Short KIR Haplotypes in Pygmy Chimpanzee (Bonobo) Resemble the Conserved Framework of Diverse Human KIR Haplotypes

Killer cell Ig-like receptors (KIRs) are expressed on NK cells and subsets of T cells, mostly CD8⁺, having activation or memory phenotype 1,2,3,4,5,6. KIR genes have been detected in several primate species but appear to be absent from rodents, including mice 7. In humans, the KIRs are encoded by a family of genes in the leukocyte receptor complex on chromosome 19 8,9,10,11,12,13. The products of these genes differ in having either two or three extracellular Ig domains and also in having either long cytoplasmic tails, associated with inhibitory signal transduction, or short tails associated with activating function 14,15,16. KIR haplotypes differ in the total number of KIR genes they contain (∼6–12) and in the relative number of genes encoding inhibitory versus activating KIRs 8,17,18. Some genes appear to be conserved features of KIR haplotypes, for example KIRCI (also called KIR3DL3), KIR2DL4, and KIR3DL2 8,17, whereas others are restricted to a subset of haplotypes, for example KIR2DL5 19. Certain human KIRs have specificity for polymorphic determinants of HLA-A, -B, or -C molecules. KIR2DL1, KIR2DL2, and KIR2DL3 are inhibitory, and KIR2DS1 and KIR2DS2 are activating receptors with HLA-C specificity 1,20,21; KIR3DL1 and KIR3DS1 are receptors with HLA-B specificity having inhibitory and activating function, respectively 22,23, and KIR3DL2 is an inhibitory receptor with HLA-A specificity 24,25. In addition, KIR2DL4 is reported to have specificity for HLA-G 26.

Population analysis and phylogenetic comparison have shown that MHC-A, -B, and -C genes evolve rapidly compared with most other genes 27,28,29,30,31. KIR genes can also evolve rapidly as shown by comparison of human and common chimpanzee KIR 32. A minority of KIR genes are conserved, whereas the majority have undergone substantial “species-specific” divergence in the ∼5 million years since chimpanzees and humans shared a common ancestor. In terms of the type and number of genes, the MHC class I gene family appears conserved in comparison to the KIR gene family: all the functional HLA class I genes have chimpanzee orthologs 31 whereas only three human KIR genes are in this category 32. Thus, from comparison of these two species, the KIR gene family is seen to have evolved faster than the MHC class I gene family. Whereas receptors of innate immunity have often been considered as being highly conserved 33,34, KIRs may provide an example where the opposite is true.

To investigate further this unusual phenomenon, we have now studied the KIR gene family of the pygmy chimpanzee (Pan paniscus), also called bonobo, a species that is estimated to have last shared an ancestor with the common chimpanzee (Pan troglodytes) some ∼2.3 million years ago 35. This study has therefore allowed an assessment of KIR divergence over a time period that is about half of that which separates humans and chimpanzees. The results highlight the evolutionary instability of the KIR gene family and have revealed a simple form of KIR haplotype that provides new insight into the basic requirement of the KIR system of NK cell receptors.

Peripheral blood was obtained from healthy chimpanzees housed at Yerkes Regional Primate Center at Emory University School of Medicine (Atlanta, GA) and at the Laboratory for Experimental Medicine and Surgery in Primates (LEMSIP) at New York University Medical Center (Tuxedo, NY). PBMCs were isolated on Ficoll-Hypaque gradients and used for the isolation of total RNA as well as for establishing EBV-transformed B lymphoblastoid cell lines.

Mitochondrial DNA typing indicated that 43 of the 48 common chimpanzees studied were of the subspecies P. troglodytes verus, 3 were of subspecies P. troglodytes troglodytes, and 2 were of subspecies P. troglodytes schweinfurthii 31,36. The individuals in this panel were chosen because they were either wild-born, or unrelated to other chimpanzees in the panel as documented by breeding records and supported by the analysis of MHC class I alleles 31,36. Within this panel, 30 different KIR genotypes are represented at relatively even frequency, of which the highest was 0.11 32. Attesting to the genetic heterogeneity within the common chimpanzee panel was that a similar sized panel of unrelated humans had 18 different genotypes, for which the most common had a frequency of 0.33 17.

