Hla
There are three HLA class I molecules that form the major ligands for KIR (HLA-A, -B, and -C). They are highly polymorphic, with over 20,000 alleles known in total 58 , 59 . The nomenclature used for HLA alleles with respect to KIR ligands is given in Figure 2 . The common HLA-A, -B, and -C allotypes are distinguished from each other by multiple amino acid substitutions. Like KIR, HLA class I polymorphism both within and outside the direct binding site, and genomic variants affecting the steady state expression level are all key modulators of NK cell sensitivity 31 , 60 - 62 . The role of HLA class I is to sample peptide fragments from intracellular protein production and present them at the cell surface for surveillance by NK or T cells. In this role, HLA class I polymorphism diversifies the specific peptide repertoires 63 . Because the binding footprint of KIR overlaps with that of the TCR, this means that KIR binding can also be dependent on variations of the presented peptide sequence 64 . This peptide specificity may increase the sensitivity for inhibitory KIR to detect infection 65 , but is likely most critical for activating KIR, which may then recognize infected cells that retain HLA class I expression but carry foreign peptides 66 , 67 . Differential peptide specificity also creates an opportunity for pathogen exploitation whereby a given pathogen peptide may bind more strongly to inhibitory KIR, protecting infected cells from being killed by cytotoxic cells 68 - 70 .
Less-polymorphic HLA class I molecules -F and -G can also form ligands for KIR3DS1 and 2DL4, respectively 71 , 72 . Although little is known of the implications for combinatorial diversity of these interactions, the receptors are also less polymorphic than other KIR. KIR2DL4 has two main phenotypes, defined by a single nucleotide deletion that prevents cell-surface expression of approximately 50% alleles 73 . Of the multiple KIR3DS1 allotypes ( Fig. 2 ), only two (3DS1*013 and *014) have been observed at appreciable frequency in any population 74 . Of interest, the amino acid substitution that distinguishes 3DS1*013 from *014 occurs in the ligand-binding region and may influence its specificity for HLA/peptide complex 75 . One further mechanism of diversity is the leader peptide from some HLA-B allotypes, in addition to -A and -C allotypes, can be presented by HLA-E. Through interaction of HLA-E with the NKG2/CD94 family of receptors, NK cells are then able to monitor polymorphic HLA class I expression through a more conserved complement to the KIR/HLA system 76 , 77 .
Kir
Methods to genotype HLA to allele level are well established and adopted by clinical laboratories worldwide, providing ample material for the multitude of candidate gene association studies 34 . Because structural complexity of the locus ( Fig. 2 ) has hindered similar attempts, gene presence/absence variation has been the method of choice 78 for most studies investigating KIR diversity in human health. Because gene-content variation is the major component of KIR mediated NK cell functional diversity, these studies have been highly informative and generally reproducible but can be conflicting, particularly across ancestry-distinct cohorts. A critical need then remains to refine this knowledge through considering gene dose numbers and allele diversity of the genes in question. As examples, an allele that is not expressed likely has the same phenotype as an absent gene, and the distinct expression phenotype, ligand affinity or even specificity that characterizes a given allele is not visible to gene-content only methods. For these reasons, methods have been developed to measure the KIR gene copy number 79 , allele sequences 80 , or both at once 81 , 82 . Methods are also described to analyze both KIR and HLA allele diversity in a combined workflow targeted to high-throughput studies 83 - 85 . For much larger studies that are currently impractical to approach through DNA sequencing, imputation from whole-genome SNP data provides an alternative to analyze KIR gene content 86 , or allele diversity 87 . As for HLA , the imputation methods are less accurate than direct sequencing, especially for some ethnicities or ancestry groups including Africans and South Asians 87 , 88 .
