Structure/function of human killer cell immunoglobulin-like receptors: lessons from polymorphisms, evolution, crystal structures and mutations

Authors


Dr K. S. Campbell, Fox Chase Cancer Center, Institute for Cancer Research, 333 Cottman Avenue, Philadelphia, PA 19111, USA. Email: kerry.campbell@fccc.edu
Senior author: Kerry S. Campbell, Ph.D.

Summay

Stimulation or tolerance of natural killer (NK) cells is achieved through a cross-talk of signals derived from cell surface activating and inhibitory receptors. Killer cell immunoglobulin-like receptors (KIR) are a family of highly polymorphic activating and inhibitory receptors that serve as key regulators of human NK cell function. Distinct structural domains in different KIR family members determine function by providing docking sites for ligands or signalling proteins. Here, we review a growing body of literature that has identified important structural elements on KIR that contribute to function through studies of engineered mutants, natural polymorphic sequence variants, crystal structure data and the conservation of protein sequences throughout primate evolution. Extensive natural polymorphism is associated with both human KIR and their ligands, MHC class I (HLA-A, -B and -C) molecules, and numerous studies have demonstrated associations between inheritance of certain combinations of KIR and HLA genes and susceptibility to several diseases, including viral infections, autoimmune disorders and cancers. In addition, certain KIR/HLA combinations can influence pregnancy and the outcome of haematopoietic stem cell transplantation. In view of the significant regulatory influences of KIR on immune function and human health, it is essential to fully understand the impacts of these polymorphic sequence variations on ligand recognition, expression and function of the receptor.

Introduction

Natural killer (NK) cells are critical effectors of the early innate immune response toward transformed and virus-infected cells.1 Whereas T and B cells are triggered upon detecting foreign invaders through the expression of antigen-specific receptors, the activation status of NK cells is regulated by a balance of intracellular signals received from an array of germ-line-encoded activating and inhibitory receptors.2 When an NK cell encounters an abnormal cell (e.g. tumour or virus-infected) and activating signals predominate, the NK cells can rapidly induce apoptosis of the target cell through directed secretion of cytolytic granules containing perforin and granzymes or engagement of death domain-containing receptors.3 Activated NK cells can also secrete cytokines, such as interferon-γ, tumour necrosis factor-α and granulocyte–macrophage colony-stimulating factor (GM-CSF), which activate both innate and adaptive immune cells.4

The human killer cell immunoglobulin-like receptors (KIR; also known as CD158) are a family of transmembrane glycoproteins expressed on NK cells and a subset of T cells.5 The KIR are key regulators of the development, tolerance and activation of NK cells.1 The major ligands for KIR are MHC class I (HLA-A, -B or -C) molecules, which are expressed on the surface of nearly every normal nucleated cell in the body, are encoded by the most polymorphic genes in humans, and define immune ‘self’. Tolerance of NK cells toward normal cells is achieved through their expression of MHC-I-binding inhibitory receptors, which include KIR, NKG2A/CD94 and CD85j (ILT2, LIR1). The NK cells preferentially attack abnormal cells that have down-regulated surface MHC-I molecules, termed ‘missing self recognition’.6 The loss of MHC-I expression is a rare event that occurs in certain tumour and virus-infected cells to avoid recognition by cytolytic T cells, which use T-cell receptor (TCR) recognition of antigenic peptides bound to MHC-I molecules to identify abnormal target cells. In this way NK cells intervene to eliminate MHC-I-deficient target cells. The KIR-mediated regulation of NK cells can significantly impact their responsiveness during viral infection, cancer, haematopoietic stem cell transplantation and pregnancy.7–11 On the other hand, the expression of self MHC-I-reactive KIR is also critical for the maturation of functionally responsive NK cells, through a process referred to as ‘education’, ‘licensing’, or ‘arming’.12–16

