Developmental and species-divergent globin switching are driven by BCL11A

Abstract

The contribution of changes in cis-regulatory elements or trans-acting factors to interspecies differences in gene expression is not well understood. The mammalian beta-globin loci have served as a model for gene regulation during development. Transgenic mice containing the human beta-globin locus, consisting of the linked embryonic (epsilon), fetal (gamma) and adult (beta) genes, have been used as a system to investigate the temporal switch from fetal to adult haemoglobin, as occurs in humans. Here we show that the human gamma-globin (HBG) genes in these mice behave as murine embryonic globin genes, revealing a limitation of the model and demonstrating that critical differences in the trans-acting milieu have arisen during mammalian evolution. We show that the expression of BCL11A, a repressor of human gamma-globin expression identified by genome-wide association studies, differs between mouse and human. Developmental silencing of the mouse embryonic globin and human gamma-globin genes fails to occur in mice in the absence of BCL11A. Thus, BCL11A is a critical mediator of species-divergent globin switching. By comparing the ontogeny of beta-globin gene regulation in mice and humans, we have shown that alterations in the expression of a trans-acting factor constitute a critical driver of gene expression changes during evolution.

Figure 1. Human γ-globin is primarily expressed in primitive erythroid cells of β-locus mice

a, Representative FACS plot showing FSC (linear scale) versus SSC (log scale) for E13.5 embryonic blood. Gating is shown to allow for the enrichment of primitive (blue population) and definitive lineages (red population). b, Relative expression of murine εy globin gene, human embryonic ε gene, and human γ-globin genes showed similar relative enrichment levels in the primitive population (P), as compared with the definitive population (D). Results are shown as mean ± standard deviation (n≥3 per group). P=0.98 for a two-sided t-test comparing the relative enrichment of εy with γ-globin. c–h, Representative immunohistochemical staining with an anti-HbF antibody from human and murine E13.5 fetal livers. All images are taken with a 60X objective. c, Human fetal livers contain numerous erythroblasts, which all stain positive for γ-globin expression. d,e, In contrast, murine fetal liver definitive erythroblasts do not show major γ-globin staining and only occasional cells with megaloblastic primitive morphology show staining (blue arrows). e,f, Many megaloblastic primitive cells in the circulation show highly positive staining (e, blue arrowheads; f, blue arrows), while smaller definitive erythrocytes are negative (f, red arrows). g, h, Staining performed on the single copy YAC lines A20 and A85 showed similar staining patterns. Positive staining was determined in comparison with background staining from transgene negative littermate controls.

Figure 2. PT-FISH analysis reveals that γ-globin expression parallels the murine embryonic globins in primitive erythroid cells

Two independent lines of transgenic YAC mice, A85 (a,c) and A20 (b,d) were analysed using four color primary transcript RNA fluorescence in situ hybridization (PT-FISH). For the first set of experiments, probes were made to target murine α-globin (mα), human β-globin (hβ), and human γ-globin (hγ). Additionally DAPI was used to identify nuclei of cells. a,b, Expression of γ-globin predominates within the two lines in the primitive populations seen circulating in primitive blood cells (PBC) from embryos E11.5 and E13.5. Minor expression is seen in the mature definitive populations from fetal liver (FL) at E13.5. Many of these cells may represent primitive cells found within the FL parenchyma. e–g, Representative images with this staining pattern of each developmental time point are shown, respectively, for PBC at E11.5, PBC at E13.5, and FL at E13.5. c,d, Probes were made to target murine α-globin (mα), murine εy globin (mεy), and human γ-globin (hγ). These data reveal parallel expression of mεy and hγ. h,i, Representative images with this staining pattern are shown for PBC at E13.5 and FL at E13.5, respectively. The color of the bars in all of the graphs corresponds to the colors of the probes that were detected for each of these primary transcripts. The graphs depict the percentage of active loci and are measured for ≥100 nuclei per probe set at each time point.

Figure 3. BCL11A expression varies between humans and mice, suggesting a model for trans-acting variation in β-globin gene expression

a, In human cells full-length proteins of BCL11A (XL/L isoforms) are reduced within cell populations that express high levels of γ-globin, including primitive and fetal liver cells. Additionally, short variant forms are present at these earlier developmental stages. All human cells were sorted for CD235 and CD71 expression. In contrast, in murine cells, full-length BCL11A protein expression is evident in all definitive progenitor populations, including sorted stage-matched E13.5 fetal liver and bone marrow erythroid cells (all populations were sorted for Ter119+/CD71+). No expression of BCL11A within murine primitive cell populations was detected. b, This model summarizes the ontogeny of β-like globin gene regulation in humans, mice, and β-locus mice^,. The ontogeny of mammalian erythropoiesis and progenitor populations is shown at the top. Progenitor populations, including primitive erythroid populations (EryP-CFC), definitive hematopoietic stem cells (HSC), and definitive erythroid burst-forming unit cells (BFU-E) are depicted. The aorto-gonado-mesonephros (AGM) and placenta are sites of definitive hematopoiesis. The patterns of β-like globin and BCL11A expression seen in the two species are shown below.

Figure 4. BCL11A −/− mice fail to silence expression of mouse embryonic β-like globins and human γ-globin genes

a, The CD71/Ter119 expression pattern is shown for fetal liver cells from E14.5 embryos, revealing grossly normal erythropoiesis with these phenotypic markers. The mean percentages for the populations in each quadrant are shown in red (n=6 for fl/+ controls and n=4 for −/− mutants). The P > 0.1 by a two-sided t-test for all gated populations analyzed. b, The expression of the embryonic globins is shown as a percentage of total mouse β-like globins for control mice (fl/+), BCL11A heterozygous (+/−), and null mice (−/−) at E14.5 (n=10,14,11 respectively). c, The expression of the embryonic globins is shown as a percentage of total mouse β-like globins at E18.5 (n=9,9,7 respectively) d, Immunohistochemistry was performed on E14.5 FLs from BCL11A fl/+ and −/− animals for the embryonic globin εy. Representative sections at 40X magnification with a 10X objective lens are shown. e, Similar IHC staining was performed for βh1 globin. In both cases robust expression is seen in the scattered erythroblasts of the FL in −/−, but not control mice. f, Expression of human β-globin locus genes is shown for animals with the various BCL11A genotypes in the presence of the β-locus YAC transgene (YAC+) at E14.5 (n=4,6,4 for the fl/+, +/−, and −/− animals, respectively) and E18.5 (n=4,7,4). All γ- and β-globin levels for the different genotypes are significantly different (P < 1×10⁻⁵ by a two-sided t-test). All data are plotted as the mean ± the standard deviation of the measurement.