Genomics and Diversity of the Common Marmoset Monkey NK Complex¹²

Natural killer cells are large granular lymphocytes that are part of the innate immune system. Upon triggering, NK cells are able to release cytokines and to kill target cells, usually virally-infected cells or tumor cells (1, 2, 3). Expressing a diverse array of activating and inhibitory receptors on their cell surface (4, 5), NK cells scan target cells for the presence of interacting ligands (6), which are in most cases members of the family of MHC class I molecules. The integration of such activating and inhibitory signals determines the functional outcome of the NK cell (7).

Almost all NK cell receptors are encoded in two gene clusters, the leukocyte receptor complex (LRC)⁵ and the NK complex (NKC), which map to human chromosomes 19q13.4 (8) and 12p13 (9), respectively. LRC-encoded receptors contain Ig-like domains, whereas the NKC encodes C-type lectin-like receptors (10). Genes mapping to the NKC include CD94 and the families of Ly49 and NK group (NKG) 2 genes (4, 11). Whereas LY49 and NKG2D molecules form homodimers, CD94 and NKG2 molecules are expressed as heterodimers at the cell surface. The Ly49 gene family is expanded and diverse in rodents (12) and includes activating and inhibitory receptors that are characterized by presence of a positively charged residue in the transmembrane region and ITIM in the cytoplasmic region, respectively (13). The ITIM-containing inhibitory receptor NKG2A and the activating receptor NKG2C form heterodimers with the CD94 molecule, which exhibits neither an activating nor an inhibitory motif. The CD94/NKG2A and the CD94/NKG2C molecules interact with the nonclassical MHC class I ligand HLA-E (14). Recently, binding of CD94 and the activating receptor NKG2E has been demonstrated (15), while the function of NKG2F, which has a truncated extracellular lectin-like domain and binds intracellularly to DAP12, is still unknown (16). Although similarly named, the NKG2D molecule is only distantly related to the other members of the NKG2 family. NKG2D homodimers interact with stress-inducible MHC class I ligands MICA, MICB, ULBP1, ULBP2, ULBP3, and ULBP4 in humans (17, 18, 19) and RAET1A, RAET1B, RAET1G, RAET1E, H60, and MULT1 in mice (20, 21). NKG2D signaling involves the ITAM-containing adaptor molecules DAP10 and DAP12. Notably, distinct splice products of NKG2D associate with DAP10 and DAP12 in mice (22, 23), but not in humans, where the NKG2D gene is not differentially spliced and its product associates with DAP10 only (24).

The CD94 gene is monomorphic and the NKG2 genes are moderately polymorphic in humans and common chimpanzees, and all genes are subject to alternative splicing (25, 26). In this study, we show the complete sequence and organization of the CD94-LY49L genomic interval in a New World monkey species, the common marmoset (Callithrix jacchus). The NKC of the common marmoset monkey encodes only a single activating CD94/NKG2 receptor, which most likely represents an ancestral form of primate-activating receptors CD94/NKG2C and CD94/NKG2E. Compared with humans, the NKG2 and CD94 genes of the marmoset are more variable.

Blood samples (1 ml each) were obtained from nine healthy unrelated common marmoset monkeys (C. jacchus) that are maintained in the German Primate Centre breeding colony. Samples were brought to 15 ml with erythrocyte lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃, and 0.1 mM EDTA) and were incubated at room temperature for 20 min. The remaining blood cells were centrifuged for 10 min at 200 × g and 7°C, washed in 15 ml of lysis buffer, and subjected to preparation of total RNA according to standard methods (27).

