Deciphering human macrophage development at single-cell resolution

Subjects

Abstract

Macrophages are the first cells of the nascent immune system to emerge during embryonic development. In mice, embryonic macrophages infiltrate developing organs, where they differentiate symbiotically into tissue-resident macrophages (TRMs)1. However, our understanding of the origins and specialization of macrophages in human embryos is limited. Here we isolated CD45+ haematopoietic cells from human embryos at Carnegie stages 11 to 23 and subjected them to transcriptomic profiling by single-cell RNA sequencing, followed by functional characterization of a population of CD45+CD34+CD44+ yolk sac-derived myeloid-biased progenitors (YSMPs) by single-cell culture. We also mapped macrophage heterogeneity across multiple anatomical sites and identified diverse subsets, including various types of embryonic TRM (in the head, liver, lung and skin). We further traced the specification trajectories of TRMs from either yolk sac-derived primitive macrophages or YSMP-derived embryonic liver monocytes using both transcriptomic and developmental staging information, with a focus on microglia. Finally, we evaluated the molecular similarities between embryonic TRMs and their adult counterparts. Our data represent a comprehensive characterization of the spatiotemporal dynamics of early macrophage development during human embryogenesis, providing a reference for future studies of the development and function of human TRMs.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Transcriptomic landscape and functional characterization of CD45+ haematopoietic cells in tissues from human embryos.
Fig. 2: Developmental trajectory of YSMPs in human embryonic liver.
Fig. 3: Two distinct waves of yolk sac-derived macrophages contribute to TRM populations in human embryos.
Fig. 4: Origin and specification of microglia in human embryos.

Data availability

Raw data from scRNA-seq analysis have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession numbers GSE133345 and GSE137010. Source Data for four Figures and eight Extended Data Figures are provided within the online content of this paper.

Code availability

All data were analysed with standard programs and packages, as detailed above. Scripts can be found at https://github.com/yandgong307/human_macrophage_project.

References

  1. 1.

    Wynn, T. A., Chawla, A. & Pollard, J. W. Macrophage biology in development, homeostasis and disease. Nature 496, 445–455 (2013).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  2. 2.

    Ginhoux, F. & Guilliams, M. Tissue-resident macrophage ontogeny and homeostasis. Immunity 44, 439–449 (2016).

    CAS  PubMed  Google Scholar 

  3. 3.

    Palis, J. et al. Spatial and temporal emergence of high proliferative potential hematopoietic precursors during murine embryogenesis. Proc. Natl Acad. Sci. USA 98, 4528–4533 (2001).

    ADS  CAS  PubMed  Google Scholar 

  4. 4.

    Dzierzak, E. & Bigas, A. Blood development: hematopoietic stem cell dependence and independence. Cell Stem Cell 22, 639–651 (2018).

    CAS  PubMed  Google Scholar 

  5. 5.

    Zhou, F. et al. Tracing haematopoietic stem cell formation at single-cell resolution. Nature 533, 487–492 (2016).

    ADS  CAS  PubMed  Google Scholar 

  6. 6.

    Samokhvalov, I. M., Samokhvalova, N. I. & Nishikawa, S. Cell tracing shows the contribution of the yolk sac to adult haematopoiesis. Nature 446, 1056–1061 (2007).

    ADS  CAS  PubMed  Google Scholar 

  7. 7.

    Ginhoux, F. et al. Fate mapping analysis reveals that adult microglia derive from primitive macrophages. Science 330, 841–845 (2010).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  8. 8.

    Yona, S. et al. Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis. Immunity 38, 79–91 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  9. 9.

    Hoeffel, G. & Ginhoux, F. Ontogeny of tissue-resident macrophages. Front. Immunol. 6, 486 (2015).

    PubMed  PubMed Central  Google Scholar 

  10. 10.

    Frame, J. M., McGrath, K. E. & Palis, J. Erythro-myeloid progenitors: “definitive” hematopoiesis in the conceptus prior to the emergence of hematopoietic stem cells. Blood Cells Mol. Dis. 51, 220–225 (2013).

    PubMed  Google Scholar 

  11. 11.

