Systemic NKG2D Down-Regulation Impairs NK and CD8 T Cell Responses In Vivo1
http://www.100md.com
免疫学杂志 2005年第14期
Abstract
The immunoreceptor NKG2D stimulates activation of cytotoxic lymphocytes upon engagement with MHC class I-related NKG2D ligands of which at least some are expressed inducibly upon exposure to carcinogens, cell stress, or viruses. In this study, we investigated consequences of a persistent NKG2D ligand expression in vivo by using transgenic mice expressing MHC class I chain-related protein A (MICA) under control of the H2-Kb promoter. Although MICA functions as a potent activating ligand of mouse NKG2D, H2-Kb-MICA mice appear healthy without aberrations in lymphocyte subsets. However, NKG2D-mediated cytotoxicity of H2-Kb-MICA NK cells is severely impaired in vitro and in vivo. This deficiency concurs with a pronounced down-regulation of surface NKG2D that is also seen on activated CD8 T cells. As a consequence, H2-Kb-MICA mice fail to reject MICA-expressing tumors and to mount normal CD8 T cell responses upon Listeria infection emphasizing the importance of NKG2D in immunity against tumors and intracellular infectious agents.
Introduction
The C-type lectin-like activating receptor NKG2D is broadly expressed on cytotoxic lymphocytes. In humans, NKG2D is present on most NK cells, CD8 T cells, and T cells in association with the adaptor protein DAP10 (1, 2). In mice, CD8 T cells express NKG2D only upon activation, whereas NK cells, NKT cells, and some T cells constitutively express NKG2D (3, 4). Different from humans, activated mouse NK cells generate a second NKG2D isoform with a shortened cytoplasmic domain (NKG2D-S), which is capable of pairing with both DAP10 and DAP12, whereas the constitutively expressed NKG2D-L isoform exclusively associates with DAP10 (5, 6). DAP10 mediates costimulation of CD8 T cells and triggers cytotoxicity by NK cells, whereas signal transduction via DAP12 augments cytotoxicity and is strictly required for activation of cytokine release (7, 8, 9).
A peculiarity of NKG2D resides in its interaction with a multitude of MHC class I-related ligands of which at least the MIC molecules are expressed inducibly in association with cell stress, infection, or malignant transformation (10, 11, 12). Whereas the ectodomain of NKG2D is fairly conserved in mouse and man, the various MHC class I-related binding partners of NKG2D are highly diverged. In humans, the MHC-encoded MIC molecules MHC class I chain-related proteins A and B (MICA and MICB)3 and five members of the UL16-binding protein (ULBP) family (ULBP1–4; RAET1G) ligate NKG2D and consequently trigger NK cells (13, 14, 15). In vitro, cell stress-inducible MIC molecules are expressed by many tumor cell lines and up-regulated upon infection with human CMV, Mycobacterium tuberculosis, and Escherichia coli (12, 16). In vivo, MIC molecules are not detectable on most healthy tissues, but are expressed on gastrointestinal epithelium, on tumors, and on human CMV-infected cells (8, 16). Recently, MICA expression was reported for tissues affected by autoimmune reactions in patients with rheumatoid arthritis and celiac disease together with evidence for an involvement of NKG2D in the autoimmune pathogenesis of these diseases (17, 18, 19).
In mice, members of the retinoic acid early transcript 1 (RAE-1) protein family, the minor histocompatibility Ag H60, and murine ULBP-like transcript 1 act as ligands of NKG2D (20, 21, 22). Similarly to ULBP molecules they all lack an 3 domain. Recently, up-regulation of RAE-1 molecules on macrophages by various ligands of TLR has been demonstrated (23). RAE-1 expression is also induced by carcinogens and stimulates antitumor activity of T cells (24). RAE-1-transduced tumor cell lines were rejected in vivo due to NK and CD8 T cell responses and induced tumor immunity against the parental cell line supporting a role for NKG2D in tumor immunity (25, 26), though no direct evidence for an involvement of NKG2D was provided.
Recent findings that tumor cells release soluble MIC molecules may account for the failure of tumor surveillance by the NKG2D system in human cancer patients. MICA molecules are shed from tumor cells by metalloproteases resulting in a reduced NKG2D ligand (NKG2DL) surface density (27). Further, soluble MICA (sMICA) was shown to down-regulate NKG2D surface expression and, thereby, to impair the antitumor reactivity of cytotoxic lymphocytes in vitro (28). Substantial levels of sMICA were detected in sera of patients with various malignancies and correlated with a systemic NKG2D down-regulation on peripheral NK and CD8 T cells (27, 29, 30). However, direct evidence for an in vivo impairment of NKG2D-mediated tumor immunosurveillance by persistent MICA expression was lacking. Another study reported NKG2D down-regulation upon coculture with NKG2DL-expressing cells that was at least in part due to signaling of the DAP10 adaptor. In addition, down-regulation of NKG2D was observed for NK cells from NOD mice and attributed to coexpression of RAE-1 (31), but consequences for NK cell activation in vivo were not investigated.
In this study, we explored implications of persistent NKG2DL expression in vivo. We investigated consequences of persistent MICA expression for NKG2D-mediated immunosurveillance as it may occur in cancer patients. In addition, we took advantage of the strongly impaired NKG2D function in H2-Kb-MICA mice to address the role of NKG2D for NK and CD8 T cell responses in vivo.
Materials and Methods
Modified full-length cDNA of H2-Kb and human 2-microglobulin were kindly provided by D. Busch (Technical University of Munich, Munich, Germany). H2-Kb/OVA257–264-tetramers were generated as described (35). For flow cytometric analysis, at day 9 postinfection, 2 x 106 splenocytes were incubated for 15 min at 4°C with rat IgG, anti-CD16/CD32 mAb, and streptavidin (Molecular Probes) in PBS with 0.5% BSA and 0.01% sodium azide. After incubation, cells were stained for 60 min at 4°C either with Cy5-conjugated anti-CD8 mAb, FITC-conjugated anti-CD62L mAb, and PE-conjugated MHC class I-OVA257–264-tetramers, or with allophycocyanin-conjugated anti-NKG2D mAb, FITC-conjugated anti-CD8 mAb, and PE-conjugated MHC class I- OVA257–264-tetramers. Subsequently, cells were washed with PBS 0.5% BSA/0.01% sodium azide and diluted in PBS. Propidium iodide was added before four-color flow cytometric analysis.
