The onset of circulation triggers a metabolic switch required for endothelial to hematopoietic transition

Abstract

Hematopoietic stem cells (HSCs) emerge during development from the vascular wall of the main embryonic arteries. The onset of circulation triggers several processes that provide critical external factors for HSC generation. Nevertheless, it is not fully understood how and when the onset of circulation affects HSC emergence. Here we show that in Ncx1^-/- mouse embryos devoid of circulation the HSC lineage develops until the phenotypic pro-HSC stage. However, these cells reside in an abnormal microenvironment, fail to activate the hematopoietic program downstream of Runx1, and are functionally impaired. Single-cell transcriptomics shows that during the endothelial-to-hematopoietic transition, Ncx1^-/- cells fail to undergo a glycolysis to oxidative phosphorylation metabolic switch present in wild-type cells. Interestingly, experimental activation of glycolysis results in decreased intraembryonic hematopoiesis. Our results suggest that the onset of circulation triggers metabolic changes that allow HSC generation to proceed.

Graphical abstract

Figure 1

Phenotypically defined pro-HSCs are found in Ncx1^−/− E9.5 embryos (A) Confocal whole mount immunofluorescence (WM-IF) of E8.25 (4-6sp) wild-type (+/+) and Ncx1^−/− embryos. Images show maximum intensity projections. Left panel: Arrowheads indicate VE-Cadherin⁺ Runx1⁺ cells in the paired aortae (pa) or vitelline artery (va). N = 6 (+/+), N = 7 (Ncx1^−/−) embryos analyzed. Right panel: N = 6 (+/+), N = 6 (Ncx1^−/−) embryos analyzed. Scale bars: 100 μm. (B) Confocal WM-IF analysis of E9.5 (22-26sp) +/+ and Ncx1^−/− embryos (maximum intensity projections). N = 6 (+/+), N = 8 (Ncx1^−/−) embryos analyzed. da: dorsal aorta; va: vitelline artery; ua:umbilical artery. Scale bars: 100 μm. (C) Confocal WM-IF analysis of E9.5 (22-26sp) +/+ and Ncx1^−/− embryos. Arrowheads indicate examples of Dll4-expressing aortic endothelial cells. N = 3 (+/+), N = 3 (Ncx1^−/−) embryos analyzed. Scale bars: 100 μm (3D), 30 μm (slice). (D) Flow cytometric analysis of E8.25-E8.5 (3-11sp) +/+ and Ncx1^−/− embryos. Embryos of the same genotype were pooled. Data are representative of 4 independent experiments of N = 3 (+/+), N = 5 (Ncx1^−/−) samples of a total of 11 (+/+), 14 (Ncx1^−/−) embryos. Endo: Ter119^- VE-Cad⁺ CD45^- CD41^- 23GFP^- endothelium; HE: Ter119^- VE-Cad⁺ CD45^- CD41^- 23GFP⁺. (E) Graph showing quantification of flow cytometric analysis in (D). Data are mean ± standard deviation (SD). (F) Flow cytometric analysis of E9.5 (21-26sp) +/+ and Ncx1^−/− embryos. Embryos of the same genotype were pooled. Data representative of 4 independent experiments with N = 8 (+/+), N = 8 (Ncx1^−/−) samples of a total of 26 (+/+), 24 (Ncx1^−/−) embryos. Endo: Ter119^- VE-Cad⁺ CD45^- CD43^- CD41^- 23GFP^- endothelium; HE: Ter119^- VE-Cad⁺ CD45^- CD43^- CD41^- 23GFP⁺; Prog: Ter119^- VE-Cad⁺ CD45^- CD43⁺ CD41⁺ progenitor cells; Pro-HSC: Ter119^- VE-Cad⁺ CD45^- CD43^- CD41^low. (G) Graphs showing quantification of flow cytometric analysis in (F). e.e.: embryo equivalent. Data are mean ± SD.

