Comparison of Chimpanzee and Human Leukocyte Ig-Like Receptor Genes Reveals Framework and Rapidly Evolving Genes¹

The leukocyte receptor complex or cluster (LRC)⁶ occupies about 1 Mb of human chromosome 19q13.4 (1, 2, 3, 4, 5). It contains families of genes that encode receptors in which the extracellular domains are made up of Ig-like domains. One family of genes is the killer cell Ig-like receptors (KIRs) expressed by NK cells (6, 7) and subpopulations of T cells (8, 9). Adjacent to the KIR genes on their centromeric side is a distinct, but related, gene family, members of which were independently discovered by different investigators and given different names: leukocyte Ig-like receptors (LIR) (10, 11), Ig-like transcripts (ILTs) (12), and monocyte Ig-like receptors (MIRs; Table I) (13). Members of this second gene family encode receptors with two or four C2-type Ig domains; most are expressed by monocytes, macrophages, and dendritic cells, but some are also found on NK, B, and T cells (14, 15). In the absence of a single, agreed nomenclature, we shall use LIR here as a general descriptor for the gene family and the receptors they encode as it more accurately describes the proteins and their tissue distributions than does ILT or MIR.

Characterization of cDNA sequences combined with genomic analysis has shown the human LIR gene family consists of 11 expressed genes and two pseudogenes (4, 5). They form two homologous blocks of genes that are in opposite orientations and separated by ∼200 kb (4). In accordance with the convention adopted in the human genomic analyses, 11 of the genes are named ILT1–11, the other two are named LIR6 and LIR8 (4, 5). A minor LIR gene haplotype is found lacking the expression of ILT6 (5, 16).

Within both the LIR and KIR families are subsets of receptors that have the potential for either activating or inhibitory function. The long cytoplasmic tails of inhibitory receptors contain immunoreceptor tyrosine-based inhibitory motifs (ITIMs) that interact with the tyrosine phosphatase SH2-containing phosphatase-1 to inhibit cell activation (17, 18). Activating receptors have short cytoplasmic tails that lack ITIMs, with their activating function being determined by the presence of a charged residue in the transmembrane domain. Such residues allow association with signaling adaptor molecules that possess immunoreceptor tyrosine-based activation motifs. For activating KIR the adaptor molecule is DAP-12 (19), whereas for activating LIR it is FcεRIγ (20).

All known ligands for KIR and LIR are MHC class I or class I-like molecules (7, 14). Different KIR genes have been shown to encode receptors for HLA-A, -B, -C, and -G, and they account for approximately half of the KIR gene family (7, 14). By contrast, only LIR1 and LIR2 are known to bind MHC class I, which they do with a broad specificity, encompassing products of both classical and nonclassical HLA class I genes (17, 21, 22, 23, 24). For other members of the LIR family, the functions and ligand specificities remain unknown.

Population studies and species comparisons have shown that MHC class I genes can be highly polymorphic and rapidly evolving. Analogous studies of the KIR gene family have revealed similar characteristics. Within the human population, KIR haplotypes vary in the number and functional type of KIR genes (25, 26), and certain of the genes are also highly polymorphic (27, 28, 29). However, the LIR gene cluster appears more conserved in its organization (30). A comparison of common chimpanzee (Pan troglodytes) and human KIRs revealed that although the two species had similar numbers of KIR genes, a minority of them appeared orthologous (31). To examine the recent evolution of the LIR gene family, we have characterized common chimpanzee LIR genes and compared them to their human counterparts.

A frozen spleen sample from the common chimpanzee (Pan troglodytes verus) Amanda was macerated with scalpels, strained through a 70-μm pore size cell strainer, and resuspended in RNAzol B (Tel-Test, Friendswood, TX). Total RNA was isolated from the lysate following the manufacturer’s recommended protocols. mRNA was purified using a poly(dT) cellulose column from the Poly(A) Quik mRNA Isolation kit following the manufacturer’s instructions (Stratagene, La Jolla, CA). Genomic DNA samples were prepared from 48 B lymphoblastoid cell lines of unrelated chimpanzees (31, 32, 33) following standard protocols (34). Samples of chimpanzee spleen and peripheral blood were purchased from the Yerkes Regional Primate Center at Emory University (Atlanta, GA).