The KIR family is encoded by 14 highly polymorphic genes (2DL1 to 2DL5, 3DL1 to 3DL3, 2DS1 to 2DS5, and 3DS1], and distinct family members can transduce either activating or inhibitory signals (Table 1). Nomenclature of KIR is based upon the number of C2-type immunoglobulin-like domains in the extracellular region (2D for two domains, 3D for three domains) and by the length of the cytoplasmic domain (L for long-tailed receptors and S for short ones) (Fig. 1).21 All inhibitory KIR have long cytoplasmic domains possessing immunoreceptor tyrosine-based inhibitory motifs (ITIMs; I/VxYxxL/V), which recruit protein tyrosine phosphatases that are critical for mediating inhibitory function.2 In contrast, KIR with short cytoplasmic domains associate with a transmembrane signalling adaptor protein, DAP12 (also called KARAP). Consistent with antigen receptor signalling, DAP12-dependent activation occurs through the recruitment of Syk/ZAP-70 tyrosine kinases by immunoreceptor tyrosine-based activation motifs [ITAM; Yxx(L/I/V)x6–8Yxx(L/I/V)].2 The only exception to this short/long-tailed rule is KIR2DL4, which is a unique long-tailed activating KIR. Compared with other KIR family members, 2DL4 is only expressed on CD56high NK cells, functions as a more potent activator of cytokine production rather than cytotoxicity, and associates with ITAM-containing FcεRI-γ adaptor instead of DAP12 (Fig. 1).22–24

Table 1. KIR gene products and their HLA ligands Thumbnail image of
Figure 1.

 The major structural differences between different killer cell immunoglobulin-like receptor (KIR) family members. The KIR can be divided into three major subgroups: the inhibitory receptors, the activating receptors, and a unique activating receptor known as KIR2DL4. Nomenclature of KIR is based upon the number of extracellular immunoglobulin-like domains (2D or 3D) and the length of the cytoplasmic domain (L for long and S for short). Receptors with long cytoplasmic domains contain one or two immunoreceptor tyrosine-based inhibitory motif sequences (orange) that provide inhibitory function. The short cytoplasmic KIR and the long-tailed KIR2DL4 are activating receptors, which contain a basic amino acid (+) within the transmembrane domain that interacts with an acidic amino acid (−) within the transmembrane domains of signalling adaptor proteins DAP12 (blue) or FcεRI-γ (green), respectively. These adaptors provide intracellular immunoreceptor tyrosine-based activator motif sequences (purple) that allow for activating function.

Individual KIR recognize distinct subsets of MHC-I allotypes (Table 1) with inhibitory KIR always having higher avidity than activating KIR for MHC-I.25 Activating KIR appear to respond best when they encounter allogeneic MHC-I, as was demonstrated by Chewning et al.,26 which could potentially be a mechanism by which NK cells promote anti-tumour responses after haematopoietic transplantation. The ligands for 2DL5, 3DL3 and 2DS5 are currently unknown. Intriguingly, CpG-oligodeoxynucleotides were recently shown to bind to 3DL2 for subsequent internalization to TLR9-containing endosomes, demonstrating that non-MHC-I molecules may also serve as KIR ligands.27

KIR evolution and conservation

The KIR genes evolved in mammals from a single ancestral gene (KIR3D) that duplicated to form 3DL and 3DX (3DL0) lineages.28–30KIR3DL subsequently evolved rapidly in response to extensive MHC-I diversity and pathogen challenges to establish four separate lineages of KIR genes in modern humans.28–30 The most conserved KIR orthologue in primates is 2DL4, which is present in most human haplotypes and found in Old World, but not New World, monkeys.28,29 In a remarkable example of convergent evolution, a functionally similar, but structurally distinct, family of type II transmembrane C-type lectin receptors, called Ly49 (also known as KLRA), are expanded in mice (approximately 15 genes) and rats (approximately 26 genes).31 Alternatively, only one to three KIR genes are found in these rodent species, and murine KIR genes are not even expressed in NK cells.32 Conversely, the KIR locus has rapidly expanded in primates, and humans carry a single remnant Ly49 gene, which does not encode a functional protein.33 In contrast, intermediate species, such as pigs and seals, have evolved with minimal expansion of either KIR or Ly49 gene loci.34,35

