Molecular and Functional Characterization of Lymphoid Progenitor Subsets Reveals a Bipartite Architecture of Human Lymphopoiesis

Article

Molecular and Functional Characterization of Lymphoid Progenitor Subsets Reveals a Bipartite Architecture of Human Lymphopoiesis Graphical Abstract

Authors Kutaiba Alhaj Hussen, Thien-Phong Vu Manh, Fabien Guimiot, ..., Jean Claude Gluckman, Ana Cumano, Bruno Canque

Correspondence [email protected]

In Brief The development pathways of human lymphoid cells remain unclear. Alhaj Hussen et al. show that human lymphoid cells stem from two functionally specialized populations of CD127 and CD127+ early lymphoid progenitors. They propose a ‘‘two-family’’ model of human lymphopoiesis that differs from the standard model of hematopoiesis established in mice.

Highlights: d

d

d

d

Human lymphopoiesis displays a bipartite architecture CD127 and CD127+ ELPs differentiate independently from LMDPs CD127 ELPs generate T cells, marginal zone B cells, NK cells, and ILCs CD127+ ELPs generate marginal zone and follicular B cells, NK cells, and ILCs

Alhaj Hussen et al., 2017, Immunity 47, 680–696 October 17, 2017 ª 2017 Elsevier Inc. https://doi.org/10.1016/j.immuni.2017.09.009

Data Resources GSE90924 GSE84366

Immunity

Article Molecular and Functional Characterization of Lymphoid Progenitor Subsets Reveals a Bipartite Architecture of Human Lymphopoiesis Kutaiba Alhaj Hussen,1 Thien-Phong Vu Manh,2 Fabien Guimiot,3 Elisabeth Nelson,1 Emna Chabaane,1 Marc Delord,4 Maxime Barbier,5 Claire Berthault,6 Nicolas Dulphy,7 Antonio Jose´ Alberdi,8 Odile Burlen-Defranoux,6 Gerard Socie´,7 Jean Christophe Bories,1 Jeroˆme Larghero,9 Vale´rie Vanneaux,9 Els Verhoeyen,10 Thierry Wirth,5 Marc Dalod,2 Jean Claude Gluckman,1 Ana Cumano,6 and Bruno Canque1,11,* U1126, Universite´ Paris-Diderot, E´cole Pratique des Hautes Etudes/PSL Research University, Institut Universitaire d’He´matologie, Hoˆpital Saint-Louis, Paris, France 2Aix Marseille Univ, CNRS, INSERM, CIML, Centre d’Immunologie de Marseille-Luminy, Marseille, France 3Service de Biologie du De ´ veloppement, INSERM UMR 1141, Universite´ Paris-Diderot, Hoˆpital Robert-Debre´, AP-HP, Paris, France 4Bioinformatique, Institut Universitaire d’He ´ matologie, Universite´ Paris-Diderot, Paris, France 5Institut de Syste ´ matique, E´volution, Biodiversite´ (ISYEB), UMR 7205, CNRS, Museum National d’Histoire Naturelle (MNHN), Universite´ Pierre et Marie Curie, and E´cole Pratique des Hautes Etudes (EPHE), Sorbonne Universite´s, Paris Sciences Lettres, Paris, France 6Unite ´ de Lymphopoı¨e`se, De´partement d’Immunologie, Institut Pasteur; Universite´ Paris-Diderot, Sorbonne Paris Cite´, Cellule Pasteur; INSERM U1223, Paris, France 7INSERM UMRS 1160, Institut Universitaire d’He ´ matologie, Universite´ Paris-Diderot, Sorbonne Paris Cite´; Laboratoire d’Immunologie et Histocompatibilite´, AP-HP, Hoˆpital Saint-Louis, 75010 Paris, France 8Plateforme d’Imagerie et de Tri Cellulaire, Institut Universitaire d’He ´ matologie, Hoˆpital Saint-Louis, Paris, France 9APHP, Ho ˆ pital Saint-Louis, Unite´ de The´rapie Cellulaire; CIC de Biothe´rapies, Universite´ Paris Diderot, Sorbonne Paris Cite´, INSERM, F-75475 Paris, France 10CIRI, International center for Infectiology Research, EVIR team, Universite ´ de Lyon, INSERM U1111, Lyon, France; and Centre Me´diterrane´en de Me´decine Mole´culaire (C3M), team ‘‘Controˆle me´tabolique des morts cellulaires’’ Inserm, U1065, 06204 Nice, France 11Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.immuni.2017.09.009 1INSERM

SUMMARY

INTRODUCTION

The classical model of hematopoiesis established in the mouse postulates that lymphoid cells originate from a founder population of common lymphoid progenitors. Here, using a modeling approach in humanized mice, we showed that human lymphoid development stemmed from distinct populations of CD127 and CD127+ early lymphoid progenitors (ELPs). Combining molecular analyses with in vitro and in vivo functional assays, we demonstrated that CD127 and CD127+ ELPs emerged independently from lympho-mono-dendritic progenitors, responded differently to Notch1 signals, underwent divergent modes of lineage restriction, and displayed both common and specific differentiation potentials. Whereas CD127 ELPs comprised precursors of T cells, marginal zone B cells, and natural killer (NK) and innate lymphoid cells (ILCs), CD127+ ELPs supported production of all NK cell, ILC, and B cell populations but lacked T potential. On the basis of these results, we propose a ‘‘two-family’’ model of human lymphoid development that differs from the prevailing model of hematopoiesis.

According to the standard model established in the mouse, the hematopoietic system displays a branched structure organized as a series of dichotomous cell hierarchies and development intermediates with gradually restricted growth and differentiation potentials. Hematopoietic stem cells (HSCs) are assumed to give rise to lineage-biased multipotent progenitors (MPPs) segregating into common myeloid progenitors (CMPs) (Akashi et al., 2000) and common lymphoid progenitors (CLPs) (Kondo et al., 1997). CLPs define a founder population undergoing further diversification into early B cell precursors (EBPs), and natural killer and innate lymphoid cell precursors (NKIPs), which continue to differentiate locally, or into early T cell precursors (ETPs) leaving the bone marrow (BM) for the thymus. Recently, substantial progress has been made in characterizing MPPs, which segregate into populations oriented toward the erythromegakaryocyte, granulocyte-macrophage, or lymphoid lineages, the latter being referred to as lymphoid-primed multipotent progenitors (LMPPs) (Pietras et al., 2015). As they progress along the lymphoid lineage, LMPPs lose myeloid potential to differentiate into lymphoid-restricted CLPs (Adolfsson et al., 2005; Mansson et al., 2007) including multi-lymphoid as well as T, B, and NK-ILC lineage-biased progenitors (Berthault et al., 2017; Ramond et al., 2014). Whether human hematopoiesis follows the same development scheme as in the mouse remains uncertain (Doulatov et al.,

680 Immunity 47, 680–696, October 17, 2017 ª 2017 Elsevier Inc.

A 10 5

10 5

10 5

10 5

10

4

10

4

10

4

10

4

10

4

10

3

10

3

10

3

10

3

10

3

10

2

10

2

10

2

10

2

10

2

0

0

0

B

10 3

10 4

10 5

0

0

10 3

10 4

10 5

10 5

0

0

10 2

10 3

10 4

10 5

0

0

10 3

10 4

10 5

0

10 2

10 3

10 4

10 5

D 3 2 1 0 −1 −2 −3

C

(legend on next page)

Immunity 47, 680–696, October 17, 2017 681

2012). Reports from the last three decades have established that acquisition of CD45RA marks a key developmental transition in human hematopoietic development (Fritsch et al., 1993). compartment comprises Whereas the CD38 /loCD45RA HSCs, MPPs, and megakaryocyte-erythroid progenitors (Notta et al., 2011; Notta et al., 2015), monocyte and dendritic cell (DC) lineages co-segregate with the lymphoid fate in the CD38lo/+CD45RA+ hematopoietic progenitor cell (HPC) population. CD7 (Canque et al., 2000; Hao et al., 2001), CD10 (Doulatov et al., 2010; Galy et al., 1995), or CD62L (Kohn et al., 2012) are commonly used to identify CD45RA+ HPC populations enriched in lymphoid potential, but because none of these are lymphoidrestricted, their value as lineage-defining markers remains contentious. Thus, whether human lymphopoiesis is initiated by a founder population orthologous to murine CLPs is still unclear. To address these issues, we developed a modeling approach in humanized mice (Parietti et al., 2012). Combining optimized immuno-phenotypic stratification of CD34hiCD45RA+ HPCs with molecular and functional characterization of more than 30 cell populations, we show that human lymphoid development stemmed from functionally specialized subsets of CD127 and CD127+ early lymphoid progenitors (ELPs), and propose an updated ‘‘two-family’’ model of human lymphoid development. RESULTS Modeling Fetal Hematopoiesis in Humanized NOD-scidIL2Rgnull (NSG) Mice We have previously shown that NSG mice engrafted with umbilical cord blood (UCB) CD34+ HPCs (referred to as NSGUCB mice hereafter) transiently support production of a BM CD34+CD7+ cell population reminiscent of human fetal thymus-colonizing cells (Parietti et al., 2012). Here, we analyzed NSG-UCB mice for 2 to 8 weeks and showed that CD34+CD7+ cells peaked by day 21, which was then defined as the reference time-point for subsequent analyses (Figures S1A–S1C). Comparable engraftment efficiencies and phenotypes were noted in mice grafted with fetal or neonatal HPCs, which contrasted with < 2% chimerism in recipient mice of adult CD34+ HPCs (data not shown). To determine how the xenogeneic setting compared with primary hematopoiesis, we analyzed CD45+ BM cells from fetal (13 to 29 gestation weeks) and postnatal (13 to 55 years old) donors by multi-parameter flow cytometry. As reported (Haddad et al., 2006), CD34+CD7+ cell percentages

reached maximum levels during gestation trimester two at which time the phenotypic profile of BM cells closely matched that of day-21 NSG-UCB mice (Figures S1D and S1E). Analysis by high-resolution fluorescence scanning of femurs from NSGUCB mice showed that human (hu-) CD45+ cells distributed unevenly in dense foci scattered throughout the trabecular bone area (Figure S1F) which suggests that, for the first 3 weeks after grafting, xenogeneic hematopoiesis proceeds in a fully humanized environment driven by founder effects. These data indicate that, at week-3 post grafting with fetal or neonatal CD34+ HPCs, the BM of NSG-UCB mice supported a multi-lineage human hematopoiesis reminiscent of the mid-fetal period. Lymphoid Cells Segregate into CD127– and CD127+ Cell Compartments Phenotyping hu-CD45+ BM cells from NSG-UCB mice using cocktails of 14–16 antibodies selected from a panel of 55 confirmed their segregation into CD45RA and CD45RA+ compartments. Sub-stratification of CD45RA+ cells showed that CD115, CD116, and CD123 expression phenotypically distinguished monocytes and DCs (data not shown). Analysis of the CD45RA+My (My : CD115 CD116 CD123 ) compartment found that lymphoid cells fell within two CD127 and CD127+ fractions, each of which comprised CD7 CD19+ and CD7+ CD19 cell populations (Figure 1A). Hu-CD45+ cells were then split into 27 CD34hi, CD34int, and CD34 fractions based on 28 parameters, and organized into a network built with a minimum spanning-tree (MSTREE) algorithm (Figures 1B, C; and Table S1). Tree-plot representation confirmed that lymphoid cells differentiated along two CD127 or CD127+ pathways originating from CD34hiCD45RA+ HPCs. MSTREE analysis also showed that granulocytes followed a CD45RA CD7loCD33+ CD66b /+ developmental trajectory directly branching from the CD34hiCD45RA HPCs. Starting from CD34hiCD45RA+ HPCs, monocyte and DC development was characterized by upregulation of CD115 followed by acquisition of CD116 and segregation into CD123+CD116+ plasmacytoid (p)DC and CD123 CD116+ monocyte-DC precursors. Multiplex PCR analysis of 11 cell populations found the expected gene expression patterns within the stem-progenitor (TAL1, GATA2, MEIS1, HLF, MPL), granulocytic (GATA1, GFI1, JAG1), monocytic (CSF1R, CEBPB, SPI1, LMO2, LYN) or pDC (IL3RA, SPIB, IRF4, MEF2C, GZMB) compartments (Figure 1D). Regarding lymphoid populations, CD34intCD127 CD7+ cells overexpressed NK-ILC lineage-associated genes (TOX, ID2,

Figure 1. Lymphoid Cells Segregate into CD127– and CD127+ Cell Compartments (A) Identification of the CD127- and CD127+ lymphoid compartments. Human cells were selected based on huCD45 expression. Gates were set on CD45RA+My (My : CD115, CD116, CD123) cells. The left dot-plot shows subdivision of the cells based on CD127 expression. Black arrows indicate the gating hierarchy. CD127 and CD127+ cells were further partitioned based on CD19 and CD24 expression. Green and red dot-plots show expression of CD7 by CD127 CD19 or CD127+CD19 cells as a function of CD34. (B) Phenotypic stratification of human cells into 27 populations distributed into 4 subgroups: Stem-Progenitor (SP1-3; black), Granulocytic (Gr1-3; red); Monocyte-Dendritic cell (MDC1-11); green); Lymphoid CD127 (L1-5; blue) or CD127+ (R7L1-5; dark blue). Within each subgroup, cells were subdivided into CD34hi (MFI R 2500), CD34int (MFI 500-2500) or CD34lo/ (MFI < 500) subsets, and quoted binarily for expression of 21 surface markers. See also Table S1. (C) MSTREE-based tree-plot representation including hypothetical nodes (small black circles) depicting the phenotypic kinships between the lympho-granulomonocytic and DC lineages. (D) Hierarchical clustering based on 71 genes expressed by 11 cell populations. Gene expression levels were normalized relative to HPRT. The black arrows indicate CD127. Colored squares show lineage-specific fingerprints as defined in Table S2. Data are means of R 6 biological replicates from 3 independent experiments.

