In Vitro Intrathymic Differentiation Kinetics of Human Fetal Liver CD34 +CD38− Progenitors Reveals a Phenotypically Defined Dendritic/T-NK Precursor Split This information is current as of September 29, 2021. Jean Plum, Magda De Smedt, Bruno Verhasselt, Fritz Offner, Tessa Kerre, Dominique Vanhecke, Georges Leclercq and Bart Vandekerckhove J Immunol 1999; 162:60-68; ; http://www.jimmunol.org/content/162/1/60 Downloaded from

References This article cites 36 articles, 25 of which you can access for free at: http://www.jimmunol.org/content/162/1/60.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 29, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. In Vitro Intrathymic Differentiation Kinetics of Human Fetal Liver CD34؉CD38؊ Progenitors Reveals a Phenotypically Defined Dendritic/T-NK Precursor Split1

Jean Plum, Magda De Smedt, Bruno Verhasselt, Fritz Offner, Tessa Kerre, Dominique Vanhecke, Georges Leclercq, and Bart Vandekerckhove2

Human CD34؉CD38؊ hematopoietic precursor cells from fetal liver are able to develop into T, NK, and dendritic cells in a hybrid human/mouse fetal thymic organ culture (FTOC). In this report, we pay particular attention to the early events in differentiation -of these precursor cells. We show that the CD34؉CD38؊ precursor cells, which are CD4؊CD7؊cyCD3؊HLA-DR؊/؉؉ (cy, cy -toplasmatic), differentiate into a CD4؉ population that remained CD7؊cyCD3؊HLA-DR؉؉ and a CD4؊ population that ex ؉ ؊ ؊ pressed CD7 and cyCD3. The CD4 CD7 cyCD3 cells differentiate into phenotypically and functionally mature dendritic cells, Downloaded from but do not differentiate into T or NK cells. The CD4؊CD7؉cyCD3؉ population later differentiates into a CD4؉CD7؉cyCD3؉HLA-DR؊ population, which has no potential to differentiate into dendritic cells but is able to differentiate into NK cells and ␥␦ and ␣␤ T . These findings support the notion that the T/NK split occurs downstream of the NK/dendritic split. The Journal of Immunology, 1999, 162: 60–68.

he thymus is the major site of T differentia- identified by the expression of CD44, high levels of class I MHC, http://www.jimmunol.org/ tion. Besides T cells, dendritic cells and NK cells can low levels of CD4 and heat stable Ag, and absence of CD3 and T complete their differentiation pathway in the thymus mi- CD8 (4, 5). In adoptive transfer experiments, these cells give rise croenvironment. Much of the knowledge of the molecular pro- to B cells as well as T cells, but not to myeloid cells. cesses and involved in thymus differentiation has come from The study of intrathymic developmental pathways in humans studying transgenic and knock-out mice (for review see Ref. relies on phenotypic analysis. Particular seminal has been the dis- 1). For obvious reasons, these studies are not applicable in humans. covery of CD34 as a stage-specific marker identifying immature However, other approaches are possible. Studies of intrathymic hematopoietic stem cells (6, 7). The expression level of CD34 and developmental pathways in the mouse relying on the use of an in its coexpression with differentiation markers as well as the capac- vitro fetal thymus organ culture (FTOC)3 system (2, 3) and intra- ity of these subsets to differentiate in FTOC allowed the delinea- by guest on September 29, 2021 thymic injection of putative progenitor cells into host mice have tion of different stages in human intrathymic development (8). defined the differentiation potential of different subsets of thymo- More recently, on the basis of their ability to reconstitute thymo- cytes. Highly purified candidate progenitor subsets are assayed by poiesis in ectopic human fetal thymus implants in immunodeficient adoptive transfer. The lineages that appear and the time course of C.B17 scid/scid (SCID) mice, the early stages of lymphoid cell appearance of more mature subsets of , such as formation of phenotypically defined subsets of CD34ϩ bone mar- CD4ϩCD8ϩ (double positive, DP) or TCR-␣␤ high, CD4ϩCD8Ϫ, row cells have been defined (9). However, although bone marrow or CD4ϪCD8ϩ (single positive, SP) cells, serve as indicators of lymphoid precursors have been carefully studied, little is known the degree of maturation of the donor population. Using these tech- about the sequential appearance of phenotypic markers during the niques, a very small population of thymocytes has been identified first 2 wk after the entrance of the human precursor cell in the that seems to include early progenitors. In mice, these cells are thymus. Phenotypic changes during the differentiation process have been proposed on the basis of multiparametric flow cytomet- ric studies of the thymus (7), or more recently by assessing the University of Ghent, University Hospital, Department of Clinical Chemistry, Micro- differentiation of sorted thymus subpopulations in human FTOC biology and Immunology, Ghent, Belgium (8, 10) or in murine FTOC (11), but data on the kinetics of the Received for publication May 22, 1998. Accepted for publication September 1, 1998. different lineages are lacking. Using a hybrid human/scid mouse The costs of publication of this article were defrayed in part by the payment of page FTOC system (12), we address in the present study the kinetics of charges. This article must therefore be hereby marked advertisement in accordance the very early steps of differentiation of CD34ϩCD38ϪLinϪ pre- with 18 U.S.C. Section 1734 solely to indicate this fact. cursor cells derived from fetal liver by analysis of the sequential 1 This work was supported by grants from the Fund for Medical Research-Flanders (FGWO-Vlaanderen), Fund for Concerted Research Program (GOA-1993-98), and appearance of surface Ags and the expression of recombination Flanders Interuniversity Institute for Biotechnology. G.L. is a senior research asso- activating gene 1 (RAG-1) and pre-TCR ␣-chain (pT␣). We report ciate, D.V. is a postdoctoral research associate, and B.V. and T.K. are research as- the existence of different phenotypic intermediates that produce sistants on the Fund for Scientific Research-Flanders (Belgium) (FWO-Vlaanderen). lymphocytes, NK cells, and dendritic cells and demonstrate that 2 Address correspondence and reprint requests to Prof. Jean Plum, University of Ghent, University Hospital, Department of Clinical Chemistry, Microbiology and the kinetics of these distinct pathways differ. Immunology, 4 Blok A, De Pintelaan 185, B-9000 Ghent, Belgium. E-mail address: [email protected] Materials and Methods 3 Abbreviations used in this paper: cy, cytoplasmatic; DP, double positive; FTOC, Animals fetal thymus organ culture; PT␣, pre-TCR ␣-chain; RAG, recombination activating gene; SP, single positive; PI, propidium iodide; HPRT, hypoxanthin CB-17 SCID mice were obtained from our own specific pathogen-free phosphoribosyltransferase. breeding facility. For timed pregnancies, females were housed in separate