Genomic DNA was isolated from B lymphoblastoid cell lines using standard methods as described by us previously 17. Genomic DNA from pygmy chimpanzees, common chimpanzees, and humans were digested with HindIII (Boehringer) and Southern blots were made using the same protocol we described previously for humans 17. The blots were hybridized with a ³²P-labeled cDNA probe encoding either a common chimpanzee KIR, Pt-KIR3DL6, or human KIR3DL1, and autoradiographed using standard protocols 37.

Total RNA isolated from PBMCs was used to synthesize first strand cDNA using previously described methods 17. KIR transcripts were PCR amplified from pygmy chimpanzee cDNA using methods that had worked for common chimpanzees and humans 32,38. The oligonucleotide primers are based upon conserved segments of human KIR sequences. The PCR products were purified from the reaction mixture using a QIAquick PCR purification kit (QIAGEN) and cloned into pCR4-TOPO vectors (Invitrogen) according to manufacturer's instructions. Partial sequences were determined on randomly picked clones using standard T7 or M13R primers and the BigDye terminator cycle sequencing kit (Applied Biosystems) in a 377 automated DNA sequencer (Applied Biosystems). Six different KIR sequences were distinguished and have been submitted to EMBL/GenBank/DDBJ under accession nos. AF258798 (Pt-KIR3DL1/2-v1), AF266729 (Pt-KIR3DL1/2-v2), AF266730 (Pt-KIR3DL1/2-v3), AF266731 (Pp-KIR3DL4), AF266732 (Pp-KIR3DLa), AF266733 (Pp-KIR3DLb), AF266734 (Pp-KIR3DLc), AF266735 (Pp-KIR3DSa), and AF266736 (Pp-KIR2DL4).

Using human KIRCI-specific primers (sense 5′-GCACTGTGGTGTCTGAAGGAC-3′, anti-sense 5′-GTAGCTCCCTCCGTGGGTCA-3′), fragments that cover exon-3 (D0 domain) through to exon-5 (D2 domain) were PCR amplified from pygmy chimpanzee Matata and common chimpanzee Alex. Primer sets designed to amplify the transmembrane region through the cytoplasmic tail of KIRCI failed in both pygmy and common chimpanzees. Using KIR2DL5-specific primers, based on human and common chimpanzee sequences, fragments covering exon-3 (D0 domain) through exon-4 (D2 domain; sense 5′-GGTGGTCAGGACAAGCCCTTG-3′, anti-sense 5′-GGTCTGACCACTCATAGGGT-3′), and exon-6 (transmembrane region) through exon-8 (cytoplasmic tail; sense 5′-TTTCTCCTTCATCGCTGC-3′, anti-sense 5′-ACCTCCTGAGGGTCTTGA-3′) were amplified from DNA of pygmy chimpanzee Matata who was typed positive by human KIR2DL5-specific primers. All PCR were carried out using the Expand Long Template PCR System (Boehringer) according to the manufacturer's instructions. The PCR conditions included 2 min initial denaturation at 94°C, 30 cycles of 20 s at 92°C, 30 s at 62°C, 8 min at 68°C, and final extension at 68°C for 10 min. PCR products were purified using the QIAEX II Gel extraction kit (QIAGEN) and the exon sequences were determined by direct sequencing.

Sequence alignments and pairwise comparisons were performed using the AutoAssembler, v2.1 (Applied Biosystems) and the Wisconsin sequence analysis software, v10.1 (Genetics Computer Group). Phylogenetic trees were constructed with PAUP 4.0b2a software (Sinauer Associates; available at http://www.sinauer.com/) using the maximum parsimony analysis 39 and neighbor-joining method 40. The level of confidence in each node of the tree was assessed from 1,000 replications by the bootstrap method 41.

