Natural selection for killer receptors and their MHC class I ligands: Pursuit of gene pairs that fit well in tandem

SUMMARY

Our understanding of the genetic basis of host resistance to viral infection and disease has progressed significantly over the last century. Numerous genes coding for modifiers of immune functions have been identified which impact a variety of critical cellular processes, including signaling via lymphocyte receptors and their ligands, signal transduction, cytokine signaling, production and release of cytotoxic effectors, transcriptional regulation and proliferation. Genome-wide association studies implicate an important role for both highly polymorphic natural killer (NK) cell receptors and their major histocompatibility complex (MHC) class I ligands in modifying host resistance. These findings indicate NK cells are critical mediators of viral control with considerable potential to affect morbidity and mortality outcomes. They further suggest that both stimulatory and inhibitory NK receptor polymorphisms alter NK cell sensing of MHC I ligands on viral targets, which influences how NK cells respond to infection. In many cases, however, the underlying causes associated with host outcomes remain elusive. Herein, we discuss several modes of NK cell sensing of MHC I and MHC I-like molecules on viral targets, and the role of genetic diversity in this evolutionarily dynamic process. We further suggest that natural selection for paired NK receptors with opposing function, but shared MHC I ligands may give rise to rare, but highly effective MHC I-dependent modes of NK cell sensing of viral targets.

Graphical Abstract

Graphical Summary: NK inhibitory and activating receptors undergo natural selection together with their MHC I ligands to provide optimal sensing of virus infected cells.

INTRODUCTION

Today, complications stemming from infectious disease account for ~25% of deaths globally [1]. Almost a century has passed since the advent of genetic approaches to explore the basis of mammalian susceptibility to viral infection [2–4]. Recent advances include identification of individual genes and the nucleotide variants within them that underpin disparity in host immune defenses and resultant impacts on viral growth and pathogenesis. A well-delineated inference of these studies focusing on the genetics of host resistance to viral infection (human and mouse) suggests that NK receptor polymorphisms are a frequent cause of variability in immune control and disease outcome. Moreover, genetic epistasis is typical of NK receptor and MHC I gene combinations affecting host morbidity and mortality outcomes [5]. These findings hint that critical interactions between NK receptors and their MHC I ligands are essential to modify NK cell control of viral spread and eventual health outcomes in infected hosts.

Genetic variation underlying polymorphic NK cell receptors for MHC I and their cognate MHC I ligands indicates that diversity favors host survival, at least at a population level. Indeed, NK cell deficiency is associated with severe host susceptibility to viral infection and premature death due to infectious causes in most affected persons [6, 7]. Whereas classical NK cell deficiency results from mutations in genes encoding GATA-2, MCM4 and IRF8 that are essential to develop mature NK cells, functional deficiency resulting in impaired NK cell activity is less well understood [6, 8, 9]. Moreover, both human killer-cell immunoglobulin-like receptor (KIR) and mouse Ly49 (killer lymphocyte receptor, subfamily a, Klra) genes are under natural selection and significant pressure to rapidly evolve [10]. While reproductive pressures exert a profound effect on gene selection, health-related impacts associated with increased fitness and host resistance to infection and pathogenesis have undoubtedly influenced these MHC I-binding NK receptors as well [11, 12].

NK cells display many different types of cell surface receptors that specifically bind classical or non-classical MHC I molecules and modify NK functional responses, including KIRs in human, Ly49 receptors in rodents, and NKG2A/C-CD94 receptors in both species that specifically bind classical and non-classical MHC I molecules. Although KIR and Ly49 receptors are structurally unrelated, both receptor types display variegated expression in NK cells, they serve analogous functional roles, and they either stimulate or inhibit NK activity upon engagement of MHC I or related cognate ligands [12, 13].

