NKG2D is the most versatile and widely distributed activating/co-stimulatory natural killer (NK)-related receptor. First identified in human NK cells in 1991 (9), NKG2D has since been discovered on numerous cell subsets including, activated (αβ and γδ) T cells, natural killer T (NKT) cells, and mucosal-associated invariant T (MAIT) cells (10–13). Increasingly, NKG2D expression is also being identified on tissue-resident innate lymphoid cells (ILC), such as certain ILC1 (14, 15), ILC2 (16) and ILC3 (17) subsets. In humans, NKG2D is encoded by the KLRK1 gene and is located within the NK gene complex (NKC) on chromosome 12p. Moreover, NKG2D is highly conserved across multiple vertebrate species (18, 19). To date, two major human haplotype alleles of NKG2D have been identified, termed LNK1 (low activity) and HNK1 (high activity) alleles (20, 21), with surface expression of NKG2D lower in carriers of the low activity LNK1/LNK1 genotype (20, 22). When expressed at the cell surface, the 42 kDa homodimeric NKG2D receptor combines with the DNAX-activating protein 10 (DAP10) homodimer and following its engagement with cognate ligands initiates a cytotoxic cellular response and/or the secretion of pro-inflammatory cytokines ( Figure 1A ) (23–25).

NKG2D is also capable of facilitating the function of other activating receptors, depending on the inflammatory milieu and/or expressing cell type. For example, in activated NK cells primed by pro-inflammatory cytokines (e.g. interleukin (IL)-2 and IL-15), NKG2D provides direct stimulatory signals (26–29), whereas in resting NK cells, it synergizes as a co-activator with other receptors, such as NKp46 and 2B4 (30, 31). In αβ T cells, NKG2D typically provides a co-stimulatory signal, acting to promote T cell receptor (TCR)-dependent cytotoxicity, production of pro-inflammatory cytokines and memory differentiation (32–37). Interestingly, prolonged exposure to IL-15 has been shown to increase expression of NKG2D in CD8⁺ T cell subsets, potentiating TCR-independent activation (38, 39). Similarly in γδ T cells, NKG2D can function as a co-stimulatory molecule (40), but may also directly trigger cytotoxicity in a TCR-independent fashion (41). Alternatively, some γδ T cells have been shown to bypass the NKG2D receptor and recognize NKG2D-L, such as ULBP4 or MICA/B, directly via their TCR, implying a TCR agonistic role (42–44). In innate-like T cells, such as invariant NKT cells, NKG2D is restricted to the CD4^- subsets, and functions to mediate direct lysis of target cells and co-stimulatory activation (12). Whereas, in MAIT cells, NKG2D is more prominent on CD8⁺ subsets, and functions as a co-stimulatory molecule (45) or, if in the presence of IL-15, exerts NKG2D-dependent innate-like cytotoxicity (13, 46). Lastly, in ILC subsets, early studies highlight that expression of NKG2D may aid in the production of pro- or anti-inflammatory mediators depending on the surrounding microenvironment (15, 16). However, further investigation is required to completely elucidate the impact of NKG2D-mediated signaling on ILC function. Altogether, although NKG2D expression, regulation and function differ across the above cell types, it undoubtedly plays a central regulatory role in the immune response and is vital for immunological surveillance against tumorigenic transformation and viral infection.

Ligands for NKG2D comprise several families of major histocompatibility complex (MHC) class I-related molecules. In humans, these include the MHC class I polypeptide-related sequence A (MICA) and B (MICB), and the human cytomegalovirus (HCMV) glycoprotein UL16-binding protein (ULBP) family (ULBP1-ULBP6) ( Figure 1B ) (47–50). NKG2D-L are, for the most part, not constitutively expressed, but instead are selectively induced upon cellular stress, damage or transformation, as is caused by events such as viral infection or oncogenesis (32, 51). Moreover, as reviewed by Lanier (52), essentially all cell types are capable of expressing one or more types of NKG2D-L if given the appropriate stimulus. For example, Fujita and colleagues (53) identified two distinct ligand expression profiles in non-neoplastic epithelial tissues: ULBP5-ULBP3-MICA/B and ULBP2/6-ULBP1-ULBP4. Moreover, in cells undergoing tumorigenic transformation, high heterogeneity in NKG2D-L expression has been reported. Notably, expression of two or more NKG2D-Ls (often MICA and MICB) is more common in solid tumors, compared to hematological tumors (shown to predominantly express MICB) (54). Furthermore, co-expression of multiple allelic forms of the same ligand has also previously been identified at the surface of cells undergoing stress, which is strongly suggestive of functional redundancy in these molecules. It is also worth noting that NKG2D-L differ in their affinity (K_D ) and avidity for NKG2D, such that ULBP1 (the only ULBP member tested to date) has the highest affinity (1.1 μM), followed by MICA (0.9-1 μM) and MICB (800 nM) (55–59). Therefore, it is speculated that NKG2D may transduce different signals or activate separate downstream pathways based on which ligand or allelic variant is bound (60), which supports the various functions of NKG2D discussed prior.

