Unparalleled Rapid Evolution of KIR Genes in Rhesus and Cynomolgus Macaque Populations

Natural killer cells provide an early defense mechanism against infectious diseases and tumor formation by their ability to recognize and kill cells with aberrant MHC class I expression (1–3). This immune surveillance is modulated by killer cell Ig-like receptors (KIR), which are expressed on NK cells and subsets of T cells (4–6). These gene products are transmembrane receptors consisting of two or three extracellular domains, which can facilitate ligand interaction, and a long or short cytoplasmic tail that can utilize an intracellular ITIM or ITAM, respectively (3, 7). In humans, the gene family encoding the KIR genes is located on chromosome 19 q13.4, and its complexity is reflected by allelic polymorphism, gene copy number variation (CNV), chromosomal recombination, and alternative splicing (8–11).

Comparison of the KIR gene cluster in humans and other primate species suggests a first round of expansion to occur between 30 and 45 million years ago (12), which involved two progenitor genes. The KIR3DX1 lineage is nowadays represented by a single copy in primates but expanded in cattle, whereas the KIR3D progenitor gene was subjected to diversification by tandem duplications, deletions, and recombinations (13, 14). This expansion resulted in a head-to-tail gene cluster encoding a broad repertoire of KIR genes, the overall architecture of which is conserved in primates. Species-specific diversification, however, may have resulted in differential lineage expansions and sequence variation, which is reflected by few KIR orthologs that are shared between distantly related primate species. Primate KIR genes are phylogenetically classified into lineages based on receptor structure and ligand specificity. In humans, lineage I includes KIR2DL4 and KIR2DL5, lineage II KIR3DL1/L2/S1, the expanded lineage III KIR2DL1-3, KIR2DS1-5, and the pseudogenes, and lineage V KIR3DL3, respectively. The initial expansion of lineage III members can be traced back to orangutans, and its emergence seems to have coevolved with the presence of HLA-C–like genes, which are present on ∼50% of the contemporary orangutan MHC haplotypes (15). In chimpanzees and humans, the lineage III KIR genes expanded further, and their genomic clusters comprise 17 and 13 KIR genes, respectively, but only four genes are considered orthologs (14). Old-World monkeys, like macaques (genus macaca), expanded mainly lineage II KIR genes (KIR3D), which may be associated with their expanded MHC class I repertoire (16, 17).

Macaques are geographically the most widespread nonhuman primates (NHP) that diversified from the human and great ape lineage ∼25 million years ago and include ∼20 species that share a habitat spanning from northeast Africa to Asia. Rhesus and cynomolgus macaques (Macaca mulatta, Macaca fascicularis) are closely related species that diverged from each other ∼1–3 million years ago. Rhesus macaques are distributed across South, East, and Southeast Asia, whereas cynomolgus macaques mainly inhabit the mainland and islands of Southeast Asia. Geographically distinct populations, such as the Indian, Burmese, and Chinese rhesus macaques and the insular cynomolgus macaques, emerged by means of natural barriers and resulted in intraspecific variation. The Isthmus of Kra, which is the narrowest part of the Malaysian peninsula, separates the cynomolgus macaques that inhabit the mainland of Southeast Asia in a northern (Cambodia, Thailand, and Vietnam) and southern (Malaysian peninsula) population, and it is suggested that this geographical barrier restricts gene flow (18, 19). In Indochina, rhesus and cynomolgus macaques may come across each other, and bidirectional introgression is substantiated by shared genetic features (20–22). For example, ancestral haplotypes of the highly polymorphic MHC class I region are encountered in rhesus macaques and cynomolgus macaques (16, 23, 24), whereas extensive allele sharing had been documented for the MHC class II genes (25).

Several sequencing platforms have been used to characterize the macaque KIR gene region, particularly in Indian rhesus macaques (10, 26–30). Data on the KIR gene cluster and repertoire in other rhesus macaque populations is limited. For cynomolgus macaques, only the Mauritian animals were characterized thoroughly (31–33). This population was founded by a few animals that were introduced to the island by human interference ∼500 y ago and, therefore, have a restricted KIR gene content.

Rhesus and cynomolgus macaques are used as preclinical models for many infectious and autoimmune diseases, as the immune response and pathologies reflect the human situation (34, 35). The origin of macaques, however, vary between different research facilities and might impact the disease phenotype, which is, for example, reported for SIV/AIDS studies in Indian and Chinese rhesus macaques (36, 37). The presence or absence of certain KIR genes, in combination with the MHC class I ligands, have been associated with disease susceptibility in both humans (6) and macaques (38, 39). A comprehensive overview of the KIR gene content and repertoire of different natural macaque populations is, however, lacking, despite the potential refinement for macaque models. Therefore, we set out to analyze the KIR transcriptomes of cohorts of rhesus and cynomolgus macaques of different geographical origins, which probably experienced varying selective pressures. Our observations illustrate in both highly related macaque species and populations an unparalleled form of rapid evolution of KIR genes that is propelled by point mutations and complex chromosomal recombinations, which generate novel gene entities and result in highly variable haplotype architectures.

