Costimulation of Multiple NK Cell Activation Receptors by NKG2D Emily L. Ho, Leonidas N. Carayannopoulos, Jennifer Poursine-Laurent, Jeremy Kinder, Beatrice Plougastel, This information is current as Hamish R. C. Smith and Wayne M. Yokoyama of September 27, 2021. J Immunol 2002; 169:3667-3675; ; doi: 10.4049/jimmunol.169.7.3667 http://www.jimmunol.org/content/169/7/3667 Downloaded from

References This article cites 71 articles, 36 of which you can access for free at: http://www.jimmunol.org/content/169/7/3667.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

*average by guest on September 27, 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 © 2002 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Costimulation of Multiple NK Cell Activation Receptors by NKG2D1

Emily L. Ho, Leonidas N. Carayannopoulos, Jennifer Poursine-Laurent, Jeremy Kinder, Beatrice Plougastel, Hamish R. C. Smith, and Wayne M. Yokoyama2

The activation of NK cells is mediated through specific interactions between activation receptors and their respective ligands. Little is known, however, about whether costimulation, which has been well characterized for activation, occurs in NK cells. To study the function of NKG2D, a potential NK costimulatory receptor, we have generated two novel hamster mAbs that recognize mouse NKG2D. FACS analyses demonstrate that mouse NKG2D is expressed on all C57BL/6 IL-2-activated NK (lymphokine- activated killer (LAK)) cells, all splenic and liver NK cells, and ϳ50% of splenic NKT cells. Consistent with limited polymorphism of NKG2D, its sequence is highly conserved, and the anti-NKG2D mAbs react with NK cells from a large number of different mouse strains. In chromium release assays, we show that stimulation of NK cells with anti-NKG2D mAb can redirect lysis. Also, Downloaded from enhanced lysis of transfected tumor targets expressing NKG2D ligand could be inhibited by addition of anti-NKG2D mAb. Interestingly, stimulation of LAK cells via NKG2D alone does not lead to release. However, stimulation of LAK via both an NK activation receptor (e.g., CD16, NK1.1, or Ly-49D) and NKG2D leads to augmentation of cytokine release compared with stimulation through the activation receptor alone. These results demonstrate that NKG2D has the ability to costimulate multiple NK activation receptors. The Journal of Immunology, 2002, 169: 3667Ð3675. http://www.jimmunol.org/ atural killer cells constitute a subset of lymphocytes that inhibitory NK receptors share many functional characteristics. contributes to the innate immune response against intra- Where defined, they interact with MHC class I on the target, and N cellular (1, 2) and may mediate tumor sur- this interaction prevents activation of NK effector function by elic- veillance (3). Unlike B cells, which produce Abs, NK cells use iting a potent inhibitory signal mediated by an immunoreceptor effector mechanisms similar to those of T cells. They engage in tyrosine-based inhibitory motif (ITIM) in the cytoplasmic domain. cell-mediated cytolysis by releasing granules containing perforin When NK inhibitory receptors are engaged, the ITIM is tyrosine and granzymes that result in target cell apoptosis (4), and secrete phosphorylated and SHP-1 phosphatase is recruited and activated, such as IFN-␥, GM-CSF, and TNF-␣ that can modulate which presumably then dephosphorylates signaling molecules in- by guest on September 27, 2021 adaptive immune responses (5). volved in the activation cascade (14Ð17). NK cells were originally characterized (and named) for their Comparatively less is known about the nature of NK activation ability to lyse target cells without prior sensitization, i.e., natural receptors. Individual NK cells express multiple different activating killing (6, 7). It is now known that NK cells use specific receptors NK receptors of both Ig and lectin-like structural families that lack to mediate killing through the recognition of distinct ligands ex- cytoplasmic ITIMs (18Ð20). These receptors contain charged pressed on target cells (8). These NK receptors fall into two func- amino acids in their transmembrane domains, permitting interac- tional types, inhibitory and stimulatory (9). Much more is known tion with distinct signaling chains containing immunoreceptor ty- about the inhibitory receptors, which in turn can be organized into rosine-based activation motifs (ITAMs). NK cells express multiple two structural categories: killer Ig-like type I integral membrane ITAM signaling chains, including KARAP/DAP12, CD3␨, and receptors, encoded in the leukocyte receptor complex, and lectin- Fc⑀RI␥ (8). Ly-49D and Ly-49H associate with KARAP/DAP12 like type II integral membrane receptors, encoded in the NK (21, 22), NKR-P1 associates with Fc⑀RI␥ (23), and CD16 associ- complex (NKC)3 (10Ð13). Despite their structural differences, the ates with CD3␨ (24, 25) and Fc⑀RI␥ (26). Although it is not yet known whether these signaling chains are functionally equivalent, when activating receptors are cross-linked, the ITAMs in the as- Division of Rheumatology/Howard Hughes Medical Institute, Washington University sociated signaling chain become tyrosine phosphorylated. These School of Medicine and Barnes-Jewish Hospital, St. Louis, MO 63110 ITAMs then recruit downstream signaling kinases such as Syk and Received for publication February 12, 2002. Accepted for publication July 29, 2002. ZAP70, triggering activation of cytotoxic, proliferative, and/or se- 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 accordance cretory responses (27, 28). Thus, NK cell activation is similar to T with 18 U.S.C. Section 1734 solely to indicate this fact. cell signaling in that both cells use comparable signaling cascades. 1 This work was supported by the Barnes-Jewish Hospital Foundation, the T cell activation is a complex process that involves activation Research Institute, and grants from the National Institutes of Health to W.M.Y. E.L.H. through the TCR and a second receptor, a costimulatory receptor, is in the Medical Scientist Training Program. L.N.C. is a physician research fellow of the Howard Hughes Medical Institute. W.M.Y. is an Investigator of the Howard which by the simplest definition, potentiates signals from the TCR. Hughes Medical Institute. Experiments using anti-TCR Abs and APCs showed that stimulation 2 Address correspondence and reprint requests to Dr. Wayne M. Yokoyama, Division through TCR alone is insufficient for activation of naive T cells (29). of Rheumatology/Howard Hughes Medical Institute, Washington University School Full activation requires simultaneous engagement of the TCR and of Medicine, Box 8045, 660 South Euclid, St. Louis, MO 63110. E-mail address: [email protected] 3 Abbreviations used in this paper: NKC, NK gene complex; ITAM, immunoreceptor tyrosine-based activation motif; ITIM, immunoreceptor tyrosine-based inhibitory mo- early; sNKG2D, soluble NKG2D; LAK, lymphokine-activated killer; CHO, Chinese tif; PI3, phosphatidylinositol 3; PIG, pMX-IRES-GFP; RAE, retinoic acid inducible hamster ovary,; dhfr, dihydrofolate reductase.

