Downloaded From

Natural Killer Cell Development and Function Precede βα T Cell Differentiation in Mouse Fetal Thymic Ontogeny

This information is current as James R. Carlyle, Alison M. Michie, Sarah K. Cho and Juan of September 24, 2021. Carlos Zúñiga-Pflücker J Immunol 1998; 160:744-753; ; http://www.jimmunol.org/content/160/2/744 Downloaded from

References This article cites 64 articles, 37 of which you can access for free at: http://www.jimmunol.org/content/160/2/744.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• 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

*average by guest on September 24, 2021

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 © 1998 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Natural Killer Cell Development and Function Precede ␣␤ T Cell Differentiation in Mouse Fetal Thymic Ontogeny1

James R. Carlyle, Alison M. Michie, Sarah K. Cho, and Juan Carlos Zu´n˜iga-Pﬂu¨cker2

Natural killer (NK) cells mediate MHC-unrestricted cytolysis of virus-infected cells and tumor cells. In the adult mouse, NK cells are bone marrow-derived lymphocytes that mature predominantly in extrathymic locations but have also been suggested to share a common intrathymic progenitor with T lymphocytes. However, mature NK cells are thought to be absent in mouse fetal ontogeny. We report the existence of thymocytes with a mature NK cell phenotype (NK1.1؉/CD117؊) as early as day 13 of gestation, approximately 3 days before the appearance of CD4؉/CD8؉ cells in T lymphocyte development. These mature fetal thymic NK cells express genes associated with NK cell effector function and, when freshly isolated, display MHC-unrestricted cytolytic activity in vitro. Moreover, the capacity of fetal thymic NK cells for sustained growth both in vitro and in vivo, in addition to their close phenotypic resemblance to early precursor thymocytes, confounds previous assessments of NK lineage Downloaded from precursor function. Thus, mature NK cells may have been inadvertently included in previous attempts to identify multipotent and bipotent precursor thymocytes. These results provide the ﬁrst evidence of functional NK lymphocytes in mouse fetal ontogeny and demonstrate that NK cell maturation precedes ␣␤ T cell development in the fetal thymus. The Journal of Immunology, 1998, 160: 744–753.

igration of fetal liver-derived hemopoietic precursors pression of the winged-helix nude (whn) gene (10, 11). However, http://www.jimmunol.org/ to the early fetal thymic rudiment occurs by day 12 of the development of NK cells is thymus independent, and these mouse gestation. The earliest hemopoietic cells to col- cells are present at normal to elevated levels in athymic nude mice M 3 onize the thymus, thymic lymphoid progenitors (TLPs), are mul- as well as in mice defective in the ability to rearrange genes en- tipotent lymphoid-committed precursors capable of giving rise to coding the Ag receptors (severe combined immune deficiency the B, T, thymic dendritic, and NK cell lineages but lack signifi- (SCID) and RAG-deficient mice) (12–15). Nonetheless, in addi- cant potential for myeloid and other hemopoietic cells (1, 2). Soon tion to peripheral sites for NK lymphopoiesis, NK cell develop- after exposure to the thymic microenvironment, these precursors ment can occur within the thymus, and these cells have been sug- rapidly lose B lymphoid potential and become committed to the gested to share a common thymic progenitor with T lymphocytes by guest on September 24, 2021 T/NK lineages (3). Subsequently, a wave of thymocyte differen- within the TLP population (1, 2, 16–19). Previous studies pro- tiation is established in the fetal thymus, marked by the ordered vided evidence for, but failed to define, a proposed bipotent thymic appearance of various developmental stages along the pathway to progenitor for T and NK cells (16, 20–22); additionally, these mature T cells (1, 2, 4). In fetal thymic ontogeny, the development reports did not outline the earliest stages of NK cell development of a defined subset of ␥␦ T cells precedes that of conventional ␣␤ in fetal ontogeny. Instead, these investigations demonstrated that T cells (5). However, the ordered appearance of NK cells remains various purified populations of thymocytes can give rise to either unknown within the context of thymocyte development, and func- T or NK cells under different in vitro or in vivo conditions (16, tional NK cells are thought to be absent in mouse fetal ontogeny. 20–22). Importantly, none of these studies addressed the possibil- NK cells are responsible for mounting MHC-unrestricted cytol- ity that NK cells derived from populations of precursor thymo- ysis of virus-infected and transformed cells (6–9). The develop- cytes, upon i.v. injection or in vitro culture, represented an out- ment of mature peripheral ␣␤ and ␥␦ T cells is thymus-dependent growth of an already existent subset of mature NK cells. and does not occur efficiently in mice that fail to develop a proper To investigate these questions, we analyzed day 13 to 15 mouse thymic epithelium (nu/nu, or nude mice) due to a defect in ex- fetal thymocytes, which contain precursors for all lymphoid lineages, but no mature ␣␤ T or B lymphocytes, and have an overall CD3Ϫ/CD4Ϫ/CD8Ϫ triple-negative (TN) phenotype (1, 2, 4). Re- Department of Immunology, University of Toronto, Toronto, Ontario, Canada cently, we reported the identification of a novel population of thy- Received for publication August 27, 1997. Accepted for publication October mocytes that serve as common committed progenitors for T and 3, 1997. NK lymphocytes (3). These precursors display both the NK1.1 The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in molecule (NKR-P1C) of NK cells (23, 24) as well as the CD117 accordance with 18 U.S.C. Section 1734 solely to indicate this fact. (c-kit) molecule characteristic of hemopoietic precursors (17, 25– ϩ 1 This work was supported by grants from the Medical Research Council of Can- 27). We now report the identification of NK1.1 thymocytes with ada (MRC) and the National Cancer Institute of Canada (to J.C.Z.-P.), and by an a mature NK cell phenotype, lacking expression of CD117. These MRC studentship (to J.R.C) and an MRC scholarship (to J.C.Z.-P.). fetal thymic NK cells are evident as early as day 13 to 14 of 2 Address correspondence and reprint requests to Dr. J.C. Zuń˜iga-Pflu¨cker, Depart- ment of Immunology, University of Toronto, Medical Sciences Building, Toronto, gestation, express genes associated with NK cell effector function, Ontario, M5S 1A8 Canada. E-mail address: jc.zuniga.pfl[email protected] and display MHC-unrestricted cytolytic activity directly ex vivo. 3 Abbreviations used in this paper: TLP, thymic lymphoid progenitor; FTLP, fetal Strikingly, despite the above functional characteristics and lack of TLP; FTNK: fetal thymic NK1.1ϩ progenitor; FTOC: fetal thymic organ culture; CD117 surface expression, these mature NK cells possess a com- HSA: heat stable antigen; RAG: recombination-activating gene; SCF: stem cell posite phenotype similar to early precursor thymocytes, including factor; TN, triple negative; FT, fetal thymocytes; Sw, Swiss.NIH; dGuo, ϩ Ϫ ϩ deoxyguanosine. a CD44 (Pgp-1), CD25 (IL-2R␣), CD16/32 (Fc␥RIII/II),