Forty-six rhesus macaques, comprising seven families, and 70 cynomolgus macaques, comprising 11 families, were selected from the self-sustaining colony housed at the Biomedical Primate Research Centre. During the annual health checks, EDTA or heparin whole blood samples were obtained, and PBMCs were isolated from the latter. PBMC samples from 16 Chinese rhesus macaques, comprising seven families, were obtained from the Biomedical Primate Research Centre Bio-bank.

The geographical origin of most rhesus macaques was known based on importation records, such as the families from the Indian, Chinese, and Burmese populations. Additional transcriptome data of Indian rhesus macaques was incorporated from a previous KIR study conducted by our laboratory (10). The geographical origin of the cynomolgus macaques was mainly deduced by phylogenetic comparison of mitochondrial 12S rRNA gene segments (40). With regard to this data, we defined three cynomolgus macaque populations, which originated from the mainland of Malaysia, from the Indonesian and Malaysian islands, and from Mauritius. The mainland population was further divided into populations north and south of the Isthmus of Kra. The origin of three cynomolgus macaques (Ji0603077, J15028, J16019) could not be determined unambiguously. In addition, previously reported KIR haplotypes of Mauritian cynomolgus macaques were added to the analysis (33).

Total RNA was extracted directly from EDTA whole blood samples or from ±15 × 10⁶ PBMCs with RNeasy Mini Kit (Qiagen, Valencia, CA) in accordance with the manufacturer’s instructions. First-strand cDNA was synthesized with the RevertAid First Strand cDNA Synthesis Kit (Invitrogen, Carlsbad, CA) using oligo(dT)18 primers. Genomic DNA was extracted from EDTA whole blood samples by a standard salting-out procedure, or from ±15 × 10⁶ PBMCs with an AllPrep RNA/DNA Mini Kit (Qiagen) according to the manufacturer’s instructions.

Full-length KIR transcripts were obtained by amplification of total cDNA with a KIR2DL04-specific and two KIR1D/KIR3D-generic primer sets in accordance with a previously reported protocol (10). These primer sets were cross-reactive for the different rhesus and cynomolgus macaque populations. PCR products were size-selected by gel electrophoresis (±1250 bp) and purified using a GeneJet Gel Extraction Kit (Invitrogen). The samples were pooled and purified twice using AMPure XP Beads (Beckman Coulter, Woerden, the Netherlands) at a 1:1 bead-to-DNA volume ratio. The DNA concentration of purified pooled samples (>1 μg total DNA) was measured using the Qubit dsDNA HS Assay Kit and Qubit 2.0 Fluorometer (Thermo Fisher Scientific).

PacBio SMRTbell libraries were generated according to Pacific Biosystems “Procedure and Checklist - Amplicon Template Preparation and Sequencing,” and sequencing was performed using a PacBio Sequel platform with a 10 h video time using sequencing kit versions 2.0, 2.1, and 3.0, which was performed by the Leiden Genome Technology Center.

Circular consensus sequences were selected for high-read quality (value of 0.99 or higher), and demultiplexed based on unique barcoding.

Geneious Prime 2019 software was used to map the circular consensus sequences to a database, consisting of reported rhesus macaque and novel cynomolgus macaque KIR sequences, to identify 100% matching reads (100% overlap, 0% mismatch, maximum ambiguity = 1). The unused reads of related animals were grouped and were de novo assembled. The consensus of each de novo contig was trimmed for the primer sequence, and phylogenetically aligned with the rhesus and cynomolgus macaque database. De novo sequences were confirmed when shown to segregate or when identified in two separate PCRs and were subsequently submitted to the European Nucleotide Archive and assigned an accession number (https://www.ebi.ac.uk/ena/).

The nomenclature of the KIR transcripts in rhesus and cynomolgus macaques follows the general guidelines of the KIR nomenclature report for NHP (41). In brief, the name of the gene indicates the number of domains (1D, 2D, or 3D) and the signaling function (S or L). The inclusion of a “W” implies a workshop gene, which indicates a gene that is divergent on the basis of phylogenetic analysis but lacks sufficient reliability because of the low frequency or because of the absence of genomic sequencing or family studies. The inclusion of “Q” indicates that it is questionable whether the transcripts are feasible. Two digits distinguish the different genes, and an asterisk followed by three digits distinguishes alleles. Two additional digits indicate synonymous variation.

