Natural killer group 2 member D (NKG2D) is a type II transmembrane receptor encoded by the killer cell lectin-like receptor subfamily K member 1 (Klrk1) gene.1,2 NKG2D is present on all murine and human NK cells. It is constitutively expressed on human CD8+ T cells, whereas its expression on CD8+ T cells occurs after activation in mice.3 In addition to NK and CD8+ T cells, NKG2D can also be expressed by natural killer T (NKT) cells, γδ T cells, activated murine macrophages and a small subgroup of CD4+ T cells.4 The functions of NKG2D, however, differ between NK cells and CD8+ T cells. In NK cells, NKG2D signaling alone is sufficient to mediate direct killing of target cells,4 whereas in CD8+ T cells, simultaneous activation of the T-cell receptor (TCR) is required for NKG2D to be functional. In these cells, NKG2D acts as a costimulatory receptor, and its signaling leads to enhanced TCR activation and T-cell function.3,4,5

NKG2D is a promiscuous receptor, binding to a variety of stress ligands. These ligands include members of the Rae-1ϵ and H60a-c families6,7 in mice and major histocompatibility complex (MHC) class-I-related chain (MIC) A/B and UL16 binding proteins 1–6 in humans.8 The expression of NKG2D ligands on the cell surface is strictly regulated in the body. Under healthy conditions, few or no NKG2D ligands are expressed.6,7 However, expression of these ligands is induced by transcriptional upregulation under stress-related conditions, such as infection and transformation.6,7 Notably, other forms of cellular insult (e.g., exposure to DNA-damaging agents, TLR signaling, cell proliferation or exposure to certain cytokines) can also result in surface expression of these ligands.5,8,9

NKG2D has a short intracellular domain that lacks any signaling motifs. In CD8+ T cells, NKG2D signals through the adaptor protein DNAX-activating protein 10 (DAP10), whereas in NK cells, it can pair with either DAP10 or its homolog DAP12.3,10,11 DAP10 contains a YINM motif, which recruits p85 to induce PI3K signaling and Grb2 to activate Vav-SOS signaling.12,13 Interestingly, the same YxNM signaling motif exists within the CD28 costimulatory receptor,14,15,16 suggesting that these motifs are the result of coevolutionary adaptation, as both NKG2D and CD28 function as costimulatory receptors in CD8+ T cells. However, it is important to note that CD28 is primarily required during the priming of naïve T cells, while NKG2D is used by effector and memory CD8+ T cells.

In addition to its costimulatory activity, NKG2D appears to be involved in non-canonical functions, such as memory formation, although the few studies that exist concerning this subject are conflicting. For instance, memory CD8+ T cells can be activated by cytokines only under specific conditions, which transform these cells into NK-like killer cells. In such cases, they eliminate target cells through an NKG2D-mediated mechanism without TCR engagement. Importantly, this TCR-independent killing has been linked to the development of several autoimmune disorders, highlighting the importance of NKG2D under pathophysiological conditions. This review will focus on the less well-described characteristics of NKG2D. The similarities and differences between NKG2D and CD28 will be discussed, as well as the potential role of NKG2D in CD8+ T-cell memory formation. Last, the involvement of NKG2D signaling under various pathological conditions, such as autoimmune diseases and cancer, will also be discussed (Figure 1).

Figure 1
figure 1

NKG2D’s role in physiology, diseases and its potential use in medical interventions: In CD8+ T cells, NKG2D signaling is thought to occur exclusively via DAP10 and requires simultaneous TCR engagement, and thus acting as an enhancer rather than an inducer of activation. DAP10 uses a YINM motif to recruit PI3K and Grb2, and activate the associated pathways. Major outcomes of the signaling is enhanced TCR signaling, cytotoxicity and enhanced T-cell survival. However, alterations in the NKG2D signaling and expression levels can lead to autoimmune diseases that are either TCR dependent: vitiligo, type I diabetes, and RA or TCR-independent: celiac disease. Based on the contribution of NKG2D signaling to CD8+ T-cell function, approaches are being investigated for the treatment of cancer, autoimmunity and some infections diseases. DAP10, DNAX-activating protein 10; NKG2D, natural killer group 2 member D; RA, rheumatoid arthritis; TCR, T-cell receptor.