Copyright © 2002 by The American Association of Immunologists, Inc. 0022-1767/02/$02.00 3668 COSTIMULATION BY NKG2D

costimulatory molecules, resulting in triggering of T cell effector func- were housed in a specific -free barrier facility at Washington tion, survival, and proliferation (30). The prototypical T cell costimu- University. latory molecule is CD28, which recognizes B7.1 and B7.2 (31, 32). Cells and cell lines CD28 recruits and activates various tyrosine-phosphorylated , Ϫ including kinases, Lck and Itk, and the adaptor , phospholipase Chinese hamster ovary (CHO)-dihydrofolate reductase (dhfr ) cells were C␥1 (33Ð35). CD28 also signals through a YxxM motif present in its previously described (54). hybridomas were grown in D10 medium (DMEM containing 10% FCS (Harlan Sprague Dawley, Indianapolis, IN), cytoplasmic domain, which is a consensus binding site for the p85 2 mM glutamine, 100 ␮M nonessential amino acids, 1 mM sodium pyru- subunit of phosphatidylinositol 3 (PI3)-kinase (36Ð38). Thus, CD28 vate, 100 U/ml penicillin, 100 ␮g/ml streptomycin, and 100 ␮M 2-ME may enhance T cell activation through distinct biochemical pathways. (Life Technologies, Carlsbad, CA)) at 37¡Cin5%CO2. The LL/2 Lewis Like T cells, NK cells may also use costimulatory pathways, but lung carcinoma line and Daudi B cell line were purchased from American Type Culture Collection (ATCC, Manassas, VA) and grown in until recently, it has not been feasible to dissect costimulation on D10 medium. The YAC-1 lymphoma line was obtained from ATCC, and NK cells because the nature of activation receptors, the molecules the IL-3-producing X63Ag8-653 was kindly provided by H. Karasuyama responsible for primary activation, was unclear. For example, the (Tokyo, Japan). Both cell lines were grown in R10 medium (RPMI con- T cell costimulatory receptor, CD28, and its ligands, B7.1 and taining 10% FCS, 2 mM glutamine, 100 U/ml penicillin, 100 ␮g/ml strep- ␮ B7.2, appear to play a role in NK cell activation against tumor tomycin, and 100 M 2-ME). Ba/F3 cells obtained from M. Thomas (Washington University) were grown in D10 containing 10% supernatant targets (39, 40). Despite this, it was unclear which activation re- from spent X63Ag8-653 cell cultures. Sf9 insect cells were kindly provided ceptor on an NK cell was costimulated by CD28. by G. Broze (Washington University) and grown in Grace’s medium (Life One recently characterized activation molecule expressed on Technologies) supplemented with 10% FCS. Hi5 insect cells were obtained both human and mouse NK cells is the lectin-like NKG2D mole- from J. Sadler (Washington University) and grown in Express5 medium Downloaded from (Life Technologies) with 9 mM glutamine. cule (41Ð44). Human NKG2D recognizes the stress-inducible Adult and neonatal splenocytes and thymocytes were prepared by pass- MIC family (MICA and MICB) (43). In addition, it interacts with ing spleen and thymus tissues through a 70-␮m cell filter (BD Biosciences, the ULBP family of molecules (ULBP-1, -2, -3), which also bind Lexington, KY) and lysing the RBCs, as described previously (55). For Ϫ Ϫ the CMV-encoded UL16 molecule (45). Mouse NKG2D binds the IL-2-activated NK cells from A/J, B6, BALB/cJ, CD16 / , and DBA/2J minor histocompatibility molecule H60, as well as the retinoic mice, nylon wool nonadherent cells were grown in R10 medium with 1000 U/ml human rIL-2 (Chiron, Emeryville, CA) for 3Ð4 days. Plastic nonad- ␣ ␤ ␥ ␦ ⑀ http://www.jimmunol.org/ acid-inducible RAE1 family (RAE1 ,- ,- ,- ,-) (44, 46, 47). herent cells were discarded, and the remaining adherent cells were cultured Mouse and human NKG2D trigger NK cytolytic killing of target in 1000 U/ml IL-2 for another 3Ð4 days before being harvested with EDTA cells transfected to express NKG2D ligands. (Versene; Life Technologies). The resulting culture resulted in an NK cell Ϫ ϩ ϩ NKG2D, however, may not directly stimulate NK cell killing population that was typically 90% CD3 and NK1.1 or DX5 . because in previous studies other activating receptors may have CDNAs, expression constructs, and transfections contributed to target killing. Moreover, unlike other activating NK receptors, NKG2D does not appear to signal through ITAM-con- The NKG2D cDNA was cloned as previously described (42). The ex- pressed sequence tag encoding for RAE1␦ was obtained from Incyte taining signaling chains. Instead, it selectively pairs with KAP/ Genomics (St. Louis, MO; accession AI785864) and cloned into the pMX- DAP10, which can recruit the p85 subunit of PI3-kinase through IRES-GFP (PIG) retroviral vector (kindly provided by T. Kitamura, Uni- the YxxM motif contained in its intracellular domain (48, 49). This versity of Tokyo, Tokyo, Japan). Full-length RAE1␦ was amplified using by guest on September 27, 2021 signaling motif is related to a similar functional motif contained in oligonucleotides (BglII-containing 5Ј-GGAAGAGATCTACCATGGCCA Ј Ј the cytoplasmic domain of CD28, suggesting NKG2D may co- AGGCAGCAGTGA-3 and XhoI-containing 5 -GGAAACTCGAGTCA CATTGCAAATGCAAATGCAA-3Ј) and cloned into PIG using BamHI stimulate rather than directly activate effector function. Recent and XhoI. PLAT-E cells (obtained from T. Kitamura, University of Tokyo studies indicate that human T cells may use NKG2D for costimu- (56)) were transfected with the retroviral vector using FuGENE6 (Roche, lation (43, 50), and biochemical results on human KARAP/DAP12 Basel, Switzerland), and the retroviral supernatants obtained were used to and KAP/DAP10 have suggested that these two signaling chains infect Ba/F3 cells. The resulting cells were analyzed by flow cytometry for green fluorescent protein expression and cloned by limiting dilution. may function cooperatively (51). However, other studies on The chimeric Ly-49A-NKG2D cDNA was created using PCR by am- NKG2D suggest that it is a primary activation receptor (47, 52, plifying the intracellular and transmembrane domains of Ly-49A using the 53). In addition, whether the signal from NKG2D can cooperate primers (5Ј-CTTTGCGGCCGCTCAATGAGGGAATTTATCCAG-3Ј and with the signals originating from NK-activating receptors associ- 5Ј-GGAATTCTATGAGTGAGCAGGAGGTCAC-3Ј). The extracellular Ј ated with signaling chains other than KARAP/DAP12 is as yet domain of NKG2D was amplified using sequence-specific primers (5 - GAGAGCGGCCGCTTACACCGCCCTTTTCATGCAGA-3Ј and 5Ј- unknown. TAGGAATTCTATGGCATTGATTCGTGATCG-3Ј). Both PCR frag- To further study the function of NKG2D, we generated two ments were then sequentially cloned into pBluescript (Stratagene, La Jolla, mAbs that specifically recognize mouse NKG2D expressed by all CA) using SalI and EcoRI, and EcoRI and XbaI, respectively. The resulting NK cells and a subset of NKT cells. Cross-linking of NKG2D by chimeric molecule was sequenced on both strands and removed from pBluescript using SalI and XbaI restriction enzymes and ligated into ligand-transfected cells stimulates target killing, which can be pH␤AP-r-1-neo-dhfr vector (54). This chimeric construct was transfected blocked by anti-NKG2D. Importantly, immobilized anti-NKG2D into CHO-dhfrϪ cells using lipofectamine (Life Technologies), and indi- does not directly stimulate NK cell activation unless another re- vidual clones were obtained by limiting dilution. Protein expression of the ceptor is coligated, including CD16, Ly-49D, or NKR-P1. These chimeric molecule by CHO clones was detected by Western blot analysis results indicate that NKG2D functions as a costimulatory molecule of cell lysates obtained from each individual cell clone using the 1298 rabbit antisera, which recognizes the Ly-49A cytoplasmic domain (57). on NK cells for NK activation receptors associated with different CHO clone F11 (CHO.F11) expresses a high level of the chimeric Ly- ITAM-containing signaling chains. 49A-NKG2D molecule, while clone E7 (CHO.E7) does not.