A common chimpanzee spleen cDNA library was constructed with the ZAP Express Gigapack III Gold kit (Stratagene) according to the manufacturer’s instructions and screened with a human ILT2 cDNA probe following standard procedures (34). Briefly, cDNA was synthesized from 5 μg mRNA, cloned into the ZAP Express λ phage vector, packaged, and used to infect the XL1-Blue strain Escherichia coli (Stratagene). For screening, 2 × 10⁶ PFU of the primary cDNA library were plated at a density of ∼1 × 10⁵ plaques/150-mm petri dish and transferred in duplicate onto Colony/Plaque Screen nylon membranes (NEN Life Science Products, Boston, MA). The membranes were hybridized under low stringency conditions (in 30% formamide, 5× standard saline sodium phosphate-EDTA (SSPE) buffer (pH 7.4), 5× Denhardt’s reagent, 5% dextran sulfate, 1% SDS, and 1 mg/ml fragmented and denatured salmon sperm DNA at 42^οC for 16 h) with a human ILT2 cDNA probe, washed four times with 2× SSPE/0.5% SDS at 42^οC, and autoradiographed. Positive phage were plaque-purified and subjected to an in vivo excision procedure according to a Stratagene protocol; the resulting phagemids contained chimpanzee cDNA within the pBK-CMV vector.

Nucleotide sequences were determined on both DNA strands using the Big-Dye Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA) and an ABI 377 DNA sequencer (Applied Biosystems). Complete DNA sequences of the plasmid inserts were obtained by using T3 and T7 universal oligonucleotide primers and internal primers based on sequences of chimpanzee LIRs. DNA sequences were assembled and analyzed using the computer program AutoAssembler (version 2.1; Applied Biosystems) and the Wisconsin Package sequence analysis software (version 10.1; Genetics Computer Group, Madison, WI) (35).

Individual cDNA sequences in the human LIR family have been named, mainly using two different nomenclatures: either ILT (12) or LIR (10, 11), followed by a number in series (Table I). These alternative and competing nomenclatures have yet to be reconciled into a single system. In human genomic analyses a hybrid nomenclature is being used (4, 5) in which the preference is for the ILT nomenclature (11 of the genes are named ILT1–11), but LIR nomenclature is retained for two genes (LIR6 and LIR8, first described by Cosman et al., who introduced the LIR nomenclature) (10, 11). Common chimpanzee LIR genes that are clearly orthologous to a human LIR gene have been given the same names as the human genes with the additional prefix Pt- for Pan troglodytes. Chimpanzee LIR genes that are paralogous to human LIR genes or for which orthology is less certain have been given provisional designations using lowercase letters: Pt-LIRa-e. The chimpanzee cDNA sequences reported here were deposited into the GenBank database (Table I).