ϩ Ϫ ϩ ϩ Ϫ CD24low (HSA), CD90 / (Thy-1), CD122 (IL-2R␤), CD2 / GCA TAC AGG; NKR-P1 (genes 2, 34, 40) 5Ј, AAG GTA CAC ATT (LFA-2), CD5Ϫ (Ly-1), and TN phenotype. Many of these char- GCC AGA CAT; NKR-P1A (gene 2) 3Ј, GTA GAC ATG GCT CAG TGA Ј acteristics have been previously used in an attempt to define bi- TTG; NKR-P1B (gene 34) 3 , GGA CAG GGG AGA GAT GGA GAT; NKR-P1C (gene 40, NK1.1) 3Ј, GAG TCA ACG AAT GGA AAG GAA; potent T/NK precursor cells as well as early T lineage precursors Ly-49A 5Ј, TTC TGC TTC CTT CTT CTG GTA; Ly-49A 3Ј, TGT GTT (16, 22). We now demonstrate directly that fetal thymic NK cells CAA GGC AAG TTT AGA; Ly-49C 5Ј, AGA CCA GAA AAA CGC are capable of substantial growth, both in vitro and in vivo, con- CAA CTT; Ly-49C 3Ј, TTC ACT GTT CCA TCT GTC CTG; perforin 5Ј, tributing significantly to the NK cell reconstitution potential of ATG TTC CCC AGT CGT GAG AGG; perforin 3Ј, AAG GTG GAG TGG AGG TTT TTG; CD95L (Fas-ligand) 5Ј, AAG AGA ACA GGA GAA precursor-phenotype thymocytes upon adoptive transfer. Thus, NK ATG GTG; CD95L 3Ј, AGA TTT GTG TTG TGG TCC TTC. cell progenitor activity reported in previous studies that failed to exclude fetal thymic NK cells may have stemmed from an out- 51Cr-release cell-mediated cytotoxicity assay growth of pre-existing mature NK cells, in addition to bona fide Single-cell suspensions from freshly isolated day 15 fetal thymocytes from NK cell precursor activity. These results indicate that the early timed-pregnant C57BL/6 mice and adult RAG-2Ϫ/Ϫ mice were sorted for Ϫ ϩ fetal thymus is completely capable of supporting NK lineage de- a CD3 /CD90 (Thy-1) phenotype with or without NK1.1 expression. 51 velopment and that NK cell maturation and function precede ␣␤ T Sorted cells were assayed for cytolytic activity using a standard Cr-release assay (28). Sorted NK1.1ϩ/CD90ϩ/CD3Ϫ or NK1.1Ϫ/CD90ϩ/CD3Ϫ cell differentiation in mouse fetal ontogeny. cells were washed twice and aliquoted at different effector to target ratios in 100 ␮l of culture medium (DMEM medium supplemented with 12% Materials and Methods FCS, 2 mM glutamine, 10 U/ml penicillin, 100 ␮g/ml streptomycin, 100 Mice ␮g/ml gentamicin, 110 ␮g/ml sodium pyruvate, 50 ␮M 2-ME, and 10 mM HEPES, pH 7.4). Target YAC-1 or EL4 cells were labeled with 51Cr for 1 h Downloaded from Timed-pregnant C57BL/6 and Swiss.NIH mice were obtained from the and used at 3 ϫ 103 cells in 100 ␮l per well (U-bottom, 96-well plates). National Cancer Institute, Frederick Cancer Research and Development Cells were mixed at the indicated E:T ratios; then plates were centrifuged Ϫ/Ϫ Center (Frederick, MD). RAG-2 mice were bred and maintained in our for 30 s and placed in culture for4hat37°C. An amount equal to 100 ␮l animal facility. of culture supernatant was collected and measured in a gamma counter. Supernatant from target cells cultured alone or target cells plus 1% SDS Flow cytometric analysis and cell sorting gave the spontaneous or maximal release counts, respectively. Spontaneous release was Ͻ10% of maximal release. Counts obtained from culture su- Single-cell suspensions were stained for surface expression of various http://www.jimmunol.org/ markers using FITC-, Cychrome-, APC-, phycoerythrin (PE)-, or Red-613- pernatants at different E:T ratios (experimental release) were used to de- conjugated mAbs obtained from PharMingen (San Diego, CA) or Life termine percent specific lysis, as previously described (28). Technologies (Bethesda, MD), respectively, in staining buffer (Hank’s- In vivo adoptive transfer buffered saline (HBS) with 1% BSA and 0.05% NaN3). Cells were stained ␮ in 100 l for 30 min on ice and washed twice before analysis. Stained cells CD24/CD25-depleted day 15 fetal thymocytes from Sw mice were sorted were analyzed with FACScan or FacsCalibur flow cytometers using Lysis for NK1.1Ϫ/CD44ϩ (CD117ϩ) and NK1.1ϩ/CD44ϩ (CD117Ϫ) cells. II or CellQuest software (Becton Dickinson; Mountain View, CA); data Sorted cells (105 of each) were washed twice and resuspended in 300 ␮lof was live-gated by size and lack of propidium iodide uptake. All plots dis- culture medium, then injected into the tail vein of sublethally irradiated play 10,000 events contoured at 50% (log), except in Figure 5, where dot (750 cGy) adult RAG-2Ϫ/Ϫ mice. Mice were killed by cervical dislocation plots show 20,000 events. Events contained in each quadrant are given as 3 wk later and tissues were harvested for analysis. Single-cell suspensions percentages in the upper right corner. by guest on September 24, 2021 low Ϫ of spleen, thymus, lymph node, and bone marrow were analyzed by flow CD24 /CD25 day 15 fetal thymocytes, fetal liver, and adult RAG- cytometry. 2Ϫ/Ϫ thymocytes were obtained by Ab/complement-mediated lysis. Single- cell suspensions were incubated on ice with 50 to 100 ␮l of J11d.2 (anti- FTOC reconstitution CD24) and 100 to 500 ␮l of 7D4 (anti-CD25) culture supernatant for 15 CD24/CD25-depleted day 15 fetal thymocytes were sorted for NK1.1Ϫ/ min, followed by the addition of medium plus a 1/10 dilution of Low-Tox ϩ ϩ Ϫ rabbit complement (Cedar Lane, Hornby, ON) to a final volume of 3 ml, CD117 and NK1.1 /CD117 cells. Lymphocyte-depleted thymic lobes and cells were incubated at 37°C for 30 min. After complement-mediated were prepared by culturing day 15 fetal thymic lobes from timed-pregnant lysis, viable cells were recovered by discontinuous density gradient cen- Sw mice in FTOC medium containing 1.35 mM dGuo, as previously de- trifugation over Lympholyte-M (Cedar Lane). CD24low/CD25Ϫ cells rep- scribed (29, 30). After 5 to 6 days, dGuo-containing medium was replaced resented ϳ4% of total day 15 fetal thymocytes. For cell sorting, fetal thy- with FTOC medium for one day; then lobes were rinsed twice, resuspended mus single-cell suspensions were prepared and stained for FACS as in 10 ␮l medium, and placed in Terasaki plates at two lobes (one thymus) per well. Sorted donor cells (1–3 ϫ 103 of each) were washed twice with described above, except that no NaN3 was added to staining buffer. Cells were sorted using a Coulter Elite cytometer (Hialeah, FL); sorted cells were medium before reconstitution, resuspended in 20 ␮l medium, and added to 98 to 99% pure, as determined by postsort analysis. dGuo-treated alymphoid fetal thymic lobes in Terasaki plates. After adding donor cells or medium alone, the plates were inverted (“hanging drop”), RT-PCR analysis and cultures were incubated at 37°C in a humidified incubator containing Ϫ 5% CO in air for 24 to 48 h. Lobes were then transferred to standard CD24low/CD25 cell suspensions from day 15 fetal thymocytes, day 15 2 Ϫ Ϫ FTOC for 10 to 12 days. Cell suspensions from reconstituted thymic lobes fetal liver, and adult RAG-2 / mice were prepared as described above. were analyzed by flow cytometry. Total RNA was obtained from cell pellets using the Trizol RNA isolation ␮ protocol (Life Technologies). RNA was resuspended in 25 l diethyl-py- OP9 stromal cell line coculture rocarbonate (DEPC)-treated (0.1%) dH2O, and residual genomic DNA was digested using RNase-free DNase (Boehringer Mannheim, Indianapolis, In parallel with FTOC reconstitutions, CD24/CD25-depleted day 15 fetal IN). RNA was re-extracted using the Trizol protocol and resuspended in 25 thymocytes were sorted for NK1.1Ϫ/CD117ϩ and NK1.1ϩ/CD117Ϫ cells. ␮ ␮ ϫ 3 l DEPC-treated dH2O. cDNA was prepared from 1 g of each RNA using Sorted cells (1–3 10 of each) were cocultured for 11 days on confluent random hexamer primers and the cDNA Cycle kit (Invitrogen, San Diego, monolayers of OP9 cells (31, 32) in medium containing IL-3, IL-6, IL-7, CA). Subsequent PCR analysis was performed using titrations of cDNA in and SCF (50 ng/ml of each cytokine), then stimulated with LPS (10 ␮g/ml)