Isolation by RT-PCR and sequencing of NKG2D cDNA alleles Materials and Methods Total RNA was prepared from lymphokine-activated killer (LAK) cells Animals using TRIzol (Invitrogen, San Diego, CA), according to the manufacturer’s instructions, and was quantitated by UV spectrophotometer. RT-PCR was C57BL/6 (B6), A, BALB/c, DBA/2, and B6.129P2S-FCGR3TM1SJV conducted using the Superscript One-Step RT-PCR (Life Technolo- (CD16Ϫ/Ϫ; C57BL/6 background) mice were purchased from The Jackson gies), according to the manufacturer’s instructions, using the following Laboratory (Bar Harbor, ME). Armenian hamsters were purchased from NKG2D-specific primers: EcoRI-containing 5Ј-TAGGAATTCTATGGC Cytogen Research and Development (West Roxbury, MA). All animals ATTGATTCGTGATCG-3Ј and NotI-containing 5Ј-GAGAGCGGCCGCT The Journal of Immunology 3669

TACACCGCCCTTTTCATGCAGA-3Ј. The PCR products were visual- 51Cr release assay and redirected lysis assay ized on an agarose gel, purified using the Qiaquick PCR Purification kit 51 51 (Qiagen, Valencia, CA), digested with restriction enzymes EcoRI and NotI, Cr release assay was performed using Cr-labeled targets, as previously described (55). YAC-1 tumor targets were used as positive controls for and cloned into pBluescript SK II (Stratagene). The DNA was transformed 51 into XL-1 Blue (Stratagene), and multiple individual colonies were picked killing. Redirected lysis assay using Cr-labeled Daudi cells were per- for further analysis. The DNA was prepared using the Spin Miniprep kit formed as previously described (57). Purified Abs were used at a final ␮ ␮ 51 (Qiagen) and sequenced using ABI Bigdye, version 3 (Applied Biosys- concentration of 10 g/ml (redirected lysis) and 30 g/ml ( Cr release) tems, Foster City, CA). and incubated with effector cells for 20 min before adding targets. Cytokine release assay Production of soluble NKG2D (sNKG2D) Stimulation of B6 and CD16Ϫ/Ϫ LAK cells was performed as previously The entire extracellular domain of mouse NKG2D was cloned into the described (61). Briefly, 96-well Immulon-2 plates (Dynatech Laboratories, pFASTBAC vector using PCR and the sequence-specific primers (EcoRI- Chantilly, VA) were coated with 20 ␮g/ml streptavidin (Sigma-Aldrich) containing 5Ј-CGGAATTCGGATCTATGCTACT-3Ј,5Ј-ATGGAGCTG overnight at 4¡C. Triplicate wells were then coated with biotinylated pu- CGGGACGGCAGTCCAGTATTGTGCAACAAGGAA-3Ј,5Ј-TTCCTT rified Ab and incubated at room temperature for 30 min. Blocking buffer GTTGCACAATACTGGACTGCCGTCCCGCAGCTCCAT-3Ј, and KpnI- (2% BSA (V fraction) in PBS; Sigma-Aldrich) was added and incubated containing 5Ј-GGGGTACCTTTTTACACCGCCCTAAACAT-3Ј). The re- for1hatroom temperature. Wells were washed three times with PBS. A sulting construct contains the gp67 secretion signal, a 6-His tag, and a BirA total of 2 ϫ 105 washed LAK cells resuspended in R10 was added to each recognition domain upstream from the intact NKG2D extracellular domain. well (200 ␮l final vol) and incubated at 37¡C for 5Ð6 h. For anti-CD16- Sf9 insect cells were transfected with this construct using Cellfectin (Life blocking studies, cells were first incubated with purified soluble 2.4G2 Technologies). Primary NKG2D baculovirus was produced in Sf9 insect mAb (10 ␮g/ml) for 15 min. For PMA/ionomycin stimulation, cells were cells using the Bac-to-Bac system (Life Technologies), and sNKG2D was incubated with PMA at 10 ng/ml and ionomycin at 0.5 ␮g/ml final con- produced in Hi5 insect cells using secondarily amplified baculovirus. centration. Culture supernatants were harvested, diluted 1/3, and assayed Downloaded from sNKG2D was purified from Hi5 insect cell supernatant using a Ni-NTA for GM-CSF by ELISA (Endogen, Woburn, MA; R&D Systems, Minne- column (Qiagen). This sNKG2D was used to immunize Armenian apolis, MN), according to manufacturer’s instructions. In each experiment, hamsters. ELISA standard curves were derived using recombinant cytokines. Subsequently, a second sNKG2D was constructed (58). Briefly, this molecule is identical with the first described sNKG2D with a factor Xa Bioinformatics from Celera Genomics database cleavage site inserted between the BirA recognition site and the NKG2D Mouse NKG2D sequences from C57BL/6 (42), 129 ϫ 1/SvJ, DBA/J, and extracellular domain sequence. This sNKG2D was biotinylated with N- http://www.jimmunol.org/ hydroxysuccinimide-biotin (Pierce, Rockford, IL), according to the man- A/J mouse strains were compared for putative NKG2D amino acid identity ufacturer’s instructions, tetramerized around streptavidin-PE, and used for between strains of mice. The data for the 129 ϫ 1/SvJ strain were gener- cell-staining purposes. ated through use of the Celera Discovery System and Celera Genomics’ associated databases (Celera Genomics, Rockville, MD; http://celera.com). Production of anti-NKG2D Abs Results Armenian hamsters were inoculated i.p. with 20Ð100 ␮g sNKG2D emul- Generation of mAbs specific for mouse NKG2D sified in synthetic trehalose dicorynomycolate-monophosphoryl lipid A ad- juvant (Corixa, Seattle, WA) and boosted three times at 3-wk intervals. To generate reagents that would aid analysis of mouse NKG2D, Splenic B cell hybridomas were produced by the Washington University we produced soluble amino-terminal His-tagged NKG2D ectodo- by guest on September 27, 2021 Hybridoma Center from one Armenian hamster after FACS analysis main. The disulfide-linked dimers were purified by nickel affinity showed that antisera from this animal stained CHO.F11, but not CHO.E7 cells. Culture supernatants from ϳ300 hybridomas were screened by flow cytometry for Ab binding to a cell population containing a 1:3 mix of CHO.E7:CHO.F11 cells in each tube. Culture supernatants that shifted 75% of the cells were presumed to contain Ab that specifically recognized NKG2D. This screen resulted in 17 independent hybridomas, which un- derwent three rounds of subcloning. Two clones, A10 and C7, were se- lected for further analysis.

Antibodies The following Abs were purchased from PharMingen: anti-CD3 APC, anti- CD3 CyChrome, anti-CD3 PE, anti-CD3 PerCP, anti-CD4 PE, anti-CD8 PE, anti-CD44 FITC, anti-␥␦-FITC, APC PK136, PE PK136 (anti-NK1.1), and anti-TNP-1 biotin (used as hamster IgG control). Hamster IgG was purchased from ICN Biomedical, Costa Mesa, CA and used as Ab control. A10, C7, and 2.4G2 Abs were purified from spent D10 culture super- natants by ammonium sulfate precipitation (55% saturation) and affinity purification on a protein A-Sepharose column (Amersham Pharmacia Bio- tech, Piscataway, NJ) using standard methods (59). Purified Abs were bi- otinylated with N-hydroxysuccinimide-biotin (Pierce), according to the manufacturer’s instructions. FITC-2C11 (anti-CD3) mAb was produced using FITC-celite (FITC isomer 1; Calbiochem, La Jolla, CA) (60).

