T Cells and NK Cells1

The Journal of Immunology

Silencing Human NKG2D, DAP10, and DAP12 Reduces Cytotoxicity of Activated CD8؉ T Cells and NK Cells1

Mobin Karimi,* Thai M. Cao,* Jeanette A. Baker,* Michael R. Verneris,‡ Luis Soares,†§ and Robert S. Negrin2*

Human CD8؉ T cells activated and expanded by TCR cross-linking and high-dose IL-2 acquire potent cytolytic ability against tumors and are a promising approach for immunotherapy of malignant diseases. We have recently reported that in vitro killing by these activated cells, which share phenotypic and functional characteristics with NK cells, is mediated principally by NKG2D. NKG2D is a surface receptor that is expressed by all NK cells and transmits an activating signal via the DAP10 adaptor molecule. Using stable RNA interference induced by lentiviral transduction, we show that NKG2D is required for cytolysis of tumor cells, including autologous tumor cells from patients with ovarian cancer. We also demonstrated that NKG2D is required for in vivo antitumor activity. Furthermore, both activated and expanded CD8؉ T cells and NK cells use DAP10. In addition, direct killing ,was partially dependent on the DAP12 signaling pathway. This requirement by activated and expanded CD8؉ T cells for DAP12 and hence stimulus from a putative DAP12-partnered activating surface receptor, persisted when assayed by anti-NKG2D Ab- mediated redirected cytolysis. These studies demonstrated the importance of NKG2D, DAP10, and DAP12 in human effector cell function. The Journal of Immunology, 2005, 175: 7819–7828.

atural killer cells are frontline guardians in surveillance identified, and it is possible that these interactions may yield ad- of pathogens and tumors, and their activation is regu- ditional activating signals (9). CD3␨, FcR␥, and DAP12 intracel- N lated by the integration of excitatory and inhibitory sig- lular ITAM motifs initiate protein tyrosine kinase-dependent signals initiated by membrane-bound receptor molecules (1, 2). In- naling pathways (10), whereas DAP10 contains a YxxM activation hibitory signals are mediated by monomeric receptors possessing motif that triggers the lipid kinase cascade (11–13). intracytoplasmic immunoreceptor tyrosine-based inhibition motifs In human NK cells, DAP10 is the exclusive binding partner and that recruit tyrosine protein phosphatases (3). Activating signals signaling intermediate for NKG2D (14). NKG2D is a potent acti- are generated from more diverse and complementary groups of vating receptor whose ligands include proteins induced by cellular membrane molecules that, upon ligand binding, can oligomerize stress and malignant transformation, such as MHC class I-related with different cytoplasmic adaptor polypeptides to form distinct by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. protein (MICA) and -B and members of the UL16-binding pro- signaling complexes. In human NK cells, these types of activating teins, and is thought to have a particularly important role in anti- receptors include NKG2D, which associate with the adaptor mol- tumor immunity (15). ecule disulphide adaptor molecule (DAP)3-10, and the low-affinity The expression and function of these activating signaling com- IgG FcR CD16, which noncovalently couples with the adaptor plexes are not limited to NK cells. DAP12 is expressed in granu- molecules CD3␨ and FcR␥. CD3␨, FcR␥, and another adaptor mol- locytes, monocytes, as well as CD4ϩ T cells, and DAP12-deficient ecule found in NK cells, named DAP12 (4, 5), have in common a mice have been shown to have defects in dendritic cells and NK membrane-bound structure with a cytoplasmic domain bearing one cells (16–19). NKG2D has been identified on ␥␦ TCRϩ T cells or more ITAM motifs. Human DAP12 is known to mediate acti- and human CD8ϩ T cells, where it appears to have costimulatory ϩ http://classic.jimmunol.org vating signals from diverse binding partners that include killer cell activity and can be induced on activated mouse CD8 T cells Ig-like receptors S such as killer cell Ig-like receptor 2DS2, the (20–22). We have recently reported that NKG2D expression is activation-induced NKp44 receptor, and HLA-E binding CD94/ up-regulated by human CD8ϩ T cells activated and expanded ex NKG2C heterodimers (5–8). DAP12 has been shown to associate vivo in the presence of IFN-␥, high-dose IL-2, and TCR cross- with a host of other NK cell surface proteins that have yet to be linking with an anti-CD3 mAb (23). Activated and expanded T cells up-regulate cytotoxic effector molecules, acquire functional

Downloaded from and phenotypic properties that resemble NK cells, and develop *Division of Blood and Marrow Transplantation and †Division of Immunology and Rheumatology, Department of Medicine, Stanford University School of Medicine, broad cytotoxicity against a variety of malignant cell targets, in- Stanford, CA 94305; ‡Division of Pediatric Blood and Marrow Diseases, Department cluding autologous leukemic blasts (24–28). This acquisition of of Pediatrics, University of Minnesota, Minneapolis, MN 55455; and §Institute of Biomedical Pharmacology, Curitiba, Brazil cytotoxicity, which is MHC unrestricted, coincides with the induction of DAP10 expression, and the majority of killing ability is Received for publication May 18, 2005. Accepted for publication September 21, 2005. accounted for by NKG2D signaling. ϩ The costs of publication of this article were defrayed in part by the payment of page Activated and expanded CD8 T cells may have considerable charges. This article must therefore be hereby marked advertisement in accordance therapeutic utility as an immune-based strategy aimed at eradica- with 18 U.S.C. Section 1734 solely to indicate this fact. tion or suppression of malignant diseases. These T cells have broad 1 This work was supported in part by Grants KO8HL04505, P01CA049605, and in vitro and in vivo biological activity against inoculated tumors R01CA080006 from the National Institutes of Health. after both syngeneic and allogeneic bone marrow transplantation 2 Address correspondence and reprint requests to Dr. Robert S. Negrin, Division of Blood and Marrow Transplantation, Stanford University, CCSR Room 2205, 269 in rodent models (29, 30). Autologous infusion of cells of this type West Campus Drive, Stanford, CA 94305-5170. E-mail: [email protected] in patients has been associated with minimal toxicity, can reduce