NKG2D and CD28 costimulatory functions in CD8+ T cells

In CD8+ T cells, engagement of the NK receptor, NKG2D, results in enhanced TCR activation and function,1,3,4,5,9 which facilitates the recognition and destruction of target cells.3 The NKG2D receptor has been defined as a costimulatory molecule in CD8+ T cells, and thus, it is often compared with CD28, an extensively studied costimulatory receptor in CD4+ and CD8+ T cells. Despite sharing many similarities, both the NKG2D and CD28 receptors possess distinct expression patterns, structures, signaling pathways and functional outcomes.

In humans, both CD28 and NKG2D are expressed in CD8+ T cells; however, the proportions of these cells expressing these receptors differ. NKG2D is expressed by all CD8+ T cells,3 whereas CD28 is expressed by ~50% of mature human CD8+ T cells, in addition to the vast majority (~95%) of human CD4+ T cells.17 The NKG2D receptor has gained increasing research interest based on the widespread expression of its ligands compared to those of CD28 (CD80/CD86), which are expressed only in antigen-presenting cells (APCs). NKG2D ligands can be expressed in any cell type upon cell stress, including viral infection and DNA damage. This is also true in cancer cells, underscoring the relevance of this receptor under pathophysiological conditions. While NKG2D ligands do not have any known association with inhibitory receptors, the CD28 ligands B7.1 and B7.2 are shared by the inhibitory receptor cytotoxic T-lymphocyte-associated protein-4 (CTLA-4), which blocks CD28-dependent T-cell responses.18,19,20

Both NKG2D and CD28 receptors use a similar YxNM motif to transduce signaling pathways. However, they differ remarkably in their structural properties and assembly, which determine the unique functional outcomes of their signals. CD28 is expressed as a homodimer on the cell surface with an extracellular ligand-binding domain, a transmembrane domain and an intracellular domain containing the YMNM signaling motif, among other motifs (Figure 2).14,15,16 NKG2D, lacking any signaling motif, relies on the YINM motif-containing adaptor protein DAP10 to initiate its signaling pathways.21,22 The NKG2D-DAP10 complex is expressed as a hexamer consisting of a homodimer of two NKG2D molecules, each bound to two DAP10 molecules (Figure 2).23 Thus, each CD28 and NKG2D-DAP10 complex consists of two and four YxNM signaling motifs, respectively. In addition to these structural and stoichiometric differences, CD28 expresses additional non-YxNM signaling domains. These include the membrane proximal proline-rich PRRP domain, which can recruit inducible-2 (IL-2) T-cell kinase, and the distal PYAP domain, which is capable of binding to Lck, Filamin A and Grb2.16

Figure 2
figure 2

Differences in the structure and signaling motifs of CD28 and NKG2D receptors: NKG2D-DAP10 complex is a hexameric structure with two NKG2D molecules linked to total of four DAP10 adaptors, whereas CD28 is expressed as a homodimer on the cell membrane. Both receptors contain YxNM signaling motif capable of recruiting Grb2 and p85 subunit of PI3K. In addition, CD28 possesses PRRP, and PYAP domains capable of recruiting Itk, Lck, Fil A and Grb2. DAP10, DNAX-activating protein 10; ltk, IL-2-inducible T-cell kinase; NKG2D, natural killer group 2 member D; PI3K, phosphatidylinositol-3 kinase.

For both CD28 and NKG2D-DAP10 receptors, activation leads to phosphorylation of tyrosine within the YxNM motif by Src family kinase, which is followed by the recruitment of Grb2 and p85 subunits to asparagine and methionine residues, respectively. This in turn results in the activation of a wide array of well-characterized Grb2→Vav and p85→PDK1→Akt signaling pathways, ultimately resulting in AP-1, NFAT and NF-κB nuclear translocation and subsequent increases in cell survival, proliferation and expression of effector molecules and cytokines, as well as the release of cytolytic granules.14,15,16,24,25

In NK cells, it has been shown that NKG2D-DAP10 specifically stimulates phospholipase C-γ2 (PLC-γ2) downstream of the Grb2-Vav1 pathway.26 For CD28, no such preference for PLC-γ1 has been described. It is thought that PLC-γ1 and PLC-γ2 have redundant functions,27,28 and thus the functional relevance of this preferential signaling of NKG2D is not clear. It is important to note that although both of these isoforms are equally expressed in NK cells, PLC-γ1 is preferentially expressed over PLC-γ2 in T cells.29 Based on previous studies that have established PLC-γ as a key signaling molecule for costimulation signaling, it is plausible that NKG2D’s costimulatory potential is reduced compared with CD28 in CD8+ T cells. The same study demonstrated that overexpression of PLC-γ2, but not PLC-γ1, led to increased NKG2D-mediated cytotoxicity in the majority of human CTL clones.26 However, the differences between NKG2D and CD28 signaling in the context of differential PLC-γ isoform expression were not addressed in this work.