Inasmuch as the genes for these NK receptors and their MHC I ligands segregate independently in both human and mouse, NK receptor diversification should increase the likelihood that at least one can bind self-MHC I in a given host. NK cell education (licensing), which requires NK inhibitory receptor interaction with cognate MHC I, contributes to self-tolerance and endows NK cells with increased effector activity in response to stimulation through activation receptors [14, 15]. Hence, licensed NK cells are prevented from killing autologous tissues, yet they aggressively attack self-MHC I-deficient targets [16]. However, many viruses deliberately interfere with antigen processing and presentation pathways in infected cells, which results in altered or inadequate cell surface MHC I display and evasion of T cell immunity (reviewed in [17, 18]). MHC I-specific NK receptors therefore are well suited to detect viral pathogens striving to prevent host expression of self-antigens on the surface of infected cells.

Recent studies have shown that polymorphism affecting both MHC I-specific NK receptors and their cognate ligands has a synergistic effect on antiviral NK cell responses and host outcomes during infection. These findings suggest that natural selection for diversity in NK receptors and their ligands underlies the potential for highly discriminatory interactions which might assist NK cell sensing of virus modified self-antigens on infected targets. This review thus focuses primarily on human and mouse studies highlighting the importance of selection for polymorphic NK receptors and MHC I which jointly shape how NK cells sense and respond to viral infection (Figure 1).

Figure 1. A proposed model for NK cell sensing of MHC I on virus infected target cells.

Proposed NK cell sensing modes involving activation (AR) or inhibitory (IR) receptor detection of viral targets with consequent effects on antiviral NK cells and host virus control (VC) are depicted. Major NK receptor signaling pathways implicated in each mode are shown. The extent of VC may vary and, in most cases, has not been compared across sensing modes. VC associated with high expression KIR3DL1^hi, however, exceeds mild HIV protection noted for low expression KIR3DL1^lo. Expression-independent KIR3DL1–47V is associated with HIV elite control (EC) in HLA-B*57 people. Self MHC I-dependent IR enhancement of AR sensing is represented by increased expansion of activated NK cells that are able to mediate highly efficient lysis of infected target cells.

MHC AND NON-MHC GENETIC CONTROL OF HOST RESISTANCE TO VIRAL INFECTION

More than forty years ago, researchers found that outbred and different inbred strains of mice differed in their susceptibility to acute lethal MCMV infection [19]. Grundy (Chalmer) and colleagues predicted that mouse strain-specific susceptibility differences could be useful to explore the genetic basis of host resistance to viral infection [20]. They found that both MHC and non-MHC genetic factors contribute to host protection during experimental MCMV infection [20, 21]. Remarkably, the H-2^k haplotype in C3H, B10.BR or BALB.K mice confers substantial host protection in comparison to congenic strains carrying H-2^b, H-2^d or H-2^g haplotypes [20–23]. Indeed, BALB.K mice withstand ~10-fold higher dose MCMV infection than their H-2 congenic counterparts. Nonetheless, non-MHC effects are clearly at play since BALB.K and B10.BR also differ in the extent of host resistance to lethal infection [21].

Similar to strain-dependent variation in MCMV-induced lethality, spleen MCMV titers vary according to the level of host resistance within 3–4 days post-infection [23–25]. Scalzo and Shellam therefore measured MCMV titers in infected spleen and liver tissues to determine acute virus control phenotypes in different offspring obtained by crossing C57BL/6 (B6; MCMV resistant) to A/J or BALB/c (both MCMV susceptible) mice. They studied several different panels of recombinant inbred mouse strains, as well as large cohorts of intercross or backcross mice that were all typed with known genetic markers, including natural killer gene complex (NKC)-linked genetic markers [25–27]. A single autosomal dominant, NKC-linked (non-MHC) locus Cmv1, later identified as Ly49h [28–30], was found to confer robust host protection in B6 mice during acute MCMV infection.