Flow cytometric analysis Adherent LAK cells, thymocytes, and splenocytes were washed three times in sorter buffer (HBSS without phenol red (Sigma-Aldrich, St. Louis, MO),

3% FCS, 0.05% NaN3). Thymocytes and splenocytes were then filtered through nylon mesh. Cells were incubated in spent 2.4G2 hybridoma cul- ture supernatant to block nonspecificFc␥RIII binding for 20 min before addition of Abs. For Ab staining, ϳ5Ð10 ϫ 105 cells were incubated with 2Ð10 ␮g/ml Abs for 20 min, washed twice, and then incubated with sec- ondary reagents, such as streptavidin-PE (0.2 ␮g/ml). Cells were gated FIGURE 1. Specificity of mAbs A10 and C7. A, CHO.F11, which ex- using forward and side scatter and cell marker staining. Dead cells were presses the chimeric Ly-49A/NKG2D molecule, stained with mAb A10 excluded by gating with propidium iodide staining. A total of 1Ð10 ϫ 104 and C7. B, mAb A10 staining of C57BL/6 IL-2-activated NK (LAK) cells. gated events was collected from each flow cytometry sample. Comparable staining was obtained with mAb C7 (data not shown). 3670 COSTIMULATION BY NKG2D chromatography (data not shown). Armenian hamsters were im- and a minor population of NK1.1ϪCD3ϩ cells by flow cytometry munized with sNKG2D, and subsequent antisera stained a stable using mAbs A10 and C7 (Fig. 2A). In the thymus, NKG2D is transfectant (CHO.F11) expressing the NKG2D ectodomain (data expressed on NK1.1ϩ thymocytes, but not on CD4ϩ or CD8ϩ not shown). Upon fusion and screening for FACS reactivity with single-positive cells, CD4ϩCD8ϩ double-positive cells, or ␥␦ T CHO.F11, we generated two novel mAbs, A10 and C7, which cells (data not shown). In addition, NKG2D expression on mouse specifically recognize the mouse NKG2D transfectant (Fig. 1A). NK cells can be detected as early as day 1 postpartum (Fig. 2B). These mAbs do not stain CHO cells expressing distantly related This is in contrast to Ly-49 receptors, which have been shown in lectin-like NK cell receptors CD94/NKG2A and CD94/NKG2C by previous studies (57) to be undetectable in neonatal NK cells wk 1 FACS analysis (data not shown). In addition, they compete with postpartum, suggesting that the function of NKG2D may differ each other for binding to CHO.F11, as assessed by FACS, indi- from that of Ly-49. cating that they may share similar epitopes or block each other by steric hindrance (data not shown). NKG2D is also expressed on C57BL/6 IL-2-activated (LAK) cells, indicating that the native conformation of NKG2D can be recognized by the A10 and C7 mAbs (Fig. 1B). These results are consistent with previously published data using polyclonal antisera that recognized NKG2D, which were performed without control- ling for potential Fc␥RIII-mediated (nonspecific) binding (62). Downloaded from Expression of NKG2D on all NK cells and a subpopulation of splenic NK T cells NKG2D is expressed on all splenic and liver NK cells (NK1.1ϩCD3Ϫ), ϳ50% of splenic NK T (NK1.1ϩCD3ϩ) cells, http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 3. NKG2D expression on NK cells from different strains of mice. A, Splenic NK cells (gated on DX5ϩCD3Ϫ), and B, IL-2-activated FIGURE 2. FACS analysis of NKG2D expression on C57BL/6 spleen NK cells from BALB/c mice stained with mAb A10. C, Alignment of and liver cells. A, FACS plot of C57BL/6 NK1.1ϩCD3Ϫ (NK) and putative NKG2D protein sequence translated from cDNA sequences of NK1.1ϩCD3ϩ (NKT) splenocytes stained with mAb A10. As indicated, LAK cells derived from C57BL/6 (42), A/J, and DBA/2J strains of mice the x-axes show control mAb and anti-NKG2D mAb staining, and the and from genomic sequences derived from the 129 ϫ 1/SvJ strain (Celera y-axes show NK1.1 staining. These results are representative of three in- Genomics database). The NKG2D transmembrane domain is indicated by dependent experiments. B, FACS analysis of day 1 and day 21 postpartum the line. The extracellular domain of this type II integral membrane recep- and adult C57BL/6 liver NK cells (NK1.1ϩCD3Ϫ) with mAb A10. tor spans residues 90Ð232. The Journal of Immunology 3671

Table I. Ab staining of splenic NK cells from various mouse strainsa obtained similar results using BALB/c LAKs, indicating that NKG2D also serves as an activating receptor in BALB/c that also Mouse Strain H-2 A10 PK136 4E4 3D10 DX-5 recognizes RAE1␦ (Fig. 4D). These results indicate differences in ϩϪϩϪϩ functional capacities of the anti-NKG2D mAbs. More importantly, 129/SVJ b ␦ A/NCR a ϩϪn/a n/a ϩ mAb C7 can specifically block the NKG2D-RAE1 interaction B10.BR k ϩϩϩϩϩ that otherwise enhances target killing. B10.HTG g ϩϩn/a n/a ϩ Many tumors constitutively express the ligands recognized by B10.RIII r ϩϩϩϩϩ NKG2D, as indicated by a sNKG2D tetramer and flow cytometry ϩϪϪϪϩ BALB/c d (data not shown) (44, 46). One such tumor is LL/2, which ex- C3H/HeJ k ϩϪϪϪϩ C57BL/6 b ϩϩϩϩϩ presses a high level of RAE1 and H60 on the cell surface. In a C57L b ϩϩn/a n/a ϩ chromium release assay, LL/2 is lysed well by C57BL/6 LAK DBA/2 d ϩϪϪϪϩ cells. This killing is significantly blocked by the addition of mAb FVB q ϩϩn/a n/a ϩ ϩϩϩϩϩ C7 (Fig. 5A). Comparable with the functional effects on Ba/F3- MA/My k ␦ NZB/B1N d ϩϩϩϩϩ RAE1 (Fig. 4C), the addition of mAb A10 does not block killing NZW/LacJ z ϩϩϩϩϩ compared with control Ab. Interestingly, the blocking effect with SWR q ϩϩϩϪϩ mAb C7 was not always observed with other tumors that also a NK cells were determined by their reactivity to DX5 and CD3 Ab. ϩ, Positive express RAE1, such as Mlg and YAC-1 (data not shown). Because staining; Ϫ, negative staining; n/a, not available. natural killing involves the interaction of many NK receptors (ac- tivation and inhibitory) with their respective ligands, the contribu- Downloaded from tion of NKG2D to tumor lysis may not always be readily apparent. Expression of NKG2D on NK cells from disparate inbred mouse Nevertheless, our data show that NKG2D does have a role in kill- strains ing by B6 LAKs of certain tumors that constitutively express In contrast to other signaling receptors on mouse NK cells, which NKG2D ligands. display significant allelic poymorphism, NKG2D expression was In redirected lysis assays, mAb A10 can stimulate NK cells to ϩ Ϫ found on BALB/c DX5 CD3 spleen cells (Fig. 3A) and LAKs kill Daudi targets (Fig. 5B). Interestingly, mAb C7 failed to stim- http://www.jimmunol.org/ (Fig. 3B), suggesting that NKG2D is expressed by all NK cells in ulate redirected lysis, again indicating different functional proper- this strain. Furthermore, the anti-NKG2D mAbs reacted with ties of the anti-NKG2D mAbs. Taken together, these data indicate ϩ Ϫ DX5 CD3 cells from a wide variety of other strains (Table I). It that NKG2D can contribute to NK cell activation in target killing. is unlikely that the mAbs are reacting with different molecules in these strains because there is only one NKG2D gene in the mouse NKG2D has costimulatory function genome (42) and there is minimal allelic polymorphism of These data and those of published studies on NKG2D function in NKG2D among C57BL/6, 129 ϫ 1/SvJ, A/J, BALB/cJ, and NK cells do not distinguish between direct activation or costimu- DBA/2J mouse strains (Fig. 3C and see below). From the full- lation of NK cells by NKG2D recognition of ligand. To address length NKG2D sequences derived from 129 ϫ 1/SvJ, A/J, BALB/ this issue, we stimulated NK cells on immobilized anti-NKG2D by guest on September 27, 2021 cJ, and DBA/2J, we observed little or no amino acid differences in mAb. Cytokine release assays revealed that stimulation via the putative NKG2D protein sequence. For instance, the 129 ϫ NKG2D using immobilized mAb A10 led to production of GM- 1/SvJ NKG2D sequence is identical with the C57BL/6 sequence. CSF by KY-1 NK clones and C57BL/6 LAK cells (data not shown In A/J, BALB/cJ, and DBA/2J, there are amino acid changes at and Fig. 6A). Comparable with the redirected lysis assays, neither residue 10, in which a tyrosine is substituted for a histidine, and at immobilized mAb C7 nor isotype controls stimulated cytokine residue 71, in which a leucine is substituted for a valine. Because production (data not shown), demonstrating the specificity of the there is minimal polymorphism in the putative cytoplasmic and mAb A10 effect. However, the addition of affinity-purified 2.4G2 transmembrane domains (residues 10 and 71, respectively), and no mAb inhibited mAb A10 stimulation of B6 LAKs, suggesting that amino acid changes were observed in the extracellular domains, stimulation by A10 may require concomitant CD16 (Fc␥RIII) en- these data indicate that NKG2D is highly conserved between in- gagement. To test this hypothesis, we were not able to produce Ј bred strains of mice. These results strongly suggest that the mol- F(ab )2 because digestion of hamster Ig is problematic. However, ecule reacting with the mAb A10 is likely to be NKG2D for all of when we stimulated LAKs from CD16Ϫ/Ϫ mice with immobilized the mouse strains tested and that NKG2D functions similarly in anti-NKG2D alone, we failed to induce GM-CSF production (Fig. different strains (see below). 6A). This result was not due to the inability of CD16Ϫ/Ϫ LAKs to be stimulated because the level of GM-CSF production stimulated Activation function of NKG2D by PMA and ionomycin was comparable for both B6 and To explore the capacity of the anti-NKG2D mAbs to block CD16Ϫ/Ϫ LAKs. Instead, these results suggest that activation of NKG2D recognition of its ligands, we infected Ba/F3 cells with KY-1 and B6 LAK cells with mAb A10 alone was due to costimu- retrovirus directing expression of RAE1␦ (Ba/F3-RAE1␦). Stain- lation of CD16 activation by NKG2D. ing with a sNKG2D tetramer revealed that the infected Ba/F3 cells To extend these observations, we examined the effect of express a high level of RAE1␦, while uninfected Ba/F3 and Ba/F3 NKG2D cross-linking on stimulation through other NK cell acti- cells expressing vector only (Ba/F3-PIG) showed nonspecific vation receptors. To avoid inadvertent FcR effects, we also used staining (Fig. 4A). The surface expression of RAE1␦ on Ba/F3 CD16Ϫ/Ϫ NK cells in these assays. The level of GM-CSF mea- resulted in a significant increase in susceptibility to lysis compared sured with the addition of both mAbs PK136 and A10 was signif- with untransfected Ba/F3 and Ba/F3-PIG in a 51Cr release cyto- icantly higher compared with stimulation with mAb PK136 alone toxicity assay (Fig. 4B). The addition of mAb C7 blocked lysis of (Fig. 6B). In contrast, addition of mAb C7 with mAb PK136 did Ba/F3-RAE1␦ to levels below that of Ba/F3-PIG by C57BL/6 not lead to cytokine production by CD16Ϫ/Ϫ LAKs, again reflect- LAK cells, implying mAb C7 may be interfering with the inter- ing differences between the anti-NKG2D mAbs in terms of func- action of another ligand recognized by NKG2D. Meanwhile, mAb tional effects in these assays and ruling out a nonspecific effect A10 did not affect lysis compared with control Ab (Fig. 4C). We simply due to mAb binding to NK cells. In addition, similar results 3672 COSTIMULATION BY NKG2D