Rapid
Structural divergence of the respective genomic regions among closely related species identifies KIR and HLA to be evolving comparatively faster than the remainder of the genome 114 . The mode of genomic expansion to contain multiple KIR genes is unique to primates, and among primates very few direct orthologues of a given KIR gene are present. For example, all but one of the KIR specific for HLA-A, -B or -C are unique to humans 115 . The impressive expansion of the KIR locus is captured in macaques, where almost 60 distinct KIR genes are known 32 , with corresponding duplication of genes encoding their ligands 116 . In the context of these expansions, gene duplication leads to sequence homology that facilitates further reshuffling of gene segments to create receptors and ligands of novel functions 117 . This structural diversification of the KIR locus is likely ongoing in humans and frequent enough to be detected in population cohorts of relatively modest size 118 - 121 . In terms of emergence, the ancestors of highly polymorphic HLA class I are ancient, tracking with jawed vertebrates, whereas KIR may be restricted to mammals, and not functional in some of them 122 - 124 . That different and sometimes overlapping gene families have expanded similarly in other mammals 125 indicates the critical need to retain the interaction, whilst presenting a moving target to any pathogen able to evolve evasion mechanisms. The intricacy of this co-evolution is likely best characterized by HLA-C, which evolved from a duplication of an HLA-B equivalent in an ancestor of hominids (humans, gorillas, chimps, orangutans) 31 . Accordingly, genomic expansion and diversification of the KIR that bind HLA-C and its orthologues is unique to hominids 126 . Although HLA-C can elicit some T cell responses, it has become specialized for interaction with KIR-expressing cells 127 . It is unknown if the specialization occurred during or since the emergence of this receptor/ligand pair. However, in modern humans, KIR interaction with HLA-C is the only KIR/HLA interaction present in every individual, with rare exceptions highly prone to virus infections 128 . Low frequency of KIR interactions with HLA-A or -B in Amerindians 129 suggest KIR interactions with HLA-C may now have a greater impact on human survival than those with HLA-A and -B.
One unique feature of human KIR locus expansion has been development of two functionally distinct families of KIR haplotypes 35 ( Fig. 2 ). The KIR-A haplotype encodes every inhibitory KIR specific for polymorphic HLA class I, KIR2DL3 (C1), KIR2DL1 (C2), KIR3DL1 (Bw4) and KIR3DL2 (A3/11). The KIR-A haplotype encodes only one activating receptor, KIR2DS4, and this is often disabled by a 22bp deletion common to multiple alleles 119 , 130 . The KIR-A haplotype therefore conveys maximal NK cell educating potential. By contrast, KIR-B haplotypes encode fewer inhibitory receptors and a greater number of activating receptors 131 , in this case the inhibitory receptors often having reduced function 53 . Their functional distinctions manifest in the form of disease association, where KIR-A haplotypes in concert with their ligands reduce impact of virus infection and cancer 132 - 134 , but predispose to preeclampsia 135 , whereas KIR-B haplotypes associate with better fetal nourishment 136 . Consequently, although there is an incredible diversity of haplotype structures within and across human populations 79 , 137 , 138 , KIR-A and -B haplotypes are represented at high frequencies in every human population 139 . That KIR-A and -B have both been carried through multiple population bottlenecks that otherwise restrict genome-wide diversity indicates they are maintained in humans through natural selection 140 , 141 .
Kir/Hla
Based on the principle that inhibitory KIR educate NK cells to expect specific HLA class I ligands, carefully directed mismatching between donor and recipient can improve the success rate of transplantation therapy for certain leukemias. Particularly for acute myelogenous leukemia (AML), donor-derived NK cells can protect from relapse when they have been educated in the donor towards HLA class I ligands absent from the patient, likely through killing leukemic cells 182 , 183 . A further advantage is that, unlike T cells, NK cells do not promote graft-vs-host disease. Multiple groups have adopted and enhanced KIR-ligand mismatching protocols 184 - 187 (a recent extensive review is given elsewhere 188 ), which have helped pave the way towards targeted NK cell therapies for the same leukemias and other cancers 189 - 191 . Nevertheless, effect differences remain across transplant centers or treatment regimens 192 , and further refinement of matching protocols will likely be aided through refinement of the genotyping methods 193 , 194 . The above findings, together with the established links between peripheral NK cell quantity and disease course, also imply that NK cells can be harnessed to treat autoimmune and other chronic diseases, as they have for malignancies 195 - 198 . That presence of KIR3DS1 is significantly associated with resistance to PD-1 blockade 199 for example, and KIR3DL1 interaction with Bw4 + HLA affects monoclonal antibody therapy 200 , 201 demonstrates that knowledge of the receptor and ligand genotype may also inform these therapy decisions 202 . Moreover, and resulting from the rapid and population-specific evolution, all the described allotype combinations are highly variable across populations. For example, KIR3DS1 is rare in Africa and Asia, yet highly prevalent in Oceania, indicating differential disease associations and therapy responses related to genomic ancestry are to be expected. Medical genetic and association studies have largely neglected non-European populations 203 - 205 , an exclusion that continues despite the knowledge that inclusion of multiple ancestries increases power of such studies 206 . For all these reasons it is imperative that we can accurately characterize combinatorial diversity of KIR and HLA at high resolution and scale 83 , 85 , 87 , whilst encompassing all human diversity. With innovative new methods of evaluation such as KIR and HLA interaction scores 29 , 148 , 174 that evaluate the strength of binding and signal transduction as a proxy to infer NK cell function, it may be possible to shape the treatment or even prevent certain immune-mediated diseases based on knowledge of individual KIR and HLA compound genotypes.
