The Rat RT1-A1^c MHC Molecule Is a Xenogeneic Ligand Recognized by the Mouse Activating Ly-49W and Inhibitory Ly-49G Receptors¹

Natural killer cells play a significant role in innate immunity against virally infected and transformed cells (1) and contribute to transplant rejection, particularly hybrid resistance to allogeneic and xenogeneic bone marrow transplants (2, 3). Effector functions of NK cells include lysis of pathologically altered cells and the production and secretion of cytokines (4). Murine NK cells express a number of activating and inhibitory receptors that regulate these effector functions, including members of the Ly-49 receptor family (5). Ly-49 receptors belong to the C-type lectin receptor superfamily and are expressed at the cell surface as disulfide-bonded homodimers (6). These type II transmembrane proteins are encoded as a Ly-49 multigene family on chromosome 6 in the NK gene complex, and inbred mouse strains can differ with respect to their complement of activating and inhibitory type Ly-49 genes (7, 8). Inhibitory Ly-49 receptors contain immunoreceptor tyrosine-based inhibitory motifs (ITIMs)³ in their cytoplasmic tails that become phosphorylated on tyrosine upon ligand recognition (9). The phosphorylated ITIMs recruit the Src homology 2 domain-containing tyrosine phosphatase-1, which disrupts membrane-proximal signaling events that would otherwise lead to NK activation (10, 11). Activating Ly-49 receptors do not possess ITIMs and, instead, contain a charged residue in their transmembrane domain that facilitates association with the immunoreceptor tyrosine-based activation motif-containing, signaling adapter molecule DAP12. Upon ligand engagement by the Ly-49 activating receptor, phosphorylation of tyrosine residues within the DAP12 immunoreceptor tyrosine-based activation motifs, leads to tyrosine kinase (Syk, Zap-70)-mediated signaling events and NK cell activation (12, 13, 14, 15).

Ly-49 receptors are more extensively characterized in the mouse than in the rat. In both species, inhibitory Ly-49 receptors recognize classical class I MHC products, allowing NK cells to detect cells with loss of class I expression, targeting them for destruction by coexpressed activating receptors (16, 17, 18, 19, 20, 21). Some activating Ly-49 receptors in the mouse recognize classical class I MHC molecules, leading to target cell destruction (21, 22, 23, 24, 25), whereas rat activating Ly-49 receptors recognize nonclassical class I molecules (26, 27, 28). In addition, an activating Ly-49 receptor that recognizes a mouse CMV-encoded ortholog of a class I MHC molecule, significantly contributes to mouse CMV resistance (29, 30, 31).

Mouse Ly-49 receptors are also capable of recognizing xenogeneic ligands. The activating Ly-49D receptor recognizes a hamster class I molecule and an unidentified MHC-encoded rat ligand of the F344 and LEW rat strains (32, 33). Furthermore, the activating Ly-49W and inhibitory Ly-49G receptors also recognize rat strain-specific xenogeneic ligands (34). Identification of xenogeneic ligands and characterization of the molecular basis of their recognition may provide additional insights into conserved or novel modes of Ly-49 receptor ligand recognition. Additionally, the identification of Ly-49 receptors and their xenoligands could aid studies investigating the role of NK cells in the generation and use of xenogeneic mixed chimeras for the induction of xenotransplantation tolerance (35). However, there are no reports of a specific rat xenogeneic ligand for any mouse Ly-49 receptor.

In this report, we identify RT1-A1^c, a rat classical class I MHC molecule, as a ligand for the closely related Ly-49W activating and Ly-49G^BALB/c inhibitory receptors. This xenogeneic class I recognition is demonstrated to be allele specific, because other class I molecules of the RT1^c haplotype and a classical class I molecule encoded by a different MHC haplotype, RT1^av1, are not recognized by Ly-49W and -G^BALB/c. Additionally, specificity for RT1-A^c can be transferred from Ly-49W to Ly-49P by substitution of three residues shared by Ly-49W and -G^BALB/c but not Ly-49P. These residues are located in the Ly-49 β4–β5 loop, which participates in mouse Ly-49 receptor interactions with mouse class I ligands (36), suggesting that mouse Ly-49 recognition of rat class I molecules follows similar principles of interaction. Preliminary analysis of conserved motifs shared by RT1-A1^c and mouse ligands of Ly-49W and -G^BALB/c implicate a conserved MHC supertype motif as a potential mediator of xenogeneic and syngeneic recognition.

