Killer cell lectin-like receptor
Immunoreceptor tyrosine-based inhibition motif
Open reading frame
Expressed sequence tag
Rapid amplification of cDNA ends
Natural killer cell gene complex
Natural killer (NK) cells recognize and kill certain tumor cells, virally infected cells and MHC class I-disparate normal hematopoietic cells. NK cell cytotoxicity is regulated by a multitude of receptors with either activating or inhibitory signaling function. We here report the molecular cloning of bovine CD94 [killer cell lectin-like receptor (KLR)-D1] and NKp46 orthologues, four members of a bovine CD158 [killer cell immunoglobulin-like receptor (KIR)] family, and a novel KLR. This novel receptor was termed KLRJ1 and is most similar to Ly-49 (KLRA). The KLRD1 and KLRJ1 loci were mapped to a bovine NK gene complex on chromosome 5 by radiation hybrid mapping, whereas KIR2DL1 and NKP46 were localized to chromosome 18. Two of the bovine KIR(KIR2DL1 and KIR3DL1) contain immunoreceptor tyrosine-based inhibition motifs (ITIM), suggesting an inhibitory function. Bovine KIR2DS1 and KIR3DS1 lack ITIM but have an arginine-containing motif in their transmembrane domain, similar to primate KIR2DL4. Thus, KIR multigene families with divergent signaling motifs do not only exist in primates. Based on sequence comparison, it appears that the primate and bovine KIR multigene families may have evolved independently.
Natural killer (NK) cells of vertebrates are large granular lymphocytes with the capacity to recognize and lyse certain tumor cells, virus-infected cells as well as MHC-disparate normal bone marrow-derived cells 1–3. Several receptor molecules underlying this recognition have recently been identified. Structurally, NK cell receptors fall into two categories 4. The first are type II transmembrane proteins with a C-terminal domain similar to C-type lectin domains, and have been termed killer cell lectin-like receptors (KLR). These include, among others, CD94 and the NKG2, NKR-P1 and Ly-49 multigene families, which reside within the NK cell gene complex 4. In contrast to classical C-type lectins, the KLR lack key amino acids necessary for calcium binding, and may not primarily recognize carbohydrate ligands 5, 6. The other category of NK cell receptors are type I transmembrane proteins, with one or several extracellular immunoglobulin superfamily (IgSF) domains, encoded by a cluster of genes referred to as the leukocyte receptor gene complex 7. These include the leukocyte immunoglobulin-like receptors (LILR), the killer cell immunoglobulin-like receptors (KIR, also termed CD158), NKp46, and others 4. NKp46 is only expressed by NK cells 8; the KIR are expressed by NK cells and subsets of T cells 9; and the LILR are expressed by several types of leukocytes 10.
The KIR and Ly-49 multigene family members bind to MHC class I or class I-like ligands, and either inhibit or activate NK effector functions 4. Inhibitory receptors have immunoreceptor tyrosine-based inhibition motifs (ITIM) in their cytoplasmic regions, which upon tyrosine phosphorylation recruit and activate the inhibitory protein tyrosine phosphatase SHP-1 10. Activating NK receptors commonly contain a charged amino acid (lysine or arginine) in their transmembrane regions, which bind to adaptor molecules with an activating function. These contain either immunoreceptor tyrosine-based activation motifs (ITAM) necessary for recruitment of Syk or ZAP70 tyrosine kinases, or a tyrosine-based motif (YxxM) that recruits the p85 subunit of phosphatidylinositol 3-kinase 11, 12. Thus, the effector functions of NK cells are controlled by a complex balance of inhibitory and activating signals generated by a variety of receptors co-expressed on the same cell.
While our understanding of NK cell recognition in humans, rats and mice has substantially increased through the molecular identification of NK receptors, very little is known about NK cell function in cattle. We here report the molecular cloning of several bovine NK cell receptors. A novel KLR family member (KLRJ1) is identified, together with bovine orthologues of CD94 and NKp46. We also demonstrate the existence of multiple KIR members in cattle. The bovine KIR contain either ITIM in their cytoplasmic regions, or instead an arginine residue in the transmembrane region, suggesting opposite signaling functions.