a 1/5 dilution series in dH2O. dH2O and RT reactions done in the absence and IL-7 for 4 to 6 days before harvesting for flow cytometry. of avian myeloblastosis virus (AMV) reverse transcriptase were included as negative controls. PCR was performed for 30 s at 94°C, 45 s at 50 to Results and Discussion 55°C (depending on primer T ’s), and 30 s at 72°C for 32 cycles, with a ϩ Ϫ m Mature NK cells (NK1.1 /CD117 ) are present early in hot start at 94°C for 2 min and a final extension at 72°C for 3 min, using annealing temperatures specific for primer pairs as determined using the mouse fetal thymic ontogeny OLIGO program (NBI Software, Plymouth, MN). All PCR reactions for To outline the developmental appearance of the NK cell lineage in each group were performed using the same cDNA batches as shown for ␤-actin, and all PCR products correspond to the expected molecular size. fetal thymic ontogeny, we analyzed expression of the NK cell Gene-specific primers used for PCR are as follows (5Ј33Ј): ␤-actin 5Ј, marker, NK1.1 (NKR-P1C) (23, 24), on day 13 to 16 fetal thy- GAT GAC GAT ATC GCT GCG CTG; ␤-actin 3Ј, GTA CGA CCA GAG mocytes and fetal liver-derived hemopoietic cells. The existence of 746 NATURAL KILLER CELL DEVELOPMENT IN THE FETAL THYMUS

anti-CD24; 3C7, anti-CD25; Fig. 2a and data not shown); CD24low/CD25Ϫ cells represented ϳ4% of total day 15 fetal thymocytes. Although CD24/CD25 depletion of fetal thymocytes and fetal liver cells enriched for CD24low cells expressing high levels of CD117 (Fig. 2a), we noticed a signiﬁcant increase in CD117Ϫ/ CD24Ϫ cells among depleted fetal thymocytes but not fetal liver cells (Fig. 2a). Therefore, depleted cells were analyzed further for expression of various lymphocyte differentiation markers. Early investigations into thymocyte differentiation revealed that precursor potential resides in a TN fraction, a designation that originally included a CD5Ϫ/low (Ly-1) phenotype (38). Recent reports have determined that the earliest precursor thymocytes are CD5low, while fetal liver-derived hemopoietic precursors possess a CD5Ϫ phenotype (39). Analysis of CD24low/CD25Ϫ day 15 fetal thymocytes and fetal liver cells for expression of CD5 vs CD117 conﬁrmed that among the CD117ϩ and CD24low populations, which have been shown to possess multilineage precursor potential, fetal thymocytes express higher levels of CD5 than fetal liver

ϩ Ϫ Downloaded from FIGURE 1. Identification of lymphocytes with an NK1.1 /CD117 cells (Fig. 2b and data not shown). However, a significant popu- (c-kit) mature NK cell phenotype during fetal thymic ontogeny. Two- lation of CD5Ϫ cells lacking expression of both CD117 and CD24 parameter flow cytometric analysis of cell surface expression of NK1.1 was present among fetal thymocytes but absent in fetal liver cells vs CD117 on fetal thymocytes from timed-pregnant C57BL/6 mice (Fig. 2b and data not shown). Remarkably, this CD5Ϫ/CD117Ϫ/ (days 13, 14, 15, 16 of gestation), fetal liver cells (day 15 of gestation), Ϫ and thymocytes from adult mice (8 wk old). NK1.1ϩ cells lacking CD24 thymocyte population accounted for approximately 30 to CD117 expression are first detectable in the fetal thymus between days 40% of CD24/CD25-depleted day 15 fetal thymocytes (Fig. 2b). http://www.jimmunol.org/ 13 and 14 of gestation. Previous studies have shown that early progenitor thymocytes express CD44 (Pgp-1), CD16/32 (Fc␥RIII/II), and low levels of CD90 (Thy-1) (16, 17). Although the roles for these markers in ϩ NK1.1 cells during fetal ontogeny has been controversial because lymphocyte development remain to be elucidated, they provide NK1.1 expression was reported to be absent in the fetal thymus useful tools for the developmental staging of distinct precursor (16, 17, 33, 34), although earlier investigations had suggested that thymocytes: CD44 is present on precursor thymocytes up to and this was not the case (28, 35). Figure 1 shows that a significant including the pro-T cell stage, when CD25 is up-regulated and percentage of total thymocytes display detectable NK1.1 expres- commitment to the T lineage occurs (4, 18, 19, 36); high level sion as early as day 13 of gestation. We recently reported the expression of CD16/32 has been associated with a putative bipo- by guest on September 24, 2021 identification of a novel precursor phenotype that marks a devel- tent T/NK precursor stage, with diminishing expression after the opmental stage of thymocyte lineage commitment to the T and NK pro-T cell stage (16, 33, 40); and the earliest precursor thymocytes cell fates in early fetal thymic ontogeny (3). These progenitors bear low levels of CD90 until the pro-T cell stage, when high level coexpress NK1.1 and the receptor for SCF, CD117 (c-kit), which expression of CD90 is attained (4). Further analysis of CD24/ is characteristic of hemopoietic precursors in the fetal liver, bone CD25-depleted day 15 fetal thymocytes revealed that a significant ϩ marrow, and thymus (17, 25–27). As shown in Figure 1, NK1.1 / population of CD117Ϫ cells expressed both CD44 and CD16/32 ϩ ϩ CD117 cells represent the majority of NK1.1 cells in the fetal (Fig. 2b); again, these cells were absent in fetal liver cell prepa- thymus by day 13 of gestation. However, between days 13 and 14 rations (Fig. 2b). Analysis of CD90 expression demonstrated that ϩ ϩ of gestation, approximately one day after the NK1.1 /CD117 these CD117Ϫ thymocytes were predominantly CD90high, while ϩ stage is first observed, there is an emergence of NK1.1 thymo- CD117ϩ precursors were CD90low to CD90Ϫ among fetal thymocytes lacking expression of CD117 (Fig. 1), corresponding to a cyte and fetal liver cells, respectively. ϩ mature NK cell phenotype. By days 14 to 15, the NK1.1 / Another marker that has recently been correlated to a subset of Ϫ CD117 population predominates, and the percentage of cells co- early fetal thymocytes is CD122 (IL-2R␤) (22, 41). CD122 ex- expressing NK1.1 and CD117 diminishes thereafter. Significant pression has been suggested to be expressed on a population of expression of NK1.1 was not detected on day 13 to 16 fetal liver putative bipotent T/NK precursors, as well as on the earliest thy- cells (Fig. 1, day 15, and data not shown), suggesting that the mic immigrant cells (22, 41). A subset of fetal liver cells, including ϩ efficient generation of NK1.1 cells in fetal ontogeny occurs dur- the earliest committed B lineage precursors (Fraction A, ing or after migration of hemopoietic precursors from the fetal CD45R(B220)ϩ/CD43ϩ/CD24Ϫ) (42), also express CD122 (41). liver to the thymus. Flow cytometric analysis of CD24/CD25-depleted thymocytes revealed that the majority of CD117Ϫ cells expressed CD122, while Fetal thymic NK cells resemble early precursor thymocytes expression of CD122 was virtually absent on fetal liver cell prep- CD24 (HSA) and CD25 (IL-2R␣) are two markers frequently used arations (Fig. 2b). Because CD122 is also present on all mature to discriminate later T lineage differentiation stages from TLPs, resting NK cells in the adult mouse, and the overall phenotype of which are CD24low/CD25Ϫ. Figure 2 shows day 15 fetal thymo- the CD117Ϫ/CD24Ϫ thymocyte population resembled that of NK cytes and fetal liver cells before and after anti-CD24 (J11d.2) and cells, we analyzed these cells for expression of NK1.1. As shown anti-CD25 (7D4) depletion. Day 15 fetal thymocytes were chosen in Figure 2b, the majority (Ͼ80%) of CD117Ϫ cells among CD24/ because these cells contain no mature ␣␤ T or B lymphocytes, CD25-depleted thymocytes expressed NK1.1, while significant ex- possess an overall CD3Ϫ/CD4Ϫ/CD8Ϫ TN phenotype (36, 37), pression of NK1.1 on depleted fetal liver cells could not be de- and contain a significant population of mature NK cells (NK1.1ϩ/ tected. These NK1.1ϩ/CD117Ϫ fetal thymocytes also express the CD117Ϫ, Fig. 1). Postdepletion analysis verified that populations novel pan-NK cell marker, DX5 (43), while fetal liver cells and were Ͼ98% depleted of both CD24high and CD25ϩ cells (M1/69, CD117ϩ fetal thymocytes, including our recently described The Journal of Immunology 747 Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021

FIGURE 2. Fetal thymic NK cells phenotypically resemble early precursor thymocytes. a, Flow cytometric analysis of cell surface expression of CD117 vs CD24 on day 15 fetal thymocytes and fetal liver cells before and after Ab/complement-mediated depletion of CD24high (HSA, J11d.2) and CD25ϩ (IL-2R␣, 7D4) cells. b, Two-parameter analysis of CD24low/CD25Ϫ fetal thymocytes and fetal liver cells for expression of various lymphocyte differentiation markers vs CD117. The majority of CD24/CD25-depleted fetal thymocytes consist of CD117ϩ multipotent progenitors and CD117Ϫ mature NK cells, the latter population being absent in the fetal liver.