    Bertrand, J. Y. et al. Definitive hematopoiesis initiates through a committed erythromyeloid progenitor in the zebrafish embryo. Development 134, 4147–4156 (2007).

    CAS  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Ivanovs, A. et al. Human haematopoietic stem cell development: from the embryo to the dish. Development 144, 2323–2337 (2017).

    CAS  PubMed  Google Scholar 

  13. 13.

    McGrath, K. E. et al. Distinct sources of hematopoietic progenitors emerge before HSCs and provide functional blood cells in the mammalian embryo. Cell Rep. 11, 1892–1904 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14.

    Hoeffel, G. et al. C-Myb+ erythro-myeloid progenitor-derived fetal monocytes give rise to adult tissue-resident macrophages. Immunity 42, 665–678 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Ivanovs, A. et al. Highly potent human hematopoietic stem cells first emerge in the intraembryonic aorta-gonad-mesonephros region. J. Exp. Med. 208, 2417–2427 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16.

    Ivanovs, A., Rybtsov, S., Anderson, R. A., Turner, M. L. & Medvinsky, A. Postmenstrual gestational age should be used with care in studies of early human hematopoietic development. Blood 121, 3051–3052 (2013).

    CAS  PubMed  Google Scholar 

  17. 17.

    Zeng, Y. et al. Tracing the first hematopoietic stem cell generation in human embryo by single-cell RNA sequencing. Cell Res. 29, 881–894 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  18. 18.

    Zeng, Y. et al. Single-cell RNA sequencing resolves spatiotemporal development of pre-thymic lymphoid progenitors and thymus organogenesis in human embryos. Immunity 51, 930–948.e936 (2019).

    CAS  PubMed  Google Scholar 

  19. 19.

    Dong, J. et al. Single-cell RNA-seq analysis unveils a prevalent epithelial/mesenchymal hybrid state during mouse organogenesis. Genome Biol. 19, 31 (2018).

    PubMed  PubMed Central  Google Scholar 

  20. 20.

    Kiewitz, R. et al. S100A1, a new marker for acute myocardial ischemia. Biochem. Biophys. Res. Commun. 274, 865–871 (2000).

    CAS  PubMed  Google Scholar 

  21. 21.

    Costa, G. et al. SOX7 regulates the expression of VE-cadherin in the haemogenic endothelium at the onset of haematopoietic development. Development 139, 1587–1598 (2012).

    CAS  PubMed  Google Scholar 

  22. 22.

    Buttgereit, A. et al. Sall1 is a transcriptional regulator defining microglia identity and function. Nat. Immunol. 17, 1397–1406 (2016).

    CAS  PubMed  Google Scholar 

  23. 23.

    Gosselin, D. et al. An environment-dependent transcriptional network specifies human microglia identity. Science 356, eaal3222 (2017).

    PubMed  PubMed Central  Google Scholar 

  24. 24.

    Mass, E. et al. Specification of tissue-resident macrophages during organogenesis. Science 353, aaf4238 (2016).

    PubMed  PubMed Central  Google Scholar 

  25. 25.

    Peschle, C. et al. Haemoglobin switching in human embryos: asynchrony of ζ→α and ε→γ-globin switches in primitive and definite erythropoietic lineage. Nature 313, 235–238 (1985).

    ADS  CAS  PubMed  Google Scholar 

  26. 26.

    Wlodarczyk, A. et al. A novel microglial subset plays a key role in myelinogenesis in developing brain. EMBO J. 36, 3292–3308 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  27. 27.

    Wirsching, H. G. et al. Thymosin β4 induces folding of the developing optic tectum in the chicken (Gallus domesticus). J. Comp. Neurol. 520, 1650–1662 (2012).

    CAS  PubMed  Google Scholar 

  28. 28.

    Enzan, H. Electron microscopic studies of macrophages in early human yolk sacs. Acta Pathol. Jpn. 36, 49–64 (1986).

    ADS  CAS  PubMed  Google Scholar 

  29. 29.

    Tavian, M., Robin, C., Coulombel, L. & Péault, B. The human embryo, but not its yolk sac, generates lympho-myeloid stem cells: mapping multipotent hematopoietic cell fate in intraembryonic mesoderm. Immunity 15, 487–495 (2001).