2.1 cDNA cloning of a bovine CD94 orthologue
A bovine expressed sequence tag (EST) (GenBank accession number BE808683) with homology to rodent 13 and primate 14 CD94 was found by searching the dbEST database. The corresponding cDNA clone was obtained and sequenced. This clone contained the entire open reading frame (ORF) of a putative bovine CD94 homologue, but also contained an intron between the first and second lectin domain-encoding exons. Based on this sequence, three independent cDNA clones were obtained from a bovine spleen cDNA library by PCR with 5′-untranslated region (UTR)- and 3′-UTR-specific primers. Sequence analysis yielded a consensus cDNA sequence (GenBank accession number AF422180) with an ORF that is 65.7%/75.2% and 52.8%/68.5% identical at the amino acid/nucleotide levels to human 14 and rat 13 CD94, respectively. Sequence comparison with the rodent and primate KLR families indicated that this consensus sequence represents a bovine CD94 orthologue (Fig. 1A, C), and the corresponding gene has been named KLRD1 (KLR family D, member 1) to comply with human gene nomenclature. The ORF encodes a putative polypeptide of 190 amino acids and a relative molecular mass of 21.9 kDa, with two potential sites for N-linked glycosylation. The short cytoplasmic region (ten residues) and the transmembrane region contain no apparent signaling motifs (Fig. 1A).
2.2 cDNA cloning of KLRJ1, a novel killer cell lectin-like receptor
Another bovine EST (BE588664) displayed some similarity to the Ly-49 multigene family 15, but was derived from a truncated cDNA clone. To obtain the entire ORF, a bovine spleen cDNA library was screened by hybridization with a probe generated by PCR. A single full-length clone of 1111 nt (GenBank accession number AY156928) was isolated and sequenced. This clone contained an ORF predicting a 275-amino acid type II transmembrane polypeptide of 31.3 kDa, containing a cytoplasmic region, a transmembrane region, a stalk region and a C-type lectin-like domain at the C-terminal end. The sequence of the lectin-like domain contained Ly-49 signature motifs, in particular paired cysteine residues, implicating a double disulfide bridge between the α1 and β5 strands, in a position unique to Ly-49 family members. On the other hand, the putative transmembrane and cytoplasmic regions showed very little similarity to Ly-49 or other KLR families.
To investigate the possibility of a library artifact, additional cDNA clones were obtained from bovine blood leukocyte RNA by nested 5′-rapid amplification of cDNA ends (RACE) with 3′-UTR primers. Sequence analysis of these clones yielded an ORF identical to the initial clone and suggested that it represented a correctly spliced full-length transcript. Notably, no ITIM motifs were present in the cytoplasmic domain, and the transmembrane region lacked charged residues. The lectin-like and stalk domain peptide sequence was 24.7% identical to human Ly-49 16 and between 24.2 % and 26.8% to those of mouse 15 and rat 17 Ly-49. Sequence comparison indicated that this bovine cDNA clone was more similar to the Ly-49 family than to other KLR families (Fig. 1B, C). However, this cDNA showed little similarity with any of the KLR families within the intracellular and transmembrane regions, and the extracellular region was less similar to the Ly-49 family than a recently identified bovine Ly-49 cDNA 18. The clone identified here should thus not be regarded as an additional bovine Ly-49, but rather as a separate KLR family, and has been named KLRJ1 (KLR family J, member 1). Bovine KLRJ1 presently lacks orthologues in primates or rodents. Southern blot analysis of bovine liver DNA produced a simple pattern with few bands with a KLRJ1 probe (data not shown), suggesting that it is a single gene without close relatives in cattle.
2.3 Isolation of a bovine NKp46 orthologue
A third bovine EST (BF890160) with homology to human 19, rat 20 and mouse 21 NKp46 was found by database searches. The corresponding cDNA clone was obtained and fully sequenced, but contained an intron. Therefore, three independent, correctly spliced clones were obtained from a bovine spleen cDNA library by PCR cloning with 5′-UTR- and 3′-UTR-specific primers. The consensus sequence ORF (GenBank AF422181) predicted a 287-amino acid mature polypeptide of 32.3 kDa, showing 62.9% and 60.0% amino acid identity to human and rat NKp46, respectively, indicating that the clone represents a bovine NKp46 orthologue (Fig. 2, 3A). The transmembrane region contains an arginine residue in a similar position to the rodent and primate orthologues, suggesting that it may have an activating role in intracellular signaling (Fig. 3B).