Results
MICA expression by H2-Kb-MICA mice
To gain insight into consequences of a persistent NKG2DL expression in vivo, we established transgenic mice constitutively and ubiquitously expressing the human NKG2DL MICA. We corroborated previous data of MIC molecules acting as a ligands of mouse NKG2D by demonstrating that MICA*07-expressing cells bind soluble mouse NKG2D (Fig. 1) (22, 36). To achieve constitutive and ubiquitous MICA expression, a transgene containing the coding sequence of MICA*07 under control of the MHC class I H2-Kb promoter was introduced into the germline of (C57BL/6 x SJL)F1/J mice (Fig. 2A). Offspring expressing MICA on PBL were selected, and a transgenic line (H2-Kb-MICA) was established that was repeatedly backcrossed with C57BL/6 mice (B6 mice).
Previous reports described NKG2D down-regulation by NK cells upon exposure to NKG2DL-expressing cells in vitro and in vivo (23, 28, 31). We observed that down-regulation of surface NKG2D on nontransgenic splenocytes was most pronounced after cocultivation with splenocytes from MICA transgenic mice in vitro, and only marginally following treatment with sera from H2-Kb-MICA mice, whereas incubation with control cells and sera from nontgLM, respectively, had no effect (Fig. 4, B and C). In vivo, NKG2D surface expression on splenocytes from B6 mice was down-regulated after adoptive transfer into H2-Kb-MICA mice, but not after transfer into nontgLM (Fig. 4D). Altogether, these data suggest that reduced surface NKG2D on H2-Kb-MICA NK cells results in NKG2D dysfunction and that NKG2D down-regulation is primarily caused by a persistent exposure to cell-bound MICA in vivo.
NKG2D-mediated natural cytotoxicity is deficient in H2-Kb-MICA mice
To address functional consequences of NKG2D down-regulation for NK cell activation in vivo, we took advantage of H2-Kb-MICA splenocytes. In contrast to Con A-activated splenocytes (Con A blasts) from B6 mice, H2-Kb-MICA Con A blasts were subject of substantial lysis by freshly isolated poly(I:C)-activated B6 NK cells, and lysis was blocked by addition of anti-NKG2D Ab (Fig. 5A). To test the efficacy of NKG2D-mediated NK cytotoxicity in vivo, we adoptively cotransferred CFSE-labeled MICA-transgenic and PKH26-labeled nontransgenic splenocytes into nontgLM. Within 6 h, the relative number of MICA-expressing splenocytes in the peripheral blood was reduced to one-third (Fig. 5B), indicating an efficient and preferential elimination of MICA-expressing cells in vivo. A similar preferential elimination of H2-Kb-MICA splenocytes was observed when lymph nodes and spleens were analyzed (Fig. 5D). However, when CFSE-labeled MICA-transgenic splenocytes were adoptively cotransferred together with PKH26-labeled nontransgenic splenocytes into H2-Kb-MICA mice, relative numbers of MICA-transgenic splenocytes were only slightly decreased indicating that NKG2D-mediated activation of NK cytotoxicity is deficient in H2-Kb-MICA mice (Fig. 5, C and D). To address functionality of NKG2D-independent NK cytotoxicity, we adoptively transferred CFSE-labeled splenocytes from 2-microglobulin-deficient mice. Both H2-Kb-MICA mice and nontgLM exhibited similar clearance rates of 2-microglobulin-deficient cells excluding a general impairment of NK cytotoxicity in H2-Kb-MICA mice (Fig. 5E).
Impaired tumor rejection by H2-Kb-MICA mice
Tumor cells ectopically expressing NKG2DL stimulate tumor immunity. For example, RAE-1-expressing RMA cells have been shown to be rejected by NK cells and/or CD8 T cells (25, 26). We adopted this experimental setting to evaluate functional consequences of NKG2D impairment in vivo and injected 105 RMA-neo and 105 RMA-MICA*07 cells into the right and left flank, respectively, of nontgLM. As expected, RMA-neo cells gave rise to tumors in all mice, but RMA-MICA*07 cells were rejected analogous to previous findings with RAE-1-expressing RMA cells (25, 26) (Fig. 6). To assess a potential contribution of T cells recognizing putative MICA peptides presented by MHC class I in the rejection of RMA-MICA cells, we tested RAG2-deficient mice. RMA-MICA*07 cells were rejected by RAG2-deficient mice, whereas RMA-neo cells expanded to tumors, demonstrating that rejection of RMA-MICA*07 cells occurred independently of T cell recognition, but rather is due to the NKG2D-mediated activation of NK cells reaffirming the functionality of MICA as surrogate ligand of mouse NKG2D. However, when we challenged H2-Kb-MICA mice with injections of RMA cells, both RMA-neo and RMA-MICA*07 gave rise to tumors demonstrating that the NKG2D-mediated tumor rejection is strongly impaired in these mice (Fig. 6).
NKG2D dysfunction compromises CD8 T cell responses to L. monocytogenes
To address the role of NKG2D in the immune defense against infectious pathogens and for the generation of T cell responses, we scrutinized the immune response of H2-Kb-MICA mice upon infection with L. monocytogenes. The control of the intracellular pathogen L. monocytogenes involves both cells of the innate and of the acquired immune system (39). To evaluate consequences of NKG2D impairment for the early control of L. monocytogenes by innate mechanisms, H2-Kb-MICA mice and nontgLM were infected with 5 x 104 listeriae, and the bacteria titers in spleen and liver were determined at days 2 and 3 postinfection. Both groups of mice did not differ significantly in their bacterial load, indicating that NKG2D-mediated effector functions of NK cells are not relevant for the early control of L. monocytogenes (data not shown). For the analysis of Listeria-specific T cell responses, we applied a L. monocytogenes strain recombinant for a secreted form of OVA (34). This strain induces a strong OVA-specific CD8+ T cell response, which can be detected with H2-Kb/OVA257–264-tetramers. At the peak of the primary T cell response against L. monocytogenes (day 9 postinfection), we analyzed tetramer-positive T cells in both H2-Kb-MICA mice and nontgLM. Costaining with anti-NKG2D mAb revealed a significant NKG2D expression on the majority of tetramer-positive cells from nontransgenic mice. This is in accord with earlier studies reporting that CD8 T cells, but not CD4 T cells, show induced NKG2D expression several days after antigenic activation (4). In contrast to nontransgenic mice, NKG2D surface expression of tetramer-positive cells from H2-Kb-MICA mice was strongly reduced (Fig. 7A). Interestingly, spleens of H2-Kb-MICA mice contained significant lower frequencies and total numbers of tetramer-positive CD8 T cells (Fig. 7B and data not shown). These results were confirmed by the analysis of frequencies and numbers of CD8+ T cells responding to in vitro peptide stimulation with IFN- production (Fig. 7C and data not shown). When we compared the frequencies of CD62Llow cells among CD8+ T cells of infected mice, H2-Kb-MICA mice had a significantly reduced percentage of CD62Llow cells (20.0 ± 3.7% and 32.8 ± 3.8% in H2-Kb-MICA mice and littermate controls, respectively), revealing a general impairment of the anti-Listeria CD8+ T cell response in H2-Kb-MICA mice. In contrast, Listeria-specific CD4+ T cell responses were only marginally affected in H2-Kb-MICA mice. Compared with littermate controls, H2-Kb-MICA mice showed similar frequencies and only slightly reduced total numbers of CD4+ T cells responding to the immunodominant Listeria epitope LLO190–201 (Fig. 7D and data not shown).