FIGURE 4. RAE1␦ activates NK cytolysis of Ba/F3 tumor targets via NKG2D. A, sNKG2D tetramer stain- ing (10 ␮g/ml) of untransfected Ba/F3 cells, Ba/F3 cells transfected with vector only (Ba/F3-PIG), and Ba/F3 transfected with RAE1␦ (Ba/F3-RAE1␦). For each FACS plot, the dark shading indicates unstained cells,

the line indicates staining with streptavidin-PE Downloaded from only, and the heavy line indicates staining with sNKG2D streptavidin-PE tetramer. B,Ina51Cr release assay using C57BL/6 LAKs as effector cells, expression of RAE1␦ on Ba/F3 cells increased lysis compared with that of untransfected Ba/F3 and Ba/F3-PIG. C, This in- crease in lysis was inhibited by the addition of mAb C7,

but not mAb A10. D, BALB/c LAKs also recognize http://www.jimmunol.org/ RAE1␦ expressed on Ba/F3 cells in a 51Cr release assay. by guest on September 27, 2021

were obtained in analysis of Ly-49D cross-linking using mAb 4E4. are coupled to KARAP/DAP12, CD3␨, and Fc⑀RI␥. Thus, these Stimulation of CD16Ϫ/Ϫ LAK cells using mAbs 4E4 and A10 led data strongly suggest that NKG2D may potentiate signals deliv- to significantly higher levels of GM-CSF production compared ered by all known NK cell-expressed ITAM-associated signaling with stimulation with 4E4 alone. Thus, our data indicate that chains. Together with the expression of NKG2D on virtually all NKG2D synergizes with various activating NK receptors such as NK cells, our data suggest that NKG2D represents a global co- NKR-P1C, Ly-49D, and CD16, to produce a potent stimulatory stimulatory receptor for NK cells. signal. In terms of the role of this costimulatory function in natural killing, we found that the C7 anti-NKG2D mAb blocks enhanced Discussion Ba/F3 target killing by transfection of the NKG2D ligand RAE1␦. To fully activate T cells, signals must be delivered through the T In addition, C7 partially blocks the killing of the LL/2 tumor target cell Ag receptor and the costimulatory molecule, CD28 (35). The that constitutively expresses NKG2D ligands. These data suggest TCR is coupled to ITAM-associated signaling chains, which re- that NKG2D synergizes with the action of a primary NK cell ac- cruit and activate the Syk family tyrosine kinase ZAP-70. The tivation receptor in the killing of Ba/F3-RAE1␦ and LL/2 tumors. relatively nonpolymorphic receptor, CD28, through recruitment of In contrast, because mAb C7 did not always block lysis of other PI3-kinase, provides additional costimulatory signals through the tumor targets expressing RAE1, it is possible that these particular Akt pathway. In this study, we have determined that all mature NK tumors express ligands to trigger other primary activation NK re- cells express the NKG2D receptor and that NK cell activation is ceptors. Alternatively, in light of the known function of CD28 on potentiated by costimulation through NKG2D. In contrast to T T cells, CD28 may also costimulate NK cell activation, particu- cells, NK cells express multiple different ITAM signaling chains, larly because previous work has shown that it is expressed on all and both Syk family tyrosine kinases (ZAP-70 and Syk). Whereas C57BL/6 NK cells (63). Although the relationship of NKG2D to the contributions of the Syk family tyrosine kinases are incom- CD28 remains to be defined, this may explain why mAb C7 only pletely understood, we have shown in this work that NKG2D can partially blocks LL/2 tumor lysis. Nevertheless, our data show that costimulate signals delivered by NK cell activation receptors that NKG2D is involved in the natural killing of some tumor targets The Journal of Immunology 3673 Downloaded from http://www.jimmunol.org/

FIGURE 5. NKG2D has activating function on NK cells. A, Lysis of RAE1-expressing LL/2 tumor target can be partially inhibited with mAb C7. B, Daudi targets can be lysed by mAb A10, but not mAb C7 in a redirected lysis assay.