Novel cynomolgus macaque KIR sequences were compared with a database of 342 reported rhesus macaque sequences (10, 26, 29, 30) and newly identified transcripts by phylogenetic analysis, using the Neighbor-Joining Tree-Building Method (best tree mode) in MacVector software (MacVector, Cambridge, U.K.). Phylogenetic clusters were confirmed by the Maximum Likelihood Comparison and Neighbor-Joining Tree-Building Methods in MEGA7 software, and all methods provided similar trees. Rhesus and cynomolgus macaque KIR sequences that clustered together with a close phylogenetic distance were considered orthologs and received matching KIR gene names. Clusters of Mafa-KIR sequences that diverged from the other sequences according to sequence comparison and phylogenetic analysis received a workshop number. In addition, workshop numbers were assigned to cynomolgus macaque KIR genes that were thought to be the result of recombination events, as these are considered novel entities. In contrast, recombinant KIR genes in rhesus macaques are named after an allele of the gene that contributes the largest gene segment, as is described for this species in the NHP nomenclature report (41). The previously reported 46 Mafa-KIR sequences (33), all of which originated from the Mauritian cynomolgus macaque population, were also named.

The origin of each KIR haplotype was categorized per macaque population. The populational origin of the rhesus macaques determined the haplotype origin, as none of the rhesus macaques had parents from different populations. The KIR haplotypes defined in cynomolgus macaques that had their roots in the mainland of Malaysia (north or south), the Malaysian/Indonesian islands, or in Mauritius were categorized based on the defined origin. In cynomolgus macaques with mixed roots (parents from the mainland and from islands), the origin of the KIR haplotypes was determined by the sequencing of parental genomic DNA, the origin of which was known, using an Mafa-KIR3DL20 exon 4–specific primer set (forward: 5′-GAAGAGACGGTCATCCTGCAGT-3′; reverse: 5′-ACTCCCCCTATGTGTTGTCAGC-3′) and an Mafa-KIR1D exon 4–specific primer set (forward: 5′-GAAGAGACGGTCATCCTGCAGT-3′; reverse: 5′-ACTCCCCCTATGTGTTGTCAGC-3′). Thermal cycling conditions were denatured at 98°C for 2 min, followed by 32 cycles of 98°C for 20 s, 63°C for 25 s, and 72°C for 1 min. Amplicons of ∼180 bp were size-selected by gel electrophoresis and purified using a GeneJet Gel Extraction Kit (Invitrogen). Sanger sequencing was used, and the populational haplotype origin could be determined on the basis of three single-nucleotide polymorphisms.

The frequency of a KIR gene in rhesus and cynomolgus macaques, or in one of the populations, was determined based on the presence of at least a single copy on a haplotype, the origin of which was determined, rather than on the presence of the gene in an individual.

The KIR transcriptomes of 62 rhesus and 70 cynomolgus macaques covering different populations were subjected to analysis (Fig. 1) (10). All macaque samples belong to families that comprised two or more individuals, which allowed us to confirm the segregation of alleles but also to define haplotypes (Figs. 2, 3). The origin of the rhesus macaques was documented thoroughly and included Burmese (n = 14), Chinese (n = 16), and Indian (n = 32) origins (Fig. 1). Based on the phylogeny of mitochondrial DNA sequences (40), origins of the cynomolgus macaques were mapped to the mainland of Southeast Asia (n = 26), the Malaysian/Indonesian islands (n = 4), or Mauritius. The mainland population could be further divided into populations north (n = 23) and south (n = 3) of the Isthmus of Kra (Fig. 1). For 19 cynomolgus macaques, a mixed origin was documented, whereas for 21 animals only the origin of a single parent could be determined. To expand our population panel, we included previously reported KIR transcriptome data from 30 Indian rhesus macaques (10) and 30 Mauritian cynomolgus macaques (33). Altogether, three rhesus and four cynomolgus macaque populations were subjected to comparison for their KIR repertoire.