Figure 2

Ncx1^−/− pro-HSCs downregulate key hematopoietic genes (A) Confocal whole mount immunofluorescence analysis (WM-IF) of E9.5 (21-26sp) wild-type (+/+) and Ncx1^−/− embryos. All panels show single 2.5 μm-thick optical slices representative of (left) N = 3 (+/+), N = 2 (Ncx1^−/−) embryos analyzed, (middle left) N = 2 (+/+), N = 2 (Ncx1^−/−) embryos, (middle right) N = 4 (+/+), N = 2 (Ncx1^−/−) embryos, and (right) N = 3 (+/+), N = 2 (Ncx1^−/−) embryos. Scale bars: 30 μm. Arrowheads indicate VE-Cad⁺ Runx1⁺/23GFP⁺ CD43^- HE or CD31⁺ Runx1⁺ HE. Arrows indicate examples of CD41⁺ Runx1⁺ hematopoietic cells. Blue arrowheads indicate examples of VE-Cad⁺ CD43⁺ Runx1⁺ hematopoietic cells in wild-type embryos; blue asterisks highlight the absence of CD43⁺ hematopoietic cluster cells in Ncx1^−/− embryos. da: dorsal aorta; va: vitelline artery. (B) Multiplexed single cell qRT-PCR analysis of HE (Ter119^- VE-Cad⁺ CD45^- CD41^- CD43^- 23GFP⁺) and pro-HSCs (Ter119^- VE-Cad⁺ CD45^- CD41^low CD43^- 23GFP⁺), isolated from E9.5 embryos (22-26sp). Sort gates as in Figure S2B. Samples from two independent experiments (5 +/+, 10 Ncx1^−/− embryos total) with 52 +/+ and Ncx1^−/− HE and 53 +/+ and Ncx1^−/− pro-HSCs analyzed. Violin plots represent the expression of selected genes; black dots indicate average values. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01; ^∗∗∗p ≤ 0.001. (C and D) Clustered heatmaps showing multiplex single cell qRT-PCR analysis of E9.5 +/+ or Ncx1^−/− HE (C) and pro-HSC (D). Columns represent single cells; rows represent genes. Column dendrograms are ordered using hierarchical clustering. Black boxes highlight groups of cells with high Runx1 expression. (E) Multiplexed single cell qRT-PCR analysis showing expression of selected genes as in (B). ^∗p ≤ 0.05; ^∗∗p ≤ 0.01; ^∗∗∗p ≤ 0.001. (F) Confocal WM-IF of E9.5 (21-24sp) +/+ and Ncx1^−/− embryos. Single 2.5 μm-thick slices are shown. N = 3 (+/+), N = 3 (Ncx1^−/−) embryos analyzed. Arrowheads indicate examples of VE-Cad⁺ Runx1⁺ Jag1⁺ cells. Asterisks highlight lack of Jag1 expression in Ncx1^−/− embryos. Scale bars: 30 μm. (G) Confocal WM-IF of E9.5 (23-25sp) wild-type (+/+) and Ncx1^−/− embryos. Single 2.5 μm-thick slices are shown. N = 4 (+/+), N = 4 (Ncx1^−/−) embryos analyzed. Arrowheads indicate examples of CD31⁺ Runx1⁺ NICD⁺ cells. da: dorsal aorta. Scale bars: 30 μm.