682 Immunity 47, 680–696, October 17, 2017

NFIL3, PRF1, CD122, CD247) while their CD127+ counterparts displayed a more composite B lineage-biased (EBF1, VPREB1, RAG1, RAG2, BCL6) gene-expression pattern. Analyzing the CD127 and CD127+CD19+ cells also confirmed their B lineage affiliation. Gene-Expression Profiling of CD34hiCD45RA+ HPCs Identifies CD127– and CD127+ ELPs To define primary lymphoid lineage-defining markers, we subjected CD34hiMy CD19- HPCs (MFI CD34 R 2500) to principal component (PC) analysis. PC1 (47% of the variance) was driven by the reciprocal modulation of surface CD45RA and CD33, while PC2 (20% of the variance) and PC3 (13% of the variance) were respectively supported by ITGB7 upregulation and differential CD127 expression by ITGB7+ cells (Figure 2A). CD34hiMy CD19 HPCs were partitioned into three ITGB7 CD127 , ITGB7+CD127 , and ITGB7+CD127+ cell subsets, with the latter two being referred to as CD127 and CD127+ ELPs hereafter. When we applied this gating scheme to progenitors isolated from the BM of NSG mice engrafted with adult or fetal CD34+ HPCs (Figures S2A and S2B), or from the BM of primary fetal or adult donors (Figures S2C and S2D), we detected the same bipartite organization. Comparing primary fetal and adult ELPs showed a preferential expression of CD7 during fetal life and disclosed a major switch from CD7 to CD10 expression from prenatal to postnatal CD127+ ELPs which confirms that lymphoid progenitors are subjected to phenotypic and functional changes across ontogeny. CD34hi HPCs were then partitioned into 13 populations, three CD45RA CD38lo/+ (A1–A3), six CD127 ITGB7 /+CD7 /+ (F1-6) and three CD127+ITGB7+CD7 /+CD2 /+ (Fa-c) CD45RA+My cell subsets, plus the CD45RA+CD115+ (B1) precursors (referred to as monocyte-DC progenitors [MDCPs] hereafter), which were analyzed by multiplex PCR (Figures 2B and 2C). The full gating scheme and nomenclature of the CD34hi cell subsets are presented in Figure S3A. Focusing on CD45RA+CD127 cells, we found that CD38loITGB7 CD7 cells (F1; referred to as LMDPs thereafter) retained an immature gene-expression profile, and that acquisition of surface ITGB7 by the CD7 (F2 and F3) or CD7+ (F5 and F6) cell fractions coincided with transcriptional activation of master regulators of lymphoid development (LEF1, ETS1, IKZF1, BCL11A) and concomitant down-modulation of stem-progenitor and granulo-monocytic genes (MPL, HLF, MEIS1, KIT, CEBPA, GFI1, GFI1B). Consistent with polarization toward the T or NK-ILC lineages, ITGB72++CD127 CD7+ cells overexpressed TOX, ID2, AHR, CCR7, CCR9, and TRGC transcripts. ITGB7+CD127+ cells did not express myeloid lineage genes but, rather, shared expression of upstream regulators of B cell development (EBF1, MEF2C, RAG1, RAG2) arguing for an intrinsic B cell lineage bias. Among them, the CD7+CD2 cell population (Fb) displayed a multilymphoid gene-expression pattern, while the CD7 CD2 (Fa) or CD7+CD2+ (Fc) cells differentially expressed B cell- (TCF3, BACH2, BCL11A) or NK-ILC-specific genes (ID2, PRF1). CD115+ MDCPs displayed a myelomonocytic gene signature (SPI1, CEBPA, CEBPB, GFI1). These results indicate that differential CD127 expression by CD45RA+ITGB7+ HPCs can be used to identify ELP subsets.

T Cell Potential Is Asymmetrically Repartitioned between the CD127– and CD127+ ELPs In vitro functional assays confirmed that CD127 and CD127+ ELPs lacked granulocyte potential and had limited monocyte and DC potentials (Figures S2E–S2H). Culturing CD45RA+ CD127 cells (F1-6) by 50-cell pools under mono-lineage lymphoid conditions showed robust expansion but limited B, NK-ILC, or T cell output of most immature CD38loITGB7 CD7 LMDPs (F1) (Figures 3A–3C, and S2I and S2J). Consistent with the molecular data, the highest lymphoid potential was detected within the ITGB72++ cells. Over a 2-week culture period, the CD7 cells (F3) efficiently differentiated into CD19+ B cells and CD56+ NK-ILCs but generated only few CD4+CD8+ T cells, which suggested that they comprised a majority of multilymphoid progenitors (MLPs). These are referred to as CD127 CD7 MLPs hereafter. In contrast, their CD7+ counterparts (F6) displayed robust T and NK-ILCs potentials but had a limited capacity to generate B cells. CD127+ ELPs displayed only marginal T potential (Figures 3D, and S2K and S2L). Instead, the CD7 CD2 (Fa) and CD7+CD2+ (Fc) cells showed differential enrichment in B cell versus NK-ILC potentials, while their CD7+ CD2 counterparts (Fb) displayed a balanced B cell and NK-ILC potential. These cell populations were therefore referred to as CD127+ EBPs (Fa), MLPs (Fb) and NKIPs (Fc) thereafter. Consistent with these data, single-cell cultures under the lympho-myeloid condition (Doulatov et al., 2010) showed LMDPs’ high cloning efficiency with a majority of mixed B-Monocyte-NK (B-M-NK) or B-M clones (Figures 3E–3G). In contrast, the CD127 (F3; F6) and CD127+ (Fa; Fb) ELPs were rather efficient at generating NK or NK-M clones. Due to the low cloning efficiency of human EBPs, only a minority of unipotent B cell clones (%2% of total) were detected under these conditions. Mixed B-NK clones represented % 10% of ELP-derived clones arguing for an early segregation of the NK and B potentials. To test the physiological relevance of these findings, primary LMDPs (F1), CD127 CD7 /+ (F3; F6), and CD127+CD7 /+ (Fa; Fb) ELPs sorted from the BM of a 23-week-old fetal donor were assayed for in vitro differentiation potential. The lymphoid potential of these populations closely overlapped with that of their counterparts isolated from the BM of NSG-UCB mice (Figures 3H–3J). These data indicate that CD127 and CD127+ ELPs comprised multilymphoid, as well as already B or NK-ILC lineage-polarized cell subsets, but differed as to T potential, which preferentially segregated with the CD127 CD7+ cell population. Functional Mapping of the Branching Point between CD127– ETPs and NKIPs To determine whether CD127 CD7+ ELPs (F6) comprised bipotent T and NK-ILC precursors or a mix of ETPs and NKIPs, we further partitioned these based on CD7 and CD2 expression into three F6a, F6b, and F6c subsets (Figure 4A), and cultured by 10-cell pools under the T or NK-ILC condition. Consistent with enrichment in ETPs, 57% of clones derived from the CD72++CD2 (F6b) cells contained R 15% CD4+CD8+ T cells versus 37% and 24% of those derived from CD7+CD2 (F6a) or CD7+CD2+ (F6c) cells, respectively (Figure 4B). T differentiation indexes (DI) reached a median of 33 for CD72++CD2 (F6b) cells versus 12 and 6, respectively, for their CD7+CD2 (F6a) and Immunity 47, 680–696, October 17, 2017 683

A

B

C

Figure 2. Upregulation of ITGB7 by LMDPs Coincides with Lymphoid Priming (A) 3D visualization of PCA of CD34hiCD19-My HPCs showing expression of five surface markers. Each cell (multicolor dots) is displayed by the coordinates of the first three principal components. Red dots indicate maximum expression levels; green and yellow dots correspond to medium or low expression levels, respectively. (B) List and phenotype of the cell populations subjected to multiplex PCR analysis. The full gating scheme for BM CD34hiCD19 HPCs from NSG-UCB mice is presented in Figure S3A. HSCs-MPPs, hematopoietic stem cell-multipotent progenitor; Gr, granulocyte progenitor; LMDP, lympho-mono-dendritic cell progenitor; MLP, multilymphoid progenitors; EBP, early B cell precursor; NKIP, NK-ILC precursor; MDCP, monocyte-DC progenitor. (C) Gene-expression profiling of the CD34hi HPCs. Hierarchical clustering is based on the expression of 56 genes. Pools of 50 cells were sorted and analyzed as above by multiplex RT-PCR. Gene expression levels are normalized relative to HPRT. The black arrows indicate CD127. Colored squares and lineage-specific fingerprints are defined as in Figure 1D and Table S2. Data are means from 6 to 12 biological replicates from 3 independent experiments.

684 Immunity 47, 680–696, October 17, 2017

A

B

C

G

E

D

F

H

I

J

(legend on next page)

Immunity 47, 680–696, October 17, 2017 685

CD7+CD2+ (F6c) counterparts (Figure 4C). Culture under the NK-ILC condition confirmed that CD72++CD2 ETPs remained fully competent for generating NK-ILCs, while CD7+CD2+ cells (subsequently referred to as CD127 NKIPs) comprised a majority of unipotent NK-ILC precursors (Figures 4D and 4E). Singlecell cultures under the T or NK-ILC condition confirmed these findings (Figures 4F and 4G). Gene-expression profiling disclosed the expected NK-ILC signature (TOX, ID2, ETS1, PRF1) in CD7+CD2+ NKIPs but T lineage-affiliated genes were not detected in CD72++CD2 ETPs, which nonetheless expressed CCR7 and CCR9 transcripts (Figure 4H). More immature CD7+CD2 cells retained expression of stem-progenitor (KIT, RUNX1, MEIS1) and B lineage-affiliated (EBF1, VPREB1) genes. To search for epigenetic priming of the T potential in ETPs, their transcriptional response to Notch1 signaling was then assessed (Figure 4I). After 48-hour culture onto Notch1 ligand Delta-like 4 (DL4)-Ig, ETPs upregulated a limited set of T lineage genes (TCF7, RAG2, RORA, TRGC) while heterologous B cell or myeloid transcripts were decreased. Notch1-activated CD7+CD2+ NKIPs underwent a profound transcriptional rewiring marked by downregulation of NK-specific transcripts (TOX, ID2, IFNG, PRF1) and concomitant induction of BCL11B as well as of upstream regulators of ILC development (ZBTB16, AHR, RORC, IL17RE). Notch1-activated LMDPs upregulated Notch1 targets DTX1 and NOTCH3, but failed to launch a T program. These data indicate the upstream divergence between ETPs and NKIPs. CD127– and CD127+ NKIPs Differ in Their Dependence on Notch1, IL-7, and IL-15 The capacity of CD127 (F6c) and CD127+ (Fc) NKIPs to generate NK cells and ILCs were next compared. Cultures onto OP9 or OP9-delta-like (DL) 4 stroma cells under the NK-ILC condition resulted in similarly high (R90%) CD56+ cell yields, but with major differences as to cell expansion and growth factor dependence. Whereas CD127 NKIPs depended exclusively on IL-15 and expanded on OP9 cells more robustly than CD127+ NKIPs (324versus16-fold), optimal growth and differentiation of the latter cells required IL-7 plus IL-15 as well as active Notch1 signaling, which increased total cell yields 4 folds (Figures 5A and 5B). Phenotypically, the CD56+CD161+NKp44+ NK-ILCs generated in co-culture with OP9 or OP9-DL4 cells retained an immature

phenotype characterized by lack of surface CD94 and CD16, and of nuclear Eomes, T-bet, and RORgt, precluding further stratification into NK cell or ILC populations (data not shown). Switching to stroma-free NK-ILC cultures resulted in the generation of CD56+ cells comprising phenotypically mature NK cells (CD7+/ CD94+Eomes+T-bet+), ILC1s (CD7+CD94 Eomes T-bet+), and ILC3s (CD7 CD94 Eomes RORC+) (Figures 5C and 5D). Culture under the basal condition showed that, despite comparable expansion (110 versus 89 folds) and similar NK cell yields, CD127 NKIPs were more efficient at generating ILC1s, while CD127+ NKIPs had higher ILC3 output (Figures 5E–5H, ‘‘BSA condition’’). Culturing CD127+ NKIPs onto Notch1 ligand DL4-Ig increased total cell yields 4-fold (89- versus 367-fold) and led to a complete shift toward ILC3s (Figures 5E–5H, ‘‘DL4-Ig condition’’). In contrast, Notch1 activation resulted in 50% decrease of CD127 NKIP growth and, while also promoting ILC3 differentiation, it did not affect ILC1s. Single-cell cultures under the basal condition confirmed these results showing higher cloning efficiency (35% versus 14%) and expansion (434- versus 106-fold) of CD127 NKIPs (Figure 5I). Clonal assays led also to similar percentages of NK- and ILC1-containing clones in CD127 and CD127+ NKIPs (Figure 5J, left and medium panels), and showed the expected enrichment of CD127+ NKIPs in ILC3 precursors (40% versus 27%; p = 0.04, c2 test) (Figure 5J, right panel). Molecular characterization of in vitro-generated NK cells, ILC1s and ILC3s found the expected gene signatures, with substantial differences according to whether cells originated from CD127 or CD127+ NKIPs (Figure 5K). CD127 NKIP-derived NK cells had higher expression of IFNG, GZMB and GZMK transcripts, while ILC3s differentiated from CD127+ NKIPs retained a B lineage trait (VPREB1, CD79B, IGK). Phenotypic analysis based on 35 NK-ILC surface markers, and assessment of activation-induced degranulation and LTi activity, failed to show differences between homologous cell subsets originating from CD127 or CD127+ NKIPs (Figure 5L, and data not shown). Cytokine production assays found that CD56+CD94+ NK cells differentiated from CD127 NKIPs produced higher interferon (IFN)-g and tumor necrosis factor (TNF)-a levels than those issued from CD127+ NKIPs (Figure 5M). Irrespective of their origin, ILC3s produced overall similar IL17A levels. ILC2-specific cytokines (IL-5, IL-13), as well as IL-22 and lymphotoxin (LT)-b, were undetectable in culture supernatants (data not shown).