Pygmy chimpanzee genomic and/or cDNA were PCR typed for 14 human KIRs and 10 common chimpanzee KIRs using the typing systems we developed previously 17,19,32. To type for KIRCI 42, which was not included in the previous typing system, an additional set of primers was included. A typing system was also developed for the nine pygmy chimpanzee KIR sequences defined in this study. Typing for common chimpanzee KIR was refined to type for Pt-KIRCI, and to distinguish the variants of Pt-KIR3DL1/2. The oligonucleotide primers and the size of the products expected in the DNA typing are as follows: Pt-KIR3DL1/2-v1: sense 5′-GTGATCCCCTGGACATCA-3′, anti-sense 5′-TGCAGGACAAGGTCACGC-3′, 1,700 bp; Pt-KIR3DL1/2-v2 and v3: sense 5′-GTGATCCCCTGGACATCA-3′, anti-sense 5′-TGACCTTGCGCACTGCAC-3′, 1,800 bp; Pp-KIR3DLa: sense 5′-ACATGCAGGGAACTACAC-3′, anti-sense 5′-GCGCAAAGTGTCCTCAAC-3′, 1735 bp; Pp-KIR3DLb: sense 5′-ATCCTCTTCTTTCTCCTTCATCA-3′, anti-sense 5′-GCTGCTGGTGCATTGGAT-3′, 945 bp; Pp-KIR3DLc: sense 5′-AACCCCAGACACCTACAT-3′, anti-sense 5′-TTCCGTGTACACGCTGGTG-3′, 865 bp; Pp-KIR3DSa: sense 5′-GTCAGTGGTCAAAATCCCTTTCAC-3′, anti-sense 5′-TCATGGTGTGAGGAAGAGCA-3′, 715 bp; Pp-KIR2DL4: sense 5′-TACAGATGTCGAGGTTTTCACCCG-3′, anti-sense 5′-TGTGGGGCCCCGCCGGGCTGTGAGT-3′, 900 bp; Pp-KIR3DL4: sense 5′-TAACGACACTTTGCGCCA-3′, anti-sense 5′-GAGCCTACGTTCATGAGA-3′, 1,575 bp; KIRCI: sense 5′-GGACCTACAGATGTTGC-3′, anti-sense 5′-TAGTTGACCTGGGAACCCG-3′, 1,575 bp; Pt-KIRCI and Pp-KIRCI: sense 5′-GGAACCTACAGATGTTGC-3′, anti-sense 5′-TAGTTGACCTGGGAACCCG-3′, 1,575 bp. Pp-KIR2DL5- and Pp-KIR3DL5-specific typing were performed using the primer sets developed for typing their common chimpanzee othologs 32. The PCR reaction mixture and the temperature conditions were the same as used for human and common chimpanzee KIR typing, with minor modification in the annealing temperature (61°C) used for the second set of cycles for primer sets specific to Pt-KIR3DL1/2 variants, Pp-KIR3DLa, Lb, and Lc 17,32. The amplified products from all pygmy chimpanzees and some common chimpanzee typing reactions were directly sequenced, confirming the fidelity of the typing reactions.

Genomic DNA from B cell lines derived from pygmy chimpanzees, common chimpanzees, and human controls were digested with HindIII and compared in Southern blotting using a common chimpanzee KIR cDNA probe (Fig. 1). All three species exhibit polymorphism in the KIR banding pattern, and in humans one such difference (the presence or absence of the ∼24-kb band numbered 1 in Fig. 1) has been correlated with differences in the number and type of KIR genes 17,19. However, the overall number of bands in the Southern blots of different humans and common chimpanzees is similar. Distinguishing the pygmy chimpanzee is the much larger extent of the differences between individuals in the Southern blot banding patterns. The number of bands varied from three (Bosondjo and Jill) through seven (Matata), with the latter pattern being of a complexity approaching that seen in humans and common chimpanzees.

Typing systems developed previously for the analysis of human and common chimpanzee KIR 17,19,32, and including an additional set of primers for human KIRCI 42, were used to type genomic DNA from 11 pygmy chimpanzee B cell lines, including the 6 analyzed by Southern blotting. The results are summarized in Fig. 2. 13 (8.4%) of the 154 typing reactions targeted at human KIR (Fig. 2 A), and 22 (20%) of the 110 typing reactions targeted at common chimpanzee KIR (Fig. 2 B), were positive. The relatively low frequency of positive reactions suggested that pygmy chimpanzee KIRs are considerably diverged from both human and common chimpanzee KIRs, although they are closer to the latter. All 11 pygmy chimpanzees typed with primers specific for Pt-KIR2DL4 and human KIRCI, whereas Pt-KIR3DL4 and Pt-KIR3DL5 were scored by four and five individuals, respectively. For each individual, the number of positive typing reactions roughly correlated with the complexity of the banding pattern in Southern blot (Fig. 1). Thus, Bosondjo and Jill, who had the simplest banding pattern, typed only for Pt-KIR2DL4 and KIRCI, whereas Matata, who had the most complicated banding pattern, typed for four different Pt-KIRs and two human KIRs: KIR2DL5 and KIRCI (Fig. 2A and Fig. B).