In 2003, it was reported that NKG2D-DAP10 does not activate LAT, an important adaptor molecule that recruits key proteins in T-cell signaling, in NK cells.30 Notably, CD28 has been shown to induce phosphorylation of LAT independent of Syk or ZAP70 kinases.31 One study reported that although CD28 is necessary to phosphorylate LAT, its signaling motif-containing cytoplasmic tail is not involved in the initiation or persistence of TCR-induced LAT phosphorylation.32 NKG2D’s effects on LAT in the context of TCR stimulation remain to be determined. As mentioned above, CD28 has two additional signaling domains that are absent in DAP10, namely, PRRP and PYAP. Itk binds to the PRRP domain, the importance of which is not yet clear in CD28 signaling. PYAP recruits Lck, Filamin A and Grb2, and activates signaling pathways that lead to cytoskeletal rearrangement and greater cytokine mRNA stability.16 Although it apparently lacks this signaling potential, NKG2D-DAP10 has been linked to other non-YINM signaling pathways. In experiments performed with human NK cells, Sutherland et al.33 demonstrated that NKG2D stimulation leads to specific phosphorylation of JAK2 and STAT5. However, the molecular basis and functional relevance of this pathway is not yet completely understood.

Both CD28 and NKG2D stimulation have been shown to have canonical costimulatory effects on T cells, including increases in cell survival, proliferation, expression of effector molecules and cytokines, and release of cytolytic granules.14,15,16,24,25 Many studies have reported that NKG2D is capable of costimulating human and mouse T cells to an extent similar to that of CD28 at the level of proliferation, cytolytic function and interferon-γ (IFN-γ) production.4,34,35,36 It is important to note that factors such as cell activation status and cytokine treatment can alter the responsiveness of CD8+ T cells to NKG2D. For instance, Lanier and co-workers37 were unable to costimulate human and murine CD8+ T cells by engaging the NKG2D receptor (unlike CD28), presumably due to different culture conditions and/or cytokine milieu. Owing to the unique characteristics of the signaling pathways discussed above, it is not altogether surprising that NKG2D and CD28 would have different functional effects on T cells. In experiments performed on DUC18 transgenic CTLs, only CD28 could promote survival upon antigen encounter, whereas only NKG2D could form immunological synapses in the absence of antigen.34 Interestingly, in another study, Barber and Sentman38 demonstrated that in either non-transduced or NKG2D-transduced human CD8+ T cells, NKG2D engagement specifically activated β-catenin, increased the expression of β-catenin-induced genes and led to reduced production of the anti-inflammatory cytokines IL-9, IL-10, IL-13 and vascular endothelial growth factor-α in a peroxisome proliferator-activated receptor-γ-dependent manner. CD28 stimulation either failed to exert these effects or caused effects opposite to that of NKG2D in this study. It was also found that both CD28 and NKG2D stimulation decreased the migration of human CD8+ T cells; however, only NKG2D activated Cdc42, a Rho GTPase involved in T-cell motility, and only NKG2D’s effects on T-cell migration were dependent on N-WASP.39 These data suggest that CD28 and NKG2D may also use different signaling pathways to produce similar functional outcomes in T cells. There is also evidence that both receptors can synergize with each other or modulate the other’s function,36,38 suggesting possible crosstalk between the two receptors at certain signaling notches. Hu et al.40 studied the effects of CD28 signaling on the expression of NKG2D and found that CD28 costimulation resulted in sustained NKG2D expression in murine and human CD8+ T cells via Lck/JAK/STAT3 signaling.

Collectively, both NKG2D and CD28, despite sharing a similar YxNM motif and associated upstream signaling events, lead to unique signaling cascades, most likely due to different structures, recruitment of adaptor proteins, downstream effectors and possible crosstalk with other immune receptors. This results in NKG2D- or CD28-mediated costimulation of CD8+ T cells with unique features/functional outcomes. In addition to the above factors, differential expression levels of these receptors on T cells as well as those of the activating ligands in different tissues and cytokine environments may dictate the relevance of NKG2D and CD28 at the physiological level.