    CAS  PubMed  Google Scholar 

  30. 30.

    Gordon, S. & Plüddemann, A. Tissue macrophages: heterogeneity and functions. BMC Biol. 15, 53 (2017).

    PubMed  PubMed Central  Google Scholar 

  31. 31.

    Lavin, Y. et al. Tissue-resident macrophage enhancer landscapes are shaped by the local microenvironment. Cell 159, 1312–1326 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Lee, C. Z. W., Kozaki, T. & Ginhoux, F. Studying tissue macrophages in vitro: are iPSC-derived cells the answer? Nat. Rev. Immunol. 18, 716–725 (2018).

    CAS  PubMed  Google Scholar 

  33. 33.

    Takata, K. et al. Induced-pluripotent-stem-cell-derived primitive macrophages provide a platform for modeling tissue-resident macrophage differentiation and function. Immunity 47, 183–198.e186 (2017).

    CAS  PubMed  Google Scholar 

  34. 34.

    Moore, K. L., Persaud, T. V. N. & Torchia, M. G. The Developing Human E-Book: Clinically Oriented Embryology With STUDENT CONSULT Online Access (Elsevier Health Sciences, 2011).

  35. 35.

    Li, L. et al. Single-cell RNA-seq analysis maps development of human germline cells and gonadal niche interactions. Cell Stem Cell 20, 858–873.e854 (2017).

    CAS  PubMed  Google Scholar 

  36. 36.

    Picelli, S. et al. Smart-seq2 for sensitive full-length transcriptome profiling in single cells. Nat. Methods 10, 1096–1098 (2013).

    CAS  PubMed  Google Scholar 

  37. 37.

    Picelli, S. et al. Full-length RNA-seq from single cells using Smart-seq2. Nat. Protocols 9, 171–181 (2014).

    CAS  Google Scholar 

  38. 38.

    Kim, D., Langmead, B. & Salzberg, S. L. HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  39. 39.

    Anders, S., Pyl, P. T. & Huber, W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166–169 (2015).

    CAS  PubMed  Google Scholar 

  40. 40.

    Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering. Nat. Methods 14, 1083–1086 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41.

    Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).

    ADS  CAS  Google Scholar 

  42. 42.

    Chambers, S. M. et al. Hematopoietic fingerprints: an expression database of stem cells and their progeny. Cell Stem Cell 1, 578–591 (2007).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  44. 44.

    Qiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. Nat. Methods 14, 979–982 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  45. 45.

    Haghverdi, L., Büttner, M., Wolf, F. A., Buettner, F. & Theis, F. J. Diffusion pseudotime robustly reconstructs lineage branching. Nat. Methods 13, 845–848 (2016).

    CAS  PubMed  Google Scholar 

  46. 46.

    Darmanis, S. et al. A survey of human brain transcriptome diversity at the single cell level. Proc. Natl Acad. Sci. USA 112, 7285–7290 (2015).

    ADS  CAS  PubMed  Google Scholar 

  47. 47.

    Cohen, M. et al. Lung single-cell signaling interaction map reveals basophil role in macrophage imprinting. Cell 175, 1031–1044.e1018 (2018).

    CAS  PubMed  Google Scholar 

  48. 48.

    MacParland, S. A. et al. Single cell RNA sequencing of human liver reveals distinct intrahepatic macrophage populations. Nat. Commun. 9, 4383 (2018).

    ADS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank F. Tang, Q. Li and Y. Hu from Peking University for technical assistance during single cell library construction, and L. Robinson for language editing of the manuscript. This study was supported by grants from the National Key Research and Development Program of China, Stem Cell and Translational Research (2017YFA0103401, 2016YFA0100601 and 2019YFA0110201), the National Natural Science Foundation of China (31425012, 31930054, 81890991, 31871173, 81800102, 81600077, 81900115 and 31800978), the Program for Guangdong Introducing Innovative and Entrepreneurial Teams (2017ZT07S347), the Key Research and Development Program of Guangdong Province (2019B020234002), the Beijing Municipal Science & Technology Commission (Z171100000417009 and Z171100001117159), the National Key Research and Development Plan Young Scientists Program (2017YFA0106000), the State Key Laboratory of Proteomics (SKLP-K201502) and the China Postdoctoral Science Foundation (2018M643373). F.G. is supported by Singapore Immunology Network (SIgN) core funding. F.G. is a European Molecular Biology Organization (EMBO) YIP awardee and is supported by the Singapore National Research Foundation Senior Investigatorship (NRFI) NRF2016NRF-NRFI001-02. C.Z.W.L. is supported by an A*STAR Graduate Scholarship.