Five- to 8-wk-old female DBA/2J (H-2^d) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Six- to 8-wk-old male and female PVG (RT1^c) and DA (RT1^av1) rats were purchased from Harlan Sprague-Dawley (Indianapolis, IN) and Harlan U.K. (Bicester, U.K.). Male and female intra-MHC recombinant rat strains PVG.R1, PVG.R19, and PVG.R20 were purchased at 6–8 wk of age from Dr. G. Butcher at Babraham Institute (Cambridge, U.K.). All animals were housed in approved animal care facilities in accordance with institutional guidelines.

The Cwy-3 (IgG1), 4D11 (rat IgG2a), and A1 (IgG2a) anti-Ly-49 hybridomas have been described (37, 38, 39). Isotype controls include BB7.1 (IgG1) anti-HLA-B7 (40), 53-5.8.3 (rat IgG2a) anti-murine Lyt-3 (41), and B27M1 (IgG2a) anti-HLA-B27, B7 (42). The Cwy-3 hybridoma was generated in this laboratory, the 53-5.8.3 hybridoma was kindly provided by Dr. L. Herzenberg (Stanford University, Stanford, CA), and all other hybridomas were obtained from American Type Culture Collection (Manassas, VA). Abs were prepared from ammonium sulfate precipitates as described (24).

YB2/0, a nonsecreting rat myeloma, was obtained from American Type Culture Collection and maintained in DMEM supplemented with 10% FCS, l-glutamine, penicillin, streptomycin, 1 mM sodium pyruvate, and 0.1 mM nonessential amino acids. RNK-16, a spontaneous F344 rat strain NK cell leukemia cell line (43), was provided by Dr. M. Nakamura at the University of California (San Francisco, CA), and maintained in RPMI 1640 supplemented with 10% FCS, l-glutamine, penicillin, streptomycin, and 5 × 10⁻⁵ M 2-ME. RNK-16 effector cells expressing mouse Ly-49 receptors were generated by this laboratory as described (24, 25, 34) and maintained under G418 selection until 48 h before cytotoxicity assays.

Ly-49W and Ly-49P, previously cloned in this laboratory from the nonobese diabetic mouse (24, 25), were mutated using the QuikChange Mutagenesis kit (Stratagene, La Jolla, CA). Mutagenic primers were designed to mutate amino acid residues N245, D247, and Q248 in Ly-49P to the corresponding residues found in Ly-49W, D250, G252, and K253. Inversely, residues D250, G252, and K253 in Ly-49W were mutated to the Ly-49P residues, N245, D247, and Q248. All mutations were verified by DNA sequencing. The cDNAs for the coding regions of the mutant Ly-49 constructs were inserted into the XhoI/XbaI sites of the mammalian expression vector BSRαEN (provided by Dr. A. Shaw (Washington University, St. Louis, MO)). RNK-16 cells were transfected with each construct as previously described (25), and maintained under G418 selection until 48 h before cytotoxicity assays.

Spleens harvested from PVG and DA rats were immediately immersed in RNAlater RNA stabilization reagent (Qiagen, Valencia, CA) and homogenized using a rotor-stator homogenizer, and total RNA was isolated using an RNeasy Protect mini-kit (Qiagen). cDNA was produced using Powerscript reverse transcriptase (Clontech, Palo Alto, CA) with an oligo(dT) primer.