NK CELL SENSING OF AN MHC I-RELATED VIRAL DECOY

The Ly49H activation receptor expressed in about one half of NK cells in B6 mice, but not in A/J or BALB/c mouse NK cells, binds the MCMV encoded m157 molecule which is structurally related to MHC I [31, 32]. Antigen-specific recognition of m157+ infected target cells causes rapid activation and expansion of Ly49H+ NK cells so that they exert profound control over viral spread during acute MCMV infection [33]. This initial expansion phase is followed closely by Ly49H+ NK cell contraction and eventual formation of NK cell memory which is marked by antigen recall responsiveness, increased effector activity and enhanced viral control in challenge experiments [34, 35].

Specific expansion of Ly49H+ NK cells during acute MCMV infection depends upon the coordinated action of several cytokines including type I interferon (IFN), IL-12, IL-15, IL-18 and IL-33 that regulate activation, proliferation, IFN-γ production, cytotoxicity and accordingly acute MCMV control [36–38]. Type I interferon or IL-12 cytokines elicited in response to MCMV infection boost granzyme B levels and Ly49H+ NK lysis of infected targets which requires both IFN-γ and perforin [39, 40]. Moreover, type I IFN α/β receptor signaling promotes expansion of Ly49H+ NK cells by protecting them from premature apoptotic death during MCMV infection [41]. Together these data establish the importance of various cytokines and effector molecules working together in support of the antiviral role of Ly49H+ NK cells.

In sharp contrast to Ly49H, MCMV m157 binds NK inhibitory receptors (e.g. Ly49I¹²⁹) and dampens NK effector function, which suggests that it evolved primarily as a decoy MHC I to counter NK cell attack in favor of viral spread within the host [31, 42]. Although the single remaining Ly49I^B6 receptor allele in inbred B6 mice lacks this specificity, selective pressure exerted by MCMV in a predecessor of C57-related strains might have given rise to m157-specific Ly49H^B6 via alteration of an inhibitory receptor allele previously targeted for viral manipulation. Such a change would have required minor nucleotide sequence variants resulting in two amino acid adaptations, one in the transmembrane domain to enable interaction with a signaling adaptor (e.g. DAP-12), and another that alters a tyrosine in the immunoreceptor tyrosine-based inhibitory motif [10]. Inbreeding of B6 and 129 (one of Castle’s strains generated by crossing with C57-related mice) strains [43] thus could have resulted in the respective loss (B6 strain) or retention (129 strain) of an m157-specific Ly49I receptor allele, thereby further contributing to strain-specific differences in host resistance to MCMV infection.

In addition to m157-dependent effects, Ly49H+ NK sensing of the viral decoy is regulated by licensing receptors that recognize self-MHC I. mAb 5E6-reactive Ly49 receptors in B6 mice, for example, interfere with m157-dependent Ly49H+ NK cell expansion, and MCMV control upon adoptive transfer into Ly49H-deficient neonates due to sustained inhibitory interactions with self-MHC I [44]. This effect presumably is due to self-MHC I D^b and K^b-licensed Ly49I receptors [45] since 5E6 does not readily bind Ly49C expressed on NK cells in B6 mice [46]. Ly49H-dependent sensing leading to NK cell expansion and virus control in MCMV-infected neonates is likewise thwarted by the NKR-P1B inhibitory NK receptor, which has Clr-b as its self-ligand [47]. Interestingly, the MCMV m12 protein, another viral decoy, binds NKR-P1B as a way to evade NK-mediated immunity [48]. The importance of licensed NK cells in viral control thus has been questioned. It remains unknown whether this supposed negative impact is restricted to NK sensing of stand-alone viral decoys. In any case, the Ly49H-m157 receptor-ligand pairing differs from other modes of NK cell sensing of viral infection that are primarily host MHC I-dependent.