The sequences of human KIR within the extracellular, transmembrane and cytoplasmic domains are remarkably conserved, yet the KIR genes have evolved to produce a highly polymorphic family of receptors. Genetic evidence indicates that the genes expanded through duplication and recombination, which was probably accelerated by their close proximity of head-to-tail orientation within the 19q13.4 chromosomal locus in humans.36 The KIR gene products attain a high level of diversity based upon four levels of variation:9 (i) the product of each KIR gene specifically recognizes a distinct subset of the available MHC-I allotypes (Table 1),5,17,37 (ii) different combinations of the 14 KIR genes are inherited as distinct haplotypes by individuals within the human population, and different haplotypes can vary in proportion of activating and inhibitory KIR,38,39 (iii) many distinct alleles of individual KIR genes have arisen through point mutations encoding minor sequence variations of one to several amino acids, which can affect receptor surface expression level, recognition by anti-KIR antibodies, and affinity/avidity for MHC-I,37,40–42 and (iv) diversity of the NK cell repertoire in peripheral blood is generated through the stochastic expression of different combinations of the available KIR gene products on the surface of individual NK cells.43–45

Here we review our current understanding of the structure/function relationships within human KIR. Our understanding of how the molecular structure of KIR influences function is derived from studies of specifically engineered mutants, natural polymorphic variants, crystal structures, and the conservation of functionally important sequences in KIR throughout mammalian evolution, particularly among primates. The structure/function relationships in KIR will be organized by focusing separately on (i) the extracellular/ligand-binding domain, (ii) the transmembrane domain and (iii) the cytoplasmic domain, which each share responsibility for a distinct aspect of KIR function.

KIR structure/function

Extracellular domain

The KIR extracellular domains are chiefly responsible for ligand recognition and comprise two or three closely related immunoglobulin-like domain structural units (designated D1 and D2 in most KIR2D receptors; D0, D1 and D2 in KIR3D receptors; and D0 and D2 in 2DL4 and 2DL5). Domains D1 and D2 form a V-shaped orientation, as revealed from crystal structures of 2DL1, 2DL2 and 2DL3,46–48 and the angle formed between D1 and D2 varies significantly in different KIR2DL structures (from 66° for 2DL1 to 81° for 2DL2). Moesta et al. showed that the higher avidity of 2DL2, compared with 2DL3, for group 1 HLA-C ligands (which include Cw1, Cw3, Cw7 and Cw8) is the result of variations at two residues that appear to interact with each other near the hinge between D1 and D2 to alter the angle between the two domains.17

KIR binding to MHC-I is dependent upon hydrogen bonding and charge complementarity, whereas TCR/MHC interactions rely more upon hydrogen bonding and van der Waal’s forces.49 The D1 and D2 domains contact the exposed surface of the HLA molecule at a site that straddles the C-terminal end of the bound antigenic peptide.49 As a result of this binding overlap, certain bound peptides can decrease the affinity of inhibitory KIR for MHC-I, and binding is especially influenced by amino acids at positions 7 or 8 of the peptides.50,51 Interestingly, Stewart et al. provided evidence that Epstein–Barr virus infection of cells promotes 2DS1 binding to HLA-C, and although the affinity was lower than that of inhibitory 2DL1 (which recognizes the same MHC-I allotypes), the interaction of both receptors was significantly impacted by alterations at position 7 or 8 in the peptide bound to HLA-C.25

Crystal structures have revealed at least 16 residues in the D1 and D2 domains of KIR2DL that provide primary contacts with HLA-C46,49 (Table 2). These residues in 2DL2 form four salt bridges and five hydrogen bonds with conserved residues in the α1 and α2 domains of group 1 HLA-C, the cognate ligands for 2DL2.46,49 Although all 16 residues are conserved between 2DL2 and 2DL3, which share common ligand specificity, these individual residues differ slightly between other KIR. In this way, the differences in these 16 residues define the individual KIR-binding specificities toward certain conserved residues on distinct subsets of the highly polymorphic MHC-I molecules.49 For example, amino acid position 80 of HLA-C is a critical KIR contact residue, which is either asparagine to define the group 1 HLA-C allotypes (recognized by 2DL2/3) or lysine in the group 2 HLA-C allotypes (targeted by 2DL1).