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00 The Journal of Immunology 61

cages from the males until mating. The appearance of vaginal plug after 2.0 software (Becton Dickinson Immunocytometry Systems). Dead cells overnight mating was labeled as day 0 of pregnancy. The 14- to 15-day were gated out by propidium iodide (PI) exclusion, except for permeabil- pregnant mice were sacrificed to obtain the embryos for preparation of the ized cells, where the dead cells were gated out according to different scatter thymic lobes. properties. In most cases, viable human cells were gated by exclusion of mouse CD45ϩ, mouse class I MHCϩ cells, and PI-positive cells. Abs and reagents Allogeneic MLC assays The mAbs used were rat anti-mouse CD45 cytochrome (CD45, 30F1 1.1; PharMingen, San Diego, CA), biotinylated rat anti-mouse H-2 class I MHC PBL were obtained by centrifugation of blood from healthy adult volun- H-2Dd (PharMingen), and the following mouse anti-human mAbs: anti- teers over Lymphoprep (Nycomed Pharma). PBL were resuspended in Ϫ glycophorin-A (10F7 MN; a kind gift of Dr. L. L. Lanier, DNAX, Palo RPMI 1640 with 10% filtered human AB serum and 5 ϫ 10 5 M 2-ME Alto, CA), CD1a (B-B5; Serotec, Oxford, U.K.), CD38 (OKT10 FITC; (complete MLC medium). MLC were set up in 0.2 ml complete MLC Ortho, Raritan, NJ), CD2 (Leu-5b FITC), CD3 (anti CD3⑀, Leu-4 FITC, medium in round-bottom 96-well trays (Falcon, Becton Dickinson, Lincoln phycoerythrin (PE)), CD4 (Leu-3a FITC or PE), CD7 (Leu-9 FITC), CD8 Park, NJ). The stimulator cells were either sorted human thymocytes from ϩ (Leu2a FITC), CD14 (LeuM3 PE), CD16 (Leu-11a FITC), CD19 (Leu-12 FTOC after 11 days of culture (2 ϫ 103 Ϫ 2 ϫ 104) or sorted CD4 HLA- ϩ PE), CD34 (HPCA-2 FITC or PE), CD45 (HLe-1 FITC), CD56 (clone DR (102–103) putative dendritic cells thereof or PBL (104 Ϫ 5 ϫ 104), NCAM16 PE), anti-TCR-␣␤ (TCR-1 FITC), and anti-TCR-␥␦ (anti-TCR- each given 30 Gy irradiation. The responder cells (5 ϫ 104/well) were PBL ␥␦/l FITC) (all from Becton Dickinson Immunocytometry Systems, Moun- from a different donor. The cultures were incubated in a humidified incu- tain View, CA). Human recombinant IL-15 (CHO-derived) was kindly bator (5% CO2 in air) for 6 days. Each culture was pulsed overnight with provided by Immunex (Seattle, WA). 1 ␮Ci [3H]TdR (Amersham, Amersham, England) per well, before har-

ϩ Ϫ Ϫ vesting of the cultures onto fiberglass filters. Tritiated thymidine incorpo- Preparation of human CD34 CD38 Lin fetal liver cells ration was determined by liquid scintillation counter fluid counting (Mi-

crobeta, Wallac, Turku, Finland). Results are expressed as cpm. Downloaded from Human fetal liver tissues were obtained and used following the guidelines of the medical ethical commission of the University Hospital of Ghent. RNA isolation, reverse transcription, and RT-PCR Human fetal liver cells were isolated by gentle disruption of the tissue in complete medium (Iscove’s modified Dulbecco’s medium/10% FCS, Life A total of 2 ϫ 104–105 cells were harvested for determination of mRNA Technologies, Paisley, Scotland) followed by centrifugation over Lym- encoding human pT␣, RAG-1, or hypoxanthin phosphoribosyltransferase phoprep (Nycomed Pharma, Oslo, Norway). Cells were washed and resus- (HPRT) by reverse transcription-PCR (RT-PCR). The procedure for RNA isolation and RT-PCR was performed with minor modifications as de- pended in 90% FCS/10% DMSO and frozen in liquid N2. After thawing, scribed (15). In brief, total RNA was isolated using Trisol (Life Technol- fetal liver cells were washed and labeled with , CD19, and http://www.jimmunol.org/ FITC-labeled mAbs CD1, CD3, CD4, CD7, CD8, and CD38. Ab-labeled ogies), and further processed following the instructions of the supplier. cells were depleted by immunomagnetic beads. For this purpose, cells were Then, 10 ␮gofEscherichia coli tRNA (Boehringer, Indianapolis, IN) was resuspended in 0.5 ml cold PBS/2% FCS and mixed with 0.5 ml prewashed added as a carrier before chloroform addition and cDNA was prepared (to remove the preservative) sheep anti-mouse Ig-coated Dynabeads (Dy- using Superscript (Life Technologies) according to the specifications of the nal AS, Oslo, Norway) to obtain a ratio of cells to Dynabeads of 1:5. After supplier. An aliquot of each cDNA sample was subsequently amplified in 30 min at 4°C, the suspension was diluted by carefully adding 5 ml twofold serial dilutions by PCR using pT␣-, RAG-1-, and HPRT-specific PBS/2% FCS, and the rosettes of cells with Dynabeads were removed by primers. In each PCR, water, SCID-mouse fetal thymus cDNA, and human placing the tube on a magnetic particle concentrator (Dynal AS). The su- and mouse genomic DNA were included as negative controls. All primer pernatant that contained the unlabeled and weakly stained cells was re- pairs amplified human cDNA only. PCR amplification using the different moved and centrifuged (500 ϫ g for 6 min). These cells were labeled with primers was performed in 50 ␮l volumes with1UofTaq polymerase ϩ Ϫ Ϫ CD34-PE and CD34 CD38 Lin sorted. The sorted cells were transferred (Perkin-Elmer, Emeryville, CA) using a 96-well thermocycler (Omnigene, by guest on September 29, 2021 to murine thymic lobes by the hanging drop method (13). Hybaid Teddington, U.K.) under the following conditions: pT␣, 35 cycles of 30 s at 94°C, 30 s at 65°C, and 1 min at 72°C; RAG-1 and HPRT, 30 FTOC cycles of 30 s at 94°C, 30 s at 55°C, and 1 min at 72°C. The sequences of the primers were as follows: for pT␣,5Ј-TCC AGC CCT ACC CAC AGG Thymic lobes were prepared from fetal day 14–15 SCID mice. Hanging TG-3Ј (sense primer) and 5Ј-TAG AAG CCT CTC CTG ACA GAT GCA ␮ drops were prepared in Terasaki plates by adding in each well 25 lof T-3Ј (anti-sense primer); for RAG-1, 5Ј-CCA AAT TGC AGA CAT CTC complete medium containing 10,000 or less sorted human Ј Ј ϩ Ϫ Ϫ AAC-3 (sense primer) and 5 -CAA CAT CTG CCT TCA CAT CGA CD34 CD38 Lin fetal liver cells and one thymic lobe. The plates were TCC-3Ј (anti-sense primer); for HPRT, 5Ј-AAT TAT GGA CAG GAC immediately inverted to form hanging drops and incubated in an humidi- TGA ACG T-3Ј (sense primer) and 5Ј-TCA AAT CCA ACA AAG TCT fied incubator for 48 h (7.5% CO2 in air, 37°C). GGC TTA-3Ј (anti-sense primer). HPRT was visualized on ethidium bro- After incubation, the lobes were removed from the hanging drop, mide stained gels. PCR products of pT␣ and RAG-1 were Southern blotted washed, and placed on the surface of a nuclepore filter in organ culture in to Nytran paper (Schleicher & Schuell, Keene, NH). The membrane was an incubator. At different time points, thymocytes were recovered by me- hybridized at 60°C using a 32P-end-labeled internal oligonucleotide, for chanical disruption of the thymic lobes with a small tissue grinder. Thy- pT␣ 5Ј-CCT TCT CTG GCC CCA CCA ATC A-3Ј and for RAG-1 5Ј- mocytes were stained with eosin and counted with a hematocytometer. At CAT CCT GTG ACA TCT GCA AC-3Ј. least 80% of the cells were viable. Flow cytometry Results Detailed phenotypic analysis of early steps of differentiation Before labeling, cells were suspended in PBS/1% BSA/0.1% NaN3, and the Fc receptor of the mouse thymocytes was blocked by preincubation for 15 We have already reported that during hanging drop approximately min with anti-Fc␥RII/III mAb (clone 2.4.G2) (14) to avoid aspecific bind- 1000 to 2000 progenitors cells, i.e., 1⁄5th to 1⁄10th of the input num- ing of Abs by the murine thymocytes. Subsequently, the cells were stained ber of cells, enter the thymic lobe (12) and that afterward the with a panel of mAb, as indicated. The biotinylated Abs were revealed with Ͻ second step streptavidin Tri-Color (Caltag, Burlingame, CA). number of human cells initially decreases ( 500 cells/lobe) (data All mAb against human Ags and second step reagents were checked for not shown) before expanding. As a consequence of this transient negative staining on SCID thymocytes after blocking with 2.4G2 mAb. decrease, the number and the frequency of human cells during the Isotype controls were also included in most staining series and were found first days of the culture is low. When we electronically gated out to be negative. the dead as well as the murine cells by eliminating the cells that For intracellular staining of CD3 (cytoplasmatic (cy) CD3), cells were first labeled with mAbs against mouse and human surface Ags, and after stained with PI and anti-mouse CD45-Tri-Color, a population re- two washes the cells were permeabilized with Ortho Permeafix (Ortho, mained that did not stain for human CD45 (Fig. 1). When anti- Raritan, NJ) according to the guidelines of the manufacturer. After two mouse class I MHC biotin-streptavidin-Tri-Color was added, we washes, cells were labeled with CD3-FITC. The cells were analyzed on a obtained a neater picture as nearly all cells that were gated out FACScan (Becton Dickinson Immunocytometry Systems) with an argon- Ϫ ion laser tuned at 488 nm. Forward scattering, orthogonal scattering, electronically now stain for human CD45. Most likely, this CD45 and three fluorescence signals were determined on 10,000 cells and stored population, which can be stained by anti-mouse class I MHC, rep- in listmode data files. Data acquisition and analysis was done with Lysis resents a population of mouse epithelial cells. By gating out all 62 KINETIC ANALYSIS OF HUMAN DIFFERENTIATION Downloaded from http://www.jimmunol.org/