even though it is not yet known what primary NK cell activation ecules, which include several family members and are among the receptor confers specificity. most polymorphic molecules encoded in the genome. The functional differences we observed between mAbs A10 and Strikingly, despite the evolving polymorphism that character- C7 are intriguing. mAb A10 can trigger NKG2D, while mAb C7 izes the NKC genetic regions flanking the NKG2D gene (M. G. by guest on September 27, 2021 cannot, and mAb C7 can block NKG2D recognition of RAE1␦, Brown, L. Nadler, A. A. Scalzo, and W. M. Yokoyama, manu- while mAb A10 cannot. These differences could be due to several script in preparation), there is little polymorphism observed for factors. For example, the epitope recognized by mAb C7 may be NKG2D. NKG2D does not appear to belong to a family of related similar to the one used by NKG2D to interact with RAE1␦.In molecules because there is evidence for only one NKG2D gene by addition, it is possible that mAb A10 recognizes a triggering Southern analysis of genomic clones from the NKC and by bioin- epitope on NKG2D or perhaps is a better cross-linker of NKG2D formatics analysis of the Celera Genomics database. Between molecules. Further studies mapping the epitope domains recog- strains, NKG2D shows minimal amino acid sequence differences nized by these two Abs should elucidate the functional differences between four mouse strains. These molecules most likely represent between mAbs A10 and C7. alleles with the same function because of conservation of the trans- In contrast to other mouse NK gene complex-encoded lectin- membrane and cytoplasmic domains. The charged amino acid re- like receptors, NKG2D is unique in that it is relatively nonpoly- sponsible for NKG2D interaction with its signaling chain DAP10 morphic. Furthermore, it does not belong to the NKG2 family of (arginine at residue 69) is conserved in all four mouse strains. molecules, despite its name. It forms a homodimer, whereas the Finally, our mAbs react with all NK cells from all strains with other NKG2 molecules heterodimerize with CD94 (64Ð66). It has similar levels of expression, and the extracellular domain se- just 25% amino acid identity to these molecules, comparable with quences are identical. Thus, NKG2D is unique among NKC-en- the relationship of NKRP1 molecules with the Ly-49 family or the coded molecules in showing little polymorphism. general relationship of NKC-encoded molecules to the lectin su- With respect to its own ligands, the limited polymorphism of perfamily. Within a family, there is ϳ85% amino acid identity. NKG2D is somewhat surprising because the ligands for NKG2D For other NKC-encoded molecules, this homology can lead to are numerous and heterogeneous, and display significant allelic significant serologic differences between mouse strains. mAbs spe- polymorphism. NKG2D binds MICA, MICB, and several ULBP cific for a given molecule in one strain may react with a function- molecules, whereas mouse NKG2D binds H60, and several RAE1 ally different family member in another mouse strain, as well as family members (45, 46, 52, 70). Other than an apparent null allele allelic forms. For example, Abs against activation receptors of H60, significant allelic polymorphisms of its ligands have not NKRP1C and Ly-49D in C57BL/6 mice react with related inhib- been described in the mouse (71). By contrast, the human MICA itory receptors in other mouse strains (67, 68), while Abs against and MICB are the most highly polymorphic of the MHC inhibitory Ly-49A in C57BL/6 react with related activation recep- class Ib loci (72, 73). Although there are differences in ligand- tors in C57L and 129/J (68, 69). These studies highlight the dis- binding affinities to mouse NKG2D, the affinities do not correlate tinct repertoire of NK receptors expressed within a strain as well as with ligand sequence relationships (58). One possible explanation allelic polymorphisms between strains. This is not surprising given may be that NKG2D possesses multiple ligand-binding sites. This that the known ligands for these receptors are MHC class I mol- has been suggested by the recent crystal structure of NKG2D (74). 3674 COSTIMULATION BY NKG2D