Up to now, 342 distinct rhesus macaque KIR alleles that were mainly isolated from Indian animals have been identified (10, 26, 29, 30). In the current cohort that comprises 32 Indian rhesus macaques, again, another 48 unreported KIR alleles were discovered, indicating extensive allelic variation within this population. All Indian rhesus macaque KIR alleles could be clustered into 22 different KIR genes (Table I). From the Burmese and Chinese cohorts, 73 and 117 novel KIR alleles were isolated, respectively, which clustered to previously reported but also newly discovered KIR gene entities (Table I). During the course of this study, 34 rhesus macaque KIR genes were defined, which comprised 238 novel alleles, and 64 reported Mamu-KIR alleles were confirmed (Supplemental Fig. 1). The emergence of one of the novel alleles was observed in rhesus macaque R04104, which is expected to be homozygous for the KIR region, as it ought to receive two copies of Mamu-KIR3DL05*006:01 via the H21-A haplotype (Fig. 2). However, one copy of the Mamu-KIR3DL05*006:01 allele shows nonsynonymous mutations at two positions in the D1 domain (T > C and G > T), thereby generating a novel allele, designated Mamu-KIR3DL05*032. This de novo allele segregated with its corresponding haplotype (H21-B) into two offspring of R04014, and its existence was further substantiated by independent Sanger sequencing (Supplemental Fig. 2).

Most allelic variation is controlled by Mamu-KIR3DL07, -KIR3DL20, -KIR3DL01, and, to a lesser extent, -KIR2DL04. The Indian and Burmese populations share four KIR alleles, whereas only a single allele was shared between the Indian and Chinese populations (Mamu-KIR3DS06*016), the Burmese and Chinese populations (Mamu-KIR3DL05*007:01), and all three populations (Mamu-KIR3DL01*019:03) (Fig. 4).

Knowledge of the KIR cluster in cynomolgus macaques is mainly confined to the artificially introduced Mauritian population, and 49 alleles are documented (31, 33). In the current cohort from different populations, we identified 267 novel alleles that clustered into 55 distinct KIR genes (Supplemental Fig. 1, Table II). In addition, 10 of the 46 previously reported Mafa-KIR sequences identified in Mauritian cynomolgus macaques were confirmed (31, 33). The highest level of allelic variation was observed for Mafa-KIR3DL20 and -KIR1D, followed by -KIR2DL04, -KIR3DL01, and -KIR3DL07. The different populations seem to have highly unique allelic KIR repertoires. A single allele was shared between the northern mainland and the Indonesian/Malaysian populations (Mafa-KIR3DLW23*001) and the southern mainland and Mauritian populations (Mafa-KIR2DL04*002), whereas two alleles were in common between the Indonesian/Malaysian islands and Mauritian populations (Mafa-KIR3DLW13*003, Mafa-KIR3DLW26*001), and the southern mainland, the Indonesian/Malaysian islands, and Mauritian populations (Mafa-KIR1D*030Q, Mafa-KIR3DL20*002) (Fig. 4).

To sum this up, 579 KIR alleles were identified in the rhesus and cynomolgus macaque populations studied. Only two alleles were shared between both highly related species: namely, Mamu-KIR3DLW12*002/Mafa-KIR3DLW12*006 and Mamu- and Mafa-KIR3DLW18*001 (Fig. 4). The low number of allele sharing between the macaque species as well as the different populations suggests fast evolution. This is within lineages mainly mediated by point mutations, and contrasts the extensive sharing documented for MHC class II and, to a much lesser extent, for MHC class I alleles (42).

Considering a shared ancestor living 1–3 million years ago, one might expect highly similar repertoires of orthologous KIR genes in rhesus and cynomolgus macaques, as is observed for the closely related Bornean and Sumatran orangutan species (15). Apparently, however, this is not the case in both macaque species, as their KIR gene repertoires possess species-specific and a differential number of KIR gene moieties. Moreover, the 34 rhesus and 55 cynomolgus macaque KIR genes that are defined by sequence comparison and phylogenetic analysis indicate a greater expansion of the KIR gene repertoire in macaques as compared with humans and other primate species, for which 17 or fewer KIR genes were identified (41, 43, 44). The question to be answered, therefore, is how are new KIR genes generated. One mechanism that might explain the expanded macaque KIR gene repertoire is the occurrence of abundant recombination events, which result in the formation of hybrid genes composed of segments from two different KIR genes (Tables III, IV). Along with others, we found evidence of similar events in humans (10, 45), although this mechanism seems to happen more frequently in macaques. In rhesus macaques, hybrid KIR genes are named after the allele that contribute the largest segment (41). For example, multiple entities have a large Mamu-KIR3DL07 segment, which is found in conjunction with a smaller segment of -KIR3DL05, -KIR3DL08, or -KIR3DSW08, but all are named and listed as alleles of Mamu-KIR3DL07 (Table III). Another peculiar recombination event resulted in Mamu-KIR3DS04*011, the extracellular domains (exons 1–5) of which originate from -KIR3DS04, whereas the cytoplasmic tail is similar to exons 6–9 of -KIR3DL07. The name is, therefore, somewhat confusing, as this gene is listed as an allele of Mamu-KIR3DS04, although it encodes an inhibitory cytoplasmic tail. In rhesus macaques, at least 19 hybrid KIR genes were generated by recombination events (Table III). It would seem that for some of these hybrids the nomenclature is in need of attention (46). From a more general and functional perspective, hybrid gene entities could encode novel genes with potentially distinct functional features, due to differential combinations of ligand-binding domains and signal transduction elements.