FIGURE 1. Flow cytometric analysis of cells isolated from hybrid hu- man-mouse FTOC: delineation of human cells. Flow cytometric analysis of cells obtained from hybrid human/mouse FTOC after four days of culture. Representative fluorescence staining profiles of PI (dead cells) and anti- mouse-CD45-cychrome (mouse leukocytes) (left, top panel) or PI and anti- mouse-CD45-cychrome and anti-mouse-Class I-biotin plus streptavidin-

Tricolor (all mouse cells) (right, top panel) vs forward scatter (FSC) by guest on September 29, 2021 profiles are shown. Gating on R1 allows the elimination of dead cells and stained mouse cells. The FSC and side scatter (SSC) profiles of the cells that were gated on R1 are shown in the middle panel. Gating on R2 allows the elimination of cells with a high side scatter. In the bottom panel, the cell surface staining of CD34 and human CD45 is shown on cells that were gated for R1 and R2. Note that nearly all cells stain for human CD45 in right dot plot of the bottom panel. The percentages of cells within the quadrants are indicated in the upper right quadrant of each plot. nonhuman cells, we could carefully trace the human cells and FIGURE 2. Flow cytometric analysis of human cells from hybrid hu- study their phenotypic changes at the early stage of differentiation, man-mouse FTOC: phenotypic changes during first weeks of culture. A, Cell surface expression of CD34 vs cytoplasmic CD3 staining of even when the total cell number is very minute. Other systems, ϩ Ϫ such as injection of mismatched HLA donor cells into transplanted CD34 CD38 progenitor cells before and at different time points of cul- ture in FTOC. Fetal liver cells were depleted and highly purified (Ͼ99, 5%) SCID-human mice, do not offer the possibility to clearly gate out ϩ Ϫ Ϫ for CD34 38 Lin progenitor cells and put in FTOC as described in Ma- donor cells, because the Ag density of the HLA-marker is rela- terials and Methods. At different time points of FTOC, cell suspensions tively weak during the first stages of differentiation (16). There- were made, and the cells were stained with anti-mouse-CD45 cychrome fore, the hybrid human/mouse FTOC system used in our study and anti-mouse-class I MHC-biotin plus streptavidin-Tri-Color and PI, and gives us a unique opportunity to address the earliest steps of gated as shown in Fig. 1. A shows Cy CD3 staining vs CD34 in bivariate differentiation. dot plot analysis and histogram of lineage markers, as indicated, including Fig. 2A shows the cell surface expression of CD34, CD7, and CD7 and CD38 on the sorted cells that were used to start the culture (row CD38 and the cytoplasmic expression of CD3 (cyCD3) in function 1). Cy CD3 staining vs CD34 in bivariate dot plot analysis (row 2) and of duration of the culture. According the purification strategy, the histograms of CD7 expression (row 3) and CD38 expression (row 4) after sorted precursor cells were CD7- and CD38-negative (Fig. 2A)at 3, 7, and 11 days of hybrid human/mouse FTOC. The percentages of cells within quadrants are indicated in the upper right quadrant of each plot. The the start of the culture. After three days of culture, most of the ϩ numbers in the histograms are the percentage of positive cells. B, Sorted CD34 cells already expressed CD7 and CD38, 70 and 85%, re- CD34ϩCD38Ϫ progenitor cells cultured for 5 and 7 days in FTOC. Dot spectively. Very rapidly some cells coexpress CD34 cells with plots of CD4-PE vs cytoplasmic CD3-FITC (top panel), CD7-PE vs cyCD3 upon culture. However, an important fraction of the cells at cyCD3-FITC (middle panel), and CD4-PE vs CD7-FITC (bottom panel) that time has down-regulated CD34 and does not express cyCD3. are shown, after exclusion by negative gating for mouse and dead cells as After 4 to 5 days of culture, ϳ1000 to 2000 human cells can be described in Fig. 1. The percentages of cells within the quadrants of interest found in one reconstituted thymic lobe. An important fraction of are indicated in the upper left quadrant. The Journal of Immunology 63

the human cells expresses CD4 (30–50%, 300–900 cells/lobe) but does not express cyCD3 nor CD7 (Fig. 2B). At that time, cyCD3ϩ cells are CD4ϪCD7ϩ (Fig. 2B) and the expression of CD4 and CD7 is mutually exclusive (data not shown). After 7 days of cul- ture, the number of human cells has increased to 3000–4000/lobe, part of the cyCD3ϩ cells starts to express CD4, but almost all the cyCD3Ϫ cells remain CD7Ϫ. From this we conclude that there is a first wave of two cell populations that are either CD4ϪcyCD3ϩCD7ϩ or CD4ϩcyCD3ϪCD7Ϫ followed by the generation of a second wave of cells that are CD4ϩcyCD3ϩCD7ϩ. As discussed below, further phenotypic analysis allowed us to de- lineate the differentiation of four different cell lineages: dendritic cells, NK cells, TCR-␥␦, and TCR-␣␤ cells. Analysis of the development of dendritic cells After 3 days of culture, some human cells had down-regulated CD34 without cyCD3 expression (Fig. 2A). We analyzed whether these cells might represent dendritic cells. Because Sotzik et al. (17) have shown that human thymic dendritic cells are large cells Downloaded from expressing high levels of CD4 and class I and II MHC but no CD7, we analyzed these markers at different time points of FTOC. As seen in Fig. 3A, at day 4 the CD4ϩ cells express HLA-DR. Later, the intensity of both CD4 and HLA-DR increases, so that after 11 days a small but distinct population of HLA-DRϩϩCD4ϩ cells can