Table 2.   Protein sequence variations in KIR3DL1 that impact receptor function Thumbnail image of

Although the structure of KIR3D receptors has not yet been solved, previous mutational analysis by Khakoo et al.52 indicated that the D1 + D2 domains of 3DL1 mediate direct binding to MHC-I, while the D0 domain enhances ligand binding. Recent work by Sharma et al., however, suggests that D0 may also participate in direct contact with MHC-I.37 Of the three immunoglobulin-like domains, D0 is the most polymorphic in 3DL1.40 Five residues in the D0 domain (positions 5, 20, 31, 32 and 51 in 3DL1) are of particular importance, because of their positive selection through hominid evolution, as well as residues 49–52, which are invariant between all human 3DL1 alleles (see Table 2).40,52 The D0 residues can also influence 3DL1 function by augmenting cell surface levels, such as natural polymorphisms at position 18 (V18L) and position 86 (L86S), which prevent surface expression of proteins encoded by 3DL1*053 and 3DL1*004 alleles, respectively.41,42

The 3DL1 gene is the most polymorphic member of the KIR family, with at least 59 alleles reported (EMBL-EBI Immuno Polymophism Database53), and studies of extracellular 3DL1 sequence variants provide significant insight into understanding structure/function relationships, as detailed in Table 2. The high polymorphism of 3DL1 is not surprising because HLA-B is the most polymorphic group of MHC-I in humans. Considered distinct alleles within the same genetic locus, the 3DL1 and 3DS1 genes (called 3DL1/S1) encompass three lineages that differ within their extracellular domains: 3DL1*015-like, 3DL1*005-like, and 3DS1*013-like.40 Most residues that distinguish 3DL1 from 3DS1 are within or near the 16 residues that directly contact with MHC-I.40 Some alleles of 3DL1 produce a protein incapable of reaching the cell surface (*004 and *053, as previously mentioned), whereas other 3DL1 gene products are expressed at high (*001, *002, *008, *015, *020) or low (*005, *006, *007) levels on the cell surface.41,42,54,55 Extensive point mutation analysis of 3DL1 by Sharma et al. identified numerous extracellular substitutions that disrupt surface expression and binding epitopes for specific monoclonal antibodies.37 Norman et al. reported that 21 positions in the 3DL1 extracellular domain demonstrate positive selection for amino acid substitution in ape and human evolution.40 Remarkably, nearly all of these positions lying within the D1 and D2 domains either encompass or surround positions corresponding to the 16 amino acids making primary contact with MHC-I (Table 2).40,49 Many polymorphic alleles and point mutants of 3DL1 that differ within these positively selected putative ligand interaction residues significantly influence KIR expression, ligand avidity and inhibitory function (see Table 2).

Transmembrane sequence elements influencing KIR functions

In contrast to inhibitory KIR, the transmembrane domains of all activating KIR contain a basic residue critical for association with DAP12 or FcεRI-γ. Specifically, a lysine in the transmembrane domains of 2DS and 3DS receptors provides a docking site for DAP12,56 and 2DL4 contains an arginine that associates with FcεRI-γ.23,57 Evidence indicates that monomers of the activating 2DS/3DS receptors assemble with disulphide-linked dimers of DAP12 upon co-translation in the endoplasmic reticulum.58 As previously mentioned, both DAP12 and FcεRI-γ unite KIR with cytoplasmic ITAMs that mediate activation signalling (Fig. 1).

As shown in Fig. 2, the basic transmembrane residues and their positions differ between the subgroups of activating KIR and ultimately dictate the accessory protein specificity. Consistent with other immune receptors (FcαRI, NKp46, ILT1, GPVI and OSCAR), an arginine near the N-terminal end of the 2DL4 transmembrane domain (position 4) is situated to interact with a similarly localized aspartic acid in FcεRI-γ. In contrast, the centrally located lysine in 2DS/3DS (position 9) and other receptors (NKG2C/CD94, NKp44, TREM-1, TREM-2, PILRβ, SIRPβ1, and IREM-2) is crucial for interaction with a centrally located aspartic acid in DAP12. Mutations of the arginine in 2DL4 or the lysine in 2DS2 abrogate associations with FcεRI-γ and DAP12, respectively.57–59 Similarly, mutation of the aspartic acid in DAP12 disrupts association with 2DS2.58 As basic residues in receptor transmembrane domains generally require assembly with acidic residues in transmembrane adaptors, it is surprising that 2DS receptors and 2DL4 can reach the cell surface (although less efficiently) in the absence of DAP12 or FcεRI-γ, respectively.23,60,61 Interestingly, the stoichiometry of FcεRI-γ association with 2DL4 appears to be low, presumably because of low expression levels of FcεRI-γ in NK cells, and forced over-expression of FcεRI-γ increases cell surface expression of 2DL4.23 Nevertheless, the basic residues in KIR and the corresponding acidic residue in accessory proteins complement each other and promote membrane stability and surface expression.