the risk of tumor recurrence in patients with hepatocellular carci- NKG2D, DAP10, and DAP12 RNAi noma after surgical resection, and may induce partial remissions in Nucleotide sequences for short hairpin RNA (shRNA) are summarized in patients with relapse of lymphoma after bone marrow transplan- Table I. A previously reported RNAi sequence specific for murine CD8␣ tation (31, 32). In this study, we extended our evaluation of the (mCD8) was used as a nonspecific control (33). Oligonucleotides were signaling requirements for induction of cytotoxicity by activated designed that incorporated these sequences within a short hairpin structure, Ј Ј and expanded human CD8ϩ T cells. We used a combination of in using the stem loop sequence 5 -TCAAGAGA-3 , which were then cloned between MluI and ClaI sites downstream of an H1 promoter in the plasmid vitro and in vivo assays against tumor cell lines and autologous pLVTHM as previously described (34). Plasmid pLVTHM, derived from tumor targets and gene targeting by RNA interference (RNAi). We pSUPERn which contains a GFP expression cassette upstream of the H1 show that both DAP10 and DAP12, in addition to NKG2D, con- promoter (35), was a gift from D. Trono (University of Geneva, Geneva, tribute major activating signals in regulating the cytotoxicity of Switzerland). For some experiments we replaced the GFP gene on ϩ pLVTHM with a DsRed reporter gene via PmeI and SpeI linkers. 293 cells activated and expanded human CD8 T cells. We show, further- were transfected using the calcium phosphate method with 10 ␮g pLVThM more, that each of these molecules, including DAP12, has a non- lentivirus, 3.5 ␮g of vesicular stomatitis virus G plasmid, and 6.5 ␮gof redundant and partially obligate signaling role in triggering tumor CMV⌬R8.74 according to standard protocols (36). After 16 h, the medium killing. was changed, and recombinant lentivirus vectors were harvested 24–48 h later. CD8ϩ T cell or NK cell infection was performed three times at 24-h intervals. For each infection, cells were plated in 48-well plates at 1 ϫ 105 Materials and Methods cells/well and infected in the presence of protamine and hexadimethrin. Spin infection was performed at 1200 rpm for 90 min at 37°C. Four days mAb reagents and flow cytometry after the first infection, transduced cells were isolated by FACS sorting for ϩ Ͼ The mAbs used in this study were the following: CD3 (UCHT1), CD8 GFP cells to 99% purity. (RPA-T8), CD16 (3G8), CD56 (NCAM16-2), and isotype control (IgG) were purchased from BD Biosciences as either FITC or PE conjugates. Cell lines and autologous tumor cell isolation Purified NKG2D were purchased from R&D Systems, and PE NKG2D-PE RPMI 1640, a human myeloma cell line cultured in cRPMI, and P815, a and NKG2D-allophycoyanin were purchased from Beckman Coulter. La- murine mastocytoma cell line cultured in complete DMEM, were both beled cells were analyzed on either a FACScan or an LSR-II instrument purchased from American Type Culture Collection. UCI101 (a gift from (BD Biosciences) maintained by the Stanford Shared FACS Facility (Stan- S. Y. Liao, University of California, Irvine, CA) is a human ovarian car- ford University). For micromagnetic cell separation, human CD8 (BW135/ cinoma cell line cultured in complete IDMEM. Luciferase expressing 80) MicroBeads, human CD16 MicroBeads (130-045-701), or human HEA UCI101 (UCI101-luc) was generated according to a modified protocol that MicroBeads (130-061-101) were used for positive selection using an Au- we have previously described (37). In brief, retrovirus generated in Phoe- toMACS device according to the vendor’s instructions (Miltenyi Biotec). nix A producer cells transfected with the pMSCVpuroBA-L2G plasmid (a Where stated, additional cell isolation by FACS sorting was performed gift from Dr. C. Contag, Stanford University, Stanford, CA) containing a with a MoFlo instrument (DakoCytomation). firefly luciferase insert cloned into the SalI and NotI sites, was used to infect UCI101 cells plated in the presence of polybrene and protamine. Isolation and generation of ex vivo expanded and activated For preparation of tumor targets used for evaluation of autologous ac- ϩ CD8ϩ T cells tivated and expanded CD8 T cell cytolysis, fresh tumor was collected from six ovarian cancer patients at the time of surgical resection and di- CD8ϩ T cells were activated and expanded as previously described (23). gested and homogenized to single-cell suspensions using a metal cell Briefly, mononuclear cells were isolated from healthy donors by Ficoll- strainer. The tumor cells were incubated on a shaker at 1200 rpm for 30 ␮ ␮ by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. Hypaque density centrifugation, washed three times with PBS, and sepa- min at 37°C with collagenase 2 (10 g/ml) or collagenase D (10 g/ml) in rated by positive selection with CD8 MicroBeads. Cells were then labeled cRPMI. The cells were then selected in T25 flasks and subsequently pos- and further purified by FACS sorting for the CD3ϩCD8ϩ population. The itively enriched using HEA MicroBeads to Ͼ80% purity. At the time of final product was resuspended at 2 ϫ 106 cells/ml in complete RPMI 1640 surgery, PBMC were concurrently obtained from each patient and used for ϩ (cRPMI) medium supplemented with 10% FCS, 2 mM L-glutamine, 100 the generation of autologous activated and expanded CD8 T cells as noted U/ml penicillin, 100 ␮g/ml streptomycin, and 50 ␮g/ml 2-ME at 37°C in above. The protocol for collection of ovarian and peripheral blood cells ␥ 5% CO2. On day 0, cells were activated with human IFN- (1000 U/ml; was approved by the institutional review board at Stanford University. Genentech). On day 1, cells were stimulated with OKT-3 (25 ng/ml; Or- thoBioTech) and human rIL-2 (300 U/ml; Chiron). On day 4, cells were divided into flasks with new medium and not restimulated with OKT-3. Every 3–5 days thereafter, fresh medium was added supplemented with 6 http://classic.jimmunol.org IL-2 (300 U/ml) to maintain a cell density of 1.5–2 ϫ 10 /ml for a total of Table I. shRNA sequences for human NKG2D, DAP10, and DAP12 14–28 days. Bulk populations of cells have been referred to as cytokine- RNAI a induced killer cells. Twenty-four to 48 h before use, cells were labeled and further purified by FACS sorting for the CD3ϩCD8ϩ population. For some b Ј3 Ј experiments CD8ϩ T cells were also activated and expanded from patients shRNA nt Position Sequence (5 3 ) with ovarian cancer according to protocols approved by the Stanford Uni- NKG2D-sh1 67–85 AGAGTTTGCTGGCCTGTGC versity administration panel on human subjects. NKG2D-sh2 254–272 GATCTTCCCTCTCTGAGCA

Downloaded from NKG2D-sh3 618–636 AGTTACTGTGGCCCATGTC Isolation of NK cells NKG2D-sh4 846–864 GATGGCTCCATTCTCTCAC NKG2D-sh5 880–898 TAATTGAAATGCAGAAGGG Mononuclear cells from healthy donors freshly isolated with Ficoll- NKG2D-sh6 1468–1486 CGGGGTCAGGGAGGTGGTG Hypaque were purified for NK cells by a two-step procedure. Cells were first positively selected with CD16 MicroBeads, followed by additional Ϫ ϩ DAP10-sh1 332–350 GAAGATGGCAAAGTCTACA purification with FACS sorting for the CD3 D56 cell population. NK cell DAP10-sh2 335–353 GATGGCAAAGTCTACATCA purity was typically Ͼ98% as determined by reanalysis of an aliquot of DAP10-sh3 343–361 AGTCTACATCAACATGCCG separated cells. During lentiviral infection for RNAi, isolated NK cells DAP12-sh1 83–101 CCCTGCAGCAGGCTCCTGC were maintained in DMEM/Ham’s F-12 medium (2/1 mixture) for 7 days DAP12-sh2 126–144 GTGGTCTCCGTCCTGTCCA supplemented with 10% human type AB serum, 600 U/ml IL-2, 100 U/ml DAP12-sh3 310–328 ACAGCGTATCACTGAGACC penicillin, 100 ␮g/ml streptomycin, 50 ␮M ME, 10 ng/ml ethanolamine, 10 DAP12-sh4 385–403 CACACAGAGGCCGTATTAC ␮g/ml ascorbate, and 50 ng/ml selenium. mCD8-sh 888–906 GCTACAACTACTACATGAC

a Abbreviations: mCD8, murine CD8␣; nt, nucleotide. 3 Abbreviations used in this paper: DAP, disulphide adaptor molecule; MICA, MHC b Nucleotide position numbered according to the following GenBank nucleotide class I-related protein; RNAi, RNA interference; cRPMI, complete RPMI 1640; m, sequence: NKG2D, X54870; DAP10, AF172929; DAP12, AF019562; and murine murine; shRNA, short hairpin RNA; siRNA, small interfering RNA. CD8␣, XM132621. The Journal of Immunology 7821