In cynomolgus macaques, at least seven hybrid KIR genes were detected (Tables II, IV). For example, the first six exons of Mafa-KIR3DSW21 are highly similar (98–99%) to those of -KIR3DL07, whereas the transmembrane region and cytoplasmic tail of -KIR3DSW21 is identical to -KIR3DSW12. This suggests that Mafa-KIR3DSW21 may interact with similar ligands as -KIR3DL07 but that it transduces activating instead of inhibitory signals. Seven Mafa-KIR3DSW21 alleles are identified (Table II), suggesting a positive selection for variation on the gene products generated by this recombination event.

The origin of both segments could not be identified for all hybrid KIR genes. Mamu-KIR3DL05*029/*030/*033, Mafa-KIR3DLW24, Mafa-KIR3DLW29, Mafa-KIR3DSW18, and Mafa-KIR3DSW20 seem to have segments of Mamu-KIR3DL05, Mafa-KIR3DLW12, Mafa-KIR3DLW13, Mafa-KIR3DSW17, and Mafa-KIR3DSW19, respectively, but it was not possible to trace the donor of the other segment (Tables III, IV). This indicates that when more sequences become available, additional hybrid KIR gene entities and segments are likely to be defined in macaques.

Within the macaque KIR repertoire studied, 24 macaque KIR genes were highly similar and were considered to be orthologs. These genes most likely represent a single locus in both species, although it is too early to elucidate their exact physical location, as the relevant genomic studies are in progress. One would expect that the number of 24 orthologs shared between two closely related macaque species reflects common ancestry, whereas the relatively high number of species-specific KIR genes indicates the rapid generation of novel gene entities, which can, in part, be explained by abundant recombination.

The family-based study design resulted in the thorough characterization of 49 rhesus and 43 cynomolgus macaque KIR haplotypes (Figs. 5, 6), which are categorized on the basis of geographical origin of the analyzed animals (Fig. 1). The rhesus macaque KIR haplotypes are referred to as Rh-H13 to Rh-H60, consecutively to the 12 previously reported haplotypes (Fig. 5) (10). Cynomolgus macaque KIR haplotypes are referred to as Cy-H1 to Cy-H43, whereas the previously reported Mauritian chromosomal KIR configurations are listed as K1–K8 (Fig. 6) (33). All these haplotypes display extensive CNV. The rhesus macaque haplotypes encoded 4–17 KIR transcripts, whereas the cynomolgus macaque equivalents encoded 3–13 KIR transcripts. KIR3DL20 was identified on most macaque haplotypes, except for haplotype Cy-H39, and seems to be absent on the haplotypes of the Mauritian animals. However, we assume that KIR3DL20 should be considered a framework gene in macaques and that a few transcripts were missed because of primer inconsistencies. For the Mauritian cynomolgus macaque, this assumption is confirmed by haplotype Cy-H9, which is identical to K3, defined by another research team, except for the presence of Mafa-KIR3DL20*002. Haplotype Cy-H9/K3 is identified in three populations and may indicate an ancestral origin. KIR2DL04 is observed on 70 and 94% of the rhesus and cynomolgus macaque KIR haplotypes, respectively, and represent the only reported macaque KIR gene that shares an apparent ortholog with humans. In humans, this gene is considered a framework gene, and there is support that this might also be the case for its cynomolgus macaque ortholog (47).

Chromosomal recombinations such as unequal crossing over, gene fusion, and gene duplications can expand or contract a KIR haplotype, thereby affecting the genetic content. Two or more apparent allelic copies of a given KIR gene were identified on 23 of the 49 rhesus and 11 of the 43 cynomolgus macaque KIR haplotypes. It is likely that such genes were once orthologs, but owing to complex recombination events, they might end up as paralogs. These duplications involved mainly lineage II inhibitory KIR genes, such as KIR3DL01, KIR3DL07, and KIR3DL11, but also Mamu-KIR3DL20 (Rh-H27), Mafa-KIR1D (Cy-H11), and Mafa-KIR3DSW12 (Cy-H1). In total, 15 different KIR genes are duplicated on the listed haplotypes, 11 of which are considered orthologs. This suggests that ancestral genes are more often subject to duplication as compared with more recently generated species-specific KIR genes. The most extensive CNV is witnessed for Mamu-KIR3DL01 on haplotype Rh-H26 (Fig. 5), on which four allelic copies exist.