Figure 2.

 Alignment of human killer cell immunoglobulin-like receptor (KIR) transmembrane domains. Consensus transmembrane domain sequences are shown for each human KIR, based upon the most common amino acid encoded in all alleles for each gene. Unique amino acids within the activating KIR (2DL4, 2DS, 3DS) are highlighted in bold. Basic amino acids within the transmembrane domains are underlined and in bold type. The dash denotes a gap in the sequence that was introduced for alignment purposes. Although some minor polymorphic substitutions occur within the transmembrane domains of some KIR alleles, they are usually conservative substitutions between similar hydrophobic amino acids. The KIR2DS2*003 sequence, however, contains significant alterations that eliminate the unique lysine and threonine residues found in the activating KIR (bold and italic). Residue positions are numbered at the top. Sequences were derived from the EMBL-EBI Immuno Polymorphism Database (http://www.ebi.ac.uk/ipd/).53

The hydrophobic amino acids found within the transmembrane domains of human inhibitory KIR are relatively conserved (Fig. 2). In contrast, a polar threonine is uniquely located and conserved within the transmembrane domains of the activating KIR (position 13), and could potentially contribute to protein interactions within the plasma membrane. The transmembrane threonine and lysine are altered to hydrophobic amino acids in 2DS2*003, and this unique natural variant exhibits significantly reduced DAP12 association and lack of activating function.62 Interestingly, proline residues are also found within the transmembrane domains of activating KIR (position 11 in 2DS/3DS receptors and position 16 in 2DL4), as well as 3DL3 (position 11). A proline would disrupt the typical α-helical structure of a transmembrane domain, thereby creating a kink. Feng et al. demonstrated that mutation of the transmembrane proline in 2DS2 greatly diminished association with DAP12.58 Nonetheless, the same group found that replacement of all transmembrane residues other than the lysine in 2DS2 with either leucine or valine did not alter association with DAP12, whereas mutation of the lysine to arginine significantly diminished association.59 In conclusion, the centrally located lysine is necessary and sufficient for association of 2DS/3DS receptors with DAP12, although other amino acids may contribute to the integrity of the transmembrane domain.

Some studies have identified conservative transmembrane polymorphic variations that significantly influence KIR function. While most 3DL1 alleles encode an isoleucine at the first position in the transmembrane domain (Fig. 2), some alleles encode valine, including 3DL1*007, which has low inhibitory function. Surprisingly, Carr et al. found that substitution of this valine to isoleucine potentiated inhibitory function of 3DL1*007 toward HLA-Bw4 ligands, indicating that even a minor change between two similar hydrophobic residues can significantly impact inhibitory function.63 Similarly, a leucine is uncharacteristically found at position 10 in the transmembrane domains of 2DS3 and 2DS5, while other KIR contain isoleucine at that position (Fig. 2). VandenBussche et al. found that surface expression of 2DS3 was significantly increased when the leucine is changed to isoleucine.61 Nonetheless, cell surface expression levels of 2DS5 are modest despite the isoleucine at position 10, and surface expression of 2DS5 is instead limited in some alleles by variations in extracellular residues.60,64 Hence, some variations in transmembrane hydrophobic residues can influence function and expression, but their impacts vary between KIR.

Cytoplasmic sequence elements influencing KIR functions

Immunoreceptor tyrosine-based inhibitory motifs in KIR cytoplasmic domains

Perhaps the most widely studied sequence elements in inhibitory KIR are the ITIMs (I/VxYxxL/V). Following engagement with MHC-I ligands, KIR ITIM tyrosines are phosphorylated by Src family kinases to create specialized docking sites for SH2 domain-containing protein tyrosine phosphatase (SHP-1) and SHP-2.2 Recruitment of SHP-1 and SHP-2 is widely believed to inhibit NK-cell activation toward target cells by dephosphorylating critical tyrosine phosphoproteins that propagate activation signalling. SHP-1 and SHP-2 have distinct requirements for recruitment to KIR, however. Mutation of the C--terminal ITIM tyrosine of 3DL1 results in a strong inhibitory function that is mediated through exclusive recruitment of SHP-2.65,66 Surprisingly, inhibition of cytotoxicity by this mutant KIR was unaffected by significant reductions in SHP-2 levels, presumably because the mutant receptor could still recruit some of the phosphatase under these limiting conditions.67 Taken together, these observations indicate that phosphorylation of both ITIMs is required to recruit SHP-1, and both SHP-1 and SHP-2 work in concert to mediate KIR inhibitory function.