2.4 Cloning of a KIR multigene family, comprising KIR2DL, KIR2DS, KIR3DL and KIR3DS variants
A bovine EST (BE758052) displayed similarity to the human KIR multigene family 22, 23, and the corresponding cDNA clone was obtained and fully sequenced. Other investigators have now published the sequence of the same cDNA clone, demonstrated that it is a bovine KIR family member, and named it KIR2DL1 18. Similar to the results of McQueen and co-workers, we found that Southern blot analysis of a DNA probe corresponding to the full ORF of this clone displayed multiple bands, suggesting the existence of several bovine KIR family members (data not shown). The KIR2DL1 probe was used to screen a bovine spleen cDNA library. Twelve independent clones were isolated and sequenced, yielding five different ORF with homology to primate KIR. None of the clones were full-length. One clone, which lacked only the first two amino acids of the signal peptide, encoded a KIR2DS variant that contained an arginine residue in the transmembrane domain and had a short cytoplasmic tail without ITIM. To obtain the whole ORF of the other variants, 5′-RACE was carried out on bovine blood leukocyte RNA using reverse nested primers specific for the 3′-UTR regions of the various clones. Three independent PCR clones of appropriate lengths were fully sequenced. This approach yielded consensus full-length ORF predicting a KIR3DS as well as a KIR3DL family member.
To comply with the suggested bovine nomenclature 18, the bovine KIR have been named KIR3DL1, KIR3DS1 and KIR2DS1, respectively (Fig. 4). This is not intended to suggest homology to particular primate KIR family members, as the full polypeptide sequence of each bovine KIR is more similar to other bovine KIR than to any of the primate KIR (Fig. 3A, 4). In addition to the four complete bovine KIR sequences, partial sequences of additional KIR3DL variants were obtained (not shown). Some of these may be allelic variants, but our present information indicates that the bovine KIR family contains additional members that have yet to be characterized.
Several sequence features distinguish bovine KIR from their primate relatives. Firstly, human KIR2DL members may dimerize by association with zinc or related metal ions (Ni2+, Cu2+, Cd2+ or Co2+) through a histidine-containing motif at their N termini 24, 25. The corresponding N-terminal sequences of bovine KIR lack histidine residues, and as such may lack the ability for cation bridging. Instead, KIR3DL1, KIR3DS1 and KIR2DS1 may form disulfide-linked dimers between cysteine residues located in their respective stem regions. KIR3DL1 has only a single ITIM, similar to inhibitory rodent Ly-49 family members. These form disulfide-linked homodimers and thus contain two ITIM in close proximity, which is probably necessary for recruitment and activation of SHP-1 26. In support of this, bovine KIR2DL1 lacks the stem region cysteine, but has two ITIM in tandem in its cytoplasmic region (Fig. 4).
KIR3DS1 and 2DS1 lack ITIM and carry a basic amino acid in their transmembrane region, suggesting an activating function. In contrast to the human activating KIR, which have a lysine residue in the middle of their trans-membrane regions that interact with an aspartate in the transmembrane region of DAP12 27, bovine 3DS1 and 2DS1 have a transmembrane arginine residue that is located closer to the extracellular side of the membrane (Fig. 3B). The only human KIR with a similar feature is KIR2DL4, the signaling mechanisms of which are not yet entirely clear 28, 29. The bovine KIR motif bears similarity to FcαR and NKp46, which do not appear to interact with DAP12 but instead with FcRγ chain homodimers 30 or FcRγ/TCRζ chain heterodimers 12, 19. However, these receptors invariably contain a glutamate residue at the cytoplasmic end of the transmembrane region. Neither bovine KIR2DS1, bovine KIR3DS1, nor primate KIR2DL4, have an acidic residue in this position (Fig. 3B). Together this suggests that bovine KIR2DS1 and KIR3DS1 may associate with other, possibly yet unknown signaling molecules.