Our next goal was to characterize cDNA encoding pygmy chimpanzee KIRs. Because of the restricted quantities of pygmy chimpanzee blood available, cDNA was made from RNA isolated from the PBMCs of Lisala, a pygmy chimpanzee for which a B cell line had not been made and frozen PBMCs were available. Analysis of Lisala's cDNA with the typing reactions targeted towards common chimpanzee or human KIR gave results similar to those obtained for Matata's genomic DNA (Fig. 2A and Fig. B). The one difference was the absence of KIRCI in Lisala's cDNA. A similar lack of KIRCI transcription was obtained in all humans 42 and most common chimpanzees tested (Rajalingam, R., unpublished observations). Complementary DNA clones encoding pygmy chimpanzee KIR were obtained after PCR using primers based upon the conserved sequences of human KIR, an approach successfully used to isolate cDNA clones encoding common chimpanzee KIR 32. Partial nucleotide sequences were determined for 186 individual cDNA clones permitting them to be sorted into six different groups.

The most abundant KIR cDNA (42 clones) corresponds to the pygmy chimpanzee ortholog of KIR2DL4, which has 96.9 and 98.7% sequence similarity with human KIR2DL4 and common chimpanzee Pt-KIR2DL4, respectively. These relationships are apparent in a phylogenetic tree of chimpanzee and human KIR (Fig. 3). Consequently, this pygmy chimpanzee KIR has been named Pp-KIR2DL4, where Pp signifies P. paniscus. Pp-KIR2DL4 and Pt-KIR2DL4 have identical amino acid sequence in the extracellular domains and differ by just four amino acid substitutions elsewhere: two in the transmembrane region and two in the cytoplasmic tail.

The least abundant pygmy chimpanzee KIR cDNA was represented by a single clone from the initial screen. Three more clones corresponding to this KIR were obtained on screening a further 384 clones. This KIR is closely related to common chimpanzee Pt-KIR3DL4, with which it has 98% sequence homology, and is likely an ortholog (Fig. 3). The gene for this pygmy chimpanzee KIR, named Pp-KIR3DL4, was probably the target for the primers specific for Pt-KIR3DL4 and which were positive with Lisala and four other pygmy chimpanzees (Fig. 2 B). Pp-KIR3DL4 differs from Pt-KIR3DL4 by 19 amino acid substitutions. Of these, only three are in the extracellular Ig-like domains; seven being present in the transmembrane region, five in the cytoplasmic domain, two in the stem, and two in the leader. In the common chimpanzee, Pt-KIR3DL4 is an inhibitory receptor for C2 type of MHC-C allotypes having the asparagine 77, lysine 80 motif, and in human KIR this C2 specificity is correlated with the presence of methionine at position 44 of the D1 domain 43,44. Pp-KIR3DL4, like Pt-KIR3DL4, has methionine at this position, raising the possibility that it too is an inhibitory C2 receptor. Excepting Lisala, all the pygmy chimpanzees studied here have been characterized previously for alleles of Papa-C, the ortholog of the human HLA-C locus 36. All the Papa-C alleles in this cohort of animals encode heavy chains having the asparagine 77, lysine 80 motif, C2-type motif.

Most of the cDNA clones (143 clones) from the initial screen were shown to represent four KIR3D, which differ from one another by 2–6% of the nucleotide sequence. They belong to the lineage of 3Ig KIR that in humans and common chimpanzees embraces the inhibitory receptors for MHC-A (KIR3DL2, Pt-KIR3DL1/2) and MHC-B (KIR3DL1, Pt-KIR3DL1/2) allotypes (Fig. 3 and Fig. 4). In comparison with complete coding region sequences, the four pygmy chimpanzee KIR3D have 91–98% sequence similarity with KIR3D of the corresponding human and common chimpanzee lineage. However, none of the four pygmy chimpanzee KIRs appears orthologous to either a common chimpanzee or human KIR (Fig. 3 and Fig. 4), and for this reason we have provisionally designated them as Pp-KIR3DLa, Pp-KIR3DLb, Pp-KIR3DLc, and Pp-KIR3DSa.