Author information

Affiliations

Authors

Contributions

B.L., F.G. and Y.L. designed the study. Z. Bian and H.S. performed sample preparation and FACS with help from Y.Z., Z. Bai and Y.N. T.H., Z. Bian, H.S., C.L. and J.H. performed scRNA-seq with help from J.Z. and X.L. Z. Bai performed cell culture with help from L.B. and Y.Z. L.B., C.M., R.Z. and L.C. collected and prepared the samples. Y.G. and T.H. performed bioinformatics analysis with help from Z.L., B.L., Y.L., Z. Bian, C.Z.W.L., J.K.Y.C. and L.G.N. Z. Bian, C.Z.W.L., T.H., Y.G., H.S., Z. Bai, Y.L., F.G. and B.L. wrote the manuscript, with contributions from all authors.

Corresponding authors

Correspondence to Yu Lan or Florent Ginhoux or Bing Liu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Hans-Reimer Rodewald and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Technical information about scRNA-seq library and overview of haematopoietic clusters.

a, Graph showing distribution of UMI per cell for STRT–seq data (n = 8 biologically independent embryo samples and 1,461 cells). Threshold for final analysis was set at >10,000. b, Percentage of UMIs mapped on to ERCC and average expression of house-keeping genes. nGene: number of genes expressed, nUMI: number of UMIs expressed. c, UMAP visualization showing exclusion of non-haematopoietic cells from final analysis based on expression of PTPRC, EPCAM, KRT8 and COL1A1. d, Numbers, location and Carnegie stage information for cells used in final analysis. e, UMAP showing minimal batch effect among single cell libraries (n = 1,231 cells). f, g, UMAP visualization of all haematopoietic cells with Carnegie stage (f) and site (g) information mapped on. h, Violin plots of average gene and UMI numbers of scRNA-seq in STRT–seq data. For box plot within each violin plot, centre black lines indicate median values, boxes range from 25th to 75th percentiles, and whiskers correspond to 1.5× IQR. i, Gene and UMI numbers of identified haematopoietic clusters. Centre black lines indicate median values, boxes range from 25th to 75th percentiles, and whiskers correspond to 1.5× IQR. j, Heat map of the top five DEGs between haematopoietic clusters. DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. k, GSEA plots of erythroid and myeloid signatures indicate significantly lower expression of erythroid signature in human YSMPs than in mouse EMPs (extracted from a published scRNA-seq dataset)19 (n = 70 human YSMP cells and 118 mouse EMP cells). P value was calculated using permutation test (one-sided) based on phenotype by GSEA 3.0 software, representing statistical significance of normalized enrichment score (NES). FDR, false discovery rate. l, Gene expression of haematopoietic progenitor feature (CD34, MYB), erythroid feature (GATA1, KLF1) and myeloid feature (SPI1, MPO, LYZ, CSF1R) in human YSMPs and mouse EMPs13,19. Source Data

Extended Data Fig. 2 Characteristics of haematopoietic progenitors in human embryos.

