KIR Polymorphisms Modulate Peptide-Dependent Binding to an MHC Class I Ligand with a Bw6 Motif

Molecular interactions between killer immunoglobulin-like receptors (KIRs) and their MHC class I ligands play a central role in the regulation of natural killer (NK) cell responses to viral pathogens and tumors. Here we identify Mamu-A1*00201 (Mamu-A*02), a common MHC class I molecule in the rhesus macaque with a canonical Bw6 motif, as a ligand for Mamu-KIR3DL05. Mamu-A1*00201 tetramers folded with certain SIV peptides, but not others, directly stained primary NK cells and Jurkat cells expressing multiple allotypes of Mamu-KIR3DL05. Differences in binding avidity were associated with polymorphisms in the D0 and D1 domains of Mamu-KIR3DL05, whereas differences in peptide-selectivity mapped to the D1 domain. The reciprocal exchange of the third predicted MHC class I-contact loop of the D1 domain switched the specificity of two Mamu-KIR3DL05 allotypes for different Mamu-A1*00201-peptide complexes. Consistent with the function of an inhibitory KIR, incubation of lymphocytes from Mamu-KIR3DL05+ macaques with target cells expressing Mamu-A1*00201 suppressed the degranulation of tetramer-positive NK cells. These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selectivity of KIRs for MHC class I-bound peptides, and identify the first functional KIR-MHC class I interaction in the rhesus macaque. The modulation of KIR-MHC class I interactions by viral peptides has important implications to pathogenesis, since it suggests that the immunodeficiency viruses, and potentially other types of viruses and tumors, may acquire changes in epitopes that increase the affinity of certain MHC class I ligands for inhibitory KIRs to prevent the activation of specific NK cell subsets.


Introduction
Natural killer (NK) cells are able to lyse infected or malignant cells without prior antigenic stimulation, and thus provide an important innate defense against infectious agents and tumors [1,2]. NK cell activation in primates is regulated in part through interactions between the highly polymorphic killer immunoglobulin-like receptors (KIRs) expressed on NK cells and their MHC class I ligands on target cells [1,2]. KIRs are type I integral membrane proteins with either two or three immunoglobulin (Ig)like extracellular domains (2D or 3D) that transduce either inhibitory or activating signals via long (L) or short (S) cytoplasmic domains, respectively. Engagement of inhibitory KIRs by MHC class I molecules on healthy cells normally suppresses NK cell activation [1,3,4]. However, if these interactions are perturbed, for instance as a result of MHC class I downregulation by HIV-1 Nef [5,6], or presentation of a peptide antagonist [7], this inhibition is lost resulting in NK cell activation and target cell lysis.
In contrast to the T cell receptor, which is highly specific for a given peptide-MHC complex, KIRs typically recognize subsets of MHC class I molecules with common amino acid motifs in their a1 domains. Based on serological epitopes that correspond to defined sequences at positions 77-83, all HLA-B molecules, and some HLA-A molecules, can be classified as either Bw4 or Bw6 allotypes [8]. Allotypes of KIR3DL1 have broad specificity for HLA-Bw4 ligands [9], whereas KIRs specific for HLA-Bw6 have not been identified. All inhibitory KIRs that have been examined thus far also exhibit selectivity for peptides bound by their MHC class I ligands [10,11,12,13,14,15,16]. These observations are consistent with crystal structures of KIR2DL1 and KIR2DL2 in complex with their HLA-C ligands showing that KIR residues contact surfaces of the HLA class I a1 and a2 domains in an orthogonal orientation across C-terminal residues of the bound peptide [17,18]. However, the molecular basis for the selectivity of KIRs for different peptides bound by a particular MHC class I ligand has not been defined.
Genetic evidence suggests that polymorphic differences in the KIR and HLA class I genes play an important role in determining the course of infection for a number of human viral pathogens, including HIV-1 [19,20], hepatitis C virus [21], human papillomavirus [22] and cytomegalovirus [23]. In the case of HIV-1, combinations of both activating and inhibitory KIR3DL1/S1 and HLA-Bw4 alleles have been associated with delayed progression to AIDS [19,20]. NK cells expressing KIR3DS1 were also shown to suppress the in vitro replication of HIV-1 in target cells expressing HLA-Bw4 [24]. While these observations point to a role for KIR-MHC class I interactions in determining the outcome of HIV-1 infection, studies to address the functional significance of these interactions have been limited, in part, by the lack of a suitable animal model.
