in erythroid cells. By generating a series of mouse models, deleting each of the five regulatory elements of the α-globin super-enhancer individually and in informative combinations, we demonstrate that each constituent enhancer seems to act independently and in an additive fashion with respect to hematological phenotype, gene expression, chromatin structure and chromosome conformation, without clear evidence of synergistic or higher-order effects. Our study highlights the importance of functional genetic analyses for the identification of new concepts in transcriptional regulation.

Figure 1. The α-globin regulatory region typifies a super-enhancer in erythroid cells

(a). Heatmap representation of DNase-seq and ChIP-seq signal +/-2kb around the DNase I peak call regions (15,849 peaks), sorted by the ratio of H3K4me1 to H3K4me3. In the side panes the annotated “putative enhancers” are marked blue, and “putative promoters” marked red. (b). Annotation of all DNaseI peak regions (black) as putative enhancers (blue) and putative promoters (red). Annotation category cut-offs are marked by cyan lines. Promoters and enhancers are identified as described. (c). All identified enhancers (n = 1,963) within 12.5kb were ‘stitched’ together, resulting in 1,268 regions that were ranked for Med1 ChIP-seq signal (input-subtracted total reads). In total, 95 stitched enhancer regions were classified as super-enhancers, including the α- and β-globin regulatory regions. (d). The number of constituent enhancers present among the 1,173 stitched regular enhancers and the 95 stitched super-enhancers. Although super-enhancers are enriched for composite enhancers with a high number of constituents, both classes contain single and composite enhancer regions. (e). Med1 binding (input-subtracted reads per million per basepair) across stitched, regular (n = 1,173) and super-enhancers (n = 95) and a region of 3kb up- and downstream. The median size of regular and super-enhancers is used to scale the region between start and end. (f). DNase-seq and ChIP-seq profiles for Med1, Gata1, Klf1, Nf-e2, Scl, and CTCF across the α-globin locus (mm9 in primary Ter119+ erythroid cells. The paired CTCF sites flanking the α-globin locus are highlighted in blue. (Coordinates, mm9: chr11:32,125,268–32,229,368.)

Figure 2. Erythroid super-enhancer constituents vary in transcription factor binding and chromatin signature

(a). Boxplot showing normalized Med1 ChIP-seq density (input-subtracted reads/basepair/million) to constituent enhancer regions as a function of the number of constituents present in the stitched enhancer region. (Box plot shows median and interquartile range; whiskers define 1.5 × IQR) The histogram displays the relative distribution of composite enhancers. (b). All identified enhancers (n = 1,963) were ranked for Med1 ChIP-seq signal (input-subtracted total reads), and 148 individual enhancers were classified as High-Med1 enhancers. The α-globin enhancers have been highlighted as red triangles. (c). PAM-clustering results for the “putative enhancers”. Clusters 2–17 are ranked by mean Med1 signal (cluster 2 highest, cluster 17 lowest signal). Raw read counts, downscaled, input-corrected, background-subtracted and normalized to Klf1 ChIP-seq data within the peak regions. Regions having >15 reads are shown in black. (d). The fraction of individual enhancers that are constituents of super-enhancers (as defined in Fig. 1C) in each cluster. Clusters are ranked by mean Med1 signal of individual enhancers within the cluster. The colour of the bars indicates the number of master erythroid transcription factors bound to the individual enhancers of the cluster. (e). Med1 ChIP-seq signal (input-subtracted total reads) at individual enhancers and the fraction of individual enhancers that are constituents of super-enhancers as a function of the number of bound transcription factors. In total 570 enhancers are bound by one factor; 459 by two factors; 237 by three factors; and 129 by four factors.

Figure 3. Enhancer assays of individual elements

(a) Reporter assays for activity of 5 α–globin enhancers during chick embryonic development. Panels A1-5: Activity of all five enhancers was detected in the developing blood islands, indicated by arrows, in the posterior part of the embryo, at Hamilton Hamburger stage 12 (HH12). Anterior is oriented to the top. Panels A6-10. HH14 reporter activity, driven by all enhancers, was also detected in circulating blood, most notably in the head and heart. Ubiquitous expression of Histone 2B-tethered RFP was used as electroporation control. BI, blood islands; OFT, outflow tract; IFT, inflow tract; H, heart; HB, hindbrain. Scale bar represents 1 mm. Further details are given in Supplementary Table 3. (b) Functional analysis of enhancer activity in E12.5 mouse embryos. The upper panels (b1-5) show a representative LacZ stained embryo for each of the enhancer constructs indicated. Lower panels (b6-10) show a section through the heart (R1) or dorsal aorta (R2-4 and Rm), showing a population of hematopoietic cells. R1 shows a low level of activity in a subset of cells (arrowed), R2 shows robust activity in the majority of hematopoietic cells, but there is no detectable activity in these cells from R3, R4 or the Rm element. Scale bar represents 1 mm in panels 1–5 and 50μm in panels 6–10. Further details are given in Supplementary Table 3.