NKG2D and CD8+ T-cell memory formation

During the normal course of an immune response, the formation of memory CD8+ T cells is one of the fundamental goals of the adaptive immune system. T-cell differentiation into effector or memory cells is a tightly regulated process that is still not fully understood. Current views suggest that T-cell memory formation involves a passive transition from effector to memory T cells and that memory is the result of initial signaling cues promoting long-term survival.41 Among the known cues, CD4+ T cells and cytokines, such as IL-15, have been shown to play an essential role in not only activating CD8+ T cells but also programing them for memory formation.42,43,44,45,46,47 However, recent work has increasingly established that other factors, including NKG2D signaling, are responsible for CD8+ T-cell memory formation.48,49,50,51,52 Mammalian target of rapamycin complex 1 (mTORC1) has been shown to play a pivotal role in deciding the fate of primed CD8+ T cells. Low mTORC1 levels favor differentiation into memory precursor cells, while high levels of mTORC1 activity leads to the development of terminally differentiated CD8+ T cells.49 Interestingly, McQueen et al.49 demonstrated that both CD28 and NKG2D signaling activated the mTORC1 pathway but to different extents. CD28 signaling induced strong mTORC1 signaling, which led to the transcription of marker genes of terminally differentiated effector CD8+ T cells, that is, T-bet, BLIMP1 or IRF-4. In contrast, activation of NKG2D weakly induced the mTORC1 signaling pathway,49 resulting in the transcription of memory-associated genes, such as CD62L, Eomes and CD127. Therefore, mTORC1 activation by CD28 and/or NKG2D mediates CD8+ T-cell differentiation into terminally differentiated cells or memory cells.

CD8+ T-cell priming in the absence of CD4+ T-cell help results in the formation of so-called ‘helpless’ or ‘unhelped’ CD8+ T cells.42,43,44 Notably, these helpless CD8+ T cells have reduced effector functions and little to no memory formation.42,43,44 Published work from our lab has demonstrated that immunological help for CD8+ T cells in memory formation can also be provided via NKG2D signaling.52 We demonstrated, using a murine vaccination model of helpless CD8+ T cells overexpressing NKG2D ligands during immunization of mice depleted of CD4+ T cells, that these mice completely restored the quantity and quality of the CD8+ T-cell memory response (Figure 3). Importantly, this rescue was dependent on NKG2D expression on CD8+ T cells and was independent of NK cells. This was shown in experiments using CD8+ T cells from NKG2D-deficient mice and NK cell-depleted animals. Further indicating the biological importance of these findings, we also determined that NKG2D engagement by antibody crosslinking was able to rescue the function of extremely suppressed CD8+ T cells from HIV+ patients with low CD4+ T-cell counts. At the molecular level, we found that this rescue was associated with the repression of T-bet, a transcription factor involved in the terminal differentiation of effector cells.52 Paradoxically, while NKG2D signaling was able to rescue CD8+ T-cell memory formation, the initial effector response remained compromised.52 These studies indicate that NKG2D signaling on CD8+ T cells extends beyond known canonical functions (e.g., facilitating target recognition and favored killing), by providing a prosurvival signal. Moreover, these data support the hypothesis that the effector and memory functions of T cells can be disconnected, and CD8+ T-cell memory formation can occur in the absence of an efficient effector response.

Figure 3
figure 3

NKG2D rescues the function of CD4+ T-cell-unhelped CD8+ T cells. (ac) depict three possible scenarios of CD8+ T cells priming by an APC, which lead to three different outcomes for the CD8+ T cells: classical CD8+ T-cell priming by p:MHC-I (peptide-MHC-I complex) on APCs in the presence of CD4+ T-cell help, which results in the optimal proliferation, CTL capacity and formation of memory (a). These properties are defective when CD8+ T cell are activated in the absence of CD4+ T-cell help (b), which can be rescued by engaging NKG2D on CD8+ T cells during their priming (c). APC, antigen-presenting cell; CTL, cytotoxic T-lymphocyte; NKG2D, natural killer group 2 member D; MHC, major histocompatibility complex.