To considerable extent the pygmy chimpanzee KIR3D consists of sequence elements present in human and common chimpanzee KIR, but in novel combination (Fig. 5 A). In the region encoding the extracellular part of the molecule (Ig domains and stem), the four Pp-KIR3D form a clade with the common chimpanzee Pt-KIR3DL3 (Fig. 5 B). Comparison of the sequences encoding individual Ig-like domains revealed the D0 domains of Pp-KIR3DLa and Pp-KIR3DSa to be identical and to share sequence segments with both human KIR3DL1 and KIR3DL2. The D0 domains of Pp-KIR3DLb and Pp-KIR3DLc are also identical and have 97.2% sequence similarity with the corresponding domain of human KIR3DL2. In the D1 domain, Pp-KIR3DSa groups with Pp-KIR3DLb and Pp-KIR3DLc, and these closely related sequences have 96% sequence similarity with the D1 domain of KIR3DL2. In the D1 domain, Pp-KIR3DLa diverges from the other Pp-KIR3D but has 98% sequence similarity with the D1 domain of common chimpanzee Pt-KIR3DL3. In the D2 domain, the four Pp-KIR3D are very similar to each other and to Pt-KIR3DL3 (98.5% sequence similarity). Unique to the carboxyl terminal half of the D2 domain of these five KIRs is an insertion of two amino acids followed by a motif of five amino acid substitutions (Fig. 4).

In the transmembrane region and cytoplasmic tail, the relationships between the four Pp-KIR3D (Fig. 5a and Fig. C) differ from those in the extracellular domains (Fig. 5a and Fig. B). Here, Pp-KIR3DLa and Pp-KIR3DLb are like each other, and similar to the common chimpanzee Pt-KIR3DL1/2 and Pt-KIR3DL3 (and to a lesser extent with human KIR3DL2). In contrast, Pp-KIR3DLc is most closely related to KIR3DL-4M1#6, a divergent human KIR3DL sequence (sequence data are available from GenBank/EMBL/DDBJ under accession no. X97230), whereas Pp-KIR3DSa is closest to human KIR3DS1. The latter two short-tailed KIR3D share the KxPxT transmembrane motif (amino acids 330–334) which is characteristic of all human short tailed KIR. The lysine residue in this motif binds to cosignaling molecules such as DAP-12 45. In the long-tailed KIR3D, the sequence and spacing of the immunoreceptor tyrosine-based inhibitory motifs are conserved in Pp-KIR3DLa, Pp-KIR3DLb, Pp-KIR3DLc, Pt-KIR3DL1/2, Pt-KIR3DL3, and human KIR3DL2 (Fig. 4)

To search for additional pygmy chimpanzee KIR, cDNA clones were derived from the PBMCs of Bosondjo and Matata. These two animals are those with the simplest (Bosondjo) and most complicated (Matata) KIR types as assessed by Southern blot (Fig. 1). No novel KIR was identified. From Bosondjo, only clones corresponding to Pp-KIR2DL4 (47 clones) and Pp-KIR3DLb (13 clones) were obtained, consistent with the typing of this individual (Fig. 2). From Matata, clones corresponding to Pp-KIR2DL4 (33 clones), Pp-KIR3DLa (7 clones), Pp-KIR3DLb (28 clones), Pp-KIR3DLc (1 clone), and Pp-KIR3DSa (6 clones) were obtained, again consistent with the typing.

The positive typing reactions of pygmy chimpanzee genomic DNA for Pt-KIR3DL4 and Pt-KIR2DL4 (Fig. 2 B) were explained by the identification of pygmy chimpanzee KIR cDNA corresponding to these genes. In contrast, the genomic typing reactions with primers specific for KIR2DL5, Pt-KIR2DL5, KIRCI, and Pt-KIR3DL5 could not be matched to any of the pygmy chimpanzee cDNA sequences.

We therefore performed further analysis of genomic DNA from Matata, who typed positively with primers for KIR2DL5 plus Pt-KIR2DL5, KIRCI, and Pt-KIR3DL5, to identify these genes. Contiguous sequences for the coding regions of pygmy chimpanzee orthologs of the KIR2DL5 and KIRCI genes were obtained from PCR-generated fragments. Pp-KIR2DL5 exhibits 97.0 and 99.2% sequence similarity with human and common chimpanzee KIR2DL5, respectively. Pp-KIRCI shows comparable levels of sequence similarity with human (97.6%) and common chimpanzee (98.5%) KIRCI in the sequence encoding the extracellular region. Attempts to amplify sequences corresponding to the transmembrane and cytoplasmic regions of Pp-KIRCI with primers based on either the human or common chimpanzee KIRCI sequences were unsuccessful. In this 3′ part of the gene, Pp-KIRCI may be more divergent from the human and common chimpanzee genes than the 5′ part encoding the extracellular domains.