be discerned. The number of these cells is ϳ200–500 cells/lobe at http://www.jimmunol.org/ day 4–6 and increases only two- to threefold until day 12; after- ward, the number remains constant. As the total number of human thymocytes increases 10- to 20-fold (20,000–40,000 human cells/ lobe), the frequency of these putative dendritic cells decreases in function of length of culture (Fig. 3A). After 12 days of FTOC, it is possible to define cells that highly express CD4, HLA-DR, and HLA-class I are negative for CD7 and cyCD3 (Fig. 3B) and have a high forward and side scatter (Fig. 3B, R6). We further corrob- orated that these cells were functional dendritic cells by assessing by guest on September 29, 2021 their ability to stimulate allogeneic T cells in an MLC. The max- imal extent of proliferation of responder lymphocytes was similar to that seen with PBL as stimulators, but 80-fold fewer stimulator cells were required for peak stimulation (Fig. 3C). To test at what stage of development the potential to generate dendritic cells is still retained, we tested the capacity of sorted human CD4ϩ HLA-DRϪ cells and CD4Ϫ cells from day 10 FTOC to develop into dendritic cells in a second human/mouse hybrid FTOC. In a representative experiment shown in Fig. 4, we purified FIGURE 3. Comparative phenotypic analysis of total human cells and these populations (which were Ͼ96% pure upon reanalysis). It is gated human dendritic cells and functional capacity of sorted human den- clear that dendritic cells and CD4ϩ precursor cells were generated dritic cells from hybrid human-mouse FTOC. A, Phenotypic analysis of starting from the CD4Ϫ population, whereas the CD4ϩHLA-DRϪ ϩ Ϫ developing human dendritic cells in FTOC. Sorted CD34 CD38 progen- population has lost its potential to develop into dendritic cells. itor cells were cultured in FTOC for 4, 6, 9, or 11 days. Dot plots of Finally, the sorted CD4ϩ HLA-DRϩ cells were end-stage cells that CD4-PE vs HLA-DR-FITC are shown on viable human cells, obtained by retained the dendritic phenotype and were not able to give rise to negative gating for mouse and dead cells (PI) as described in Materials and Methods. B, Comparative flow cytometric analysis of the expression of T cells (data not shown). These results demonstrate that the lym- phoid/dendritic common precursor is CD4Ϫ, whereas the CD4ϩ CD4 vs HLA-DR, class I MHC, CD7, or cyCD3 on total human cells and Ϫ gated human cells with large side scatter. Top panel (left) shows the FSC (HLA-DR ) cell populations contain no dendritic precursors. vs SSC of human viable cells after 10 days of FTOC. R6 defines a gate on Taken together, these results indicate that during differentiation, cells with high side scatter distinguishing human cells with dendritic mor- dendritic cells down-regulate CD34, express HLA-DR and CD4 phology. SSC on all human cells and on R6 gated cells is shown in top and do not express cyCD3 and CD7. panel (right). The dot plots show the pattern of CD4 against HLA-DR, class I MHC, CD7, or cyCD3 of total thymocytes (left) or of R6 gated Analysis of the development of NK cells thymocytes (right). Cells were stained with CD4-PE and either HLA-DR- As only a small number of NK cells develop spontaneously in FITC, class I MHC-FITC, CD7-FITC, or cyCD3-FITC. The rectangular FTOC (Fig. 5), we took advantage of the property of IL-15 as a region in the plot defines a population with characteristics of dendritic cells. C, Stimulation of allogeneic PBL with the putative dendritic cells strong stimulator of NK cell development (18, 19) to study whether different thymic subsets were able to generate NK cells. (closed squares) isolated as defined by the rectangular region of the dot ϩ Ϫ Ϫ plot. The stimulation achieved is compared with that given by PBL (open Starting with CD34 Lin CD38 cells, we were not able to gen- circles) or total thymocytes (open squares). Results are presented as cpm/ erate NK cells with IL-15 in cell suspension (data not shown). well. A second experiment gave similar results. However, when put into FTOC and cultured in the presence of 64 KINETIC ANALYSIS OF HUMAN THYMOCYTE DIFFERENTIATION

FIGURE 4. Dendritic cell developmental potential of sorted total, CD4ϩHLA-DRϪ, and CD4Ϫ human cells from 10 day FTOC. Cell sus- Downloaded from pensions were prepared, and the cells were stained as described in Fig. 1. Each sorted population was cultured in a second FTOC. After an incuba- ϩ Ϫ FIGURE 5. NK cell developmental potential of CD34 CD38 fetal tion of 12 days, cell suspensions were made. Flow cytometric analysis was liver progenitor cells in FTOC. The progenitor cells were highly purified performed after staining the cells with mAbs against human CD4-PE and (Ͼ99, 5%) as described in Materials and Methods. The cells were cultured HLA-DR-FITC. It is obvious that the CD4Ϫ cell population generated cells for 5, 12, and 20 days without (control, left panel) and with 20 ng/ml IL-15 with features of dendritic cells, whereas the CD4ϩHLA-DRϪ cells did not (right panel). Afterward, single-cell suspensions were prepared and ana- http://www.jimmunol.org/ generate this population. Cells with characteristics of dendritic cells, based lyzed for the indicated surface Ags. The quadrants were set to contain at on expression of CD4 and HLA-DR are highlighted with a rectangular least 98% of the isotype-matched control mAbs in the lower left quadrant. region in the dot plots. This figure shows a representative experiment of three independent exper- iments. The percentages of cells within quadrants of interest are indicated.