As previously mentioned, activating KIR possess truncated cytoplasmic domains, while inhibitory KIR have long cytoplasmic domains. KIR2DL4 is a notable exception of an activating KIR containing a long cytoplasmic domain. As shown in Fig. 3, the length of the cytoplasmic domains of long-tailed KIR is somewhat variable, although most contain two ITIMs, which are necessary to mediate inhibitory function. Importantly, the truncations of activating KIR disrupt the first cytoplasmic ITIM, which eliminates inhibitory function. It is also important to note that a common frameshift found in about half of the alleles of 2DL4 encodes a truncated cytoplasmic domain that prevents cell surface expression and function.22,69,70

Figure 3.

 Alignment of human killer cell immunoglobulin-like receptor (KIR) cytoplasmic domains. Consensus cytoplasmic domain sequences are shown for each human KIR, based upon the most common amino acid encoded in all alleles for each gene. Sequences were derived from the EMBL-EBI Immuno Polymorphism Database (http://www.ebi.ac.uk/ipd/).53 Serine/threonine phosphorylation sites (colour-coded and in bold): CK1, casein kinase 1; CK2, casein kinase 2; PKC, protein kinase C; PDK, proline-directed kinase.68 Immunoreceptor tyrosine-based inhibitory motif and immunoreceptor tyrosine-based switch motif sequences are underlined and in bold.

The cytoplasmic domains of four long domain KIR (2DL4, 2DL5, 3DL2, and 3DL3) contain only one classical ITIM. The C-terminal tyrosine in both 2DL5 and 3DL2 is instead embedded in a sequence characteristic of an immunoreceptor tyrosine-based switch motif (ITSM; T/SxYxxV/I/L; Fig. 3), which is a sequence element known to recruit SHP-2 and the adaptor SAP.71,72 2DL5 can recruit both SHP-1 and SHP-2, but not SAP,73,74 although inhibitory function of 2DL5 was preferentially blocked by dominant negative SHP-2 protein, but not dominant negative SHP-1.74 The inhibitory capacity of the ITSM-containing the 2DL5 cytoplasmic domain was also shown to be less potent than that of 3DL1.74 These data indicate that the ITSM forces 2DL5 to rely more on SHP-2 recruitment (rather than SHP-1) to mediate weaker inhibitory function. In accordance with these results, recent work by Norman et al. demonstrated that the ITSM-containing cytoplasmic domain of 3DL2 provides weaker inhibition than that of 3DL1.39 2DL4 and 3DL3 contain only the amino-terminal ITIM (Fig. 3). A chimeric receptor containing the extracellular and transmembrane domains of 3DL1 and the cytoplasmic domain of 2DL4 demonstrated a potent inhibitory capacity mediated through selective recruitment of SHP-2.75 Interestingly, however, full-length 2DL4 is an activating receptor that does not appear to be negatively influenced by the ITIM,76 leaving the functional significance of SHP-2 recruitment to 2DL4 unclear. At least some of the activation signalling through 2DL4 is independent of association with FcεRI-γ and appears to be mediated through the extended cytoplasmic domain.57 This unique signalling may involve activation of DNA-PK by internalized 2DL4 within early endosomes.77 As 3DL3 contains only the amino-terminal ITIM, it is expected to function as an inhibitory receptor through selective recruitment of SHP-2, although this has not been formally tested.