We performed 5′ RACE analysis to obtain the 5′ coding region sequences of certain incomplete chimpanzee LIR cDNAs. First-strand cDNA was synthesized using 1 μg mRNA isolated from the chimpanzee Amanda with the SMART RACE kit according to the manufacturer’s instructions (Clontech, Palo Alto, CA). Two specific reverse oligonucleotide primers were designed for each chimpanzee LIR gene based on sequences determined for partial cDNAs isolated from library screening. These oligonucleotide primers were used for amplifying the 5′ cDNA of desired genes in two rounds of PCR amplifications. The PCR conditions were 94^οC for 30 s; 25 cycles of 94^οC for 5 s, 66^οC for 10 s, and 72^οC for 3 min; and 72^οC for 3 min for the initial reactions and 94^οC for 30 s; 15 cycles of 94^οC for 5 s, 65^οC for 10 s, 72^οC for 3 min; and 72^οC for 3 min for the second-round nested amplifications. Antisense oligonucleotide primers used in the initial amplification were 5′-GGG CTC AGA TCA CAG GAC TCA CG-3′ (Pt-LIRc), 5′-CGG GCA TGG GAA TGG GAG TTC AGA C-3′ (Pt-LIRd), and 5′-CTG TCG GTC AGG GCG CTG GGC G-3′ (Pt-ILT7). Antisense primers used in the second-round nested PCR were 5′-CCA CAC CTG GTC GTT GTA AGT AT-3′ (Pt-LIRc), 5′-CTT GCG TGT TCC CAG GTG ATG GAC G-3′ (Pt-LIRd), and 5′-GTC AGA TTC TCT CCG GGG GTC ACA A-3′ (Pt-ILT7). Amplified cDNA fragments were separated by electrophoresis in 1% agarose gel, purified using the QIAEX II gel extraction kit (Qiagen, Chatsworth, CA), cloned into the pCR-TOPO plasmid vector (Invitrogen, Carlsbad, CA), and sequenced.

Phylogenetic analyses were performed separately with nucleotide sequences encoding for the entire mature proteins and for only the Ig-like domains, using the PAUP 4.0 (Phylogenetic Analysis Using Parsimony) software package (version 4.0b4a, Sinauer Associates, Sutherland, MA) and other methods. Phylogenetic trees were constructed using the neighbor-joining method (36), and confidence values for each tree branch were estimated from 1000 replicates using the bootstrap technique (37). Gaps were counted as single mutational events in pairwise comparisons. Initial analysis to assess recombination between LIR genes was performed using the Partimatrix program (38), including all chimpanzee and human LIR gene sequences. Further analyses were made with groups of four sequences (two each from chimpanzee and human) and also with clusters of LIR sequences identified from phylogenetic analyses.

Rates of d_N (nonsynonymous) and d_S (synonymous) substitutions were determined using the synonymous/non-synonymous analysis program (SNAP, available at http://hiv-web.lanl.gov/SNAP/WEBSNAP/SNAP.html) based on the method of Nei and Gojobori (39) and incorporating a statistic developed by Ota and Nei (40). This analysis was performed separately on complete LIR mature protein-coding sequences, on sequences encoding all Ig-like domains, on individual Ig-like domains, or on contiguous small regions (60 nt) of each Ig-like domain. Each chimpanzee LIR gene was compared with the most closely related human LIR sequence.

To type for the presence of LIR genes in a panel of 48 unrelated chimpanzees, we designed two pairs of gene-specific PCR oligonucleotide primers for each of the nine chimpanzee LIRs that anneal to exon sequences encoding the first and second Ig-like domains (Table II). PCRs were conducted with ∼100–200 ng genomic DNA, 25 pmol of each primer, 2.5 mM of each dNTP, 1.5 mM MgCl₂, 1× AmpliTaq buffer (Applied Biosystems), and 1 U AmpliTaq DNA polymerase (Applied Biosystems). The PCR parameters were 94^οC for 10 s; 30 cycles of 94^οC for 40 s, 68^οC for 30 s, and 72^οC for 30 s; and 72^οC for 10 min. Amplified genomic DNA fragments were analyzed on ethidium bromide-stained 1% agarose gels. Typing reactions were tested on genomic DNA of the common chimpanzee Amanda from whom the spleen cDNA library was constructed; the PCR-amplified DNA were directly sequenced to confirm specificities. Subsequently, genomic DNA from a panel of 48 unrelated chimpanzees were typed, including individuals from three common chimpanzee subspecies: P. troglodytes schweinfurthii, P. troglodytes troglodytes, and P. troglodytes verus.