H-2D^k, previously cloned in this laboratory (34), was directionally cloned, in frame, into the XhoI/XmaI sites of pEGFP-N2ML (provided by Dr. D. Burshtyn (University of Alberta, Edmonton, Canada)). The pEGFP-N2ML vector has the EGFP initiation codon mutated to ATA, creating a C-terminal EGFP-fusion protein. Because the majority of published rat class I MHC sequences do not include the leader sequence, silent mutations (QuikChange; Stratagene) creating an MluI site at the leader/coding region junction of H-2D^k allowed cloning of rat class I MHC coding sequences fused to this mouse class I leader sequence. Each rat class I MHC transcript, without leader sequence, was amplified with Advantage-HF 2 polymerase mix, digested with MluI/XmaI and ligated into the H-2D^k leader-EGFP fusion vector. Successful generation of in-frame H-2D^k leader-rat class I MHC-EGFP constructs was verified by DNA sequencing. YB2/0 cells were transfected with each construct as previously described (24) and maintained under G418 selection until 48 h before cytotoxicity assays.

Expression of rat class I MHC constructs in YB2/0 cells was determined by EGFP fluorescence intensity relative to untransfected YB2/0 cells using a FACScan flow cytometer (BD Biosciences, Mountain View, CA). Expression of mouse Ly-49 receptors on RNK-16 transfectants was monitored with Cwy-3, 4D11, and A1 Abs as previously described (24, 25, 34).

Con A-activated T cell blasts were prepared from spleens of DBA/2 mice and PVG and DA rat strains. Spleen cells were cultured at 5 × 10⁶ cells/ml in RPMI 1640 supplemented with 10% FCS, 2-ME, and 3 μg/ml Con A (Sigma-Aldrich, St. Louis, MO) for 48 h. Blast cells were recovered after washing in RPMI 1640 medium. No difference in blast formation or RNK-16 effector recognition was observed between spleen cells of male or female origin.

Target cells were labeled with 100–150 μCi of Na⁵¹CrO₄ (⁵¹Cr) (Mandel, Guelph, Canada) at 37°C for either 1 h (YB2/0 and YB2/0 transfectants) or 1.5 h (Con A blasts). Targets were washed extensively, and 1 × 10^{4 51}Cr-labeled cells were incubated with RNK-16 or transfected RNK-16 cells for 4 h at 37°C in V-bottom microtiter plates at various E:T ratios in triplicate. After incubation, supernatant samples were collected and counted in a MicroBeta TriLux liquid scintillation counter (PerkinElmer, Wellesley, MA). Percent specific lysis was determined as follows: (experimental release − spontaneous release)/(maximal release − spontaneous release) × 100%. In Ly-49-specific Ab-blocking assays, mAbs were incubated for 30 min with 1 μg of soluble protein A and protein G (Calbiochem, La Jolla, CA) for each microgram of mAb before addition to effector cells to prevent reverse Ab-dependent cell-mediated cytotoxicity. Effector cells were incubated with the mAb-protein A/G mixture for 30 min before addition of target cells. All cytotoxicity experiments were repeated a minimum of three times.