Simian immunodeficiency virus (SIV) infection of the rhesus macaque is an important animal model for lentiviral pathogenesis and for AIDS vaccine development [25]. Rhesus macaques express MHC class I molecules that correspond to products of the classical HLA-A and -B genes (Macaca mulatta; Mamu-A and -B), but not the HLA-C gene [26,27]. Consistent with the co-evolution of KIR and MHC class I molecules, genes for the two-domain KIRs specific for HLA-C have not been identified in macaques [28,29]. Instead, macaques have an expanded repertoire of KIR3DL genes characterized by extensive polymorphism and gene duplication [28,29,30,31,32].
Here we identify Mamu-A1*00201, a common rhesus macaque MHC class I molecule with a Bw6 motif, as a ligand for multiple allotypes of Mamu-KIR3DL05. We show that the binding of Mamu-KIR3DL05 to Mamu-A1*00201 is peptide-dependent, and that the relative avidity and peptide-selectivity of binding is determined by polymorphisms in the D0 and D1 domains. We also demonstrate that target cells expressing Mamu-A1*00201 suppress the degranulation of primary Mamu-KIR3DL05 + NK cells. These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selective recognition of MHC class I-bound peptides by KIRs, and define the first functional KIR-MHC class I interaction in the rhesus macaque.

Results
Peptide-dependent tetramer staining of primary rhesus macaque NK cells Samples of peripheral blood from Mamu-A1*00201 + rhesus macaques were stained with Mamu-A1*00201 tetramers folded with SIV peptides to establish baseline CD8 + T cell responses prior to beginning a vaccine study. To our surprise, Mamu-A1*00201 in complex with the Gag 71-79 GY9 peptide stained a subset of CD8 + CD3lymphocytes from one animal (Mm 337-07). Plasma from this animal tested negative for SIV RNA and for antibodies to viral antigens, indicating that this animal had not been previously exposed to SIV. The majority of tetramerpositive cells expressed CD8a and CD16, characteristic of NK cells that are capable of mediating cytolytic activity [33], as well as additional NK cell markers including NKp46, NKG2A, and NKG2D (Fig. 1A). A subset of these cells also cross-reacted with an antibody to human KIR2D (Fig. 1A). Although most of the tetramer-positive cells were CD16 + CD3 -NK cells, staining was also observed for CD8 + CD3 + T cells (Fig. 1B). A longitudinal comparison of the frequency of tetramer-positive CD8 + CD3 + versus CD16 + CD3lymphocytes revealed that these two populations were relatively stable in this animal over more than a year, ranging from 0.16% to 0.69% for CD8 + T cells and from 5.1% to 9.8% for CD16 + NK cells. To investigate the contribution of the peptide bound by Mamu-A1*00201 to this unusual pattern of tetramer staining, whole blood was stained with Mamu-A1*00201 tetramers folded with peptides corresponding to eight different CD8 + T cell epitopes of SIV [34]. In addition to Gag 71-79 GY9, staining was also observed for Env 788-795 RY8, but not for any of the other tetramers (Fig. 1C). Thus, the tetramer staining observed for primary NK cells and CD8 + T cells from Mm 337-07 was dependent on the peptide bound by Mamu-A1*00201.
Identification of Mamu-KIR3DL05 as a receptor for Mamu-A1*00201 Since KIRs are known to be expressed on subsets of human NK cells and CD8 + T cells [4,35,36], we hypothesized that this pattern of tetramer staining might reflect Mamu-A1*00201 binding to a rhesus macaque KIR. Full-length KIR cDNA sequences were therefore cloned from the PBMCs of Mm 337-07 and sequenced. Six KIR3DL alleles, three KIR3DS alleles and two KIR2DL04 alleles were identified in this animal, and their predicted amino acid sequences are shown in Fig. 2.