51Cr release cytotoxicity assay Statistics Tumor targets were labeled with 51Cr (DuPont-NEN) by incubating 1 ϫ Differences in the mean observations between experimental groups were 6 ␮ 51 10 cells in 300 Ci (11.1 MBq) of Cr for at 37°C for2hin5%CO2. determined with Student’s t test. Statistical significance was conferred at The labeled cells were washed three times with PBS, resuspended in the p Ͻ 0.05 level. cRPMI, and plated in 96-well plates at a concentration of 1 ϫ 104 cells/ml in triplicate. Effector cells were added at specified ratios (10:1 or 40:1) and Results incubated at 37°C for4hin5%CO2. At the completion of each assay, RNAi with NKG2D-specific shRNA suppresses cytotoxicity of supernatant was collected and counted using a gamma counter (Cobra/AII; activated and expanded CD8ϩ T cells Packard). The percent-specific 51Cr lysis was calculated with the following equation: percent-specific lysis ϭ 100 ϫ (test release) Ϫ (spontaneous We have recently demonstrated that human CD8ϩ T cells activated Ϫ release)/(maximal release) (spontaneous release). and expanded during culture in the presence of IFN-␥, IL-2, and Redirected cytotoxicity assay OKT3 lose cytolytic potential when NKG2D is silenced by tran- For redirected killing assay, 1 ϫ 106 P815 cells were labeled with 51Cr as sient transfection with a small interfering RNA (siRNA) duplex described above. Excess 51Cr was removed by washing, and the cells were oligonucleotide (23). To complement these studies and to develop resuspended in 500 ␮l of medium. Anti-NKG2D (5 ␮g) was added and a model for in vivo analysis of NKG2D function, we sought to incubated for 30 min. The P815 target cells were then used as a target in establish stable NKG2D knockdown by RNAi in activated CD8ϩ 51 the Cr release assay as described. T cells via transduction of shRNA with the pLVTHM lentiviral Cytokine production vector. NKG2D-sh6, which incorporates nucleotide sequences tar- Activated and expanded CD8ϩ T cells at 21 days of culture were removed, geting a portion of the NKG2D gene transcript previously shown washed, and incubated with 20 ng/ml PMA (Sigma-Aldrich) and 1 ␮M functional for silencing NKG2D by siRNA transfection, was eval- ionomycin or medium alone in a total volume of 200 ␮l/well in a 96-well uated along with five other NKG2D-specific shRNA. NKG2D-sh6 plate and cultured for 48 h. The supernatant was harvested and assayed for was again found to most effectively silence NKG2D gene expres- cytokine production by ELISA for IL-2 and IFN-␥ (eBioscience) according ϩ to the vendor’s instructions. sion in CD8 T cells activated and expanded after 14–28 days in culture (Fig. 1, A and B). FACS analyses of transduced cells, iden- Northern blotting tified by positive GFP reporter gene expression, demonstrate that RNA was prepared with the RNeasy kit (Qiagen) according to the vendor’s infection with empty pLVTHM vector does not affect NKG2D instructions, and RNA blotting was performed according to standard pro- surface levels (Fig. 1, C and D). At the same time, NKG2D RNAi ␣ 32 tocols (38). Three - P-labeled probes were constructed with the follow- with NKG2D-sh6 is specific insofar as it has no effect on other ing PCR primers: NKG2D, 5Ј-CTGGGAGATGAGTGAATTTCATA-3Ј Ј Ј surface receptors, including CD3, CD8, CD16, and CD56. Acti- and 5 -GACTTCACCAGTTTAAGTAAATC-3 (417-bp fragment); ϩ DAP10, 5Ј-CATCTGGGTTCACATCCTCTT-3Ј and 5Ј-CAGAAGTCAA vated and expanded CD8 T cells can lyse RPMI tumor targets AGGTCCAAGC-3Ј (306-bp fragment); and DAP12, 5Ј-CCGCAAAGAC (Fig. 1E) and retain this ability when RNAi is attempted with CTGTACGCCA-3Ј and 5Ј-TGGACTTGGGAGAGGACTGG-3Ј (650-pb NKG2D-specific shRNA that do not effectively silence NKG2D. fragment). PCR products were cloned by TA cloning into pCR2.1 (Invitro- Cytolysis is similarly potent after RNAi with mCD8-sh, an shRNA gen Life Technologies), digested with EcoRI, separated by 1.5% agarose ␣ gel electrophoresis, and purified with the QiaQick gel extraction kit specific for murine CD8 (33), and after transduction by empty (Qiagen). pLVTHM vector without RNAi. Cytolysis is suppressed to a level by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. Immunoprecipitations and Western blotting comparable with background cytolysis of P815 targets, however, if NKG2D RNAi is induced by NKG2D-sh6 ( p Ͻ 0.0001). P815 is ϩ CD8 T cells or NK cells were lysed using freshly prepared lysis buffer (10 a murine mastocytoma cell line known to be resistant to killing by mM Tris (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.5% ϩ deoxycholate, 0.1% SDS, complete Protease Inhibitor Cocktail (Roche), activated and expanded human CD8 T cells. Thus, we confirmed and 500 ␮M PMSF) and clarification at 12,000 ϫ g for 10 min at 4°C. our prior observations that NKG2D is required for in vitro tumor ϩ Seventy micrograms of each cell lysate was separated on an 18% denatur- lysis by activated CD8 T cells. ing polyacrylamide gel and transferred to polyvinylidene difluoride membranes for immunoblot analysis. Membranes were blocked with 5% nonfat NKG2D is required for cytolysis of autologous tumor by milk and probed with monoclonal anti-NKG2D (M585, mouse IgG) or activated and expanded CD8ϩ T cells polyclonal DAP10 (C-20, goat IgG; Santa Cruz Biotechnology), followed http://classic.jimmunol.org by incubation with a species-specific, secondary, HRP-conjugated Ab We and others have shown that human plasmacytoma cell lines, (Santa Cruz Biotechnology). Blots were developed using ECL reagent such as U266, and plasmacytoma cells from patients with multiple (Amersham Biosciences). myeloma express MICA and ULBP3, and that cytotoxicity by ac- ␮ For immunoprecipitation, clarified lysate was labeled with 1 g of anti- tivated CD8ϩ T cells and NK cells can correlate with the expres- DAP12 polyclonal rat IgG Ab (FL-113; Santa Cruz Biotechnology) for 1 h at 4°C, followed by cross-linking to protein G-agarose beads (Pierce) pre- sion of these known NKG2D ligands (23, 40). We sought to extend equilibrated with lysis buffer, for1hat4°Conarotator. After washing, these observations by examining requirements for NKG2D signal- ϩ Downloaded from immunoprecipitated lysate was resuspended in Laemmli buffer with 5% ing in killing of autologous tumor by CD8 T cells activated and 2-ME and boiled for 5 min before separation and immunoblotting as de- expanded from human patients undergoing cancer therapy. A scribed above. novel NKG2D ligand, termed Letal, has recently been found to be SCID/Hu tumor model and bioluminescent imaging expressed by ovarian carcinoma patients (41). Patients with ovar- C.B-17 SCID mice, 8–12 wk old, were obtained from the colony at the ian cancer undergoing surgical resection of their tumor were se- Stanford University Research Animal Facility and irradiated with a single lected as sources of tumor targets for in vitro cytolysis studies. A 2-Gy fraction using a Phillips Unit Irradiator (250 kV, 15 mA) at a dose PBMC sample was simultaneously obtained at the time of tumor rate of 100 rad/min. After irradiation, mice were maintained on antibiotic collection for activation and expansion of CD8ϩ T cells. After water containing trimethoprin/sulfmethoxole. Recipient mice received 1 ϫ ϩ 106 luciferase-expressing UCI101-luc cells (1 ml/injection) via i.p. injec- expansion, NKG2D expression in CD8 T cells from these pa- tion. Intraperitoneal injection with UCI101-luc into immune-deficient mice tients can be functionally silenced using NKG2D-sh6 and establishes tumors with peritoneal implants within 1 wk and eventually pLVTHM lentiviral transduction (Fig. 2, A and B). In 51Cr release massive ascites and death (39). On days 6 and 12 after tumor inoculation, assays, we then observed that activated CD8ϩ T cells can effec- each mouse was given an injection with 40 ϫ 106 activated and expanded CD8ϩ T cells. Mice were evaluated every second day from the time of tively lyse labeled tumor cell targets isolated from the same patient tumor injection for 30 days by bioluminescence imaging with the IVIS 200 when transduced without RNAi or with nonspecific mCD8-sh (Fig. Imaging System (Xenogen) as previously described (37). 2C). In contrast, cytolysis of each ovarian tumor target paired with 7822 NKG2D, DAP10, AND DAP12 IN CD8ϩ T AND NK CELL CYTOTOXICITY

FIGURE 1. NKG2D RNAi with shRNA by lentiviral transduction in activated and expanded human CD8ϩ T cells. Human CD8ϩ T cells activated and expanded after 14–28 days of culture in the presence of IFN-␥, IL-2, and OKT3 were infected with pLVTHM lentivirus expressing one of six shRNA-targeting NKG2D. As shown by RNA (A) and Western blotting (B) of transduced cells, NKG2D silencing is most effective with NKG2D-sh6, which targets a sequence (nt 1468–1486) in the 3Ј untranslated region of the NKG2D transcript. C and D, FACS analysis of transduced cells. NKG2D RNAi with NKG2D-sh6 in- duces complete silencing of NKG2D while preserving surface expression of CD3, CD8, CD16, and CD56. E, Cytolysis of 51Cr-labeled RPMI human plasmacytoma cell line by activated and expanded CD8ϩ T cells is suppressed Ͼ90% by NKG2D RNAi with NKG2D-sh6 (p Ͻ 0.0001), whereas RNAi with nonspeciﬁc mCD8 shRNA (mCD8-sh) and transduction with empty pLVTHM (No RNAi) had no effect. Cytolysis of murine P815 mastocytoma was not observed. Shown are representative results from one experiment repeated three times.

patient activated CD8ϩ T cells was strongly suppressed when and ionomycin failed to augment levels of IFN-␥, and production NKG2D was silenced by NKG2D-sh6 ( p Ͻ 0.0001). of these cytokines was not inﬂuenced by NKG2D silencing. NK cells can kill tumor targets through a combination of cell- Similar amounts were obtained for IL-2. These results suggest that by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. mediated and humoral cytotoxicity mechanisms. We previously the production of cytotoxic cytokines does not contribute to killing showed that activated human CD8ϩ T cells require cell-cell con- by activated and expanded CD8ϩ T cells.