6. Kiessling, R., E. Klein, and H. Wigzell. 1975. Natural killer cells in the mouse. I. Cytotoxic cells with specificity for mouse Moloney cells: specificity and distribution according to genotype. Eur. J. Immunol. 5:112. 7. Herberman, R. B., M. E. Nunn, and D. H. Lavrin. 1975. Natural cytotoxic reac- tivity of mouse lymphoid cells against syngeneic and allogeneic tumors. I. Dis- tribution of reactivity and specificity. Int. J. Cancer 16:216. 8. Lanier, L. L. 2000. Turning on natural killer cells. J. Exp. Med. 191:1259. 9. Lanier, L. L. 1997. Natural killer cells: from no receptors to too many. Immunity 6:371. 10. Ciccone, E., D. Pende, O. Viale, A. Than, C. Di Donato, A. M. Orengo, R. Biassoni, S. Verdiani, A. Amoroso, A. Moretta, and L. Moretta. 1992. In- volvement of HLA class I alleles in natural killer (NK) cell-specific functions: expression of HLA-Cw3 confers selective protection from lysis by alloreactive NK clones displaying a defined specificity (specificity 2). J. Exp. Med. 176:963. 11. Moretta, A., M. Vitale, C. Bottino, A. M. Orengo, L. Morelli, R. Augugliaro, M. Barbaresi, E. Ciccone, and L. Moretta. 1993. P58 molecules as putative re- ceptors for major histocompatibility complex (MHC) class I molecules in human natural killer (NK) cells: anti-p58 antibodies reconstitute lysis of MHC class I-protected cells in NK clones displaying different specificities. J. Exp. Med. 178:597. 12. Colonna, M., and J. Samaridis. 1995. Cloning of immunoglobulin-superfamily members associated with HLA-C and HLA-B recognition by human natural killer cells. Science 268:367. 13. Karlhofer, F. M., R. K. Ribaudo, and W. M. Yokoyama. 1992. MHC class I alloantigen specificity of Ly-49ϩ IL-2-activated natural killer cells. Nature Downloaded from 358:66. 14. Burshtyn, D. N., A. M. Scharenberg, N. Wagtmann, S. Rajagopalan, K. Berrada, T. Yi, J. P. Kinet, and E. O. Long. 1996. Recruitment of tyrosine phosphatase HCP by the killer cell inhibitor receptor. Immunity 4:77. 15. Campbell, K. S., M. Dessing, M. Lopez-Botet, M. Cella, and M. Colonna. 1996. Tyrosine phosphorylation of a human killer inhibitory receptor recruits protein tyrosine phosphatase 1C. J. Exp. Med. 184:93. 16. Nakamura, M. C., E. C. Niemi, M. J. Fisher, L. D. Shultz, W. E. Seaman, and http://www.jimmunol.org/ J. C. Ryan. 1997. Mouse Ly-49A interrupts early signaling events in cytotoxicity and functionally associates with the Shp-1 tyrosine phosphatase. J. Exp. Med. 185:673. 17. Vely, F., S. Olivero, L. Olcese, A. Moretta, J. E. Damen, L. Liu, G. Krystal, J. C. Cambier, M. Daeron, and E. Vivier. 1997. Differential association of phos- phatases with hematopoietic co-receptors bearing immunoreceptor tyrosine-based FIGURE 6. NKG2D costimulates CD16, NK1.1, and Ly-49D activa- inhibition motifs. Eur. J. Immunol. 27:1994. tion. A, C57BL/6 LAKs are stimulated by plate-bound mAb A10 (1 ␮g/ml 18. Biassoni, R., C. Cantoni, M. Falco, S. Verdiani, C. Bottino, M. Vitale, R. Conte, final concentration). This activation is inhibited by addition of soluble anti- A. Poggi, A. Moretta, and L. Moretta. 1996. The human leukocyte antigen (HLA)-C-specific “activatory” or “inhibitory” natural killer cell receptors display CD16 (2.4G2) mAb. B, CD16Ϫ/Ϫ LAKs are not activated with A10 or C7 highly homologous extracellular domains but differ in their transmembrane and by guest on September 27, 2021 Ab alone (5 ␮g/ml). Addition of mAb PK136 (anti-NK1.1) with mAb A10 intracytoplasmic portions. J. Exp. Med. 183:645. (5 ␮g/ml) leads to GM-CSF production at a significantly higher level com- 19. Bottino, C., S. Sivori, M. Vitale, C. Cantoni, M. Falco, D. Pende, L. Morelli, pared with that of PK136 or PK136 with control Ab alone. mAb 4E4 R. Augugliaro, G. Semenzato, R. Biassoni, et al. 1996. A novel surface molecule Ϫ/Ϫ homologous to the p58/p50 family of receptors is selectively expressed on a (anti-Ly-49D)-mediated activation of CD16 LAKs is also augmented subset of human natural killer cells and induces both triggering of cell functions by addition of mAb A10, but not by mAb C7 or control mAb. For this and proliferation. Eur. J. Immunol. 26:1816. particular experiment, the GM-CSF produced by cells stimulated with con- 20. Smith, H. R. C., F. M. Karlhofer, and W. M. Yokoyama. 1994. Ly-49 multigene trol mAb alone in the PK136 assay (left panel) was unusually high. Usu- family expressed by IL-2-activated NK cells. J. Immunol. 153:1068. ally, control mAb leads to GM-CSF production at background or below (as 21. Smith, K. M., J. Wu, A. B. Bakker, J. H. Phillips, and L. L. Lanier. 1998. Cutting seen in the 4E4 assay (right panel)). edge: Ly-49D and Ly-49H associate with mouse DAP12 and form activating receptors. J. Immunol. 161:7. 22. Gosselin, P., L. H. Mason, J. Willette-Brown, J. R. Ortaldo, D. W. McVicar, and S. K. Anderson. 1999. Induction of DAP12 phosphorylation, calcium mobiliza- Each one of these domains may provide an evolutionary advan- tion, and cytokine secretion by Ly49H. J. Leukocyte Biol. 66:165. 23. Arase, N., H. Arase, S. Y. Park, H. Ohno, C. Ra, and T. Saito. 1997. Association tage, again highlighting the functional importance of NKG2D. with FcR-␥ is essential for activation signal through NKR-P1 (CD161) in natural killer (NK) cells and NK1.1ϩ T cells. J. Exp. Med. 186:1957. Acknowledgments 24. Lanier, L. L., G. Yu, and J. H. Phillips. 1989. Co-association of CD3␨ with a receptor (CD16) for IgG Fc on human natural killer cells. Nature 342:803. We gratefully acknowledge Adrian Davies and Dr. Mark Soloski for help 25. Anderson, P., M. Caligiuri, C. O’Brien, T. Manley, J. Ritz, and S. F. Schlossman. in determining the specificities of the anti-NKG2D; Erika Blattenberger 1990. Fc␥ receptor type III (CD16) is included in the ␨ NK receptor complex and Chad Dubbelde for expert technical assistance; Kim Marlotte for an- expressed by human natural killer cells. Proc. Natl. Acad. Sci. USA 87:2274. imal care; Laura Belorgey for initial studies on NKG2D expression in 26. Wirthmueller, U., T. Kurosaki, M. S. Murakami, and J. V. Ravetch. 1992. Signal ␥ ␥ neonatal mice; and Drs. Marco Colonna, Susan Gilfillan, Jonathan Heusel, transduction by Fc RIII (CD16) is mediated through the chain. J. Exp. Med. 175:1381. and Sungjin Kim for helpful discussions and critical reading of the 27. Lanier, L. L., B. C. Cortiss, J. Wu, C. Leong, and J. H. Phillips. 1998. Immu- manuscript. noreceptor DAP12 bearing a tyrosine-based activation motif is involved in acti- vating NK cells. Nature 391:703. References 28. McVicar, D. W., L. S. Taylor, P. Gosselin, J. Willette-Brown, A. I. Mikhael, R. L. Geahlen, M. C. Nakamura, P. Linnemeyer, W. E. Seaman, S. K. Anderson, 1. Biron, C. A., K. S. Byron, and J. L. Sullivan. 1989. Severe herpesvirus et al. 1998. DAP12-mediated signal transduction in natural killer cells: a domi- in an adolescent without natural killer cells. N. Engl. J. Med. 320:1731. nant role for the syk protein-tyrosine kinase. J. Biol. Chem. 273:32934. 2. Ballas, Z. K., J. M. Turner, D. A. Turner, E. A. Goetzman, and J. D. Kemp. 1990. A patient with simultaneous absence of “classical” natural killer cells (CD3Ϫ, 29. Schwartz, R. H. 1996. Models of T cell anergy: is there a common molecular CD16ϩ, and NKH1ϩ) and expansion of CD3ϩ, CD4Ϫ, CD8Ϫ, NKH1ϩ subset. mechanism? J. Exp. Med. 184:1. J. Allergy Clin. Immunol. 85:453. 30. Bretscher, P. 1992. The two-signal model of lymphocyte activation twenty-one 3. Trinchieri, G. 1989. Biology of natural killer cells. Adv. Immunol. 47:187. years later. Immunol. Today 13:74. 4. Berke, G. 1994. The binding and lysis of target cells by cytotoxic lymphocytes: 31. Aruffo, A., and B. Seed. 1987. Molecular cloning of a CD28 cDNA by a high- molecular and cellular aspects. Annu. Rev. Immunol. 12:735. efficiency COS cell expression system. Proc. Natl. Acad. Sci. USA 84:8573. 5. Perussia, B. 1991. Lymphokine-activated killer cells, natural killer cells and cy- 32. June, C. H., J. A. Bluestone, L. M. Nadler, and C. B. Thompson. 1994. The B7 tokines. Curr. Opin. Immunol. 3:49. and CD28 receptor families. Immunol. Today 15:321. The Journal of Immunology 3675