Ly-49D and Ly-49G^BALB/c recognize hamster and rat xenogeneic ligands (32, 33, 34). Features that are shared by or distinguish xenogeneic and mouse ligands may offer novel insights into the molecular basis of Ly-49 ligand recognition. Additionally, the identification of Ly-49 receptors and their xenogeneic MHC ligands could serve in the construction of xenogeneic mixed chimeras that do not require depletion of host NK cells in xenotransplant tolerance models (44). Ly-49D recognizes a xenogeneic ligand on Chinese hamster ovary cells (33, 45) that has been identified as the class I MHC molecule Hm1-C4 (32). As a xenogeneic ligand, Hm1-C4 has offered limited insight into Ly-49 recognition motifs, primarily due to the phylogenic distance between hamster and mouse, resulting in relatively low sequence identity between their respective class I MHC molecules (32). Furthermore, hamsters are not commonly used in studies of xenogeneic mixed chimerism. In contrast, the rat is more closely related to mouse and expresses class I MHC molecules that are relatively high in sequence identity to mouse MHC while displaying evolutionary distance, and rat/mouse xenogeneic mixed chimeras are a traditional model within transplantation studies. These properties and circumstances make rat a more informative partner to mouse in which to examine shared Ly-49 recognition motifs across species. The Ly-49D receptor mediates cytolysis of Con A blasts from both the F344 and LEW rat strains in an MHC haplotype-specific manner (33), but no specific xenogeneic ligand has been identified. Our laboratory has also demonstrated differential recognition of xenogeneic Con A blasts from the PVG and LEW rat strains by the Ly-49G inhibitory receptor (34). As recognition of rat Con A blasts by mouse Ly-49 receptors is strain specific, and no rat xenogeneic ligands have been definitively identified for mouse Ly-49 receptors, we set out to identify rat ligand(s) mediating the strain-specific recognition by Ly-49G^BALB/c and the related Ly-49W activating receptor.

RNK-16 cells, which are not intrinsically cytolytic toward Con A blasts of rat strains used (Fig. 1,A), were transfected with mouse Ly-49 receptors and tested for cytotoxicity against Con A blasts prepared from two rat strains expressing different MHC haplotypes, PVG (RT1^c) and DA (RT1^av1). The specificity of the Ly-49G^BALB/c inhibitory receptor for Con A blasts was examined through the use of a chimeric receptor, Ly-49WG^BALB/c, described previously (34). This chimeric activating receptor maintains the specificity of the Ly-49G^BALB/c inhibitory receptor ectodomain, while possessing the transmembrane and cytosolic portions of the activating receptor Ly-49W. As previously demonstrated, RNK-16 cells transfected with the chimeric Ly-49WG^BALB/c receptor, specifically recognize and lyse Con A blasts prepared from the PVG rat strain, which expresses MHC molecules of the RT1^c haplotype (Fig. 1,B). Two different activating Ly-49 receptors originally cloned from the nonobese diabetic mouse, Ly-49P and Ly-49W (24, 25), were also tested for their ability to trigger lysis of xenogeneic rat Con A blasts when transfected into RNK-16 cells. The ectodomain of Ly-49P closely resembles that of the inhibitory receptor Ly-49A, whereas that of Ly-49W resembles that of the inhibitory Ly-49G receptor. Ly-49W was also seen to trigger cytolysis of Con A blasts specifically from the PVG rat strain (Fig. 1,C). In contrast, Ly-49P was incapable of triggering cytolysis of blasts from the PVG rat strain, despite being competent in triggering lysis of Con A blasts from the control DBA/2 mouse strain (H-2^d) (Fig. 1 D) (24). Thus, Ly-49G and Ly-49W, which share 97% amino acid identity in their ectodomains, recognize PVG strain rat ligand(s), whereas the Ly-49A-related Ly-49P does not.

Because xenogeneic recognition of rat targets by mouse Ly-49 receptors occurs in both a strain- (33, 34) and an MHC haplotype-specific manner (33), we aimed to determine whether the xenogeneic recognition of PVG rat strain by Ly-49W and Ly-49G^BALB/c could be attributed to a specific region of the MHC within this strain.

The rat MHC is encoded on chromosome 20 (46). It contains a classical class I MHC-encoding region, RT1-A, a class II MHC-encoding region, RT1-B/D, and a nonclassical class I MHC-encoding region, RT1-C/E/M (Fig. 2,A). The RT1-A region of the RT1 complex encodes between one and three classical class I MHC molecules, depending on the rat strain. In the PVG rat (haplotype c), the RT1-A region encodes two molecules, RT1-A1^c and RT1-A2^c, whereas the DA rat strain (haplotype av1) only encodes one molecule in this region, RT1-A^a (Fig. 2,B) (47). The nonclassical class I MHC region, RT1-C/E/M, encodes numerous molecules, many of which have yet to be fully characterized. The more centromerically encoded molecules in this region, are known to be classical-like in both structure and function (46, 48). Of these classical-like molecules, some have been cloned from the PVG (GenBank accession nos. CAD60946, CAA06296, and CAA74192) and DA rat strains (GenBank accession no. CAD60945) (Fig. 2 B).