Serine and theonine phosphorylation sites in KIR cytoplasmic domains

Although it was initially assumed that tyrosine phosphorylation of the two cytoplasmic ITIMs was the only functionally relevant kinase modification event on KIR, several cytoplasmic serine and threonine phosphorylation sites have also been identified.68 In fact, basal level serine phosphorylation was significantly higher than basal tyrosine phosphorylation in an NK cell line and serine phosphorylation was significantly induced in primary NK cells by protein kinase C stimulation with phorbol ester.68 The major constitutive phosphorylation sites in 3DL1 are two serines immediately upstream of the first ITIM (S364 and S367), as well as distinct serine and threonine sites positioned between the two ITIMs (S394 and T399) (Fig. 3).68 Further in vitro and in vivo analyses established that these individual sites could be phosphorylated by casein kinase 1 (CK1; S364), CK2 (S367), protein kinase C (PKC; S394), and an unidentified kinase presumed to be a proline-directed kinase (PDK; T399), based upon positioning of the threonine adjacent to a downstream proline.68 Well-known PDKs include mitogen-activated protein kinases and calmodulin-dependent kinase, although direct evidence of T399 phosphorylation by any of these kinases is lacking. In vitro studies showed that S394 could be phosphorylated by several PKC isoenzymes.68 To clarify subsequent discussion, these phosphorylation sites will be referred to by the kinase for which they serve as substrates (CK1, CK2, PKC and PDK).

Remarkably, three of these four serine and threonine phosphorylation sites exhibit extensive conservation within KIR sequences from humans and other mammals (Figs 3 and 4). When comparing the human KIR repertoire, the CK2 site is present in all KIR, except 3DS1. Similarly, the PKC site is found in all human KIR with long cytoplasmic domains, including 2DL4, but is absent in short-domain KIR, because of the upstream truncation. In contrast, the PDK site is conserved in all long-domain KIR, except 2DL4. On the other hand, the CK1 site is variably present among KIR. Notably, the CK2, PKC and PDK sites are also broadly conserved in most primate KIR, whereas their presence varies in KIR from other mammals (Fig. 4). Interestingly, the PDK site in some KIR from cow and rat is substituted with glutamic acid, which is known to exhibit ‘phosphomimetic’ properties, thereby suggesting a functional need for such a charged modification at that position.

Figure 4.

 Alignment of mammalian killer cell immunoglobulin-like receptor (KIR) cytoplasmic domains. Consensus cytoplasmic domain sequences are shown for each KIR, based upon the most common amino acid encoded in all alleles for each gene. Serine/threonine phosphorylation sites (colour-coded and in bold):68 CK1, casein kinase 1; CK2, casein kinase 2; PKC, protein kinase C; PDK, proline-directed kinase. Immunoreceptor tyrosine-based inhibitory motif, immunoreceptor tyrosine-based switch motif, and related tyrosine-containing sequences are underlined and in bold. The dash denotes a gap inserted into the sequence for alignment purposes. Substitutions in the PDK site that could act as phosphomimetic residues (glutamic acid, E) are underlined. Species designations and sequence sources: human (Homo sapiens) (EMBL-EBI Immuno Polymorphism Database53 and http://www.uniprot.org); common chimpanzee (Pan troglodytes)78; pigmy chimpanzee (Pan paniscus)79; gorilla, Western gorilla (Gorilla gorilla)80; orangutan, Bornean orangutan (Pongo pygmaeus)81; rhesus macaque monkey (Mucaca mulatta)82; African sabaeus monkey (Chlorocebus sabaeus)83; horse (Equus caballus); cow (Bos taurus)84; mouse (Mus musculus)85,86; rat (Rattus norvegicus)86.

Despite the many sequence polymorphisms identified over a broad spectrum of positions in human KIR, only a single polymorphism has been reported within the CK2, PKC and PDK positions. This polymorphism is a substitution of the PKC serine phosphorylation site to cysteine in 3DL1*060, which would eliminate the phosphorylation site.39Hence, it is readily apparent that the CK2, PKC and PDK phosphorylation sites on KIR cytoplasmic domains are extremely conserved, especially within primates, strongly implying that they have evolved to play important roles in KIR function.

Current evidence in NK cell lines indicates that at least some serine and threonine phosphorylation events can modulate KIR surface expression levels by influencing internalization and turnover. First, mutation of 3DL1 serines at either the CK1 and CK2 sites together or the PKC site alone to alanine increased KIR surface expression by about 50%. On the other hand, mutation of the PKC site to phosphomimetic aspartic acid did not change surface expression levels. Both alanine mutants (CK1 and CK2 together or PKC alone) exhibited substantially better inhibition toward HLA-Bw4 ligand-bearing target cells, which was attributed to the higher receptor expression levels. On the other hand, phosphomimetic substitution at the PKC site on 3DL1 resulted in a slightly reduced inhibitory capacity, despite equivalent surface expression levels when compared with wild-type 3DL1. These data indicate that serine phosphorylation has the potential to ultimately influence KIR inhibitory function.