Surprisingly, the role of NKG2D-L extends beyond providing a signal for cellular stress. In cells of the myeloid lineage, these ligands can mediate lymphocyte activation leading to cytotoxicity, cytokine production and proliferation. For instance, expression of MICA/B on dendritic cell-derived exosomes plays an important role in promoting NK cell differentiation and proliferation (61). Furthermore, ULBP2/3 expression levels are increased during CD34⁺ hematopoietic progenitor commitment to the granulomonocyte lineage, suggesting that NKG2D-L play a role in promoting myeloid differentiation (62). Alternatively, NKG2D-L expression on myeloid cells can lead to lymphocyte inactivation and maintenance of immune homeostasis. For example, persistent expression of membrane bound ULBP1 and MICB on myeloid cells induces NKG2D internalization and desensitization of NK cells (63). Moreover, overexpression of MICA on activated CD8⁺ T cells makes them susceptible to NK cell lysis, indicating that NKG2D-L may participate in immune homeostasis during ongoing immune responses (64). During infection with Mycobacterium tuberculosis, heightened ULBP1 expression on expanded T regulatory cells (Tregs) facilitates NK cell-mediated killing of these cells, thereby enhancing the overall immune response (65). Ultimately, NKG2D-L expression is associated with both cytotoxic and regulatory processes, as is reflected by the diverse roles played by these molecules in host immunity.

The MICA and MICB genes were originally described as stress-induced MHC class I polypeptide-related sequences and are located in the MHC region on the short arm of chromosome 6 (47, 66). These genes are highly polymorphic (67), with, to date, over 100 described alleles (allelefrequencies.net). The MIC alleles have variations that are, to a certain extent, concentrated in the extracellular domains as well as truncated forms due to coding frame-shifts (e.g. MICA 5.1). Specific alleles have been associated with disease outcomes (68–70), influence the amount of soluble protein and impact binding affinity to its cognate receptor, NKG2D (71). Moreover, numerous MICA splice variants have been documented thus far ( Figure 1C ), with the majority binding NKG2D similarly as their wildtype counterparts, highlighting that lack of a domain (e.g. α3 domain) does not necessarily reduce binding affinity (60). Interestingly, there is a naturally occurring MICA-MICB null combination (deletion of MICA and premature stop codon for MICB) that exists on the HLA-B48 haplotype found in East Asian and South American populations (72–74). Furthermore, the MICA-MICB genes are merged in chimpanzees resulting in a hybrid form (75), while they are absent in mice (76). The presence of the null haplotype without obvious phenotypic consequences suggests redundancy in the NKG2D receptor-ligand axis but its overall importance is highlighted by the overlapping mechanisms exhibited by cancers and viruses to evade it (77).

Although the ULBPs are distantly related to MICA/B in sequence, they differ in their location, mapping instead to the opposite (long) arm of chromosome 6 (78, 79). Emerging data suggests that the extensive diversity seen in the ULBP family may be due to the functional or locational specialties of each ligand, as is evidenced with ULBP4 and its predominant expression in skin (80–82). Splice variants exist (ULBP4 (83), and ULBP5 (83, 84); Figure 1C ) providing significant within locus diversity. Furthermore, as reviewed by Carapito & Bahram (19), clear differences in allele frequencies between geographically distinct populations exist for the ULBP family, which suggest that polymorphisms in ULBP may be a consequence of divergent selective pressures. Moreover, the possession of a large ULBP family in humans and other species is thought to provide a selective host advantage in the evasion of viruses and tumors. However, overall, the ULBP family appears to be less polymorphic than the MIC genes, albeit studies of ULBP gene polymorphisms and haplotypes remain limited (50, 85–88).

