Tracing Hematopoietic Stem Cell Formation At Single-Cell Resolution

Hematopoietic stem cells(HSCs) continuously regenerate all blood and immune cell types throughout life and are also capable of self-renewal. As a key event of developmental hematopoiesis, HSC emergence and maturation at embryonic stage received wide attention in the past few years. Although endothelium-hematopoiesis transition(EHT) has been widely accepted for long time, the details of endothelium to HSC remained largely unknown since it occurs within a narrow time window, while HSC-competent cells at this stage such as pre-HSCs are so rare and lack specific phenotype for isolating them at high purity.Here, by using a single-cell resolution of OP9-DL1 co-culture and serial transplantation strategy, we successfully explored a new marker, CD201, to enrich both T1 pre-HSCs and T2 pre-HSCs at a high frequency of >30%. Then we applied single-cell RNA-Seq technique to analyse five populations closely related to HSC formation: endothelial cells, CD45- and CD45+ pre-HSCs in E11 aorta-gonad-mesonephros(AGM) region, and mature HSCs in E12 and E14 foetal liver. In comparison, the pre-HSCs showed unique features in transcriptional machinery, apoptosis, metabolism state, signalling pathway, transcription factor network, and lnc RNA expression pattern. By comparing with proximal populations without HSC potential, the core molecular signature of preHSCs was identified. Collectively, our work paves the way for dissection of complex molecular mechanisms regulating the step-wise generation of HSCs in vivo, informing future efforts to engineer HSCs for clinical application.Hematopoietic stem cells(HSCs) continuously regenerate all blood and immune cell types throughout life and are also capable of self-renewal. As a key event of developmental hematopoiesis, HSC emergence and maturation at embryonic stage received wide attention in the past few years. Although endothelium-hematopoiesis transition(EHT) has been widely accepted for long time, the details of endothelium to HSC remained largely unknown since it occurs within a narrow time window, while HSC-competent cells at this stage such as pre-HSCs are so rare and lack specific phenotype for isolating them at high purity.Here, by using a single-cell resolution of OP9-DL1 co-culture and serial transplantation strategy, we successfully explored a new marker, CD201, to enrich both T1 pre-HSCs and T2 pre-HSCs at a high frequency of >30%. Co-culture plus transplantation revealed robust short-term reconstitution occurred only in CD201 high but not CD201low/-group within the CD31+CD45-CD41lowc- Kit+ population in E11 AGM region. To further functionally quantify the CD31+ CD45-CD41lowc-Kit+CD201high population, we performed co-cultures seeded with 3-cell and single-cell, respectively, followed by transplantation. 81.0% and 31.3% of the recipients showed apparent long-term multilineage reconstitution, respectively. Secondary transplantation further confirmed the self-renewal ability of donor HSCs from 3 primary repopulated recipients receiving single-cell-initiated cultures. These in vivo functional data validated that the CD31+CD45-CD41lowc-Kit+CD201high subset in E11 AGM region highly enriched T1 pre-HSCs at a frequency of 1:2.3, as calculated by limiting dilution analyses. Therefore, the highly purified CD45- T1 pre-HSC population enabled, for the first time, further analyses of the elusive nascent HSC precursor at single-cell resolution.Principal component analysis(PCA) and unsupervised hierarchical clustering analyses of single-cell RNA-Seq data revealed that T2 CD41 low population was clearly separated into two distinct subpopulations. And marker gene expression analysis reveals that CD201 might be used for enriching T2 pre-HSC. we performed co-cultures seeded with 3-cell and single-cell in this population, respectively, followed by transplantation. Secondary transplantation further confirmed the self-renewal ability of donor HSCs These in vivo functional data validated that the CD31+CD45+cKit+CD201high subset in E11 AGM region highly enriched T2 pre-HSCs at a frequency of 1:2.1.Then we applied single-cell RNA-Seq technique to analyse five populations closely related to HSC formation: endothelial cells, CD45- and CD45+ pre-HSCs in E11 aortagonad-mesonephros(AGM) region, and mature HSCs in E12 and E14 foetal liver. In comparison, the pre-HSCs showed unique features in transcriptional machinery, apoptosis, metabolism state, signalling pathway, transcription factor network, and lnc RNA expression pattern. By comparing with proximal populations without HSC potential, the core molecular signature of pre-HSCs was identified.Unsupervised hierarchical clustering analysis, PCA and t-SNE analyses revealed three clusters: ECs, pre-HSCs, and