To date, few studies have investigated the role of NKG2D signaling in CD8+ T-cell memory formation. A study by Wensveen et al.51 showed that NKG2D signaling enhanced IL-15-mediated PI3K activity in activated CD8+ T cells, thus assisting in memory commitment. In a series of adoptive transfer experiments, the authors found that NKG2D deficiency in OT-I cells resulted in impaired immunity against challenge with LCMV expressing OVA. Moreover, in vitro-activated T cells upregulated the antiapoptotic Mcl-1 protein, which was found to be downregulated in NKG2D-KO CD8+ T cells,51 suggesting that NKG2D induces memory formation by increasing levels of survival proteins, such as Mcl-1. In contrast to these findings, a study by Andre et al.48 suggested that NKG2D had no major role in the generation and expansion of memory CD8+ T cells, but instead substantially enhanced the cytolytic effector responses of reactivated memory T cells, thereby contributing to efficacious tumor rejection. In this study, the authors used a transgenic mouse model in which the human NKG2D ligand MHC class-I chain-related proteins A (MICA) was ubiquitously and constitutively expressed, resulting in severe dysfunction of NKG2D signaling due to negative feedback loops. These mice were used to assess the rejection of OVA-expressing tumors (EG7) by tumor-specific memory CD8+ T cells. The authors observed that the generation and magnitude of memory CD8+ T-cell responses were independent of NKG2D function, whereas in the effector phase, efficient tumor rejection by CD8+ T cells was dependent on NKG2D. However, there may be a possible relationship between strong antigens and the need for NKG2D signaling. In this case, OVA antigen could be considered a strong TCR agonist, and OVA-reactive CD8+ T cells may therefore be NKG2D-independent. Moreover, in these studies, NKG2D signaling was lacking throughout the life of the CD8+ T cells (e.g., CD8+ NKG2D-KO T cells or dysfunctional NKG2D), which may have unexpected effects on their peripheral behavior.

These studies indicate that CD8+ T-cell memory formation is a dynamic process in which multiple factors have distinct and redundant contributions. Based on these data, the role of NKG2D signaling in memory formation may be to transcriptionally certify CD8+ T cells that have actually recognized and killed stressed cells. While this may not apply to CD8+ T cells with high-affinity TCRs, this may play an important role in CD8+ T cells that have weaker TCRs or are responding in a subinflammatory environment. Moreover, these studies raise important questions, such as whether the reduced quality of a less-potent memory response is due only to the lack of NKG2D signaling or whether it is a reflection of a poor effector phase. Future studies will be required to fully delineate the role of NKG2D in CD8+ T-cell memory formation.

NKG2D ligands and their regulation

In humans, NKG2D ligands include two families: MICA and MICB and HCMV UL16-binding proteins (ULBP1–6). While these ligands are HLA class I homologs, they do not associate with β-2m and play no known role in antigen presentation. In mice, there are three families of ligands for NKG2D: the Rae-1 family, which includes five members (Rae-1α, Rae-1β, Rae-1γ, Rae-1δ and Rae-1ϵ); the H60 family, which contains three members (H60a, H60b, H60c); and the UL16-binding protein-like transcript 1 family, consisting only of MULT1.

The expression of NKG2D ligands is strictly regulated in normal cells and little or no NKG2DL is expressed in healthy adult tissues.6 NKG2D ligands can be upregulated following activation of the ATM/ATR (ataxia telangiectasia mutated/ATM- and Rad3-related) DNA damage repair pathway. Studies have shown that DNA-damaging agents (fluorouracil or cisplatin) or local ionizing radiation can induce NKG2D ligand surface expression in an ATM/ATR-dependent manner.53 In addition, activation of other pathways, for example, by TLR4 and TLR7/8 agonists, also induces NKG2D ligand expression.54 Moreover, cytokines such as IFN-γ and tumor growth factor-β can affect the expression of NKG2D ligands. IFN-γ downregulates the expression of NKG2D ligands in murine and human melanoma cells,55,56 while transforming growth factor-β has also been shown to decrease the transcription of several NKG2D ligands in glioma cells.57 Interestingly, studies have demonstrated that inhibition of the proteasomal pathway can also induce the expression of NKG2D ligands.58 Previous research has shown that heat shock protein (HSP70) and several NKG2D ligands share the same promoter sequence, which serves as a docking site for HSP 1 (HSF1).59 In response to cellular stress, HSF1 is released in the cytoplasm. Following nuclear translocation, HSF1 binds to the promoter regions of both HSP70i and NKG2D ligands, initiating their transcription.59 Expression of NKG2D ligands acts as a clear indicator of cell damage, which can be interpreted as suicide by proxy (in NKG2D-expressing cells). However, the spatiotemporal control of HSP70i (prosurvival) and NKG2D ligand (suicide by proxy) expression in stressed cells remains to be elucidated, representing an important area of translational research.