NK1.1ϩ/CD117ϩ fetal thymic NK1.1ϩ progenitor (FTNK) cells, termined (Fig. 2). Thus, the fetal thymus contains a subset of lack expression of DX5 (data not shown; J. R. Carlyle et al., manu- mature NK lymphocytes that displays some markers typical of script in preparation). Further characterization of the overall phe- precursor thymocytes, such as CD44 and CD16/32, yet lacks ex- notype of these cells was performed by multiparameter ﬂow cy- pression of numerous differentiation markers, including CD3, tometric analysis of sorted NK1.1ϩ day 15 fetal thymocytes. A CD4, CD8 (i.e., TN), CD5, CD24, CD25, and other lineage (Lin) detailed summary of the composite phenotype of the NK1.1ϩ/ markers, such as B220, Mac-1, Gr-1, and Ter-119. To determine CD117Ϫ fetal thymic NK cell population is outlined in Table I, in whether fetal thymic NK cells were indeed functionally mature, we comparison to our recently described fetal thymic NK1.1ϩ/ characterized their properties further. CD117ϩ (FTNK) T/NK-committed progenitors, and NK1.1Ϫ/ CD117ϩ fetal TLPs (3). Fetal thymic NK cells express genes associated with NK cell Taken together, our results show that mature (NK1.1ϩ/ effector function CD117Ϫ) NK cells are highly enriched among precursor-pheno- The NK1.1 molecule (NKR-P1C) is a member of the NKR-P1 type thymocyte populations in which CD117 expression is not de- gene family (23, 24) and forms part of a proposed NK receptor 748 NATURAL KILLER CELL DEVELOPMENT IN THE FETAL THYMUS

Table I. Phenotypic characterization of fetal thymocyte subsets grouped according to expression of NK1.1 and CD117a

FTLP,b FTNK, Mature NK, Marker NK1.1Ϫ/CD117ϩ NK1.1ϩ/CD117ϩ NK1.1ϩ/CD117Ϫ

NK1.1 (NKR-P1C) Ϫϩϩ CD117 (c-kit) ϩϩϪ CD44 (Pgp-1) ϩϩϩ CD25 (IL-2R␣) ϪϪϪ CD2 (LFA-2) ϪϪϩ/Ϫc CD5 (Ly-1) low low low/Ϫ CD3⑀ ϪϪϪ CD4 ϪϪϪ CD8 ϪϪϪ CD16/32 (Fc␥RIII/II) Ϫϩϩ CD24 (HSA) low low low/Ϫ CD90 (Thy-1) low ϩ/low ϩ/Ϫ CD122 (IL-2R␤) Ϫ low/Ϫϩ MHC Class I high high high ␣␤TCR ϪϪϪFIGURE 3. NK-enriched day 15 fetal thymocytes express character-

␥␦TCR ϪϪϪistic gene products of functional NK cells. RT-PCR analysis for the Downloaded from Lin ϪϪϪexpression of NK cell effector function-associated genes on RNA iso- DX5d ϪϪϩlated from (a) total day 15 Sw fetal thymocytes (FT) and fetal liver (FL) a Cell surface expression was determined by ﬂow cytometric analysis of day cells, and (b) CD24/CD25-depleted Sw and C57BL/6 (B6) day 15 fetal Ϫ/Ϫ 13 (FTNK-enriched) or day 15 (mature NK-enriched) FT after CD24/CD25 de- thymocytes and RAG-2 adult thymocytes. CD24/CD25-depleted pletion and sorting for cells with or without NK1.1 expression. (NK-enriched) day 15 fetal thymocytes express members of the b Abbreviations: FTLP, fetal thymic lymphoid progenitor (B/T/NK multipoten- NKR-P1 and Ly-49 gene families, perforin, and Fas ligand (CD95L). tial); FTNK, fetal thymic NK1.1ϩ progenitor (T/NK bipotential); mature NK, mature NK cell; low, low but signiﬁcant staining detected; high, high level staining; http://www.jimmunol.org/ ϩ, positive staining; Ϫ, negative staining; Lin, lineage commitment markers other than those mentioned above (B220, Mac-1, Gr-1, Ter-119). c Multiple designations indicate heterogeneous expression on cell populations. d (Fig. 3a vs Fig. 3b, respectively, FT). As expected, CD24/CD25- DX5 is a novel pan-NK cell marker that binds an as yet unknown surface Ϫ/Ϫ molecule on NK cells from mice of all strains tested to date (43). depleted adult RAG-2 thymocytes (Fig. 3b, adult thymocytes (AT)) and splenocytes (data not shown) also express these NK function-associated genes. Among CD24/CD25-depleted fetal liver cells, only background expression was detectable for any of

the NK-related gene products tested (data not shown); the inability by guest on September 24, 2021 gene complex that identifies TN lymphocytes with NK cell func- to phenotypically identify significant numbers of NK cells among tion (28, 44–46). To determine whether the mature NK cell phe- fetal liver suspensions limits further attempts to enrich for such ϩ Ϫ notype of NK1.1 /CD117 fetal thymocytes was indicative of NK cells or their immediate precursors, without employing in vitro cell function, we assessed the expression of various genes associ- culture techniques. Nevertheless, day 15 fetal thymocytes express ated with NK cell effector function by performing RT-PCR on numerous gene products typically associated with NK cell effector RNA isolated from total and CD24/CD25-depleted (NK-enriched) function. day 15 fetal thymocytes and fetal liver cells. As a positive control, RT-PCR analysis of NK-enriched cell populations, although not RNA was also isolated from CD24/CD25-depleted adult RAG- quantitative nor conclusive, revealed a number of significant find- 2Ϫ/Ϫ thymuses, which lack thymocytes beyond the early pre-T cell ings. In comparison to NKR-P1A and NKR-P1C levels, NKR-P1B stage (CD44Ϫ/CD25ϩ), yet contain normal to elevated numbers of expression was quite high among Sw fetal thymocytes, while re- mature NK cells (15). Ϫ Ϫ maining very low in fetal B6 and adult RAG-2 / thymocytes Consistent with the finding that cells with a mature NK pheno- (Fig. 3b). This difference appears to be strain-specific because the type are virtually absent among fetal liver suspensions (Figs. 1 and same trend was observed for adult spleen cells from each of the 2), no significant expression of these genes could be detected by strains (data not shown). These data suggest that strain-specific RT-PCR on RNA isolated from day 15 fetal liver cells (Fig. 3a, expression of NKR-P1 gene family members may be controlled at fetal liver (FL)). In contrast, among total day 15 fetal thymocytes, low level expression of NK-related genes could be detected (Fig. the transcriptional level. In contrast, expression of Ly-49 gene 3a, FT), including products of the NKR-P1 gene family (NKR- family members appears to be developmentally regulated in the P1A; NKR-P1B; and NKR-P1C, NK1.1) (23, 24), the Ly-49 gene thymus. Expression levels of both Ly-49A and Ly-49C, in com- family (Ly-49A; Ly-49C) (44, 47), Fas ligand (CD95L) (48, 49), parison with the other genes tested, were higher in adult (RAG- Ϫ/Ϫ and the cytolytic pore-forming molecule, perforin (50–52). In ad- 2 ) thymocytes than in fetal (Sw and B6) thymocytes. This does dition, it has been previously demonstrated that immature fetal not appear to be due to strain-specific differences, because Ly- thymocytes express the cell death-associated protease/caspase 49A/C expressions were comparable in all three strains among (53), granzyme B, at an early stage in ontogeny (54). To further adult splenocytes (data not shown). Thus, it appears that in the characterize expression of these NK-related gene products, day 15 fetal thymus, expression of Ly-49 genes may be developmentally thymocyte suspensions from two unrelated (albeit NK1.1-express- delayed, while NKR-P1 molecules appear quite early in ontogeny. ing) strains of mice, Sw and C57BL/6 (B6), were isolated and This may have important functional consequences because these enriched for mature NK cells by depleting for CD24/CD25 before two families of molecules are postulated to possess opposing reg- RNA isolation. Consistent with the degree of NK cell enrichment ulatory roles in NK cell effector function, with Ly-49 acting as an observed phenotypically in Figures 1 and 2, expression of NK- inhibitory receptor (44, 47, 55), and NKR-P1 acting as a positive related genes was dramatically enhanced by CD24/CD25 depletion modulator of NK cell activity and lytic function (45, 55). The Journal of Immunology 749