The crystal structures of human KIR2DL1 and KIR2DL2 in complex with their HLA-C ligands have revealed the individual residues that make-up the receptor-ligand contacts 31, 32. These residues are only conserved to a small extent between primates and cattle (Fig. 4), and the ligands for bovine KIR remain to be identified.
The individual IgSF domains of bovine KIR show homology to either the D0-, D1- or D2-type domains of primate KIR (Fig. 3C) 9. Bovine KIR3DL1 and KIR3DS1, like primate KIR3D members, have a D0-D1-D2 configuration. Bovine KIR2DL1 has a D0-D2 configuration 18. This differs from most primate KIR2D molecules, which have a D1-D2 configuration, but is similar to primate KIR2DL4 and KIR2DL5 receptors that lack the D1 domain due to a deletion involving exon 4 9. The bovine KIR2DS1 cDNA has a D0-D1 configuration and lacks a D2-type IgSF domain (Fig. 3C). This is highly unusual, and suggests that bovine KIR2DS1 may be a splice variant of a gene that may also encode a splice variant with three IgSF domains.
2.5 Bovine NK cell receptors are clustered in two separate gene complexes on chromosomes 5 and 18
A bovine/hamster radiation hybrid (RH) panel 33 was used to determine the chromosomal localization of KIR2DL1, NKP46, CD94 (KLRD1) and KLRJ1. Bovine exon-specific PCR primer pairs were used to score each hybrid cell line for presence or absence of the bovine gene. Two-point analysis was carried out with resulting data against the RH data for markers whose chromosomal locations were already known from linkage or physical mapping. KLRD1 was localized to chromosome 5 (BTA 5) and linked most tightly to BMS1658 (Lod score: 13.6/10 centiRay). KLRJ1 mapped to the same position, indicating the existence of a bovine NK gene complex. The BTA 5 region containing the bovine NKC is homologous to HSA12p13, RNO4 and MMU6, where the NKC of the three species are located 17, 34, 35. Computer searches of additional bovine EST, together with published cDNA sequences, suggests the existence of bovine NKR-P1, NKG2 and CD69 orthologues 36, 37 in addition to bovine KLRA1 18 and the two KLR families described in this report. These putative bovine KLR genes are likely to also be localized within the bovine NKC.
The RH results for KIR2DL1 and NKP46 localized these genes on chromosome 18, most closely linked to the microsatellite marker TGLA132 (KIR2DL1 –7 centiRay/ Lod: 16.5 – TGLA132 –21 centiRay/Lod: 10.4 NKP46) (Fig. 5). This is similar to the situation in primates and rodents, where immunoglobulin-like regulatory receptors are clustered in a leukocyte receptor gene complex located on HSA19q13 7, RNO1 38 and MMU7 39, respectively.
In this report, we describe several bovine receptor genes, homologous to NK cell receptor genes in rodents or primates. From our data and the work of others, a large proportion of the primate and rodent NK receptor repertoire appears to have orthologues or homologues in cattle 18, 36, 37. One possible exception is bovine KLRJ1. We have not been able, by computer searches of EST or genomic databases, to identify possible KLRJ1 orthologues in primates or rodents. Its close kinship with theLy-49 family suggests a role in NK cell recognition of target cells, and functional studies of this novel KLR family member are warranted. However, the lack of NK-specific antibodies has hampered the functional characterization of bovine NK cells. Our cloning of NKp46, a pan-NK cell surface marker in humans 8 as well as in rodents (S. F. Berg, I. H. Westgaard, E. Dissen et al., unpublished), paves the way for generation of NK-specific mAb in cattle. The identification of NK cell receptors in cattle also facilitates molecular studies of NK cell function in this agriculturally important species.