tact and rely on granule release to effect cytolysis (42, 43). We ϩ evaluated whether cytokine production may contribute to killing DAP10 RNAi in activated and expanded human CD8 T cells by activated CD8ϩ T cells by ELISA for IL-2 and IFN-␥ suppresses in vitro cytolysis (Fig. 2D). After 21 days of activation and expansion in culture, NKG2D is a type II homodimeric, C-type, lectin-like transmem- these cells constitutively produce IFN-␥. Stimulation with PMA brane molecule with a short cytoplasmic tail and by itself lacks http://classic.jimmunol.org

FIGURE 2. NKG2D RNAi suppresses cytolysis of autologous ovarian tumor by activated and expanded CD8ϩ T cells. CD8ϩ T cells activated and expanded from patients were evaluated for cytolysis of autologous ovarian tumors isolated at the time of surgical re-

Downloaded from section. RNA (A) and Western blots (B) from a representative patient show that RNAi with NKG2D-sh6 by pLVTHM lentivirus effectively silences NKG2D expression. C, Cytolysis of 51Cr-labeled ovarian tumor targets by paired patient activated and expanded CD8ϩ T cells (E:T cell ratio, 10:1 or 40:1) is significantly reduced by NKG2D RNAi (p Ͻ 0.0001) compared with nonspecific RNAi and mock pLVTHM lentiviral transduction. Shown are representative results from one experiment repeated four times with other ovarian cancer autologous T cells paired samples. ELISA for IFN-␥ (D) by activated CD8ϩ T cells showing that production of this cytokine is not influenced by NKG2D silencing and does not significantly increase with stimulation by PMA and ionomycin. The Journal of Immunology 7823

intrinsic signaling capabilities (44). Instead, NKG2D forms a sta- tors. Similar to DAP10, we found levels of DAP12 expression in ble complex with the predominantly intracellular DAP10 adaptor activated and expanded CD8ϩ T cells comparable to those ob- protein to deliver activating signals via recruitment of the p85 served in freshly isolated NK cells (Fig. 4A). Of four DAP12- subunit of the PI3K. We therefore used RNAi to evaluate the role specific shRNA evaluated, we determined that DAP12-sh3 in- of DAP10 signaling in activated and expanded CD8ϩ T cells. We duced the most suppression of DAP12 gene expression in activated first determined that DAP10 is expressed by activated T cells and CD8ϩ T cells (Fig. 4B). The DAP12-sh3 nucleotide sequence is the CD8ϩ T cell subset (Fig. 3A). We then generated three DAP10- complementary to an entirely translated portion of exon 4 of the specific shRNA and found that DAP10-sh3, which targets exon 4 DAP12-coding sequence. In accordance with our speculation, we of the DAP10 mRNA near the translation stop codon, most effec- observed a partial and significant reduction in cytotoxicity by ac- tively silenced DAP10 gene expression when transduced with len- tivated and expanded CD8ϩ T cells when DAP12 was silenced by tivirus (Fig. 3, B and C). Activated CD8ϩ T cells subjected to RNAi using DAP12-sh3 ( p ϭ 0.002; Fig. 4C). Although less pro- DAP10 RNAi using the DAP10-sh3 lost ϳ75% of cytolytic ac- found than requirements for NKG2D- and DAP10-mediated sig- tivity ( p ϭ 0.0003) compared with transduction using ineffective naling, these results suggest that activated human CD8ϩ T cells DAP10 shRNA, nonspecific mCD8-sh, and empty pLVTHM also rely on DAP12-dependent signaling pathways for optimal tu- vector. mor cytolysis.

DAP12 silencing also suppresses cytolysis by activated and Double-knockdown of DAP10 and DAP12 in activated CD8ϩ T expanded CD8ϩ T cells cells DAP10 gene knockdown suppressed cytolysis by activated and The proximal signaling pathway for DAP10 includes Grb-2, phos- expanded CD8ϩ T cells less potently than did silencing of pholipase C-2, SLP-76, and PI3K (45). For DAP12, initial signal- NGK2D, suggesting that additional signaling pathways contribute ing involves activation of Src protein kinases, which phosphorylate to maximum cytolysis triggered by NKG2D-mediated activation. the ITAM on tandem tyrosine residues, followed by second-line No association between NKG2D and DAP12, another intracyto- recruitment of Syk family protein kinases. It is thought that these plasmic activating adaptor protein, has been observed in humans disparate signals then culminate in a common pathway leading to (9, 14). DAP12 is a promiscuous binding partner, however, and activation of Rac, MAPK kinase, and ERK and, ultimately, cytol- thus may contribute signals important for cytolysis in activated ysis. To gain insight into whether the integration of DAP10 and CD8ϩ T cells through pairing with other activating surface recep- DAP12 signaling contributes to induction of optimal tumor killing, we generated activated CD8ϩ T cells silenced for both DAP10 and by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. http://classic.jimmunol.org Downloaded from

FIGURE 4. DAP12 RNAi in activated and expanded CD8ϩ T cells. A, FIGURE 3. DAP10 RNAi in activated and expanded CD8ϩ T cells. A, RNA blot showing that DAP12 is expressed by activated and expanded RNA blot showing that DAP10 is constitutively expressed by freshly iso- human T cells at a level comparable to that in freshly isolated NK cells. B, lated human NK cells and all activated and expanded T cells including the RNA blot; DAP12 silencing using four DAP12 shRNA in activated CD8ϩ T CD8ϩ subset. By RNA (B) and Western blotting (C), DAP10 silencing by cells by pLVTHM lentivirus was functionally achieved with DAP12-sh3. three DAP10 shRNA in activated CD8ϩ T cells by pLVTHM lentivirus DAP12-sh3 (nt 310–328) targets exon 4, an entirely translated portion of the was most effective with DAP10-sh3. DAP10-sh3 (nt 343–361) targets exon DAP12 coding sequence. C, Cytolysis of RPMI 1640 targets in a 51Cr release 4 of the DAP10 transcript, nine nucleotides upstream of the translation stop assay by activated and expanded CD8ϩ T cells is partially suppressed by codon. D, Cytolysis of 51Cr-labeled RPMI 1640 targets by activated and DAP12 RNAi with DAP12-sh3 (p Ͻ 0.002) and to a lesser extent by DAP12- expanded CD8ϩ T cells is suppressed by ϳ75% (p ϭ 0.003) after DAP10 sh4 at the higher E:T cell ratio of 40:1 compared with negative mCD8-sh and RNAi with DAP10-sh3. Shown are representative results from one exper- empty pLVTHM vector controls. Shown are representative results from one iment repeated three times. experiment repeated three times. 7824 NKG2D, DAP10, AND DAP12 IN CD8ϩ T AND NK CELL CYTOTOXICITY

FIGURE 5. Double transduction with DAP10-sh3 and DAP12-sh3 for RNAi. A, The GFP cassette in pLVTHM was replaced by a DsRed reporter gene for transduction of DAP10-sh3 to confirm by double DAP10 and DAP12 knockdown FACS analysis in activated T cells as shown by GFP and DsRed fluorescence, also indicating no reduction in NKG2D expression. B, RNAi of DAP10 and DAP12 is specifically mediated by, respectively, DAP10- sh3 and DAP12-sh3, as shown in this RNA blot of activated and expanded CD8ϩ T cells. C, Western blot for DAP10 and DAP12 after transduction for DAP10-sh3 and DAP12-sh3. D, Suppression of 51Cr-labeled RPMI 1640 cytolysis by activated and expanded T cells in the presence of simultaneous DAP10 and DAP12 RNAi (p Ͻ 0.001) compared with mCD8-sh and empty pLVTHM vector negative controls as shown. Shown are representative results from one experiment repeated three times.