Discussion
A hallmark of the NKG2D/NKG2DL-system is the inducible surface expression of at least some of the MHC class I-related NKG2DL in response to cell stress, microbial infection, or malignant transformation, thereby marking dysfunctional cells for elimination by cytotoxic lymphocytes via NKG2D-mediated mechanisms ("induced-self" hypothesis) (10, 11, 40). Conversely, a sustained NKG2DL expression as described in patients with malignant or autoimmune diseases may desensitize NKG2D-mediated immune responses. Here, we analyzed transgenic mice constitutively expressing MICA to address functional consequences of persistent NKG2DL expression in vivo. MICA was chosen because it is the best-characterized NKG2DL with regard to expression in normal and diseased tissues, regulation of expression and soluble release, and availability of biochemical and structural data (10, 12). Although the amino acid sequence of the MICA ectodomain is fairly divergent from any mouse NKG2DL, crystal structures of MICA and RAE1 1/2 platform domains interacting with NKG2D are highly related as are crystal structures of human and mouse NKG2D ectodomains (41, 42, 43). Structural homology is reflected in functional equivalence of MICA/B and RAE1 with regard to mouse NKG2D ligation and NKG2D-mediated activation of mouse NK cells qualifying MICA as a bona fide surrogate ligand of mouse NKG2D (Fig. 1) (22, 36).
MICA expression in H2-Kb-MICA mice generally parallels H2-Kb expression with a strong expression by B and NK cells and an intermediate to low expression by thymocytes and hepatocytes. A disparate expression of Kb and MICA molecules was only observed for peripheral CD4 and CD8 T cells that remains to be addressed. MICA-expressing splenocytes were used to evaluate the efficiency of NKG2D-mediated activation of natural cytotoxicity in vivo. Hitherto, NKG2D-mediated cytotoxicity has been extensively demonstrated in vitro using tumor cell lines expressing NKG2DL. Previous studies also demonstrated that ectopic expression of NKG2DL by tumors induces strong NK and CTL responses in vivo and that rejection of NKG2DL-expressing tumor cells is perforin-dependent (25, 26, 44).
By adoptive transfer of MICA-expressing splenocytes, we here demonstrate that ectopic expression of NKG2DL renders "normal" syngeneic cells highly susceptible to cytotoxicity in vivo, vividly underscoring that NKG2DL-expression potently overrides inhibitory signals by MHC class I molecules for NK cell activation.
To our surprise, H2-Kb-MICA mice were vital, fertile, and did not exhibit any overt signs of autoimmunity despite a strong MICA surface expression. A previous study reported hyperkeratosis and leukocytosis in transgenic mice with a MICB cDNA under control of a chicken -actin promoter (45). The difference in phenotype of these mice as compared with the H2-Kb-MICA mice may be due to a different tissue expression of the MIC molecules. It turned out that in H2-Kb-MICA mice "tolerance" toward MICA-expressing cells is established by down-modulation of NKG2D, which is due to permanent exposure to MICA. In fact, down-modulation of surface NKG2D by NK cells after exposure to NKG2DL-expressing cells in vitro and in vivo has been previously reported (23, 31). It remains to be investigated whether a high local NKG2D ligand expression also leads to a systemic NKG2D down-regulation and dysfunction or rather stimulates local immune reactions. Previous studies on rejection of NKG2DL-expressing tumor cell lines and on the role of NKG2D in the pathogenesis of diabetes in NOD mice, respectively, suggest the latter (25, 26, 46).
A down-modulation of NKG2D has also been reported in human cancer patients, where tumors express and release substantial amounts of sMICA (28). Interestingly, H2-Kb-MICA mice also contain high serum levels of sMICA that exceed levels observed in cancer patients at least 10-fold (29) and thus is not a peculiarity of malignant cells. This is in line with the detection of sMICA in patients with autoimmune diseases and raises the question whether MICA shedding is of physiological relevance, e.g., in regulating MIC cell surface levels. Characterization of the MIC shedding activity may resolve this issue. Conflicting results exist regarding the down-regulation of NKG2D by sMICA. Whereas MICA-containing sera of patients with malignancies reportedly cause systemic NKG2D down-regulation correlating with reduced NKG2D expression on CD8 T cells and NK cells in cancer patients (28, 30, 47), NKG2D surface expression is not altered in patients with rheumatoid arthritis and celiac disease despite similar sera levels of sMICA (17, 18). Our in vitro experiments indicate that NKG2D down-regulation in H2-Kb-MICA mice is mainly due to engagement of cell-bound MICA as sMICA had little effect on NKG2D surface expression of nontransgenic NK cells. Possibly, MICA released from tumor cells and benign cells, respectively, may be subjected to different posttranslational modifications differentially affecting NKG2D down-regulation.
Irrespective of the exact molecular mechanism, our results now provide direct in vivo evidence that persistent MICA expression results in down-modulation of NKG2D and impacts tumor immunity as H2-Kb-MICA mice failed to reject MICA-expressing RMA cells in contrast to nontgLM or RAG2-deficient mice. Though our mouse model does not mirror the localized MICA expression by tumors, it does strongly support the notion that ligand-induced NKG2D down-modulation is detrimental for tumor immunosurveillance.
In H2-Kb-MICA mice, down-modulation of surface NKG2D on NK cells was not complete raising the possibility that the observed dysfunction may in part also be due to an impaired signal transduction. However, lysis of CHO cells by H2-Kb-MICA NK cells was similar to lysis by nontransgenic NK cells suggesting that the DAP12 signaling pathway is not affected. Upon in vitro cultivation, H2-Kb-MICA NK cells acquired higher NKG2D expression levels and regained functional activity demonstrating that NKG2DL-induced silencing of NKG2D is reversible. Failure to completely restore NKG2D surface expression levels may also be due to a MICA-NKG2D cis-interaction within same cells as postulated for activated NK cells from NOD mice (31).
