Redundant mechanisms driven independently by RUNX1 and GATA2 for hematopoietic development

Abstract

RUNX1 is essential for the generation of hematopoietic stem cells (HSCs). Runx1-null mouse embryos lack definitive hematopoiesis and die in mid-gestation. However, although zebrafish embryos with a runx1 W84X mutation have defects in early definitive hematopoiesis, some runx1W84X/W84X embryos can develop to fertile adults with blood cells of multilineages, raising the possibility that HSCs can emerge without RUNX1. Here, using 3 new zebrafish runx1-/- lines, we uncovered the compensatory mechanism for runx1-independent hematopoiesis. We show that, in the absence of a functional runx1, a cd41-green fluorescent protein (GFP)+ population of hematopoietic precursors still emerge from the hemogenic endothelium and can colonize the hematopoietic tissues of the mutant embryos. Single-cell RNA sequencing of the cd41-GFP+ cells identified a set of runx1-/--specific signature genes during hematopoiesis. Significantly, gata2b, which normally acts upstream of runx1 for the generation of HSCs, was increased in the cd41-GFP+ cells in runx1-/- embryos. Interestingly, genetic inactivation of both gata2b and its paralog gata2a did not affect hematopoiesis. However, knocking out runx1 and any 3 of the 4 alleles of gata2a and gata2b abolished definitive hematopoiesis. Gata2 expression was also upregulated in hematopoietic cells in Runx1-/- mice, suggesting the compensatory mechanism is conserved. Our findings indicate that RUNX1 and GATA2 serve redundant roles for HSC production, acting as each other's safeguard.

Graphical abstract

Figure 1.

runx1^−un mutants survive to adult and recover multilineage hematopoiesis. (A) Schematic representation of the runx1 gene showing the regions (orange bars) targeted with engineered TALENs that generated the runx1^del8, runx1^del25 mutant lines and the CRISPR targets, T1 and T2 (blue bars), used to generate the runx1^del(e3-8) mutant line as shown in the lower panel. The cyan box in the lower panel indicates a 66-bp insertion between the 2 cut sites. Percentage of gata1:dsRed⁺ (B) and cd41:GFP^low or cd41:GFP^high (C) in Tg(gata1:dsRed; cd41:GFP) runx1^del8/del8 runx1^+/del8 controls at 10 and 17 dpf analyzed by flow cytometry (^##P < .01, ^####P < .0001). (D) Kidney histology and blood smear of a representative 22 dpf runx1^del8/del8 appear similar to a runx1^+/del8 control sibling. (E) Stacked bar chart showing the percentage of adult runx1^del/del8, runx1^del2/del25, and runx1^{del(e3-8)/del(e3-8)} recovered from inbred heterozygous parents; red dashed lines indicate the expected Mendelian ratio. (F) Volcano plot showing the differentially expressed genes in the kidneys between adult runx1^−un and wild-type zebrafish. The red dots identify known hematopoietic markers for thrombocytes, myeloid cells, and HSCs that are differentially expressed at FC > 2 and P_adj < .05. (F′) Gene ontology enrichment analysis results for panel F.

Figure 2.

Development of cd41:GFP⁺ hematopoietic precursors in runx1^del8/del8 AGM and CHT. (A) Snapshots of the AGM region of runx1^del8/del8 and control sibling Tg(cd41:GFP); Tg(kdrl:mCherry) at 52 hours postfertilization (hpf), obtained from time-lapse recordings on a confocal microscope (supplemental Movies 1 and 2). cd41:GFP⁺ mCherry⁺ double-positive cells are indicated with white asterisks. (B-C) Live confocal imaging of the CHT (B) and AGM (C) regions in the Tg(cd41:GFP); (B) Tg(kdrl:mCherry) embryos at 3 dpf. cd41:GFP⁺ cells are found in the CHT. (C) Residual cd41:GFP⁺ cells (white asterisks) are still detectable in the AGM of runx1^del8/del8.

Figure 3.

Single-cell gene expression profile of wild-type and runx1^del8/de cd41:GFP^low hematopoietic precursors at 2.5 dpf. (A) UMAP of freshly FACS-isolated cd41:GFP^low cells from wild-type and runx1^del8/del8 embryos at 2.5 dpf. Colored clusters represent hematopoietic cells; gray clusters are nonhematopoietic (based on expression profile). (B) UMAP depicting the genotypes of the cd41:GFP^low cells. Mint dots, wild type; magenta dots, runx1^del8/del8. (C) Heat map depicting the expression of signature genes representative of different cell identities. The horizontal bars on the top correspond to the clusters identified in panel A (Identity) and show the distribution of the 2 genotypes (Genotype) across the clusters. One hundred representative cells per clusters are shown. (D) Feature plots depicting the expression level of the HSC markers c-myb and lmo2 (purple is high, gray is low). See also supplemental Figure 2 for additional scRNA-seq data from 2.5-dpf embryos.