Hybrid KIR genes (Tables III, IV) are associated with chromosomal recombination events and might mark expanded and contracted KIR haplotypes. For example, two hybrid Mamu-KIR3DL20 genes, composed of exons encoding the extracellular domains of Mamu-KIR3DL20 (exons 1–7), and the cytoplasmic tail of Mamu-KIR1D (Rh-H38) or Mamu-KIR2DL04 (Rh-H14, -H18), seem to coincide with a centromeric haplotype contraction. Also, haplotype Rh-H27 carries an example of a gene that consists of the first four exons of Mamu-KIR3DL01 and the last five exons of Mamu-KIR3DL08 (Fig. 5, Table III). The formation of this gene probably resulted in another contracted haplotype, as only four KIR genes are present at the telomeric end. In the cynomolgus macaque, haplotypes Cy-H6 and -H21 seem to be expanded, marked by the presence of the recombinant genes Mafa-KIR3DLW27 and Mafa-KIR3DSW20, whereas the relatively short haplotype Cy-H38 contains another hybrid gene Mafa-KIR3DSW21, the emergence of which might have resulted in a contraction (Fig. 6, Table IV).

This in-frame fusion mechanism occurs rather frequently, as 21 rhesus macaque and 21 cynomolgus macaque haplotypes contain a recombinant gene (Figs. 5, 6), although not each hybrid gene seems to mark contraction or expansion. Thus, the KIR gene cluster in both macaque species seems to be subjected to frequent gene duplications and chromosomal recombination events, which not only generate novel gene entities, but also result in a differential KIR haplotype architecture.

The occurrence of at least 24 orthologs in both macaque species is most likely due to the sharing of a common ancestor, but introgression between the two species may also have an impact. The frequency of these orthologs, however, differs considerably between both macaque species (Fig. 7). The orthologous genes that are encountered more frequently on rhesus than on cynomolgus macaque haplotypes are KIR3DL01, KIR3DLW03, KIR3DL05, KIR3DL06, KIR3DL07, KIR3DS02, KIR3DS06, and KIR3DSW08. It is noted that for these genes the allelic variation is higher in rhesus macaques than in cynomolgus macaques (Tables I, II).

Other orthologs are more often present in cynomolgus macaques, such as KIR3DL11, KIR3DLW12, KIR3DLW14, KIR3DLW25, KIR3DSW10, KIR3DSW20, and KIR3DSW21, that, with the exception of KIR3DL11, display greater allelic variation compared with rhesus macaques (Tables I, II). An exceptional example is formed by Mafa-KIR1D, which is present on 82% of the haplotypes in cynomolgus monkeys but only on 22% of the haplotypes in rhesus macaques. Moreover, the allelic variation of Mafa-KIR1D exceeds that of Mamu-KIR1D, despite the difference in the number of animals studied per species (Tables I, II). These differences may indicate that KIR1D in cynomolgus macaques executes a more essential role.

On average, one more inhibitory KIR gene was present on haplotypes of rhesus macaques, whereas an additional activating KIR was encoded on cynomolgus macaque haplotypes. The differential gene and allele frequencies are indicators for species-specific selection and might involve different infectious pathogen encounters due to varying habitats.

The populations of rhesus and cynomolgus macaques (Fig. 1) parade differences in KIR gene content and gene frequency. Rhesus macaques from the Burmese population encoded, on average, one and two additional KIR3DL and KIR3DS receptors, respectively, as compared with the haplotypes that stem from the Indian and Chinese populations. Approximately 70% of the haplotypes contained at least one Mamu-KIR3DL01 and/or Mamu-KIR3DL07 copy, regardless of the origin, whereas multiple other KIR genes were differently distributed over the rhesus macaque populations (Fig. 8). For example, Mamu-KIR1D is located on 56% of the haplotypes from Burmese animals, whereas it is present on only 16 and 14% of the Indian and Chinese rhesus macaque KIR haplotypes, respectively. Eleven KIR genes were identified in a single rhesus macaque population, including newly defined activating KIR3DS genes that were only encountered in the Burmese cohort studied.

In cynomolgus macaques, animals that originate from the mainland populations seem to have, on average, one additional inhibitory KIR receptor (KIR3DL) per haplotype as compared with the subjects that inhabit the Indonesian/Malaysian islands. Differential gene distribution trends are observed for several KIR genes (Fig. 9). For example, Mafa-KIR3DL01 and -KIR3DLW12 were more frequently identified in the northern-mainland population, whereas -KIR3DSW13 and -KIR3DLW28 were found present only in the Mauritian population. Activating KIR genes with orthologs in Indian rhesus macaques were mainly identified in the northern-mainland population, including KIR3DS02 and KIR3DSW07, whereas the other cynomolgus macaque populations have species-specific activating KIR genes.