In humans, MICA expression has also been reported for thymic epithelial cells raising the possibility that NKG2D is involved in thymic selection of T cells (48). Because MICA costimulates T cells via NKG2D and represents a ligand of some human TCR, constitutive MICA expression might also affect the generation and/or expansion of lymphocyte subpopulations bearing NKG2D or TCR (1, 8, 15, 49, 50). However, we did not detect any major alterations in the total number or composition of lymphocyte subpopulations in naive H2-Kb-MICA mice. In particular, numbers of splenic NK cells, CD8 T cells, and T cells were not significantly altered as compared with nontgLM. Recently, a transgenic mouse with a gut-specific MICA expression directed by the T3b-promoter was described with expansions of CD4+CD8+ intraepithelial T cells (37). But a functional involvement of NKG2D was not examined and also no other mechanisms presented that may account for this expansion. In H2-Kb-MICA mice, frequencies of CD4+CD8+ intraepithelial T cells did not significantly differ from nontgLM.
Because H2-Kb-MICA mice exhibit a profound NKG2D dysfunction, we implemented these mice to address the relevance of NKG2D in staging immune responses toward infectious pathogens. Infection with L. monocytogenes was chosen, because listeriae activate both innate and adaptive immune responses. Further, induction of RAE1-expression by macrophages incubated with L. monocytogenes has recently been reported (23). When we compared the pathogen load of H2-Kb-MICA mice and nontgLM at days 2 and 3 postinfection, we found no significant differences, indicating that NKG2D dysfunction on NK cells had no major impact at an early stage of infection. However, NKG2DL expression induced by L. monocytogenes may trigger cytokine release by NK cells that in turn may contribute to the generation of an anti-Listeria-specific T cell response. NK cells of H2-Kb-MICA mice showed an impaired IFN- production upon NKG2D triggering ex vivo that may also generate a suboptimal generation of an anti-Listeria Th1 response in vivo. However, H2-Kb-MICA NK cells should still be able to respond to other IFN- inducing stimuli (i.e., cytokines such as IL-12 and IL-18) present during listeriosis. T cells are of major importance for the immune control and clearance of Listeria (39). We analyzed the anti-Listeria T cell response at day 9 postinfection and found the number and frequency of Listeria-specific CD8 T cells strongly reduced in H2-Kb-MICA, while Listeria-specific CD4 T cells were not affected. At present, it is unclear why the CD8 T cell response is impaired in Listeria-infected H2-Kb-MICA mice. Down-regulation of NKG2D on activated CD8 T cells may reduce costimulatory signals for proliferation and cell survival. In fact, it has been reported that NKG2D-mediated signal transduction via DAP10 also involves activation of the serine/threonine kinase Akt that promotes cellular proliferation (51). Alternately, an impaired activation of NK cells by NKG2D may result in reduced cytokine secretion and/or cell lysis generating a suboptimal Th1 response.
In summary, we demonstrate that the persistent expression of MICA results in a pronounced down-modulation of NKG2D on NK cells and activated CD8 T cells in vivo. The resulting NKG2D dysfunction strongly impacts immunity against NKG2DL-expressing tumor cells and impairs the generation and/or expansion of Listeria-specific CD8 T cells emphasizing an important function of NKG2D for immunity against tumors and intracellular pathogens.
Acknowledgments
We gratefully acknowledge the excellent technical assistance by Beate P?mmerl and Jessica Bigott, and the experimental assistance by Anouk Feitsma. We thank Keesook Li of the transgenic mouse facility of the Fred Hutchinson Cancer Research Center. We are grateful to Adrian Hayday and colleagues for sharing unpublished data and thank Stefan Bauer for critical reading of the manuscript. The pHSE vector is a kind gift of Hanspeter Pircher.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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 with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from Interdisziplin?res Zentrum für Klinische Forschung Tübingen (Project IIA1; to A.S.), Deutsche Krebshilfe (10-1921-Sa I; to A.S.), Deutsche Forschungsgemeinschaft (SFB633; to H.-W.M.) and the National Institutes of Health (IA-30581; to T.S.).
2 Address correspondence and reprint requests to Dr. Alexander Steinle, Institute for Cell Biology, Department of Immunology, Eberhard Karls University of Tübingen, Auf der Morgenstelle 15, 72076 Tübingen, Germany. E-mail address: alexander.steinle@uni-tuebingen.de
3 Abbreviations used in this paper: MICA, MHC class I chain-related protein A; MICB, MHC class I chain-related protein B; ULBP, UL16-binding protein; RAE-1, retinoic acid early transcript 1; NKG2DL, NKG2D ligand; poly(I:C), polyinosinic-polycytidylic acid potassium salt; sMICA, soluble MICA; nontgLM, nontransgenic littermate; LLO190–201, listeriolysin O.
Received for publication December 29, 2004. Accepted for publication April 30, 2005.
References
Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, T. Spies. 1999. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 285: 727-729.
Wu, J., Y. Song, A. B. Bakker, S. Bauer, T. Spies, L. L. Lanier, J. H. Phillips. 1999. An activating immunoreceptor complex formed by NKG2D and DAP10. Science 285: 730-732.
Ho, E. L., L. N. Carayannopoulos, J. Poursine-Laurent, J. Kinder, B. Plougastel, H. R. Smith, W. M. Yokoyama. 2002. Costimulation of multiple NK cell activation receptors by NKG2D. J. Immunol. 169: 3667-3675.
Jamieson, A. M., A. Diefenbach, C. W. McMahon, N. Xiong, J. R. Carlyle, D. H. Raulet. 2002. The role of the NKG2D immunoreceptor in immune cell activation and natural killing. Immunity 17: 19-29
Diefenbach, A., E. Tomasello, M. Lucas, A. M. Jamieson, J. K. Hsia, E. Vivier, D. H. Raulet. 2002. Selective associations with signaling proteins determine stimulatory versus costimulatory activity of NKG2D. Nat. Immunol. 3: 1142-1149.