To identify the receptor for Mamu-A1*00201, Jurkat cells were transfected with constructs expressing each of the KIR alleles cloned from Mm 337-07 and stained with Mamu-A1*00201 tetramers. To differentiate transfected from untransfected cells, the KIR alleles were expressed from a bicistronic vector that coexpresses enhanced green fluorescent protein (eGFP). Since not all KIRs are well expressed on the cell surface, and antibodies are not available to macaque KIRs, an HA tag was introduced at the N-terminus of the D0 domain of each KIR. Our rationale for introducing the HA tag at this position is based on a recent threedimensional model showing that the N-terminus of KIR3DL1 is free and oriented away from surfaces that are predicted to contact the peptide-MHC class I complex [37], and experiments demonstrating that the introduction of an epitope tag at the N-terminus of the D0 domain does not interfere with ligand recognition [38]. Following the electroporation of Jurkat cells with these KIR expression constructs, the cells were stained with Mamu-A1*00201 tetramers and with a monoclonal antibody to the HA tag. Transfected cells were identified by gating on the Author Summary NK cells provide an important first line of defense against infectious diseases and tumors by virtue of their ability to kill infected or malignant cells without prior sensitization. NK cell activation is regulated in part through interactions between KIRs expressed on the surface of NK cells and their MHC class I ligands on target cells. Here we identify Mamu-A1*00201 (Mamu-A*02), a common MHC class I molecule in the rhesus macaque, as a ligand for Mamu-KIR3DL05. We show that this interaction is peptidedependent, since soluble Mamu-A1*00201 tetramers folded with certain SIV peptides, but not others, stained cells expressing Mamu-KIR3DL05. Differences in binding avidity were associated with polymorphisms in the D0 and D1 domains of Mamu-KIR3DL05, whereas differences in peptide-specificity mapped to the D1 domain. These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selectivity of KIRs for MHC class I-bound peptides, and identify the first functional KIR-MHC class I interaction in the rhesus macaque. These observations suggest that SIV, and potentially also HIV-1, may acquire changes in epitopes that increase the avidity of MHC class I ligands for inhibitory KIRs as a mechanism of immune evasion to prevent the activation of certain NK cell subsets.
eGFP + population, the surface expression of each KIR was verified by HA staining, and binding to Mamu-A1*00201 in complex with Gag 71-79 GY9 versus Nef 159-167 YY9 was assessed by tetramer staining.
All of the KIRs were expressed on the cell surface under the conditions of this assay, as indicated by HA staining (Fig. 3). However, only Mamu-KIR3DL05*008 resulted in a detectable level of staining with the Gag 71-79 GY9 tetramer (Fig. 3A). At higher levels of surface expression, staining was also observed for Nef 159-167 YY9, indicating that this tetramer can bind to Mamu-KIR3DL05*008 under conditions of protein over expression (Fig. 3A). These results identify Mamu-KIR3DL05*008 as a receptor for Mamu-A1*00201, and indicate that the peptide bound by Mamu-A1*00201 can modulate this interaction.
Mamu-KIR3DL05*003, -KIR3DL05*008 and -KIR3DL05*010, were indistinguishable in their pattern of tetramer staining (Fig. 4A). This is reflected by the similarity in their values for the mean fluorescence intensity (MFI) of tetramer staining divided by the MFI of HA staining, which are provided in Table 1 as a quantitative comparison of tetramer binding corrected for differences in surface expression for each KIR. In accordance with the rank order of tetramer staining observed for primary NK cells (Fig. 1C), staining was highest for Gag 71-79 GY9, followed by Env 788-795 RY8, and then Nef 159-167 YY9 ( Fig. 4A and Table 1). With the exception of a single amino acid difference in the first position of the D0 domain of Mamu-KIR3DL05*010, each of these KIRs have identical Ig-like domains (Fig. 4B).
Relative to Mamu-KIR3DL05*008, Mamu-KIR3DL05*004 exhibited an increase in the intensity of tetramer staining for Gag 71-79 GY9 (1.3 fold), Env 788-795 RY8 (2.0 fold) and Nef 159-167 YY9 (4.3 fold) ( Fig. 4A and Table 1). Since the Ig-like domains of Mamu-KIR3DL05*004 and Mamu-KIR3DL05*008 only differ by a single amino acid at position 138 ( Fig. 4B), the histidine residue at this position accounts for the increase in Mamu-KIR3DL05*004 binding to Mamu-A1*00201. Based on a recently proposed three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 [37], this residue is predicted to lie at the base of the second MHC class I-contact loop of the D1 domain, and may alter the conformation of this loop in a way that enhances binding to Mamu-A1*00201.