Genes that were identified in either three rhesus or four cynomolgus macaque populations mainly encode inhibitory receptors (Figs. 8, 9). The activating receptor are more often observed in two or one populations (Figs. 8, 9). Overall, the observed variable gene content and gene frequency in the different macaque populations support evidence pointing to rapid evolution of the KIR genes at the population level.

An essential step in the evolution of the primate KIR cluster started with the initial expansion of a lineage II KIR gene progenitor. Subsequently, other KIR lineages seem to have emerged through deletion and recombination events. In macaques, lineage II KIR genes (KIR3D) were subjected to substantial expansion (10, 17, 32), which coincides with an extended MHC class I gene repertoire (16, 24). The present study involves the comparative analysis of rhesus and cynomolgus macaque populations from distinct geographic areas. The KIR gene repertoires were found to reflect rapid evolution. Our data illustrate that not only within these closely related species, but even within their populations, new KIR gene entities are generated by complex recombination processes resulting in the formation of hybrid genes. In addition, a high level of allelic polymorphism was encountered in both macaque species, but the sharing of alleles was virtually absent. Moreover, recombination resulted in marked differences in the KIR haplotype architecture of both species, again testifying the rapid evolution of the macaque KIR genes, which has not been described in other NHP species.

In humans, the KIR gene cluster mainly diversifies at the allelic level, whereas gene expansion is modest and mainly confined to lineage III genes (43). Two major haplotype configurations are recognized, for which a trade-off has been suggested based on differential haplotype frequencies in human populations (47). The A haplotype configurations standardly express seven receptors and have an inhibitory profile, whereas the B haplotypes show moderate gene content variability, including multiple activating KIR genes (7–13 KIR genes) (Fig. 10) (44). Chimpanzees (Pan troglodytes) and humans diverged from a common ancestor ∼5 million years ago, and, although the complexity of the KIR clusters is, to some extent, comparable, species-specific diversification is observed in receptor structure, haplotype architecture, MHC class I recognition potential, and gene content (48). The chimpanzee KIR region mainly comprises inhibitory genes and resemble human A haplotypes. Several chimpanzee KIR genes are actually recombinant genes (48). The repertoire, however, is limited to 13 KIR genes, four of which are orthologous framework genes that are shared with humans (Fig. 10) (41, 48). Although little is known about the allelic variation in chimpanzees, the limited KIR gene repertoire might suggest that the ancient selective sweep, which targeted the ancestral chimpanzee MHC class I region and was likely caused by a retroviral infection (49, 50), may also have had an indirect impact on its ligands within the KIR gene region. Bonobos (Pan paniscus) and chimpanzees shared a common ancestor ∼2.3 million years ago. In this species, only seven KIR genes are reported, five and three genes of which are orthologs, shared with common chimpanzees and humans, respectively (51). The short bonobo KIR haplotypes, the limited KIR gene repertoire, and the reduced bonobo MHC class I content (52–55) may result from subsequent selective sweeps (56). The expansion of lineage III KIR genes, which in macaques are represented by KIR1D, correlates with the emergence of HLA-C–like genes in orangutans (57–59). In orangutans, 13 KIR genes are identified, of which the framework genes share orthologs with humans (15, 41, 57, 58). Two sibling orangutan species inhabit Sumatra (Pongo abelii) and Borneo (Pongo pygmaeus) (60). Only one gene is species-specific (KIR2DS15 in Bornean orangutans), whereas all other genes are orthologs. Ten of the 130 KIR alleles that were identified are shared between both sister species. For the human and great ape species discussed above, rapid evolution is mainly reflected by the gain in allelic variation, whereas the generation of novel gene entities and the formation of complex haplotype architectures seem to be relatively limited (Fig. 10).

The present communication sheds light on the evolution of the KIR region in two highly related Old-World monkey species, which share an introgression zone. In rhesus and cynomolgus macaques, the massive expansion of the lineage II KIR genes exceeds the modest lineage III expansion in great apes and humans. The rapid evolution of the macaque KIR region is reflected not only by allelic polymorphism, but even more prominently by the large number of species-specific recombinant genes and haplotypes with a complex architecture in different macaque populations. In humans and other hominids, these events seem to be less abundant. Moreover, the allelic variation in macaques seems to exceed the numbers that are encountered in humans and higher primates. We recorded a total number of 579 KIR alleles, and it is important to realize that the number of samples that were analyzed is relatively small as compared with the human situation (43). The reason for the extensive expansion of the KIR gene cluster in macaques is of interest. Whereas in humans, the less variable KIR gene content and haplotype configurations seem to be the result of a trade-off, such an indication appears to be absent in macaques. The most plausible driving forces of the rapid KIR cluster evolution in macaques might involve coevolution with the extended MHC class I region, differential infectious pathogen encounters, a discontinuous habitat, and susceptibility to chromosomal recombination. Nonetheless, we cannot rule out that the extensive expansion of the macaque KIR gene system may have evolved due to the lack of evolutionary pressure on this system. The ligand of only a few receptors have been identified and, therefore, the functional impact of the expanded macaque KIR repertoire remains largely unclear. However, diversification of ligand interactions is suggested by overlapping, but nonredundant, MHC class I specificity of multiple KIR (61–66). The extensive diversity of the macaque MHC and KIR clusters might facilitate interactions with allele-level specificity, differential affinity, and peptide dependency and may contribute to rapid adaption driven by environmental conditions.