IL-15, large numbers of CD3ϪCD56ϩ cells were generated. Fig. 5 Analysis of the development of T cells: purified CD34ϩCD38Ϫ shows that the addition of 20 ng/ml IL-15 to FTOC resulted in a fetal liver cells develop into DP thymocytes in FTOC and dramatic increase in CD56ϩ cells. These cells had the appearance subsequently complete their differentiation of large granular lymphocytes (data not shown). Further pheno- We have previously shown that 28 days of FTOC started with by guest on September 29, 2021 typic analysis showed that part of the cells coexpresses CD16 (Fig. 10,000 CD34ϩ fetal liver cells resulted in the appearance of 5). In addition, we performed sorting experiments to obtain CD4ϩCD8ϪCD3ϩCD1ϩ human thymocytes (12). Here we show ϩ Ϫ Ϫ CD4 HLA-DR and CD4 subsets from day 10 FTOC, and both that after 40 days of culture CD1ϪTCR-␣␤ϩ and CD1ϪTCR␥␦ϩ subsets were able to generate NK cells when reintroduced in thymocytes were generated by CD34ϩCD38Ϫ fetal liver cells and FTOC in the presence of IL-15 (Fig. 6). However, the starting that the thymocytes had partially down-regulated CD45RO and CD34ϩCD38ϪLinϪ population, as well as CD4ϩHLA-DRϪ cells acquired CD45RA (Fig. 7). At that time point of the culture, from day 10 FTOC, were not able to generate NK cells in suspen- 50,000–150,000 human cells are found per lobe. In addition, part ϩ Ϫ ϩ ϩ Ϫ sion cultures with IL-15. In contrast, CD4Ϫ cells from day 10 of the CD3 CD1 TCR-␣␤ cells were either CD4 CD8 or Ϫ ϩ FTOC generated NK cells when cultured in suspension in the pres- CD4 CD8 (data not shown). We have shown previously that Ϫ ϩ ϩ Ϫ ϩ ϩ ence of IL-15 (data not shown). This apparent paradox might be CD1 CD3 CD4 or CD1 CD3 CD8 cells have acquired func- ascribed to a difference in sensitivity of the assays and that a small tional capacity and are considered as complete differentiated thy- contamination of the CD4Ϫ cells in the CD4ϩ cells are able to mocytes (20). In addition, we have also demonstrated that Ϫ ␥␦ϩ generate NK cells in FTOC, whereas this does not occur in sus- CD1 TCR- cells have functional characteristics of mature T cells (21). pension. In this respect, the purity of the sorted cell populations was at least 96%, and the input of cells was 5000. Therefore, the Analysis of mRNA expression for RAG-1 and pT␣ absolute number of cells with the NK phenotype CD56ϩCD3Ϫ To assess the immaturity of the starting population, we analyzed recovered per lobe, which is ϳ8,000–12,000 after 10 days of cul- the presence of RAG-1 and pT␣ mRNA transcripts. Expression of ture, does not rule out the possibility of outgrowth of contaminat- RAG genes is absolutely required for rearrangement of TCR ing cells. Alternatively, the fact that with IL-15 CD4ϩ cells can Ϫ genes, but is also present in developing B cells as it is required for generate NK cells in FTOC and not in suspension, whereas CD4 the rearrangement of Igs. pT␣ is restricted to cells of the cells generate NK cells in FOTC and in suspension, can be inter- ␣ Ϫ lineage (22, 23). The expression of RAG-1 and pT mRNA has preted as the CD4 populations contain more mature NK precur- already been addressed in the analysis of human thymic subsets sor cells (day 10). Taken together, these results indicate that up- (24). In Fig. 8, it is shown that pT␣ is already present in the Ϫ regulation of the CD4 marker on HLA-DR precursor cells is CD34ϩCD38ϪLinϪ starting population and that the expression is accompanied with a loss of differentiation potential for dendritic higher after 3 days of culture in FTOC. In contrast, although cells, but that NK and T potential is still preserved. Therefore, RAG-1 expression is sometimes weakly present in the starting these results suggest that the branching point of dendritic cells is population, it is not detectable in day ϩ3 samples. RAG-1 mRNA earlier in the developmental scheme than the T/NK cell branching expression could be clearly demonstrated in human thymocytes point. after 11 and 14 days of FTOC. The Journal of Immunology 65

FIGURE 7. Flow cytometry analysis of human cells after 40 days of culture in hybrid human-mouse FTOC. Cultures were initiated with CD34ϩCD38Ϫ progenitor cells from fetal liver. After 40 days of culture, cell suspensions were prepared, and cells were stained with PI and anti- mouse CD45cychrome and anti-mouse MHC class I MHC-biotin plus FIGURE 6. NK cell developmental potential of sorted CD4ϩHLA-DRϪ streptavidin-Tri-Color and PE or FITC directly conjugated mAbs as indi- Downloaded from Ϫ and CD4 human cells from 10 day FTOC. Cell suspensions were pre- cated. This allowed us to gate out dead cells and mouse cells and to analyze pared and the cells were stained with anti-CD4-PE and anti-HLA-DR- for the coordinate expression of CD4 vs CD8, CD4 vs CD3 (top panel), FITC and PI and anti-mouse CD45-cychrome and anti-mouse MHC class CD1 vs TCR-␥␦, CD1 vs TCR-␣␤, and CD45RA vs CD45RO (bottom I MHC-biotin plus streptavidin-Tri-Color. This allowed us to gate out dead panel) on human cells. Representative results of one of three experiments ϩ Ϫ Ϫ cells and mouse cells and select for CD4 HLA-DR and CD4 human is shown. cells. Each sorted population was cultured in a second FTOC in the pres-

ence of 20 ng/ml IL-15. After an incubation of 12 days, cell suspensions http://www.jimmunol.org/ were prepared. Flow cytometric analysis was performed after staining the have used has potential shortcomings. First, we have introduced hu- cells with mAbs against human CD2-PE and CD56-PE. It is obvious that man pluripotent fetal liver cells with myeloid differentiation capacity both populations generated NK cells based on expression of CD56 and (data not shown) in FTOC. This might not represent the default path- absence of CD2. This experiment is representative for two independent way, because it remains an open question whether an uncommitted experiments. The percentages of cells within the quadrant of interest is pluripotent cell seeds the thymus or a more committed cell. Secondly, indicated. although the differentiation capacity of the fetal liver cells in the xe- nogeneic environment is impressive, we cannot exclude that critical factors or ligands that are species-specific are lacking and might in-

Discussion fluence the default differentiation pathway. Nevertheless, our experi- by guest on September 29, 2021 In this study, we show that in a hybrid human/mouse FTOC, imma- ence with cord blood-derived hematopoietic precursors shows that the ture human stem cells from fetal liver spontaneously develop into at differentiation obtained in in vitro xenogeneic FTOC (25) is very sim- least four different lineages: dendritic cells, NK cells, TCR-␥␦, and ilar to that obtained by precursor cells that were injected in vivo in TCR-␣␤ T cells. This system allows a detailed kinetic analysis of the human thymic tissue transplanted under the kidney capsule of SCID/ different cell populations, because a cohort of CD34ϩCD38Ϫ stem human reconstituted mice (our unpublished observations). Therefore, cells starts to develop resembling a synchronized culture. This is in we consider that our approach gives us an unique opportunity to study contrast to analysis of different subsets in a human thymus, where all the T developmental pathway of prenatal fetal liver precursor cells. If different cell populations are already present. the human/mouse FTOC recapitulates the normal differentiation of Moreover, in this study we are able to address the differentiation of the human thymus, our results imply that immature CD34ϩCD38Ϫ pluripotent hemapoietic precursor cells from fetal liver, whereas most human stem cells develop very rapidly into dendritic cells, that the data on human T cell differentiation are founded on studies with post- differentiation toward NK cells takes somewhat longer in time, and, natal thymus. However, the human/mouse FTOC method that we finally, that TCR-␥␦ cells are generated before TCR-␣␤ cells.

FIGURE 8. Analysis of pT␣ and RAG-1 mRNA expression in fetal liver CD34ϩCD38Ϫ stem cells and in FTOC cells. Expression of mRNA was determined by semiquantitative RT-PCR analysis, in which the level of HPRT was used as a con- trol for the amount of mRNA isolated. Five times more cDNA was used to determine RAG-1 and pT␣ compared with HPRT. HPRT is visualized on ethidium bromide stained gels, and PCR products of RAG-1 and pT␣ are visualized by Southern blots. The four bands on a gel represent a twofold dilution of one sample. 66 KINETIC ANALYSIS OF HUMAN THYMOCYTE DIFFERENTIATION