A cDNA library was made from spleen cells of an individual chimpanzee and screened with a human ILT2 cDNA probe. The screen yielded 83 hybridizing cDNA clones, which were further characterized by DNA sequencing using the T3 and T7 phagemid-based oligonucleotide primers. On the basis of partial sequences in the 5′ and 3′ ends of the cDNAs, the clones were sorted into nine groups, of which four were well represented. The other five groups were represented by only one clone each, three of these being incomplete clones lacking the 5′ end. Of the 83 clones 41 appeared to be full-length cDNA clones; the remaining 42 clones appeared either as partial clones lacking 5′ regions or products of alternatively spliced variants. A total of 44 cDNA clones representing all nine groups were completely sequenced. For the three groups represented by only single partial clones, complete coding region sequences were characterized using 5′ RACE analysis. All nine sequences were novel, but clearly belong to the gene family defined by the human LIR sequences.

Eight of the nine chimpanzee LIR genes encode proteins with four extracellular Ig-like domains (D1–D4); the ninth encodes a protein with two Ig-like domains. This strong bias toward receptors with four Ig-like domains parallels that seen in the human LIR family (14). The positions of cysteine residues that make up the intradomain disulfide bonds of the Ig-like domains are conserved in chimpanzee and human. Aligning and comparing the chimpanzee and human LIR as a group reveals sequence variability of ∼78% in D1, ∼72% in D2, ∼49% in D3, and ∼64% in D4. Variability is well spread within all Ig-like domains, although D3 appears to the most conserved (data not shown).

To identify affinities between individual chimpanzee and human LIR, we performed a systematic pairwise comparison of amino acid sequences. Four of the chimpanzee sequences had particular affinity with one human sequence, consistent with them being pairs of orthologous genes (Fig. 1 and identified by arrows). Within each of the four orthologous pairs, the amino acid sequences were <5% different, whereas differences between the four chimpanzee orthologs and other human paralogs were >17%. The human LIR shown to have chimpanzee orthologs were ILT1, ILT3, ILT7, and LIR8; we therefore gave these chimpanzee LIR the same name as their human orthologs and also the prefix Pt- for Pan troglodytes. For the other five chimpanzee LIRs, the patterns of pairwise difference are distinct from those seen for the four with obvious human orthologs (Fig. 1). There is a more continuous range of differences, in which no single human LIR is as closely related as seen with the orthologous pairs. To indicate this difference, these chimpanzee LIRs have been given names that do not correspond to human LIRs: Pt-LIRa, -b, -c, -d, and -e. These five chimpanzee LIRs can be further divided into two groups according to the groups of human LIR to which they are more closely related. Thus, Pt-LIRa, -b, and -c are more closely related to human ILT1, -2, -4, and -6 and LIR6, whereas Pt-LIRd and -e are closer to human ILT5, -8, and -9 (Figs. 1 and 2). These affinities could reflect orthologous or paralogous relationships. None of the chimpanzee LIRs shows particular affinity with ILT11 or the pseudogene ILT10 (Fig. 2). Within the population of cDNA clones we sequenced, the relative abundance of chimpanzee LIRs were: 32% for Pt-LIRa, 32% for Pt-LIRb, 14% for Pt-LIR8, 12% for Pt-ILT1, and 2% each for Pt-ILT3, ILT7, LIRc, LIRd, and LIRe.