FIGURE 4. Freshly isolated fetal thymic NK cells mediate MHC-unrestricted cytotoxicity in vitro. Fresh day 15 fetal thymocytes (B6) and RAG-2Ϫ/Ϫ adult thymocytes were sorted for CD3Ϫ/CD90ϩ (Thy-1) cells, with or without NK1.1 expression, and tested for cytotoxicity against NK-sensitive YAC-1 cells or the NK-insensitive cell line, EL-4. Sorted NK1.1ϩ day 15 fetal thymocytes mediate MHC-unrestricted cytotoxicity of YAC-1 targets ex vivo, in the absence of exogenous cytokine treatment, while sorted NK1.1Ϫ or total (unsorted) day 15 fetal thymocytes fail to lyse either target. Downloaded from

Freshly isolated fetal thymic NK cells display cytes were capable of giving rise to donor-derived NK cells (Fig. MHC-unrestricted cytotoxicity in vitro 5b, NK1.1ϩ/H-2Kqϩ) in the spleen. However, only the multipotent NK1.1Ϫ subset (CD117ϩ, Fig. 2b) was capable of generating B To determine whether fetal thymic NK cells are functional, we Ϫ ϩ Ϫ cells, as determined by CD45R (B220) expression on NK1.1 do- tested the ability of freshly sorted NK1.1 (CD117 ) fetal thy- ϩ qϩ http://www.jimmunol.org/ 51 nor-derived progeny (Fig. 5b, CD45R /H-2K ). These donor- mocytes to perform MHC-unrestricted cytolysis of Cr-labeled ϩ YAC-1 target cells. Thymocytes obtained at day 15 of fetal ges- derived CD45R cells also expressed surface IgM (data not tation were sorted by FACS for a CD3Ϫ/CD90ϩ phenotype, with shown). Cells from nonreconstituted (Control) mice showed no background staining for donor class I (H-2Kqϩ) expression (Fig. or without NK1.1 expression. As shown in Figure 4, freshly sorted Ϫ NK1.1ϩ day 15 fetal thymocytes, without a requirement for pre- 5b). The reconstitution potential of the NK1.1 subset is not lim- exposure to cytokines such as IFN-␥, IL-2, IL-12, or IL-15, were ited only to B and NK cell lineages because these cells are also capable of lysing NK-sensitive YAC-1 target cells but failed to capable of giving rise to T cells in FTOC reconstitutions (3). How- lyse the NK-insensitive EL-4 cell line. As expected, freshly iso- ever, we could find no evidence of T lineage reconstitution in the ϩ Ϫ Ϫ lated NK1.1 thymocytes from adult RAG-2 / mice also lysed thymus upon in vivo adoptive transfer of either subset. This ob- by guest on September 24, 2021 YAC-1 targets, while failing to lyse EL-4 cells (Fig. 4). In contrast, servation is consistent with previous studies assessing the precur- fetal thymocytes lacking expression of the NK1.1 marker failed to sor potential of fetal thymocytes upon adoptive transfer into adult lyse either target, as did total (unsorted) fetal thymocytes (Fig. 4). host mice (18, 56). It has been suggested that this may be due to The latter observation is consistent with previous attempts to de- a developmental stage difference; fetal thymocytes exhibit a re- duction in thymic reconstitution potential compared with their tect NK cell function in freshly isolated fetal thymocytes (16, 28, low 34, 35); in the absence of purification, the high frequency of analogous “CD4 ” adult counterpart upon i.v. adoptive transfer NK1.1Ϫ thymocytes could inhibit NK cell cytotoxic function. Im- and may have difficulty homing to the adult thymus microenvi- portantly, we show that freshly isolated fetal thymic NK cells pos- ronment (16, 40, 56). Therefore, we employed an in vitro FTOC sess cytolytic function at a developmental stage before the appear- reconstitution assay for T and NK cell potential. Additionally, for ance of CD4ϩ/CD8ϩ DP cells in fetal thymic ontogeny. Thus, detecting B and NK cell potential, we employed a sensitive in vitro functional NK cell development precedes ␣␤ T cell differentiation coculture assay using the bone marrow-derived stromal cell line, in mouse fetal thymic ontogeny. OP9 (3). Fetal thymic NK cells are capable of sustained growth In vivo adoptive transfer of precursor-phenotype thymocytes in vitro The early developmental maturity of fetal thymic NK cells, com- To address the growth potential of fetal thymic NK cells in a T bined with their close phenotypic resemblance to early progenitor lineage assay, we assessed their ability to reconstitute dGuo-de- thymocytes (Fig. 2 and Table I), implies that previous descriptions pleted FTOCs. CD24/CD25-depleted day 15 fetal thymocytes of purported multipotent, bipotent T/NK, or unipotent NK lineage were sorted for NK1.1Ϫ/CD117ϩ (FTLP) and NK1.1ϩ/CD117Ϫ “precursor” thymocytes, in particular those involving populations (NK) cells, and 1 to 3 ϫ 103 donor cells were used for FTOC not defined according to CD117 or NK1.1 expression, may have reconstitution. As previously demonstrated (3), sorted NK1.1Ϫ/ inadvertently included pre-existing mature NK cells (16, 17, 22). CD117ϩ (FTLP) cells gave rise to immature CD4/CD8 double- To address this issue directly, we used in vivo adoptive transfers. positive and mature CD4 and CD8 single-positive cells (Fig. 6a). CD44ϩ/CD25Ϫ cells, which have been thought to contain multi- In addition, FTLP cells were capable of generating mature T cells, potent precursors for the T, B, and NK lineages (4), were sorted as determined by high-level expression of ␣␤TCR on NK1.1Ϫ from CD24/CD25-depleted day 15 fetal thymocytes (Sw mice, cells (Fig. 6a ␣␤TCRϩ/NK1.1Ϫ), as well as a few NK cells (Fig. H-2q) and subdivided according to NK1.1 expression (Fig. 5a). 6a, ␣␤TCRϪ/NK1.1ϩ) (3). In contrast, NK1.1ϩ/CD117Ϫ (NK) Three weeks after i.v. injection into sublethally irradiated (750 cells remained double-negative for both CD4 and CD8 and exclu- cGy) RAG-2Ϫ/Ϫ (H-2b) host mice, tissues were examined for ev- sively gave rise to an outgrowth of NK cells (Fig. 6a). These in idence of donor-derived (H-2Kqϩ) progeny. As shown in Figure 5, vitro-cultured NK cells are large granular lymphocytes. The low both the NK1.1ϩ and NK1.1Ϫ subsets of CD44ϩ/CD25Ϫ thymo- level staining observed for ␣␤TCR is due to increased background 750 NATURAL KILLER CELL DEVELOPMENT IN THE FETAL THYMUS Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021