Dendritic cells comprise a family of professional APC respon- stimulus for differentiation of thymic precursors toward phenotyp- sible for the activation of primary T cell responses. Large numbers ically and functionally mature NK cells. of functional dendritic/Langerhans cells can be generated by treat- The precise role of NK development in the thymus remains ing dendritic cell precursors from several sources with granulo- unknown. Because the main physiological maturation of NK cells cyte-macrophage-CSF alone or in combination with other growth occurs in the bone marrow, it is possible that NK maturation in the factors. More recently, it was also shown that ligation of CD40 thymus is not relevant. In particular, the fact that only CD94 is induces the proliferation and differentiation of CD34 hematopoi- up-regulated and other HLA class-I specific inhibitory receptors etic stem cells into dendritic cells (26). Sotzik et al. (17) have are not expressed (data not shown, and Ref. 30) may support the shown that dendritic cells in the human thymus can be identified as hypothesis that this maturation is incomplete and not physiological. CD4highCD44ϩCD1ϪCD2ϪCD3ϪCD7ϪCD8ϪCD34Ϫ class I The first T cells that develop in the mouse thymus are TCR-␥␦ MHChigh class II MHChigh cells with efficient stimulator capacity cells. Indirect evidence that this is also the case for the human thymus Ϫ for allogeneic T cells. We show here that dendritic cells are gen- is based on studies on T cell acute lymphoblastic leukemia. As CD3 erated spontaneously within the thymus microenvironment from T cell acute lymphoblastic leukemias have been found with rear- CD34ϩCD38Ϫ progenitor cells within 8–10 days. These cells are ranged TCR-␦ genes and germline TCR-␤ and TCR-␥ genes (31), it the earliest cells to be generated and to be functionally mature. The is probable that TCR-␦ genes rearrange early during T cell develop- functional relevance of this population within the thymus remains ment before other TCR genes. In addition, RT-PCR analysis per- formed by Ktorza et al. (32) showed expression of full-length TCR-␦ to be established. There is no doubt from mouse studies that den- ϩ Ϫ dritic cells are essential for the negative selection of T cells, transcripts in the CD34 CD1 TN thymocyte subset, before expres- ␤ Downloaded from whereas epithelial cells mediate positive selection (27). However, sion of full length TCR- transcripts. In the earliest phase of fetal at present it is not clear whether dendritic cells in the thymus thymus colonization at 8.2 wk of gestation, by immunohistochemistry ␤ originate from precursor cells that differentiate in the thymus or 5% of the thymocytes stained intracellularly with -F1 mAb, sug- ␤ ␥␦ whether mature dendritic cells migrate from the periphery to the gesting expression of TCR- -chain, whereas only a few TCR- cells ␦ thymus. This is of interest because in the latter situation, Ags that were found. The time of first appearance of TCR- is at 9.5 wk (33). are loaded in the periphery can participate in the negative selection In our study we do not have direct proof that differentiating thymo-

␥␦ http://www.jimmunol.org/ process, whereas otherwise these Ags have to be produced locally cytes start earlier in their attempt to rearrange the TCR- genes than TCR-␣␤ genes. We consistently found that cells with cell surface or have to reach the thymus. Our findings show that dendritic cells expression of TCR-␥␦ are present before we can detect T cells with are able to differentiate in the thymus from precursor cells and that TCR-␣␤. However, as it is known that TCR-␣␤ cells first have to the kinetics are faster than for the other cell types. This allows that express TCR-␤-chain with pT␣ before completion of the differenti- negative selection can be mediated very rapidly when the T cells ation and expression of TCR-␣␤, the difference in time of appearance are at a very immature stage and supports the assumption that a of TCR-␥␦ and TCR-␣␤ cells might rather reflect a different kinetic linked development of T cells and dendritic cells may ensure that in maturation of the two cell types. developing T cells are negatively selected predominantly by self- Our finding that RAG-1 and pT␣ mRNA is present in the start-

Ags presented on newly formed thymic dendritic cells. This idea ϩ Ϫ by guest on September 29, 2021 ing CD34 CD38 fetal liver cell population differs from the fail- was already proposed for the murine thymus, as it was shown that ϩ Ϫ ure to detect expression of pT␣ and RAG-1 in the CD34 CD38 after irradiation bone marrow cells injected intrathymically restore cell population of cord blood, as reported by Blom et al. (34), and simultaneously T cells and dendritic cells (28, 29). ϩ Ϫ in the CD34 CD38 population of fetal liver, as reported by Ja- Another cell type that develops spontaneously albeit in low num- leco et al. (19). In contrast, Ramiro et al. were able to demonstrate bers in human/mouse FTOC are NK cells (Fig. 5). It is obvious that pT␣ mRNA in CD34ϩ progenitors in fetal liver, cord blood, and most NK cells are generated outside the thymus as nude mice have a adult bone marrow (24). This could be due to differences in the normal NK activity. However, the thymus provides a microenviron- sensitivity of the test or in the strategies used to purify the pro- ment that allows stem cells to differentiate toward the NK cell path- ϩ Ϫ genitor cells. In this regard we used FITC-labeled Abs directed way. This is shown by the fact that the CD34 Lin population from against CD38 instead of PE-labeled CD38 reagents. As PE stains fetal liver did not differentiate toward NK cells in cell suspension the cells brighter, it is possible that cells with a weak expression of when IL-15 was added, even in the presence of stem cell factor and CD38 were present in our starting population. It is still a matter of IL-7 (data not shown), whereas in FTOC IL-15 resulted in a massive ϩ Ϫ debate whether an uncommitted pluripotent precursor cell seeds differentiation toward NK cells. Similarly, CD4 and CD4 precur- the thymus and becomes T cell committed afterward or whether sor cells from day 10 FTOC also had the capacity to develop into NK commitment occurs extrathymically and committed cells enter the cells in FTOC in presence of IL-15. However, only day 10 FTOC thymus to complete their differentiation. Definitive proof awaits Ϫ ϩ Ϫ CD4 precursor cells, in contrast to the CD34 CD38 start popula- study of cells at the clonal level, but our findings suggest that tion, were able to develop into NK cells in the presence of IL-15 in CD34ϩCD38Ϫ fetal liver express RAG-1 and pT␣ mRNA remi- suspension cultures. This means that the stromal environment and/or niscent of committed T cells or that uncommitted cells could start cytokines present in the thymus can allow the differentiation of the the transcription of a variety of genes. Single-cell PCR will inform IL-15 refractory CD34 precursor cell to a cell population that re- us on the frequency of these cells. sponds vigorously to IL-15 by cell proliferation and differentiation toward NK cells. It is clear from our studies that IL-15 is a potent Proposed model for development of different lineages in human promoter for the growth and differentiation of human NK cells. We thymus also found that part of the NK cells, generated in IL-15-supple- The kinetic study examining the acquisition of different lineage mented FTOC, coexpress CD16 (Fig. 5) and CD94 (data not markers in our in vitro system, with particular emphasis on the shown). We have already shown previously the role of IL-15 as earliest steps in differentiation, revealed the sequential expression NK cell promoter in mouse fetal thymus (18). Our present obser- of different surface Ags, shown in Fig. 9. vations point to an essential role of IL-15 in human NK develop- The starting population of fetal liver CD34ϩCD38Ϫ progenitor ment and are in agreement with the observation of Mingari et al. cells does not express the surface Ags CD1, CD2, CD3, CD4, (30), who have been able to show that IL-15 provides an efficient CD5, or CD7. Analysis of this population revealed that the cells The Journal of Immunology 67 Downloaded from