Figure 3. Asymmetric Repartition of the T Potential between CD127– and CD127+ ELPs (A–C) CD45RA+CD127 cells were seeded by FACS in 96-well plates (50 cells/well) and cultured for 2 weeks under the (A) T (OP9-DL4 cells with SCF, FL, IL-7), (B) NK-ILC (OP9 cells with SCF, FL, IL-7, IL-15), or (C) B (MS5 cells with SCF, TPO, IL-7) conditions. Emerging clones were analyzed for T (CD4+CD8+), NK-ILC (CD56+) or B (CD19+) cell yields. The bi-dimensional scatterplots show percentages of CD4+CD8+, CD56+, or CD19+ cells (y axis) as a function of log10 growth indexes (GI) (x axis). Results were pooled from three independent experiments. See also Figures S2E–S2J. The gating procedure defining the corresponding cell fractions is presented in Figure S3A. (D) CD45RA+CD127+ cells were seeded, cultured and analyzed as above. Results were pooled from three independent experiments. See also Figures S2E-H and S2K-L. (E–G) Cells were deposited individually in 96-well plates and cultured for 2 weeks under the lympho-myeloid condition (MS5 cells with SCF, TPO, IL2, IL-7). Cellcontaining wells (R5 hu-CD45+cells/well) were analyzed by FACS for detection of CD56+ NK cells, CD19+ BLs and CD14+ monocytes. (E) Cloning efficiencies and (F) growth indexes of 44 (F1), 99 (F3), 105 (F6), 83 (Fa), and 62 (Fb) clones recovered from 60 (F1) or 300 (F3, F6, Fa, Fb) seeded cells. Gray bars indicate median values; circles correspond to individual clones. (G) Bar plots show the percentages of mixed B-M-NK, B-M, B-NK, NK-M clones, and pure B, NK, and M clones recovered from each cell type seeded. Results were pooled from two independent experiments. (H–J) CD127 CD7 /+ (F3; F6) and CD127+CD7 /+ (Fa; Fb) ELPs, as well as LMDPs (F1) sorted from the BM of a 23-week-old fetal donor were cultured as above (25 cells/well) under the (H) T, (I) NK-ILC, or (I) B condition. The bi-dimensional scatterplots show percentages of CD4+CD8+, CD56+ or CD19+ cells (y axis) as a function of log10 growth indexes (GI) (x axis). Red circles indicate T cell clones containing R 40% of CD4+CD8+ T cells.

686 Immunity 47, 680–696, October 17, 2017

A

B

C

D

F

E

H

I

G

(legend on next page)

Immunity 47, 680–696, October 17, 2017 687

These data indicate that, although sharing NK, ILC1, and ILC3 potentials, CD127 and CD127+ NKIPs differed by their dependence on Notch1, IL-7, or IL-15. CD127– and CD127+ proB Cells Differ in Maturation Patterns, Dependence on IL-7, and Differentiation Potentials Culture under the B condition found that CD127+CD7 EBPs (Fa) expanded less than CD127 MLPs (F3) but displayed accelerated B cell development (Figures 6A and 6B). Omitting IL-7 decreased 7-fold the percentage of sIgM+ B cells in cultures of CD127+CD7 EBPs but did not affect overall B cell output (Figure 6C). Further phenotypic profiling of the CD127 CD7 MLPs led us to identify a minor CD92++ subset (2%–3%) strongly expressing EBF1 and PAX5 transcripts (Figures 6D and 6E). To ascertain that they corresponded to EBPs, we assayed CD127 CD7 CD92++ cells for acquisition of surface CD19 in short-term cultures. At variance with LMDPs and CD127 CD7 CD9 /+ MLPs, but like their CD127+CD7 counterparts, about 20% of CD127 CD7 CD92++ cells reached the CD34 CD19+ stage after 96 hr (Figure 6F). Phenotypic characterization of downstream CD34int and CD34 CD19+ proB cells (referred to as proB1s and proB2s thereafter) confirmed that they lacked surface (s-) and cytoplasmic (cyt-) IgM (Figure S3B). Like EBPs, the more immature CD34int proB1s segregated into CD127 and CD127+ cell subsets that also differed as to CD9, CD10, CD20, CD24, CD27, and ITGB7 expression (Figures 6G and S3C). Conversely to CD127 proB2s that retained a stable CD9+CD20 phenotype, the CD127+ proB2s segregated into three CD9+CD20 , CD9intCD20+, and CD9 CD20+ cell subsets (Figure S3D). When cultured for 2 weeks under the B condition (Figures 6H–6J), CD127+ proB1s expanded up to 8-fold (median: 4.3-fold) in an IL-7 dependent manner and they generated a mix of CD9+CD20 /+ and CD9-CD20 /+ B cells from which only a minority acquired s-IgM (median: 12%). In contrast, CD127 proB1s showed limited survival but retained high CD9 expression. Molecular profiling disclosed high expression of regulators of B cell differentiation (BCL11A, IRF4, BACH2, FOXO1, FOXP1) and IgH rearrangements (RAG1, RAG2, IGHM) by CD127 proB1s and proB2s, the latters harboring a marginal zone

(MZ) B cell signature (SOX7, Notch2, DTX1) (Descatoire et al., 2014) (Figures 6K and 6L). In contrast, CD127+ proB1-2 s retained an immature gene-expression profile precluding assignment to the MZ or follicular B cell lineages. To bypass this limitation, CD9hiCD20 , CD9intCD20+, and CD9 CD20+ B cells differentiated in vitro from the CD127+ proB1s were then subjected to multiplex PCR analysis (Figure 6M). We found that CD9hiCD20 B cells no longer expressed CD127 transcripts but displayed upregulated SOX7 and Notch2 expression indicating that they corresponded to MZ B cells. Consistent with follicular B cells, CD9 CD20+ B cells retained high CD127 expression and upregulated SPIB, ETS1, COCH, and BCL6 transcripts. These findings indicate that CD127 and CD127+ EBPs and proB cells differed as to growth factor requirements, maturation patterns and differentiation potentials. CD127– and CD127+ ELPs Undergo Distinct Modes of Lineage Restriction Culturing LMDPs for 96 hr on OP9 cells without exogenous growth factors confirmed their capacity to generate the CD127 and CD127+ ELPs as well as CD115+CD116 /+ MDCPs (Figure S4A). Maintaining LMDPs for a longer period of culture (7 days) increased the myeloid output without affecting ELP differentiation (Figure S4B, upper panel). Adding SCF increased total cell yields > 3-fold without interfering with ELP differentiation (Figure S4B, medium panel). Culture under a multi-cytokine condition (SCF + FL + TPO) increased 25-fold total cell yields but inhibited CD127+ ELP differentiation (Figures S4B, lower panel, and S4C). When the CD127 CD7 /+ ELPs (Figure S4D) were cultured in the absence of growth factors, about half of them diverged toward the CD115+CD116 /+ monocyte-DC lineage indicating the lack of lymphoid restriction. Consistent with their intrinsic B cell lineage bias, CD127+CD7 /+ ELPs retained only marginal myeloid potential and, irrespective of CD7 expression, 60%–70% of them reached the CD19+CD127+ proB cell stage after 96-hr culture (Figure S4E). Collectively, these data indicate that CD127 and CD127+ ELPs emerged independently from LMDPs, and underwent distinct modes of lineage restriction.

Figure 4. Early Divergence between CD127– ETPs and NKIPs (A) Gating scheme defining the F6a, F6b, and F6c cell populations. The full gating procedure is presented in Figure S3A. (B and C) Assessment of the T potential of the above defined populations deposited by pools of 10 cells in 96-well plates. (B) The bi-dimensional scatterplots show percentages of CD4+CD8+ (y axis) as a function of log10 growth indexes (GI) (x axis); red circles indicate clones containing R 15% CD4+CD8+ T cells. (C) Quantification of the T potential expressed as differentiation indexes (log10 growth index (GI) 3 CD4+CD8+ cell percentage). Results were pooled from two experiments. (D and E) Assessment of the NK-ILC potential of F6a-c cells deposited by 10-cell pools in 96-well plates. Bi-dimensional scatterplots and quantification of the CD56+ NK-ILCs are presented as above. Results were pooled from two experiments. (F and G) F6b or F6c cells were deposited individually in 96-well plates and cultured under the (F) T or (G) NK-ILC condition. Cell-containing wells (R20 hu-CD45+cells/well) were analyzed as above. Left y-axes indicate cloning efficiencies; right y-axes show differentiation indexes. Quantification of T lineage commitment is based on quantification of CD7+CD5+ T precursors; NK differentiation indexes are defined as above. Results are pooled from two independent experiments. For (C), (E), (F), and G, asterisks indicate statistical significance based on the Mann-Whitney test (p < 0.0001***; p < 0.001**; p < 0.01*). (H) Hierarchical clustering based on the expression of 48 genes by cells of the corresponding A1, F1, and F6a-c populations. Data are means of 4–6 biological replicates from two independent experiments. Colored squares show HSC-MPP, myeloid, T & NK-ILC, and B fingerprints as defined in Table S2. The black arrows indicate CCR7 and CCR9. (I) Effect of Notch1 signaling on the transcriptional profile of F1, F6b, and F6c cells. The corresponding populations were sorted in wells pre-coated with DL4-Ig or bovine serum albumin (10 mg/ml), cultured for 48 hours in complete medium (SCF, FL, IL-7), and sorted again by 50-cell pools for multiplex RT-PCR analysis. Hierarchical clustering is based on the expression of 59 genes. Gene-expression levels are normalized relative to GAPDH. Data are means of four biological replicates from one experiment. Yellow and gray squares correspond to genes with concordant Notch1-dependent variations in F6b and F6c cells; blue boxes indicate genes whose variations are opposite in F6b and F6c cells. The black arrows indicate CCR7 and CCR9.

688 Immunity 47, 680–696, October 17, 2017

A

C

B

D

E

H

L

K

F

G

I

J

M

(legend on next page)

Immunity 47, 680–696, October 17, 2017 689

Notch1 Differently Affects the Emergence of CD127– and CD127+ ELPs In Vivo Whether inhibition of IL-7R signaling affected the emergence of CD127 and CD127+ ELPs was next examined, but treating NSG-UCB mice with JAK3 inhibitor, CP-690550, did not affect lymphoid development (data not shown). To investigate the possible involvement of Notch signaling in this process, we grafted NSG mice with CD34+ HPCs transduced with lentiviral vectors driving expression of a Notch1-specific small hairpin RNA (shNotch1). As reported in the mouse (Radtke et al., 1999), reducing Notch1 expression led to a profound inhibition of intrathymic development in NSG-UCB mice (Figures S5A and S5B). At week 3 after grafting, Notch1 knockdown resulted in a 2-fold increase in the frequency of the most immature CD45RA HPCs, and CD38 upregulation was limited relative to controls (Figure S5C). Restricting the analysis to CD34hiCD45RA+ HPCs found a selective impairment of CD127+ ELP differentiation ( 40%) and a corresponding increase of LMDP frequency (+40%) in shNotch1 mice (Figures S5D and S5G). Sub-stratification of the CD127+ ELPs showed that Notch1 knockdown modulated the balance between CD7+ CD2 MLPs ( 30%) and CD7 CD2 EBPs (+30%), but spared CD7+CD2+ NKIP differentiation (Figures S5E and S6H). As to CD127 ELPs, Notch1 knockdown selectively impaired the emergence of ETPs, whose percentages decreased by 60%. Conversely to Notch1, knocking down Notch2 did not interfere with xenogeneic hematopoiesis (data not shown). These data indicate that CD127 and CD127+ ELPs were differently regulated by Notch1. CD127– and CD127+ ELPs Generate Distinct Lymphoid Subsets In Vivo CD127 and CD127+ ELPs, as well as LMDPs were then transferred intravenously to sublethally irradiated newborn mice and analyzed 4 weeks after grafting (Figure 7A). LMDPs yielded a

detectable BM progeny in 7 of 12 mice, with hu-CD45+ cells representing 0.4%–1% of BM mononuclear cells (median: 0.65%). The engraftment capacity of CD127 and CD127+ ELPs was more limited since hu-CD45+ BM cells were recovered in only 4 of 12 mice, with percentages of 0.05 to 0.1% (median: 0.07%). Phenotyping of hu-CD45+ BM cells confirmed that LMDPs displayed multilineage potential and found limited myeloid output of the CD127 and CD127+ ELPs (Figure 7B). Consistent with the in vitro data, CD127 ELPs generated a majority of CD127 B cells while CD127+ ELPs gave rise to both CD127 and CD127+ B cells. According to their CD127 or CD127+ origin, BM NK-ILCs differed as to CD7 and CD127 expression but lacked EOMES and RORC expression (data not shown). Characterization of hu-CD45+ cells isolated from the spleen of positive mice engrafted with LMDPs (median: 0.08%) or CD127 (median: 0.03%) and CD127+ (median: 0.01%) ELPs confirmed that the CD127+ ELPs mainly produced ILC3s (Figure 7C). Thymus colonization (R100 hu-CD45+ cells/thymus) and intra-thymic T cell development was noted in all positive mice engrafted with CD127 ELPs, whereas only rare hu-CD45+ cells (< 10 hu-CD45+ cells/thymus) could be detected in the thymi of mice receiving LMDPs or CD127+ ELPs (Figure 7D). Thus, CD127 and CD127+ ELPs differed as to the capacity to generate specific B and NK-ILCs subsets, and to ensure thymus colonization in vivo. CD127– and CD127+ ELPs Harbor Family-Specific Gene Signatures Global gene-expression analyses were then performed to compare the gene signatures of LMDPs, CD127 CD7 /+ and CD127+CD7 /+ ELPs sorted from the BM of NSG-UCB mice with those of their primary counterparts from a 21-week-old fetal donor. Principal component and ingenuity pathway analyses confirmed that a mitotic signature distinguished regenerative

