SUPPLEMENTARY INFORMATION Articles https://doi.org/10.1038/s41590-017-0001-2 In the format provided by the authors and unedited. Single-cell analysis reveals the continuum of human lympho-myeloid progenitor cells Dimitris Karamitros 1,2, Bilyana Stoilova 1,2, Zahra Aboukhalil 1, Fiona Hamey 3, Andreas Reinisch 4, Marina Samitsch 1, Lynn Quek 1,2,5, Georg Otto 1, Emmanouela Repapi 1, Jessica Doondeea 1,2, Batchimeg Usukhbayar 1,2, Julien Calvo 6, Stephen Taylor 1, Nicolas Goardon 1, Emmanuelle Six 7, Francoise Pflumio 6, Catherine Porcher 1, Ravindra Majeti 4, Berthold Göttgens 3 and 1,2,5 Paresh Vyas * 1 MRC Molecular Haematology Unit, MRC Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, UK. 2 Oxford Biomedical Research Centre, University of Oxford, Oxford, UK. 3 Department of Haematology and Wellcome Trust Medical Research Council Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK. 4 Division of Hematology, Stanford Institute for Stem Cell Biology and Regenerative Medicine, Stanford, California, USA. 5 Department of Hematology, OUH NHS Trust, Oxford, UK. 6 UMR967 INSERM/CEA/Université Paris 7/Université Paris 11, Paris, France. 7 UMR1163, Paris Descartes Sorbonne Paris Cité University, Imagine Institute, Paris, France. Dimitris Karamitros and Bilyana Stoilova contributed equally to this work. *e-mail: paresh.vyas@imm.ox.ac.uk Nature Immunology www.nature.com/natureimmunology 2017 Nature America Inc., part of Springer Nature. All rights reserved.
Supplementary Figure 1 Immunophenotypic separation of eight distinct human hemopoietic stem progenitor cell (HSPC) populations and in vitro functional lympho-myeloid potential of LMPPs, MLPs and GMPs. (a) Representative plot of flow sorting of human CB HSPCs. Population frequencies shown are the mean from 44 CB donors calculated as a percentage of Lin - CD34 + compartment. HSC, hemopoietic stem cell; MPP, multipotent progenitor; LMPP, lymphoid-primed multipotent progenitor; MLP, multi-lymphoid progenitor; CMP, common myeloid progenitor; GMP, granulocyte-monocyte progenitor; MEP, megakaryocyte-erythroid progenitor; B/NK, B-NK cell progenitor. (b) Representative flow cytometric analysis of the B cell, NK, monocytic and granulocytic output of CB LMPP, MLP and GMP cultured for 2 weeks on SGF15/2. Frequencies shown are mean from 3 CB donors calculated as a percentage of human CD45 + cells. (c) Representative flow cytometric analysis of T cell output of CB LMPP, MLP and GMP cultured for 5 weeks on SF7a. Frequencies shown are average from 5 CB donors calculated as a percentage of human CD45 + cells.
Supplementary Figure 2 LMPPs and GMPs are lympho-myeloid progenitors, while MLPs are mainly lymphoid progenitors in quantitative in vitro assays. (a) Experimental strategy for sorting, culture and analysis for quantitative in vitro assays in SGF15/2. Similar strategy was used for the other culture conditions. (b) LDA and single cell in vitro culture conditions used in the study. (c) Representative flow cytometric profiles of the outputs from the quantitative in vitro assays. Plots are gated on human CD45 + cells. (d) LDA plots showing the frequency (f: 1 in X cells can give rise to) of LMPP, MLP and GMP cells with B-cell, NK cell, monocytic and granulocytic potential. Plots are generated in R and the lines represent the estimates calculated using ELDA software. (e) Total cloning efficiency (left) of single LMPP, MLP and GMP in S7T2GM/G/M culture (LMPP: 69/96 cells, MLP: 4/52, GMP: 98/110). Significance defined using Fisher s exact test. Cloning efficiency of lymphoid (Ly, middle) and myeloid lineages (My, right). Bars indicate total cloning efficiency; filled portion indicates the proportion of lymphoid (lymphoid plus mixed clones) or myeloid potential (myeloid plus mixed clones). Mean ± SD is shown. Significance is defined using student s t-test. (f) Single-, (g) bi- and (h) multi-lineage single cell outputs in S7T2GM/G/M culture, presented as percentage of the positive wells. (i) Lymphoid (Ly), myeloid (My) and lympho-myeloid (Ly-My) outputs presented as percentage of all plated single cell LMPP, MLP and GMP wells in S7T2GM/G/M culture. (j) Summary of lineage outputs from single LMPP, MLP and GMP. For the LDA data were obtained from 4 CB donors, for the single cell assay in S7T2GM/G/M culture data were obtained from 2 CB donors.
Supplementary Figure 3 Human LMPPs, MLPs and GMPs have distinct differentiation potential in vivo. (a) Experimental strategy for generation of human ossicles in NSG mice and human progenitor transplantation. (b) Gating strategy for the sorting of human progenitors for in vivo transplantation. (c) Number of cells of LMPP, MLP and GMP populations injected per ossicle. (d) Correlation between the number of transplanted cells and the myeloid/lymphoid ratio of the output cells.