Although various sets of primers were designed to amplify pygmy chimpanzee sequences related to Pt-KIR3DL5, only those used in the initial typing analysis (Fig. 2 B) gave a pygmy chimpanzee product. This gene fragment gave 423 nucleotides of sequence encoding the D1 and D2 domains. In this sequence, Pp-KIR3DL5 differs by three nucleotide substitutions from the corresponding Pt-KIR3DL5. The 99.3% sequence similarity of these sequences is consistent with them being derived from orthologous genes.

From the pygmy chimpanzee KIR sequences, a PCR typing system was developed and the genomic DNA of the 11 pygmy chimpanzees were analyzed (Fig. 2 C). They all typed positively for three genes, Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb. In the Southern blot, the only HindIII band present in all individuals of the three species is the one numbered 6 in Fig. 1. Thus, this ∼6-kb band is a candidate for containing the KIR2DL4 and/or the KIRCI gene. Typing for Pp-KIR3DL4 revealed that only 5 of 12 pygmy chimpanzees have this KIR (Fig. 2 C). Bosondjo and Jill, who both have the simplest Southern blot pattern, were among those lacking Pp-KIR3DL4 (Fig. 1 and Fig. 2 C). Typing for the presence of the four Pp-KIR3D related to KIR3DL2 revealed three individuals having one Pp-KIR3D, four having two, three having three, and two having all four of them (Fig. 2 C).

Bosondjo, Jill, and Zalia have only Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb, consistent with Bosondjo and Jill having the simplest banding pattern in Southern blot (Fig. 1; Zalia was not analyzed by blot). Kitty has Pp-KIR3DLa and Pp-KIR3DL4 in addition to Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb, correlating with an additional ∼15-kb band in the blot (band 3 in Fig. 1). Brian has Pp-KIR3DLc and Pp-KIR3DL5, in addition to Pp-KIR3DLa, Pp-KIR3DL4, Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb, correlating with additional bands of ∼20 and ∼7 kb in the blot (bands 2 and 5 in Fig. 1, respectively). Finally, the additional presence of Pp-KIR3DSa and Pp-KIR2DL5 in Matata is associated with an additional band of ∼11 kb in the blot (band 4 in Fig. 1).

Nine of the pygmy chimpanzees are members of a family within which we could trace the segregation of KIR and infer possible haplotypic associations (Fig. 6 A). A minimum of six different haplotypes is required to explain the observed genotypes. The most common haplotype in the family (labeled “a” in Fig. 6 A) appears to contain just three genes: Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb with Bosondjo, Jill, and Zalia being homozygous for this haplotype (Fig. 6 A). Less frequent are five haplotypes containing two Pp-KIR3D: Lb plus La, Lc plus La, and Lc plus Sa; data raising the possibility that Lb and Lc are alleles of one locus, with La and Sa being alleles of a second locus. The results demonstrate the existence of two genes encoding Pp-KIR3D with some haplotypes having both genes and others only one. This is analogous to the human situation, where some haplotypes have KIR3DL1 and KIR3DL2 and others have only KIR3DL2 8.

In the common chimpanzee, four Pt-KIR3D have also been found to be part of the KIR3D lineage that includes all genes encoding KIR specific for MHC-A and -B 32. Three of these (Pt-KIR3DL1/2v1, Pt-KIR3DL1/2v2, and Pt-KIR3DL1/2v3) are closely related variants having extracellular regions similar to human KIR3DL1 and transmembrane and cytoplasmic regions similar to human KIR3DL2 (Fig. 4). The fourth common chimpanzee KIR3D (Pt-KIR3DL3) diverges from Pt-KIR3DL1/2 in the D1 and D2 domains, where we now see that it has sequence similarity with the pygmy chimpanzee KIR3DL (Fig. 3,Fig. 4,Fig. 5). DNA typing of a panel of 48 unrelated common chimpanzees shows that the three variants of Pt-KIR3DL1/2 and Pt-KIR3DL3 segregate as alleles of one locus (Table, and Fig. 6 B). Thus, no chimpanzee with Pt-KIR3DL3 carries two variants of Pt-KIR3DL1/2, whereas many individuals who lack Pt-KIR3DL3 are heterozygous for Pt-KIR3DL1/2 variants. Thus, in common chimpanzee there is a single gene encoding this lineage of KIR3DL, a situation contrasting with that in the pygmy chimpanzees and humans, where KIR haplotypes having two genes in this lineage are both present and frequent.