DAP12. To isolate double transductants for in vitro cytolysis as- by DAP10 and DAP12 double knockdown was signiﬁcant, with a says, we replaced the GFP reporter gene on the pLVTHM lenti- Ͼ80% reduction ( p Ͻ 0.0001) in cytotoxicity compared with neg-

by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. viral vector with DsRed for transduction of DAP10-sh3. Thus, ative controls (Fig. 5D), reflecting more potent suppression than DAP10-sh3-transduced cells can be FACS sorted based on DsRed silencing of DAP10 or DAP12 alone. fluorescence, as opposed to GFP fluorescence for successful Human NKG2D has been shown to assemble as a hexameric DAP12-sh3 transduction, and fluorescence of both DsRed and complex consisting of the NKG2D homodimer interacting with GFP in DAP10-sh3 and DAP12-sh3 double transductants (Fig. 5, four DAP10 molecules via a three-helix interface in the transmem- A–C). NKG2D expression was preserved even with simultaneous brane domains (46). In DAP10-deficient mice, NKG2D surface silencing of DAP10 and DAP12 after activation in 14–28 days of assembly and signaling can be delivered through noncovalent culture, as determined by FACS. Suppression of in vitro cytolysis binding with DAP12 via expression of an alternative splice variant http://classic.jimmunol.org

FIGURE 6. NKG2D pairs with DAP12 in activated ϩ

Downloaded from CD8 T cells. Western blot of whole cell lysates from activated and expanded CD8ϩ T cells (A) or purified NK cells (B) confirming silencing of DAP10 by DAP10-sh3, whereas NKG2D protein expression is preserved. C, Receptor complexes were immunoprecipitated with a polyclonal anti-DAP12 Ab and blotted for NKG2D, showing pairing of DAP12 with NKG2D regardless of DAP10 silencing in activated and expanded CD8ϩ T cells but not in purified NK cells (D). E, Binding of NKG2D and DAP12 in activated CD8ϩ T cells was confirmed by immunoprecipitation with anti- NKG2D, followed by blotting for DAP12. The Journal of Immunology 7825

FIGURE 7. NKG2D, DAP10, and DAP12 RNAi in human NK cells. CD3ϪCD56ϩ NK cells freshly isolated from normal human donors were evaluated for cytolysis after RNAi for NKG2D, DAP10, and DAP12. RNA (A) and Western blots (B) demonstrate that functional silencing of NKG2D can be achieved in fresh polyclonal human NK cells using NKG2D-sh6 and pLVTHM lentiviral vector. Similar DAP10 and DAP12 RNAi was observed by RNA blot (C) after transduction with DAP10-sh3 and DAP12-sh3. Cytolysis of 51Cr- labeled RPMI 1640 targets by fresh NK cells is suppressed with NKG2D RNAi (D; p ϭ 0.002) and to a similar extent with either DAP10 (p ϭ 0.005) or DAP12 (p ϭ 0.009) RNAi (E). Shown are representative results from one experiment repeated three times.

(22). Because no human NKG2D splice variant has been identified dency on NKG2D for NK tumor cytolysis in these different ex- to date, we were surprised to find that NKG2D surface expression perimental circumstances (23, 47). We tested whether a strong is maintained in activated CD8ϩ T cells despite DAP10 silencing. NKG2D stimulus, serving as the dominant activating signal, can It is possible that human NKG2D isoforms may exist at low levels overcome suppression of killing due to DAP12 knockdown by or only in certain conditions or cell types, however, and thus we performing redirected cytolysis of Fc␥R-bearing P815 cell targets examined whether NKG2D can interact with DAP12 in activated preincubated with an agonistic anti-NKG2D mAb. Vigorous Ab- and expanded CD8ϩ T cells. We first confirmed by Western blot- mediated redirected killing was observed by freshly isolated hu- ting that DAP10-sh3 effectively silences DAP10 expression in man NK cells when transduced with empty pLVTHM vector or both CD8ϩ T cells and NK cells (Fig. 6, A and B). Activated CD8ϩ with control mCD8-sh RNAi, but not after transduction with T cells differ from NK cells, however, in that we observed that NKG2D-sh6 or DAP10-sh3 (Fig. 8A). In contrast to cytolysis of NKG2D can be coimmunoprecipitated with DAP12 regardless of RPMI targets, we observed little suppression of NK cell-redirected whether DAP10 is silenced, which is lost when DAP12 RNAi is killing with DAP12-sh3-mediated RNAi. Suppression of redi- induced (Fig. 6, C–E). These results indicate that human NKG2D rected cytolysis by activated and expanded CD8ϩ T cell persisted, expression can be accounted for by pairing with DAP12 when however, when DAP12 was silenced by RNAi (Fig. 8B). DAP10 is not present in activated and expanded CD8ϩ T cells. NKG2D RNAi in an SCID/Hu tumor model by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. NKG2D, DAP10, and DAP12 RNAi in human NK cells Using NKG2D-sh6 and the lentiviral vector to induce stable gene Using the NKG2D-sh6, DAP10-sh3, and DAP12-sh3 constructs, silencing, we evaluated the role of NKG2D signaling in an in vivo we examined the roles of these signaling molecules in primary NK tumor model. We generated a luciferase-expressing human ovarian cells isolated from normal human donors. These three shRNA and carcinoma cell line, UCI101-luc, and inoculated these cells in the pLVTHM lentivirus were able to elicit high level functional C.B-17 SCID mice, followed by i.v. injections of activated and silencing of NKG2D, DAP10, and DAP12 gene expression in NK expanded human CD8ϩ T cells on days 6 and 12 after tumor seed- cells (Fig. 7, A–C). NK cell killing of RPMI 1640 target cells was ing. Tracking tumor growth by bioluminescence imaging, we ob- not compromised by lentiviral transduction, but was potently sup- served that mice treated with expanded and activated CD8ϩ T cells http://classic.jimmunol.org pressed by NKG2D silencing, which was not seen with RNAi by transduced with NKG2D-sh6 had markedly faster tumor progres- control shRNA (Fig. 7D). This finding is in agreement with ob- sion than animals treated with CD8ϩ T cells not containing the servations by others who have established the general dependency NKG2D-sh6 siRNA (Fig. 9). These results indicate that suppres- by NK cells on NKG2D for in vitro killing of NKG2D ligand- sion of NKG2D expression significantly impacts the in vivo bio- bearing tumor targets (15). Because NKG2D-activating signals ap- logical capability of expanded CD8ϩ T cells in controlling tumor pear to be exclusively transduced by DAP10 in human NK cells growth and progression.

Downloaded from (9), we were not surprised to find a similar degree of NK cell cytolysis suppression with DAP10 gene silencing (Fig. 7E). Un- Discussion expectedly, however, we observed that DAP12 RNAi also sub- In this report we dissect the molecular signaling events leading to stantially suppressed NK cell killing compared with control lenti- cytolysis of tumor cells by activated and expanded human CD8ϩ viral transductions. T cells and NK cells. We developed shRNA specific for human NKG2D, DAP10, and DAP12 for lentiviral transduction that were NKG2D, DAP10, and DAP12 RNAi in redirected cytolysis of ϩ effective for stable gene expression silencing. Using these shRNA NK cells and activated CD8 T cells in RNAi experiments, we confirmed our previous observation that Elegant studies conducted by Rosen et al. (14) provided compel- activated and expanded human CD8ϩ T cells depend on NKG2D ling evidence suggesting that DAP10 is entirely sufficient, and, for in vitro killing and demonstrated that NKG2D is also required hence, DAP12 dispensable, for NKG2D function in human NK for in vivo antitumor activity. This dependency on NKG2D and cells insofar as direct killing of MICA-transduced BaF/3 cell tar- consequently DAP10 signaling, although obligate, is not sufficient gets. The surface density of NKG2D ligands where natively ex- for optimal cytolysis, because partial suppression of killing solely pressed by tumor cell lines can be at low levels, however, which by DAP12 RNAi suggests a nonredundant role for activation ini- may account for conflicting observations of only partial depen- tiated by the DAP12 pathway. 7826 NKG2D, DAP10, AND DAP12 IN CD8ϩ T AND NK CELL CYTOTOXICITY