NKG2D and autoimmunity

The prevention of autoimmunity is regulated by both central and peripheral mechanisms. The central mechanisms involve thymic education, in which early self-reactive T cells that strongly react with peptide:MHC complexes undergo negative selection in the thymus, during which cell death is induced. Nevertheless, many T cells expressing low-affinity TCRs against self-antigen are able to populate the periphery. In most cases, these self-reactive T cells remain inactive. However, under certain pathological conditions, they can become activated and mediate host tissue destruction, causing autoimmunity. There is emerging evidence that NKG2D signaling is involved in certain forms of T-cell-mediated autoimmunity.52,60,61

Although interactions between autoreactive TCRs and peptide:MHC complexes are normally of low affinity, the outcome of these interactions can be influenced by factors outside the TCR:peptide:MHC complex. NKG2D is an activating receptor that enhances TCR signaling,3 potentially reducing the threshold for T-cell activation via the TCR. In the great majority of cases, autoimmunity does not occur during CD8+ T-cell responses against intracellular infections. However, in certain instances, TCR-independent cytotoxic functions of NKG2D have been observed with higher expression levels on CD8+ T cells when NKG2D is engaged by its ligands.

The involvement of NKG2D in potentiating CD8+ T-cell responses against low-affinity peptides has been demonstrated by our group and others in the context of antitumor immunity and vitiligo.62 In the latter, there is an association between early inflammatory events (e.g., exposure to UV light, a bleaching agent, phenols or trauma) and the onset of disease.63 This is significant, as the expression of NKG2D ligands is upregulated in response to cell stress, rendering these cells visible to NKG2D-expressing immune cells.1,3,4,5,9 Consistent with these observations, we demonstrated that engagement of NKG2D, but not 2B4 (another activating receptor), resulted in the exacerbation of CD8+ T-cell-mediated vitiligo using vaccine mouse models.62 These data indicate that autoimmune vitiligo is limited by insufficient signaling, despite plentiful self-reactive T cells in the peripheral immune system. Importantly, studies have also shown a direct role for CD8+ T cells expressing NKG2D in another skin-related autoimmune disease, alopecia areata,64,65,66 indicating strong similarities between both diseases.

A similar situation has been reported for type I diabetes, in which mouse studies from the Lanier group demonstrated that NKG2D is involved during the onset of the disease. The authors found that Rae-1 is present in prediabetic pancreas islets of non-obese diabetic mice and that autoreactive CD8+ T cells infiltrating the pancreas express NKG2D, resulting in the destruction of insulin-producing β-cells by T cells.67 Additionally, treatment with a non-depleting anti-NKG2D monoclonal antibody (mAb) during the prediabetic stage completely prevented disease by impairing the expansion and function of autoreactive CD8+ T cells.

In chronic inflammatory rheumatoid arthritis (RA), data from Andersson et al.60 work has demonstrated the expression of NKG2D and its ligands on human RA synovial cells and in the paws of arthritic mice. The authors demonstrated that, in established collagen-induced arthritis, in vivo blockade of NKG2D with an anti-NKG2D mAb ameliorated the disease and protected against pathologic joint damage. In this study, histologic analysis of arthritic joints from anti-NKG2D-treated mice demonstrated significant joint protection compared with that of control mice.60 These studies presented compelling data demonstrating that self-reactive CD8+ T cells can be activated by NKG2D-mediated mechanisms that can lead to the destruction of host tissues. Considering that tumors are stressed self-tissues, it would be expected that NKG2D signaling plays a role in the recognition of NKG2D ligand-expressing cancer cells.

While NKG2D is a potent costimulator of TCR-mediated effector functions, its physiological behavior is tightly associated with TCR function, and NKG2D is not expected to function independently of TCR signaling.4,30,68 However, in celiac disease, an autoimmune disease elicited by gluten intolerance characterized by the destruction of the small intestine, studies by Bana Jabri and co-workers61,69 have shown that IL-15 drives the upregulation of NKG2D, ultimately enabling CD8+ T cells to kill in a TCR-independent manner through NKG2D-mediated mechanisms. The authors found that NK cells deficient in NKG2D-DAP10 expression were unable to respond to IL-15. Alternatively, the same study reported that IL-15-activated Jak3 could phosphorylate DAP10 and thus prime the NKG2D-DAP10 signaling pathway. In this way, IL-15 was capable of converting effector T cells into NK-like ‘lymphokine-activated killers’ (LAK cells) both in vitro and in vivo in celiac patients. Based on these data, it has been proposed that DAP10 couples with the IL-15 receptor. Notably, in this study, it was shown that the IL-15/NKG2D/DAP10 pathway conferred TCR-independent cytolytic functions on T cells in the intestinal epithelium.