FIGURE 5. Precursor-phenotype fetal thymocytes sorted according to NK1.1 expression show distinct reconstitution potential upon in vivo adoptive transfer. a, FACS of CD24/CD25-depleted day 15 Sw (H-2q) FT for NK1.1Ϫ/CD44ϩ (R1, 27%) and NK1.1ϩ/CD44ϩ (R2, 12%) cells. Cells (1 ϫ 105) were injected i.v. into sublethally irradiated (750 cGy) adult RAG-2Ϫ/Ϫ mice (H-2b). b, Spleen cells from reconstituted or nonreconstituted (Control) mice were analyzed 3 wk after i.v. injection of sorted NK1.1ϩ and NK1.1Ϫ subsets of CD44ϩ/CD25Ϫ thymocytes. Flow cytometric analysis of NK1.1 vs H-2Kq and CD45R (B220) vs H-2Kq reveals that both populations of precursor-phenotype thymocytes give rise to donor-derived (H-2Kqϩ) NK cells upon adoptive transfer, yet only the multipotent NK1.1Ϫ subset is capable of generating B lymphocytes, as revealed by CD45R expression on NK1.1Ϫ donor cells.

staining because we failed to detect DJ␤ rearrangement on their DX5 mAb (data not shown). Although we have not quantitated the DNA by PCR (data not shown). Cell yields from FTOCs indicated extent of growth potential exhibited by fetal thymic NK cells in that fetal thymic NK cells have the capacity to expand at least this assay (at least 50-fold), they continue to divide in culture and 10-fold in this assay, depending on the length of the culture period. can be maintained for at least 1 to 2 mo in vitro. However, they To examine the growth potential of fetal thymic NK cells further must be continually passed onto fresh OP9 cells because the OP9 in a B cell assay, we seeded sorted cells onto OP9 bone marrow- cells are lysed due to the cytolytic activity of the NK cells. In derived stromal cells. In parallel with FTOC reconstitutions, 1 to addition, the growth of fetal thymic NK cells is enhanced with 3 ϫ 103 sorted NK1.1Ϫ/CD117ϩ (FTLP) and NK1.1ϩ/CD117Ϫ exogenous IL-2. Therefore, the reported existence of NK cell pre- (NK) cells were cocultured with conﬂuent OP9 cells, as described cursors in the fetal thymus that can grow out in vitro in the pres- previously (3). Again, as recently shown (3), sorted NK1.1Ϫ/ ence of IL-2 was most likely due to an expansion of this pre- CD117ϩ (FTLP) cells gave rise to mature B cells, indicated by existing subset of mature NK cells (22, 34, 57). expression of surface IgM on CD45Rϩ cells (Fig. 6b, CD45Rϩ/ The in vivo and in vitro data shown in Figures 5 and 6, together IgMϩ), and a few NK cells (Fig. 6b, NK1.1ϩ/CD90ϩ/Ϫ). How- with our previous evidence (3), suggest that the NK1.1Ϫ ever, NK1.1ϩ/CD117Ϫ (NK) cells remained negative for both (CD117ϩ) subset of CD44ϩ/CD25Ϫ thymocytes represents mul- CD45R and IgM and again gave rise to an outgrowth of mature tipotent lymphoid-restricted precursors, while the NK1.1ϩ popu- NK cells, the majority of which expressed CD90 (Fig. 6b, lation contains mature NK cells (CD117Ϫ) that are capable of sus- NK1.1ϩ/CD90ϩ/Ϫ); these NK cells also stained positive for the tained growth after adoptive transfer in vivo or in vitro. Thus, there The Journal of Immunology 751 Downloaded from

FIGURE 6. Sustained growth of fetal thymic NK cells upon in vitro culture. CD24/CD25-depleted day 15 fetal thymocytes were sorted for NK1.1Ϫ/CD117ϩ and NK1.1ϩ/CD117Ϫ cells, which were cocultured under differential conditions. a, dGuo-depleted fetal thymic lobes were reconstituted with 1 to 3 ϫ 103 cells, either NK1.1Ϫ/CD117ϩ or NK1.1ϩ/CD117Ϫ, and cultured for 12 days in FTOC. Flow cytometric analysis ␣␤ Ϫ for CD4 vs CD8 and NK1.1 vs TCR expression reveals that both subsets give rise to NK cells, yet only the NK1.1 subset is capable of generating http://www.jimmunol.org/ conventional ␣␤ T cells, as indicated by high level expression of ␣␤TCR on NK1.1Ϫ cells. b, In parallel with FTOC reconstitutions, 1 to 3 ϫ 103 sorted NK1.1Ϫ/CD117ϩ or NK1.1ϩ/CD117Ϫ cells were cocultured for 11 days on confluent monolayers of OP9 bone marrow-derived stromal cells in the presence of cytokines (IL-3, -6, -7, and SCF), then stimulated with LPS and IL-7 for an additional 6 days before analysis. Flow cytometric analysis of CD45R (B220) vs IgM and NK1.1 vs CD90 reveals that both populations give rise to NK cells, yet only the NK1.1Ϫ subset is capable of generating B lymphocytes, as revealed by IgM expression on CD45Rϩ cells. is a population of pre-existing NK1.1ϩ cells among precursor- not reach significant levels in the circulation until the neonatal phenotype fetal thymocytes that contain mature and functional NK stage (28, 35, 46, 61), when hemopoietic function shifts from the by guest on September 24, 2021 cells capable of confounding lineage potential assays if not prop- fetal liver to the neonatal/adult bone marrow, a site that is known erly distinguished from multipotent CD117ϩ/CD44ϩ/CD25Ϫ/ to be capable of supporting NK lineage maturation (62, 63). It may NK1.1Ϫ precursors. In light of these findings, the purported dis- be that the bone marrow is primarily responsible for peripheral NK covery of bipotent T/NK precursors requires reassessment. cell production whereas the fetal liver may be incapable of supporting NK lineage differentiation, possibly due to the absence of Conclusions particular cytokines or stromal microenvironments (64, 65). Al- Our data provide the first evidence of the development of mature though the fetal thymus is capable of supporting complete NK cell and functional NK cells in mouse fetal ontogeny. The fact that NK maturation, thymus-derived NK cells may be locally involved in cell maturation initially occurs within the early fetal thymus, to- regulation of thymopoiesis (66) and may not reach the periphery. gether with the recent description of NK1.1ϩ ␣␤ T cells (58) and Consistent with this, we have failed to detect significant numbers the NK1.1ϩ/CD117ϩ (FTNK) bipotent T/NK progenitor stage of of NK cells in the fetal blood and spleen until day 16 of gestation thymocyte development (3), further reinforces the developmental (J. R. Carlyle, manuscript in preparation). Nonetheless, mature NK and lineage relationships between T and NK cells. Moreover, we cells differentiate early during fetal thymic ontogeny, exhibit gene show that NK cells are phenotypically present in the fetal thymus expression patterns consistent with NK cell effector function, and by day 13 of gestation, before the onset of VDJ rearrangement of display MHC-unrestricted cytotoxicity ex vivo, without a require- the TCR␤ locus; as well, NK cell function is detectable by day 15, ment for pre-exposure to cytokines. before the appearance of CD4ϩ/CD8ϩ cells in fetal thymic ontog- Importantly, the close phenotypic resemblance of fetal thymic eny. This indicates that NK cell development precedes that of ␣␤ NK cells to early precursor thymocytes implies that previous de- T lymphocytes. Although an analogous subset of NK cells scriptions of purported NK precursor and bipotent T/NK precursor (CD56ϩ/CD5Ϫ) was observed in the human fetal thymus (59, 60), potentials may have been contaminated with these pre-existing the earliest stages of NK cell development were not outlined, and mature NK cells. Indeed, our in vivo transfer experiments provide it remains unknown during human fetal thymic ontogeny whether direct evidence that the NK1.1ϩ subset of CD44ϩ/CD25Ϫ fetal NK cells are present and/or functional before ␣␤ T cell differen- thymocytes (which also expresses CD16/32) can reconstitute NK tiation. Thus, our data are the first to show that the maturation of cells upon adoptive transfer. Therefore, previous findings that have functional NK cells, like that of the canonical V␥3ϩ ␥␦ T cells, used CD44, CD16/32, and/or CD122 to identify progenitor thy- precedes ␣␤ T cell development (5). mocytes, in particular where characterization of NK1.1 and/or Our identification of fetal thymic NK cells, together with the CD117 is lacking, may have inadvertently included mature NK inability to detect significant NK1.1 expression in the fetal liver cells within a putative precursor population. Fetal thymic NK cells (Figs. 1, 2, and 4), suggests that fetal NK cell differentiation may are capable of sustained outgrowth, both in vitro and in vivo, po- be restricted to the thymus until the establishment of peripheral tentially obscuring bona fide multipotent, bipotent, and unipotent sites of NK lymphopoiesis. This could explain why NK cells do NK lineage precursor activity. Hence, investigations that failed to 752 NATURAL KILLER CELL DEVELOPMENT IN THE FETAL THYMUS exclude pre-existing NK cells before assessing NK lineage poten- 25. Ogawa, M., Y. Matsuzaki, S. Nishikawa, S. Hayashi, T. Kunisada, T. Sudo, tial (16, 17, 22, 33), including the purported discovery of bipotent T. Kina, H. Nakauchi, and S. Nishikawa. 1991. Expression and function of c-kit in hemopoietic progenitor cells. J. Exp. Med. 174:63. T/NK precursors, must now be re-evaluated in light of our obser- 26. deVries, P., K. A. Brasel, H. J. McKenna, D. E. Williams, and J. D. Watson. vations. Indeed, the early developmental expression of NK1.1 and 1992. Thymus reconstitution by c-kit-expressing hematopoietic stem cells puri- other members of the NKR-P1 gene family suggest that such NK fied from adult mouse bone marrow. J. Exp. Med. 176:1503. 27. Godfrey, D. I., A. Zlotnik, and T. Suda. 1992. Phenotypic and functional char- cell molecules might be included as lineage (Lin) differentiation acterization of c-kit expression during intrathymic T cell development. J. Immu- markers for future hemopoietic precursor evaluations, both intra- nol. 149:2281. thymic and extrathymic. Our identification of mature and func- 28. Hackett, J., Jr., M. Tutt, M. Lipscomb, M. Bennett, G. Koo, and V. Kumar. 1986. Origin and differentiation of natural killer cells. II. Functional and morphologic tional NK cells in fetal ontogeny sheds new light on our under- studies of purified NK-1.1ϩ cells. J. Immunol. 136:3124. standing of NK lineage development and function and could aid in 29. Jenkinson, E. J., L. Franchi, R. Kingston, and J. J. T. Owen. 1982. Effects of the derivation of long-lived mouse NK cell lines. deoxyguanosine on lymphopoiesis in the developing thymus rudiment in vitro: application in the production of chimeric thymus rudiments. Eur. J. Immunol. 12:583. Acknowledgments 30. Jenkinson, E. J., and J. J. T. Owen. 1990. T cell differentiation in thymus organ culture. Semin. Immunol. 2:51. We thank Drs. Michael Julius, Michael Lenardo, Richard G. Miller, Phil- 31. Nakano, T., H. Kodam, and T. Honjo. 1994. Generation of lympho-hematopoietic ippe Poussier, and Hergen Spits for discussions and for critically reading cells from embryonic stem cells in culture. Science 265:1098. the manuscript, and Cheryl Smith for technical assistance with cell sorting. 32. Nakano, T. 1995. Lymphohematopoietic development from embryonic stem cells in vitro. Semin. Immunol. 7:197. 33. Rodewald, H.-R. 1995. Pathways from hematopoietic stem cells to thymocytes.