This study has examined the KIR gene family of the Pygmy chimpanzee or Bonobo (P. paniscus) and compared it to that of its closest relative, the common chimpanzee (P. troglodytes), and its second most close relative, the human species (Homo sapiens). We characterized sequences for nine pygmy chimpanzee KIRs (Pp-KIR), representing at least seven different genes. Each Pp-KIR belongs to one of the three lineages of KIR defined from study of human and common chimpanzee KIR and all three lineages are represented in the pygmy chimpanzee KIR 15,32,42. All three species have orthologs for KIRCI, KIR2DL4, and KIR2DL5, and their pairwise comparison indicates that pygmy chimpanzee KIR are more similar to common chimpanzee KIR than human KIR, consistent with the estimated separations from these species of ∼2.3 and ∼5 million years, respectively 35. Also supporting this hierarchy is the presence of pygmy chimpanzee orthologs (Pp-KIR3DL4 and Pp-KIR3DL5) for two common chimpanzee KIRs (Pt-KIR3DL4 and Pt-KIR3DL5), for which there are no human orthologs.

The two pygmy chimpanzee KIR genes without obvious orthologs in either common chimpanzees or humans are those of the KIR3D lineage that in common chimpanzees and humans is characterized by genes encoding receptors specific for MHC-A and MHC-B 22,23,24,25,32. In common chimpanzee, this lineage is represented by a single gene (Pt-KIR3DL1/2) which encodes a receptor that binds both MHC-A and -B allotypes, whereas in humans there are two genes, KIR3DL1 that encodes an MHC-B receptor and KIR3DL2 that encodes an MHC-A receptor. All the evidence points to this lineage of KIR genes having undergone much recombination, both to change the number of genes as well as to produce allelic variation. Thus, it is difficult without direct comparison of haplotype sequences to discern whether particular pygmy chimpanzee, common chimpanzee, and human genes in this KIR3D lineage have orthologous or paralogous relationships. Within this lineage, the pattern of polymorphism within a species and of species-specific divergence is consistent with coevolution of these KIRs with MHC class I polymorphism.

The striking feature that distinguishes the pygmy chimpanzee KIR system from its common chimpanzee and human counterparts is the presence of small KIR haplotypes and their preponderance in the cohort of animals we studied. These small Pt-KIR haplotypes give relatively simple patterns on Southern blotting, three bands, and appear to contain just three KIR genes: Pp-KIRCI, Pp-KIR2DL4, and Pp-KIR3DLb. Of importance is that each of these genes is either orthologous or paralogous to one of the three genes that is a conserved component of otherwise divergent human KIR haplotypes, and which Wilson et al. have called “framework genes” 8. Thus, Pp-KIR2DL4 and Pp-KIRCI are orthologous to human KIR2DL4 and KIRCI (also designated KIR3DL3) and Pp-KIR3DLb is of the same lineage and most closely related to KIR3DL2. In the ∼100-kb human KIR gene family, the KIRCI and KIR3DL2 genes define the two ends, and the KIR2DL4 gene is placed in the middle. In each of the two intervals defined by this framework the two human haplotypes sequenced by Wilson et al. 8 differ in having between one and five genes.

3 of the 12 pygmy chimpanzees studied (Bosondjo, Jill, and Zalia) are homozygotes for short Pp-KIR haplotypes. Bosondjo is the father of five of the other animals and by the criteria of fecundity in captivity, homozygosity for short KIR haplotypes does not seem an impairment. Similarly, in terms of health and longevity the three animals homozygous for the short haplotype do not appear compromised in comparison to the others. In conclusion, the short KIR haplotypes appear to provide the minimal essential functions of the KIR system that are needed to generate functional NK cells, live, survive, and reproduce. Whether the preponderance of the short Pp-KIR haplotypes seen in the animals studied here is representative of the natural situation is uncertain and difficult to assess because of the very small numbers of animals in captivity and the endangered status of the species.