Figure 4. Hematologic impact of single and double enhancer knockouts

(a). Mouse α-globin locus (chromosome 11), illustrating the α-globin genes (Hba-a1 and Hba-a2, blue highlight); the five α-globin enhancers (R1, R2, R3, Rm and R4, grey) and the regions deleted in each enhancer knockout models (green) in relation to multispecies conserved regions (red). (b). Hemoglobin was measured in adult blood from mice homozygous for each individual enhancer knockout, and for homozygotes for the R2/R3^-/- double deletion. None of the enhancer knockout models exhibits hemoglobin levels outside of the normal range (red shaded box). Hematologic parameters for the R1/R2^-/- double knockout could not be analysed due to its embryonic lethality and hematologic data shown for this model are from adult heterozygotes only. (c). Total reticulocyte count for each model. A significantly elevated reticulocyte count is observed in the R1^-/- and R2^-/- knockout models. All other models fall within the normal range (red shaded box). (d). Blood films (top panel) and brilliant cresyl blue (BCB) stained blood (lower panel) from the blood of wild type (WT), R1^-/- and R2^-/- mice. Whilst the blood films from the three genotypes are essentially identical, increased reticulocytes can be observed in BCB films from the R1^-/- and (to a lesser extent) R2^-/- mice. Data shown represent means and standard deviation. All data are from independent biological replicates, with numbers per group given in Supplementary Table 5. Statistical analysis was performed using a one-way ANOVA with Dunnett’s correction. No randomization was required for any mouse analysis.

Figure 5. α-globin transcription in single and double enhancer knockouts

(a) NanoString quantification of the ratio between α- and β-globin transcripts in steady state RNA isolated from primary mouse fetal liver cells from homozygote mice at E12.5. Samples were taken 12 hours after exposure to high levels of Epo (intermediate erythroblasts). A variable but significant reduction in the α:β-transcript ratio is observed in the R1^-/-, R3^-/-, R4^-/- and R2^-/-/R3^-/- knockouts. A 90% reduction is seen in the R1^-/-/R2^-/- double knockout. (b). NanoString quantification of the α:β-globin transcript ratio in nascent RNA isolated from primary mouse fetal liver cells at the same stage as Figure 5a. Modest effects are observed in the R1^-/-, R2^-/- R4^-/- and R2^-/-/R3^-/- models, with the greatest reduction in α:β-transcript ratio observed in the R1^-/-/R2^-/- double knockout. All data are from a minimum of 3 independent biological replicates, and are shown as mean with standard deviation. Statistical analysis is by one-way ANOVA with Dunnett’s correction (**** reflects p<0.0001). (c). Embryos from WT (++), heterozygote (+/-) and homozygotes (-/-) for the R1/R2 knockout taken at E14.5 Homozygotes are smaller and paler with hemorrhagic areas and evidence of nuchal edema (arrowed). (d). RNA sequencing of nascent RNA obtained from primary fetal liver cultures using metabolic labeling. Unspliced directional transcripts are shown across the α–globin cluster

Figure 6. Analysis of chromatin structure.

(a). Open chromatin landscape (ATAC-seq) at the α-globin cluster in wild type, five individual single enhancer knockout and two double enhancer knockout mice. Formation of the elements in the cluster is not impaired by deletion of any individual enhancers, nor by the double deletions. The position of each individual element of the predicted α globin super-enhancer is named and highlighted by red dashed lines. (Coordinates, mm9: chr11:32,123,000–32,209,000.) (b). Comparison of the interaction profiles from the Hba-a1&2 promoters in primary erythroid cells from WT mice, each of the single and double enhancer knockout models and ES cells (E14). NG Capture-C was performed using the Hba-a1&2 promoters as a viewpoint (since the genes are virtually identical this represents a composite interaction profile from both promoters). The X-axis displays the number of unique interactions from the promoter fragments with each DpnII fragment genome-wide, normalised for total number of interactions. DpnII fragments overlapping the deleted regions removed for visual clarity. The region displayed in panel a is indicated by a black dashed line below the interaction profiles. (Coordinates, mm9: chr11:32,032,001–32,332,000.)