FIGURE 9. Proposed model for development of different cell lineages in human thymus. Dotted line indicate two possible pathways. http://www.jimmunol.org/ express CD45 but do not express CD45RO and a minority express is an early T/NK marker in the thymus. The next Ag that is expressed CD45RA (Ͻ10%). The progenitor cells express predominantly c- is CD4. It is obvious, as shown in Fig. 3, that CD4 expression occurs (CD117) (Ͼ85%) and do not express CD10 and Thy-1 (CD90) in the absence of cyCD3 in these cells that will become dendritic cells. (data not shown). CD3 is also not demonstrable in the cytoplasm. HLA-DR, which is already present from the start of the culture, is HLA-DR was already present on the starting population (Ͼ90%). up-regulated during differentiation toward dendritic cells. We could Sorting for HLA-DRϪ cells did not result in a different differen- never demonstrate CD5 on these cells, whereas a small fraction of tiation pattern, although the differentiating HLA-DRϪ cells remain these cells expresses CD1a (data not shown). Therefore, we can con- by guest on September 29, 2021 HLA-DRϪ (our unpublished observations), with the exception of clude that during differentiation toward dendritic cells, progenitor the dendritic cells that become HLA-DRϩ. The presence of pT␣ in cells down-regulate CD34, possibly transiently express low levels of the starting population is consistent with the concept of T-com- CD7, up-regulate HLA-DR and CD4, and never express cyCD3. mitment of stem cells before the entry in the thymus or could The next CD7ϩ population that can be detected expresses cyCD3, indicate the presence of cells engaged in the extrathymic T cell which is indicative for cells engaged in the T/NK cell pathway. At that pathway. At present, there is no direct evidence to conclude that time, CD4 is not expressed. When CD4 is expressed, HLA-DR is one single progenitor cell in the thymus can differentiate into T, down-regulated (data not shown). Resort experiments have shown NK, and dendritic cells or whether there are different progenitor that these cells have lost the potential to generate dendritic cells, cells. Our observation that RAG-1 mRNA expression is absent whereas they can generate both NK and T cells. after 3 days of FTOC is compatible with the view that the precur- These data indicate that in humans the T/NK split occurs down- sor cells with RAG-1 mRNA transcripts are engaged in stream of the NK/dendritic cell split. These data are compatible with differentiation and are not able to develop in FTOC. the finding in human bipotential T/NK cells that have been demon- It is clear that the first Ags that appear on the cell membrane of strated by Sanchez et al. (11). Others have shown that the human differentiating CD34ϩCD38Ϫ LinϪ are CD38 and CD7 (Fig. 2). thymus contain CD34ϩ intrathymic precursors that develop into T These markers are not lineage specific and are considered as activa- and dendritic cells (35, 36). In our study, there is no evidence that tion markers. However, CD7 is not expressed by all cells, and we find dendritic T cells are tightly linked to T cell development beyond a evidence that strong expression of CD7 never occurs during differ- common precursor that is multipotent. This is in contrast to the ob- entiation of dendritic cells. This assumption is based on the fact that servation in the mouse system, where a common T/dendritic cell pro- cyCD3 is very rapidly expressed within 3 days of culture on part of genitor was found (28). This could be due to species differences, dif- the CD34ϩ cells and that this cyCD3 expression is almost exclusively ference between adult and fetal precursors, or to our model where an found on cells with a strong CD7 expression. Cells with a lower CD7 early precursor with myeloid potential is introduced. However, post- expression might be precursor cells of cyCD3-positive cells. How- natal thymocytes sorted according to phenotypic differentiation mark- ever, we cannot exclude that part of these cells express CD7 tran- ers and tested for their differentiation capacity in FTOC are consistent siently and will never express cyCD3 and are cells at the early stages with the notion that dendritic cells branch off earlier that the T/NK of the dendritic cell development. In this respect, we have been able bipotential cell progenitors. The CD34ϩCD1Ϫ subset, which is con- to generate dendritic cells in FTOC starting from sorted sidered as the most immature thymic progenitor, gives rise to HLA- CD34ϩCD7lowϩ cells from fetal liver (data not shown). It is generally DRϩCD4ϩCyCD3Ϫ, putative dendritic cells, T cells, and in the pres- considered that expression of cyCD3 is an early marker for both NK ence of IL-15 to cells with the CD56ϩ NK cell phenotype (data not and T cell commitment. As CD7 expression correlates with cyCD3 shown). On the contrary, CD34ϩCD1ϩ thymic progenitors, consid- expression, CD7, in absence of membrane CD3 and CD4 expression, ered to be at a later step of differentiation than the CD34ϩCD1Ϫ cells, 68 KINETIC ANALYSIS OF HUMAN THYMOCYTE DIFFERENTIATION develop into T cells but not into cells with the dendritic phenotype. In differentiation of fetal liver CD34(ϩ) lineage(Ϫ) cells in human thymic organ the presence of IL-15, the CD34ϩCD1ϩ cells develop into CD56ϩ culture. J. Exp. Med. 180:123. 11. Sanchez, M. J., M. O. Muench, M. G. Roncarolo, L. L. Lanier, and J. H. Phillips. cells (data not shown). 1994. Identification of a common T/ progenitor in human fetal From day 6–7 of FTOC, CD5 expression appears on human thymus. J. Exp. Med. 180:569. thymocytes (data not shown). On day 6, besides the dendritic cell 12. Plum, J., M. De Smedt, M. P. Defresne, G. Leclercq, and B. Vandekerckhove. 1994. Human CD34ϩ fetal liver stem cells differentiate to T cells in a mouse population that expresses CD4 and lacks cyCD3, we find a cyCD3- thymic microenvironment. Blood 84:1587. positive population that is still CD4-negative but already expresses 13. De Smedt, M., G. Leclercq, B. Vandekerckhove, and J. Plum. 1994. Human fetal CD5. This indicates that the expression of CD5 either precedes liver cells differentiate into thymocytes in chimeric mouse fetal thymus organ culture. Adv. Exp. Med. Biol. 355:27. CD4 expression or that CD4-negative cells acquire CD5 and could 14. Unkeless, J. 1979. Characterization of a monoclonal against mouse be precursors of NK cells, whereas T-committed cells first express macrophage and lymphocyte Fc receptors. J. Exp. Med. 150:580. CD4 and later acquire CD5. However, as NK cells can be gener- 15. Vanhecke, D., B. Verhasselt, M. Desmedt, B. Depaepe, G. Leclercq, J. Plum, and B. Vandekerckhove. 1997. MHC class II molecules are required for initiation of ated from CD4-positive cells, we assume that at least part of the positive selection but not during terminal differentiation of human CD4 single NK cells are derived from a precursor with dim CD4 expression. positive thymocytes. J. Immunol. 158:3730. 16. Galy, A. H. M., D. Cen, M. Travis, S. Chen, and B. P. Chen. 1995. Delineation Based on the findings of Sanchez et al. (11), who showed that ϩ low of T-progenitor cell activity within the CD34 compartment of adult bone mar- CD5 cells are able to generate NK cells, we propose also an row. Blood 85:2770. intermediate CD5-positive stage. 17. Sotzik, F., Y. Rosenberg, A. W. Boyd, M. Honeyman, D. Metcalf, R. Scollay, Finally, for the development of TCR-␣␤ and TCR-␥␦ cells, our L. Wu, and K. Shortman. 1994. Assessment of CD4 expression by early T pre- model is based on our ability to show the intracellular presence of the cursor cells and by dendritic cells in the human thymus. J. Immunol. 152:3370. 18. Leclercq, G., V. Debacker, M. de Smedt, and J. Plum. 1996. Differential effects Downloaded from ␤-chain of the TCR after 10 days of culture. At that time, the cells did of interleukin-15 and interleukin-2 on differentiation of bipotential T/natural not stain for TCR-␣␤ on the cell surface. The first TCR that could be killer progenitor cells. J. Exp. Med. 184:325. demonstrated on the cell surface was TCR-␥␦ on day 11–14, whereas 19. Jaleco, A. C., B. Blom, P. Res, K. Weijer, L. L. Lanier, J. H. Phillips, and H. Spits. 1997. Fetal Liver contains committed NK progenitors, but is not a site ␣␤ ϩ on day 17–20 TCR- cells appeared (data not shown and Ref. 12). for development of CD34 cells into T cells. J. Immunol. 159:694. Ϫ Ϫ Finally, we find mature TCR-␥␦ CD1 and TCR-␣␤ CD1 20. Vanhecke, D., G. Leclercq, J. Plum, and B. Vandekerckhove. 1995. Character- ϩ ϩ CD4ϩCD45RAϩCD45ROϪ cells after 14–16 and 35–40 days, re- ization of distinct stages during the differentiation of human CD69 CD3 thy- mocytes and identification of thymic emigrants. J. Immunol. 155:1862.

