MHC class I molecules play a central role in both innate and adaptive immunity. In addition to presenting peptides derived from intracellular antigens on the cell surface for recognition by CD8⁺ T cells, MHC class I molecules serve as ligands for killer cell Ig-like receptors (KIRs) on natural killer (NK) cells (1, 2). Whereas the extensive polymorphism of classical MHC class I molecules ensures the presentation of a diverse repertoire of peptides derived from intracellular pathogens to CD8⁺ T cells, the limited polymorphism of nonclassical MHC class I molecules reflects their more specialized functions.

Macaques and other Old World monkeys express an expanded array of polymorphic MHC class I genes that correspond to the classical HLA-A and -B genes of humans (3–5). However, these species lack an ortholog of HLA-C as a consequence of the recent evolutionary origin of HLA-C as a duplication of an ancestral MHC-B gene that occurred after the divergence of apes and Old World monkeys (3, 6). Old World monkeys have accordingly expanded their lineage II KIR, which encode KIR3D receptors for MHC-A and -B, but lack lineage III KIR that encode KIR2D receptors for HLA-C in humans (7–13). Old World monkeys also have orthologs of each of the human nonclassical HLA-E, -F and -G genes. MHC-E and -F are well-conserved and broadly expressed in primates (14–16). However, the MHC-G genes of Old World monkeys have accumulated multiple stop codons and no longer encode functional proteins (17, 18). In the absence of a functional MHC-G locus, Old World monkeys have evolved another nonclassical MHC class I gene, designated MHC-AG, which appears to serve a similar function (19–22).

Macaques, vervets, and baboons express MHC-AG molecules with similar molecular features and tissue distribution as HLA-G (22, 23). MHC-AG sequences are more similar to MHC-A than to MHC-G as a consequence of their evolutionary origins as an MHC-A gene duplication (19–22). However like HLA-G, MHC-AG exhibits limited polymorphism, a truncated cytoplasmic tail, and is only expressed on placental trophoblasts that form the barrier between the mother and the developing fetus (19, 20, 23). Hence, Mamu-AG is believed to serve a similar function as HLA-G, namely contributing to pregnancy success through interactions with NK cells and myeloid cells that promote placental vascularization, tolerance to the haploidentical fetus, and protect the maternal-fetal interface from invading pathogens (23, 24).

The rhesus macaque (Macaca mulatta) is an important animal model for infectious diseases such as HIV/AIDS, CMV, Zika virus and SARS-CoV-2 (25–35), transplantation (36, 37), and reproductive biology (23, 38). We previously identified Mamu-AG as a ligand for KIR3DL05 (39). Here we show that KIR3DL05 recognizes certain allotypes of Mamu-AG, but not others, and define residues in the α1- and α2-domains of Mamu-AG essential for interactions with this receptor. We further identify Mamu-AG as a ligand for the activating receptor KIR3DS02. These observations provide an essential foundation for investigating the role of NK cells in the rhesus macaque as a nonhuman primate model for infectious diseases and reproductive biology. These findings also illustrate how polymorphisms in the α1-domain can lead to variation in KIR recognition of closely related MHC class I ligands.

Definition of the MHC class I ligands of rhesus macaque KIRs is foundational for studying NK cell responses in this species as an animal model for infectious diseases, transplantation and reproductive biology. In a previous survey of KIR-MHC class I interactions in the rhesus macaque, Mamu-AG2*01:01 was identified as a ligand for KIR3DL05*008 (39). Although Mamu-AG does not have a direct human ortholog, it shares similar molecular features and tissue distribution as HLA-G (19, 20, 23). However, Mamu-AG is considerably more polymorphic than HLA-G owing to allelic variation at multiple loci (44). KIR3DL05 is also polymorphic with at least 33 different KIR3DL05 alleles identified so far (45). To gain a better understanding of the determinants of KIR3DL05 specificity for Mamu-AG, we investigated the impact of Mamu-AG polymorphisms on this receptor-ligand interaction.

