Integrative cross-species transcriptome analysis reveals earlier occurrence of myelopoiesis in pre-circulation primates compared to mice

Abstract

Hematopoiesis originates in the yolk sac, which forms prior to the establishment of blood circulation and exhibits distinct developmental processes between primates and mice. Despite increasing appreciation of yolk sac hematopoiesis for its lifelong contribution to the adult hematopoietic system and its regulatory roles in organogenesis, cross-species differences, particularly before the onset of blood circulation, remain incompletely understood. In this study, we constructed an integrative cross-species transcriptome atlas of pre-circulation hematopoiesis in humans, monkeys ( Macaca fascicularis), and mice. This analysis identified conserved populations between primates and mice, while also revealing more differentiated myeloid, erythroid, and megakaryocytic lineages in pre-circulation primates compared to mice. Specifically, SPP1-expressing macrophages were detected in primates before the onset of blood circulation but were absent in mice. Cell-cell communication analysis identified CSF1 ⁺ extraembryonic mesoderm cells as a potential supportive niche for macrophage generation, with ligand-receptor interactions between macrophages and other cell populations in the human yolk sac. Interestingly, pre-circulation SPP1 ⁺ macrophages exhibited hallmark signatures reminiscent of a macrophage subset that positively regulates hematopoietic stem cell generation. Our findings provide a valuable cross-species resource, advancing our understanding of human pre-circulation yolk sac hematopoiesis and offering a theoretical basis for the regeneration of functional blood cells.

Keywords: Hematopoiesis; Integrative analysis; Macrophage; Single-cell RNA sequencing; Yolk sac.

Figure 1

Construction of transcriptome atlas of pre-circulation hematopoiesis across humans, monkeys, and mice at single-cell resolution A: Schematic of cross-species transcriptome integration and comparison of pre-circulation hematopoiesis among humans, monkeys, and mice. CS, Carnegie stage; dpf, days post-fertilization; E, embryonic day; CCA, canonical correlation analysis. B: UMAP visualization of scRNA-seq dataset of hematopoietic-related cells derived from human CS7 embryo. Colors represent cell clusters. ExM, extraembryonic mesoderm; EC, endothelial cell; HPC, hematopoietic progenitor cell; MkP, megakaryocyte progenitor; Ery, erythroblast; MyP, myeloid progenitor; Mac, macrophage. C: Dot plot showing scaled mean expression (color) and proportion of cells expressing genes of cell clusters shown in B from human scRNA-seq dataset. D: UMAP visualization of scRNA-seq dataset of endothelial and hematopoietic-related cells derived from CS8–9 monkey embryos. Colors represent cell clusters. HAB, hemogenic angioblast; HEC, hemogenic endothelial cell. E: Dot plot showing scaled mean expression (color) and proportion (size) of cells expressing genes of cell clusters shown in D from monkey ( in vivo) scRNA-seq dataset. F: UMAP visualization of integrative cross-species transcriptome atlas of pre-circulation yolk sac hematopoiesis. Colors of each dot represent cell clusters. Four major categories were labeled in UMAP (mesoderm; extraembryonic mesoderm, ExM; endothelial cell, EC; hematopoietic populations, including myeloid, megakaryocytic, and erythroid populations). Cell populations in dashed box are cells related to mesoderm and ExM cells. Cells marked by gray background are hematopoietic populations. Mk, megakaryocytic populations; aEC, arterial endothelial cell; Neu, neutrophil; YSEC, yolk sac endothelial cell; hematopoietic-endothelial progenitor, Hema; BP, blood progenitor; EMP, erythromyeloid progenitor. G: UMAP visualization of cells in F divided by datasets. Colors of each dot represent cell clusters in each dataset and share the same cell annotation in F. H: UMAP visualization of cells in F divided by datasets, dot colors in UMAP represent sampling stages of each cell. I: UMAP visualization showing expression levels of indicated genes in integrated datasets.