The similarities and differences in the chimpanzee and human LIR proteins are depicted in the schematic comparison of Fig. 2. The orthologous LIRs include two with four Ig-like domains and short cytoplasmic tails (Pt-ILT1, Pt-ILT7), one with four Ig-like domains and a long cytoplasmic tail (Pt-LIR8), and one with two Ig-like domains and a long cytoplasmic tail (Pt-ILT3). Each group of paralogous LIR contains members with long and short cytoplasmic tails (Fig. 2, B and C). All short-tailed chimpanzee LIR contains an arginine residue in the transmembrane region homologous to those found in human short-tailed LIRs and which is implicated in association with an activating adaptor molecule. The three or four ITIMs present in the long-tailed chimpanzee LIR could be important for inhibitory function. The chimpanzee LIRs with three ITIMs (Pt-ILT3, Pt-LIR8, Pt-LIRb) lack either the first or the second ITIM compared with human and chimpanzee LIRs that have four ITIMs. With one exception, the first pair of ITIMs has YxxV motifs, and the second pair has YxxL motifs; the exception is the last ITIM of the chimpanzee Pt-ILT3, which has a YxxV motif. Pt-ILT3 also has a deletion in the cytoplasmic tail that is not present in its human ortholog, but is identical with that present in human ILT4 and LIR8. Sequence analysis by PCR amplification of this region of the Pt-ILT3 gene showed the presence of a pseudoexon that encodes the missing part of the cytoplasmic tail (data not shown). That the pseudoexon is not incorporated into mRNA was confirmed by RT-PCR analysis of Pt-ILT3 from several individual chimpanzees (data not shown).

Phylogenetic trees were constructed from the human and chimpanzee LIR nucleotide sequences. Shown in Fig. 3 are two trees, one made from nucleotide sequences encoding the mature proteins, and the other made only from sequences encoding the extracellular Ig-like domains. Both trees have a similar topology consisting of five main clusters of LIR, each containing both chimpanzee and human LIR, and two additional branches represented by single human sequences (ILT10 and ILT11). Comparison of the amino acid sequences also gave trees with similar topology (data not shown). Three of the clusters (III, IV, and V in Fig. 3) contain only a pair of orthologous LIRs: ILT3/Pt-ILT3, LIR8/Pt-LIR8, and ILT7/Pt-ILT7. Cluster I in Fig. 3 involves four chimpanzee and five human LIRs, including the orthologs ILT1/Pt-ILT1. Cluster II of two chimpanzee and three human sequences contains no orthologous LIRs. In both trees Pt-LIRb and ILT4 (LIR2) form a separate branch, raising the possibility that they are orthologs. However, they are more diverged from each other than is the case for the other orthologous pairs, and in both trees the confidence in the ILT4/Pt-LIRb branch (86 and 83%) is considerably less than for ILT1/Pt-ILT1, ILT3/Pt-ILT3, ILT7/Pt-ILT7, and LIR8/Pt-LIR8 (100%).

To assess the role of recombination in generating LIR gene diversity, we used the Partimatrix program described by Jakobsen et al. (38). The extent to which individual LIR genes have been involved in recombination varies. Those genes for which human/chimpanzee orthologies were readily identified (Fig. 1) and which were defined with deeper branches and high confidence in the phylogenetic analyses (Fig. 3) have not been involved in recombination: these comprise ILT3/Pt-ILT3, ILT7/Pt-ILT7, and LIR8/Pt-LIR8. An example of output given for such genes by the Partimatrix program is shown for ILT3/Pt-ILT3 in Fig. 4,A. It shows that there has been little recombination between genes in two major clusters of the phylogenetic trees (ILT5/Pt-LIRd vs ILT3/Pt-ILT3; Figs. 3 and 4 A).

Partition matrix analysis of all five LIRs in cluster II (Fig. 3), the cluster containing no obviously orthologous genes, shows that recombination has played a considerable role in the diversification of these five LIRs (Fig. 4,B). For example, ILT5 and ILT8 are very similar in the region encoding the extracellular part of the molecule (column 3 in Fig. 4,B), but they diverge at the 3′ exons encoding the transmembrane and cytoplasmic domains. Here, ILT5 groups with Pt-LIRd (column 2 in Fig. 4,B), and ILT8 with Pt-LIRe (column 1 in Fig. 4,B), reflecting the divergence of sequences encoding the long-tail and short-tail LIRs. That recombination between exons encoding the Ig-like domains and the 3′ exons has been instrumental in diversifying cluster II LIRs is further supported by differences in the fine structures of this cluster in phylogenetic trees made from sequences encoding mature proteins (Fig. 3,A) and only exons encoding the Ig-like domains (Fig. 3,B). A more complicated pattern of recombination is revealed in the Partimatrix analysis of six members from cluster I (Fig. 3): ILT2, -4, and -6 and Pt-LIRa, -b, and -c (Fig. 4 C).