High-density colony filters of BAC library CHORI-259 were obtained from Children’s Hospital Oakland Research Institute (http://bacpac.chori.org/home.htm). Filters were hybridized with probes of the human CD94, NKG2A, and LY49L genes according to recommendations in filter-accompanying sheets. Probes for these genes were obtained by PCR with primers hCD94-5′-TGTTTAAGACCACTCTGTGG and hCD94-3′-CTTTCTGGAGTTCTATGTTG, hNKG2A-5′-AGTACAGTCCCTGACATCAC, hNKG2A-3′-AGGGAATAACAACTATCGTT as well as hLy49L-5′-CAGTGCCCTGGCACCTCATT and hLy49L-3′-TCTCTATGACATCTGTTCTT, respectively. Briefly, 100 ng of human genomic DNA, 0.3 μM each of forward and reverse primer, 166 μM of each dNTP, and 1 U BioTherm TaqDNA polymerase (Genecraft) were used for PCR. Cycling conditions were: initial denaturation at 94°C for 2 min, 30 cycles of 94°C for 1 min, 51°C for 1 min, and 72°C for 1 min, with a final elongation of 5 min at 72°C. Reactions were performed in Applied Biosystems 2700 (Applied Biosystems) and LabCycler (Sensoquest) thermal cyclers. Hybridization of the BAC filters as well as preparation of BAC DNA and Southern blotting was performed essentially as described in the study of Walter et al. (28).

For the identification of exon 2 of the marmoset LY49L gene, primer Ly49-ex2fwd1 (Table I) derived from to the human LY49L exon 2 sequence was used along with primer Caja-Ly49-ex3rev (Table I) derived from the exon 3 sequence of the marmoset LY49L gene and BAC DNA of clone 162P15. The respective PCR product was sequenced to confirm its specificity.

BAC clone DNA was purified by CsCl density centrifugation and isolated DNA was sheared by sonification. DNA fragments of 1.5–3.5 kb were selected and cloned into plasmid pUC19. Inserts were subsequently amplified by PCR using insert-flanking M13 forward and reverse primers. Amplificates were sequenced with Applied Biosystems BigDye terminator chemistry and analyzed in Applied Biosystems 3730 sequencers (Applied Biosystems). Raw sequences were processed by Phred (available from Phil Green, University of Washington (www.phrap.org)) and assembled into a contiguous sequence by Phrap (available from Phil Green, University of Washington (www.phrap.org)) (29). BLAST and FGENESH-2 algorithms (http://www.ncbi.nlm.nih.gov/BLAST/; http://www.softberry.com/all.htm) identified exons and introns of genes in finished BAC clone sequences and annotations of genes were done manually. All BAC clone sequences have “finished sequencing” quality (1 putative mistake/100,000 bases) and the data have Phred values of >90. No gaps are present in the sequence.

Two micrograms of total RNA was reverse transcribed for 1 h at 42°C using oligo(dT) primer and 200 U of Moloney mouse leukemia virus reverse transcriptase (Promega) in a total volume of 25 μl. For PCR, 1 μl of cDNA was used in standard PCR according to the different primers (Table I) and expected lengths of the amplificates. PCR products were separated in 1% agarose gels and visualized with ethidium bromide staining. Bands were excised under UV light and DNA was recovered using standard silica adsorption. Isolated DNA was sequenced with respective PCR primers and Applied Biosystems BigDye terminator chemistry. Alternatively, RT-PCR products were cloned in a pGEM-T Easy PCR cloning vector according to the supplier’s manual (Promega) and single clones were sequenced.

The 5′ ends of cDNAs were experimentally determined by 5′RACE with the GeneRacer kit (Invitrogen Life Technologies), which was used according to the recommendations of the manufacturer. PCR products obtained with various primers (Table I) were cloned in the pCR4-TOPO vector (Invitrogen Life Technologiers) and at least 10 clones/gene were sequenced.

NKC gene sequences from various primates as well as mouse (outgroup) were extracted from GenBank (Table II) and were aligned by ClustalX software (http://www.embl.de/∼chenna/clustal/darwin/) (30) at the amino acid level, and the alignment was imposed on the DNA sequences and adjusted manually. Phylogenetic trees were reconstructed using the neighbor-joining method (31) based on the following distances: 1) the number of nucleotide substitutions per site estimated by the logdet method (32); 2) dN; and 3) the number of amino acid replacements per site estimated by the JTT model (33). The reliability of branching patterns in phylogenetic trees was assessed by bootstrap analysis (1000 pseudo samples) as implemented in the MEGA software, version 3.1 (http://www.megasoftware.net/) (34).