References Curr. Opin. Immunol. 7:176. Downloaded from 34. Brooks, C. G., A. Georgiou, and R. K. Jordan. 1993. The majority of immature 1. Shortman, K., and L. Wu. 1996. Early T lymphocyte progenitors. Annu. Rev. fetal thymocytes can be induced to proliferate to IL-2 and differentiate into cells Immunol. 14:29. indistinguishable from mature natural killer cells. J. Immunol. 151:6645. 2. Zuń˜iga-Pflu¨cker, J. C., and M. J. Lenardo. 1996. Regulation of thymocyte de- 35. Koo, G. C., J. R. Peppard, and A. Hatzfeld. 1982. Ontogeny of Nk-1ϩ natural velopment from immature progenitors. Curr. Opin. Immunol. 8:215. killer cells. I. Promotion of Nk-1ϩ cells in fetal, baby, and old mice. J. Immunol. 3. Carlyle, J. R., A. M. Michie, C. Furlonger, T. Nakano, M. J. Lenardo, C. J. Paige, 129:867. and J. C. Zuń˜iga-Pflu¨cker. 1997. Identification of a novel developmental stage 36. Godfrey, D. I., J. Kennedy, T. Suda, and A. Zlotnik. 1993. A developmental marking lineage commitment of progenitor thymocytes. J. Exp. Med. 186:173. pathway involving four phenotypically and functionally distinct subsets of 4. Godfrey, D. I., and A. Zlotnik. 1993. Control points in early T-cell development. CD3ϪCD4ϪCD8Ϫ triple negative adult mouse thymocytes defined by CD44 and http://www.jimmunol.org/ Immunol. Today 14:547. CD25 expression. J. Immunol. 150:4244. 5. Havran, W. L., and J. P. Allison. 1988. Developmentally ordered appearance of 37. Godfrey, D. I., J. Kennedy, P. Monbaerts, S. Tonegawa, and A. Zlotnik. 1994. thymocytes expressing different T-cell antigen receptors. Nature 335:443. Onset of TCR-␤ rearrangement and role of TCR-␤ expression during 6. Trinchieri, G. 1989. Biology of natural killer cells. Adv. Immunol. 47:187. Ϫ Ϫ Ϫ CD3 CD4 CD8 thymocyte differentiation. J. Immunol. 152:4783. 7. Welsh, G., and M. Vargas-Cortes. 1992. Natural killer cells in virus infection. In The 38. Fowlkes, B. J., L. Edison, B. J. Mathieson, and T. M. Chused. 1985. Early T Natural Immune System: The Natural Killer Cell. C. E. Lewis and J. O’ D. Gee, eds. lymphocytes: differentiation in vivo of adult intrathymic precursor cells. J. Exp. Oxford University Press, New York, pp. 108–150. Med. 162:802. 8. Moretta, L., E. Ciccone, A. Poggi, M. C. Mingari, and A. Moretta. 1994. Origin and functions of human natural killer cells. Int. J. Clin. Lab. Res. 24:181. 39. Hozumi, K., A. Kobori, T. Sato, T. Nishimura, and S. Habu. 1996. Transcription and demethylation of TCR beta gene initiate prior to the gene rearrangement in 9. Biron, C. A. 1997. Activation and function of natural killer cell responses during ϩ viral infections. Curr. Opin. Immunol. 9:24. c-kit thymocytes with CD3 expression: evidence of T cell commitment in the 10. Nehls, M., D. Pfeifer, M. Schorpp, H. Hedrich, and T. Boehm. 1994. New mem- thymus. Int. Immunol. 8:1473. by guest on September 24, 2021 ber of the winged-helix protein family disrupted in mouse and rat nude mutations. 40. Rodewald, H.-R., K. Kretzschmar, S. Takeda, C. Hohl, and M. Dessing. 1994. Nature 372:103. Identification of pro-thymocytes in murine fetal blood: T lineage commitment can 11. Boehm, T., M. Nehls, and B. Kyewski. 1995. Transcriptional factors that control precede thymus colonization. EMBO J. 13:4229. development of the thymic microenvironment. Immunol. Today 16:555. 41. Reya, T., J. A. Yang-Snyder, E. V. Rothenberg, and S. R. Carding. 1996. Reg- 12. Garni-Wagner, B. A., P. L. Witte, M. M. Tutt, W. A. Kuziel, P. W. Tucker, ulated expression and function of CD122 (interleukin-2/interleukin-15R-␤) dur- M. Bennett, and V. Kumar. 1990. Natural killer cells in the thymus: studies in ing lymphoid development. Blood 87:190. mice with severe combined immune deficiency. J. Immunol. 144:796. 42. Hardy, R. R., C. E. Carmack, S. A. Shinton, J. D. Kemp, and K. Hayakawa. 1991. 13. Mombaerts, P., J. Iacomini, R. S. Johnson, K. Herrup, S. Tonegawa, and Resolution and characterization of pro-B and pre-pro-B cell stages in normal V. E. Papaioannou. 1992. RAG-1-deficient mice have no mature B and T lym- mouse bone marrow. J. Exp. Med. 173:1213. phocytes. Cell 68:869. 43. Moore, T. A., U. von Freeden-Jeffry, R. Murray, and A. Zlotnik. 1996. Inhibition 14. Shinkai, Y., G. Rathbun, K.-P. Lam, E. M. Oltz, V. Stewart, M. Mendelsohn, of ␥␦ T cell development and early thymocyte maturation in IL-7 Ϫ/Ϫ mice. J. Charron, M. Datta, F. Young, A. M. Stall, and F. W. Alt. 1992. RAG-2- J. Immunol. 157:2366. deficient mice lack mature lymphocytes owing to inability to initiate V(D)J re- 44. Yokoyama, W. M. 1995. Natural killer cell receptors. Curr. Opin. Immunol. arrangement. Cell 68:855. 7:110. 15. Diamond, R. A., S. B. Ward, K. Owada-Makabe, H. Wang, and 45. Ryan, J. C., E. C. Niemi, R. D. Goldfien, J. C. Hiserodt, and W. E. Seaman. 1991. Ϫ Ϫ E. V. Rothenberg. 1997. Different developmental arrest points in RAG-2 / and NKR-P1, an activating molecule on rat natural killer cells, stimulates phospho- SCID thymocytes on two genetic backgrounds: developmental choices and cell inositide turnover and a rise in intracellular calcium. J. Immunol. 147:3244. death mechanisms before TCR gene rearrangement. J. Immunol. 158:4052. 46. Koo, G. C., F. J. Dumont, M. Tutt, J. Hackett, Jr., and V. Kumar. 1986. The 16. Rodewald, H.-R., P. Moingeon, J. L. Lucich, C. Dosiou, P. Lopez, and NK-1.1(Ϫ) mouse: a model to study differentiation of murine NK cells. J. Im- ␥ E. L. Reinherz. 1992. A population of early fetal thymocytes expressing Fc RII/ munol. 137:3742. III contains precursors of T lymphocytes and natural killer cells. Cell 69:139. 47. Smith, H. R., F. M. Karlhofer, and W. M. Yokoyama. 1994. Ly-49 multigene 17. Matsuzaki, Y., J. Gyotoku, M. Ogawa, S. Nishikawa, and Y. Katsura. 1993. family expressed by IL-2-activated NK cells. J. Immunol. 153:1068. Characterization of c-kit positive intrathymic stem cells that are restricted to 48. Suda, T., T. Takahashi, P. Golstein, and S. Nagata. 1993. Molecular cloning and lymphoid differentiation. J. Exp. Med. 178:1283. expression of the Fas ligand, a novel member of the tumor necrosis factor family. 18. Zuń˜iga-Pflu¨cker, J. C., D. Jiang, and M. J. Lenardo. 1995. Requirement for Cell 75:1169. TNF-␣ and IL-1␣ in fetal thymocyte commitment and differentiation. Science 49. Nagata, S. 1994. Fas and Fas ligand: a death factor and its receptor. Adv. Immu- 268:1906. nol. 57:129. 19. Moore, T. A., and A. Zlotnik. 1995. T-cell lineage commitment and cytokine response of thymic progenitors. Blood 86:1850. 50. Lee, R. K., J. Spielman, D. Y. Zhao, K. J. Olsen, and E. R. Podack. 1996. 20. Sanchez, M. J., M. O. Muench, M. G. Roncarolo, L. L. Lanier, and J. H. Phillips. Perforin, Fas ligand, and tumor necrosis factor are the major cytotoxic molecules 1994. Identification of a common T/natural killer cell progenitor in human fetal used by lymphokine-activated killer cells. J. Immunol. 157:1919. thymus. J. Exp. Med. 180:569. 51. Liu, C. C., C. M. Walsh, and J. D. Young. 1995. Perforin: structure and function. 21. Spits, H., L. L. Lanier, and J. H. Phillips. 1995. Development of human T and Immunol. Today 16:194. natural killer cells. Blood 85:2654. 52. Kagi, D., B. Ledermann, K. Burki, P. Seiler, B. Odermatt, K. J. Olsen, 22. Leclercq, G., V. Debacker, M. De Smedt, and J. Plum. 1996. Differential effects E. R. Podack, R. M. Zinkernagel, and H. Hengartner. 1994. Cytotoxicity medi- of interleukin 15 and interleukin 2 on differentiation of bipotential T/natural killer ated by T cells and natural killer cells is greatly impaired in perforin-deficient progenitor cells. J. Exp. Med. 184:325. mice. Nature 369:31. 23. Giorda, R., and M. Trucco. 1991. Mouse NKR-P1: a family of genes selectively 53. Miller, D. K. 1997. The role of the Caspase family of cysteine proteases in coexpressed in adherent lymphokine-activated killer cells. J. Immunol. 147:1701. apoptosis. Semin. Immunol. 9:35. 24. Ryan, J. C., J. Turck, E. C. Niemi, W. M. Yokoyama, and W. E. Seaman. 1992. 54. Ebnet, K., C. N. Levelt, T. T. Tran, K. Eichmann, and M. M. Simon. 1995. Molecular cloning of the NK1.1 antigen, a member of the NKR-P1 family of Transcription of granzyme A and B genes is differentially regulated during lym- natural killer cell activation molecules. J. Immunol. 149:1631. phoid ontogeny. J. Exp. Med. 181:755. The Journal of Immunology 753