Both the classical RT1-A and nonclassical RT1-C/E/M regions of the PVG rat are known to encode molecules recognized by the rat Ly-49 receptors, Ly-49i2 and Ly-49s3, respectively (18, 26). Therefore, we wished to examine each region separately to determine whether the recognition of PVG Con A blasts by Ly-49W and Ly-49WG^BALB/c, could be attributed to molecules encoded in either the RT1-A region or RT1-C/E/M region of the PVG rat. We obtained PVG congenic, intra-MHC recombinant rat strains between the nonrecognized RT1^av1 haplotype and the recognized RT1^c haplotype (Dr. G. Butcher (Babraham Institute, Cambridge, U.K.)) that have a recombination between classical class I and class II regions, or class II and nonclassical class I regions (Fig. 2,B). These recombinants allowed us to examine the requirement for the RT1^c haplotype at each RT1 region, for recognition and lysis of targets by RNK-16 cells expressing Ly-49W and Ly-49WG^BALB/c. Untransfected RNK-16 cells were unable to mediate cytolysis of Con A blast targets from any of the recombinant strains (Fig. 3,A). RNK-16 cells transfected with the activating Ly-49W receptor mediated cytolysis of the PVG.R20 (c-c-a) recombinant strain but not the PVG.R1 (a-c-c) or PVG.R19 (a-a-c) intra-MHC recombinants (Fig. 3,B), indicating that a molecule encoded in the classical class I RT1^c region is required for recognition by Ly-49W. Similarly, RNK-16 cells expressing the Ly-49WG^BALB/c activating chimera were also seen to mediate cytolysis of Con A blasts from the PVG.R20 (c-c-a) strain, which expresses the c haplotype in the classical class I region, but not of Con A blasts from either PVG.R19 (a-a-c) or PVG.R1 (a-c-c) strains (Fig. 3 C), which only express the c haplotype in the nonclassical class I region or nonclassical class I and class II MHC regions, respectively. Therefore, both the Ly-49W receptor and Ly-49WG^BALB/c activating chimera recognize a ligand(s) encoded in the classical class I region of the RT1^c haplotype.

Because xenogeneic recognition by Ly-49W and Ly-49WG^BALB/c was dependent on expression of the c haplotype at the classical class I region of the RT1 complex in our recombinant studies, we examined whether recognition was mediated specifically by one of the two classical class I molecules encoded in the RT1-A region of the PVG rat, either RT1-A1^c or RT1-A2^c (Fig. 2 B).

Rat MHC molecules were cloned from splenic cDNA libraries of DA and PVG strain rats. As specific Abs are not available to all of the MHC molecules assayed, each molecule was expressed as a C-terminally fused EGFP protein. The fusion of EGFP to the cytoplasmic tail of each MHC molecule allowed detection of expression through green fluorescence, while conserving the extracellular domain structure for examination of xenorecognition by mouse Ly-49 receptors. Three MHC-EGFP fusion constructs were prepared from the PVG rat strain: the two classical class I molecules RT1-A1^c and RT1-A2^c, as well as the nonclassical class I molecule RT1-U2^c. An MHC-EGFP fusion construct of the classical class I molecule RT1-A^a was also prepared from the control DA rat strain, which does not have a ligand for Ly-49W or Ly-49G^BALB/c. Each of the MHC-EGFP fusion constructs was transfected into the rat myeloma cell line YB2/0 (RT1^u), and expression was determined by EGFP fluorescence intensity relative to untransfected YB2/0 cells (Fig. 4).