Figure 5. CD127– and CD127+ NKIPs Share NK, ILC1, and ILC3 Potentials (A) CD127 (F6c) and CD127+ (Fc) NKIPs were cultured for 2 weeks by 50-cell pools on OP9 or OP9-DL4 cell layers under the NK-ILC condition, and examined for cell growth and CD56+ cell output (gate: huCD45+CD56+CD14 cells). Gray bars indicate median values. Black (CD127 ) or empty (CD127+) circles correspond to individual clones. The data were pooled from two independent experiments. (B) CD127 NKIPs were cultured on OP9 cells, and CD127+ NKIPs on OP9-DL4 cells, under the NK-ILC condition with or without IL-7 or IL-15, and analyzed as above. The data were pooled from two independent experiments. (C and D) Phenotypic characterization of in vitro-generated NK cells, ILC1s and ILC3s. CD127 NKIPs were seeded (50 cells/well) in U-bottom wells, and cultured for 3 weeks with SCF, FL, IL-7, and IL-15. (C) The bi-dimensional dot-plot shows segregation of the in vitro generated CD56+ cells into three CD7 /+CD94+ (NK), CD7+CD94 (ILC1) and CD7 CD94 (ILC3) cell subsets. (D) Intracellular expression of Eomes, T-bet, and RORgt by NK (EOMES+ T-bet+ RORgt ), ILC1s (EOMES T-bet+ RORgt ), and ILC3s (EOMES T-bet RORgt+) defined above. (E–H) Effect of Notch1 ligation on (E) cell growth, as well as (F) NK cell, (G) ILC1, or (H) ILC3 yields. Cells were seeded by 50-cell pools into BSA or DL4-Ig precoated U-bottom wells, cultured and analyzed as above. The data were pooled from two independent experiments. Asterisks indicate statistical significance based on the Mann-Whitney test (p < 0.0001***; p < 0.001**; p < 0.01*). P values ranging from 0.05 to 0.01 are indicated. (I and J) Single-cell cultures: 360 cells of each cell subset were cultured individually in 96-well plates and analyzed as above. (I) Cloning efficiencies (positive wells: R 40 CD56+ cells; left axis) and growth indexes (right axis) are based on analysis of 128 (CD127 NKIPs) and 49 (CD127+ NKIPs) clones. Bars indicate median values; black (CD127 ) or empty (CD127+) circles correspond to individual clones. Assessment of statistical differences is based on the Mann-Whitney test. (J) Bar plots show the percentages of NK cell-, ILC1-, or ILC3-containing clones. Cells are defined as tripotent when they give rise to three cell types; bipotent when generating two cell types, unipotent when giving rise to a unique cell type. Statistical significance of the differences is based on the Fischer’s exact test. (K) Hierarchical clustering based on the expression of 58 genes by NK cells, ILC1s, and ILC3s differentiated from CD127 or CD127+ NKIPs. Gene expression levels are normalized relative to GAPDH. Data are means from 3–6 biological replicates from one experiment. Colored squares show NK, ILC3, and ILC1 fingerprints as defined in Table S2. The black arrows indicate CD127. (L) Boxplots show percentages of CD107+ cells within NK cells, ILC1s, and ILC3s generated from CD127 (gray) or CD127+ (empty) NKIPs after 3-hr culture in the presence (upper boxes) or absence (lower boxes) of K562 cells (ratio 1:1). The data represent means and SD from two independent experiments performed in triplicates. (M) Levels of IFN-g, TNF-a, and IL-17A produced by NK cells, ILC1s, or ILC3s differentiated from CD127 or CD127+ NKIPs stimulated for 3 hr with PMA and ionomycin. The data show means and SD of four biological replicates from one representative experiment.

690 Immunity 47, 680–696, October 17, 2017

A

D

B

C

K

E L

M

F H

I

J

G

Figure 6. CD127– and CD127+ proB Cells Differ in Their Capacity to Generate MZ or Follicular B Cells (A)-(C) In vitro assessment of the B potential of CD127 CD7 MLPs (F3) and CD127+CD7 EBPs (Fa). Cells were cultured by 50-cell pools for 2 weeks under the B condition, with or without IL-7, and analyzed for (A) cell growth, (B) CD19+ cell yields, (C) and sIgM expression. Results were pooled from two independent experiments. Grays bars indicate median values. Black (CD127 ) and empty (CD127+) circles correspond to individual clones. Asterisks indicate statistical significance based on the Mann-Whitney test (p < 0.0001***; p < 0.001**; p < 0.01*). (D–F) Characterization of in vivo-generated CD127 CD92+ EBPs. (D) Gating strategy for isolation of CD127 CD92+ EBPs. Dot-plots show the subdivision of the CD127 CD7 MLP and CD127+CD7 EBP cell populations according to CD9 expression levels. See also Figure S3A for the complete gating procedure. (legend continued on next page)

Immunity 47, 680–696, October 17, 2017 691

NSG-UCB from steady-state fetal hematopoiesis (Figures S6A–S6C). High-throughput gene set enrichment analyses (GSEA) and cross-dataset enrichment analysis (Spinelli et al., 2015) using cell subset-specific gene signatures revealed the correspondence between phenotypically identical NSG-UCB and primary populations, albeit with limited overlap for CD127 CD7 MLPs (Table S3 and Figures S6D–S6F). These data indicate that, during the first 3 weeks after grafting, the xenogeneic setting did not introduce major distortions in the transcription profile of human LMDPs and ELPs. LMDPs were then used as reference to generate consensus relative signatures shared by homologous NSG-UCB and primary ELP subsets (Table S4). Direct comparison of CD127 CD7 and CD7+ ELPs found the expected segregation of T and NK-ILC lineage genes in the latters and disclosed a more composite gene expression pattern in their multilymphoid counterparts (Figures S6G and S6H; and Table S5 left column). In the same manner, CD127+CD7 and CD7+ ELPs differentially expressed B versus NK-ILC lineage transcripts (Figure S6I and Table S5 right column). To define molecular identities on a family basis, we generated consensus gene signatures specifying the CD127 (CD7+ and CD7 ; 225 annotated genes) or CD127+ (CD7+ and CD7 ; 178 annotated genes) ELPs (Figure S7A left panel, and Table S6 left columns). Consistent with an intrinsic B lineage bias, the gene signature of CD127+ ELPs was selectively enriched in B lineage genes (EBF1, PAX5, CD79A, CD79B). Analyzing CD127 ELPs confirmed the higher expression of genes linked to NK-ILC development (AHR, ID2, NFIL3) or thymus colonization (CCR7, CCR9), and disclosed persistence of an original myeloid trait (CCR1, CCR2, CD36, CD68, MPEG1, IL6ST, IRF8, RUNX2). Collectively, these results demonstrate that CD127 and CD127+ ELPs displayed family-specific gene signatures. Family-Specific ELP Gene Signatures Are Evolutionary Conserved Cross-species GSEA were finally performed to compare the family-specific gene signatures of human ELPs with those of recently characterized immature (HSAint), B (HSAhi), or T and NK-ILC (HSAlow) lineage-biased CLP subsets isolated from the fetal liver (FL) of E13 mouse embryos (Berthault et al., 2017) (Figure S7B). To examine the homologies between human ELPs and mouse CLP subsets, we generated gene signatures defining the HSAlow (799 annotated genes) and HSAhi (383 annotated genes) mouse CLPs (Table S6 right columns), and then tested for enrichment in human CD127 or CD127+ ELPs. Cross-dataset

GSEA showed selective enrichment of the HSAlow CLP signature in human CD127 ELPs, and, reciprocally, of the HSAhi CLP signature in CD127+ ELPs (Figures S7C and S7D). Performing GSEA with the gene signatures defining human CD127 and CD127+ ELPs confirmed these results by showing a differential enrichment between HSAhi and HSAlow mouse CLPs (Figures S7E and S7F). Genes contributing to the GSEA leading edges were then extracted to generate cross-species gene signatures depicted as heatmaps of genes differentially expressed in CD127 - HSAlow versus CD127+ - HSAhi cell populations (Table S7 and Figures 7E and 7F). Human CD127 ELPs and HSAlow murine CLPs shared expression of upstream regulators of NK-ILC differentiation or thymus homing as well as of the above mentioned myeloid trait. Consistent with their B lineage bias, human CD127+ ELPs and HSAhi murine CLPs shared the same core B lineage signature. The gene signature specifying CD127 ELPs and HSAlow CLPs was also enriched in the most immature HSAint murine CLP fraction. Subjecting single CD127 CD7+ ELPs and CD127+CD7+ MLPs to multiplex PCR for expression of a set of 17 genes differentially expressed by the CD127 - HSAlow and CD127+ - HSAhi cellular fractions confirmed that CD127 CD7+ ELPs expressed TOX, NFIL3, CCR7, and CCR9 transcripts at higher levels, and retained a myeloid trait (CSFR2A, IL3RA, IL6R, RUNX2, GFI1). Multiplex PCR also disclosed selective expression of EBF1, PAX5, RAG1, and RAG2 transcripts in CD127+CD7+ ELPs (Figure 7G). These results indicate that, despite divergent lymphoid architectures, the gene regulatory networks governing lymphoid specification have been conserved from mice to humans. DISCUSSION Here, we provide a roadmap of human lymphopoiesis that supports a ‘‘two-family’’ model of lymphoid development (Figure 7H). A sorting scheme of CD34hiCD45RA+ HPCs, which resolves monocytic-DC and lymphoid lineages, identified two founder populations of CD127 and CD127+ ELPs originating from multipotent LMDP intermediates. Although LMDPs were functionally and phenotypically close to previously reported human MLPs (Doulatov et al., 2010) or murine LMPPs (Adolfsson et al., 2005), our results do not support the idea that this population was already lymphoid primed. Instead, they indicate that lymphoid priming coincided with ITGB7 upregulation by LMDPs. Consistent with this view, LMDPs were poorly efficient at generating lymphoid cells in mono-lineage differentiation assays, but displayed high cloning efficiency and generated a majority of

(E) Relative expression of mRNAs coding for EBF1 and PAX5 transcription factors by LMDPs, and the CD127 CD7 and CD127+CD7 cell populations fractionated according to CD9 expression levels. Expression values are normalized relative to HPRT. Results are means + SD of 6 biological replicates from two independent experiments. (F) The CD127 CD7 and CD127+CD7 cell populations fractionated as above were cultured for 96 hours by 25 cell-pools under the B condition, before assessing CD19+ B cell yields. The data are from 1 experiment out of two performed in triplicates. (G–I) In vitro growth and differentiation of CD127 (CD9+CD10+CD20 ) and CD127+ (CD9 CD10 CD20+) proB1 cells. (G) Phenotypic characterization of the CD127 and CD127+ proB1 cells (see also Figures S3C and S3D). Cells were cultured as above by 500-cell pools for 2 weeks under the B condition, with or without IL-7, and analyzed for (H) growth, (I) sIgM expression, and (J) phenotype. Grays bars indicate median values. Black (CD127 ) and empty (CD127+) circles correspond to individual clones (n = 30). Asterisks indicate statistical significance based on the Mann-Whitney test (p < 0.0001***; p < 0.001**; p < 0.01*). (K) List and phenotype of the cell populations subjected to multiplex PCR analysis. Full gating schemes are presented in Figure S3D and S3E. (L and M) Heatmaps showing expression of a set (L) of 48 genes by CD127 and CD127+ proB1-3 cells isolated from the BM of NSG-UCB mice or (M) of 45 genes by CD19+CD9 /+CD20 /+ BLs differentiated from the CD34intCD127+ proB1 cells after 2-week culture period under the B condition. Gene expression levels are normalized relative to GAPDH. Data are means from three or four biological replicates from one experiment.