Screening of the marmoset BAC library with probes for human genes CD94, NKG2A, and LY49L identified more than a dozen BAC clones. We analyzed 12 BAC clones that reacted with more than one probe (Fig. 1). Notably, clones 188C15, 191M1, 196P12, and 467G5 reacted with all three probes, indicating that the CD94-LY49L genomic interval in the common marmoset monkey is present on single BAC clones and, thus, is more compact compared with the corresponding human region. Based on sequencing of a CD94 PCR product (data not shown), BAC clones 188C15, 196P12, 209K11, 467G5, and 44K16 could be assigned to one haplotype and BACs 191M1 and 61L5 to a further haplotype. A single clone from each haplotype was chosen for complete sequencing. Clones 188C15 and 191M1 encompass 186,125 bp and 228,774 bp, respectively. Southern blot analysis as well as inspection of the clone sequences identified the following genes on both haplotypes: CD94, NKG2D, NKG2F, NKG2CE, NKG2A, and LY49L (Fig. 1). The order and transcriptional orientation of the genes are the same as in humans. Marmoset NKG2F represents a pseudogene due to the presence of a preliminary stop codon in exon 1. BAC clones 191M1 and 188C15 both end at exactly the same site in intron 2 of LY49L, and exon 3 of this gene could be identified. However, LY49L exons 4–7 (for exon numbering see Ref35) are obviously deleted since they are not included in the sequenced BAC clones. Exon 2, which harbors the translational start codon, could be identified by PCR in the overlapping BAC clone 162P15 (data not shown). Sequencing of this PCR product identifies a frameshift mutation in exon 2. Therefore, LY49L represents a fragmentary pseudogene in the common marmoset monkey genome. In conclusion, the marmoset monkey and human CD94-LY49L intervals are colinear, except for the NKG2C and NKG2E loci, which is a single gene in the marmoset that we have designated NKG2CE according to phylogenetic analysis (see below).

The deduced amino acid sequences of the marmoset CD94, NKG2A, NKG2CE, and NKG2D genes revealed the typical structures of type II membrane proteins (Fig. 2). Marmoset NKG2A represents an inhibitory receptor according to the presence of two ITIMs in the cytoplasmic region, whereas NKG2CE and NKG2D are activating receptors due to lack of ITIMs and presence of a lysine and an arginine residue, respectively, in the transmembrane region (Fig. 2). Similar to other mammalian species, the marmoset CD94 molecule does not contain any of these functionally relevant motifs or amino acid residues. These data suggest that the NKC of the common marmoset monkey encodes a single inhibitory (CD94/NKG2A heterodimer) and two activating (CD94/NKG2CE heterodimer and NKG2D homodimer) NK cell receptors.

Specific primers were constructed based on the genomic sequence of the marmoset CD94, NKG2A, NKG2CE, NKG2D, and NKG2F genes determined here. RNA was extracted from PBMC of nine unrelated marmoset individuals and was used for RT-PCR analysis. Full-length transcripts could be obtained from all genes analyzed (Fig. 3,A) except for the NKG2F pseudogene, which could not be amplified (data not shown). Faint bands of shorter size were also visible (Fig. 3,A) and could be verified as alternatively spliced transcripts upon cloning and sequencing. Several alternatively spliced transcripts were found for the marmoset CD94 gene (Fig. 3). Thus, CD94B (splice variant 1) shows an additional codon (CAG) due to the presence of an alternative splice acceptor site at the 5′ end of exon 5 (Fig. 3,B). Such a splice variant was also found in humans, chimpanzees, and the rhesus macaque (26, 36). Further splice variants of the marmoset CD94 gene could be identified that lack exons 3, 4, or 5 (coding for the stalk or part of the lectin-like domain, respectively) and were further diversified by inclusion of the additional CAG codon at the 5′ end of exon 5 (Fig. 3,B). A cryptic splice donor site was used in splice variants 6, 7, and 8, reducing the length of exon 4 from 152 to 90 bp (Fig. 3,B). No splice variant lacking the transmembrane-coding exon as described in humans, chimpanzees, and rhesus macaques could be identified for the marmoset monkey CD94 gene. Altogether, 11 splice variants were found besides the full-length CD94 transcript (Fig. 3,B). A stalk-lacking CD94 isoform that corresponds to marmoset splice variants 2 and 3 (Fig. 3 B) was found in humans to be expressed on the cell surface along with either NKG2A or NKG2B and transmitting inhibitory signals (37). The conservation of the stalk-lacking CD94 isoforms in common marmosets and humans points to potential biological significance. Although in the stalk-lacking isoforms the open reading frame of the carbohydrate recognition domain does not change, this is not the case for the marmoset CD94 splice variants 4–11. Frameshifts lead to premature termination in these variants, resulting in truncated proteins that lack substantial or even any part of the carbohydrate recognition domain. Additionally, splice variants 4, 5, and 10 do not encode amino acid residue Cys⁵⁸, which is expected to participate in the interchain disulfide bond between CD94 and NKG2 (38, 39). Therefore, splice variants 4–11 are probably not able to pair with NKG2. The truncated isoforms might either recognize ligands other than MHC class I molecules or are simply not functional.