33. August, A., S. Gibson, Y. Kawakami, T. Kawakami, G. B. Mills, and B. Dupont. NKG2D in tumor cell lysis mediated by human NK cells: cooperation with nat- 1994. CD28 is associated with and induces the immediate tyrosine phosphory- ural cytotoxicity receptors and capability of recognizing tumors of nonepithelial lation and activation of the Tec family kinase ITK/EMT in the human Jurkat origin. Eur. J. Immunol. 31:1076. leukemic T-cell line. Proc. Natl. Acad. Sci. USA 91:9347. 54. Wang, L. L., D. T. Chu, A. O. Dokun, and W. M. Yokoyama. 2000. Inducible 34. August, A., and B. Dupont. 1994. CD28 of T lymphocytes associates with phos- expression of the gp49B inhibitory receptor on NK cells. J. Immunol. 164:5215. phatidylinositol 3-kinase. Int. Immunol. 6:769. 55. Idris, A. H., K. Iizuka, H. R. C. Smith, A. A. Scalzo, and W. M. Yokoyama. 1998. 35. Chambers, C. A., and J. P. Allison. 1997. Co-stimulation in T cell responses. Genetic control of natural killing and in vivo tumor elimination by the Chok Curr. Opin. Immunol. 9:396. locus. J. Exp. Med. 188:2243. 36. August, A., and B. Dupont. 1994. Activation of src family kinase lck following 56. Morita, S., T. Kojima, and T. Kitamura. 2000. Plat-E: an efficient and stable CD28 crosslinking in the Jurkat leukemic cell line. Biochim. Biophys. Acta 199: system for transient packaging of retroviruses. Gene Ther. 7:1063. 1466. 57. Smith, H. R., H. H. Chuang, L. L. Wang, M. Salcedo, J. W. Heusel, and 37. Prasad, K. V., Y. C. Cai, M. Raab, B. Duckworth, L. Cantley, S. E. Shoelson, and W. M. Yokoyama. 2000. Nonstochastic coexpression of activation receptors on C. E. Rudd. 1994. T-cell antigen CD28 interacts with the lipid kinase phospha- murine natural killer cells. J. Exp. Med. 191:1341. tidylinositol 3-kinase by a cytoplasmic Tyr(P)-Met-Xaa-Met motif. Proc. Natl. 58. Carayannopoulos, L. N., O. V. Naidenko, J. Kinder, E. L. Ho, D. H. Fremont, and Acad. Sci. USA 91:2834. W. M. Yokoyama. 2002. Ligands for murine NKG2D display heterogeneous 38. Truitt, K. E., C. M. Hicks, and J. B. Imboden. 1994. Stimulation of CD28 triggers binding behavior. Eur. J. Immunol. In press. an association between CD28 and phosphatidylinositol 3-kinase in Jurkat T cells. 59. Harlow, E., and D. Lane. 1988. Antibodies: A Laboratory Manual. Cold Spring J. Exp. Med. 179:1071. Harbor Laboratory, Cold Spring Harbor. 39. Azuma, M., M. Cayabyab, D. Buck, J. H. Phillips, and L. L. Lanier. 1992. In- 60. Coligan, J. E., A. M. Kruisbeek, D. H. Margulies, E. M. Shevach, and W. Strober. volvement of CD28 in MHC-unrestricted cytotoxicity mediated by a human nat- 1996. Current Protocols in Immunology. Wiley, New York. ural killer leukemia cell line. J. Immunol. 149:1115. 61. Wang, L. L., J. Blasioli, D. R. Plas, M. L. Thomas, and W. M. Yokoyama. 1999. 40. Chambers, B. J., M. Salcedo, and H. G. Ljunggren. 1996. Triggering of natural Specificity of the SH2 domains of SHP-1 in the interaction with the immunore- killer cells by the costimulatory molecule CD80 (B7-1). Immunity 5:311. ceptor tyrosine-based inhibitory motif-bearing receptor gp49B. J. Immunol. 162: 41. Houchins, J. P., T. Yabe, C. McSherry, and F. H. Bach. 1991. DNA sequence 1318. analysis of NKG2, a family of related cDNA clones encoding type II integral 62. Diefenbach, A., A. M. Jamieson, S. D. Liu, N. Shastri, and D. H. Raulet. 2000. Downloaded from membrane proteins on human natural killer cells. J. Exp. Med. 173:1017. Novel ligands for the murine NKG2D receptor: expression by tumor cells and 42. Ho, E. L., J. W. Heusel, M. G. Brown, K. Matsumoto, A. A. Scalzo, and activation of NK cells and . Nat. Immun. 1:119. W. M. Yokoyama. 1998. Murine Nkg2d and Cd94 are clustered within the natural 63. Nandi, D., J. A. Gross, and J. P. Allison. 1994. CD28-mediated costimulation is killer complex and are expressed independently in natural killer cells. Proc. Natl. necessary for optimal proliferation of murine NK cells. J. Immunol. 152:3361. Acad. Sci. USA 95:6320. 64. Phillips, J. H., C. W. Chang, J. Mattson, J. E. Gumperz, P. Parham, and 43. Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, and T. Spies. L. L. Lanier. 1996. CD94 and a novel associated protein (94ap) form a NK cell 1999. Activation of NK cells and T cells by NKG2D, a receptor for stress- receptor involved in the recognition of HLA-A, HLA-B, and HLA-C allotypes. inducible MICA. Science 285:727. Immunity 5:163. 44. Diefenbach, A., A. M. Jamieson, S. D. Liu, N. Shastri, and D. H. Raulet. 2000. http://www.jimmunol.org/ Ligands for the murine NKG2D receptor: expression by tumor cells and activa- 65. Lazetic, S., C. Chang, J. P. Houchins, L. L. Lanier, and J. H. Phillips. 1996. tion of NK cells and macrophages. Nat. Immun. 1:119. Human natural killer cell receptors involved in MHC class I recognition are 45. Cosman, D., J. Mullberg, C. L. Sutherland, W. Chin, R. Armitage, W. Fanslow, disulfide-linked heterodimers of CD94 and NKG2 subunits. J. Immunol. 157: M. Kubin, and N. J. Chalupny. 2001. ULBPs, novel MHC class I-related mole- 4741. cules, bind to CMV glycoprotein UL16 and stimulate NK cytotoxicity through 66. Carretero, M., C. Cantoni, T. Bellon, C. Bottino, R. Biassoni, A. Rodriguez, the NKG2D receptor. Immunity 14:123. J. J. Perezvillar, L. Moretta, A. Moretta, and M. Lopez-Botet. 1997. The CD94 46. Cerwenka, A., A. B. H. Bakker, T. McClanahan, J. Wagner, J. Wu, J. H. Phillips, and NKG2-A C-type lectins covalently assemble to form a natural killer cell and L. L. Lanier. 2000. Retinoic acid early inducible genes define a ligand family inhibitory receptor for HLA class I molecules. Eur. J. Immunol. 27:563. for the activating NKG2D receptor in mice. Immunity 12:721. 67. Kung, S. K., R. C. Su, J. Shannon, and R. G. Miller. 1999. The NKR-P1B gene 47. Girardi, M., D. E. Oppenheim, C. R. Steele, J. M. Lewis, E. Glusac, R. Filler, product is an inhibitory receptor on SJL/J NK cells. J. Immunol. 162:5876. P. Hobby, B. Sutton, R. E. Tigelaar, and A. C. Hayday. 2001. Regulation of 68. Makrigiannis, A. P., P. Gosselin, L. H. Mason, L. S. Taylor, D. W. McVicar,

cutaneous malignancy by ␥␦ T cells. Science 294:605. J. R. Ortaldo, and S. K. Anderson. 1999. Cloning and characterization of a novel by guest on September 27, 2021 48. Wu, J., Y. Song, A. B. Bakker, S. Bauer, T. Spies, L. L. Lanier, and J. H. Phillips. activating Ly49 closely related to Ly49A. J. Immunol. 163:4931. 1999. An activating immunoreceptor complex formed by NKG2D and DAP10. 69. Mehta, I. K., H. R. C. Smith, J. Wang, D. H. Margulies, and W. M. Yokoyama. Science 285:730. 2000. A “chimeric” C57L-derived Ly49 inhibitory receptor resembling the 49. Chang, C., J. Dietrich, A. G. Harpur, J. A. Lindquist, A. Haude, Y. W. Loke, Ly49D activation receptor. Cell. Immunol. 209:29. A. King, M. Colonna, J. Trowsdale, and M. J. Wilson. 1999. Cutting edge: 70. Groh, V., S. Bahram, S. Bauer, A. Herman, M. Beauchamp, and T. Spies. 1996. KAP10, a novel transmembrane adapter protein genetically linked to DAP12 but Cell stress-regulated human major histocompatibility complex class I gene ex- with unique signaling properties. J. Immunol. 163:4651. pressed in gastrointestinal epithelium. Proc. Natl. Acad. Sci. USA 93:12445. 50. Groh, V., R. Rhinehart, J. Randolph-Habecker, M. S. Topp, S. R. Riddell, and 71. Malarkannan, S., P. P. Shih, P. A. Eden, T. Horng, A. R. Zuberi, G. Christianson, T. Spies. 2001. Costimulation of CD8␣␤ T cells by NKG2D via engagement by D. Roopenian, and N. Shastri. 1998. The molecular and functional characteriza- MIC induced on -infected cells. Nat. Immun. 2:255. tion of a dominant minor H antigen, H60. J. Immunol. 161:3501. 51. Wu, J., H. Cherwinski, T. Spies, J. H. Phillips, and L. L. Lanier. 2000. DAP10 72. Fodil, N., L. Laloux, V. Wanner, P. Pellet, G. Hauptmann, N. Mizuki, H. Inoko, and DAP12 form distinct, but functionally cooperative, receptor complexes in T. Spies, I. Theodorou, and S. Bahram. 1996. Allelic repertoire of the human natural killer cells. J. Exp. Med. 192:1059. MHC class I MICA gene. Immunogenetics 44:351. 52. Diefenbach, A., E. R. Jensen, A. M. Jamieson, and D. H. Raulet. 2001. Rae1 and 73. Bahram, S. 2000. MIC genes: from genetics to biology. Adv. Immunol. 76:1. H60 ligands of the NKG2D receptor stimulate tumor immunity. Nature 413:165. 74. Li, P., D. L. Morris, B. E. Willcox, A. Steinle, T. Spies, and R. K. Strong. 2001. 53. Pende, D., C. Cantoni, P. Rivera, M. Vitale, R. Castriconi, S. Marcenaro, Complex structure of the activating immunoreceptor NKG2D and its MHC class M. Nanni, R. Biassoni, C. Bottino, A. Moretta, and L. Moretta. 2001. Role of I-like ligand MICA. Nat. Immun. 2:443.