a, Heat map showing differential regulon expression between haematopoietic progenitor clusters (YSMP, n = 116; ErP, n = 72; MkP, n = 30; GMP, n = 45; myeloblast, n = 100; CD7lo progenitor (CD7loP), n = 140; CD7hi progenitor (CD7hiP), n = 104; HSPC, n = 33) generated by SCENIC and clear sets of cell-type specific regulons that may play critical roles in the development of each progenitor population. The number of genes associated with each regulon is listed in parentheses. ErP and MkP signatures were very similar, although MkP appeared to have downregulated expression of KLF1and up-regulated expression of other platelet-related transcription factors such as TAL1 and NFIB. CD7loP had overlapping modules with myeloblast and GMP, sharing the myeloid-restricted TFEC pathways, but lacked expression of more myeloid-committed CEBPs. CD7hiP showed many signatures typical of lymphoid potential, such as the activation of LEF1 and TCF4 signals. HSPC were characterized by activation of the HOXA10 module, as well as higher levels of lymphoid-associated BCL11A when compared to YSMP. b, GSEA plots of the top two differentially expressed regulons between YSMP (n = 116 cells) and HSPC (n = 33 cells). GSEA analysis revealed that YSMPs highly enriched the SPR and ZFP64 regulons, while HSPCs had higher expression of the EOMES and POU3F2 modules. P value was calculated using permutation test (one-sided) based on phenotype by GSEA 3.0 software, representing the statistical significance of enrichment score. c, Volcano plot of DEGs between YSMP (n = 116 cells) and HSPC (n = 33 cells), with the top 10 genes for each cluster indicated. Although the regulon landscape was similar between these two groups, we identified 110 DEGs (Supplementary Table 3). There were more upregulated genes in HSPC (86, red) than in YSMP (24, blue), with HSPC expressing genes related to antigen presentation including CD74 and HLA-DRA as well as lymphoid-related genes including IGLL1. DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.25 and adjusted P < 0.05 were selected. d, Proportion changes of YSMPs and HSPCs in the haematopoietic progenitor populations of yolk sac and liver between CS11 and CS23 (n = 8 biologically independent embryo samples). The proportion of the YSMP population peaked at CS11 before steadily decreasing, while that of the HSPC population expanded between CS17 and CS20 before reducing to about 10% at CS23. e, Proportion changes of different haematopoietic progenitor clusters from CS12 to CS23 in the liver (n = 6 biologically independent embryo samples and 259 cells), and CS13 to CS20 in the blood (n = 5 biologically independent embryo samples and 131 cells). f, Heat map showing expression levels of the top five differentially expressed transcription factors between YSMP, ErP, MkP, myeloblast, HSPC, CD7loP and CD7hiP cells. DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. Source Data

Extended Data Fig. 3 Validation of STRT–seq macrophage clustering in yolk sac by 10x Genomics.

a, Quality control for 10x Genomics data by UMI and gene numbers (n = 2 biologically independent embryo samples and 11,944 cells). The threshold for final analysis was set as gene number >1,000 per cell. b, UMAP visualization of total haematopoietic clusters generated via STRT–seq with cells from yolk sac (n = 5 biologically independent embryo samples and 238 cells) mapped on and coloured by stage information, which were extracted for further analysis. c, PCA of cells from yolk sac in STRT–seq with re-clustering (left) and stage (right) information mapped on (n = 5 biologically independent embryo samples and 238 cells). These cells were re-clustered into three clusters and annotated by gene expression profiles. The YS-Mac1 cluster mainly consisted of cells from CS11, while the YS-Mac2 cluster mainly consisted of cells from CS15. Based on these findings, we selected the 10x Genomics data from the CS11 and CS15 yolk sacs to validate our clustering. d, UMAP visualization of 10x Genomics data from CS11 and CS15 yolk sacs (n = 2 biologically independent embryo samples and 9,565 cells). The Mac cluster was extracted for further analysis. e, PCA of Mac cluster in 10x Genomics data with re-clustering (left) and stage (right) information mapped on. f, Expression profile of top ten DEGs between YS-Mac1 and YS-Mac2 identified by STRT–seq (n = 238 cells) and projected onto the 10x Genomics data (n = 9,565 cells). Even though 10x Genomics data have lower depth compared to STRT–seq, similar expression profiles can be seen for the majority of genes. g, UMAP visualization of integrated yolk sac data from STRT–seq and 10x Genomics analysis (n = 9,803 cells). There is less overlap in the mesenchymal (Mes1 and Mes2) and epithelial (Epi1) clusters because the STRT–seq data were only from CD45+ haematopoietic cells, whereas the 10x Genomics data were from all yolk sac cells. Note that YSMPs from both datasets are well merged. h, Bar plot showing putative surface markers of YSMPs. On the basis of these data, CD34 and CD44 were selected for functional assays. DEGs were identified from combined STRT–seq and 10x Genomics data using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test with P value adjusted for multiple testing using Bonferroni correction), and surface marker genes with fold change >1.25 and adjusted P < 0.05 were selected. Source Data