In the case of mmKIR3DL05x, tetramer staining was observed for Nef 159-167 YY9, but not for Gag 71-79 GY9 or Env 788-795 RY8 (Fig. 4A). This shift in the pattern of Mamu-A1*00201 tetramer staining almost certainly reflects differences in D1, since mmKIR3DL05x has a unique D1 domain, but nearly identical D0 and D2 domains to other allotypes of Mamu-KIR3DL05 (Fig. 4B). Using cryopreserved PBMCs from the original source of mmKIR3DL05x, we verified that mmKIR3DL05x represents a bona fide allele, and not a PCR artifact, by independently cloning and confirming the cDNA sequence for this allele, and by PCR amplification of a 2.0 kb region spanning intron 4 from genomic DNA with primers to unique sequences in exons 4 and 5. Additional sequence comparisons revealed that the D1 domain of mmKIR3DL05x, as well as the leader peptide and the D0 domain, are identical to Mamu-KIR3DS02*00402 and mmKIR3DHa (Fig. S2). Thus, mmKIR3DL05x appears to be the product of a recombination event in which exon 4 (encoding D1) was acquired, either by the introduction of exons 1-4 of a Mamu-KIR3DS gene into -KIR3DL05 or by the introduction of exons 5-9 of Mamu-KIR3DL05 into a -KIR3DS gene.
A closer examination of mmKIR3DL05x revealed that seven of the thirteen differences in the D1 domain coincide with, or are immediately adjacent to, loops predicted to contact surfaces of the peptide-MHC class I complex [37]. These include a charge difference at position 144 in the second loop (L2) and a cluster of six residues at positions 164-170 in the third loop (L3) (Fig. 4B and Fig. 5A). To determine if these differences account for the unique binding pattern exhibited by mmKIR3DL05x, we constructed recombinants in which these sequences were exchanged with the corresponding sequences of Mamu-KIR3DL05*008, and tested them for binding to Gag 71-79 GY9 versus Nef 159-167 YY9. Reciprocal L2 substitutions affected the avidity, but not the specificity, of tetramer binding (Fig. 5B). In contrast, exchanging L3 residues switched the specificity, and altered the avidity, of binding to the Mamu-A1*00201 tetramers. The 3DL05*008/xL3 recombinant bound Nef 159-167 YY9, but not Gag 71-79 GY9, and the 3DL05x/*008L3 recombinant bound both Gag 71-79 GY9 and Nef 159-167 YY9 (Fig. 5B). Hence, these results reveal a role for polymorphisms in the third predicted contact loop of the D1 domain in determining the selective recognition of different peptides bound by the same MHC class I molecule. tetramers. The KIRs were expressed from a bicistronic vector designed to introduce a common leader peptide followed by an HA tag at the Nterminus of the D0 domain, and to co-express eGFP from a downstream internal ribosomal entry site. Jurkat cells were electroporated with the KIR expression constructs and stained the following day with APC-conjugated Mamu-A1*00201 tetramers, followed by a PE-conjugated antibody to the HA tag. Tetramer versus HA staining was determined after gating on the eGFP + cell population. Quadrant gates were set using empty vectortransfected controls stained with tetramer and antibody to the HA tag. doi:10.1371/journal.ppat.1001316.g003 Additional Mamu-KIR3DL05 + rhesus macaques were identified by sequence-specific PCR and screened for tetramer-positive NK cells and CD8 + T cells. The Gag 71-79 GY9 tetramer stained subsets of CD8 + CD3and CD8 + CD3 + lymphocytes in in peripheral blood from each of the Mamu-KIR3DL05 + animals, but not from Mamu-KIR3DL05animals ( Fig. 6A and 6B). In accordance with the complex regulation of KIR expression, which is influenced by a number of factors including differences in gene content on different KIR haplotypes, differences in the repertoire of MHC class I genes and polymorphic differences in KIR genes [40,41,42,43], there was considerable animal-to-animal variation in the frequency and intensity of tetramer staining (Fig. 6B). Variation in the frequency of tetramer-positive CD8 + CD3 + lymphocytes, particularly in Mamu-A1*00201 2 animals that do not have Gag 71-79 -specific CD8 + T cells, may also reflect changes in KIR expression on memory CD8 + T cells related to age and/or prior exposure to infectious agents, since similar changes have been associated with age and encounters with viral pathogens in humans [36,44,45]. Overall, these results demonstrate that the presence of the Mamu-KIR3DL05 gene is predictive of Gag 71-79 GY9 staining in peripheral blood, that this pattern of tetramer staining is independent of Mamu-A1*00201 and SIV infection, and that the variability of staining is typical of the heterogeneity of KIR expression on human NK cells and CD8 + T cells [36,46].