In chronic inflammatory RA, data from Andersson and co-workers’60 work has demonstrated the expression of NKG2D and its ligands on human RA synovial cells and in the paws of arthritic mice. The authors demonstrated that, in established collagen-induced arthritis, in vivo blockade of NKG2D with an anti-NKG2D mAb ameliorated the disease and protected against pathologic joint damage. In this study, histologic analysis of arthritic joints from anti-NKG2D-treated mice demonstrated significant joint protection compared with that of control mice. These studies presented compelling data demonstrating that self-reactive CD8+ T cells can be activated by NKG2D-mediated mechanisms that can lead to the destruction of host tissues. Considering that tumors are stressed self-tissues, it would be expected that NKG2D signaling plays a role in the recognition of NKG2D ligand-expressing cancer cells. Owing to its role in enhancing the cytotoxic response of CD8+ T cells (in a TCR-dependent and TCR-independent manner) and its pathophysiological role, NKG2D may be able to be exploited as a target for autoimmune disease treatment. Notably, since antitumor T-cell responses can be viewed as a form of autoimmunity, the relationship between these two different diseases may provide an opportunity for the development of novel cancer therapies.

NKG2D and cancer immunity

Initial support for a role for NKG2D in tumor immunity was provided by studies using tumor cell lines engineered to express NKG2D ligands. In these studies, mice were able to reject tumor cells expressing NKG2D ligands but not unmanipulated counterparts.70 The mechanisms mediating rejection were based on NK cells and CD8+ T cells. Subsequent studies by Smyth et al.71 using methylcholanthrene-treated mice demonstrated that in vivo NKG2D neutralization resulted in a higher incidence of sarcoma. This observation was followed by studies conducted by Guerra et al.70 using NKG2D-deficient mice crossed with TRAMP mice (SV40 large T-driven prostate cancer). In this case, the absence of NKG2D in TRAMP mice resulted in early tumor development and growth compared with TRAMP mice expressing NKG2D. These observations demonstrated that tumor onset and progression were accelerated in the absence of NKG2D or NKG2D ligands.

Importantly, studies by Liu et al.72 demonstrated dual and opposing roles for NKG2D in cancer. The authors used mice modified to express human MICB and a shedding-resistant mutant MICB variant under the control of a prostate-specific promoter. These mice were then crossed to TRAMP mice. TRAMP/MICB mice (MICB is shed) developed prostate tumors at a faster rate than TRAMP mice. However, TRAMP/shedding-resistant-MICB mice remained tumor-free. This interesting dichotomy was explained by elevated levels of soluble MICB and the associated depletion of NK cells in TRAMP/MICB mice.

Challenging several of the conclusions drawn from these observations, studies using mice with severe dysfunction of NKG2D signaling due to systemic expression of MICA revealed that these mice were able to reject EG7 tumors with similar efficacy as their wild-type counterparts.48 However, that EG7 cells are a thymoma cell line engineered to express the highly immunogenic antigen chicken ovalbumin. Thus, it is possible that CD8+ T-cell responses against OVA-derived peptide will be mediated by high-affinity TCRs, which would be sufficient to mediate tumor killing in an NKG2D-independent manner.

Despite their differing conclusions, these studies emphasize the important role of NKG2D in immune responses against tumors and suggest the possibility of exploiting this antitumor role in the design of novel therapeutic approaches against cancer.

Final remarks

Based on the physiological role of NKG2D in immune responses against stressed cells (both infected and cancerous cells) and the broad expression of its ligands on cancer cells, NKG2D is an attractive target for clinical development of novel therapeutic agents. Current strategies are focused on inducing NKG2D signaling during CD8+ T-cell priming (i.e., vaccine approaches), engaging NKG2D receptors on effector cells (i.e., CD8+ T cells), targeting NKG2D ligands on cancer cells (i.e., NKG2D-CAR T cells), inducing ligands for NKG2D in cancer cells (i.e., radiotherapy) and preventing the suppressive effects of shed-off NKG2D ligands on immune cells (i.e., blocking antibodies).

We have shown using a DNA vaccination approach that genes encoding NKG2D ligands can be incorporated in a vaccine, along with the desired antigen. In this study, vaccination with genes encoding Rae-1ϵ or H60 and the melanocyte shared antigen Trp1 resulted in numerically and functionally enhanced CD8+ T-cell immune responses against Trp1-derived peptides compared to vaccination with Trp1 alone.62 Other approaches are being considered, such as in vitro activation of NKG2D and TCR pathways in CD8+ T cells before adoptive cell transfer. This approach would provide transferred CD8+ T cells with enhanced cytolytic and survival capacities optimal for antitumor immune responses. One additional benefit may be that increasing or maintaining the expression of NKG2D in CD8+ T cells could result in resistance to immunosuppressive factors in the tumor environment.