Our findings reveal three general patterns of Mamu-AG recognition among the KIR3DL05 allotypes selected for this study. The most common pattern exhibited by KIR3DL05*004, *005 and *008 includes dominant responses to Mamu-AG2*01:01 and lower, more variable responses to Mamu-AG3*03:01, -AG3*03:02, -AG3*03:03 and -AG3*03:04. The nearly identical responses of KIR3DL05*004 and *008 reflect the overall similarity of their extracellular domains, which only differ by a single amino acid in the D1 domain (Q139H). KIR3DL05*005 differs from each of these KIRs by seven residues in D0 domain and two in the D1 domain. However, only one of these polymorphisms (Y29H) is located in a predicted MHC class I contact site. In contrast to the other allotypes, KIR3DL05*001 responded strongly to Mamu-AG1*05:01, weakly to Mamu-AG2*01:01, and not at all to Mamu-AG3. Since KIR3DL05*001 and *008 share identical D1 and D2 domains, but differ by nine amino acids in the D0 domain, including three residues in predicted MHC class I contact sites, this shift in Mamu-AG recognition is a function of polymorphisms in the D0 domain. KIR3DL05*X exhibited a more complex pattern that included dominant responses to both Mamu-AG2*01:01 and -AG1*05:01, and detectable responses to Mamu-AG3*03:01, -AG3*03:02, -AG3*03:03 and -AG3*03:04. Whereas KIR3DL05*X differs from KIR3DL05*008 by only two amino acids in the D0 domain, there are thirteen differences in the D1 domain. The more extensive differences in D1 reflect the hybrid origins of this KIR, which is comprised of the leader peptide, D0 and D1 domains of KIR3DS02 (exons 1-4) and the D2, stem, transmembrane and cytoplasmic domains of KIR3DL05 (exons 5-9) (40). Overall, these observations reveal differences in KIR3DL05 recognition of Mamu-AG molecules that are determined by polymorphisms in their D0 and D1 domains.

We also tested an allotype of KIR3DS02 with identical D0 and D1 domains to KIR3DL05*X. KIR3DS02*004 recognized Mamu-AG2*01:01, but did not respond to other Mamu-AG allotypes. These results identify Mamu-AG2*01:01 as a ligand for this activating KIR and indicate that differences in the D2 domains of KIR3DS02*004 and KIR3DL05*X account for differences in their responses to other Mamu-AG molecules.

An analysis of single amino acid variants identified Mamu-AG polymorphisms that affect KIR3DL05 recognition. These include differences in the α-helical region of the α1 domain (R75Q & V76E), an N-terminal loop of the α1 domain (W18G & E19Q), and the β-sheet that forms the floor of the peptide binding pocket (S11A & I95F). Residues 18, 19, 75 and 76 are located on the same face of the α1 domain and coincide with predicted KIR contact sites (41). Thus, the effects of polymorphisms at these positions may be attributed to interactions with KIR3DL05. On the other hand, residues 11 and 95 are not predicted to contact the KIR, but may alter the conformation of the bound peptide or adjacent residues in the α1 domain in a manner that indirectly affects interactions with KIR3DL05. These observations illustrate how a few amino acid differences in surfaces of the α1 domain can result in differential recognition of closely related MHC class I molecules and are consistent with the co-evolution of the D0 and D1 domains of KIRs in concert with the α1 domain of MHC class I ligands.

In humans, HLA-G is expressed by fetal trophoblasts that invade the uterine endometrium during early stages of pregnancy and on extravillous trophoblasts that remodel spiral arteries of the developing placenta (46). HLA-G is a ligand for KIR2DL4 (47, 48), which is an unusual 2 domain KIR with both activating and inhibitory features. KIR2DL4 has a positively charged arginine residue in the transmembrane domain for pairing with FcεRIγ to transduce activating signals and a long cytoplasmic tail with an immunotyrosine-based inhibitory motif (ITIM) (49). This receptor interacts with a soluble isoform of HLA-G in endosomes to stimulate the release of pro-inflammatory and pro-angiogenic factors by decidual NK cells that promote placental vascularization (50–52). Although the presence of HLA-E on extravillous trophoblasts is thought to play a dominant role in maintaining fetal tolerance via inhibitory interactions with CD94/NKG2A on decidual NK cells, HLA-G may facilitate tolerance through additional interactions with inhibitory receptors and by serving as a source of leader peptides bound by HLA-E (52, 53).