HOST MHC I-DEPENDENT NK CELL SENSING OF VIRAL INFECTION

The H-2^k haplotype imparts essential control of acute MCMV replication, viral spread and lethal infection [26, 49, 50]. As with B6 mice, depletion of NK cells in C57-related MA/My (H-2^k) mice prior to infection fully eliminates MCMV control [26, 43, 49, 50]. A combination of genes within the NKC-Ly49 gene cluster and the MHC is essential to acute viral control in MA/My x BALB/c cross offspring [49], much like genetic epistasis relating defined pairs of KIR and HLA afford increased viral control and protection from disease progression in persons with chronic viral infection (discussed below). The MA/My Ly49P activation receptor which specifically binds MHC I D^k plus MCMV gp34 complexes in vitro, thus, is predicted to deliver antigen-specific NK cell control of MCMV [51]. The Ly49L activation receptor expressed on NK cells in BALB.K mice also binds MHC I D^k plus MCMV gp34 complexes in vitro. Although in vivo evidence of a role for Ly49P is limited, Ly49L+ NK cells expand in vivo during MCMV infection, and adoptive transfer of mature Ly49L+ NK cells protects neonates during infection [52]. These findings establish that antigen-specific activation receptors expressed by NK cells contribute to acute virus control and critical host resistance.

In separate work, genetic analysis of MA/My x C57L (MCMV resistant and susceptible, respectively) cross offspring identified MHC I D^k as a primary genetic factor influencing MCMV control [50, 53]. MHC I D^k-dependent MCMV resistance is equally abolished in these mice by depleting either total NK cells or a select subset expressing Ly49G2, a cognate inhibitory receptor for self-MHC I D^k [53, 54]. In bone marrow chimeric mice expressing MHC I D^k solely in hematopoietic or non-hematopoietic cells, Ly49G2+ NK cells display impaired stimulatory activity, less effective lysis of MHC I-deficient target cells and diminished MCMV control [54]. Both the frequency and numbers of resting Ly49G2+ NK cells in naïve mice are regulated by MHC I D^k [55], and they additionally undergo specific expansion and accumulation in response to MCMV in MHC I D^k mice, but not in mice lacking their cognate self-MHC I ligand [56]. Moreover, Ly49G2^{MA/My or C57L}+ NK cell responsiveness to MCMV is specifically dependent on the MHC I D^k locus [55]. NK cells bearing certain Ly49G2 receptor allotypes become licensed when MHC I D^k is expressed in both hematopoietic and non-hematopoietic cells. Incomplete education due to absent or lineage-restricted self-MHC I exposure during development, however, results in inferior Ly49G2+ NK cells and less effective viral control. The importance of NK cell education in host resistance to infection, however, remains to be established in animals deficient for a single NK licensing receptor. These animals should provide an outstanding model system to explore how NK inhibitory receptor signaling contributes to the role of NK cells in viral immunity, which might further yield insights into the basis for elite virus control observed in HIV-infected people (discussed below).

Self-MHC regulated NK cell expansion in response to MCMV thus is not a common feature since Ly49G2^B6+ NK cells expand irrespective of host H-2 haplotype in B6 and B10.D2 mice during Listeria or MCMV infection [57]. However, H-2-dependent expansion, tissue localization and virus control is characteristic of licensed NK cells responding to MCMV in B6 mice depleted of regulatory T cells, or in chimeric B6 hosts following hematopoietic stem cell transplantation [58, 59]. Licensed NK cells thus mediate critical virus control in different genetic backgrounds presumably due to rapid detection of viral targets and consequent expansion during infection. Unlicensed NK cells in these settings, in contrast, fail to respond similarly to infection, thereby delineating NK cell sensing modes based on detection of a viral decoy versus virus modified host MHC I molecules.

HUMAN KIR AND HLA GENE PAIRS IMPART FITNESS: HOST MHC I-DEPENDENT NK CELL SENSING OF VIRAL INFECTION INFERRED

More than a decade ago, Carrington and coworkers found that specific KIR and HLA genetic pairings or epistatic interactions correspond with delayed progression toward disease in HIV-infected persons [60]. Individuals with both KIR3DS1 and HLA-B Bw4 allotypes experienced delayed disease progression which suggests that KIR3DS1 NK activation receptor recognition of HIV-infected target cells might coincide with better NK cell clearance of viral infection and less disease.