55. Yokoyama, W. M., and W. E. Seaman. 1993. The Ly-49 and NKR-P1 gene interleukin 2 receptor ␤ chain monoclonal antibody in mice. J. Exp. Med. 178: families encoding lectin-like receptors on natural killer cells: the NK gene com- 1103. plex. Annu. Rev. Immunol. 11:613. 62. Moore, T. A., M. Bennett, and V. Kumar. 1996. Murine natural killer cell dif- 56. Antica, M., L. Wu, K. Shortman, and R. Scollay. 1993. Intrathymic lymphoid ferentiation: past, present, and future. Immunol. Res. 15:151. precursor cells during fetal thymus development. J. Immunol. 151:5887. 63. Moore, T., M. Bennett, and V. Kumar. 1995. Transplantable NK cell progenitors 57. Skinner, M., G. Le Gros, J. Marbrook, and J. D. Watson. 1987. Development of in murine bone marrow. J. Immunol. 154:1653. fetal thymocytes in organ cultures: effect of interleukin 2. J. Exp. Med. 165:1481. 64. Manoussaka, M., A. Georgiou, B. Rossiter, S. Shrestha, J. A. Toomey, 58. Bendelac, A. 1995. Mouse NK1ϩ T cells. Curr. Opin. Immunol. 7:367. P. V. Sivakumar, M. Bennett, V. Kumar, and C. G. Brooks. 1997. Phenotypic and 59. Sanchez, M. J., H. Spits, L. L. Lanier, and J. H. Phillips. 1993. Human natural functional characterization of long-lived NK cell lines of different maturational killer cell committed thymocytes and their relation to the T cell lineage. J. Exp. status obtained from mouse fetal liver. J. Immunol. 158:112. Med. 178:1857. 65. Puzanov, I. J., M. Bennett, and V. Kumar. 1996. IL-15 can substitute for the 60. Phillips, J. H., T. Hori, A. Nagler, N. Bhat, H. Spits, and L. L. Lanier. 1992. marrow microenvironment in the differentiation of natural killer cells. J. Immu- Ontogeny of human natural killer (NK) cells: fetal NK cells mediate cytolytic nol. 4282. function and express cytoplasmic CD3 epsilon, delta proteins. J. Exp. Med. 175: 66. Murphy, W. J., A. Raziuddin, L. Mason, V. Kumar, M. Bennett, and D. L. Longo. 1055. 1995. NK cell subsets in the regulation of murine hematopoiesis. I. 5E6ϩ NK 61. Tanaka, T., F. Kitamura, Y. Nagasaka, K. Kuida, H. Suwa, and M. Miyasaka. cells promote hemopoietic growth in H-2d strain mice. J. Immunol. 155:2911– 1993. Selective long-term elimination of natural killer cells in vivo by an anti- 2917. Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021