Each chimpanzee LIR was compared with the most closely related human LIR(s), and the relative frequencies of nonsynonymous (d_N, coding) and synonymous (d_S, silent) substitutions were calculated. Overall comparison of the sequences encoding the mature protein revealed an excess of synonymous substitutions for all LIRs, consistent with the action of purifying selection upon LIR genes (Table III). When sequences encoding only the Ig-like domains were compared, some ratios were higher and closer to 1.00, the value expected under neutral evolution. For the Pt-LIRa/LIR6 and Pt-LIR8/LIR8 comparisons the values were in slight excess of 1.00 (Table III). To refine the analysis we divided the sequence of each Ig-like domain into five similarly sized fragments of ∼60 nt and computed the d_N and d_S values for these fragments. The values are consistent with purifying selection for almost all regions of the Ig-like domains of ILT1/Pt-ILT1, ILT3/Pt-ILT3, and ILT7/Pt-ILT7, with a few isolated regions containing only nonsynonymous substitutions (Fig. 5,A). In contrast, the values for LIR8/Pt-LIR8 provide evidence for positive selection throughout the fourth Ig-like domain and in the carboxyl-terminal part of the first and second Ig-like domains (Fig. 5,A). Comparisons of Pt-LIRd and Pt-LIRe with human ILT5, -8, and -9 gave no clear evidence for positive selection overall; however d_N was locally larger than d_S in segments 10 and 20 of the Pt-LIRd analysis and segment 9 of the Pt-LIRe analysis (Fig. 5 B). Comparison of Pt-LIRa, Pt-LIRb, and Pt-LIRc with human ILT2, ILT4, and LIR6 also revealed no evidence for positive selection (data not shown).

A characteristic of chimpanzee and human KIR haplotypes is variation in the number and content of genes. To determine whether this is also a feature of the chimpanzee LIR gene family, we developed a PCR-based system for typing the nine chimpanzee LIRs and used it to type genomic DNA from a panel of 48 common chimpanzees that had already been typed for Pt-KIR (31). Whereas 30 different Pt-KIR genotypes are represented in this panel, only two different LIR genotypes were identified (Fig. 6). The common genotype consisted of all nine Pt-LIRs and was observed in 42 individuals representing two subspecies, P. troglodytes schweinfurthii and P. troglodytes verus. The minority genotype was defined by negative typing for Pt-LIRa and was seen in six individuals representing three subspecies: schweinfurthii, troglodytes, and verus. Thus there is evidence for some heterogeneity in Pt-LIR haplotypes, but it is modest compared with the variability seen with Pt-KIRs.

LIR and KIR are related gene families closely linked in the human genome, and both encode activating and inhibitory leukocyte receptors in which the extracellular regions are made up of Ig-like domains. Neither of these gene families appears to be present in the mouse genome, at least not as a recognizable ortholog, suggesting that these receptor systems have undergone considerable change and diversification during the course of mammalian evolution (41). Investigation of this proposition requires comparison of species more closely related than mice and men; therefore, we have been studying chimpanzees, the living species most closely related to humans. This study aimed to identify the expressed LIR genes of the common chimpanzee (Pan troglodytes) and to determine their structural and evolutionary relationships with human LIRs.

Thirteen human LIR genes (ILT1–11, LIR6, and LIR8) have been defined, of which two appear to be pseudogenes (ILT9 and ILT10) (4, 5). From the analysis of cDNA clones obtained from a splenocyte library we have characterized nine different chimpanzee LIRs, all expressed in the cells of one individual. The nine chimpanzee and 11 human expressed LIRs exhibit a similar range of diversity, as is clearly seen in phylogenetic trees (Fig. 3). In these trees are five LIR clusters, of which two contain only long-tailed LIRs (III and IV), one contains only short-tailed LIRs (V), and two contain both long- and short-tailed LIRs (I and II). All these clusters and subclusters contain both chimpanzee and human LIR. Thus, the set of nine chimpanzee LIRs defined here should provide good representation of the total set of chimpanzee LIR genes, although the possibility that additional family members await discovery must not be ruled out.