The role of NK cells and CD8 + T cells that express Mamu-KIR3DL05 in SIV-infected animals remains to be determined. However, among the eight animals represented in Fig. 6B, there were no obvious differences in the percentage of tetramer-positive cells for either population that could be attributed to the status of SIV infection. Indeed, of the two uninfected animals (Mm 177-05 and Mm RHAX18), the two animals infected with attenuated SIV mac 239 Dnef (Mm 350-04 and Mm 376-04), and the two animals infected with pathogenic SIV mac 239 (Mm R03035 and Mm 20-05), each pair had among the lowest and the highest frequencies of tetramer-positive lymphocytes (Fig. 6B). Some of the tetramer-positive CD8 + CD3 + lymphocytes in the Mamu-A1*00201 + animals probably represent virus-specific CD8 + T cells, since we cannot differentiate binding of the Gag 71-79 GY9 tetramer to Mamu-KIR3DL05 versus the T cell receptor. Nevertheless, in two of the three SIV-infected animals (Mm 350-04 and Mm R02020), the percentage of tetramer-positive cells was actually higher for the CD8 + CD3 2 population than for the CD8 + CD3 + population (Fig. 6B). Although the explanation for this is presently unclear, it is possible that Mamu-KIR3DL05 interactions with Mamu-A1*00201 may suppress CD8 + T cell responses to the Gag 71-79 GY9 epitope in Mamu-KIR3DL05 + animals, which could explain the inconsistent, and often weak, CD8 + T cell responses to Gag 71-79 GY9 that we and others have observed in SIV-infected, Mamu-A1*00201 + macaques.
To investigate the functional consequences of NK cell recognition of Mamu-A1*00201, PBMC from four Mamu-KIR3DL05 + animals were stimulated with the MHC class Ideficient 721.221 cell line [47], or with 721.221 cells expressing either Mamu-A1*00201, -A1*01101 (Mamu-A*11) or -B*010101 (Mamu-B*01), and stained for CD107a as a degranulation marker. The cells were also stained with Gag 71-79 GY9 to differentiate Mamu-KIR3DL05 + NK cells from Mamu-KIR3DL05 2 NK cells. CD107a was upregulated on the surface of both tetramer-positive and tetramer-negative NK cells in response to parental 721.221 cells and 721.221 cells expressing Mamu-A1*01101 or -B*010101 (Fig. 6C). In contrast, CD107a was suppressed on tetramerpositive NK cells, but not on tetramer-negative NK cells, in the presence of target cells expressing Mamu-A1*00201 (Fig. 6C). The same pattern of NK cell activation/inhibition was also observed by intracellular cytokine staining for IFNc (Fig. S3). Moreover, CD107a was suppressed on tetramer-positive NK cells from both Mamu-A1*00201 + and -A1*00201 2 animals (Fig. 6C), indicating that these cells were responsive to Mamu-A1*00201 whether or not they were educated in the presence of this MHC class I molecule. These results are therefore consistent with the functional inhibition of Mamu-KIR3DL05 + NK cells by Mamu-A1*00201.

Discussion
Polymorphic differences in the KIR and HLA class I genes play an important role in determining the course of infection for HIV-1 and for a number of other viral pathogens [19,20,21,22,23,24]. However, studies to address the functional significance of KIR-MHC class I interactions have been hampered by the lack of a suitable animal model. In the present study, we identify Mamu-A1*00201, an MHC class I molecule present in approximately 20% of Indian origin rhesus macaques [48], as a ligand for multiple allotypes of Mamu-KIR3DL05. Although the frequency independent of Mamu-A1*00201, reflecting the segregation of KIR and MHC class I genes on different chromosomes, and was detectable regardless of the status of SIV infection. Moreover, the variability in the frequency and intensity of tetramer staining among Mamu-KIR3DL05 + animals was typical of the heterogeneity of KIR expression on human NK cells and CD8 + T cells [36,46]. Although tetramer staining has been reported for NK cell clones and for transfected cells expressing human KIRs [11,12], to our knowledge this is the first report of direct ex vivo tetramer staining of primary NK cells.