Figure 3

Ncx1^−/− embryos display an aberrant peri-aortic microenvironment (A) Confocal WM-IF analysis of E9.5 (22-25sp) +/+ and Ncx1^−/− embryos. Left panels show maximum intensity projections. Boxed area is magnified in the middle and right panels (single 2.5 μm-thick slices). Arrowheads indicate α-SMA⁺ peri-aortic SMCs, absent from Ncx1^−/− embryos (asterisks). Yellow dashed arrow: distance between dorsal aorta (da) and vitelline artery (va). N = 3 (+/+), N = 3 (Ncx1^−/−) embryos analyzed. Scale bars: 300 μm (3D), 50 μm (slice). (B) Distance between dorsal aorta and vitelline artery as a measurement of the sub-aortic mesenchyme thickness. Measurements done on images from N = 7 (+/+), N = 5 (Ncx1^−/−) different embryos (1-4 images/embryo; 5 measurements / image; 16 (+/+), 17 (Ncx1^−/−) different images used. Data are mean ± SD. (C) Flow cytometric analysis of macrophages (Ter119^- CD45⁺ F4/80⁺ CD11b⁺) in E9.5 (21-25sp) +/+ and Ncx1^−/− caudal part (CP). N = 3 (+/+), N = 3 (Ncx1^−/−) embryos were analyzed individually in 2 independent experiments. (D) Quantification of flow cytometric analysis in (C). Data are mean ± SD. (E) Confocal WM-IF of E9.5 (21-24sp) +/+ and Ncx1^−/− embryos (single 2.5 μm-thick slice representative of N = 4 (+/+), N = 4 (Ncx1^−/−) embryos). Arrowheads: peri-aortic F4/80⁺ macrophages. Scale bars: 30 μm. (F) Uniform Manifold Approximation and Projection (UMAP; (Becht et al., 2018)) of the E9.5 (20-23sp) +/+ and Ncx1^−/− PAS scRNA-Seq dataset. Cells were isolated from 4 embryos for each genotype. (G) Percentage of cells in each PAS scRNA-Seq cluster. (H) Bubble plot showing marker genes for each PAS scRNA-Seq cluster. Dot size indicates the percentage of expressing cells; color intensity indicates expression level. (I) Bubble plot showing expression of genes encoding for hematopoietic niche signals in niche cell subsets. Expression is shown separately for +/+ and Ncx1^−/− cells.

Figure 4

Ncx1^−/− pro-HSCs fail to mature into functional HSCs ex vivo (A) Schematic of OP9 co-aggregate culture experiments. (B) Flow cytometric analysis of OP9 co-aggregates with CP of E9.5 (23-28sp) control (+/+ or Ncx1^+/−) or Ncx1^−/− embryos. Co-aggregates were analyzed individually. N = 11 (+/+ or Ncx1^+/−), N = 7 (Ncx1^−/−); 4 independent experiments. (C) Quantification of flow cytometric analysis in (B). Data are mean ± SD. (D) Repopulation analysis of irradiated CD45.1 syngeneic mice transplanted with 1-3 e.e. of E9.5 (21-27sp) control (+/+ or Ncx1^+/−) or Ncx1^−/− CD45.2⁺ CP (left) or YS (right) cells after culture. Graphs shows peripheral blood (PB) chimerism represented as % donor cells (CD45.2⁺) among total CD45⁺ cells, 16 weeks after transplant. Data from 5 independent experiments. Lines show the mean. (E) CFU-C per embryo equivalent (e.e.) of control (+/+ or Ncx1^+/−) or Ncx1^−/− E9.5 (21-27sp) CP and YS, after culture (left) or uncultured (right). N = 5 (+/+ or Ncx1^+/−), N = 5 (Ncx1^−/−) from 3 independent experiments (co-aggregate). N = 6 (+/+ or Ncx1^+/−), N = 7 (Ncx1^−/−) CP; N = 6 (+/+ or Ncx1^+/−), N = 6 (Ncx1^−/−) YS from 2 independent experiments (uncultured). GEMM: granulocyte, erythroid, monocyte/macrophage, megakaryocyte; G/M/GM: granulocyte, monocyte/macrophage; Ery: erythroid. Data are mean ± SD.

Figure 5

Single cell RNA-seq of wild-type and Ncx1^−/− cells undergoing EHT (A) UMAP of 736 EHT cells (362 wild-type and 374 Ncx1^−/−) analyzed by Smart-Seq2 scRNA-seq. Cl: cluster (B) Expression of the venous marker Nr2f2 (COUP-TFII), the arterial marker Dll4, the HE marker CD44 and the master hematopoietic transcription factor Runx1, super-imposed on the UMAP. (C) Gene Ontology (GO) biological processes and KEGG pathways enriched in upregulated (red bars) or downregulated (blue bars) DE genes between cluster 1 (Ncx1^−/−) and 2 (+/+). (D) Gene set enrichment analysis (GSEA) of DE genes (FDR < 0.1) upregulated in cluster 1 compared to cluster 2. The top two gene sets are shown. NES: normalized enrichment score.