Supplementary Figure 4 Transcriptional profiling of human HSPCs in bulk shows the distinct transcriptional patterns of human progenitor populations. (a) Heatmap showing hierarchical clustering and the expression of all genes by HSPC populations. Expression values are normalized per gene. (b) Plot comparing the eigenvalues for each principle component (PC) for the PC analysis (PCA) of HSPC using the top 300 ANOVA genes (black) and 300 randomly selected genes from a randomized expression matrix (red). (c) 3D PCA plot showing HSPC populations using the top 300 ANOVA genes. Percentage variance for each PC is shown. (d) PCA plots showing HSPC populations using 10,000-1,000 genes with the highest variance across all HSPC populations. Percentage variance for each PC is shown. (e) Table of differentially expressed genes in one versus one comparisons of HSPCs. Genes are upregulated in a population column versus row. (f) Heatmap showing the expression of genes recently identified as being expressed by lineage primed hemopoietic cells 25. Genes are color-coded according to their classification. Expression values are normalized per gene. (g-h) Heatmap showing the expression of (g) hemopoietic-related transcription factors that were differentially expressed between the MLP and GMP and (h) cytokine and chemokine related genes by the LMPP, GMP and MLP. Genes affiliated with the lymphoid or myeloid lineages are color-coded (lymphoid: orange, myeloid: green) and genes associated with immune function are labeled in black. Expression values are normalized per gene.
Supplementary Figure 5 Transcriptional heterogeneity of single lympho-myeloid progenitor cells. (a) Heatmap showing hierarchical clustering on genes (rows) and single cells (columns), based on qrt-pcr gene expression analysis. (b) Top, Cell type composition of the 3 clusters identified in the dendrogram in (a). Middle, differentially expressed genes between cluster pairs. Bottom, expected number of cells per cluster, based on the functional assays. (c) Diffusion map dimensionality reductions colored in by cell type (left) and cluster membership (right), based on qrt-pcr gene expression analysis. (d) Heatmap showing clustering of single LMPP, GMP and MLP using the 55 most highly and variably expressed genes between clusters. Single cell RNA-sequencing data from two donors were used to generate the clusters. The heat map shows clustering from one of two. Data from the other donor is in Fig. 5b. Log-normalised gene expression (rows) for each single cell (columns) is shown. (e) Cell type composition of the 3 clusters identified in the heatmap in (d) and Fig. 5b. (f) PCA plot colored by cell type (top) or cluster membership (bottom). For single cell qrt-pcr gene expression analysis: data from 4 CB donors. For single cell RNA sequencing analysis: data from 2 donors (donor 1 in Fig. 5, donor 2 in current figure).
Supplementary Figure 6 Further functional purification of the lympho-myeloid progenitor populations. (a) CD10 fluorescence intensity in LMPPs in the 3 clusters identified from clustering of single cell RNA Seq data presented Fig. 5b. Wilcoxon rank sum test is used to define significance. (b) Histograms showing expression of CD10 and CD45RA for LMPP with different functional output. The percentage cut-offs used to define the new sorting strategy for LMPP ly and LMPP mix are shown. Thresholds were defined based on maximum CD10 and CD45RA expression of LMPPs with myeloid output. (c) Total cloning efficiency (left) of single MLP, LMPP, LMPP ly, LMPP mix and GMP in SF7b culture (LMPP ly : 26/108 cells, LMPP mix : 48/100). Significance defined using Fisher s exact test. Cloning efficiency of lymphoid (Ly, middle) and myeloid lineages (My, right). Bars indicate total cloning efficiency; filled portion indicates the proportion of lymphoid (lymphoid plus mixed clones) or myeloid potential (myeloid plus mixed clones). Mean ± SD is shown. Significance is defined using students t-test. (d) Single-, (e) bi- and (f) multi-lineage outputs from single cells in SF7b culture, presented as percentage of the positive wells. (g) CD38 fluorescence intensity in GMPs in the 3 clusters identified from clustering of single cell RNA Seq data presented Fig. 5b. Wilcoxon rank sum test is used to define significance. (h) Correlation between expression of selected genes and cell surface marker expression in GMPs. ρ indicates Spearman s rank correlation coefficient values and p is corresponding p-value from cor.test function in R. (i) Histograms showing expression of CD38 for GMP with different functional output. The percentage cut-off used to define the new sorting strategy for GMP CD38 hi and CD38 mid are shown. For functional assay based on the revised LMPP sorting strategy (SF7b culture) data are from 6 CB donors; for LMPP, MLP and GMP controls data from 9 CB donors (including 3 CB donors used for Fig. 2f-j).
Supplementary Figure 7 Models of human lympho-myeloid differentiation. (a) Lineage output of lympho-myeloid progenitors (b-c) Models of human hemopoietic differentiation and lineage diversification. Dashed arrows - hierarchical relationships not experimentally validated. (d) Novel model of human lympho-myeloid differentiation. Multiple, rare, functionally distinct bi- and multi-lineage lymphomyeloid progenitors (LMP) could differentiate either directly from hematopoietic stem cells (HSC) or multipotent progenitors (MPP). We have used the term LMP, rather than LMPP (lymphoid primed multi-potential progenitor) to describe these lympho-myeloid progenitors, as not all LMP will be lymphoid biased. Multi-lineage LMPs are rare. Lymphoid only and myeloid only progenitors are shown below. Bi-lineage progenitors are more frequent and uni-potent progenitors are most common. The hierarchical relationships between LMP and lymphoid and myeloid only bi- and uni-lineage progenitors remain to be determined. This model also leaves open the question of when commitment to either lymphoid or myeloid fate occurs. Erythro-megakaryocytic differentiation leading to a MEP (megakaryocyte-erythroid progenitor), erythroid (BFU-E) and megakaryocytic progenitor (MkP) could occur directly from either HSC or MPP.