To date, adoptive cell transfer (ACT) therapies, whereby peripheral blood mononuclear cells are collected, edited, expanded and reinfused back into patients, have successfully been developed to express NKG2D (largely on T cells) and shown considerable anti-tumoral and anti-viral potential in vitro (183–185). However, in vivo responses to NKG2D-focused chimeric antigen receptor (CAR) ACT therapy to date do not appear to be as robust, suggesting that inducing NKG2D alone may not be sufficient (186). To our knowledge, transduction of NKG2D in combination with other receptors within the same cell for improved CAR therapy has only recently been applied clinically, with one (1/9) active registered trial (as of 27 July 2021 on ClinicalTrials.gov) combining transduction of NKG2D and ACE2 for treatment of SARS-CoV-2 (NCT04324996). Most currently active trials (8/9) aim to solely target the NKG2D receptor, and this is partially to reduce the risk of insertional mutagenesis and gene dysregulation (187, 188). CRISPR-Cas9, however, provides a novel avenue for ACT therapy by providing a means to conduct targeted gene insertion in a multiplexed fashion to maximize host immunity (176, 177). Strong evidence suggests that CRISPR-based knockout of receptors responsive to immunosuppressive mediators, such as transforming growth factor beta receptor 2 (TGFβR2), or immune checkpoints (dampeners of cellular activation), such as programmed cell death 1 (PD-1), which play key roles in physiological immune homeostasis, are likely to improve NKG2D-mediated cellular cytotoxicity (189–191). Simultaneous editing of multiple loci, particularly immune checkpoints, in autologous or allogenic cells in this way is also likely to complement immune checkpoint inhibitor (ICI) therapy, such as anti-PD-L1. Moreover, CRISPR-Cas9 expands the potential of ACT therapeutics by facilitating targeted insertion of gene sequences (192), such as KLRK1, in cell subsets that otherwise have no or low expression of the NKG2D receptor. Similarly, knockout of inhibitory receptors within the same effector cell using CRISPR-based therapies is predicted to be an excellent starting point in improving ACT therapy outcomes via CRISPR-based methods. Notably, enhancement of NKG2D-dependent immune responses in this way requires careful consideration and evaluation of the potential for collateral adverse autoimmune reactions. Indeed, given the central role of the NKG2D receptor-ligand axis in autoimmune conditions, such as Crohn’s disease (36), coeliac disease (193), and rheumatoid arthritis (194), care needs to be taken to not generate autoreactive lymphocytes.

Apart from direct gene editing to induce or improve NKG2D-mediated immunity, enhancing NKG2D-L on tumor or virus-infected cells represents both an alternative and complementary strategy to augment ACT therapy, ICIs and overall host immunity. In this approach, CRISPR-dCas9-based transcriptional activation or repression may be used to directly activate NKG2D-L loci and ensure their surface expression. Recently, Sekiba et al. (195), have shown this to be possible in vitro by applying CRISPR-dCas9 to transcriptionally activate MICA in Huh7 and HepG2 human hepatocellular carcinoma cell lines. Although no other NKG2D-L have been targeted in this way to date, it is predicted that multiplexed activation of several types of NKG2D-L is likely to be most effective in promoting NKG2D-mediated immunity and may resolve the low response rate of NKG2D-ACT therapy in vivo. However, our group and others (178, 196, 197) have extensively reviewed that multiplexed CRISPR-based editing, particularly in vivo within the tumor or virally infected cell, is best conducted with an optimized set of targets, rather than a large panel, so as to avoid reduced editing efficiency (retroactivity) and extensive off-target effects. Therefore, an accurate understanding of individual ligands and outcome of their binding to NKG2D is strongly recommended. For instance, given the locational and functional specialties within the ULBP family (80, 81), we suggest that activation of these ligands, either individually or in combination, is needed to elucidate their contribution to the NKG2D pathway in different tumoral/viral contexts. Moreover, it is likely that only a subset of ULBP or MIC members need to be targeted, with some unable to engage NKG2D effectively, as previously reported (81). Notably, given the higher affinity of ULBP1 for the NKG2D ligand, compared to MICA and MICB (59), it may serve as a better therapeutic target. However, further investigation is needed to elucidate the affinities of the remaining ULBP family members, and whether higher affinity to the NKG2D ligand directly translates to improved cytotoxicity. CRISPR-dCas9 can also theoretically be applied to other levels of NKG2D-L regulation ( Table 1 ) for improved expression and immunity. An obvious example of this is in the transcriptional repression of genes involved in NKG2D-L shedding, such as ADAM10/17 or MMP9/14, which are commonly hijacked during viral infection and tumorigenesis, and are known to be targetable using CRISPR-based technologies ( Table 1 ) (143, 165). Targeting the molecules responsible for proteolytic shedding of NKG2D-L in this way is likely to be beneficial both in the clinic and in furthering our basic biological understanding of the mechanisms driving proteolytic shedding. Although CRISPR-dCas9 epigenetic editing is transient and does not induce permanent genetic alterations, significant care needs to be taken to deliver this technology specifically to the target tissue or cell using precise delivery systems, so as to avoid inducing a severe systemic inflammatory state, particularly if used in combination with NKG2D-ACT therapeutics.