Figure 2

Cross-species transcriptomic comparison of pre-circulation hematopoiesis A: Heatmap visualization of Pearson correlation coefficients between each pair of cell types across datasets from humans, monkeys ( in vivo and in vitro), and mice. Colors represent Pearson correlation coefficients. Hierarchical clustering of cell clusters from integrative cross-species atlas, based on Euclidean distance, is shown on the top of the heatmap. B: Violin plots showing signature scores based on top 60 DEGs between progenitors during the first (BP, left panel) and second (EMP, right panel) hematopoietic wave of the yolk sac in cell clusters from human and monkey datasets. For box plots within each violin plot, center black lines indicate median values, boxes range from the 25th to 75th percentiles. ns: Not significant; ^**: P<0.01; ^***: P<10 ^-8. C: Pseudotime trajectory showing molecular progression from mesoderm precursors to myeloid (green), megakaryocytic (blue), and erythroid (red) cells (diff. is abbreviation of differentiation). Common precursors (yellow) indicate shared cells in both myeloid and megakaryocytic differentiation. Arrowhead indicates common root point of three hematopoietic trajectories for downstream distribution analysis. D: Ridge plots showing cross-species distribution along pseudotime orders of myeloid (green), megakaryocytic (blue), and erythroid differentiation (red). Colors represent species. While human (CS7) and monkey (CS8–9) datasets were from pre-circulation embryos, the mouse dataset included both pre-circulation (E6.75–E8.0) and post-circulation (E8.25–E8.5) embryos. Dashed lines represent end point of mouse pre-circulation distribution. E: Dot plot showing scaled mean expression (color) and proportion (size) of cells expressing myeloid-related DEGs of human HPC, monkey ( in vivo) HPC2, and mouse EMP. Pre-macrophage feature genes are labeled in red.

Figure 3

Cell-cell communication analysis of cross-talk between SPP1 ⁺ Mac and other populations in human yolk sac A: UMAP visualization of co-expression levels of SPP1 and C1QC in scRNA-seq datasets of humans, monkeys ( in vivo and in vitro), and mice, respectively. B: UMAP visualization showing comparative view of expression levels of macrophage-related genes ( CD86, CD163, HLA-DMB, TLR1, MAF, and MAFB) between primates and mice. C: Network plot showing scaled interaction weight between macrophages and other clusters in human CS7 yolk sac, predicted by CellChat. D: Dot plot showing contribution probability of outgoing ligand-receptor signaling from each human yolk sac cell population in C. Sizes of each dot represent contribution probability, while colors represent cell clusters. E: Heatmap showing importance scores of cell populations from human yolk sac in C contributing to CSF and CCL signaling. In CSF signaling, ExM cells play an exclusive sender role, while macrophages are the major receiver. In CCL signaling, macrophages are the major sender and receiver, indicating a self-regulatory pattern. F: UMAP visualization of the distribution of CD24-expressing and CSF1-expressing subsets of ExM cells from human yolk sac dataset. Dot colors represent newly defined cell clusters, gray dots other cell populations. G: Dot plot showing average expression (color) and proportion (size) of cells expressing top DEGs between CD24 ⁺ ExM and CSF1 ⁺ ExM cells. H: Violin plot showing cell orders ranked by differentiation state (1 to 0 represent less-differentiation to more-differentiation states) predicted by cytoTRACE. For box plots within each violin plot, center black lines indicate median values, boxes range from 25th to 75th percentiles. ^***: P<10 ^-8. I: Dot plot showing communication probability (dot color) and P-value (dot size) of indicated ligand-receptor interactions between macrophages (sender) and other cell populations (receptor) in C. J: Heatmap visualization of expression levels of ligand and receptor genes in I. K: Violin plot showing expression levels of CSF1R, CX3CR1, and MRC1 (encoding CD206) in myeloid-related cell clusters in human dataset. L: Heatmap visualization of expression levels of gene sets related to pro-inflammatory signaling (from Hallmark) in cell clusters from human yolk sac. M: Bar plot showing proportions of sampling stages in each myeloid population from Bian et al. ( 2020). Macrophages (Mac) were sampled from yolk sac (CS11-CS17), blood (CS13-CS20), liver (CS12-CS23), lung (CS20 and CS23), skin (CS20 and CS23), and head (CS11-CS23). YSMP, yolk sac-derived myeloid-biased progenitor; GMP, granulocyte–monocyte progenitor. N: Violin plot showing expression level of SPP1 gene in myeloid-related cell clusters in M. Statistical testing was performed between indicated pairs of cell clusters (YS_Mac1 and YS_Mac2, Head_Mac1 and Head_Mac2, Head_Mac2 and Head_Mac3, Head_Mac3 and Head_Mac4). ns: Not significant; ^**: P<0.01; ^***: P<10 ^-8.

Figure 4

Schematic showing earlier myelopoiesis and occurrence of macrophages in primates versus mice before blood circulation onset A: Gradient of colors in different cell populations, including myeloid, megakaryocyte, and erythroid lineages, showing earlier myelopoiesis in primates compared to mice before blood circulation onset (less-diff. and more-diff. represent less- and more-differentiated states, respectively). B: Earlier occurrence of macrophages in primates but not in mice before blood circulation onset. C: Potential cell-cell cross talk between SPP1 ⁺ macrophages and other cellular components in human CS7 yolk sac.