Gilfillan, S., E. L. Ho, M. Cella, W. M. Yokoyama, M. Colonna. 2002. NKG2D recruits two distinct adapters to trigger NK cell activation and costimulation. Nat. Immunol. 3: 1150-1155.(Katrin Wiemann*, Hans-Wil)
The immunoreceptor NKG2D stimulates activation of cytotoxic lymphocytes upon engagement with MHC class I-related NKG2D ligands of which at least some are expressed inducibly upon exposure to carcinogens, cell stress, or viruses. In this study, we investigated consequences of a persistent NKG2D ligand expression in vivo by using transgenic mice expressing MHC class I chain-related protein A (MICA) under control of the H2-Kb promoter. Although MICA functions as a potent activating ligand of mouse NKG2D, H2-Kb-MICA mice appear healthy without aberrations in lymphocyte subsets. However, NKG2D-mediated cytotoxicity of H2-Kb-MICA NK cells is severely impaired in vitro and in vivo. This deficiency concurs with a pronounced down-regulation of surface NKG2D that is also seen on activated CD8 T cells. As a consequence, H2-Kb-MICA mice fail to reject MICA-expressing tumors and to mount normal CD8 T cell responses upon Listeria infection emphasizing the importance of NKG2D in immunity against tumors and intracellular infectious agents.
Introduction
The C-type lectin-like activating receptor NKG2D is broadly expressed on cytotoxic lymphocytes. In humans, NKG2D is present on most NK cells, CD8 T cells, and T cells in association with the adaptor protein DAP10 (1, 2). In mice, CD8 T cells express NKG2D only upon activation, whereas NK cells, NKT cells, and some T cells constitutively express NKG2D (3, 4). Different from humans, activated mouse NK cells generate a second NKG2D isoform with a shortened cytoplasmic domain (NKG2D-S), which is capable of pairing with both DAP10 and DAP12, whereas the constitutively expressed NKG2D-L isoform exclusively associates with DAP10 (5, 6). DAP10 mediates costimulation of CD8 T cells and triggers cytotoxicity by NK cells, whereas signal transduction via DAP12 augments cytotoxicity and is strictly required for activation of cytokine release (7, 8, 9).
A peculiarity of NKG2D resides in its interaction with a multitude of MHC class I-related ligands of which at least the MIC molecules are expressed inducibly in association with cell stress, infection, or malignant transformation (10, 11, 12). Whereas the ectodomain of NKG2D is fairly conserved in mouse and man, the various MHC class I-related binding partners of NKG2D are highly diverged. In humans, the MHC-encoded MIC molecules MHC class I chain-related proteins A and B (MICA and MICB)3 and five members of the UL16-binding protein (ULBP) family (ULBP1–4; RAET1G) ligate NKG2D and consequently trigger NK cells (13, 14, 15). In vitro, cell stress-inducible MIC molecules are expressed by many tumor cell lines and up-regulated upon infection with human CMV, Mycobacterium tuberculosis, and Escherichia coli (12, 16). In vivo, MIC molecules are not detectable on most healthy tissues, but are expressed on gastrointestinal epithelium, on tumors, and on human CMV-infected cells (8, 16). Recently, MICA expression was reported for tissues affected by autoimmune reactions in patients with rheumatoid arthritis and celiac disease together with evidence for an involvement of NKG2D in the autoimmune pathogenesis of these diseases (17, 18, 19).
In mice, members of the retinoic acid early transcript 1 (RAE-1) protein family, the minor histocompatibility Ag H60, and murine ULBP-like transcript 1 act as ligands of NKG2D (20, 21, 22). Similarly to ULBP molecules they all lack an 3 domain. Recently, up-regulation of RAE-1 molecules on macrophages by various ligands of TLR has been demonstrated (23). RAE-1 expression is also induced by carcinogens and stimulates antitumor activity of T cells (24). RAE-1-transduced tumor cell lines were rejected in vivo due to NK and CD8 T cell responses and induced tumor immunity against the parental cell line supporting a role for NKG2D in tumor immunity (25, 26), though no direct evidence for an involvement of NKG2D was provided.
Recent findings that tumor cells release soluble MIC molecules may account for the failure of tumor surveillance by the NKG2D system in human cancer patients. MICA molecules are shed from tumor cells by metalloproteases resulting in a reduced NKG2D ligand (NKG2DL) surface density (27). Further, soluble MICA (sMICA) was shown to down-regulate NKG2D surface expression and, thereby, to impair the antitumor reactivity of cytotoxic lymphocytes in vitro (28). Substantial levels of sMICA were detected in sera of patients with various malignancies and correlated with a systemic NKG2D down-regulation on peripheral NK and CD8 T cells (27, 29, 30). However, direct evidence for an in vivo impairment of NKG2D-mediated tumor immunosurveillance by persistent MICA expression was lacking. Another study reported NKG2D down-regulation upon coculture with NKG2DL-expressing cells that was at least in part due to signaling of the DAP10 adaptor. In addition, down-regulation of NKG2D was observed for NK cells from NOD mice and attributed to coexpression of RAE-1 (31), but consequences for NK cell activation in vivo were not investigated.
In this study, we explored implications of persistent NKG2DL expression in vivo. We investigated consequences of persistent MICA expression for NKG2D-mediated immunosurveillance as it may occur in cancer patients. In addition, we took advantage of the strongly impaired NKG2D function in H2-Kb-MICA mice to address the role of NKG2D for NK and CD8 T cell responses in vivo.
Materials and Methods
Modified full-length cDNA of H2-Kb and human 2-microglobulin were kindly provided by D. Busch (Technical University of Munich, Munich, Germany). H2-Kb/OVA257–264-tetramers were generated as described (35). For flow cytometric analysis, at day 9 postinfection, 2 x 106 splenocytes were incubated for 15 min at 4°C with rat IgG, anti-CD16/CD32 mAb, and streptavidin (Molecular Probes) in PBS with 0.5% BSA and 0.01% sodium azide. After incubation, cells were stained for 60 min at 4°C either with Cy5-conjugated anti-CD8 mAb, FITC-conjugated anti-CD62L mAb, and PE-conjugated MHC class I-OVA257–264-tetramers, or with allophycocyanin-conjugated anti-NKG2D mAb, FITC-conjugated anti-CD8 mAb, and PE-conjugated MHC class I- OVA257–264-tetramers. Subsequently, cells were washed with PBS 0.5% BSA/0.01% sodium azide and diluted in PBS. Propidium iodide was added before four-color flow cytometric analysis.