In summary, combinatorial diversity of KIR and HLA class I is driven by diversifying selection through confrontation with pathogens. The exceptional variation, both within and of interactions between them, being reinforced by population demography, including admixture events that may also be adaptive. Coexistence of functionality to support healthy pregnancies and prevent the unnecessary destruction of tissues has resulted in evolutionary patterns constrained by tradeoffs between these functions and controlling infection. For human health, this means that medical interventions and disease severity can differ across individuals and ancestries; a quandary that can be mitigated by characterizing the functional diversity of KIR and HLA class I and understanding how this diversity impacts disease and the efficacy of medical interventions.
Evolution
The evolutionary mechanism that maintains KIR-A and -B in humans is often termed balancing selection. Balancing selection embodies frequency fluctuations of genetic variants accompanying the relative selective advantage of their respective phenotypes. In this respect KIR-A haplotypes have arisen to very high frequency in East Asian and Amerindian populations through positive natural selection 41 , 129 , 142 , 143 , presumably in response to specific pathogens. However Hiby et al. first noticed that the preeclampsia pregnancy syndrome is most prevalent in women who are homozygous for KIR-A and who also carry a C2-ligand fetus 135 . This disadvantage to reproductive fitness, replicated in multiple populations 136 , 144 - 146 , has produced an inverse correlation of KIR-A and C2 + HLA-C frequencies across these populations 135 . The assumption is that KIR2DL1, which is carried by KIR-A haplotypes and interacts strongly with C2 + HLA-C 147 , is driving the phenomenon. Indeed, the observations become more pronounced when known distinctions of ligand binding and signal transduction strength across KIR2DL1 allotypes are considered 148 , 149 . Inverse correlation of allele frequencies is observed also for genetic variants impacting the relative cell surface expression levels of KIR2DL1 and C2 + HLA-C 150 . Finally, the haplotypes that offer protection from preeclampsia vary across populations, but all of them carry activating KIR that can bind C2 + HLA-C 144 , 146 , 151 , 152 . Indicating a tenuous balance, too many activating KIR may predispose to further pregnancy complications such as acute atherosis 153 . These findings identify a unifying concept across infection and reproduction, whereby specific receptors or ligands that increase in frequency due to a selective advantage may then become a disadvantage due to the high frequency of the respective pairing. This form of co-evolution through balancing selection manifests in multiple guises across the distinct pairs of KIR and HLA class I ligands 66 , 140 , 154 . In populations with high pathogen exposure, natural selection acting for and against specific KIR and HLA class I allotype pairs likely remains ongoing 129 , 155 - 157 .
Adaptation
As the C ligand evolved with the ancestors of modern hominids, it split into C1 and C2 forms 31 . In humans C1 and C2 are defined by a single amino acid substitution at residue 80, where C1 have asparagine and C2 have lysine. A subset of HLA-B molecules (B*46 and B*73) also carry the C1 motif, making them good at interacting with NK cells as well as presenting unique peptide repertoires to T cells 49 , 163 . The motif was obtained through genomic recombination, which may have occurred in ancient humans prior to admixture with modern humans 41 , 115 , 160 . Subsequently, the amino acid residues that interact with KIR drive balancing selection of HLA-C that is stronger than that observed for HLA-A or -B 154 , 164 . The C2-specific KIR emerged on multiple occasions 115 , and are similarly defined by substitutions at a single amino acid residue, this time at position 44. The human inhibitory KIR specific for C1 + HLA-C have lysine, and those specific for C2 + HLA-C have methionine 165 . Phylogenetic based molecular diversity analyses, in conjunction with species divergence time estimates, show that residue 44 of C-ligand specific KIR evolves under positive diversifying selection 115 . Again, this mechanism of natural selection is evident in modern humans. Studies of indigenous Southern Africans identified a variant of KIR2DL1, frequent in the ≠Khomani population, that has lysine instead of methionine at residue 44 166 . The allotype (2DL1*022) is then able to bind C1 + HLA-C, instead of C2 + HLA-C that is recognized by other KIR2DL1 allotypes. In the neighboring Nama population, a different frequently-occurring substitution disables KIR2DL1 expression 148 . For reasons unknown, the frequency of C2 + HLA-C is uniquely high in southern Africa, suggesting the convergent emergence and natural selection of the two variants 148 occurred here to restore the evolutionary balance in favor of reproduction. Similar analyses of other KIR have identified examples of positively-selected single amino acid substitutions affecting interactions with their HLA class I ligands 74 , 140 , 167 .