The identification of multiple KIR genes in cattle disproves the idea that KIR receptors originated at the level of primates, and instead indicates the existence of one or several ancestral KIR genes prior to the split between cattle and primates. All primate species investigated contain multiple KIR genes; about 12 genes appear to exist in humans, with considerable haplotype variation 7, 9. Four bovine KIR genes have been identified in this study, and additional partial cDNA sequences indicate that the bovine KIR family may be of about the same size as that of primates. The KIR multigene families in these species could have evolved either from a single ancestral gene (by a series of gene duplication events) or from several KIR genes predating the split in evolution. By comparing all primate and cattle KIR peptide sequences, the bovine KIR are more similar to each other than to any individual primate KIR, and vice versa. Although thisfavors the former hypothesis, a multigene heritage could have been masked by extensive homogenization.
However, breaking the KIR down to individual exons and comparing sequences did not provide evidence for the existence of a family of ancestral KIR. The IgSF domains of both bovine and primate KIR can be grouped into D0-, D1- and D2-type, indicating that a single ancestral gene must have been a KIR3D with a D0-D1-D2 configuration. Although it lacks the D1 domain due to a genomic deletion,primate KIR2DL4 stands out as possibly most related to a single ancestral gene for several reasons. It is the only KIR found in all primates investigated, and is expressed by most NK cells 9, 40. More importantly, the transmembrane regions of primate KIR2DL4 contain an arginine residue in exactly the same position as bovine KIR2DS1 and KIR3DS1, with some ofthe flanking amino acids also conserved. Although a role for this residue in signaling has not yet been demonstrated, the presence of this motif only in the bovine KIR lacking ITIM suggests that itmediates binding to an adaptor protein with activating function. If indeed this transmembrane motif existed prior to the primate/cattle split, activating KIR may have a long history in evolution. In addition, the presence of ITIM motifs in tandem orientation in bovine KIR is similar to the inhibitory KIR in primates.
The question thus arises whether at least two KIR co-existed before the split, one with ITIM and the other with an arginine residue in the transmembrane region, or whether both motifs were present in the same KIR, as is the case with human KIR2DL4. Alternatively, these signaling motifs may have been "imported" from other receptor genes by exon shuffling, favored by a common selective pressure. It seems unlikely, however, for this to have taken place in parallel in cattle and primates. Based on our current information, it seems possible that KIR multigene families may have separately evolved in primates and cattle, from a single ancestral KIR3D gene with both ITIM and an arginine residue in the transmembrane region, and that selective pressure has favored segregation into inhibitory (KIR2DL or KIR3DL) and activating (KIR2DS and KIR3DS) members. Further investigation of other mammalian species, and of the functional role of bovine KIR2DS1 and KIR3DS1, is needed to conclude on these matters.
4 Materials and methods
4.1 Cloning of CD94 cDNA
The full, correctly spliced ORF of CD94 cDNA was amplified by PCR from a bovine spleen cDNA library (Uni Zap XR Library, Stratagene, La Jolla, CA) using gene-specific primers (5′-TTAGCTCCATGGACTAACCTCACTAT-3′ and 5′-TCGAAGATCCCAGAGGTAGGTAATA-3′) and PfuTurbo DNA polymerase for 35 cycles (Stratagene) followed by a second PCR using Taq DNA polymerase (Promega, Madison, WI) for ten cycles. The PCR product was ligated into pCR2.1-TOPO vector (Invitrogen, Groningen, The Netherlands), and three individual clones were sequenced on both strands (Medigenomix, Martinsried, Germany).
4.2 Cloning of KLRJ1 cDNA
Using primers (5′-TGTGGATTCTTTGGTTACCAG-3′ and 5′-TGAGGCCAGAAGGAAGAGAA-3′) derived from a bovine EST (GenBank accession number BE588664), a 0.5-kb product was amplifiedfrom a bovine spleen cDNA library (Stratagene), cloned, and used as a probe to screen the same library for a full-length clone. Plaques were lifted onto nylon membranes and hybridized to probe radiolabeled with [32P]dCTP (Megaprime DNA labeling system, Amersham International, Little Chalfont, GB) at 42°C for 20 h in 50% formamide, 5× SSC, 50 mM sodium phosphate pH 6.5, 250 μg/ml sonicated salmon testis DNA (Sigma, St. Louis, MO), 5× Denhardt's solution and 0.1% SDS. Membranes were washed 4×5 min in 2× SSC, 0.1% SDS at room temperature, then 2×30 min at 50°C in 0.5× SSC, 0.1% SDS and subjected to autoradiography. One clone was isolated and sequenced. According to the manufacturers' instructions, 5′-RACE (GeneRacer, Invitrogen) was performed on total RNA 13 isolated from bovine blood leukocytes using the adaptor-specific 5′ primer paired with the gene-specific 3′-UTR primer 5′-GGTTCTGAGTGTCTATGCTTTGTGTG-3′ for ten cycles followed by 35 cycles using the inner adaptor-specific 5′ primer combined with the gene-specific primer 5′-GGCTTTAACTCTAGACTCTGGATTCTTC-3′. A 1.1-kb PCR product was excised from a low-melting-point agarose gel, purified (Wizard PCR Prep, Promega), TA-cloned and sequenced as described above.