Figure 6

Mapping of Ncx1^−/− cells on wild-type EHT diffusion trajectories shows their failure to develop past the Cd44⁺Foxc2⁺Sox17⁺Runx1^lo stage and to switch from glycolysis to OxPhos (A) scRNA-seq diffusion maps of wild-type cells, showing expression of selected genes. Each dot represents an individual cell. Gene expression levels are shown as Log (normalized counts). DC: diffusion component. (B) Diffusion map of wild-type cells indicating 13 Louvain clusters. Clusters were assigned to cell types based on gene expression (Table S2). Clusters 1, 3, 13, 8, 7, 5 were used to compute EHT trajectory. A: arterial; V: venous; EC: endothelial cells; HE: hemogenic endothelium. (C) Projection of Ncx1^−/− cells on the wild-type diffusion map, computed using the Nearest Neighbors regression algorithm. The size of the red dots shows the number of Ncx1^−/− cells localizing to a particular point of the diffusion map. (D) Beeswarm plots of wild-type and Ncx1^−/− cells along the EHT differentiation trajectory, ordered by diffusion pseudotime. Each dot represents an individual cell. (E and F). Scatterplots showing expression of selected genes along diffusion pseudotime. Genes in (E) are Runx1, the hypoxia responsive gene Adm and its receptors; Genes in (F) are genes involved in glycolysis. Gene expression is shown in the y axis as Log (normalized counts). Lines fitting the expression of genes over pseudotime were obtained by locally weighted linear regression.

Figure 7

In vivo and in vitro activation of hypoxia and glycolysis during EHT reduce the hematopoietic output (A) Schematic of dofetilide experiments. (B) Schematic of explant cultures under hypoxia (1% O₂) or normoxia (20% O₂). (C) Multiplexed mini-bulk qRT-PCR analysis on pools of 25 cells. Top row: data from Ter119^- VE-Cad⁺ 23GFP⁺ cells isolated from E9.5 (21-27sp) embryos harvested from control and dofetilide-treated females. N = 5 (control), N = 6 (dofetilide) from 2 independent experiments. Bottom row: data from Ter119^- VE-Cad⁺ 23GFP⁺ cells isolated from 13-24sp CP explants cultured for 24h under hypoxia (1% O₂). N = 5 (20% O₂), N = 5 (1% O₂), 4 independent experiments; samples from the same experimental groups were pooled for analysis. Color code indicates LogFC (fold change) of dofetilide treated versus control (untreated) or 1% O₂ versus 20% O₂. (D) Expression of endothelial and hematopoietic genes in Ter119^- VE-Cad⁺ 23GFP⁺ cells sorted from control and dofetilide-treated E9.5 (21-27sp) embryos, analyzed by multiplexed mini-bulk qRT-PCR on replicates of 25 cells. Data (mean ± SD) from the same 2 independent experiments as in (C). (E) Schematic of explant cultures of wild-type E9.5 (19-26sp) CPs and YSs cultured in presence of DMOG. Explants were analyzed individually. (F) Flow cytometry analysis of DMOG CP explant cultures. N = 11 (DMSO); N = 9 (DMOG 0.1 mM); N = 7 (DMOG 0.5 mM); N = 7 (DMOG 1mM). Data from 4 independent experiments. Data are mean ± SD. (G) Analysis of CFU-C in CP DMOG explant cultures. N = 14 (DMSO); N = 12 (DMOG 0.1 mM); N = 7 (DMOG 0.5 mM); N = 10 (DMOG 1mM), 5 independent experiments. Data are mean ± SD. (H) Flow cytometry analysis of YS DMOG explant cultures. YS: N = 8 (DMSO); N = 7 (DMOG 0.1 mM); N = 7 (DMOG 0.5 mM); N = 4 (DMOG 1mM). Data from 4 independent experiments. Data are mean ± SD. (I) Analysis of CFU-C in YS DMOG explant cultures. Replicates are the same as (H). Data are mean ± SD.