HLA-B Bw4 defense against HIV intriguingly extends to persons carrying specific KIR3DL1 allotypes encoding inhibitory NK receptors [61]. When sorted by NK cell surface expression level, highly expressed KIR3DL1 allotypes best correspond with delayed disease progression and lower viral loads in persons carrying HLA-B*57 with an isoleucine at position 80 (80I) of the heavy chain. On the other hand, low expression KIR3DL1 allotypes are beneficial in HLA-B*27 persons with a threonine residue at position 80 (80T), albeit this pairing is less protective than KIR3DL1^hi in HLA-B*57-80I persons. Although a basis of KIR3DL1 protection is tenuous, varied expression corresponds to inhibitory activity in NK cells [62, 63], which may selectively modify NK cell sensing of distinct HLA-B Bw4 ligands on infected targets. KIR3DL1 discernment of HLA-B*57 presented peptides [64] additionally suggests it may be sensitive to HIV-modified peptide cargo due to changes in the overall conformation of host MHC antigens in infected cells. In effect, less inhibitory KIR3DL1 signaling during infection should promote NK cell lysis of infected targets and HIV control via stimulation through a still unknown NK activation receptor. KIR3DS1 is an interesting candidate in this regard as it specifically binds HLA-B*57 presenting peptides with discrepant P8 residues, as well as two different HIV peptides [65].

Amongst HIV-infected persons, some that are serologically positive for anti-HIV antibodies exhibit slower progression towards clinical immunodeficiency while maintaining a relatively low viral burden. Many of these slow progressors eventually succumb to HIV infection without anti-retroviral treatment. A small fraction of this group referred to as ‘elite controllers’ maintain remarkably low to undetectable (plasma HIV RNA < 50–75 copies / ml) viral loads without therapy and do not progress to AIDS for still unknown reasons. These individuals represent < 1.0% of all HIV-infected persons [66, 67]. Elite control (EC) in these individuals appears to be highly durable and can extend beyond 25 years [66].

HLA-B*57 which confers significant HIV protection is highly represented amongst elite controllers. Carrington’s research team therefore recently sifted HLA-B*57+ controller and non-controller genome sequences for host genetic modifiers underpinning its effect in HIV-infected persons. Remarkably, only a single KIR3DL1 nucleotide variant (position 47 valine; 47V) was significantly associated with EC [68]. This KIR3DL147V effect was not strictly aligned with higher or lower KIR3DL1 expression, but it was HLA-B*57:01-dependent as EC was not evident in HLA-B*57:03 persons with two disparate amino acids at the base of the peptide binding groove [68]. Taken together, these data suggest that mere inhibitory NK receptor expression is not a key determinant of EC. Subtle structural variations in the NK receptor and its MHC I ligand might favor NK cell discernment of peptide cargo changes and resultant NK cell responses during viral infection.

Interactions between KIRs and their MHC I ligands differ from those between the T cell receptor and MHC I, which underlies KIR specificity for HLA alleles (e.g. position 80 in the alpha 1 helix of HLA-C) [12]. Nonetheless, KIRs display peptide-specific interactions with MHC I as in the case of KIR3DL2 which binds HLA-A3 or HLA-A11 and an EBNA3A peptide fragment from Epstein-Barr virus [69]. More recently, MHC I-restricted and peptide-specific recognition has been shown for the KIR2DS2 activation receptor that recognizes Hepatitis C Virus (HCV)-infected cells. Naiyer et al found that an HCV NS3 helicase 1a ‘LNP’ peptide fragment, conserved amongst the vast majority of sequenced HCV isolates, stimulates KIR2DS2 activity when presented by HCV-protective HLA-C*0102 molecules [70]. Though other flaviviruses lack expression of NS3 LNP peptides, 61 of 63 members of the genus flaviviruses instead express NS3 RNA helicases including a highly conserved KIR2DS2-binding peptide domain that fits the HLA-C*0102 peptide-binding groove [70]. Recent work from Hammer and colleagues further shows that human cytomegalovirus (HCMV)-responsive NKG2C+ adaptive NK cells expand and differentiate differently in response to HCMV UL40-derived peptides [71]. These findings together demonstrate that stimulatory NK receptors mediate antigen-specific recognition of viral infection, similar to other pattern recognition receptors (e.g. Toll-like receptors), via surveillance for conserved pathogen associated molecular patterns (PAMPs), which coincides with NK-mediated control of viral replication and host protection.