An alternative approach is the use of NKG2D-CAR T cells. Here, the rationale is that these engineered T cells can eliminate tumors that express ligands for NKG2D. Studies in mouse models of lymphoma, ovarian cancer and multiple myeloma have shown that targeting NKG2D ligands with NKG2D-CAR T cells results in effective antitumor responses.73,74,75 However, NKG2D ligand expression on non-tumor cells, as part of the response to activation of DNA repair mechanisms involving the ATM/ATR repair pathways, represents a safety concern. This could result in ‘on-target, off-tumor’ responses, leading to significant and potentially lethal toxicity. Indeed, murine studies have demonstrated that adoptive transfer of NKG2D-CAR T cells resulted in fatal responses.76,77 Importantly, investigators have established that the severity of such toxicity varied, depending on the mouse strain used as the recipient. Specifically, BALB/c mice were more sensitive to toxicity than C57BL/6 mice.77 These findings are of significance and must be considered when attempting to treat human patients, who are highly heterogeneous. Thus, approaches that target tumor cells expressing NKG2D ligands must be considered with caution. Notably, the use of NKG2D-CAR T cells is currently being tested in a phase I clinical trial, and safety and toxicity results are expected soon.

NKG2D ligands represent an attractive target for therapies, and inducing their expression represents a viable approach for the induction of antitumor immune responses. Irradiation induces activation of the ATM/ATR repair pathways, resulting in the expression of NKG2D ligands. To take advantage of this, one could envision the combined use of local (tumor) radiation and immune checkpoint therapies, namely, anti-PD-1 mAbs. Moreover, several drugs can trigger the expression of NKG2D ligands. Among these are DNA alkylating agents, proteasome inhibitors58 and HDAC inhibitors.78,79,80 Again, however, a key consideration must be that the expression of NKG2D ligands is limited to cancer cells; otherwise, the development of autoimmune responses would be a likely outcome.

In addition to cancer, NKG2D’s role has also been evaluated in anti-viral immunity in various studies. In patients with chronic hepatic viral infections, such as HBV and HCV, NKG2D expression was found to be higher on intrahepatic CD8+ T cells and was able to costimulate HBV-specific CD8+ T cells.81 Similar costimulatory roles of NKG2D have been reported for virus-specific CD8+ T cells in cytomegalovirus.68 Furthermore, in a study focusing on virus-induced encephalitis in mice, blocking NKG2D receptors resulted in a reduction of the cytolytic capacity of the virus-specific CD8+ T cells and an increase in viral titers, as well as an increase in associated mortality.82 Along similar lines, blockade of NKG2D signaling led to reduced clearance of Theiler's murine encephalomyelitis virus in the murine CNS.83 Recently, Kavazović et al.84 showed that a lack of NKG2D on virus-specific murine CD8+ T cells attenuated their cytokine production and ability to clear cytomegalovirus infection. Taken together, these data demonstrate the important role of NKG2D in enhancing the effector function of CD8+ T cells against viral infections. Based on these and our findings, one can speculate that NKG2D-based approaches may be used for the treatment of intracellular pathogens. For example, in the case of HIV infection, which results in low CD4+ T-cell levels, exogenous NKG2D activation would be expected to provide a useful substitute for CD4+ T-cell assistance to CD8+ T cells. An additional benefit of NKG2D signaling would be that the functionally exhausted phenotype that characterizes CD8+ T cells from HIV patients may be reversed. Similar assumptions can also be made regarding the use of NKG2D in other chronic infections, such as HCV and HBV. Based on our current understanding of NKG2D in autoimmunity, studies investigating how NKG2D contributes to the onset of these diseases and its role in autoimmune disease perpetuation are warranted.

Therapeutically, one can envision the development of NKG2D blocking reagents that could prevent the progression of disease by interfering with the recognition of NKG2D ligands by CD8+ T cells. Based on the discussed studies, we strongly believe that continued efforts towards enhancing immunity in cancer treatment and controlling autoimmunity will follow NKG2D-based approaches.

In summary, although much remains to be elucidated concerning NKG2D’s diverse roles in the immune system, significant evidence suggests that it may represent a novel therapeutic target for multiple historically intransigent diseases.