A vision of KIR variation at super resolution

A challenge that faces natural killer (NK) cells in fulfilling their myriad functions is adapting to the evolution of their MHC class I ligands. The selection pressure imposed on polymorphic MHC class I molecules by the need to adapt to pathogens tests the ability of the killer cell immunoglobulin-like receptors (KIR) to maintain a functional relationship with these ligands. They have met this challenge, in part, by rapidly diversifying. The diversity of KIR proteins was originally described in the 1990s, following analysis of NK cells expressing the DX9 epitope.1 Subtle polymorphisms in KIR genes were noted in cDNA sequencing studies from several laboratories,2-4 but it was not until a genomic DNA analysis of 68 individuals had been performed that the extent of diversity of the KIR gene content began to be appreciated.5 In this study a 24-kb band on a Southern blot divided the KIR haplotypes into A (encoding multiple inhibitory and one activating KIR) and B (encoding multiple inhibitory and activating KIR) types, with subsequent further refinement into centromeric A and B, and, telomeric A and B haplotypes.6 Disease association studies, especially for infectious disease and pregnancy syndromes, have highlighted the role of diversity in KIR gene content in human disease. The recent and dramatic improvements in high-throughput and high-resolution sequencing technologies are providing new insight into the extent of diversity of the KIR gene family in humans. The 2015 KIR Polymorphism Workshop hosted KIR biologists from around the world to discuss the state-of-the art for KIR biology, embracing KIR genetics, evolution, function and disease association.

The structure of KIR haplotypes, with their repetitive nature containing highly polymorphic genes with frequent copy number variation, has made sequence assembly and allele-level typing challenging. As a consequence, knowledge of structural KIR diversity remains coarse, despite the need for better typing methods, particularly in the context of clinical transplantation. This state of affairs was highlighted by Vierra-Green, who showed how typing of common KIR haplotypes was largely successful. However, this could also be challenging due to variations in the copy number of specific KIR genes, and more difficult in specific populations, such as northern Native Americans. KIR typing of German bone marrow donor samples using amplicon sequencing shows promise, both in being cost effective and having the potential to yield allele information in the future (Wagner and Lange). However, high throughput methods may require refinement to accurately type more diverse KIR genes including KIR3DL1/S1 (Alicata and Moretta). The improvement of KIR typing resolution is clearly an area where new technologies offer much potential. The ability of Pacific Biosystems sequencing to resolve the maternal and paternal KIR haplotypes, based on a fosmid library, was illustrated (Hall and Ranade) as well as more traditional methods like LinkSeq, a quantitative PCR KIR-specific detection mechanism (Russnak and Antovich).

Analysing what can be very complex sequence data sets is often a greater challenge than generating the data in the first place. Using pre-existing sequencing data sets offers an enormous resource. Nemat-Gorgani and Parham exemplified some of the analysis issues with a study of divergent African populations. These populations contain a high frequency of novel KIR alleles and haplotypes that often precludes accurate genotyping. A combination of pyrosequencing, Sanger sequencing and Illumina technology enabled the discovery of several novel KIR variants predicted to have altered functionality, on a background of conserved telomeric B but highly diverse centromeric B haplotypes. Two methods were presented that use short read data to call KIR haplotypes by recognizing KIR-specific reads, which are then used to call genotype and allele. A method named SeRRAMC was applied to a cancer cohort, correlating the absence of functional KIR2DS4 with greater risk of developing cancer (Rossof and Lee). Of particular note was the extremely high accuracy and rigour of using short reads in an in silico sequence-specific oligonucleotide probe approach, both for identifying known haplotypes and alleles but also in accurately constructing novel alleles sequences (Norman and Parham).

As diversification of KIR is not limited to humans, comparative studies to other catarrhine primates have considerable potential to inform evolutionary mechanisms and the scope of KIR diversity. The orang-utan occupies an intermediate position in MHC and KIR evolution, with the main KIR ligand, MHC-C, emerging just before orangutan speciation. The structure of orang-utan KIR haplotypes is similar to that in other great apes, but with greater diversity in the centromeric region, and relatively few KIR alleles shared by the Sumatran and Bornean orang-utan species (Guethlein and Parham). Apart from catarrhine primates, the only other animals shown to have a diversified KIR gene complex are domestic cattle and their wild ancestor, the aurochs, in which a completely independent expansion of KIR genes occurred. Nevertheless, the similarities between cattle and primate KIR are striking, including significant polymorphism and a dominance of inhibitory receptors (Gibson and Hammond). However, cattle also have a highly diverse NK cell receptor repertoire based upon the C-type lectin-like receptors and it remains to be determined how this diversity translates into function (Schwartz and Hammond). In terms of the evolution of individual KIR genes, they come and go, as illustrated by the resurrection of the human-specific and highly polymorphic pseudogene KIRDP1 (Parham).

KIR genes are expressed in a variegated fashion by NK cells, with both the frequency and levels of KIR expression exhibiting substantial donor variation. The control of variegated expression is an elaborate process, associated with a ‘probabilistic switch’ in the KIR promoter. In a further refinement of his original model, Anderson demonstrated the influence of Pro1 elements on further controlling KIR expression in a tissue-specific manner. Control of cell-surface levels of KIR proteins can also be related to differences in the protein-coding region of the KIR2DL5 gene (Vilches) and their context within the KIR locus (Roberts and Riley). Similarly, allelic variants of KIR2DL1 are expressed on the cell surface at different levels, and they segregate on different KIR haplotypes (Falco and Moretta). The importance of the surface levels of KIR is that they affect NK cell function, with KIR analysis by stochastic optical reconstruction microscopy revealing that these levels can influence KIR clustering at the NK cell synapse (Kennedy and Davis). Furthermore, NK cell inhibition is further impacted by the ability of zinc ions to mediate KIR polymerization (Rajagopalan and Long). KIR expression can also affect disease outcomes. Hence the chemotherapeutic agent 5-azacytidine used to treat myelodysplastic syndrome affects KIR expression and as a consequence augments the anti-leukaemic effect of NK cells (Sohlberg and Malmberg).