For the inhibitory NKG2A gene, two splice variants were found in addition to the full-length transcript (Fig. 3). Variant 1 uses an alternative splice acceptor site at the 5′ end of exon 2. This does not change the open reading frame, as it affects only the 5′ untranslated region. Splice variant 2 lacks exon 3 that codes for the transmembrane region. The transcriptional variant NKG2B, which lacks exon 4 (stalk) and is present in humans, chimpanzees, and rhesus macaques, could not be found in marmoset monkeys (Fig. 3 B). Thus, NKG2B either might not be expressed in marmosets or the NKG2B expression was down-regulated, similarly to what has been demonstrated in rhesus macaque individuals (40).

The activating NKG2D gene exhibited two splice variants (Fig. 3). Variant 1 lacks the transmembrane-coding exon (Fig. 3,B) and variant 2 lacks exons 3–7 (Fig. 3 B). A biological function of the latter appears rather unlikely because most parts of the NKG2D protein is absent. Transmembrane-deleted isoforms of NKG2A and NKG2D have also been found in rhesus macaques, suggesting some biological significance of these products of alternative splicing (36). Because these molecules cannot be expressed on the cell surface due to the absence of the transmembrane domain, they probably reside in the cytosol and may act to regulate the expression of the full-length receptor, similarly to what has been proposed for transmembrane-lacking CD94 and NKG2 isoforms in humans (41) and rhesus macaques (42).

In contrast to other primate species, the activating NKG2CE gene of the marmoset monkey does not show any alternatively spliced transcripts (Fig. 3). The AluJb repeat, which is partially used as exon 7 in human NKG2E, is also present in the marmoset and maps ∼3.1 kb downstream of the NKG2CE exon 6. Because no NKG2E-orthologous gene is present in the marmoset NKC, we wondered whether this particular “exon 7” might be included in the NKG2CE mRNA. However, a mutation of the splice acceptor site (AG to GG) of “exon 7” is present in the marmoset (data not shown). Consistent with this observation, no RT-PCR product could be obtained with primers cjNKG2CE-fw and cjNKG2E-rev (data not shown).

The cytoplasmic and transmembrane regions essentially determine the inhibitory or activating function of the NK receptors. The respective coding exons of the type II receptors analyzed here are located at the 5′ end of the genes. Differentially spliced transcripts of the mouse (but not humans) Nkg2d gene code for receptors that can associate with either the DAP10 or the DAP12 adaptor molecule (22, 23). Thus, we determined the 5′ ends of the NKC gene transcripts experimentally by 5′RACE. The marmoset NKG2D gene transcripts originate from different transcriptional start points (Fig. 4,A). Transcriptional start codons (ATG) could be identified upstream of the canonical ATG. However, the translated open reading frames starting from these upstream ATG codons show premature terminations and they do not translate into a proper NKG2D molecule. Similar to humans, the marmoset monkey does not show a NKG2D isotype with a shorter cytoplasmic tail and, therefore, is expected to associate with DAP10 only. Interestingly, only one additional untranslated exon could be identified for marmoset NKG2D transcripts, as opposed to up to three untranslated exons in humans (43). Variable transcriptional start sites were also found for the marmoset CD94, NKG2CE, and NKG2A genes (Fig. 4, B–D). No longer or shorter cytoplasmic regions are evident for CD94 and NKG2CE. A putatively longer open reading frame can be constructed for NKG2A, but no known functionally relevant amino acid motif (e.g., ITIM) is evident. Yet, it is unknown whether translation indeed starts from this upstream ATG codon.