Extended Data Fig. 4 In vitro functional assay of YSMPs.

a, Gating strategy for sorting of the YSMPs (CD45+CD34+CD44+) from a CS12 yolk sac. b, Representative morphologies of bulk cultures (100 cells per well) of negative control cells (CD45+CD34CD44, n = 5 replication wells) and YSMPs (CD45+CD34+CD44+, n = 16 replication wells) after 14 days of culture on MS5 feeder layer. n = 3 independent experiments from one sample of CS11 yolk sac and two samples of CS 12 yolk sac for YSMPs. Scale bars, 100 μm. c, Representative FACS analysis of cells collected from bulk cultures of YSMPs. Note that the myeloid cells (CD33+) are predominant, in contrast to a small number of erythroid cells (CD235a+) detected (n = 3 independent experiments). d, Representative morphologies of haematopoietic cells generated by a single YSMP from a CS13 yolk sac after 3 and 10 days of culture on MS5 feeder layer. In total, 184 YSMPs were individually cultured and 67 of them generated morphologically typical haematopoietic clusters. Scale bars, 100 μm. e, Representative FACS analysis of four kinds of distinct differentiation potential of single YSMPs. Cells were collected from the single-cell YSMP cultures and in total 39 wells were individually analysed. Lineage differentiation potentials are indicated in red for each clone. Mo/Mac, monocytes/macrophages (CD45+CD33+CD14+); Gr, granulocytes (CD45+CD33+CD66b+); Ery, erythrocytes (CD235a+); Mk, megakaryocytes (CD41a+). Source Data

Extended Data Fig. 5 Developmental trajectory of YSMP in human embryonic liver.

a, PCA of YSMP, GMP, myeloblast and monocyte populations sampled from CS12 to CS17 livers using PC1 and PC2 (n = 4 biologically independent embryo samples and 88 cells). b, PCA matrix of YSMP, GMP, myeloblast and monocyte populations. c, Monocle visualization of YSMP, GMP, myeloblast and monocyte populations sampled from CS12 to CS17 livers with the expression of the indicated genes mapped on. d, Heat map showing scaled expression of branching curated genes of monocyte and neutrophil fates ordered by pseudotime. e, Six main patterns of gene expression compared between YSMP, GMP, myeloblast and monocyte clusters. The expression levels of all pattern genes (coloured lines) and the average expression of each pattern (black line) are shown. The complete list of genes can be found in Supplementary Table 5. f, g, Heat maps displaying expression of monocyte specification-related surface markers (f) and transcription factors (g) along pseudotime. Source Data

Extended Data Fig. 6 Two distinct waves of yolk sac-derived macrophages contribute to TRM populations in human embryos.

a, The area coloured lavender on the UMAP visualization highlights the myeloid groups selected for re-analysis (n = 8 biologically independent embryo samples and 1,231 cells). b, UMAP visualization of all the haematopoietic cells with re-clustered myeloid and macrophage clusters (n = 782 myeloid cells) mapped on. c, Heat map showing scaled expression of the top five DEGs for each re-clustered myeloid and macrophage population. DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. d, UMAP visualization of myeloid cells with stage (top) and site (bottom) information mapped on. e, Heat map showing scaled expression of curated TRM signature genes from a previous mouse study24 in the re-clustered macrophage populations (n = 450 cells: 7 Liver_Mac cells, 51 Blood_Mac cells, 71 Lung_Mac cells, 46 Skin_Mac cells, 61 YS_Mac1 cells, 29 YS_Mac2 cells, 73 Head_Mac1 cells, 38 Head_Mac2 cells, 41 Head_Mac3 cells, and 33 Head_Mac4 cells). Source Data

Extended Data Fig. 7 Monocyte distribution and macrophage specification in human embryonic head, lung, liver and skin.