The importance of specific KIR and HLA genetic pairings in host protection from viral infection is not limited to HIV. Indeed, individuals carrying homozygous KIR2DL3 and HLA-C1 allotypes manifest significant resistance to Hepatitis C Virus (HCV)-induced disease [72]. Despite that an underlying mechanism for KIR2DL3 and HLA-C1 synergy during HCV infection remains elusive, it is possible that KIR2DL3 can detect subtle changes in virus-modified HLA-C1 since it displays some degree of peptide-selectivity in discriminating MHC I ligands [73, 74]. Moreover, KIR2DL3 is sensitive to variation in HCV core-derived peptides presented by HLA-C*03:04 which results in NK cell functional differences [75]. These data suggest that KIR2DL3 recognition of virus-modified HLA-C molecules favors NK-mediated HCV control, perhaps by promoting signaling through an activation receptor due to the loss of negative signaling through a licensing receptor. Notably, KIR2DS2 also binds HLA-C*03:04 although this interaction is not NS3 LNP peptide-specific [70]. An intriguing question is whether a combination of certain KIR2DL3 and KIR2DS2 alleles further augment the host response in HLA-C1 persons during HCV infection.

CONCLUDING REMARKS

NK cell surveillance for virus-induced changes in host cells during infection relies on polymorphic NK receptor sensing of MHC I and related antigens (Figure 1), and transduction of varied, and often opposing, functional signals in distinct, but overlapping NK cell subsets. Genetic analysis of host resistance to viral infection indicates that polymorphism in NK receptors and cognate MHC I significantly affect antiviral NK cell responses and ensuing outcomes in the infected host.

Increased fitness and host protection associated with MHC I-specific NK receptor and self-MHC I gene pairs suggests that specific receptor-ligand interactions synergistically modify NK cells, including target cell sensing, cytotoxic effector functions, activation, cytokine production, proliferation and/or antiviral activity. Inasmuch as both activation and inhibitory receptor genes show linkage to disease protection, opposing receptors working separately, or in tandem, apparently is essential to NK sensing of virus-modified host MHC I determinants and resultant antiviral NK effector functions.

A rare percentage of HIV infected people with antibody reactivity to HIV antigens intriguingly display overt and durable resistance (i.e. elite control) for years in the complete absence of anti-retroviral therapy. Recent work demonstrates that HLA-B*57+ persons carrying the KIR3DL147V allotype are significantly protected from HIV in comparison to KIR3DL147I persons. This finding suggests KIR3DL1–47V+ NK cells are equipped for specific sensing of HIV infected targets, although a basis of KIR3DL1–47V receptor protection remains unclear. Perhaps HIV-specific KIR3DS1 jointly expressed in KIR3DL1–47V+ NK cells promote sensing and clearance of HIV infected targets. Related to this, licensed NK cells displaying certain Ly49G2 allotypes provide essential MCMV control in MHC I D^k mice, which represents a novel and important model system to explore the role of licensed NK cells in viral immunity and possibly elite virus control.

These studies raise a number of important questions requiring further attention. Are the observed genetic associations in human studies related to NK cells directly, their effect on adaptive immunity, or some combination of the two? What’s the role of the inhibitory receptor? Are they sensitive to fine differences in peptide cargo (host v. viral) when carried by specific HLA alleles? Does inhibitory receptor sensing result in more pronounced effector NK cell stimulation? A better understanding of the role of both activation and inhibitory receptor sensing of viral infection should reveal if and how different receptors with divergent functions might partner to deliver optimal antiviral NK activities and increased host resistance during viral infection.