FIGURE 9. NKG2D RNAi suppresses the antitumor activity of activated and expanded CD8ϩ T cells in an SCID/hu model. The luciferase- expressing ovarian carcinoma cell line UCI101-luc (1 ϫ 106) was im- planted in C.B-17 SCID mice and followed for 30 days by bioluminescence imaging. On days 6 and 12 after tumor inoculation, mice received injec- FIGURE 8. NKG2D, DAP10, and DAP12 RNAi in redirected cytolysis. tions with 40 ϫ 106 activated and expanded human CD8ϩ T cells trans- P815, which is an FcR-bearing murine plasmacytoma cell line, was used as duced with either NKG2D-sh6 or pLVTHM empty vector (No RNAi). A, a target in 51Cr release assays after incubation with anti-NKG2D (10 ␮g/ Bioluminescence imaging on day 30 in representative mice from one ex- ml) mAbs. A, Freshly isolated NK cells effectively kill 51Cr-labeled anti- periment. Little tumor was established after i.p. seeding if mice were sub- NKG2D-coated P815 targets regardless of whether DAP12 is silenced by sequently given injections with activated CD8ϩ T cells not subjected to RNAi with DAP12-sh3 transduction, but not after either NKG2D or NKG2D RNAi (upper right panel). In contrast, i.p. UCI101-luc leads to ϩ DAP10 RNAi transduction. B, Activated and expanded CD8 T cells cy- bulky tumor on all peritoneal surfaces and massive ascites in mice given by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. tolysis is suppressed after NKG2D or DAP10 RNAi transduction; however, NKG2D-silenced activated CD8ϩ T cells (upper left panel). B, Quantifi- in contrast to NK cells, it remains suppressed in the presence of DAP12 cation of tumor signal after injection of expanded CD8ϩ T cells treated ϩ RNAi. Neither NK cells nor activated CD8 T cells kill P815 targets or with NKG2D-sh6 siRNA f, vector; ‚, without RNAi (n ϭ 25 mice). those preincubated with anti-NKG2D. Shown are representative results from one experiment repeated three times. direct killing of NKG2D ligand-bearing tumor as opposed to anti- NKG2D mediated indirect killing of FcR-bearing tumor targets by Although selective and robust inhibition of protein function by freshly isolated polyclonal and unstimulated NK cells. In both in- RNAi has lead to its increasing exploitation as a tool for scientific stances, cytolysis is NKG2D and DAP10 dependent, because http://classic.jimmunol.org investigation, some limitations of the technique merit emphasis knockdown of either gene by RNAi suppressed killing. We found with regard to the current study. Foremost among these concerns that in vitro lysis of RPMI targets was also substantially sup- is the confounding effect of transcript variation, mainly alternative pressed by DAP12 RNAi, however, indicating that signals medi- splice variants, on the effectiveness of RNAi with the shRNA that ated by other activating NK receptors are additionally required for we used. Numerous RNA splice variants of human NKG2D, the greatest killing. This finding is consistent with reports by others DAP10, and DAP12 have been described (12, 48, 49), and a query who observed that lysis of tumor cells naturally expressing

Downloaded from of the public nucleotide sequence databases revealed many more NKG2D ligands is only partially inhibited by blocking with that have not been fully characterized. We have not excluded the NKG2D-specific Abs even at saturating doses (47, 51). Con- possibility that alternative expression of these variants between versely, some target tumor cells that lack expression of NKG2D NK cells vs activated and expanded CD8ϩ T cells or from one ligands can be sensitive to NK cell-mediated cytotoxicity, again normal donor compared with another may influence the extent of suggesting alternative activating receptors (51). This contrasts with gene knockdown with our RNAi. We have otherwise controlled for indirect killing of anti-NKG2D-coated P815 targets, where we the mode of RNAi delivery and the physiologic consequences of found that lysis by NK cells was efficient despite DAP12 silencing. RNAi induction, including those involved in viral host defense This may be a reflection of conditions where selective NKG2D (50), with a nonspecific shRNA approximating the thermodynamic signaling is the predominant stimulus for activation, akin to en- profile expected of an active siRNA. hancement of cytotoxicity by tumor cell lines transfected with The regulation of NK cell activation has been an area of inten- NKG2D ligands that are otherwise resistant to NK cell killing (14, sive investigation, and our results highlight the complexity of the 20, 52). Whether there is a threshold of NKG2D signaling ampli- mechanisms that underlie regulation of tumor cytolysis. Using tude that must be exceeded to effect cytolysis is not known, and RNAi to target signaling molecules involved in NK cell stimula- this may be influenced by differences in the surface density of tion, we observed subtle differences in signaling requirements for NKG2D ligands, as previously suggested (47). Alternatively, The Journal of Immunology 7827

NKG2D binds with its various ligands at markedly different bind- study clearly show that activated and expanded CD8ϩ T cells rely ing affinities (44); thus, heterogeneous tumor cell expression of on NKG2D and DAP10 signaling to acquire cytolysis of tumor NKG2D ligands may influence signaling. targets, but we have yet to understand the role of DAP12 in the We unexpectedly observed that NKG2D expression is main- activation cascade. We have also not clarified the distinction be- tained in activated and expanded CD8ϩ T cells despite DAP10 tween cytolysis as opposed to proliferation and cytokine secretion silencing and showed that binding with DAP12 may account for and the dependency on each of these proximal signaling pathways, this phenomenon. The receptor specificity of DAP10 and DAP12 although our results suggest that cytokine production does not con- resides entirely in their transmembrane domains (9, 14, 46). It is tribute to killing in activated CD8ϩ T cells. We have observed that possible that there are structural variants in the transmembrane activated and expanded murine CD8ϩ T cells efficiently eradicate domain of either NKG2D or the adaptor proteins permissive for tumor, but do not cause graft-vs-host disease (30), and it is pos- pairing of NKG2D with DAP12 in activated CD8ϩ T cells, al- sible that subtle differences in these activation pathways, compared though evidence supporting this has not been found in studies of with those used by unmanipulated alloreactive T cells, may be polyclonal NK cells and IL-2-activated blood mononuclear cells crucial for avoidance of deleterious immune effects. (53). This consideration is further complicated by our observation Ultimately, we aim to test the antitumor activity of activated and that NKG2D is expressed by activated CD8ϩ T cells even when expanded CD8ϩ T cells in rationally designed clinical studies in- both DAP10 and DAP12 are simultaneously silenced. Similarities volving patients with malignant diseases. We provide additional can be drawn to observations made in Jurkat cell lines, which do evidence that CD8ϩ T cells activated and expanded from normal not express DAP12 and express DAP10 only at low levels (5, 12), donors have in vivo tumor activity using an SCID/hu model. We whereby NKG2D expression can be induced by transfection with furthermore show that these activated cells can be generated from an NKG2D expression vector alone (20). Although binding of patients with ovarian cancer, which gain the ability to lyse autol- NKG2D with adaptor proteins other than DAP10 and DAP12 has ogous tumor targets isolated from the same patient. Findings from not been reported, we can speculate on novel associations between this study provide us with a better understanding of the broad NKG2D and yet to be identified receptor complex subunits. Dou- antitumor potential of these cells, enhance our understanding of ble knockdown of DAP10 and DAP12 abolishes much of the kill- regulatory mechanisms supporting cytolysis, and will aid in the ing by activated and expanded CD8ϩ T cells, however, implying implementation of activated and expanded CD8ϩ T cells as a po- that these novel NKG2D receptor complexes are unlikely to be tentially efficacious form of cellular immunotherapy. important in signaling for activation of cytolysis. Unlike NK cells, activated and expanded CD8ϩ T cells require Acknowledgments simultaneous stimulation mediated by DAP10 as well as DAP12 We thank Dr. John Chan for assistance with acquisition of PBMC and pathways for all aspects of in vitro killing evaluated in this study. tumor samples from ovarian cancer patients, and the patients themselves From previous studies we had characterized cytolysis by activated for participating in our study. We also thank Dr. Tim Doyle for providing and expanded T cells as being MHC unrestricted and TCR inde- us with the pMSCVpuroBA-L2G retroviral vector, and Dr. Christopher H. pendent, requiring cell-cell contact, and relying on an intact gran- Contag for his thoughtful advice. Many thanks to Dr. Leon Su for help with vector cloning and discussion of experimental results. We thank Dr. Ruby ule release pathway; more recently, we identified NKG2D as the Wong for assistance with statistical analyses.

