Although still located within the LRC 314, the human KIR3DL0 locus maps approximately 180 kb centromeric to the known KIR cluster (Figure 1A). This location in between the two related leukocyte immunoglobulin-like receptor (LILR) clusters may explain why this gene has escaped discovery for so long. Despite its location outside of the KIR cluster, KIR3DL0 is clearly more related phylogenetically to KIR genes than to any other Ig-containing gene within the LRC (Figure 1B).

All KIR family members share a comparable gene structure; discrete exons encode the signal peptide, either two or three extracellular immunoglobulin domains, a stem, transmembrane region, and a cytoplasmic tail. As we show below, the genomic sequence for KIR3DL0 is characteristic of KIR and highly conserved in all primates (Figure 2), but, in human, a single deletion of seven nucleotides at the end of the third immunoglobulin domain (IgD2) in exon 5 leads to a frameshift, preventing it from being expressed on the cell surface. Because of the deletion, it uses alternative reading frames for the end of exon 5 and exons 6/7, resulting in a potentially secreted form of KIR3DL0 with no TM and cytoplasmic domains. The only possibility of the KIR3DL0 receptor remaining membrane-bound would be an association with the adaptor molecule DAP12 mediated by a charged amino acid in the TM. DAP12 is able to stimulate cytotoxic activity as it contains an immunoreceptor tyrosine-based activation motif and is the mode in which activating KIRs have an effect. Although the frameshift has resulted in the presence of a charged amino acid (K) in the terminal exon (exon 7), the deduced amino acid sequence does not correspond to a TM or core hydrophobic region, which makes the association with DAP12 less likely in this case. Human KIR3DL0 is based on a full-length cDNA isolated from a human NK cell line (NK-92), and is further supported by a partial cDNA sequence [GenBank:BC033195] from the mammalian gene collection (MGC), isolated from a different human NK cell line (NIH_MGC_106). In order to check for polymorphism, 86 unrelated healthy individuals were sequenced across the deleted region and the same frameshift causing deletion was present in all individuals.

The relatedness of KIR3DL0 to the KIR gene family is further confirmed by gene structure analyses. Using FINEX, a method that identifies gene families based on the conservation of gene structure (intron/exon phases) rather than sequence similarity 15, the closest relative of KIR3DL0 is the prototypical human KIR gene KIR3DL1 with a significant z-score of +10.88 (data not shown). Furthermore, using VISTA 16, a global genomic comparison tool, significant conservation in both coding and non-coding sequences is present between all primate KIR3DL0 sequences and KIR3DL1 [see Additional file 1].

We show KIR3DL0 to be highly conserved in primates with the identification of orthologues in the common chimpanzee (Pan troglodytes), gorilla (Gorilla gorilla), rhesus monkey (Macaca mulatta) and common marmoset (Callithrix jacchus). The two copies in marmoset are arranged in a head-to-head orientation and are likely to be the result of a species-specific duplication event. These non-human primates diverged from humans approximately 5, 7, 25 and 50 million years ago, respectively. In all cases, KIR3DL0 maps to the syntenic region between LAIR2 and LILRA2. Details of the genomic LRC organization in these primates will be described elsewhere.

The primate KIR3DL0 sequences share the same exon-intron configuration and are predicted to encode functional proteins (Figure 2). In addition to the human cDNA, full length cDNA sequences were isolated from chimpanzee and rhesus monkey. Variations between the genomic and cDNA sequences may indicate KIR3DL0 to be polymorphic. Although only two sequences could be compared per species, we identified five non-synonymous substitutions in chimpanzee and one synonymous substitution in human. The conservation of two immunoreceptor tyrosine-based inhibitory motifs, present in the cytoplasmic tail of the predicted KIR3DL0 protein sequences from all species studied, indicates an inhibitory function for these receptors. These motifs are present in only one copy in marmoset, which has the complete nine-exon gene structure. The short form is missing the last three exons encoding the TM and cytoplasmic domains, resulting in a potentially secreted form, which is similar, but not identical, to the KIR3DL0 cDNA in human (Figure 2). While three N-linked glycosylation sites (NXS/T) are conserved in the majority of the sequences analyzed, one additional site is present in the IgD2 of the two KIR3DL0 copies in marmoset. For the short marmoset form in particular, the extra glycosylation site increases the potential level of glycosylation. High levels of glycosylation have been shown to be essential for secretion of some immunoglobulins 17.