692 Immunity 47, 680–696, October 17, 2017

A

E

F

G

B

C

H

D

Figure 7. Evolutionary Conservation of Family-Specific ELP Gene Signatures (A) Schematic diagram of in vivo transfer experiments. Sublethally irradiated newborn mice were injected with LMDPs, CD127 , or CD127+ ELPs sorted from the BM of NSG-UCB mice and analyzed 4 weeks later. To improve cell engraftment efficiency and lymphoid reconstitution, mice received weekly intra-peritoneal injections of SCF, IL-2, and IL7 (0.2 mg each). (legend continued on next page)

Immunity 47, 680–696, October 17, 2017 693

mixed lympho-myeloid clones under the lympho-myeloid culture condition. Our assumption that CD127 and CD127+ ELPs define independent lymphoid founder populations relies on multiple lines of evidence. At the phenotypic level, it is based on the observation that the bipartite organization was conserved across development stages since, as for ELPs, downstream lineagespecified B and NK-ILC precursors segregated into distinct CD127 and CD127+ cell subsets. At the molecular level, we show that, irrespective of their multilymphoid or lineage-polarized status, CD127 and CD127+ ELPs harbored family-specific gene signatures that also conditioned their respective lineage restriction modes. Conversely to CD127 ELPs which retained a myeloid trait and could still diverge toward the monocyte-DC lineage, CD127+ ELPs were almost devoid of myeloid potential but intrinsically biased toward the B lineage. Though our data are in line with a recent report in the mouse (Luis et al., 2016), they do not support the assumption that thymus colonizers should correspond to ‘‘lympho-myeloid’’ progenitors. Rather, they indicate that persistence of a myeloid trait is a hallmark of the human CD127 ELPs and HSAlow murine CLPs. At the functional level, the two-family model is further supported by our finding that the T cell potential segregated preferentially with CD127 ELPs, and that CD127 and CD127+ NKIPs or proBs differed as to both growth factor requirements and differentiation potentials. CD127 NKIPs were particularly efficient at generating NK cells in the sole presence of IL-15, while their CD127+ counterparts required Notch1, IL-7, and IL-15 for optimal differentiation into ILC3s. In the same manner, CD127 and CD127+ proB1s differed as to maturation patterns, MZ versus follicular B cell differentiation potential and dependence on IL-7. At the developmental level, we provide evidence that CD127 and CD127+ ELPs emerged independently from LMDPs, underwent divergent modes of lineage restrictions and were subjected to differential Notch1 regulation. Whereas Notch1 drove the emergence of CD127+CD7+ MLPs and prevented their subsequent B lineage diversion, within the CD127 ELP compartment it primarily acted by promoting ETP differentiation. Our finding that ETPs display a CD127 origin confirm our previous report that fetal thymus immigrants lack surface CD127, whose upregulation depends on intrathymic Notch signaling (Haddad et al., 2006). Our results are also consistent with recent reports that CD127 and CD127+ NKIPs are detected in human blood (Lim et al., 2017), tonsils or the intestinal lamina propria (Bernink et al., 2015; Crellin et al., 2010), but at variance with others

claiming that NK cells and ILCs share a single common precursor (Renoux et al., 2015; Scoville et al., 2016). Finally, our results should contribute to a better understanding of the molecular and cellular basis of inherited immune diseases affecting the lymphoid lineage, and help to identify the cells of origin of lymphoid malignancies. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d d

d

d

KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS B Human Sample Collection MICE METHODS DETAILS B Processing of Human Tissues and Cell Separation B Xenotransplantations B Flow Cytometry and Cell Sorting B MsTree-Based Analyses B Cell Cultures and Analysis B Lentiviral Vectors B Fluidigm Analyses B Processing of Microarray Data QUANTIFICATION AND STATISTICAL ANALYSIS B Generation of Human Fingerprints B Microarray Analysis and Bioinformatics DATA AND SOFTWARE AVAILABILITY B Data Resources

SUPPLEMENTAL INFORMATION Supplemental Information includes seven figures and seven tables and can be found with this article online at https://doi.org/10.1016/j.immuni.2017.09.009. AUTHOR CONTRIBUTIONS K.A.H. designed and performed most experiments and wrote the paper; E.V. and E.N. produced the lentiviral vectors; A.J.A. generated the microarray data; T.-P.V.M., M. Dalod, and M. Delord performed the bioinformatics analyses; M.B. and T.W. produced the MStree representation of human hematopoietic architecture; C.B., O.B.-D., and A.C. conducted the mouse studies; N.D. performed the functional characterization of NK-ILCs; E.C. contributed to the fate mapping studies; F.G. provided the fetal BMs; J.L. and V.V. provided the UCB and adult BM samples; J.C.B. and G.S. contributed to the

(B–D) Phenotypic characterization of human cells isolated from the BM, spleen, and thymus of recipient mice. (B) FACS analysis of hu-CD45+ BM cells for lymphoid and myeloid markers. (C) Phenotypic characterization of CD19 CD33 CD3 NK-ILCs from spleen using CD7, CD127, Eomes, and RORC expression. (D) Analysis of hu-CD45+ thymic cells based on CD4 and CD8 expression. Blue arrows show the gating strategy. For each cell population a total of 12 mice from 2 independent experiments were analyzed. Data are representative of at least 3 mice in each group. (E and F) Heatmaps illustrating the differential expression values of selected genes across (E) human and (F) mouse cell populations. FBM: fetal BM; FL: fetal liver. See also Figures S6 and S7 and Tables S3, S4, S5, S6, and S7. The black arrows indicate CCR7 and CCR9. Genes linked to stem-progenitor or myeloid cell development are in purple. (G) Heatmap showing expression of a set of 17 genes defining the CD127 and CD127+ ELP families in single CD127 CD7+ (n = 72) and CD127+CD7+ ELPs (n = 96). Gene-expression levels were normalized relative to GAPDH. Results are means of three or four replicates. Genes linked to T and NK-ILC, multipotent progenitor, and myeloid (MPP and My), or B cell development are indicated. (H) The two-family model of lymphoid development. Plain and dashed arrows indicate predominant and accessory cell type outputs. TLs, T lymphocytes; MZBLs, marginal zone B lymphocytes; FBLs, follicular B lymphocytes.

694 Immunity 47, 680–696, October 17, 2017

experimental design; J.C.G. and B.C. ensured the scientific supervision of the project and wrote the paper.

ACKNOWLEDGMENTS The authors thank Caroline Costa (Plateau de re´trovirologie, Centre international de recherche en infectiologie Lyon, France) for production of the lentiviral vectors. We are grateful to Christelle Doliger, Sophie Duchez, Niclas Setterblad, Guylaine Henry (Plateforme d’Imagerie et de Tri Cellulaire and INSERM UMRS 1160, Institut Universitaire d’He´matologie, Paris France), and Alaa Shawket (CIML, Marseille, France) for technical support. We also thank Sophie Ezine, Michele Goodhardt, and David Garrick for critical discussions. This work was supported by the Agence de la Biome´decine, Universite´ Paris-Diderot, the Comite´ de Paris de la Ligue contre le Cancer, the Institut National du Cancer, the Fondation Ramsay Ge´ne´rale de Sante´, and the Fondation pour la Recherche Me´dicale. Received: December 16, 2016 Revised: April 6, 2017 Accepted: September 17, 2017 Published: October 17, 2017

REFERENCES Adolfsson, J., Ma˚nsson, R., Buza-Vidas, N., Hultquist, A., Liuba, K., Jensen, C.T., Bryder, D., Yang, L., Borge, O.J., Thoren, L.A., et al. (2005). Identification of Flt3+ lympho-myeloid stem cells lacking erythro-megakaryocytic potential a revised road map for adult blood lineage commitment. Cell 121, 295–306. Akashi, K., Traver, D., Miyamoto, T., and Weissman, I.L. (2000). A clonogenic common myeloid progenitor that gives rise to all myeloid lineages. Nature 404, 193–197. Bernink, J.H., Krabbendam, L., Germar, K., de Jong, E., Gronke, K., KofoedNielsen, M., Munneke, J.M., Hazenberg, M.D., Villaudy, J., Buskens, C.J., et al. (2015). Interleukin-12 and -23 Control Plasticity of CD127(+) Group 1 and Group 3 Innate Lymphoid Cells in the Intestinal Lamina Propria. Immunity 43, 146–160. Berthault, C., Ramond, C., Burlen-Defranoux, O., Soubigou, G., Chea, S., Golub, R., Pereira, P., Vieira, P., and Cumano, A. (2017). Asynchronous lineage priming determines commitment to T cell and B cell lineages in fetal liver. Nat. Immunol. 18, 1139–1149. Canque, B., Camus, S., Dalloul, A., Kahn, E., Yagello, M., DezutterDambuyant, C., Schmitt, D., Schmitt, C., and Gluckman, J.C. (2000). Characterization of dendritic cell differentiation pathways from cord blood CD34(+)CD7(+)CD45RA(+) hematopoietic progenitor cells. Blood 96, 3748–3756.

Fritsch, G., Buchinger, P., Printz, D., Fink, F.M., Mann, G., Peters, C., Wagner, T., Adler, A., and Gadner, H. (1993). Rapid discrimination of early CD34+ myeloid progenitors using CD45-RA analysis. Blood 81, 2301–2309. Galy, A., Travis, M., Cen, D., and Chen, B. (1995). Human T, B, natural killer, and dendritic cells arise from a common bone marrow progenitor cell subset. Immunity 3, 459–473. Haddad, R., Guimiot, F., Six, E., Jourquin, F., Setterblad, N., Kahn, E., Yagello, M., Schiffer, C., Andre-Schmutz, I., Cavazzana-Calvo, M., et al. (2006). Dynamics of thymus-colonizing cells during human development. Immunity 24, 217–230. Hao, Q.L., Zhu, J., Price, M.A., Payne, K.J., Barsky, L.W., and Crooks, G.M. (2001). Identification of a novel, human multilymphoid progenitor in cord blood. Blood 97, 3683–3690. Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U., and Speed, T.P. (2003). Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249–264. Kohn, L.A., Hao, Q.L., Sasidharan, R., Parekh, C., Ge, S., Zhu, Y., Mikkola, H.K., and Crooks, G.M. (2012). Lymphoid priming in human bone marrow begins before expression of CD10 with upregulation of L-selectin. Nat. Immunol. 13, 963–971. Kondo, M., Weissman, I.L., and Akashi, K. (1997). Identification of clonogenic common lymphoid progenitors in mouse bone marrow. Cell 91, 661–672. Lim, A.I., Li, Y., Lopez-Lastra, S., Stadhouders, R., Paul, F., Casrouge, A., Serafini, N., Puel, A., Bustamante, J., Surace, L., et al. (2017). Systemic Human ILC Precursors Provide a Substrate for Tissue ILC Differentiation. Cell 168, 1086–1100 e1010. Luis, T.C., Luc, S., Mizukami, T., Boukarabila, H., Thongjuea, S., Woll, P.S., Azzoni, E., Giustacchini, A., Lutteropp, M., Bouriez-Jones, T., et al. (2016). Initial seeding of the embryonic thymus by immune-restricted lympho-myeloid progenitors. Nat. Immunol. 17, 1424–1435. Mansson, R., Hultquist, A., Luc, S., Yang, L., Anderson, K., Kharazi, S., AlHashmi, S., Liuba, K., Thore´n, L., Adolfsson, J., et al. (2007). Molecular evidence for hierarchical transcriptional lineage priming in fetal and adult stem cells and multipotent progenitors. Immunity 26, 407–419. Notta, F., Doulatov, S., Laurenti, E., Poeppl, A., Jurisica, I., and Dick, J.E. (2011). Isolation of single human hematopoietic stem cells capable of longterm multilineage engraftment. Science 333, 218–221. Notta, F., Zandi, S., Takayama, N., Dobson, S., Gan, O.I., Wilson, G., Kaufmann, K.B., McLeod, J., Laurenti, E., Dunant, C.F., et al. (2015). Distinct routes of lineage development reshape the human blood hierarchy across ontogeny. Science 351, aab2116. Parietti, V., Nelson, E., Telliam, G., Le Noir, S., Pla, M., Delord, M., Vanneaux, V., Mohtashami, M., Macintyre, E.A., Gluckman, J.C., et al. (2012). Dynamics of human prothymocytes and xenogeneic thymopoiesis in hematopoietic stem cell-engrafted nonobese diabetic-SCID/IL-2rgnull mice. J. Immunol. 189, 1648–1660.

Crellin, N.K., Trifari, S., Kaplan, C.D., Cupedo, T., and Spits, H. (2010). Human NKp44+IL-22+ cells and LTi-like cells constitute a stable RORC+ lineage distinct from conventional natural killer cells. J. Exp. Med. 207, 281–290.

Pietras, E.M., Reynaud, D., Kang, Y.A., Carlin, D., Calero-Nieto, F.J., Leavitt, A.D., Stuart, J.M., Go¨ttgens, B., and Passegue´, E. (2015). Functionally Distinct Subsets of Lineage-Biased Multipotent Progenitors Control Blood Production in Normal and Regenerative Conditions. Cell Stem Cell 17, 35–46.

Delaney, C., Varnum-Finney, B., Aoyama, K., Brashem-Stein, C., and Bernstein, I.D. (2005). Dose-dependent effects of the Notch ligand Delta1 on ex vivo differentiation and in vivo marrow repopulating ability of cord blood cells. Blood 106, 2693–2699.

Radtke, F., Wilson, A., Stark, G., Bauer, M., van Meerwijk, J., MacDonald, H.R., and Aguet, M. (1999). Deficient T cell fate specification in mice with an induced inactivation of Notch1. Immunity 10, 547–558.

Descatoire, M., Weller, S., Irtan, S., Sarnacki, S., Feuillard, J., Storck, S., Guiochon-Mantel, A., Bouligand, J., Morali, A., Cohen, J., et al. (2014). Identification of a human splenic marginal zone B cell precursor with NOTCH2-dependent differentiation properties. J. Exp. Med. 211, 987–1000. Doulatov, S., Notta, F., Eppert, K., Nguyen, L.T., Ohashi, P.S., and Dick, J.E. (2010). Revised map of the human progenitor hierarchy shows the origin of macrophages and dendritic cells in early lymphoid development. Nat. Immunol. 11, 585–593. Doulatov, S., Notta, F., Laurenti, E., and Dick, J.E. (2012). Hematopoiesis: a human perspective. Cell Stem Cell 10, 120–136.