␣␤ http://www.jimmunol.org/ spectively. This indicates that it takes longer for TCR- cells to 21. Offner, F., K. Vanbeneden, V. Debacker, D. Vanhecke, B. Vandekerckhove, achieve full maturation end-stages as compared with TCR-␥␦ cells. J. Plum, and G. Leclercq. 1997. Phenotypic and functional maturation of TCR ␥␦ This is could be due to a more elaborated selection process. These cells in the human thymus. J. Immunol. 158:4634. 22. Saint Ruf, C., K. Ungewiss, M. Groettrup, L. Bruno, H. J. Fehling, and results show that our experimental model is able to support full mat- H. von Boehmer. 1994. Analysis and expression of a cloned pre-T cell receptor uration of the different cell populations. Our model was based on gene. Science 266:1208. kinetic studies compiling at least 25 different cultures, which gave a 23. Fehling, H. J., A. Krotkova, C. Saint Ruf, and H. von Boehmer. 1995. Crucial ␣ ␣␤ ␥␦ consistent picture. Moreover, several experiments starting from role of the pre-T-cell receptor gene in development of but not T cells. ϩ Ϫ Nature 375:795. CD34 CD38 populations from postnatal thymus gave the same ki- 24. Ramiro, A. R. C., C. Trigueros, C. Marquez, J. L. San Millan, and M. Toribio. netics and sequential order of phenotypes (data not shown). These 1996. Regulation of pre-T cell receptor (pT␣-TCR␤) during findings strongly support that the hybrid mouse-human FTOC is a human thymic development. J. Exp. Med. 184:519. by guest on September 29, 2021 25. Verhasselt, B., M. De Smedt, R. Verhelst, E. Naessens, and J. Plum. 1998. Ret- reliable model system for human thymocyte differentiation. rovirally transduced CD34ϩϩ human cord blood cells generate T cells expressing high levels of the retroviral encoded green fluorescent marked in vitro. Blood 91:431. Acknowledgments 26. Flores-Romo, L., P. Bjo¨rk, V. Duvert, C. van Kooten, S. Sealand, and ϩ We thank Elaine Thomas (Immunex) and L. L. Lanier (DNAX) for pro- J. Banchereau. 1997. CD40 ligation on human cord blood CD34 hematopoietic progenitors induces their proliferation and differentiation into functional dendritic viding reagents, Christian De Boever for artwork, and Evelien Naessens cells. J. Exp. Med. 185:341. and Veronique Debacker for excellent technical assistance. 27. Brocker, T., M. Riedinger, and K. Karjalainen. 1997. Targeted expression of major histocompatibility complex (MHC) class II molecules demonstrates that dendritic cells can induce negative but not positive selection of thymocytes in References vivo. J. Exp. Med. 185:541. 1. Fisher, A., and B. Malissen. 1998. Natural and engineered disorders of lympho- 28. Ardavin, C., L. Wu, C. L. Li, and K. Shortman. 1993. Thymic dendritic cells and cyte development. Science 280:237. T cells develop simultaneously in the thymus from a common precursor popu- 2. Jenkinson, E. J., P. Jhittay, R. Kingston, and J. J. T. Owen. 1985. Studies of the lation. Nature 362:761. role of the thymic environment in the induction of tolerance to MHC antigens. 29. Ardavin, C. 1997. Thymic dendritic cells. Immunol. Today 18:350. Transplantation 39:331. 30. Mingari, M. C., C. Vitale, C. Cantoni, M. Ponte, F. Schiavetti, S. Bertone, 3. Kingston, R., E. J. Jenkinson, and J. J. T. Owen. 1985. A single stem cell can A. Moretta, and L. Moretta. 1997. Interleukin-15-induced maturation of human nat- recolonize an embryonic thymus, producing phenotypically distinct T-cell pop- ural killer cells from early thymic precursors: selective expression of CD94/NKG2-A ulations. Nature 317:811. as the only class I-specific inhibitory receptor. Eur. J. Immunol. 27:1374. 4. Wu, L., R. Scollay, M. Egerton, M. Pearse, G. J. Spangrude, and K. Shortman. 31. van Dongen, J. J., W. M. Comans Bitter, I. L. Wolvers Tettero, and J. Borst. 1991. The earliest T-lineage precursor cells in the adult murine thymus expresses 1990. Development of human T lymphocytes and their thymus-dependency. Thy- low levels of CD4. Nature 349:71. mus 16:207. 5. Wu, L., M. Antica, G. R. Johnson, R. Scollay, and K. Shortman. 1991. Devel- 32. Ktorza, S., S. Sarun, F. Rieux Laucat, J. P. de Villartay, P. Debre, and C. Schmitt. opmental potential of the earliest precursor cells from the adult mouse thymus. 1995. CD34-positive early human thymocytes: T cell receptor and cytokine re- J. Exp. Med. 174:1617. ceptor gene expression. Eur. J. Immunol. 25:2471. 6. Civin, C., L. Strauss, C. Brovall, M. Fackler, J. Schwartz, and J. Shaper. 1984. Antigenic analysis of hematopoiesis. III. A hematopoietic progenitor cell surface 33. Haynes, B. F., and C. S. Heinly. 1995. Early human T cell development: analysis antigen defined by a monoclonal antibody raised against KG-1a cells. J. Immunol. of the human thymus at the time of initial entry of hematopoietic stem cells into the fetal thymic microenvironment. J. Exp. Med. 181:1445. 133:157. ϩ 7. Terstappen, L. W., S. Huang, and L. J. Picker. 1992. Flow cytometric assessment 34. Blom, B., P. Res, E. Noteboom, K. Weijer, and H. Spits. 1997. Prethymic CD34 of human T-cell differentiation in thymus and bone marrow. Blood 79:666. progenitors capable of developing into T cells are not committed to the T cell 8. Galy, A., S. Verma, A. Barcena, and H. Spits. 1993. Precursors of CD3ϩ CD4ϩ lineage. J. Immunol. 158:3571. CD8ϩ cells in the human thymus are defined by expression of CD34: delineation 35. Res, P., E. Martinez Caceres, A. Cristina Jaleco, F. Staal, E. Noteboom, ϩ of early events in human thymic development. J. Exp. Med. 178:391. K. Weijer, and H. Spits. 1996. CD34 CD38 dim cells in the human thymus can 9. Galy, A., M. Travis, D. Cen, and B. Chen. 1995. Human T, B, natural killer, and differentiate into T, natural killer, and dendritic cells but are distinct from pluri- dendritic cells arise from a common bone marrow progenitor cell subset. Immu- potent stem cells. Blood 87:5196. nity 3:459. 36. Marquez, C., C. Trigueros, E. Fernandez, and M. L. Toribio. 1995. The Devel- 10. Barcena, A., A. H. M. Galy, J. Punnonen, M. O. Muench, D. Schols, opment of T and non-T lineages from CD34ϩ human thymic precursors can be M. G. Roncarolo, J. E. De Vries, and H. Spits. 1994. Lymphoid and myeloid traced by the differential expression of CD44. J. Exp. Med. 181:475.