The gross specificity of the inhibitory KIR that recognize HLA-A, -B and –C epitope alleles has been known for two decades, but the fine specificity of the inhibitory KIR for peptide–MHC complexes and the specificity of the activating KIR continue to attract the attention of KIR researchers. Parham reported that KIR2DS5 can be an activating receptor specific for the C2 epitope of HLA-C, but this ligand specificity is exhibited only by some KIR2DS5 allotypes. Hence, it is possible to correlate the protective effect of KIR2DS5 against pre-eclampsia in Ugandans,7 with its ligand-binding specificity, demonstrating the importance of allele-level resolution KIR typing in studies of disease. The KIR3DL1 gene is most diverse, and this diversity impacts upon its ligand binding (Brooks). Conversely the activating receptor KIR3DS1, which segregates as an allele of the inhibitory receptor KIR3DL1, was found not to share an HLA-Bw4 specificity with KIR3DL1, but to bind to HLA-F, one of three non-classical MHC class I molecules (Garcia-Beltran and Altfeld). Structural studies have given deeper insights into allelic diversity, in terms of the subtle differences in the orientation of binding to HLA-C by KIR2DL2 and KIR2DL3 (Pymm and Rossjohn) and also the gross structural difference between KIR2DL4 and the other human KIR2D (Moradi and Vivian).

Challenging the concept of a simple motif-based ligand-binding model of the inhibitory KIR, Sim, Boyton and Long, showed data indicating that the HLA-C2-specific inhibitory receptor KIR2DL1 can bind group 1 HLA-C allotypes if an appropriate peptide is presented. Furthermore the 26 allotypes of KIR2DL1 and KIR2DS1 comprise four distinct phylogenetic clades (Hilton and Parham). With few exceptions, the members of each clade segregate with either A or B haplotypes and their ligand specificity for C1 or C2 allotypes correlates with the haplotype on which they are located. The enigmatic activating receptor KIR2DS2 can also bind HLA-C1 allotypes, but only in the context of specific peptides (Khakoo). Peptides derived from hepatitis C virus can also be presented by MHC class I to NK cells and hence induce KIR binding (Lunemann and Altfeld). This could be relevant in vivo, because the peptide-binding repertoire of MHC class I affects NK cell inhibition (Mbiridindi and Khakoo). For example, in human immunodeficiency virus (HIV) infection, Holzemer and Altfeld showed from sequence data analysis that in vivo an HLA-C-binding peptide epitope mutates in a manner consistent with selection pressure by NK cells on HIV. Consistent with this model, lower levels of NK cell inhibition lead to increased killing of HIV-infected cells (Korner and Altfeld). The impact of the NK cell response on HIV is determined at the genetic level. Using a whole genome sequencing approach, Carrington described how specific KIR3DL1 alleles are associated with controlling HIV in the presence of one of its Bw4⁺ MHC class I ligands, HLA-B*57. Analysis of high-resolution genetic data from a cohort of individuals chronically infected with hepatitis B, presented by Hollenbach, revealed a protective effect for KIR3DL2 that was mediated only by some allelic variants.

Chronic viral infection, especially cytomegalovirus, can lead to expansion of NK cells expressing particular KIR. Uhrberg showed how the KIR A and KIR B haplotypes can have differential expression of a shared KIR gene, again indicating the importance of the haplotypic context of each KIR gene in determining the expressed KIR repertoire. This has implications for bone marrow transplantation, as cytomegalovirus-driven adaptive NK cells can have an augmented anti-leukaemic effect (Miller). New insights into the protective and susceptibility effects of KIR in a range of inflammatory and autoimmune disorders were described by Rajalingam, Wellington and Elliott, Kollnberger and Bowness, and, Traherne and Trowsdale. Kusnierczyk, Chazara and Moffett demonstrated that KIR genes are also important in the outcome of pregnancy. Bjorkstrom, Sharkey, Ivarsson and Moffett complemented these genetic models with phenotypic studies demonstrating the uniqueness of the uterine NK cell repertoire and the potential for differences in the education of uterine NK cells compared with peripheral blood NK cells. In particular, these studies have validated researchers' efforts to obtain difficult-to-collect human samples by giving novel insights into basic NK cell biology and opening up new ideas for NK cell-associated diseases.

Conclusion

For many years mechanistic understanding of the KIR variation underlying disease associations has been hampered by the low resolution gene-content typing of the extraordinarily complex KIR genes. This KIR workshop marked an exciting moment in time, by clearly demonstrating that recent gains in sequencing capacity and allele resolution now make population level KIR and HLA genotyping possible at the highest resolution. Combining these technological advances with fine-resolution studies of KIR specificity and function now provides human immunogeneticists with the opportunity to define in complete detail the role of KIR, HLA class I and NK cell biology in the pathogenesis of infectious, inflammatory, neoplastic and pregnancy-associated diseases. Converting these ideas into treatment opportunities will be our next major challenge. The 10th KIR workshop will be organized by Francesco Colucci and will take place in Cambridge, UK in Spring 2017.

Disclosures

The authors have no competing interests to declare.