Ramond, C., Berthault, C., Burlen-Defranoux, O., de Sousa, A.P., Guy-Grand, D., Vieira, P., Pereira, P., and Cumano, A. (2014). Two waves of distinct hematopoietic progenitor cells colonize the fetal thymus. Nat. Immunol. 15, 27–35. €lsson, J., Friberg, D., Soneji, S., Renoux, V.M., Zriwil, A., Peitzsch, C., Michae and Sitnicka, E. (2015). Identification of a Human Natural Killer Cell LineageRestricted Progenitor in Fetal and Adult Tissues. Immunity 43, 394–407. Scoville, S.D., Mundy-Bosse, B.L., Zhang, M.H., Chen, L., Zhang, X., Keller, K.A., Hughes, T., Chen, L., Cheng, S., Bergin, S.M., et al. (2016). A Progenitor Cell Expressing Transcription Factor RORgt Generates All Human Innate Lymphoid Cell Subsets. Immunity 44, 1140–1150.

Immunity 47, 680–696, October 17, 2017 695

Spinelli, L., Carpentier, S., Montan˜ana Sanchis, F., Dalod, M., and Vu Manh, T.P. (2015). BubbleGUM: automatic extraction of phenotype molecular signatures and comprehensive visualization of multiple Gene Set Enrichment Analyses. BMC Genomics 16, 814. Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and

696 Immunity 47, 680–696, October 17, 2017

Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550. Subramanian, A., Kuehn, H., Gould, J., Tamayo, P., and Mesirov, J.P. (2007). GSEA-P: a desktop application for Gene Set Enrichment Analysis. Bioinformatics 23, 3251–3253.

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

FITC anti-human CD1c (clone_ AD5-8E7)

Miltenyi

Cat#130-098-009

PerCPCy5.5 anti-human CD2 (clone_ RPA-2.10)

Biolegend

Cat#300216

FITC anti-human CD4 (clone_ RPA-T4)

Biolegend

Cat#300506

APC anti-human CD5 (clone_ UCHT2)

BD Biosciences

Cat#555355

FITC anti-human CD7 (clone_ 8H8.1)

Beckman Coulter

Cat#A07755

Antibodies

BV510 anti-human CD7 (clone_ M-T701)

BD Biosciences

Cat#563650

PE-CF594 anti-human CD7 (clone_ M-T701)

BD Biosciences

Cat#562541

PE-Cy7 anti-human CD8 (clone_ RPA-T8)

Biolegend

Cat#301012

FITC anti-human CD9 (clone_ HI9a)

Biolegend

Cat#312103

PE-CF594 anti-human CD9 (clone_ M-L13)

BD Biosciences

Cat#563641

BV650 anti-human CD10 (clone_ HI10a)

BD Biosciences

Cat#563734

APC-H7 anti-human CD10 (clone_ HI10a)

BD Biosciences

Cat#655404

PE/Dazzle 594 anti-human CD11b (clone_ ICRF44)

Biolegend

Cat#301347

PE-Cy7 anti-human CD11c (clone_ Bu15)

Biolegend

Cat#337215

PE-CF594 anti-human CD14 (clone_ MOP9)

BD Biosciences

Cat#562335

PE anti-human CD15 (clone_ HI98)

Biolegend

Cat#301905

BUV395 anti-human CD16 (clone_ 3G8)

BD Biosciences

Cat#563785

APC-Cy7 anti-human CD16 (clone_ 3G8)

BD Biosciences

Cat#557758

BV711 anti-human CD19 (clone_ HIB19)

Biolegend

Cat#302245

APC anti-human CD19 (clone_ HIB19)

Biolegend

Cat#302211

APC anti-human CD20 (clone_ 2H7)

Biolegend

Cat#302309

APC anti-human CD22 (clone_ S-HCL-1)

Biolegend

Cat#363505

PE-CF594 anti-human CD24 (clone_ ML5)

BD Biosciences

Cat#562405

APC anti-human CD27 (clone_ M-T271)

Biolegend

Cat#356409

BV510 anti-human CD27 (clone_ M-T271)

Biolegend

Cat#356420

PE anti-human CD31 (clone_ WM59)

Biolegend

Cat#303105

APC anti-human CD32 (clone_ FLI8.26)

BD Biosciences

Cat#559769

PE anti-human CD33 (clone_ WM53)

BD Biosciences

Cat#555450

Pacific Blue anti-human CD34 (clone_ 581)

Biolegend

Cat#343512

BV421 anti-human CD34 (clone_ 581)

Biolegend

Cat#343610

BV510 anti-human CD38 (clone_ HIT2)

BD Biosciences

Cat#563251

APC anti-human CD41a (clone_ HIP8)

BD Biosciences

Cat#559777

PE anti-human CD43 (clone_ 10G7)

Biolegend

Cat#343204

AF700 anti-human CD45 (clone_ HI30)

Biolegend

Cat#304024

PE anti-human CD45RA (clone_ HI100)

BD Biosciences

Cat#555489

APC anti-human CD49d (clone_ 9F10)

Biolegend

Cat#304307

BUV737 anti-human CD49d (clone_ 9F10)

BD Biosciences

Cat#565296

PE-Cy7 anti-human CD49f (clone_ GoH3)

Biolegend

Cat#313621

PE anti-human CD53 (clone_ HI29)

Biolegend

Cat#325406

PE anti-human CD54 (clone_ HA58)

Biolegend

Cat#353105

PE-Cy7 anti-human CD56 (clone_ MEM-188)

Biolegend

Cat#304628

AF700 anti-human CD56 (clone_ 5.1H11)

Biolegend

Cat#362522

BV605 anti-human CD57 (clone_ NK-1)

BD Biosciences

Cat#563895

APC anti-human CD62L (clone_ DREG-56)

Biolegend

Cat#304809 (Continued on next page)

Immunity 47, 680–696.e1–e8, October 17, 2017 e1

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

PE anti-human CD66a,c,e (clone_ ASL32)

Biolegend

Cat#342304

BUV737 anti-human CD69 (clone_ FN50)

BD Biosciences

Cat#564439

PE-Cy5 anti-human CD85j (clone_ GHI/75)

BD Biosciences

Cat#551054

APC anti-human CD94 (clone_ DX22)

Biolegend

Cat#305508

FITC anti-human CD94 (clone_ HP-3D9)

BD Biosciences

Cat#555888

PerCPCy5.5 anti-human CD94 (clone_ HP-3D9)

BD Biosciences

Cat#562361

APC anti-human CD96 (clone_ NK92.39)

Biolegend

Cat#338409

PE anti-human CD106 (clone_ STA)

Biolegend

Cat#305805

FITC anti-human CD107a (clone_ H4A3)

BD Biosciences

Cat#555800

APC anti-human CD114 (clone_ LMM741)

Biolegend

Cat#346107

APC anti-human CD115 (clone_ 9-4D2-1E4)

Biolegend

Cat#347306

APC-Vio770 anti-human CD116 (clone_ REA211)

Miltenyi

Cat#130-100-992

BV605 anti-human CD117 (clone_ 104D2)

Biolegend

Cat#313218

PE anti-human CD122 (clone_ Mik-b2)

BD Biosciences

Cat#18735B

BV786 anti-human CD123 (clone_ 7G3)

BD Biosciences

Cat#564196

PE-Cy5 anti-human CD127 (clone_ A019D5)

Biolegend

Cat#351324

PE anti-human CD127 (clone_ A019D5)

Biolegend

Cat#351304

BV711 anti-human CD132 (clone_ AG184)

BD Biosciences

Cat#563129

PE anti-human CD135 (clone_ 4G8)

BD Biosciences

Cat#558996

PE anti-human CD141 (clone_ AD5-14H12)

Miltenyi

Cat#130-098-841

PerCP-Vio770 anti-human CD158e1/e2 (clone_ REA168)

Miltenyi

Cat#130-104-883

APC anti-human CD161 (clone_ DX12)

BD Biosciences

Cat#561729

PE anti-human CD162 (clone_ KPL-1)

Biolegend

Cat#328806

PE anti-human CD166 (clone_ 3A6)

Biolegend

Cat#343904

PE anti-human CD180 (clone_ MHR73-11)

Biolegend

Cat#312906

PE-Cy5 anti-human CD184 (clone_ 12G5)

Biolegend

Cat#306508

APC anti-human CD186 (clone_ K041E5)

Biolegend

Cat#356005

APC anti-human CD196 (clone_ G034E3)

Biolegend

Cat#353415

BV650 anti-human CD196 (clone_ 11A9)

Biolegend

Cat#353426

PE anti-human CD197 (clone_ GO43H7)

Biolegend

Cat#353203

BV786 anti-human CD197 (clone_ 3D12)

BD Biosciences

Cat#563710

PE anti-human CD199 (clone_ LO53E8)

Biolegend

Cat#358903

PE-Vio770 anti-human CD226 (clone_ DX11)

Miltenyi

Cat#130-099-966

FITC anti-human CD268 (clone_ 11C1)

Biolegend

Cat#316904

PE-Cy7 anti-human CD269 (clone_ 19F2)

Biolegend

Cat#357507

FITC anti-human CD294 (clone_ BM16)

BD Biosciences

Cat#561659

APC anti-human CD294 (clone_ BM16)

Biolegend

Cat#350110

BV605 anti-human CD303 (clone_ AC144)

Biolegend

Cat#354223

BV650 anti-human CD314 (clone_ 1D11)

BD Biosciences

Cat#743559

APC anti-human CD314 (clone_ 1D11)

Biolegend

Cat#320808

BV786 anti-human CD335 (clone_ 9-E2)

BD Biosciences

Cat#563329

PerCP-eFluor 710 anti-human CD336 (clone_ 44.189)

eBioscience

Cat#46-3369-42

BV421 anti-human CD337 (clone_ p30-15)

BD Biosciences

Cat#563385

APC anti-human CCR10 (clone_ 6588-5)

Biolegend

Cat#341505

BUV395 anti-human CCR10 (clone_ 1B5)

BD Biosciences

Cat#565322

APC anti-human CLEC9A (clone_ 8F9)

Biolegend

Cat#353805

PE anti-human EOMES (clone_ WD1928)

eBioscience

Cat#12-4877-42

PE-Cy7 anti-human IgD (clone_ IA6-2)

Biolegend

Cat#348209

FITC anti-human IgL kappa (clone_ MHK-49)

Biolegend

Cat#316506 (Continued on next page)

e2 Immunity 47, 680–696.e1–e8, October 17, 2017

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

PE anti-human IgL lambda (clone_ MHL38)

Biolegend

Cat#316607

BV421 anti-human IgM (clone_ MHM-88)

Biolegend

Cat#314516

PE-Cy7 anti-human ITGB7 (clone_ FIB504)

eBioscience

Cat#25-5867-42

APC-Vio770 anti-human KLRG1 (clone_ REA261)

Miltenyi

Cat#130-103-642

PE anti-human PanKIR2D (clone_ NKVFS1)

Miltenyi

Cat#130-092-688

APC anti-human Perforin (clone_ dG9)

Biolegend

Cat#308111

APC anti-human RORgt (clone_ AFKJS-9)

eBioscience

Cat#17-6988-82

BV421 anti-human T-bet (clone_ 4B10)

Biolegend

Cat#644815

APC anti-human TSLPR (clone_ 1F11/TSLPR)

BD Biosciences

Cat#563153

Adult bone marrow

Unite´ de The´rapie Cellulaire of Hoˆpital Saint-Louis (Paris)

N/A

Fetal bone marrow

Service de Biologie du De´veloppement, Hoˆpital Robert Debre´, Paris

N/A

Umbilical cord blood

Unite´ de The´rapie Cellulaire of Hoˆpital Saint-Louis (Paris)

N/A

Biological Samples

Chemicals, Peptides, and Recombinant Proteins BIT 9500

StemCell Technologies

Cat#09500

Bovine serum albumin

Eurobio

Cat#GAUBSA0165

Dimethyl sulfoxide

Sigma Aldrich

Cat#D4540

Fc receptor-binding inhibitor

eBioscience

Cat#16-9161-73

Fetal calf serum

Life Technologies

Cat#10270106

Ficoll-Hypaque

Pancoll, PAN Biotech GmbH

Cat#P04-60500

Human FLT3L, premium grade

Miltenyi

Cat#130-096-479

Human G-CSF, premium grade

Miltenyi

Cat#130-093-861

Human GM-CSF, premium grade

Miltenyi

Cat#130-093-866

Human IL-2 IS, premium grade

Miltenyi

Cat#130-097-744

Human IL-3, premium grade

Miltenyi

Cat#130-095-069

Human IL-7, premium grade

Miltenyi

Cat#130-095-363

Human IL-15, premium grade

Miltenyi

Cat#130-095-765

Human M-CSF, premium grade

Miltenyi

Cat#130-096-492

Human SCF, premium grade

Miltenyi

Cat#130-096-695

Human TPO, premium grade

Miltenyi

Cat#130-095-752

Ionomycine

Sigma Aldrich

Cat#0634

Phorbol 12-Myristate 13-acetate (PMA)

Sigma Aldrich

Cat#P-8139

RNase-free DNase I

Roche

Cat#1284932

SUPERase-In RNase Inhibitor

Ambion

Cat#AM2694

2-mercaptoethanol

Life Technologies

Cat#31350010

CD34 Microbead kit

Miltenyi Biotech

Cat#130-046-702

CellsDirect one-step qRT-PCR kit

Invitrogen

Cat#11753500

Foxp3 staining buffer kit

eBioscience

Cat#00-5523-00

Luminex screening assay

R&D Systems

N/A

Mouse Cell Depletion Kit

Miltenyi Biotech

Cat#130-104-694

RNeasy Plus Micro Kit

QIAGEN

Cat#74034

Zombie Violet Fixable Viability Kit

Biolegend

Cat#423114

Critical Commercial Assays

Deposited Data Raw and normalized data

(Berthault et al., in revision)

GSE84366

Raw and normalized data

This paper

GEO: GSE90924 (Continued on next page)

Immunity 47, 680–696.e1–e8, October 17, 2017 e3

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

Mouse: OP9

A. Cumano, Institut Pasteur

N/A

Mouse: OP9-DL4

A. Cumano, Institut Pasteur

N/A

Mouse: MS5

DMSZ

N/A

Jackson Laboratory

N/A

Life Technologies

See Table S8

Experimental Models: Cell Lines

Experimental Models: Organisms/Strains Mouse: NSG (NOD.Cg-PrkdcscidIL2RGtm1wjl/SzJ Oligonucleotides Taqman probes Recombinant DNA PRRL lentiviral backbone

Institut Gustave Roussy, France

N/A

shNotch1 (GGAGCATGTGTAACATCAACA)

Delaney et al., 2005

N/A

shNotch2 (GCAGAGGACTCTTCTGCTAAC)

Delaney et al., 2005

N/A

Software and Algorithms FlowJo software (9.9.4)

TreeStar Inc.