a, UMAP visualization of myeloid cells with monocytes (n = 64 cells) coloured by site information mapped on. Bar plot shows cell numbers at different sites. b, UMAP visualization of myeloid cells with monocytes and macrophages from human embryonic head (n = 176 cells) mapped on. Cluster (left) and stage information (right) are indicated by colours. c, UMAP visualization of myeloid cells with monocytes and macrophages from human embryonic lung (n = 64 cells) mapped on. Cluster (top) and stage information (bottom) is indicated by colours. d, UMAP visualization of the myeloid cells with macrophages from human embryonic liver (n = 41 cells) coloured by stage information mapped on. These cells were used to study Kupffer cell specification in situ. e, DiffusionMap visualizing differentiation trajectory of embryonic Kupffer cells with stage information (left) and pseudo-order (right) mapped on. Note that the cells also lined up in a continuum from CS12 to CS23, suggesting the gradual and sequential acquisition of TRM identity. f, Heat maps showing scaled expression of DEGs (left) and transcription factors within DEGs (right) in embryonic Kupffer cells across stages with three main gene expression patterns identified. DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. Complete gene list can be found in Supplementary Table 9. g, DiffusionMap visualizing differentiation trajectory of embryonic Kupffer cells with expression levels of the indicated genes mapped on. Expression of C1QB, a gene associated with macrophage tissue residency, was gradually upregulated, while genes related to Kupffer cell function such as CD5L, SPIC and VCAM1 were expressed only at the end of the developmental pathway, suggesting that specialized Kupffer cells began to appear after CS17. Many of the downregulated genes are inflammation- or migration-related, such as CCR2, S100A4 and IL17RA, while the expression of residency and Kupffer cell identity genes such as CD163, TIMD4 and VSIG4 was increased. Many of the signature genes, such as SPIC and VCAM1, have been previously reported in TRMs using animal models, which confirms that these cells were moving towards a more differentiated tissue-resident state. h, DiffusionMap visualization of macrophages from human embryonic skin (n = 49 cells) with stage information (left) and the expression levels of the indicated genes (right) mapped on. Source Data

Extended Data Fig. 8 Characteristics of human embryonic TRMs versus conventional TRMs.

a, b, UMAP visualization of embryonic TRMs and their conventional TRM counterparts with site and stage (a) and cluster (b) information mapped on (n = 464 cells: 20 adult head macrophages, 39 embryonic head macrophages, 97 adult liver macrophages, 9 embryonic liver macrophages, 160 adult lung macrophages, 59 embryonic lung macrophages, 37 paediatric skin macrophages, and 43 embryonic skin macrophages). We performed combined analysis including the four embryonic TRM populations in the present study (Head_Mac4, Liver_Mac, Lung_Mac, and Skin_Mac) and the corresponding conventional TRMs in adults (head, liver and lung from public scRNA-seq data)46,47,48 and children (skin). In total, five main macrophage clusters (head, liver, lung, skin and unspecified) were identified by unsupervised clustering. Two of the embryonic TRM populations (head and liver) clustered with their corresponding adult counterparts. The embryonic skin TRMs distributed into both specified and unspecified clusters, with the former cluster together with those from paediatric skin. The embryonic lung macrophages did not cluster at all with those in the adult lung, which indicated that the differentiation and specification of these TRMs had not yet occurred. c, UMAP visualizations of all TRM clusters with the expression levels of Langerhans (CD207, CD1A and CD1C), microglial (SALL1, CX3CR1 and TMEM119) and Kupffer (ID1, VCAM1 and TIMD4) cell-related genes mapped on. d, Heat map showing scaled expression of the top ten DEGs between the five identified human macrophage clusters (n = 464 cells: 58 head macrophages, 104 liver macrophages, 156 lung macrophages, 45 skin macrophages, and 101 unspecified macrophages). DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. Note that they were distinguished by the expression of TRM genes that have been well described in previous animal and human studies, such as CD207 for the skin, VCAM1 for the liver and P2RY12 for the head. e, Heat map showing scaled expression of DEGs between the embryonic TRMs and their conventional TRM counterparts in each tissue (head, liver, lung and skin). DEGs were detected using FindAllMarkers function in Seurat (one-sided Wilcoxon rank-sum test, with P value adjusted for multiple testing using Bonferroni correction), and genes with fold change >1.5 and adjusted P < 0.05 were selected. The complete list of genes can be found in Supplementary Table 8. Many of the upregulated genes in the embryonic TRMs are related to cell cycle or tissue development, whereas the upregulated genes in the conventional TRMs are more related to immune function. For example, in the head, the embryonic TRMs expressed the neurodevelopmental gene TMSB4X as well as the cell cycle-related gene EEF1A1, whereas conventional TRMs expressed the immune-related gene ITGAX. Embryonic skin macrophages expressed the chemokine SPP1, indicating that they are either cells in transition or have newly arrived in the niche, further supporting our prediction that skin TRM specification has just begun at this time-point. Source Data