Figure 4.

Single-cell analysis of wild-type and runx1^del8/del8 cd41:GFP^low at 6, 10, and 16 dpf show only few overlapping populations. UMAP of freshly FACS-isolated cd41:GFP^low cells from wild-type and runx1^del8/del8 embryos at 6 (A), 10 (D), and 16 dpf (G). Colored clusters represent hematopoietic cells; gray clusters are nonhematopoietic (based on expression profile). UMAP depicting the genotypes of the cd41:GFP^low cells at 6 (B), 10 (E), and 16 (H). Mint dots, wild type; magenta dots, runx1^del8/del8 (B,E). At 16 dpf, mint dots represent wild type; magenta dots, runx1^del8/del8 with circulating blood cells (recovered); blue dots, bloodless runx1^del8/del8. (C,F,I) Heat maps depicting the expression of signature genes representative of different cell identities in each of the clusters identified, respectively, in panels A, D, and G. The horizontal bars on the top correspond to the clusters identified the correspondent UMAP (Identity) and show the distribution of the 2 genotypes (Genotype) across the clusters. One hundred representative cells per clusters are shown. See also supplemental Figures 3 and 4 for additional scRNA-seq data.

Figure 5.

Gata2 is upregulated in the HSPCs of Runx1 zebrafish and mice knockouts. (A) Stream plot representing the pseudotime trajectory projection of wild-type cd41:GFP^low cells at 6, 10, and 16 dpf and their different identities. (B) Subway map depicting the distribution of wild-type cd41:GFP^low cells from different time points to the different branches. (C) Stream plot representing the pseudotime trajectory projection of runx1^del8/del8 cd41:GFP^low cells at 6, 10, and 16 dpf and their different identities. (D) Subway map illustrating the contribution of runx1^del8/del8 cd41:GFP^low at different time points to the different branches. (E) Venn diagram representing the number of upregulated genes in runx1^del8/del8 HSC/HSPCs vs wild-type at 6, 10, and 16 dpf (P_adj < .05, FC > 1.5). Sixty-one genes were commonly upregulated, of which 8 were transcription factors (listed). (F) Stream plot depicting the expression of gata2b in the runx1^del8/del8 larval cd41-GFP^low pseudotime development (from panels C and D). See supplemental Figure 5 for additional analyses of hematopoietic differentiation trajectories in runx1^del8/del8 and wild-type larvae at 6, 10, and 16 dpf. (G) Unsupervised hierarchical clustering of wild-type and Runx1^−un RNAseq samples and heat map depicting the differentially expressed genes between wild-type and Runx1^−un c-Kit⁺ HSPCs (940 downregulated and 414 upregulated in the Runx1^−un; P_adj < .05, FC > 2). (H) Top transcription factors enriched for regulating the differentially expressed genes. The x axis of the heatmap lists the transcription factors, and the y axis shows the DE genes sorted in the same order of the y axis in panel B. Bar plot on the top showed the enrichment P values for each transcription factor. (I) Gata2 expression in c-kit⁺ bone marrow cells in control and Runx1^−/− mice, measured by RNA-seq. Round and triangle points mark transcripts per kilobase million values of each mouse in control and Runx1^−/− mice, respectively. Red ticks mark the average transcripts per kilobase million value. See supplemental Figure 6 for additional data on Gata2 expression in Runx1 conditional knockout mice.

Figure 6.

gata2b and gata2a are required for the development of definitive hematopoiesis in runx1 mutants. (A) Schematic representation of the gata2b gene showing the CRISPR targets T1 and T2 (red bars) used to generate the gata2b^ins5 and the gata2b^del7 mutant lines. (B) Survival of adult runx1^del8/gata2b^ins5 double mutants obtained from the incross of runx1^+/del8; gata2b^ins5/ins5 fish. Red dashed lines indicate the expected ratio of runx1^−un recovery based on our previous experimental data (Figure 1E). Each bar segment shows the percentage and number of fish recovered for each genotype. (C) Schematic representation of gata2a genomic structure. The cyan box represents the 150-bp i4 enhancer that was removed using 2 CRISPR guides (red letters). (D) Photographs of representative 1-month-old runx1^+/−gata2b^−atgata2a^+/− and runx1^−ungata2b^−atgata2a^−at fish, with the latter more pale and smaller than the former. (E) Histologic analysis of kidney marrow of triple mutants runx1^del8;gata2b^ins5;gata2a^i4del1 at 1 or 2 months. Kidneys from runx1 wild-type and runx1^+/− fish (panels II, IV, and VI) are similar to the kidney from the wild-type control (panel I) with marrows populated by blood progenitors. Kidneys from runx1^del8/del8 fish with gata2b^−atgata2a^−at (III), gata2b^+/−gata2a^−at (V), or gata2b^+/−gata2a^+/− (VII) completely lack blood progenitors. See supplemental Figure 7 for additional characterizations of the zebrafish gata2b and gata2a mutants.