The two allelic BAC sequences were analyzed for the occurrence of differences. A total of 200 polymorphic sites was identified in the overlapping sequences (186,125 bp) and include 148 single nucleotide polymorphisms (SNP), 7 dinucleotide polymorphisms, 31 insertions/deletions, and 14 complex polymorphisms (supplementary Fig. 1⁶). No significant clustering of polymorphic sites was observed in coding sequences. These polymorphisms can be used for analysis of genetic associations of the NKC in future studies with marmosets as animal models of human diseases.

The two CD94 alleles (CD94*01, CD94*02) present in the BAC clonal contig encompass 179 and 180 codons, respectively. An additional threonine residue (position 46) is present in the stalk region (Table III) of the CD94*02 allele identified in BAC clone 191M1. The stalk region has been shown to be important for dimer formation (38, 44). Thus, the additional threonine residue in the marmoset CD94 stalk might effect pairing with NKG2 molecules. Further allelic substitutions were identified in the nine marmoset monkeys that were analyzed from our breeding colony: two synonymous substitutions in codons 7 and 22 and three nonsynonymous substitutions in codons 72, 79, and 123 (Table III). Codon 123 is part of the putative ligand binding site (39) and, hence, polymorphism might affect ligand binding. Thus, similar to the orangutan (45), the CD94 gene is rather polymorphic in the common marmoset monkey compared with humans and chimpanzees where CD94 is strictly monomorphic (25, 26).

For the inhibitory NK receptor gene NKG2A, 5 SNPs were identified in the 10 common marmoset monkey individuals (9 from our colony and 1 from the BAC library). Interestingly, all SNPs are nonsynonymous (Table III) and map to exons encoding the cytoplasmic region, stalk and the carbohydrate recognition domain (CRD). A nonsynonymous SNP is found for codon 199 of marmoset NKG2A (Table III). Interestingly, amino acid residue 199 is part of the putative HLA-E-binding surface (39). Thus, this polymorphism might change the binding affinity to the putative ligand of marmoset NKG2A, the Caja-E molecule. The two ITIMs in the cytoplasmic region of marmoset NKG2A are not affected by allelic substitutions. Only a single nonsynonymous SNP is known in human NKG2A, but the majority of SNPs map to introns and untranslated regions (25). Two synonymous and two nonsynonymous NKG2A SNPs were identified in four common chimpanzee individuals (26) and seven NKG2A alleles were found in five orangutans (45). In rhesus macaques, NKG2A is considerably polymorphic (36). Consistent with this NKG2A polymorphism, the rhesus macaque class Ib gene Mamu-E is known to be polymorphic as well (46). This suggests that the NKG2A gene is diverse in certain Old World and New World primates and might be subject to positive (diversifying) selection.

Four SNPs of the activating NKG2CE gene were identified in the 10 common marmoset individuals, of which 3 are nonsynonymous and map to the stalk and CRD (Table III). No clues for amplification/deletion of the NKG2CE gene were obtained. Two nonsynonymous SNPs were reported for the human NKG2C gene and one synonymous and one nonsynonymous SNP for the human NKG2E gene (26, 47). Interestingly, ∼4% of the Japanese and Dutch population exhibit a homozygous NKG2C deletion (25, 48) and the chimpanzee NKG2C gene is duplicated (26). Four alleles of Pt-NKG2CI and two of Pt-NKG2CII were described in four chimpanzee individuals and five Pt-NKG2E alleles in three individuals. Only a single NKG2CE gene was described in the orangutan and five alleles were identified in five individuals (45). Similar to NKG2A, also the NKG2C gene is diverse in the rhesus macaque (36).

The activating NKG2D gene shows a low level of polymorphism in humans (26) and orangutans (45) and no polymorphism in chimpanzees. Similarly, no polymorphism of NKG2D could be noticed in the marmoset monkeys in our analysis (Table III).