Genotyping 48 unrelated common chimpanzees showed 42 to have genes for all nine chimpanzee LIRs; the remaining six individuals had just eight of the genes. The negative typing reactions for Pt-LIRa could be due to absence of the gene or to Pt-LIRa gene polymorphism that affected the oligonucleotide priming sites used in the typing. Favoring the former interpretation is the fact that two different sets of Pt-LIRa-specific oligonucleotide primers gave negative results. Either way, there is clear evidence for two different chimpanzee LIR haplotypes with frequencies of 35% (Pt-LIRa⁻) and 65% (Pt-LIRa⁺) assuming Hardy-Weinberg equilibrium. Two human LIR haplotypes have also been defined, one containing all the genes and one containing an ILT6 psuedogene with a ∼6.7-kb deletion (5, 16); the latter haplotype appeared to be have a frequency of 18–28% in the populations examined (5, 16). The persistence of such deletion haplotypes is evident by their presence in common chimpanzee subspecies that diverged about 1.6 million years ago (42). The modest variation in LIR genotypes observed within chimpanzee and human contrasts dramatically with the extensive variation in the KIR genotype observed in both species (Fig. 6) (25, 31).

Pairwise comparison of chimpanzee and human LIR allowed four pairs of orthologs to be assigned with confidence (Fig. 1): ILT1/Pt-ILT1, ILT3/Pt-ILT3, ILT7/Pt-ILT7, and LIR8/Pt-LIR8. These pairs of nucleotide sequences differ by about 2%, comparable to estimates of the overall genome sequence similarity of ∼98.8% between human and chimpanzee (43). These genes show little evidence of having undergone recombination with other LIR genes, and in large part the differences between the species do not appear to be the result of natural selection; exceptional in this regard are LIR8 and Pt-LIR8 (Fig. 5 A). On the basis of cDNA sequence comparison, orthologous relationships cannot be assigned between five chimpanzee and nine human LIRs. In the pairwise comparisons of these five chimpanzee LIR sequences, they differ by 9–14% in nucleotides with the closest human LIRs, values much greater than the genome average and which appear largely to be a consequence of recombination between LIR genes. Attempts to identify chimpanzee genes more closely related to human ILT2, -4, -5, -8, and -11 were made by PCR amplification of chimpanzee genomic DNA with primers specific for these human genes. No positive reaction was obtained (data not shown) consistent with the absence of conserved, orthologs for these genes.

Comparison of chimpanzee and human LIR allows us to divide the LIR gene family into a group of four genes that have been relatively stable during the ∼5.5 million yr of evolution that separate modern chimpanzees and humans (44), and an additional group of genes that have been evolving more rapidly through intergenic recombination, gene duplication, and gene deletion. In the human LRC, LIR genes are organized as two similarly sized regions (4). These arose by en bloc duplication of an ancestral region, which gave rise to two daughter blocks in opposite transcriptional orientations and separated by an intervening region of about 200 kb (4, 16). We now see that the four human LIR genes having chimpanzee orthologs are evenly distributed between the two blocks, and each block contains one long-tailed and one short-tailed LIR (LIR8 and ILT7 in one block, and ILT1 and ILT3 in the other; Fig. 7).