N/A

Prism 6

Prism-Graphpad

www.graphpad.com

R

N/A

https://www.r-project.org/

BioNumerics software package v7.5

N/A

http://www.applied-maths.com/ news/bionumerics-version-75-released

Biomark qPCR analysis software

Fluidigm

N/A

Morpheus software

N/A

https://software.broadinstitute.org/ morpheus/

BubbleGUM software

Spinelli et al., 2015

N/A

CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Bruno Canque ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS Human Sample Collection Human fetal (gestational ages 13 to 29 weeks, n = 21) or adult (13-55 years, n = 7) tissues from male or female donors were collected after informed consent according to institutional guidelines and the French bioethical legislation. Umbilical cord blood (UCB) was provided by the Unite´ de The´rapie Cellulaire of Hoˆpital Saint-Louis (Paris). Adult bone marrow (BM) samples were from healthy graft donors (Unite´ de The´rapie Cellulaire, Hoˆpital Saint-Louis, Paris). Fetal BMs were from spontaneously terminated pregnancies (Service de Biologie du De´veloppement, Hoˆpital Robert Debre´, Paris). MICE NOD.Cg-PrkdcscidIL2RGtm1wjl/SzJ (005557) mice known as NOD scid gamma (NSG) mice (Jackson Laboratory, Bar Harbor, MI) were housed in the pathogen-free animal facility of Institut Universitaire d’He´matologie (Paris). Male or female NSG mice (n > 400) were xenografted at 2 months of age. The Ethical Committee at Paris Nord University approved all performed experiments. METHODS DETAILS Processing of Human Tissues and Cell Separation Femurs from fetal donors were sterilely excised and processed in RPMI 1640 medium supplemented with 10 U/ml RNase-free DNase I, before single-cell suspensions were filtered through a cell strainer (70 mm; BD biosciences). Blood or BM mononuclear cells were separated by Ficoll-Hypaque centrifugation (Pancoll, PAN Biotech GmbH) before processing for flow cytometry, cell sorting or CD34+ HPC isolation. CD34+ HPCs were isolated with the CD34 Microbead kit (Miltenyi Biotech; purity > 90%), frozen in heat-inactivated fetal calf serum (FCS) supplemented with 10% DMSO, and stored in liquid nitrogen until use.

e4 Immunity 47, 680–696.e1–e8, October 17, 2017

Xenotransplantations Mice were irradiated with 2.25 Gy 24 h before injection of 2.5 3 105 CD34+ HPCs in the caudal vein, and sacrificed 2 to 8 weeks later. Single-cell suspensions recovered from femurs and tibias were filtered through a cell strainer (70 mm; BD biosciences) and depleted in mouse cells using the Mouse Cell Depletion Kit (Miltenyi) before being processed for Flow Cytometry or cell sorting. For secondary transfer experiments, 3-5 3 104 (CD127- ELPs, LMDPs) to 2 3 105 (CD127+ ELPs) cells sorted from the BM of NSG-UCB mice were injected in the facial vein of sublethally irradiated newborn recipients together with 0.2 mg of SCF, IL-2 and IL-7 (all from Miltenyi Biotech). Mice received SCF, IL-2 and IL7 (0.2 mg each) by intraperitoneal injection weekly and were analyzed four weeks post-transplantation. Recipient mice were considered as positive for cell engraftment when hu-CD45+ cells constituted R 0.01% of total BM mononuclear cells. Flow Cytometry and Cell Sorting Single-cell suspensions were incubated with human Fc receptor-binding inhibitor (Fc Block, eBioscience) before surface staining with anti-human monoclonal antibodies (mAbs). Fluorescence minus one (FMO) controls with isotype controls were used to define positive signals for flow cytometry or cell sorting. Dead cells were excluded with the Zombie Violet Fixable Viability Kit (Biolegend). For labeling, cells were resuspended in PBS, 2% FCS (1 to 5 x107 cells/500 ml) and incubated with the following mAbs: CD45 A700 (Biolegend, clone HI30), CD34 PB (Biolegend, clone 581), CD38 BV510 (BD Bioscience, clone HIT2), CD117 BV605 (Biolegend, clone 104D2), CD45RA PE (BD Bioscience, clone HIT100), CD7 FITC (Beckman Coulter, clone 8H8.1), CD2 PerCPCy5.5 (Biolegend, clone RPA2.10), CD115 APC (Biolegend, clone 9-4D2-1E4), CD116 APC-vio770 (Miltenyi, clone REA211), CD123 BV786 (BD Bioscience, clone 7G3), CD127 PC5 (Biolegend, clone A019D5), ITGB7 PC7 (eBioscience, clone FIB504), CD10 BV650 (BD bioscience, clone HI10a), CD19 BV711 (BD bioscience, clone HIB19), CD24 PE-CF594 (BD bioscience, clone ML5). Flow cytometry analyses and cell sorting were performed with a BD Fortessa Analyzer or a BD FACSAria IIII sorter (BD Biosciences; (purity R 95%). MsTree-Based Analyses Human cells harvested from the BM of NSG-UCB mice were labeled with antibody panels against CD34, CD117, CD38, CD33, CD45RA, ITGB7 for identification of stem/progenitor cells; CD66, CD115, CD116, CD123, CD11b, CD14, CD1c, CD141, CD303 for isolation of the myeloid differentiation pathways; CD127, CD19, CD10, CD24, CD7 and CD2 for visualization of the lymphoid pathways. Human cells distinguished based on lineage affiliation (STEM/PROGENITOR, Gr, MONO-DC, LYMPHOID) were subdivided into CD34hi (MFI R 2500), CD34int (MFI 500-2500) or CD34lo/- (MFI < 500) cell populations, each of which being binary quoted for expression of the other markers. The resulting digital matrix was analyzed with Minimum Spanning Tree with hypothetical nodes algorithm implemented in BioNumerics software package v7.5 to produce a tree-plot representation of the phenotypic network. Hypothetical nodes were produced if the total network length was decreased, and if there were at least 2 neighbors at a distance > 2. Determination of branch length is based on a logarithmic scaling. Cell Cultures and Analysis For assessment of granulo-macrophagic and dendritic cell potentials, CD34+ HPCs were sorted by pools of 50 cells in 96-well U-bottom plates in RPMI medium, 10% FCS, 1% penicillin-streptomycin, 1% glutamin supplemented with SCF (10 ng/ml), FLT3L (10 ng/ml), IL-3 (10 ng/ml), G-CSF (10 ng/ml), GM-CSF (10 ng/ml), M-CSF (10 ng/ml). Cells were allowed to differentiate for 7 days before flow cytometry with a BD Fortessa Analyzer using the following antibodies: CD45 AF700 (Biolegend, clone HI30), CD14 PE-CF594 (BD bioscience, clone MOP9); CLEC9A APC (Biolegend, clone 8F9); CD66a,c,e (PE, Biolegend, clone ASL32); BDCA3/CD141 (PE-Vio770 Miltenyi, clone AD5-14H12); BDCA1/CD1c FITC (Miltenyi, clone AD5-8E7); CD123 BV786 (BD Bioscience, 7G3). Greater than 100 cells were required to define positive wells. For T cell differentiation assays, OP9-DL4 cells (1.5 3 103 cells/well in 200 ml; gift of A. Cumano, Institut Pasteur, Paris) were seeded in Opti-MEM-Glutamax supplemented with 10% FCS, 1% penicillin-streptomycin and 1/1000 b-mercapto-ethanol (Life Technologies) in 96-well flat-bottom plates 24 hour prior to co-culture. Before cell sorting, medium was removed and 200 ml of complete medium supplemented with SCF (10 ng/ml), FLT3L (5 ng/ml), and IL-7 (2 ng/ml) (all from Miltenyi Biotech) was added. Cells were sorted individually or by pools of 10 or 50 directly into the wells, and allowed to differentiate for 2 weeks. On culture day 7, 50 ml of complete medium supplemented with 4X cytokine concentrations were added to each well. At culture day 14, the well content was harvested and labeled with CD45 AF700 (Biolegend, clone HI30), CD34 PB (Biolegend, clone 581), CD7 BV510 (BD Bioscience, clone M-T701), CD5 APC (BD Bioscience, clone UCHT2), CD4 FITC (Biolegend, clone RPA-T4), and CD8 PC7 (Biolegend, clone RPA-T8) Mabs. Flow cytometry was performed as above. Day 14 was chosen as culture endpoint for quantification of the T potential based on preliminary experiments showing that this time point is highly discriminant for identication of T cell precursors. For assessment by RT-PCR of the transcriptional responses to Notch1 ligation, the corresponding cell subsets (0.5 to 1 3 103 cells/well) were sorted directly in 96-well flat-bottom plates pre-coated with 10 mg/ml Delta4ext-IgG or bovine serum albumin, and cultured for 48h with the same medium and cytokine cocktail as above. At that time, living cells (7AAD-) were sorted again and deposited directly by pools of 50 into 96-well PCR plates containing the RT-PCR mastermix (see below). Gene expression levels were normalized to HPRT or GAPDH and expressed as means of 4 replicates. For NK/ILC differentiation assays, OP9 or OP9-DL4 cells were seeded as above in 96-well flat-bottom plates 24 hour prior to culture initiation. Before cell sorting, medium was removed and 200 ml of complete medium supplemented with SCF (10 ng/ml), Immunity 47, 680–696.e1–e8, October 17, 2017 e5