by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. critical activating intermediary (23, 29, 42, 43). Our finding that ϩ killing by activated and expanded CD8 T cells is sensitive to Disclosures DAP10 silencing by RNAi is predictable given the exquisite rela- The authors have no financial conflict of interest. tionship between cytolysis and Ab-mediated NKG2D blockade that we had observed from previous studies. That cytolysis can References also be inhibited by DAP12 RNAi was unexpected and raises 1. Lanier, L. L. 2005. NK cell recognition. Annu. Rev. Immunol. 23: 225–274. questions about whether there are key DAP12-binding receptor 2. Yokoyama, W. M., S. Kim, and A. R. French. 2004. The dynamic life of natural ϩ killer cells. Annu. Rev. Immunol. 22: 405–429. partners expressed by activated and expanded CD8 T cells in- 3. Vely, F., and E. Vivier. 1997. Conservation of structural features reveals the fluencing cytolysis. Of clinical relevance, DAP12 loss-of-function existence of a large family of inhibitory cell surface receptors and noninhibitory/ http://classic.jimmunol.org mutations have been identified (49, 54), and it would be important activatory counterparts. J. Immunol. 159: 2075–2077. ϩ 4. Wirthmueller, U., T. Kurosaki, M. S. Murakami, and J. V. Ravetch. 1992. Signal to understand whether CD8 T cells activated and expanded from transduction by Fc␥RIII (CD16) is mediated through the ␥ chain. J. Exp. Med. such individuals would be impaired with regard to tumor killing. 175: 1381–1390. 5. Lanier, L. L., B. C. Corliss, J. Wu, C. Leong, and J. H. Phillips. 1998. Immu- Many emerging lines of evidence now point toward remarkably noreceptor DAP12 bearing a tyrosine-based activation motif is involved in acti- convergent evolution between NK cells and T cells, and our results vating NK cells. Nature 391: 703–707. in this study lend additional support to this idea. Although NK 6. Lanier, L. L., B. Corliss, J. Wu, and J. H. Phillips. 1998. Association of DAP12 with activating CD94/NKG2C NK cell receptors. Immunity 8: 693–701. Downloaded from cells used a strategy of balancing selection to diversify its reper- 7. Vitale, M., C. Bottino, S. Sivori, L. Sanseverino, R. Castriconi, E. Marcenaro, toire of recognition receptors as opposed to T cells, which acquired R. Augugliaro, L. Moretta, and A. Moretta. 1998. NKp44, a novel triggering diversity through germline recombination of the TCR, both center surface molecule specifically expressed by activated natural killer cells, is involved in non-major histocompatibility complex-restricted tumor cell lysis. on interaction with ligands bearing the general structure of MHC J. Exp. Med. 187: 2065–2072. molecules for activation (1, 55). NK cells constitutively express 8. Cantoni, C., C. Bottino, M. Vitale, A. Pessino, R. Augugliaro, A. Malaspina, ␥ S. Parolini, L. Moretta, A. Moretta, and R. Biassoni. 1999. NKp44, a triggering granzyme, perforin, and IFN- , whereas cytotoxic T cells must receptor involved in tumor cell lysis by activated human natural killer cells, is a undergo proliferation and de novo transcription of these gene prod- novel member of the immunoglobulin superfamily. J. Exp. Med. 189: 787–796. ucts to become competent effector cells, yet both ultimately use the 9. Wu, J., H. Cherwinski, T. Spies, J. H. Phillips, and L. L. Lanier. 2000. DAP10 and DAP12 form distinct, but functionally cooperative, receptor complexes in same defense mechanisms to defend against invading pathogens or natural killer cells. J. Exp. Med. 192: 1059–1068. tumor (56). It is interesting to speculate about whether physiologic 10. McVicar, D. W., L. S. Taylor, P. Gosselin, J. Willette-Brown, A. I. Mikhael, conditions mimicking our in vivo activation and expansion proto- R. L. Geahlen, M. C. Nakamura, P. Linnemeyer, W. E. Seaman, S. K. Anderson, et al. 1998. DAP12-mediated signal transduction in natural killer cells: a domi- col exist and how these could be beneficial for effective immune nant role for the Syk protein-tyrosine kinase. J. Biol. Chem. 273: 32934–32942. function, or why a T cell would maintain, but keep dormant, path- 11. Chang, C., J. Dietrich, A. G. Harpur, J. A. Lindquist, A. Haude, Y. W. Loke, A. King, M. Colonna, J. Trowsdale, and M. J. Wilson. 1999. Cutting edge: ways for activation of cytolysis shared with NK cells and, more- KAP10, a novel transmembrane adapter protein genetically linked to DAP12 but over, use it preferentially over its native TCR. Our results in this with unique signaling properties. J. Immunol. 163: 4651–4654. 7828 NKG2D, DAP10, AND DAP12 IN CD8ϩ T AND NK CELL CYTOTOXICITY