In humans, two clusters of rapidly evolving genes in each duplicated block are separated from each other by the orthologous or framework genes (Fig. 7). Regarding the nonorthologous genes, ILT4 and related members (cluster I) lie between LIR8 and ILT7 in one block, and between ILT1 and ILT3 in the other. In contrast, ILT5 and related genes (cluster II) are downstream of LIR8 and ILT3. Thus the organization of the LIR regions is such that conserved, orthologous framework genes alternate with more rapidly evolving genes. This situation is like that found in the KIR gene family where three framework genes define two intervals of high gene variability (5, 31, 45). Also comparable in the two gene families are the percent sequence identities between human and chimpanzee orthologs (∼2–5%) and more rapidly evolving genes (>8%). Furthermore, the numbers of inhibitory, long-tailed KIR and LIR are similar in humans and chimpanzees, but humans have more activating, short-tailed KIR and LIR than chimpanzees. Thus, for both families the inhibitory receptors appear more conserved.

Fig. 7 shows a working model for the organization of the chimpanzee LIR complex and its comparison to human LIR genes. Given the good representation of human and chimpanzee LIR genes in all major clusters, the chimpanzee LIR gene family is probably organized similarly to the human LIR genes in two homologous blocks (Fig. 7). Chimpanzee genes have been placed in the same intervals as human genes. In this scheme one can have some confidence in the positions of the orthologous or framework genes, but the positions assigned to the faster evolving clusters of ILT4- and ILT5-related genes chimpanzee genes remain speculative. Consider first the cluster consisting of ILT4 and related genes (cluster I in Fig. 3). As Pt-LIRb groups with human ILT4 in both phylogenetic analyses (Fig. 3), and since they both encode long-tailed LIRs, we assigned them to the same interval. As Pt-LIRc is related to human ILT6 (Fig. 3B), and as they are the only genes encoding soluble proteins (Fig. 2), Pt-LIRc could be assigned to the same interval as ILT6. Alternatively, ILT6 and Pt-LIRa could be placed at the same interval, as they are both the missing genes on the minority haplotypes. Of the human genes in the ILT5-related cluster (II in Fig. 3), ILT5 and ILT8 are together in one gene block and represent a closely related pair encoding an inhibitory and an activating four-Ig-domain receptor. In contrast, the ILT5-related gene of the other block is ILT9, an unusual LIR with three Ig-like domains. As Pt-LIRd and Pt-LIRe encode a conventional pair of inhibitory and activating LIR, such as ILT5 and ILT8, these genes have been assigned to the same interval as ILT5 and ILT8, respectively. These tentative assignments provide a working model for future direct analysis of the genomic organization of the chimpanzee LIR gene family.

In both humans and chimpanzees, their KIR haplotypes are much more diverse in gene number and content than are the LIR haplotypes. The propensity for new KIR haplotypes to evolve by asymmetric recombination is explained by the close proximity of the KIR genes and their separation by short, highly homologous intergenic sequences (5). This is not the situation for the human LIR gene family (4, 5). Due to this difference we anticipated that the LIR genes would be phylogenetically more conserved than the KIR genes. This assumption was proven wrong, with the interspecies variation in KIR and LIR genes being comparable. Both families contain some genes that are relatively stable and recognizable as orthologs, whereas other genes have been rapidly evolving through recombination. Thus, for some genes in both the KIR and LIR families there appears to be natural selection for new variants. The KIR genes known to encode receptors with specificity for classical MHC class I molecules are ones that have rapidly evolved since divergence of human and chimpanzee ancestors (31). Similarly, the LIRs with known specificity for MHC class I are encoded by genes, ILT2 (LIR1) and ILT4 (LIR2), with no clear-cut chimpanzee orthologs and that are members of a cluster of rapidly evolving genes. In conclusion, this comparison of chimpanzee and human species shows that both the LIR and KIR gene families evolve rapidly, making the LRC a likely hotspot of difference between chimpanzee and human genomes.

We thank the Yerkes Regional Primate Center at Emory University for providing chimpanzee samples. B.P.S. thanks particularly the staff at Yerkes for their hospitality and for their assistance in obtaining chimpanzee spleen samples. We also thank Dr. Erin Adams for the generation of the chimpanzee B cell lines.