The NKC genomic organization (Fig. 1) and phylogenetic analyses (data not shown) indicate orthologous relationships of primate CD94, NKG2D, and LY49L genes. In contrast, the phylogenetic relationship is less clear for primate NKG2 genes due to duplication, deletion, and homogenization of genes. Gene tree analysis performed with the cytoplasmic, transmembrane, and stalk regions of primate NKG2 amino acid sequences showed separate branching of NKG2A and NKG2C/NKG2E genes (Fig. 5,A). Interestingly, the marmoset NKG2CE gene clustered at the bases of the NKG2C/NKG2E branch, suggesting that it may represent an ancestral form of NKG2C and NKG2E genes of higher primates. In contrast, a tree based on the lectin-like domain showed branching by species and not by loci (Fig. 5 B). This implies that lectin-like domains have been homogenized within species by interlocus recombination (“gene conversion”). Furthermore, within-species homogeneity of lectin-like domains suggests that they may bind to the same ligand, which is probably the Caja-E class I molecule in the case of the marmoset CD94/NKG2A and CD94/NKG2CE molecules. This latter finding of a more species-specific evolution of the lectin-like domain is in accord with the evolution of its binding partner, the MHC class I ligands, which evolve rapidly (49, 50).

The common marmoset (C. jacchus) is a Neotropical primate that belongs to the New World monkeys (Platyrrhini). Due to their small size and similar physiology to humans, marmoset monkeys are increasingly used as nonhuman primate models in biomedical research, e.g., in studies of infectious (51) and autoimmune diseases (52) as well as in transplantation research (53). Marmosets usually born twins or even triplets. Interestingly, the offspring are natural bone marrow chimeras because their blood system is connected due to placental vascular anastomoses (54, 55). Yet, despite its importance in immunological research, only very few data on NK cell receptors are known for the common marmoset monkey (56), and for New World monkeys in general.

The corresponding human and marmoset CD94-LY49L intervals comprise 274 and 171 kb, respectively. This 1.5-fold difference is due to the absence of one NKG2 gene in the marmoset and to larger intergenic regions in humans, particularly between the NKG2A and LY49L genes. Nevertheless, the NKC regions of humans and marmosets are colinear and the genes are orthologous, suggesting similar functions of human and marmoset NKC genes. Exceptions are the human activating NKG2C and NKG2E genes, which are paralogs of the activating NKG2CE gene. The marmoset NKG2CE gene branches basal to all catarrhine primate NKG2C and NKG2E genes in phylogenetic trees and, thus, might represent an ancestral form of Old World primate NKG2C and NKG2E genes. It has to be noted that the activating NKG2CE gene has the same designation in the orangutan and marmoset. However, the orangutan NKG2CE gene is not an ortholog of marmoset NKG2CE and probably represents an intermediate between the chimpanzee and human NKG2C and NKG2E genes (45). Thus, repeated gene duplications and deletions have shaped the present-day NKC genomic organization in primates. The reason for this more plastic NKC gene content in nonhuman primates compared with humans may lie in the different MHC class I gene contents, which is particularly evident between rhesus macaques and humans (57), or in a different contribution of NKC-encoded molecules to resistance against pathogens in nonhuman primates.

The LY49L gene of the marmoset is fragmentary and is regarded as a pseudogene. This is similar to other primates such as the humans, chimpanzee, and gorilla, where LY49L is assumed to be nonfunctional as well (35, 58). In contrast, a correctly spliced mRNA with an intact open reading frame was observed in the baboon and orangutan and, therefore, LY49L is likely functional in these primates (45, 58). Notably, the LY49L mutations are different in the human/chimpanzee/gorilla compared with the common marmoset monkey, indicating that inactivation of LY49L occurred repeatedly in primate evolution. Thus, it might be interesting to analyze in future studies whether the presence or absence of LY49L molecules correlates with the absence or presence of other NK cell receptors.