Incubation of peripheral blood lymphocytes from Mamu-KIR3DL05 + macaques with target cells expressing Mamu-A1*00201 specifically suppressed the degranulation of tetramerpositive NK cells. These results are consistent with the functional inhibition of primary NK cells expressing Mamu-KIR3DL05 by  [37]. The residues indicated in yellow are located in surface-exposed loops in close proximity to the bound peptide. The residues indicated in red represent differences in the D1 domain at sites that do not contribute directly to interactions with MHC class I ligands. The residues highlighted in magenta represent positions 77-83 of the a1 domain corresponding to the Bw6 of Mamu-A1*00201. (B) Jurkat cells were electroporated with constructs expressing recombinants of Mamu-KIR3DL05*008 and mmKIR3DL05x, for which residues of the second (L2) and third (L3) predicted MHC class I-contact loops in D1 were exchanged, and stained with APCconjugated tetramer (Gag 71-79 GY9 or Nef 159-167 YY9) followed by a PE-conjugated antibody to the HA tag. Tetramer versus HA staining was determined after gating on the eGFP + cell population and quadrant gates were set using control cells transfected with an empty vector. doi:10.1371/journal.ppat.1001316.g005 Mamu-A1*00201. Furthermore, this inhibition was observed for tetramer-positive NK cells from Mamu-A1*00201 2 as well as from Mamu-A1*00201 + animals, indicating that these cells were responsive to Mamu-A1*00201, whether or not they were educated in animals that express this ligand. Although the mechanisms of NK cell education are not fully understood [52], there is evidence that the maturation of NK cells expressing inhibitory KIRs is dependent on interactions with self-MHC class I molecules, and that NK cells expressing a particular inhibitory KIR in the absence of an appropriate MHC class I ligand are rendered hyporesponsive [3,53,54]. Thus, the in vitro suppression of tetramer-positive NK cells from Mamu-A1*00201 2 animals by target cells expressing Mamu-A1*00201 implies that these cells were educated for recognition of another MHC class I ligand. This is perhaps not surprising given the complexity of the rhesus macaque MHC class I genes [55,56], and the ability of KIRs to recognize multiple MHC class I ligands with common amino acid motifs in their a1 domains [9,57].
Based on haplotype modeling and phylogenetic comparisons, Mamu-KIR3DL05 is predicted to represent a single genetic locus [29]. Although KIR3DL05 is not orthologous to any of the human KIR genes, interactions between Mamu-KIR3DL05 and Mamu-A1*00201 resemble features of KIR3DL1 binding to HLA-Bw4. A three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 was recently constructed based on a crystal structure of KIR2DL1 in complex with HLA-C*04 [17,37]. This model predicts that surface-exposed loops in each of the three Ig-like domains of KIR3DL1 contact the HLA class I molecule over the C-terminus of the bound peptide, and that the specificity of KIR3DL1 for HLA-Bw4 is dependent on a salt bridge between glutamate 282 in the D2 domain of KIR3DL1 and arginine 83 in the a1 domain of HLA-Bw4 [37]. Consistent with this model, polymorphisms in the Ig-like domains of Mamu-KIR3DL05 were associated with differences in binding to Mamu-A1*00201.
Polymorphisms in D1 altered the selective binding of Mamu-KIR3DL05 to Mamu-A1*00201 in complex with different SIV peptides. In contrast to other allotypes of Mamu-KIR3DL05, mmKIR3DL05x preferentially bound to Mamu-A1*00201 folded with Nef 159-167 YY9 rather than Gag 71-79 GY9. This difference in peptide preference mapped to six amino acids in the third D1 loop predicted to contact surfaces of the peptide-MHC class I complex. These results support a recent three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 [37], and reveal a role for polymorphisms in the D1 domain in determining the selectivity of KIRs for MHC class I-bound peptides. Interestingly, mmKIR3DL05x appears to be the product of a recombination event in which exon 4 sequences coding for the D1 domain were derived from a KIR3DS gene; an observation that is consistent with domain shuffling as a mechanism of KIR evolution in primates [58].
Unlike previously identified ligands for human KIRs, the a1 domain of Mamu-A1*00201 contains a Bw6 motif. In contrast to Bw4, the Bw6 motif has a glycine rather than an arginine at position 83 (N 77 LRNLRG 83 ). Yet, Mamu-KIR3DL05 retains a glutamate at position 285, which corresponds to glutamate 282 of KIR3DL1. Since the peptides recognized by Mamu-KIR3DL05 each contain a positively charged residue at position 6 or 8 (Gag 71-79 GSENLKSLY, Env 788-795 RTLLSRVY and Nef 159-167 YTSGPGIRY), it is conceivable that glutamate 285 may form an alternative salt bridge with the peptide that accounts for the peptide-dependence of Mamu-KIR3DL05. However, a charge at this position does not appear to be sufficient for binding, since the Vif 89-97 IW9 peptide, which also contains a lysine at position 6 (ITWYSKNFW), did not result in detectable Mamu-A1*00201 tetramer staining. While the molecular interactions underlying the binding of Mamu-KIR3DL05 to Mamu-A1*00201 remain to be fully defined, these observations offer a potential explanation for the contribution of the peptide to this interaction, and perhaps suggest a more prominent role for certain peptides in KIR recognition of other Bw6 ligands.