Results
MICA expression by H2-Kb-MICA mice
To gain insight into consequences of a persistent NKG2DL expression in vivo, we established transgenic mice constitutively and ubiquitously expressing the human NKG2DL MICA. We corroborated previous data of MIC molecules acting as a ligands of mouse NKG2D by demonstrating that MICA*07-expressing cells bind soluble mouse NKG2D (Fig. 1) (22, 36). To achieve constitutive and ubiquitous MICA expression, a transgene containing the coding sequence of MICA*07 under control of the MHC class I H2-Kb promoter was introduced into the germline of (C57BL/6 x SJL)F1/J mice (Fig. 2A). Offspring expressing MICA on PBL were selected, and a transgenic line (H2-Kb-MICA) was established that was repeatedly backcrossed with C57BL/6 mice (B6 mice).
Previous reports described NKG2D down-regulation by NK cells upon exposure to NKG2DL-expressing cells in vitro and in vivo (23, 28, 31). We observed that down-regulation of surface NKG2D on nontransgenic splenocytes was most pronounced after cocultivation with splenocytes from MICA transgenic mice in vitro, and only marginally following treatment with sera from H2-Kb-MICA mice, whereas incubation with control cells and sera from nontgLM, respectively, had no effect (Fig. 4, B and C). In vivo, NKG2D surface expression on splenocytes from B6 mice was down-regulated after adoptive transfer into H2-Kb-MICA mice, but not after transfer into nontgLM (Fig. 4D). Altogether, these data suggest that reduced surface NKG2D on H2-Kb-MICA NK cells results in NKG2D dysfunction and that NKG2D down-regulation is primarily caused by a persistent exposure to cell-bound MICA in vivo.
NKG2D-mediated natural cytotoxicity is deficient in H2-Kb-MICA mice
To address functional consequences of NKG2D down-regulation for NK cell activation in vivo, we took advantage of H2-Kb-MICA splenocytes. In contrast to Con A-activated splenocytes (Con A blasts) from B6 mice, H2-Kb-MICA Con A blasts were subject of substantial lysis by freshly isolated poly(I:C)-activated B6 NK cells, and lysis was blocked by addition of anti-NKG2D Ab (Fig. 5A). To test the efficacy of NKG2D-mediated NK cytotoxicity in vivo, we adoptively cotransferred CFSE-labeled MICA-transgenic and PKH26-labeled nontransgenic splenocytes into nontgLM. Within 6 h, the relative number of MICA-expressing splenocytes in the peripheral blood was reduced to one-third (Fig. 5B), indicating an efficient and preferential elimination of MICA-expressing cells in vivo. A similar preferential elimination of H2-Kb-MICA splenocytes was observed when lymph nodes and spleens were analyzed (Fig. 5D). However, when CFSE-labeled MICA-transgenic splenocytes were adoptively cotransferred together with PKH26-labeled nontransgenic splenocytes into H2-Kb-MICA mice, relative numbers of MICA-transgenic splenocytes were only slightly decreased indicating that NKG2D-mediated activation of NK cytotoxicity is deficient in H2-Kb-MICA mice (Fig. 5, C and D). To address functionality of NKG2D-independent NK cytotoxicity, we adoptively transferred CFSE-labeled splenocytes from 2-microglobulin-deficient mice. Both H2-Kb-MICA mice and nontgLM exhibited similar clearance rates of 2-microglobulin-deficient cells excluding a general impairment of NK cytotoxicity in H2-Kb-MICA mice (Fig. 5E).
Impaired tumor rejection by H2-Kb-MICA mice
Tumor cells ectopically expressing NKG2DL stimulate tumor immunity. For example, RAE-1-expressing RMA cells have been shown to be rejected by NK cells and/or CD8 T cells (25, 26). We adopted this experimental setting to evaluate functional consequences of NKG2D impairment in vivo and injected 105 RMA-neo and 105 RMA-MICA*07 cells into the right and left flank, respectively, of nontgLM. As expected, RMA-neo cells gave rise to tumors in all mice, but RMA-MICA*07 cells were rejected analogous to previous findings with RAE-1-expressing RMA cells (25, 26) (Fig. 6). To assess a potential contribution of T cells recognizing putative MICA peptides presented by MHC class I in the rejection of RMA-MICA cells, we tested RAG2-deficient mice. RMA-MICA*07 cells were rejected by RAG2-deficient mice, whereas RMA-neo cells expanded to tumors, demonstrating that rejection of RMA-MICA*07 cells occurred independently of T cell recognition, but rather is due to the NKG2D-mediated activation of NK cells reaffirming the functionality of MICA as surrogate ligand of mouse NKG2D. However, when we challenged H2-Kb-MICA mice with injections of RMA cells, both RMA-neo and RMA-MICA*07 gave rise to tumors demonstrating that the NKG2D-mediated tumor rejection is strongly impaired in these mice (Fig. 6).
NKG2D dysfunction compromises CD8 T cell responses to L. monocytogenes
To address the role of NKG2D in the immune defense against infectious pathogens and for the generation of T cell responses, we scrutinized the immune response of H2-Kb-MICA mice upon infection with L. monocytogenes. The control of the intracellular pathogen L. monocytogenes involves both cells of the innate and of the acquired immune system (39). To evaluate consequences of NKG2D impairment for the early control of L. monocytogenes by innate mechanisms, H2-Kb-MICA mice and nontgLM were infected with 5 x 104 listeriae, and the bacteria titers in spleen and liver were determined at days 2 and 3 postinfection. Both groups of mice did not differ significantly in their bacterial load, indicating that NKG2D-mediated effector functions of NK cells are not relevant for the early control of L. monocytogenes (data not shown). For the analysis of Listeria-specific T cell responses, we applied a L. monocytogenes strain recombinant for a secreted form of OVA (34). This strain induces a strong OVA-specific CD8+ T cell response, which can be detected with H2-Kb/OVA257–264-tetramers. At the peak of the primary T cell response against L. monocytogenes (day 9 postinfection), we analyzed tetramer-positive T cells in both H2-Kb-MICA mice and nontgLM. Costaining with anti-NKG2D mAb revealed a significant NKG2D expression on the majority of tetramer-positive cells from nontransgenic mice. This is in accord with earlier studies reporting that CD8 T cells, but not CD4 T cells, show induced NKG2D expression several days after antigenic activation (4). In contrast to nontransgenic mice, NKG2D surface expression of tetramer-positive cells from H2-Kb-MICA mice was strongly reduced (Fig. 7A). Interestingly, spleens of H2-Kb-MICA mice contained significant lower frequencies and total numbers of tetramer-positive CD8 T cells (Fig. 7B and data not shown). These results were confirmed by the analysis of frequencies and numbers of CD8+ T cells responding to in vitro peptide stimulation with IFN- production (Fig. 7C and data not shown). When we compared the frequencies of CD62Llow cells among CD8+ T cells of infected mice, H2-Kb-MICA mice had a significantly reduced percentage of CD62Llow cells (20.0 ± 3.7% and 32.8 ± 3.8% in H2-Kb-MICA mice and littermate controls, respectively), revealing a general impairment of the anti-Listeria CD8+ T cell response in H2-Kb-MICA mice. In contrast, Listeria-specific CD4+ T cell responses were only marginally affected in H2-Kb-MICA mice. Compared with littermate controls, H2-Kb-MICA mice showed similar frequencies and only slightly reduced total numbers of CD4+ T cells responding to the immunodominant Listeria epitope LLO190–201 (Fig. 7D and data not shown).