Transfection of YB2/0 cells with the classical class I region-encoded MHC molecule RT1-A1^c dramatically increased lysis by RNK-16 cells transfected with either Ly-49W or Ly-49WG^BALB/c (Fig. 5). Lysis of RT1-A1^c-transfected YB2/0 cells was specific, because YB2/0 cells expressing the other classical class I region-encoded MHC molecule found in the c haplotype, RT1-A2^c, were not lysed above background levels by either Ly-49W- or Ly-49WG^BALB/c-transfected RNK-16 cells (Fig. 5,A). As expected, YB2/0 cells expressing a nonclassical class I MHC molecule from the PVG rat, RT1-U2^c, were not recognized by Ly-49W- or Ly-49WG^BALB/c-expressing RNK-16 cells (Fig. 5,B). Additionally, YB2/0 cells expressing RT1-A^a, from the DA rat strain, did not trigger lysis through recognition by Ly-49W or Ly-49WG^BALB/c (Fig. 5 C).

We wanted to ensure that recognition of RT1-A1^c-transfected YB2/0 cells by the activating chimera Ly-49WG^BALB/c is also mediated by the native Ly-49G^BALB/c inhibitory receptor. RNK-16 cells expressing Ly-49G^BALB/c exhibit moderate background cytolysis of untransfected YB2/0 cells. This cytolysis is completely inhibited when YB2/0 cells are transfected with RT1-A1^c but not with other class I molecules from the PVG and DA rat strains (Fig. 6,A). The 4D11 Ab, which recognizes Ly-49G^BALB/c, reverses the inhibition of the Ly-49G^BALB/c-expressing RNK-16 transfectants by RT1-A1^c, whereas the isotype control Ab does not (Fig. 6 B). These results demonstrate that, like the chimeric receptor, Ly-49WG^BALB/c, the native inhibitory Ly-49G^BALB/c receptor mediates specific recognition of RT1-A1^c. Thus, expression of individual RT1 molecules from the recognized PVG rat strain and nonrecognized DA rat strains in YB2/0 cells allowed identification of the classical class I MHC molecule RT1-A1^c as a xenogeneic ligand capable of triggering NK cell effector functions through recognition by the mouse Ly-49 receptors Ly-49W and Ly-49G^BALB/c.

Analysis of the Ly-49A/H-2D^d cocrystal (49) implicates participation of Ly-49 β4–β5 loop residues, among others, in class I MHC recognition. Furthermore, nonconserved residues in this loop may be responsible for the specificity of Ly-49 receptors for particular class I alleles (34). Because we observed class I allele-specific rat xenogeneic ligand recognition by Ly-49W and -G^BALB/c, we tested for the potential participation of the Ly-49 β4–β5 loop in determining specificity for the rat RT1-A1^c class I ligand. Both Ly-49W and Ly-49G^BALB/c contain the residues DCGK in their predicted β4–β5 loop, whereas Ly-49P contains the residues NCDQ in this region. To determine the involvement of DCGK in recognition of PVG Con A blasts by Ly-49W, we mutated the Ly-49P receptor, which does not recognize the PVG rat strain (Fig. 7,A), to express β4–β5 loop residues from Ly-49W and saw a gain of PVG rat strain target recognition by the mutant Ly-49P receptor (B). Conversely, when we mutated the Ly-49W receptor, which recognizes the PVG rat strain (Fig. 7 C), to express residues found in the β4–β5 loop of Ly-49P, we saw no loss of PVG Con A blast recognition by the mutant Ly-49W receptor (D).