mature HSCs. The T1 and T2 pre-HSCs were relatively similar to each other whereas E12 and E14 HSCs were partially mixed with each other. We also performed pseudotime analysis by Monocle method, and the pseudotime developmental path gave the same order as that by PCA, showing continuous development from ECs to HSCs through T1 and T2 pre-HSCs. The most dramatic change occurred between ECs and T1 pre-HSCs. The genes downregulated clearly enriched for terms related to cell migration, vasculature development, and blood vessel morphogenesis, while the genes upregulated strongly enriched for hematopoietic or lymphoid organ development and intracellular signalling cascade. We examined the expression patterns of artery and vein-specific genes during HSC specification. At the individual cell level, the arterial and venous features were segregated in E11.0 AGMderived ECs. T1 pre-HSCs expressed evident arterial markers but much lower levels of venous markers as compared to ECs, suggesting that pre-HSCs should have a more intimate lineage relationship with arterial ECs than with venous ones. Supportively, there might be a common endothelial precursor that is specified to the arterial or hemogenic lineage during embryogenesis.We explored the pattern genes, namely the developmental-stage as well as cell-type specifically expressed genes, during HSC formation using PCA. We analysed the dynamic TF network within the pattern genes between ECs and pre-HSCs. Some TFs essential for hematopoietic specification from mesodermal precursors were detected in all five populations. Albeit at varying frequencies and levels, the T1 pre-HSCs and HSCs expressed all the eight reprogramming TFs, the combination of which could confer committed progenitors with HSC potential.Given the five populations defined according to surface markers, we analysed their epitope features using 264 recently reported candidate molecule. Then Cd47 was selected to address this issue, which was expressed in approximately half of putative ECs at E10, the earliest time point when HSC-competent ECs and CD45- pre-HSCs could be detected in AGM region. Upon induction by OP9-DL1 co-culture, only cells derived from CD47+ subset of both ECs and CD45- pre-HSCs could significantly longterm multi-lineage engraft irradiated recipients. Using direct transplantation, long-term chimerism was detected only in c-Kit+CD47+ but not c-Kit+CD47- subset of the E11.5 AGM region. This suggested that 259 the HSC-competent ECs, CD45- pre-HSC, and mature HSCs were exclusively positive for CD47 in E10-E11 AGM region.The molecular signature of pre-HSCs was obtained by comparing with closely related populations without pre-HSC potential. The 98 genes that were shared between T1 and T2 pre-HSCs were designated as pre-HSC signature genes. Therefore, most of the preHSC signature genes deserve comprehensive analyses and evaluation in hematopoietic system of both embryos and adults. In contrast to adult HSCs in a consistently static state as expected, the pre-HSC populations showed pronounced divergence in the expression pattern of 304 cell division related genes, suggesting existence of both actively proliferative and relatively quiescent states. Furthermore, we analysed the cell cycle status of functional T1 pre-HSCs. The cells in S/G2/M phases exhibited the most remarkable proliferation in co-cultures and also most notable reconstitution potential. The T1 cells in G0 phase showed less robust proliferation in vitro and successful repopulation with lower chimerism. In striking contrast, none of the recipients were repopulated by T1 cells in G1 phase, which hardly proliferated in the co-cultures. Collectively, the results imply a previously unrecognized expansion of a proportion of pre-HSCs, prior to their maturation and migration to FL.Next, the gene set enrichment analyses(GSEA) coupled with the pathway gene set data of Kyoto Encyclopedia of Genes and Genomes(KEGG), were applied by pair-wise comparison to explore signalling pathways potentially involved in HSC formation. The result by GSEA was verified by hypergeometric testing based on differentially expressed genes. Compared to ECs, oxidative phosphorylation and glycolysis pathways were both overrepresented in T1 population, implying that mitochondrial aerobic respiration was specifically activated at this stage. In contrast, quiescent HSCs reside in hypoxic BM niches and rely heavily upon glycolysis for energy production. Additionally, we found that the mechanistic targets of rapamycin(m TOR) signaling pathway was highly enriched in T1 pre-HSCs compared with ECs.Collectively, our work paves the way for dissection of complex molecular mechanisms regulating the step-wise generation of HSCs in vivo, informing future efforts to engineer HSCs for clinical application.