The ortholog of HLA-G (Mamu-G) in rhesus macaques is a pseudogene (17). However, macaques express another nonclassical MHC class I molecule with similar tissue distribution (19). Like HLA-G, Mamu-AG is expressed on fetal cytotrophoblasts that invade the endometrium and on syncytiotrophoblasts of the placental chorionic villi (23). While it has been difficult to experimentally verify the function of HLA-G in humans, passive transfer of a Mamu-AG-specific monoclonal antibody to pregnant macaques was shown to disrupt leukocyte responses to implantation, vascularization, and placental development (38). Thus, Mamu-AG appears to serve a similar function as HLA-G.

The identification of Mamu-AG as a ligand for KIR3DL05 and KIR3DS02 reveals that certain macaque KIRs are capable of mediating functional interactions with this nonclassical molecule. While macaques express a recognizable ortholog of KIR2DL4 (KIR2DL04) (7, 54), there is no reason to believe that this receptor interacts with Mamu-AG, which bears no homology to HLA-G. Nevertheless, it is likely that there are other receptors for Mamu-AG that are yet to be identified. These may include additional KIRs or other types of receptors that react more broadly with MHC class I molecules, such as the Ig-like transcripts 2 and 4 (ILT-2 and ILT-4) that are known to serve as receptors for HLA-G in humans (55).

A complex and precarious balance of NK cell activating and inhibitory receptor interactions is essential for normal placental development, fetal tolerance, and immune surveillance of the maternal-fetal interface. Precisely how KIR3DL05 and KIR3DS02 interactions with Mamu-AG contribute to this balance in macaques is presently unclear. The ligation of activating receptors on decidual NK cells may stimulate the release of pro-inflammatory and pro-angiogenic factors that promote placental vascularization and increase blood supply to the developing fetus, similar to the interactions of KIR2DL4 with soluble HLA-G or lineage III KIRs (KIR2Ds) with HLA-C in humans (52, 56). In the absence of HLA-G or HLA-C, it is conceivable that macaques depend on Mamu-AG recognition by activating KIRs such as KIR3DS02 to stimulate placental vascularization. Mamu-AG may also facilitate fetal tolerance through interactions with inhibitory KIRs such as KIR3DL05 or by providing leader peptides that stabilize Mamu-E on the surface of trophoblasts for inhibitory signaling through CD94/NKG2A (57).

Recent studies have demonstrated a role for decidual NK cells in protecting the maternal-fetal interface. Extravillous trophoblasts are susceptible to a number of pathogens that are transmitted in utero, including Zika virus (ZiKV), cytomegalovirus (CMV), and Toxoplasma gondii (58–63). In the case of ZiKV, infection of extravillous trophoblasts causes ER stress that results in the downregulation of HLA-C and -G, which renders these cells susceptible to killing by decidual NK cells (60). Similar antimicrobial functions of decidual NK cells have been reported for CMV and Listeria monocytogenes (64–66). Thus, disruption or alteration of Mamu-AG expression on macaque trophoblasts may trigger decidual NK cell responses through inhibitory or activating KIRs. KIR3DL05 and KIR3DS02 may thereby contribute to pregnancy success by protecting the maternal-fetal interface from invading microorganisms. Furthermore, polymorphic differences in these KIRs and Mamu-AG could account for individual variation in resistance to placental transmission of viral and bacterial pathogens.

The rhesus macaque has become an increasingly valuable animal model for studying viral pathogens such as CMV and Zika virus that are frequently transmitted to the developing fetus during pregnancy (67–70). Hence, the molecular interactions between macaque NK cell receptors and their MHC class I ligands that occur at the maternal-fetal interface may be especially important for understanding viral transmission and pathogenesis in this context. The present study defines many of the Mamu-AG ligands recognized by diverse allotypes of KIR3DL05 and identifies KIR3DS02*004 as an activating receptor for Mamu-AG2*01:01. These findings afford greater insight into NK cell recognition of a nonclassical molecule implicated in pregnancy and provide a broader foundation for investigating NK cell biology in the rhesus macaque as a nonhuman primate model for infectious diseases and reproductive biology.

The GenBank accession numbers for the sequences presented in this study are provided in the Materials and Methods. All of the data presented in this study are included in the article. Further inquires can be directed to the corresponding author.

RN, JA, WS, MW, and PB performed the experiments and analyzed the data. KS and SJ designed the experiments. DE conceived the project and wrote the manuscript. All of the authors contributed to the manuscript and approved the submitted version.

This work was supported by Public Health Service Grants AI095098, AI161816, AI121135, AI148379 and AI098485 (to DE) and OD011106 (to the WNPRC).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.