The extent to which KIR recognition of Bw6 ligands has been elaborated in the rhesus macaque is presently unclear. However, since this motif is retained in the MHC class I molecules of humans and macaques, the absence of human KIRs that recognize HLA-Bw6 appears to reflect the loss of receptors of this specificity during the course of human evolutionary history. While the reason for this is not understood, it may be related to the expansion of the lineage III KIR genes coding for KIR2DL/S receptors with a D1-D2 configuration, and a greater dependence on the regulation of NK cell activation through interactions with their HLA-C ligands.
The identification of inhibitory KIRs that bind with high avidity to a common MHC class I molecule in the rhesus macaque in complex with SIV-derived peptides suggests a potential mechanism of immune evasion. The Nef proteins of HIV-1 and SIV selectively downregulate MHC class I molecules from the surface of infected cells to evade destruction by virusspecific CD8 + T cells [6,59]. However, the removal of these molecules from the cell surface increases the susceptibility of infected cells to elimination by NK cells [6]. By acquiring changes in CD8 + T cell epitopes that increase the binding of MHC class I ligands to inhibitory KIRs, the virus may prevent the activation of NK cells under conditions of incomplete downregulation by Nef. This possibility is supported by recent evidence that peptides can modulate NK cell activation by varying the affinity of HLA ligands for inhibitory KIRs [7]. Whereas Fadda et al. show that antagonistic peptides that disrupt MHC class I interactions with inhibitory KIRs leads to NK cell activation [7], our data suggests that viruses may acquire changes in epitopes that stabilize these interactions to suppress NK cell activation in a way that favors virus replication.
KIRs are also expressed on subsets of memory CD8 + T cells in HIV-1 infected individuals, and have been associated with a decrease in the responsiveness to TCR-dependent stimulation [44,60]. Thus, peptides that stabilize interactions with inhibitory KIRs may also suppress CD8 + T cell activation. Deleterious combinations of KIR and MHC class I alleles may therefore select for changes in epitopes of HIV-1 and SIV that inhibit certain NK cell and CD8 + T cell responses; a scenario that may further undermine the host's ability to contain virus replication. Consistent with this hypothesis, a single nucleotide polymorphism was recently identified as a marker for two Mamu-KIR3DL05 alleles that were more prevalent among SIV-infected rhesus macaques with high viral loads in animals [61].
The identification of Mamu-A1*00201 as a ligand for Mamu-KIR3DL05 now affords an opportunity to investigate the functional implications of KIR-MHC class I interactions. Using KIRand MHC class I-defined animals, experiments can now be designed to examine the phenotypic changes that occur in a specific population of NK cells during the course of virus infection in a way the was previously only possible for CD8 + T cells. Characterization of the molecular interactions underlying the binding of Mamu-KIR3DL05 to Mamu-A1*00201 also promises to yield fundamental insights regarding the role of viral peptides in modulating KIR recognition of MHC class I ligands. The binding of Mamu-KIR3DL05 to Mamu-A1*00201 in complex with SIV peptides suggests that these interactions may be particularly important in determining the course of SIV infection.

Ethics statement
All of the animals used for these studies were Indian origin rhesus macaques (Macaca mulatta). These animals were housed at the New England Primate Research Center (NEPRC) and were maintained in accordance with standards of the Association for Assessment and Accreditation of Laboratory Animal Care and the Harvard Medical School Animal Care and Use Committee. Animal experiments were approved by the Harvard Medical Area Standing Committee on Animals and conducted according to the principles described in the Guide for the Care and Use of Laboratory Animals [62].

KIR nomenclature and Genbank accession numbers
Rhesus macaque KIR sequences were submitted to Genbank and to the Immuno-Polymorphism Database (www.ebi.ac.uk/ipd/ kir/) [63]. Sequences that have been assigned official names are indicated with the prefix Mamu-KIR. In cases where official names have not yet been assigned, sequences are referred to using a provisional nomenclature indicated by the prefix mmKIR. The names and Genbank accession numbers for each of the KIR alleles in this study are listed in Table S1.