Mutation of the PKC site on 3DL1 to alanine increased surface receptor turnover, whereas phosphomimetic substitution at the same site decreased turnover.68 Furthermore, the phosphomimetic PKC site mutant was internalized at a slower rate than either wild-type 3DL1 or an alanine mutant. Taken together, these data indicate that PKC phosphorylation of KIR at a highly conserved site between the two ITIMs (S394 in 3DL1) acts to stabilize the receptor on the cell surface. The functional impact of the serine to cysteine polymorphism in the PKC site of 3LD1*060 has not been tested,39 but is predicted to be analogous to the alanine mutant.68 Interestingly, an earlier report demonstrated that TCR engagement can stabilize KIR expression on CD8 T cells,87 suggesting that TCR-mediated PKC activation may induce KIR phosphorylation to elicit this effect. Moreover, Chwae et al. found that surface levels of a chimeric receptor containing the KIR cytoplasmic domain were increased upon stimulation of PKC through a mechanism involving surface recycling.88 This PKC-mediated up-regulation was at least partially abrogated by mutation of the PKC phosphorylation site (S394 equivalent).89

In summary, mechanisms that maintain surface stability of inhibitory KIR are essential to maintain tolerance of NK cells and the KIR+ subset of CD8 T cells, and a growing body of evidence indicates that the cytoplasmic PKC site plays a role in regulating surface expression. These observations in inhibitory KIR are consistent with serine and threonine phosphorylation events known to regulate the surface expression, internalization and intracellular trafficking of other receptors.90,91 Further studies are needed to elucidate the functional impacts of KIR phosphorylation at the CK1, CK2 and PDK sites, as well as other cytoplasmic serine and threonine residues, which could also potentially be phosphorylated.89 Pharmacological manipulation of these phosphorylation events could ultimately influence KIR expression levels, thereby altering NK cell activity.

Concluding remarks

In conclusion, the KIR are a highly polymorphic family of human receptors, and many naturally occurring sequence variations can significantly impact their expression, ligand affinity and function. The KIR are key regulators of immune tolerance, especially on NK cells, and the level of KIR surface expression can significantly influence the balance of signals regulating NK cell maturation and activation. A growing literature indicates that distinct allelic combinations of KIR and HLA class I genes can contribute to disease susceptibility, reproductive fitness, and the outcome of haematopoietic stem cell transplantation.7–11 Although we have a basic understanding of how major KIR structural elements contribute to function, our present knowledge of the wide array of polymorphic influences is limited. Further factors contributing to our ignorance are the multifaceted influences of the diverse polymorphism among HLA class I ligands and the unidentified ligands for several KIR. These uncharacterized factors are currently limiting our ability to fully establish many clinically relevant contributions of complex KIR/HLA class I interactions on human health. Therefore, we must achieve a more comprehensive understanding of the influences of polymorphic variations in KIR sequences on surface expression, signalling capacity and HLA recognition. In addition to the complex influences of polymorphism, we are only beginning to understand how post-translational protein folding, glycosylation and phosphorylation can impact KIR expression and function. Better understanding of these post-translational mechanisms may provide opportunities to therapeutically modulate surface expression of KIR and thereby regulate NK cell function. It is becoming increasingly clear that the complex interactions between polymorphic KIR and HLA ligands are contributing more to our immune responses to infections, cancer and transplantation than is currently understood.

Acknowledgements

We thank Drs Paul Norman, Dan McVicar, Geraldine O’Connor, Christoph Seeger, Alexander MacFarlane IV and Margaret Joyce for critical reading of this manuscript. The authors apologize for our inability to reference all relevant publications because of imposed limitations. This work was supported by R01 grants CA083859 and CA100226 (K.S.C), training grant CA009035-32 (A.K.P) and partially by Centers of Research Excellence grant CA06927 (FCCC) from the National Institutes of Health. The research was also supported in part by an appropriation from the Commonwealth of Pennsylvania. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

Disclosures

The authors have no financial disclosures.

Ancillary