Discussion
A hallmark of the NKG2D/NKG2DL-system is the inducible surface expression of at least some of the MHC class I-related NKG2DL in response to cell stress, microbial infection, or malignant transformation, thereby marking dysfunctional cells for elimination by cytotoxic lymphocytes via NKG2D-mediated mechanisms ("induced-self" hypothesis) (10, 11, 40). Conversely, a sustained NKG2DL expression as described in patients with malignant or autoimmune diseases may desensitize NKG2D-mediated immune responses. Here, we analyzed transgenic mice constitutively expressing MICA to address functional consequences of persistent NKG2DL expression in vivo. MICA was chosen because it is the best-characterized NKG2DL with regard to expression in normal and diseased tissues, regulation of expression and soluble release, and availability of biochemical and structural data (10, 12). Although the amino acid sequence of the MICA ectodomain is fairly divergent from any mouse NKG2DL, crystal structures of MICA and RAE1 1/2 platform domains interacting with NKG2D are highly related as are crystal structures of human and mouse NKG2D ectodomains (41, 42, 43). Structural homology is reflected in functional equivalence of MICA/B and RAE1 with regard to mouse NKG2D ligation and NKG2D-mediated activation of mouse NK cells qualifying MICA as a bona fide surrogate ligand of mouse NKG2D (Fig. 1) (22, 36).
MICA expression in H2-Kb-MICA mice generally parallels H2-Kb expression with a strong expression by B and NK cells and an intermediate to low expression by thymocytes and hepatocytes. A disparate expression of Kb and MICA molecules was only observed for peripheral CD4 and CD8 T cells that remains to be addressed. MICA-expressing splenocytes were used to evaluate the efficiency of NKG2D-mediated activation of natural cytotoxicity in vivo. Hitherto, NKG2D-mediated cytotoxicity has been extensively demonstrated in vitro using tumor cell lines expressing NKG2DL. Previous studies also demonstrated that ectopic expression of NKG2DL by tumors induces strong NK and CTL responses in vivo and that rejection of NKG2DL-expressing tumor cells is perforin-dependent (25, 26, 44).
By adoptive transfer of MICA-expressing splenocytes, we here demonstrate that ectopic expression of NKG2DL renders "normal" syngeneic cells highly susceptible to cytotoxicity in vivo, vividly underscoring that NKG2DL-expression potently overrides inhibitory signals by MHC class I molecules for NK cell activation.
To our surprise, H2-Kb-MICA mice were vital, fertile, and did not exhibit any overt signs of autoimmunity despite a strong MICA surface expression. A previous study reported hyperkeratosis and leukocytosis in transgenic mice with a MICB cDNA under control of a chicken -actin promoter (45). The difference in phenotype of these mice as compared with the H2-Kb-MICA mice may be due to a different tissue expression of the MIC molecules. It turned out that in H2-Kb-MICA mice "tolerance" toward MICA-expressing cells is established by down-modulation of NKG2D, which is due to permanent exposure to MICA. In fact, down-modulation of surface NKG2D by NK cells after exposure to NKG2DL-expressing cells in vitro and in vivo has been previously reported (23, 31). It remains to be investigated whether a high local NKG2D ligand expression also leads to a systemic NKG2D down-regulation and dysfunction or rather stimulates local immune reactions. Previous studies on rejection of NKG2DL-expressing tumor cell lines and on the role of NKG2D in the pathogenesis of diabetes in NOD mice, respectively, suggest the latter (25, 26, 46).
A down-modulation of NKG2D has also been reported in human cancer patients, where tumors express and release substantial amounts of sMICA (28). Interestingly, H2-Kb-MICA mice also contain high serum levels of sMICA that exceed levels observed in cancer patients at least 10-fold (29) and thus is not a peculiarity of malignant cells. This is in line with the detection of sMICA in patients with autoimmune diseases and raises the question whether MICA shedding is of physiological relevance, e.g., in regulating MIC cell surface levels. Characterization of the MIC shedding activity may resolve this issue. Conflicting results exist regarding the down-regulation of NKG2D by sMICA. Whereas MICA-containing sera of patients with malignancies reportedly cause systemic NKG2D down-regulation correlating with reduced NKG2D expression on CD8 T cells and NK cells in cancer patients (28, 30, 47), NKG2D surface expression is not altered in patients with rheumatoid arthritis and celiac disease despite similar sera levels of sMICA (17, 18). Our in vitro experiments indicate that NKG2D down-regulation in H2-Kb-MICA mice is mainly due to engagement of cell-bound MICA as sMICA had little effect on NKG2D surface expression of nontransgenic NK cells. Possibly, MICA released from tumor cells and benign cells, respectively, may be subjected to different posttranslational modifications differentially affecting NKG2D down-regulation.
Irrespective of the exact molecular mechanism, our results now provide direct in vivo evidence that persistent MICA expression results in down-modulation of NKG2D and impacts tumor immunity as H2-Kb-MICA mice failed to reject MICA-expressing RMA cells in contrast to nontgLM or RAG2-deficient mice. Though our mouse model does not mirror the localized MICA expression by tumors, it does strongly support the notion that ligand-induced NKG2D down-modulation is detrimental for tumor immunosurveillance.
In H2-Kb-MICA mice, down-modulation of surface NKG2D on NK cells was not complete raising the possibility that the observed dysfunction may in part also be due to an impaired signal transduction. However, lysis of CHO cells by H2-Kb-MICA NK cells was similar to lysis by nontransgenic NK cells suggesting that the DAP12 signaling pathway is not affected. Upon in vitro cultivation, H2-Kb-MICA NK cells acquired higher NKG2D expression levels and regained functional activity demonstrating that NKG2DL-induced silencing of NKG2D is reversible. Failure to completely restore NKG2D surface expression levels may also be due to a MICA-NKG2D cis-interaction within same cells as postulated for activated NK cells from NOD mice (31).