12. Wu, J., Y. Song, A. B. Bakker, S. Bauer, T. Spies, L. L. Lanier, and J. H. Phillips. 34. Wiznerowicz, M., and D. Trono. 2003. Conditional suppression of cellular genes: 1999. An activating immunoreceptor complex formed by NKG2D and DAP10. lentivirus vector-mediated drug-inducible RNA interference. J. Virol. 77: Science 285: 730–732. 8957–8961. 13. Billadeau, D. D., J. L. Upshaw, R. A. Schoon, C. J. Dick, and P. J. Leibson. 2003. 35. Brummelkamp, T. R., R. Bernards, and R. Agami. 2002. A system for stable NKG2D-DAP10 triggers human NK cell-mediated killing via a Syk-independent expression of short interfering RNAs in mammalian cells. Science 296: 550–553. regulatory pathway. Nat. Immunol. 4: 557–564. 36. Zufferey, R., D. Nagy, R. J. Mandel, L. Naldini, and D. Trono. 1997. Multiply 14. Rosen, D. B., M. Araki, J. A. Hamerman, T. Chen, T. Yamamura, and attenuated lentiviral vector achieves efficient gene delivery in vivo. Nat. Biotech- L. L. Lanier. 2004. A structural basis for the association of DAP12 with mouse, nol. 15: 871–875. but not human, NKG2D. J. Immunol. 173: 2470–2478. 37. Edinger, M., Y. A. Cao, M. R. Verneris, M. H. Bachmann, C. H. Contag, and 15. Raulet, D. H. 2003. Roles of the NKG2D immunoreceptor and its ligands. Nat. R. S. Negrin. 2003. Revealing lymphoma growth and the efficacy of immune cell Rev. Immunol. 3: 781–790. therapies using in vivo bioluminescence imaging. Blood 101: 640–648. 16. Bakker, A. B., E. Baker, G. R. Sutherland, J. H. Phillips, and L. L. Lanier. 1999. 38. Brown, T. 2004. Analysis of RNA by Northern and slot-blot hybridization. In Myeloid DAP12-associating lectin (MDL)-1 is a cell surface receptor involved in Current Protocols in Immunology, Vol. 3. J. E. Coligan, B. E. Bierer, the activation of myeloid cells. Proc. Natl. Acad. Sci. USA 96: 9792–9796. D. H. Margulies, E. M. Shevach, and W. Strober, eds. Wiley & Sons, Hobokin, pp. 10.12.11–10.12.14. 17. Aoki, N., S. Kimura, Y. Takiyama, Y. Atsuta, A. Abe, K. Sato, and M. Katagiri. 39. Fuchtner, C., D. A. Emma, A. Manetta, G. Gamboa, R. Bernstein, and S. Y. Liao. 2000. The role of the DAP12 signal in mouse myeloid differentiation. J. Immunol. 1993. Characterization of a human ovarian carcinoma cell line: UCI 101. Gy- 165: 3790–3796. necol. Oncol. 48: 203–209. 18. Tomasello, E., P. O. Desmoulins, K. Chemin, S. Guia, H. Cremer, J. Ortaldo, 40. Carbone, E., P. Neri, M. Mesuraca, M. T. Fulciniti, T. Otsuki, D. Pende, V. Groh, P. Love, D. Kaiserlian, and E. Vivier. 2000. Combined natural killer cell and T. Spies, G. Pollio, D. Cosman, et al. 2005. HLA class I, NKG2D, and natural dendritic cell functional deficiency in KARAP/DAP12 loss-of-function mutant cytotoxicity receptors regulate multiple myeloma cell recognition by natural Immunity mice. 13: 355–364. killer cells. Blood 105: 251–258. 19. Snyder, M. R., M. Lucas, E. Vivier, C. M. Weyand, and J. J. Goronzy. 2003. 41. Conejo-Garcia, J. R., F. Benencia, M. C. Courreges, P. A. Gimotty, E. Khang, Selective activation of the c-Jun NH2-terminal protein kinase signaling pathway ϩ R. J. Buckanovich, K. A. Frauwirth, L. Zhang, D. Katsaros, C. B. Thompson, et by stimulatory KIR in the absence of KARAP/DAP12 in CD4 T cells. J. Exp. al. 2004. Ovarian carcinoma expresses the NKG2D ligand Letal and promotes the Med. 197: 437–449. survival and expansion of CD28Ϫ antitumor T cells. Cancer Res. 64: 2175–2182. 20. Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, and T. Spies. 42. Schmidt-Wolf, I. G., P. Lefterova, B. A. Mehta, L. P. Fernandez, D. Huhn, 1999. Activation of NK cells and T cells by NKG2D, a receptor for stress- K. G. Blume, I. L. Weissman, and R. S. Negrin. 1993. Phenotypic characteriza- inducible MICA. Science 285: 727–729. tion and identification of effector cells involved in tumor cell recognition of 21. Groh, V., R. Rhinehart, J. Randolph-Habecker, M. S. Topp, S. R. Riddell, and cytokine-induced killer cells. Exp. Hematol. 21: 1673–1679. T. Spies. 2001. Costimulation of CD8␣␤ T cells by NKG2D via engagement by 43. Mehta, B. A., I. G. Schmidt-Wolf, I. L. Weissman, and R. S. Negrin. 1995. Two MIC induced on virus-infected cells. Nat. Immunol. 2: 255–260. pathways of exocytosis of cytoplasmic granule contents and target cell killing by 22. Diefenbach, A., E. Tomasello, M. Lucas, A. M. Jamieson, J. K. Hsia, E. Vivier, cytokine-induced CD3ϩCD56ϩ killer cells. Blood 86: 3493–3499. and D. H. Raulet. 2002. Selective associations with signaling proteins determine 44. O’Callaghan, C. A., A. Cerwenka, B. E. Willcox, L. L. Lanier, and stimulatory versus costimulatory activity of NKG2D. Nat. Immunol. 3: P. J. Bjorkman. 2001. Molecular competition for NKG2D: H60 and RAE1 com- 1142–1149. pete unequally for NKG2D with dominance of H60. Immunity 15: 201–211. 23. Verneris, M. R., M. Karami, J. Baker, A. Jayaswal, and R. S. Negrin. 2004. Role 45. Vivier, E., J. A. Nunes, and F. Vely. 2004. Natural killer cell signaling pathways. of NKG2D signaling in the cytotoxicity of activated and expanded CD8ϩ T cells. Science 306: 1517–1519. Blood 103: 3065–3072. 46. Garrity, D., M. E. Call, J. Feng, and K. W. Wucherpfennig. 2005. The activating 24. Brooks, C. G., D. L. Urdal, and C. S. Henney. 1983. Lymphokine-driven “dif- NKG2D receptor assembles in the membrane with two signaling dimers into a ferentiation” of cytotoxic T-cell clones into cells with NK-like specificity: cor- hexameric structure. Proc. Natl. Acad. Sci. USA 102: 7641–7646. relations with display of membrane macromolecules. Immunol. Rev. 72: 43–72. 47. Pende, D., P. Rivera, S. Marcenaro, C. C. Chang, R. Biassoni, R. Conte, 25. Schmidt-Wolf, I. G., R. S. Negrin, H. P. Kiem, K. G. Blume, and I. L. Weissman. M. Kubin, D. Cosman, S. Ferrone, L. Moretta, et al. 2002. Major histocompat- 1991. Use of a SCID mouse/human lymphoma model to evaluate cytokine-in- ibility complex class I-related chain A and UL16-binding protein expression on duced killer cells with potent antitumor cell activity. J. Exp. Med. 174: 139–149. tumor cell lines of different histotypes: analysis of tumor susceptibility to 26. Lu, P. H., and R. S. Negrin. 1994. A novel population of expanded human NKG2D-dependent natural killer cell cytotoxicity. Cancer Res. 62: 6178–6186. CD3ϩCD56ϩ cells derived from T cells with potent in vivo antitumor activity in 48. Shum, B. P., L. R. Flodin, D. G. Muir, R. Rajalingam, S. I. Khakoo, S. Cleland, by guest on October 1, 2021. Copyright 2005 Pageant Media Ltd. mice with severe combined immunodeficiency. J. Immunol. 153: 1687–1696. L. A. Guethlein, M. Uhrberg, and P. Parham. 2002. Conservation and variation J. Immunol. 27. Hoyle, C., C. D. Bangs, P. Chang, O. Kamel, B. Mehta, and R. S. Negrin. 1998. in human and common chimpanzee CD94 and NKG2 genes. 168: Expansion of Philadelphia chromosome-negative CD3ϩCD56ϩ cytotoxic cells 240–252. from chronic myeloid leukemia patients: in vitro and in vivo efficacy in severe 49. Toyabe, S., U. Kaneko, and M. Uchiyama. 2004. Decreased DAP12 expression combined immunodeficiency disease mice. Blood 92: 3318–3327. in natural killer lymphocytes from patients with systemic lupus erythematosus is associated with increased transcript mutations. J. Autoimmun. 23: 371–378. 28. Alvarnas, J. C., Y. C. Linn, E. G. Hope, and R. S. Negrin. 2001. Expansion of ϩ ϩ 50. Huppi, K., S. E. Martin, and N. J. Caplen. 2005. Defining and assaying RNAi in cytotoxic CD3 CD56 cells from peripheral blood progenitor cells of patients mammalian cells. Mol. Cell 17: 1–10. undergoing autologous hematopoietic cell transplantation. Biol. Blood Marrow 51. Jamieson, A. M., A. Diefenbach, C. W. McMahon, N. Xiong, J. R. Carlyle, and Transplant. 7: 216–222. D. H. Raulet. 2002. The role of the NKG2D immunoreceptor in immune cell 29. Verneris, M. R., M. Ito, J. Baker, A. Arshi, R. S. Negrin, and J. A. Shizuru. 2001. activation and natural killing. Immunity 17: 19–29. Engineering hematopoietic grafts: purified allogeneic hematopoietic stem cells 52. Diefenbach, A., A. M. Jamieson, S. D. Liu, N. Shastri, and D. H. Raulet. 2000. plus expanded CD8ϩ NK-T cells in the treatment of lymphoma. Biol. Blood http://classic.jimmunol.org Ligands for the murine NKG2D receptor: expression by tumor cells and activa- Marrow Transplant. 7: 532–542. tion of NK cells and macrophages. Nat. Immunol. 1: 119–126. 30. Baker, J., M. R. Verneris, M. Ito, J. A. Shizuru, and R. S. Negrin. 2001. Expan- ϩ 53. Andre, P., R. Castriconi, M. Espeli, N. Anfossi, T. Juarez, S. Hue, H. Conway, sion of cytolytic CD8 natural killer T cells with limited capacity for graft- F. Romagne, A. Dondero, M. Nanni, et al. 2004. Comparative analysis of human ␥ versus-host disease induction due to interferon production. Blood 97: NK cell activation induced by NKG2D and natural cytotoxicity receptors. Eur. 2923–2931. J. Immunol. 34: 961–971. 31. Takayama, T., T. Sekine, M. Makuuchi, S. Yamasaki, T. Kosuge, J. Yamamoto, 54. Paloneva, J., M. Kestila, J. Wu, A. Salminen, T. Bohling, V. Ruotsalainen, K. Shimada, M. Sakamoto, S. Hirohashi, Y. Ohashi, et al. 2000. Adoptive im- P. Hakola, A. B. Bakker, J. H. Phillips, P. Pekkarinen, et al. 2000. Loss-of- munotherapy to lower postsurgical recurrence rates of hepatocellular carcinoma: function mutations in TYROBP (DAP12) result in a presenile dementia with bone Downloaded from a randomised trial. Lancet 356: 802–807. cysts. Nat. Genet. 25: 357–361. 32. Leemhuis, T., S. Wells, C. Scheffold, M. Edinger, and R. S. Negrin. 2005. A 55. Norman, P. J., M. A. Cook, B. S. Carey, C. V. Carrington, D. H. Verity, phase I trial of autologous cytokine-induced killer cells for the treatment of re- K. Hameed, D. D. Ramdath, D. Chandanayingyong, M. Leppert, H. A. Stephens, lapsed Hodgkin disease and non-Hodgkin lymphoma. Biol. Blood Marrow Trans- et al. 2004. SNP haplotypes and allele frequencies show evidence for disruptive plant. 11: 181–187. and balancing selection in the human leukocyte receptor complex. Immunoge- 33. Rubinson, D. A., C. P. Dillon, A. V. Kwiatkowski, C. Sievers, L. Yang, netics 56: 225–237. J. Kopinja, D. L. Rooney, M. M. Ihrig, M. T. McManus, F. B. Gertler, et al. 2003. 56. Stetson, D. B., M. Mohrs, R. L. Reinhardt, J. L. Baron, Z. E. Wang, L. Gapin, A lentivirus-based system to functionally silence genes in primary mammalian M. Kronenberg, and R. M. Locksley. 2003. Constitutive cytokine mRNAs mark cells, stem cells and transgenic mice by RNA interference. Nat. Genet. 33: natural killer (NK) and NK T cells poised for rapid effector function. J. Exp. Med. 401–406. 198: 1069–1076.