4.3 Cloning of NKp46 cDNA
Three individual clones of 1.3-kb were amplified from the bovine spleen cDNA library using a vector-specific 5′ primer combined with a gene-specific 3′-UTR primer (5′-CAGGAAAGATGGGGGTGGTTGA-3′), separated on agarose gel, purified, cloned and sequenced as described above.
4.4 Cloning of KIR cDNA
A cDNA corresponding to an EST with homology to primate KIR (GenBank accession number BE758052) was obtained and sequenced. To search for additional KIR cDNA, a 1.0-kb probe corresponding to the ORF of this sequence was used to screen the bovine spleen cDNA library (using the same conditions as described above). Eleven clones were isolated and sequenced. As most clones did not contain complete ORF, nested 5′-RACE was performed on bovine peripheral blood leukocyte RNA using adaptor-specific forward primers paired with different combinations of reverse primers generated from the 3′-UTR of the isolated KIR clones (KIR3DS1, outer: 5′-CTCAAAGGATGGGGCCAGGTCA-3′, inner: 5′-TGTCTCAGAGAGGGTCCTGTGGT-3′; KIR3DL1, outer: 5′-GGGCTGCTTTGGAGGAGAAGGAA-3′, inner 5′-GGTCCTTGATTTGTAAGACTCTGTGTGC-3′). The products were excised from agarose gel, purified, cloned and sequenced as described above.
4.5 RH mapping
The Roslin/Cambridge 3,000-rad bovine/hamster RH cell panel 33 (ResGen; Invitrogen) was screened with the following exon-specific primer pairs: KIR2DL1: 5′-CGTGGTTCCTCAAGGACAGCAT-3′ and 5′-CAGATTGGGGAGGTGACTGTAGTG-3′ (68°C annealing temperature); NKP46: 5′-GCAGATGCTCTCGAAACCTG-3′ and 5′-GGTAGGAGCACCTGTATTGCC-3′ (64°C annealing temperature); CD94: 5′-TATTACCTACCTCTGGGATCTTCG-3′ and 5′-TGTTGATGAATTAGGCATACTGTC-3′ (63°C annealing temperature); and KLRJ1: 5′-CTGTGTGAGAAGCCTGTCTGTG-3′ and 5′-GGTTCTGAGTGTCTATGCTTTGTGTG-3′ (63°C annealing temperature).
Thirty-five cycles of amplification were carried out in 25 μl volume with 20 ng template DNA, 6 pmol of each primer, 0.5 U of DyNAzyme II in 1× buffer (Finnzymes, Espoo, Finland) and 50 μM of each dNTP, using a hot start protocol and between four and eight initial touch-down cycles (1°C reduction for each cycle). PCR products were separated on a 2% agarose gel. Two investigators carried out scoring of PCR products independently. Two-point analysis against bovine microsatellite markers previously typed on the panel was performed using the RHMAP3.0 package 41.
4.6 Sequence analysis
GenBank databases were searched using tBLASTn 42, and sequence analysis was performed using software available through the Norwegian EMBNet node. Protein alignments and treegeneration was performed with CLUSTAL X 43 and trees visualized with the NJplot or TreeView 44 programs.
We thank Marianne Lauritzen for technical assistance and Sigbjørn Fossum for critical reading of the manuscript. This work was supported by the Norwegian Cancer Society, the Norwegian Research Council, and Bergljot and Sigurd Skaugen's Fund.