The KIR genes identified to date have been divided into five distinct lineages 891113: lineage I comprises genes with specific domain configurations (IgD0 and IgD2 domains), lineage II is defined by ligand specificity (HLA-A and HLA-B), lineage III contains a mixture of KIR genes with two or three Ig domains, lineage IV genes are only found in rhesus monkeys, and lineage V encompasses KIR genes lacking the stem. Having established that KIR3DL0 is conserved across species at the gene and protein levels (e.g. 33–41% similarity to human KIR3DL1 [see Additional file 2], we used phylogenetic analysis to determine whether or not the new KIR3DL0 genes also represent a novel KIR lineage. Using the deduced full-length protein sequences of known primate KIR genes, the tree shown in Figure 3 was constructed. The novel KIR3DL0 sequences all cluster separately and with high bootstrap confidence at the base of the tree, demonstrating that KIR3DL0 represents not only a novel but also the ancestral lineage of primate KIR genes. In order to confirm this finding and to assess whether, because of its location, KIR3DL0 arose by non-homologous recombination between KIR and LILR sequences, separate trees were also constructed for each of the KIR3DL0 domains and the corresponding domains of KIR and LILR. Under the conditions used here (see Materials and Methods for details), all KIR3DL0 domains clustered at the base of the KIR clade, separated from the LILR domains [see Additional file 3]. The IgD1 domain of KIR3DL0 is more closely related to the IgD0 domains of the KIR gene family (including the IgD0 of KIR3DL0) in comparison to all other domains used in the analysis.

We confirmed expression of KIR3DL0 by cDNA cloning in human, chimpanzee and rhesus monkey. The human cDNA was cloned from an NK cell line (NK-92), and preliminary analysis shows expression in five other NK cell lines but not in Jurkat cells (T cells) (data not shown). To evaluate the pattern of expression in more detail, we chose the rhesus monkey. Peripheral blood mononuclear cells (PBMC) were obtained and NKp46+ cells, previously shown to be enriched for NK cytolytic activity 18, were isolated by immunoaffinity methods. Firstly, using RT-PCR, a 1.4 kb cDNA was identified that encodes a protein with 458 amino acids, and verifies the in-silico structure of primate KIR3DL0. A second fragment was also observed and represents an alternative splice form missing exon 5; this variant has only two Ig domains. Secondly, levels of KIR3DL0 mRNA in these cell populations were quantified by a real time RT-PCR assay. In view of data indicating that IL-2 can increase expression of KIRs, rhesus NKp46+ cells were also cultured in the presence of recombinant human IL-2 for one week, and tested for KIR3DL0 mRNA expression at the end of this period. As shown in Figure 4, levels of KIR3DL0 message were modestly enriched in NKp46+ cells compared to unfractionated PBMC for the three rhesus monkeys tested, although the relatively low number of copies per NKp46+ cell suggests that the receptor may be expressed by only a subset of NKp46+ cells. Culture in the presence of IL-2 resulted in substantial upregulation of expression of KIR3DL0 in cells from all three animals.

BAC contigs covering the KIR region were generated and sequenced as described previously 11. KIR3DL0 genomic sequences were obtained as follows: for human, using ENSEMBL 26 assembly NCBI-35; for chimpanzee, using ENSEMBL assembly CHIMP-1; for gorilla, by BAC sequencing [EMBL:CR759947 and EMBL:CR759950]; and for marmoset, by BAC sequencing [EMBL:CR925829]. The single sequence gap that covers the transmembrane region in the CHIMP-1 assembly was amplified and sequenced using forward (5'-caccaacattctttggagcaagtt-3') and reverse (5'-aggctgaggtggaagaatggc-3') primers. The frameshift region in the human KIR3DL0 gene was sequenced in 86 healthy individuals (81 Caucasian, 3 African American, 1 Asian, 1 unknown ethnicity) using forward (5'-ttccaggccaacttttctgtgg-3'), and reverse (5'-tctgtgatccagtgggcacca-3') primers. KIR3DL0 cDNA sequences were obtained by reverse transcription of total RNA using Superscript (Invitrogen). Total RNA was isolated using Trizol (Invitrogen). The macaque and chimpanzee KIR3DL0 cDNAs [GenBank:DQ224422 and GenBank:DQ157756] were cloned by two rounds of PCR from peripheral blood lymphocyte cDNA. The first round of PCR was performed with primers KIR3DL0-F2, 5'-ctgtgtcctgcccaatagaag-3', and KIR3DL0-R2, 5'-cctcctaggaatagatccagg-3'. For the second round, we used primers KIR3DL0-F2 and KIR3DL0-R1, 5'-aggaatagatccaggtccttg-3'. The human KIR3DL0 cDNA [GenBank:DQ224421] was cloned by one round of PCR using the KIR3DL0-F2 and KIR3DL0-R1 primers.

Genomic sequences were analyzed using ARTEMIS 27, ACT 28, and BLAST 29. Multiple sequence alignments were carried out in ClustalX 30, and edited manually to maximize the alignment. Trees were constructed using the Neighbour Joining method 31 in MEGA version 2.1 32, using human KIR3DL0 derived from genomic sequence, with the frameshift in IgD2 manually corrected to bring the sequence back into frame. Trees were rooted at midpoint, using a complete deletion (Figure 1B) or pairwise deletion (Figure 3), the Poisson correction option, and 500 bootstrap replicates. Global genomic comparisons were made using VISTA 16. FINEX 15 was used in conjunction with its corresponding gene structure database derived from EMBL version 181 (eukaryotes only).