Comparison of the KIR gene families in the two chimpanzee species and humans indicates that KIRCI, KIR2DL4, and a KIR3D gene have been conserved as framework genes of KIR haplotypes since divergence of the human and chimpanzee lines ∼5 million years ago 35. The three genes of the short haplotypes represent all three KIR lineages, and include ones encoding receptors for nonclassical and classical MHC class I (KIR2DL4 is a receptor for MHC-G 26, and KIR3D includes receptors for MHC-A and -B [23–25]). For KIRCI, neither its function nor its pattern of expression have been defined 42. KIRCI is in the lineage of KIR that includes those human and common chimpanzee KIRs that have specificity for MHC-C determinants 32. This lineage of KIR appears to have been the most rapidly evolving during the last 5 million years, a possibility being that KIRCI was the first gene of this type and that other members of the lineage are derivatives of it.

We have been unable to assess functionally the MHC class I specificity of pygmy chimpanzee KIR because of the small quantities of pygmy chimpanzee blood available. However, some inferences as to the possible receptor specificities can be made from structural comparison with human and common chimpanzee KIR. Based on their phylogenetic conservation, Pp-KIR2DL4 is a candidate MHC-G receptor and Pp-KIR3DL4 a candidate receptor for the C2 MHC-C specificity. By analogy with their paralogs in the other species, Pp-KIR3DLa, Pp-KIR3DLb, Pp-KIR3DLc, and Pp-KIR3DSa are candidates for MHC-A and -B receptors. In the D1 domain, Pp-KIR3DLa is distinguished from the other Pp-KIR3DL by several residues (E21, D48, T49, E54, and H55) which it shares with human KIR2DL2. In the crystallographic structure of the complex of KIR2DL2 with HLA-Cw3, these residues contribute to the interaction surface 46, raising the possibility that Pp-KIR3DLa may have affinity for MHC-C allotypes with the C1 motif. No Papa-C alleles encoding the C1 motif have been found in the pygmy chimpanzees studied here 36, but the small number of animals does not mean that such allotypes are not present in the population at large.

The similarities in the KIR in the two chimpanzee species serve to emphasize how different they both are from human KIR. First, a major component of the human KIR family is a set of KIR2D with D1 plus D2 configuration, which are related to chimpanzee KIR3D of a lineage different from that containing MHC-A– and -B–specific KIRs, and which have exons 3 that are not used 8,12,47,48. The number of these KIR2Ds is much reduced in common chimpanzee and we have no evidence for such KIRs in pygmy chimpanzee, although they were deliberately sought 32. Because of the number of pygmy chimpanzees studied (12 individuals), we cannot rule out that such genes do not exist in this species, but genes encoding KIR2D of D1 plus D2 configuration appear to be represented at low frequency in common chimpanzee and none of them are invariant components of human KIR haplotypes, although all haplotypes have at least one of them 8,17,18,32. Second, the numbers of activating KIRs, as assessed from the size and sequence of the cytoplasmic tail, is considerably greater in humans compared with either chimpanzee species 15,32. For pygmy chimpanzee only one activating receptor has been defined, Pp-KIR3Sa.

The presence of KIR genes in primates and their absence in rodents led to the hypothesis that the KIR gene family is of recent origin and perhaps specific to the primates 7,49. This view is supported by the presence within the KIR region of Alu sequences that are mostly of a type that originated only 31–55 million years ago 8. As for other multigene families 50,51, the modern KIR gene family is envisioned to have originated with duplication of a single gene followed by successive expansions in gene number. Thus, in these formative times there was a trend in which the size of the KIR family increased. Accordingly, it is possible that all the haplotypes now present in chimpanzees and humans are derived from an older form of haplotype containing just the three framework genes and that the short haplotypes present in the pygmy chimpanzee retain this ancestral configuration. The genomic structure of the modern KIR gene family is unusual in that the genes are closely juxtaposed and separated by short homologous sequences 8. As well as reciprocal recombination, this arrangement is particularly favorable for unequal crossing over, a process that can delete, expand, and hybridize members of the gene family. The evidence that such mechanisms are active is the extent and type of diversity seen in both human and chimpanzee KIR haplotypes 17,18,32,52,53. Thus, it is alternatively possible that the short haplotypes present in the pygmy chimpanzee are derived from more complicated haplotypes in which inessential genes were deleted by unequal recombination.

We thank the staff at the Yerkes Regional Primate Center and those at the Laboratory for Experimental Medicine and Surgery in Primates (LEMSIP) for their assistance in obtaining chimpanzee samples.

This work was supported by National Institute of Health grants AI31168 and AI22039 (to P. Parham).