Cloning and sequencing of rhesus macaque KIR alleles
Peripheral blood lymphocytes were isolated over Ficoll (Sigma) and aliquots of 2-10 million PBMC were frozen in Trizol (Invitrogen). Total RNA was extracted using the RNeasy kit (Qiagen) according to the manufacturer's instructions. KIR cDNAs were amplified by reverse transcription-polymerase chain reaction (RT-PCR) using the Superscript III One-Step RT-PCR kit (Invitrogen) with modified versions of the Ig3Up and Ig3Down primers [64]. Cycling conditions included an RT step at 55uC for 30 min, a denaturation step at 94uC for 2 min, followed by 40 cycles of denaturation (94uC for 15 sec), annealing (55uC for 30 sec) and extension (68uC for 90 sec), and a final extension step at 68uC for 5 min. PCR products were cloned into the pGEM-T Easy vector (Promega) and sequenced with T7 and SP6 sequencing primers. Sequences were analyzed using Sequencher 4.8 (Gene Codes Inc.) and MacVector 9.5.2 (MacVector Inc.) software packages. At least three identical cDNA clones were identified for each KIR allele.

KIR expression and tetramer staining of transfected Jurkat cells
Rhesus macaque KIRs were PCR amplified from cDNA clones using primers to introduce an HA tag at the N-terminus of the D0 domain. The KIR cDNAs were then cloned into pCGCG, a bicistronic vector that co-expresses eGFP, in frame with an upstream sequence for the leader peptide of Mamu-KIR3DL05*008. Jurkat cells (1610 7 cells) were electroporated (250V, 975mF) with plasmid DNA (40 mg) in serum-free RPMI (400 ml) in a 0.4 cm cuvette (BioRad). After resting (10 min, 20uC), the cells were re-suspended in RPMI medium (9 ml) with 10% FBS and incubated overnight at 37uC, 5% CO 2 . After 22 hours, the cells were stained with APC-conjugated tetramers (30 min, 37uC), followed by PE-conjugated anti-HA PE (GG8-IF3.3, Miltenyi Biotec Inc.) (20 min, 20uC). The cells were washed and fixed in 2% paraformaldehyde PBS. At least 200,000 events were acquired using a FACSCalibur flow cytometer (BD Biosciences) and the data was analyzed using FlowJo 8.8.6.

Mamu-KIR3DL05 genotyping
Genomic DNA was extracted from 1-2 million PBMC using the DNAeasy kit (Qiagen, Valencia, CA), and 10 ng was used as template in a 25 ml PCR reaction with forward and reverse primers (GAGACCCATGAACTTAGGCTTC & GCAGTG-GGTCACTGGGGA) for amplification of a 156 bp sequence in exon 5 specific to Mamu-KIR3DL05. Primers specific for a conserved 300 bp region of Mamu-DRB were included as an internal control [48]. Cycling conditions included a denaturation step at 96uC for 2 min followed by 30 cycles of denaturation (94uC for 30 sec), annealing (63uC for 45 sec) and extension (72uC for 45 sec), and a final extension step at 72uC for 10 min. PCR products were separated on a 1% agarose gel containing ethidium bromide and visualized by UV transillumination.
NK cell suppression by specific MHC class I molecules PBMC (1610 6 cells) were stimulated for 18 hours with 721.221 cells, or with 721.221 cells expressing rhesus macaque MHC class I molecules, at a 5:1 ratio in the presence of anti-CD107a PE-Cy5 (clone H4A3, BD Pharmingen), Golgi-Stop and Golgi-plug (BD Pharmingen). The cells were then stained with APC-conjugated tetramers (30 min, 37uC), followed by anti-CD16-FITC, anti-NKG2A-PE, anti-CD8a-Alexa 700 and CD3 APC-Cy7 (20 min, 20uC). The cells were then permeabilized and stained for 30 min with anti-IFN-c-PE-CY7 (Clone 4S.B3, BD Pharmingen). Samples were washed and fixed in 2% paraformaldehyde PBS. At least 200,000 lymphocyte events were collected using an LSRII flow cytometer, and the data was analyzed using FlowJo 8.8.6.  Figure S3 Target cells expressing Mamu-A1*00201 suppress the production of IFNc by tetramer-positive NK cells. PBMC from two Mamu-A1*00201 + and two Mamu-A1*00201macaques were incubated overnight at a 5:1 effector to target cell ratio with parental 721.221 cells, or with 721.221 cells expressing individual rhesus macaque MHC class I molecules. Following stimulation, the samples were stained with Gag 71-79 GY9 tetramer, followed by antibodies to CD3, CD8, CD16 and NKG2A. The samples were then fixed, permeabilized and stained with an IFNc-specific monoclonal antibody. After gating on CD3-NK2GA + lymphocytes, the frequency of tetramer-positive versus tetramer-negative NK cells expressing IFNc was determined.