Besides NKG2D, the common marmoset monkey possesses only a single-activating NK cell receptor gene in the NKC. This is different from most catarrhine primates, which have several activating NKC-encoded NK cell receptors. Thus, the genetic diversity in terms of NKC gene numbers appears to be reduced in the common marmoset monkey. From the LRC, only a single KIR gene, KIR3DL0, has so far been described in common marmosets and appears to represent an ancestral inhibitory KIR gene in primates (56). It is unknown so far whether marmosets have more KIR genes in addition to KIR3DL0 and whether the overall genetic equipment of NK cell receptors is rather low in this New World monkey species. Interestingly, the ligands of NK receptors, the MHC class I molecules, exhibit also a low degree of diversity in the subfamily Callitrichidae (tamarins and marmosets) (59, 60). Because NK cell receptors have to keep up with the rapidly evolving MHC class I ligands, it might not be surprising to find a confined NK cell receptor gene repertoire in a species with limited MHC diversity. The question then arises: how do these primates cope with pathogens? It has been reported that captive cotton-top tamarins and common marmoset monkeys are prone to certain viral and bacterial infections, allowing for the implementation of these nonhuman primates as appropriate disease models in the study of human infectious diseases (51). Yet, whether such disease associations do occur also in free-living tamarins and marmosets is largely unknown (51). Our data indicate that the marmoset CD94, NKG2A, and NKG2CE genes show a moderate degree of allelic diversity among a total of 10 individuals analyzed. Due to the naturally occurring bone marrow chimerism in tamarins and marmosets, theoretically up to four different alleles of a single gene can be identified in bone marrow-derived cells of a single individual. It appears plausible that this chimerism allows for a more efficient use of the set of alleles or genes present in a population, thereby counteracting restricted diversity at MHC and NK cell receptor loci. The availability of the genomic sequence and SNPs reported here are helpful in future studies to elucidate potential associations of NKC-encoded genes and infectious and autoimmune diseases using the marmoset monkey as model.

Expression analysis of the NKC genes in the common marmoset revealed the presence of alternatively spliced transcripts of the CD94, NKG2A, and NKG2D genes, but not of NKG2CE. Although products of differential splicing have already been described for NKC genes in other primate species (26, 36, 40, 42), the complexity of marmoset CD94 alternative transcripts is surprising. Besides the well-known CD94B alternative transcript that has an additional CAG codon (Fig. 3 B, glutamine residue) in the CRD region, two CD94 isoforms might be expressed that lack the stalk region. Lieto et al. (37) have described corresponding CD94 isoforms in humans. Interestingly, in vitro analyses indicated a preferential association of both stalk-lacking isoforms of human CD94 (CD94-T4) and NKG2A (NKG2B), which could functionally inhibit TCR-mediated signals in Jurkat cells (37). It has been speculated that these stalk-lacking isoforms might segregate to other surface microdomains than the full-length CD94/NKG2A receptor, adding a further level of complexity to the inhibitory NK cell immunological synapse. As a result of the lack of suitable isoform-discriminating Abs, it remains unknown so far whether the alternative isoforms of the marmoset monkey are indeed expressed as proteins in vivo. However, stalk-lacking isoforms are identified in humans (37) and Old World (36, 40) and New World monkeys (this article) and, thus, are evolutionarily conserved for >43 million years (61). Further support for functional significance of alternatively spliced transcripts comes from studies that revealed substantial variability in mRNA expression of NKG2 genes lacking the stalk and/or the transmembrane region. The expression level of these isoforms was variable in different rhesus macaques as well as in a single rhesus macaque individual surveyed over a period of 1.5 years (40).

Functionally relevant alternative splicing was found at the 5′ end of the mouse Nkg2d transcripts, which code for NKG2D that associate differently with either the DAP10 or DAP12 adaptor proteins (22, 23). 5′RACE analysis of the marmoset NKG2D gene did not identify such alternatively spliced transcripts and it is therefore hypothesized that the marmoset monkey NKG2D molecule associates with DAP10 only.

In conclusion, our analyses of the NKC in a New World monkey species, the common marmoset, revealed a condensed size of this genomic region, but a colinear and conserved genomic organization. The NKC-encoded NK cell receptor genes are moderately polymorphic in the common marmoset monkey and the transcripts of CD94, NKG2A, and NKG2D, but not the NKG2CE gene, are alternatively spliced. Experimental determination of the 5′ end of all studied marmoset NKC gene transcripts indicated variable transcriptional start sites, with no functional changes to be expected.

We are grateful for the expert technical assistance of Christiane Schwarz and Nico Westphal.

The authors have no financial conflict of interest.