Supplementary information

Gating strategy for FACS

Supplementary Figure 1. a, Gating strategy for CD45+CD235a- cells for STRT-seq. b, Gating Strategy for pediatric Langerhans Cells (skin tissue resident macrophages).

Reporting Summary

Supplementary Table 1

Sample Information and cells enrolled. This table shows the Carnegie staging, anatomical location of sample extraction and cell number pre and post quality control.

Supplementary Table 2

Cell information and DEG of all clusters of STRT-seq. This table shows the differentially expressed genes between the clusters obtained by STRT-seq. A total of n = 8 biologically independent embryo samples and 1231 hematopoietic cells were included in this analysis.

Supplementary Table 3

DEG of progenitors and SCENIC binary matrix. This table shows the differentially expressed genes between all progenitor populations, as well as between specifically YSMP and HSPC. The SCENIC binary matrix is also included. This table including n = 640 progenitor cells, which consits of 116 YSMP cells, 72 ErP cells, 30 MkP cells, 45 GMP cells, 100 Myeloblast cells, 33 HSPCs, 104 CD7hiP cells and 140 CD7loP cells, for the sheet of “DEG of all clusters”; 116 YSMPs and 33 HSPCs for the sheet of “DEG of between YSMP and HSPC” and “scenic binary matrix”.

Supplementary Table 4

Cell information and DEG of all clusters of 10x Genomics. This table shows the annotation and differentially expressed genes of the data obtained from CS11 and CS15 yolk sac by 10x Genomics. The sheet1, DEGs_all_clusters_10x, including n = 9803 cells which consists of 6449 cells in CS 11 YS from 10x data, 3116 cells in CS 15 YS from 10x data. The sheet2, DEGs of integrated data, including 9803 cells from 10x and 238 YS cells from STRT-seq.

Supplementary Table 5

Gene patterns of YSMP specification in liver. This table shows the genes that correspond to the patterns identified in Extended Data Fig. 5e.

Supplementary Table 6

DEG of myeloid group subclusters. This table shows the differentially expressed genes between the myeloid clusters identified in Fig. 3a, including n = 782 myeloid cells.

Supplementary Table 7

Gene patterns of head macrophages specification. This table shows the genes that correspond to the patterns identified in Fig. 4c. This table including n = 155 macrophage cells sampling in head, which consists of 9 Head_Mac0 cells, 45 Head_Mac1 cells, 29 Head_Mac2 cells, 39 Head_Mac3 cells and 33 Head_Mac4 cells.

Supplementary Table 8

TRM comparison between embryo and adult (pediatric). This table shows the differentially expressed genes between embryonic and adult/pediatric tissue resident macrophages in Extended Data Fig. 8. This table including n = 464 macrophage cells, which consists of 20 Adult head macrophage cells, 39 embryonic head macrophage cells, 97 adult liver macrophage cells, 9 embryonic macrophage cells, 160 adult lung macrophage cell, 59 embryonic lung macrophage cells, 37 child skin macrophage cells, 43 embryonic skin macrophage cells.

Supplementary Table 9

Gene patterns of liver macrophages specification. This table shows the genes that correspond to the patterns identified in Extended Data Fig. 7f, including n = 41 macrophage cells sampling in liver.

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bian, Z., Gong, Y., Huang, T. et al. Deciphering human macrophage development at single-cell resolution. Nature 582, 571–576 (2020). https://doi.org/10.1038/s41586-020-2316-7

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.