The general high levels of allelic polymorphism detected in the KIR region in primates might indicate that it is more prone to generate mutations than other regions of the genome. Mutation rates are elevated in CpG islands, which are genomic regions that are enriched for CpG sites with an observed-to-expected ratio >60%. All KIR genes indeed carry CpG islands (67). CpG site mutations, however, mainly involve cytosine to thymine transitions, whereas mutations in the generation of the novel Mamu-KIR3DL05 allele involved T > C and G > T transitions, which are not commonly observed transition events. The two-point mutations are separated by only two nucleotides, which suggests that one mutation initiated the other and, perhaps, was caused by the recruitment of error-prone repair mechanisms (68). In addition to CpG islands and error-prone repair, other factors that might enhance the regional mutation rate may include recombination events, deletion and insertion events, chromatin configurations, distance to the telomere, and replication time (69, 70). Furthermore, relatively more single-nucleotide polymorphisms were observed in regions that were homologous in humans, chimpanzees, and macaques, which substantiates the extensive variation of the KIR gene region (69). The birth of novel KIR alleles has been described previously in human families (71), and together with the event recorded in macaques, this might suggest that point mutations substantially contribute to the extensive allelic KIR variation. Of course, it is clear that the generation of mutations is only one side of the coin, and that selection determines which polymorphisms will be enriched in the populations or are eventually rooted out.

In humans, genetic KIR variation is documented for over 250 populations and mainly records allelic variation and differential haplotype distribution in relation to gene frequencies (47, 72). Similar observations were made for the different macaque populations. Genes that are shared in all three rhesus or four cynomolgus macaque populations mainly involved inhibitory KIR genes (Figs. 8, 9). The conserved nature of these genes suggests an impact on essential functions, such as NK cell education, a process for which the involvement of inhibitory KIR is well established in humans (73–75). The role of activating KIR in humans is less understood, but associations with disease progression or protection are described, and in vitro studies demonstrate specific binding to certain peptide–MHC class I complexes (76–81). In macaques, activating KIR genes were mainly identified in one or two populations and may substantiate more specialized functions, like pathogen recognition (Figs. 8, 9). For instance, Mamu-KIR3DSW18 is encountered only in Burmese rhesus macaques, and a similar observation was made for Mafa-KIR3DSW08 in the northern-mainland cynomolgus macaques. In addition, the majority of the KIR alleles appear not to be shared at the population level (Fig. 4). Again, this hints at a speedy generation of allelic polymorphism. For the Indian rhesus macaque population, a genetic bottleneck is evident (82), but it did not result in a reduced KIR gene content, which might indicate that the rapid evolution of the KIR repertoire erased the genetic footprint of a bottleneck.

Rhesus and cynomolgus macaques are widely used as preclinical models in translational biomedical research to further a better understanding of human diseases and the development of vaccines and therapies (34, 35). The genetic makeup of the different macaque species, however, can vary considerably and might potentially influence the outcome of studies. Even at the population level, a differential disease susceptibility has been reported: for example, in SIV/AIDS-related experiments in rhesus macaques of Indian and Chinese origin (36, 37, 83). It is possible that the KIR repertoire may be one of the factors that have an impact on disease outcome, as correlations between KIR gene content and disease phenotypes in humans (6) and macaques are documented (38, 39).

This study design, including rhesus and cynomolgus macaque families from different geographical origin, allowed the transcriptomic characterization of the complex KIR cluster. The high level of allelic polymorphism, the number of novel gene entities, the plastic haplotype architecture, and the diversification at the species and population levels illustrate the unparalleled rapid evolution of the KIR gene region in macaques. This communication paves the way to study the impact of KIR genes in NHP models for human health and disease, but also may help in selecting animals with particular genetic markers for studies in the area of personalized medicine.

We thank D. Devine for editing the manuscript and F. van Hassel for preparing the figures.

The authors have no financial conflicts of interest.