These simple nucleotide mutations can affect immunity to infection. It has long been known that HIV infected individuals who possess the HLA-B*57:01 allotype develop T cell immunity, but only some are protected from progression to AIDS 168 . Somewhat independently, allotype specific interactions of KIR3DL1 with Bw4 + HLA-B also affect disease progression 134 . Among KIR3DL1 allotypes, those having valine at residue 47 were recently shown to distinguish the B*57:01 non-progressors from progressors 169 . Valine 47 allotypes had no effect on carriers of HLA-B*57:03, which differs by two amino acids from B*57:01 169 . Likely explaining this difference, the two substitutions mean that HLA-B*57:01 can present the same peptides as HLA-B*57:03, but at an orientation that enhances interaction with KIR3DL1 60 . This complex scenario shows how single amino acid variations of receptor or ligand can underlie subtle functional changes that have dramatic effect on control of infection.
Combinatorial
Effective fetal trophoblast invasion that occurs in the early stages of placentation is mediated by maternal uterine NK (uNK) cells. These cells are distinct from peripheral blood NK cells 110 , 111 , exhibiting little cytotoxicity, and interact with the fetal cells to mediate maternal spiral artery remodeling 112 . Inefficiency at this stage can lead to malnourished fetus, and preeclampsia in the latter stages of pregnancy. Of the three highly polymorphic HLA class I, only HLA-C is expressed by fetal extravillous trophoblasts. As evidenced through highly reproducible studies of life-threatening pregnancy disorders, including preeclampsia 10 , KIR interaction with HLA-C is therefore a second target for natural selection acting on the combinatorial diversity of receptor and ligand allotypes. Indeed, specific alleles and combinations of KIR and HLA class I provide the most consistent genome-wide determinants of preeclampsia 113 .
Diversification
Exemplifying the inverse correlation between KIR-A and C2 + HLA are East Asians, where the frequency of C2 + HLA-C is low and, although the preeclampsia risk remains, incidence is also low 145 . It is likely that the lack of C2 + HLA, whether through selection or genetic drift, has allowed interactions of HLA-A and -B with KIR to proliferate to an unusually high level in East Asia 41 . Here, there are no ‘null’ alleles of inhibitory KIR and the allotypes that have attenuated function are rare 158 . Accompanying this fully equipped KIR locus is a distinctly high ratio of HLA class I haplotypes carrying KIR ligands at either HLA-A or -B or both 41 . The HLA-A and -B alleles most frequent in Chinese Southern Han have ancestry distinct from their flanking genomic sequence, showing they were obtained relatively recently from admixture with neighboring populations, before rising to high frequency through natural selection 41 . Because HLA-A and -B are specialized for diversification of peptide presentation, this amplification likely serves to enhance both T cell and NK cell responses to intracellular pathogens. That adaptive introgression of HLA haplotypes is seen in other populations 159 , suggests the phenomenon is also widespread and ongoing. Introgression from ancient humans was likely a major contributor to the current HLA class I allele spectrum chiefly of East Asia and Oceania 160 . Although the ancestral populations that KIR alleles have been obtained from are more difficult to trace, in many cases the receptors have accompanied the ligands during the admixture events 41 , 160 . An example may be the recently identified open reading frame variants of HLA-H pseudogene, which were obtained from admixture with ancient humans 161 . These variants are expressed and functional and have a presently unidentified receptor expressed by NK cells 162 .
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.