FLT3L (10 ng/ml), IL-7 (10 ng/ml) and IL-15 (10 ng/ml) was added. Cells were sorted directly into the wells individually or by pools of 10 to 50 cells, and allowed to differentiate for 2-3 weeks. On culture days 7 and 14, 50 ml of fresh medium supplemented with 4X cytokine concentrations were added to each well. Alternatively, cells were deposited directly in 96-well U-bottom plates and cultured for 2-3 weeks in Opti-MEM, Glutamax, 10% human AB serum and 1% penicillin-streptomycin and 1/1000 b-mercaptoethanol (Life Technologies) supplemented with the same cytokine cocktail. Alternatively, 96-well U-bottom plates were pre-coated for 2 h at 37 C with 10 mg/ml Delta4ext-IgG or bovine serum albumin before addition of complete medium and cell sorting. Cell surface staining was performed with the following mAbs: CD45 AF700 (Biolegend, clone HI30), CD7 FITC (Beckman Coulter, clone 8H8.1), CD56 PC7 (Biolegend, clone MEM-188), CD14 PE-CF594 (BD Bioscience, clone MOP9), CD94 APC (Biolegend, clone DX22), CD117 BV605 (Biolegend, clone 104D2), CD127 PC5 (Biolegend, clone A019D5), and NKp44/CD336 PE (Biolegend, clone p44-8), or with CRTH2/CD294 FITC or PE (Biolegend, BM16). Intracellular staining was performed using the Foxp3 staining buffer kit (eBioscience) with the following mAbs: RORC APC (eBioscience, clone ASKJS-9), EOMES PE (eBioscience, clone WD1928), T-bet BV421 (Biolegend, clone 4B10). Flow cytometry was performed as above. NK/ILCs were defined as the CD45+CD14-CD56+ cell population. NK/ILC cytotoxic activity was assessed in 4-hour co-cultures with K562 cells (effector/target cell ratio: 1:1). Cell degranulation was quantified based on cell surface acquisition of CD107a. After washing, cells were labeled with CD56 PC7, CD7 BV510, CD94 PerPC (see above), CD107a (Biolegend, clone H4A3), DNAM1 FITC (Biolegend, clone 11A8) and NKG2D APC (Biolegend, clone 1D11) mAbs. For assessment of cytokine production cells were incubated for 3 hours with PMA (25 ng/mL) and ionomycine (1 mg/ml). Multiplex cytokine dosages (IFN-g, IL-13, IL-17, IL-22, TNF-a, LT-a) in culture supernatants were performed using a custom made Luminex screening assay (R&D Systems). For B cell differentiation assays, MS5 cells (1.5 3 104 cells/well) were seeded 24 hours prior to co-culture initiation in 96-well flat bottom plates with RPMI medium supplemented with 10% FCS and 1% penicillin-streptomycin and 1/1000 b-mercapto-ethanol (Life Technologies). Before cell sorting, medium was removed and 200 ml of complete medium supplemented with SCF (10 ng/ml), TPO (10 ng/ml) with or without IL-7 (10 ng/ml). Cells were sorted directly into the wells by pools of 50 to 500 cells, and allowed to differentiate for 2 weeks. On culture days 7, 50 ml of fresh medium + 4X growth factors were added to each well. The following antibodies were used to assess BL differentiation: CD45 AF700 (Biolegend, clone HI30), CD19 APC (Biolegend, clone HIB19), CD9 PE-CF594 (BD biosciences clone M-L13), CD7 FITC (Beckman Coulter, clone 8H8.1), CD10 BV650 (BD bioscience, clone HI10a), CD20 APC (Biolegend, clone 2H7), CD24 BV785 (BD biosciences clone ML5), CD127 PC5 (Biolegend, clone A019D5), CD45RB PE (Biolegend, clone MEM-55), CD27 BV510 (Biolegend, clone M-T271), CD86 BV605 (Biolegend, IT2.2), IgM BV421 (Biolegend, clone MHM-88), IgD PC7 (Biolegend, clone IA6-2), Ig light chain k FITC (Biolegend, clone MHK-49), Ig light chain l PE (Biolegend, clone MHL38). Flow cytometry was performed as above with a BD Fortessa Analyzer. For lympho-myeloid differentiation assay, MS5 cells (1.5 3 104 cells/well) were seeded as above 24 hours prior to co-culture initiation in 96-well flat bottom plates with RPMI medium supplemented with 10% FCS and 1% penicillin-streptomycin and 1/1000 b-mercapto-ethanol (Life Technologies). Before cell sorting, medium was removed and 200 ml of complete medium supplemented with SCF (10 ng/ml), TPO (10 ng/ml), IL2 (10 ng/ml) and IL-7 (10 ng/mL) was added. Cells were sorted individually into the wells, and allowed to differentiate for 2 weeks. On culture days 7 and 14, 50 ml of fresh medium + 4X growth factors were added to each well. The following antibodies were used to assess lympho-myeloid differentiation: CD45 AF700 (Biolegend, clone HI30), CD19 APC (Biolegend, clone HIB19), CD56 PC7 (Biolegend, clone MEM-188) and CD14 PE-CF594 (BD Bioscience, clone MOP9). Flow cytometry was performed as above with a BD Fortessa Analyzer. Lentiviral Vectors The PRRL lentiviral backbone was kindly provided by Hana Raslova (Institut Gustave Roussy, France). Knockdown experiments were performed by using the PRRL-EF1a/EGFP vector with H1 driving hairpin expression. shNotch1 (GGAGCATGTGTAACAT CAACA), shNotch2 (GCAGAGGACTCTTCTGCTAAC) have been described previously (Delaney et al., 2005). shNotch1 and shNotch2 knockdown vectors reduce Notch1/2 levels by R 80% in UCB CD34+ HPCs (data not shown). UCB CD34+cells (purity > 90%) were transduced 6-8 hours at a multiplicity of infection of 20-25 in RPMI medium supplemented with 20% BIT 9500 (StemCell Technologies) and SCF (50 ng/ml), FLT3L (50 ng/ml), TPO (20 ng/ml) and IL-3 (10 ng/ml; all from Milteynyi) before injection to irradiated mice. Fluidigm Analyses Gene expression analyses were performed with the Fluidigm 96.96 Dynamic Array IFC and TaqMan Gene Expression Assays (Life Technologies). Fifty cells were sorted directly into 96-well PCR plate containing 2.5 ml TaqMan specific gene assay mix (Applied Biosystems), 5 ml of CellsDirect 2x Reaction mix, 0,2 ml SuperScriptTM III RT/PlatinumR Taq Mix (Invitrogen, CellsDirect one-step qRT-PCR kit), 1.2 mL TE buffer, and 0.1 ml SUPERase-In RNase Inhibitor (Ambion). Reverse transcription was performed for 15 min at 50 C followed by 2 min at 95 C for RT inactivation. The corresponding cDNAs were then preamplified for 21 cycles at 95 C for 15 s and 60 C for 4 min. Preamplified products were diluted 1:5 in TE buffer and analyzed on a Biomark system (Fluidigm) with the following PCR cycling condition: 95 C for 10 min and 40 cycles at 95 C for 15 s and 60 C for 60 s. Data were analyzed using the Biomark qPCR analysis software (Fluidigm). For gene expression quantification, data were exported as an Excel file and analyzed by the DDCt method. Results were normalized to HPRT or GAPDH expression, and expressed as mean expression levels of 3 to 12 biological replicates. Hierarchical clustering was performed on standardized means of gene expression levels using the Euclidean or the Manhattan distance using the ‘‘pheatmap’’ R package. For single-cell analyses, cell were were sorted individually into 96-well e6 Immunity 47, 680–696.e1–e8, October 17, 2017

PCR using the Index Sorting module of the FACSDiva V8.01 software and processed as above except that 29 cycles of pre-amplification were performed. Results are expressed as mean of 3-4 technical replicates and normalized relative to GAPDH expression. Hierarchical clustering was performed with the ‘‘morpheus’’ software (https://software.broadinstitute.org/morpheus/). References of the TaqMan primers used for the analysis can be communicated on demand. Processing of Microarray Data For microarray analyses of human cells, 1 or 2 biological replicates (3-5000 cells/sample) were used for each population sorted from NSG-UCB or the BM of gestation week 21 fetal donor (GEO: GSE90924). Inasmuch as the analysis focuses on the earliest stages of lymphoid ontogeny, CD7+CD2+ NKIPs were excluded from the gating procedure. Cell isolation, labeling and subsequent sorting were performed as described above in ‘‘Flow cytometry and cell sorting.’’ Cells were sorted in RLT Plus Buffer before total RNA purification using the RNeasy Plus Micro Kit (QIAGEN). RNAs quantification and quality controls were performed using the HT RNA Pico sensitivity LabChip Kit and the Caliper LabChip Microfluidic System (Perkin Elmer). For each sample 1 ng of total RNA was amplified, labeled, and fragmented using GenChip WT Pico (Affymetrix). Each sample was hybridized onto GeneChip human transcriptome array 2.0 (Affymetrix), washed, and stained with the Affymetrix Fluidics Station 450. Array Scanning was performed with the Affymetrix GeneChip Scanner 3000 7G using the Command Console software (Affymetrix) and then analyzed using the Affymetrix APT rma-sketch routine and gene-level analysis library files downloaded from the affymetrix web site. Based on the density plot of the normalized expression values, probes not reaching a minimal value of 3 (log2 scale) across all arrays were removed from the analysis. Transcriptome data from lymphoid precursors isolated from E13 mouse fetal liver (3 biological replicates for each population; GeneChip Mouse Gene 1.0 ST) were generated by A. Cumano (GEO: GSE84366). Based on HSA expression levels, fetal liver CD127+ CLPs were subdivided into HSAlow (CD127+Flk2+a4b7-; T-NK/ILC-polarized) HSAint (CD127+Flk2+a4b7-; immature) and HSAhi (CD127+Flk2+a4b7-; B lineage-polarized) cell subsets. Data from fetal liver Multilineage Progenitors (Lin-AA4.1+CD1172++ IL7R-B220-; GSM791131; GSM791132; GSM791133) were downloaded from the Immgen consortium (Immgen.org). The Mouse Gene 1.0 ST CEL files were processed through Bioconductor in the R statistical environment (version 3.0.2). Quality control of the array hybridization (NUSE plot) and normalization of the raw Affymetrix expression data with RMA (Irizarry et al., 2003) were performed using the Oligo package. Partial Least Square (PLS) correction was then performed to remove the data source effect. Based on the density plot of the normalized expression values, probes not reaching a minimal value of 6 (log2 scale) across all arrays were removed from the analysis. QUANTIFICATION AND STATISTICAL ANALYSIS Generation of Human Fingerprints To decipher the development stage or lineage affiliation of the cell subsets subjected to Fluidigm analysis, we generated cell-type and/or cell-lineage specific fingerprints by analyzing the expression pattern of the genes of interest across public datasets downloaded from GEO (GSE72642, GSE28490, GSE78896, GSE42414, GSE74246) or from ArrayExpress (E-TABM-34), encompassing various cell populations from early progenitors to terminally differentiated cells. Each dataset was processed through Bioconductor in the R (version 3.3.2) statistical environment. Using the limma package, Affymetrix genechips and Illumina expression beadchips were pre-processed and normalized using the Robust Multi-chip Average (RMA) procedure and the quantile normalization method, respectively. Illumina RNA sequencing raw counts data were processed and RPKM values were obtained using the edgeR and annotate packages. Log2 expression values were then submitted to GeneSign (BubbleGUM) in order to extract the cell-specific and/or cell-lineage fingerprints, using the Min(test)/Max(ref)>1.2x method. Moreover, since the identity of a cell population and/or a cell lineage can rely not only on the expression of genes more highly expressed exclusively in that population and/or lineage as compared to other populations, but also on groups of genes that, when expressed together at high level, constitute the fingerprint of the population of interest, we developed a strategy to identify such groups of genes by merging all the data together in such a way that the expression of the genes were comparable, independently of whether they were assessed using gene chips or RNA sequencing. To do so, we focused on the projects that encompassed CD4+ T cell expression data (GSE72642 and E-TABM-34 normalized together with RMA, GSE28490, GSE78896, and GSE74246). We used this common reference population to scale the expression of the genes in each cell samples. For microarray data, for each single gene, only the probe having the highest expression across all arrays was kept in order to remove redundancy. We then merged the data from the different projects and performed a hierarchical clustering on both the samples and the genes in order to detect groups of genes highly expressed in specific groups of cell populations. The result of the analysis for fingerprint generation is summarized in Table S2. The corresponding genes are referred to as ‘‘Computationally defined genes’’ (left column). Experimentally validated upstream regulators of hematopoietic development that did not fall in Mathematically defined fingerprints are referred to as ‘‘Experimentally defined genes’’ (right column). Microarray Analysis and Bioinformatics For principal component analysis of human transcriptome data, PCA of human transcriptome data was performed in R using the ade4 package. Probesets contributing the most to PCs 1 and 2 were selected as having loadings (correlation between a gene and a PC) above 0.5 or below 0.5 based on examination of the loading density plot for each PC. This allowed selection of 1216 probesets as drivers for PC1, and of 539 probesets for PC2. Ingenuity pathway analysis (IPA) was performed on these probesets.

Immunity 47, 680–696.e1–e8, October 17, 2017 e7

The BubbleGUM software (Spinelli et al., 2015) was used to generate cell-specific transcriptomic signatures from a given dataset and to assess their enrichment across cell types from another dataset (http://www.ciml.univmrs.fr/applications/BubbleGUM/index. html). BubbleGUM is composed of two modules; i) GeneSign, which generates transcriptomic signatures and ii) BubbleMap, which automatically assesses the enrichment of input gene signatures between all possible pairs of conditions from independent datasets, based on gene set enrichment analysis (GSEA) methodology (Subramanian et al., 2007; Subramanian et al., 2005). The output is an integrated graphical display. Using GeneSign, gene signatures of each population sorted from NSG-UCB or primary fetal BM were generated for each dataset, i.e., the lists of genes that are more highly expressed in the cell population of interest (test classes) as compared to other cell populations (reference classes). To do this, we used the ‘‘Min(test)/Max(ref)>1x’’ calculation method on the probes that consistently changed expression at least 2 folds across all cell populations of the NSG-UCB or FBM datasets. To run BubbleMap, we used these in-house signatures of cell populations devised in GeneSign as well as GeneSets randomly generated in order to increase statistical power. BubbleMap was run with 1000 geneset-based permutations, and with ‘‘signal to noise’’ as a metric for ranking the genes. The results are displayed as a BubbleMap, where each bubble is a GSEA result and summarizes the information from the corresponding enrichment plot. The color of the Bubble corresponds to the subset from the pairwise comparison in which the signature is enriched. The bubble area is proportional to the GSEA normalized enrichment score (NES). The intensity of the color corresponds to the statistical significance of the enrichment measured by the false discovery rate (FDR) derived by computing the multiple testing-adjusted permutation-based p value using the Benjamini–Yekutieli correction. For generation of relative gene signatures, we used the ‘‘Min(test)/Max(ref)>1x’’ calculation method in GeneSign by selecting the probes that were more highly expressed in the various lymphoid cells considered separately (CD127+CD7-, CD127+CD7+, CD127-CD7- and CD127-CD7+) or grouped into families (CD127+ and CD127-) relative to LMDPs from the NSG-UCB or the FBM dataset. Human cell specific relative signatures then consisted in the top 6000 probes having the highest fold changes according to this analytical approach. The same method was applied to murine HSAlo or HSAhi populations relative to MLPs for the generation of relative murine gene signatures. For generation of consensus gene signatures, probes that were found in the intersection of the Venn diagram obtained by overlapping the equivalent relative signatures from the NSG-UCB and FBM datasets were then subjected to an extra filter that consisted in selecting only the probes that were consistently (in both NSG-UCB and FBM datasets) higher in a population of interest (ex: CD127-CD7-) as compared to another population of interest (ex: CD127-CD7+) using the « Min(test)/Max(ref)>1x » calculation method, in order to extract the final consensual gene signature. Cross-dataset or cross-species transcriptomic comparisons were performed using BubbleMap with the settings defined above or using classical GSEA in the case of the family (CD127- versus CD127+ and HSAlo versus HSAhi) comparisons. In the case of the crossspecies comparisons, the symbols of the genes belonging to the relative species-specific gene signatures were converted into the gene symbols of the orthologous genes in the other species, using Ensembl Biomart, before application of BubbleMap or GSEA. In order to extract the final conserved family gene signatures, the leading edge of each GSEA analysis was extracted. Corresponding leading edges were pooled together to obtain the final CD127+/HSAhi and CD127-/HSAlo conserved gene signatures. Heatmaps were generated using Morpheus from the Broad Institute (https://software.broadinstitute.org/morpheus/). Data are represented as Median unless specified. The sample size for each experiment and the replicate number of experiments are included in the figure legends. For two-group comparisons p values are calculated with the unpaired Student’t test or the Mann-Whitney test as applicable with Prism. DATA AND SOFTWARE AVAILABILITY Data Resources The accession numbers for the microarray data reported in this paper are Gene Expression Omnibus (GEO): GSE90924 and GSE84366.

e8 Immunity 47, 680–696.e1–e8, October 17, 2017