In humans, MICA expression has also been reported for thymic epithelial cells raising the possibility that NKG2D is involved in thymic selection of T cells (48). Because MICA costimulates T cells via NKG2D and represents a ligand of some human TCR, constitutive MICA expression might also affect the generation and/or expansion of lymphocyte subpopulations bearing NKG2D or TCR (1, 8, 15, 49, 50). However, we did not detect any major alterations in the total number or composition of lymphocyte subpopulations in naive H2-Kb-MICA mice. In particular, numbers of splenic NK cells, CD8 T cells, and T cells were not significantly altered as compared with nontgLM. Recently, a transgenic mouse with a gut-specific MICA expression directed by the T3b-promoter was described with expansions of CD4+CD8+ intraepithelial T cells (37). But a functional involvement of NKG2D was not examined and also no other mechanisms presented that may account for this expansion. In H2-Kb-MICA mice, frequencies of CD4+CD8+ intraepithelial T cells did not significantly differ from nontgLM.
Because H2-Kb-MICA mice exhibit a profound NKG2D dysfunction, we implemented these mice to address the relevance of NKG2D in staging immune responses toward infectious pathogens. Infection with L. monocytogenes was chosen, because listeriae activate both innate and adaptive immune responses. Further, induction of RAE1-expression by macrophages incubated with L. monocytogenes has recently been reported (23). When we compared the pathogen load of H2-Kb-MICA mice and nontgLM at days 2 and 3 postinfection, we found no significant differences, indicating that NKG2D dysfunction on NK cells had no major impact at an early stage of infection. However, NKG2DL expression induced by L. monocytogenes may trigger cytokine release by NK cells that in turn may contribute to the generation of an anti-Listeria-specific T cell response. NK cells of H2-Kb-MICA mice showed an impaired IFN- production upon NKG2D triggering ex vivo that may also generate a suboptimal generation of an anti-Listeria Th1 response in vivo. However, H2-Kb-MICA NK cells should still be able to respond to other IFN- inducing stimuli (i.e., cytokines such as IL-12 and IL-18) present during listeriosis. T cells are of major importance for the immune control and clearance of Listeria (39). We analyzed the anti-Listeria T cell response at day 9 postinfection and found the number and frequency of Listeria-specific CD8 T cells strongly reduced in H2-Kb-MICA, while Listeria-specific CD4 T cells were not affected. At present, it is unclear why the CD8 T cell response is impaired in Listeria-infected H2-Kb-MICA mice. Down-regulation of NKG2D on activated CD8 T cells may reduce costimulatory signals for proliferation and cell survival. In fact, it has been reported that NKG2D-mediated signal transduction via DAP10 also involves activation of the serine/threonine kinase Akt that promotes cellular proliferation (51). Alternately, an impaired activation of NK cells by NKG2D may result in reduced cytokine secretion and/or cell lysis generating a suboptimal Th1 response.
In summary, we demonstrate that the persistent expression of MICA results in a pronounced down-modulation of NKG2D on NK cells and activated CD8 T cells in vivo. The resulting NKG2D dysfunction strongly impacts immunity against NKG2DL-expressing tumor cells and impairs the generation and/or expansion of Listeria-specific CD8 T cells emphasizing an important function of NKG2D for immunity against tumors and intracellular pathogens.
Acknowledgments
We gratefully acknowledge the excellent technical assistance by Beate P?mmerl and Jessica Bigott, and the experimental assistance by Anouk Feitsma. We thank Keesook Li of the transgenic mouse facility of the Fred Hutchinson Cancer Research Center. We are grateful to Adrian Hayday and colleagues for sharing unpublished data and thank Stefan Bauer for critical reading of the manuscript. The pHSE vector is a kind gift of Hanspeter Pircher.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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 with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from Interdisziplin?res Zentrum für Klinische Forschung Tübingen (Project IIA1; to A.S.), Deutsche Krebshilfe (10-1921-Sa I; to A.S.), Deutsche Forschungsgemeinschaft (SFB633; to H.-W.M.) and the National Institutes of Health (IA-30581; to T.S.).
2 Address correspondence and reprint requests to Dr. Alexander Steinle, Institute for Cell Biology, Department of Immunology, Eberhard Karls University of Tübingen, Auf der Morgenstelle 15, 72076 Tübingen, Germany. E-mail address: alexander.steinle@uni-tuebingen.de
3 Abbreviations used in this paper: MICA, MHC class I chain-related protein A; MICB, MHC class I chain-related protein B; ULBP, UL16-binding protein; RAE-1, retinoic acid early transcript 1; NKG2DL, NKG2D ligand; poly(I:C), polyinosinic-polycytidylic acid potassium salt; sMICA, soluble MICA; nontgLM, nontransgenic littermate; LLO190–201, listeriolysin O.
Received for publication December 29, 2004. Accepted for publication April 30, 2005.
References
Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, T. Spies. 1999. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 285: 727-729.
Wu, J., Y. Song, A. B. Bakker, S. Bauer, T. Spies, L. L. Lanier, J. H. Phillips. 1999. An activating immunoreceptor complex formed by NKG2D and DAP10. Science 285: 730-732.
Ho, E. L., L. N. Carayannopoulos, J. Poursine-Laurent, J. Kinder, B. Plougastel, H. R. Smith, W. M. Yokoyama. 2002. Costimulation of multiple NK cell activation receptors by NKG2D. J. Immunol. 169: 3667-3675.
Jamieson, A. M., A. Diefenbach, C. W. McMahon, N. Xiong, J. R. Carlyle, D. H. Raulet. 2002. The role of the NKG2D immunoreceptor in immune cell activation and natural killing. Immunity 17: 19-29
Diefenbach, A., E. Tomasello, M. Lucas, A. M. Jamieson, J. K. Hsia, E. Vivier, D. H. Raulet. 2002. Selective associations with signaling proteins determine stimulatory versus costimulatory activity of NKG2D. Nat. Immunol. 3: 1142-1149.
Gilfillan, S., E. L. Ho, M. Cella, W. M. Yokoyama, M. Colonna. 2002. NKG2D recruits two distinct adapters to trigger NK cell activation and costimulation. Nat. Immunol. 3: 1150-1155.(Katrin Wiemann*, Hans-Wil)