Although the wild-type Ly-49P receptor does not recognize any ligand from RT1^av1/c intra-MHC recombinant rat strains (Fig. 8,A), the mutant Ly-49P receptor gained the ability to recognize a ligand localized to the RT1-A region of the rat c haplotype (B). In contrast, when comparing the recognition of the wild-type form of the Ly-49W receptor (Fig. 8,C) to mutant Ly-49W (D), both recognized a ligand in the RT1-A region of the rat c haplotype, indicating no loss of function for the Ly-49W β4–β5 loop mutant. We could not confirm RT1-A1^c as opposed to RT1-A2^c as the recognized ligand due to high background lysis of YB2/0 RT1-A1^c and A2^c transfectants by the mutant Ly-49 RNK-16 transfectants (data not shown). Nevertheless, a classical class I ligand of the c haplotype is recognized and is most likely RT1-A1^c, considering our results using wild-type Ly-49W and chimeric -G^BALB/c receptors (Fig. 5). These results indicate that specificity for xenogeneic rat class I can be transferred to Ly-49P by substitution with specific residues from the β4–β5 loop of Ly-49W. They also reveal that these residues cannot be solely responsible for xenogeneic ligand specificity, as the Ly-49W mutant substituted with β4–β5 loop residues from Ly-49P retains its recognition of a PVG ligand.

In this study, we demonstrate that the rat class I MHC molecule RT1-A1^c is a xenogeneic ligand for the Ly-49W activating and Ly-49G^BALB/c inhibitory receptors, a surprising result given that ∼30–40 million years have transpired since the evolutionary divergence of mice and rats. Furthermore, rodent class I MHC and Ly-49 genes are evolving rapidly with significant species-specific evolution, resulting in no evident class I MHC orthologs (50), nor clear rat orthologs of Ly-49W or -G^BALB/c (Fig. 9). All known rat Ly-49 receptors have at most 67% amino acid identity with the mouse Ly-49W and -G^BALB/c receptors. The rat Ly-49i2, an inhibitory receptor, also recognizes RT1-A1^c (18), but it is distantly related to Ly-49W and -G^BALB/c (Fig. 9), sharing only 49–50% amino acid identity. With the absence of a clear Ly-49 or MHC ortholog between mouse and rat and their relatively low sequence identity, an alternative explanation appears to be necessary for this xenorecognition besides simple nondivergence of sequence. One possibility would be the convergent evolution of MHC through common selective pressures upon these two species. This phenomenon can be demonstrated with the MHC of mice and distantly related primate species for T cell recognition (51), and is strongly suggested for the MHC of rhesus monkeys and humans (52), species that have diverged to a similar extent as mice and rats. The convergent evolution of class I MHC may have resulted in RT1-A1^c recognition as a syngeneic ligand by rat Ly-49i2 and as a xenogeneic class I ligand, which may share properties with mouse class I ligands of unrelated mouse receptors, Ly-49W and -G^BALB/c.

It is interesting to consider whether the molecular basis of Ly-49W and -G^BALB/c recognition of RT1-A1^c resembles Ly-49 recognition of mouse class I molecules. The cocrystal of Ly-49A bound to H-2D^d serves as a model for mouse Ly-49-class I interactions. Two independent sites of Ly-49A interaction with H-2D^d are identified: site 1 at the N terminus of the α₁ helix and C terminus of the α₂ helix, and site 2 in the cleft bordered by the α₁/α₂ domains, the conserved α₃ domain, and β₂-microglobulin (49). We demonstrate that Ly-49W and Ly-49G^BALB/c, but not Ly-49P, recognize RT1-A1^c, and a previous report indicates that Ly-49D does not recognize PVG strain ligands including RT1-A1^c (33). Because Ly-49A, -D, -G, -P, and -W share significant sequence identities and belong to the Ly-49A-G subfamily (53), the differences in class I recognition are presumed to involve polymorphic residues of the Ly-49 receptors, particularly in the ligand binding C-terminal lectin-like domain. Using the Ly-49A/H-2D^d cocrystal for modeling mouse Ly-49 interactions with RT1-A1^c, indicates that polymorphic residues of Ly-49 receptors are concentrated in the β4–β5 loop, which can contact class I at either site 1 or site 2, and thus are candidates for determining xenogeneic class I specificity. Ly-49W and -G^BALB/c share the sequence DCGK at the β4–β5 loop and recognize RT1-A1^c, whereas Ly-49D is YCDQ and Ly-49P is NCDQ at this location. Interestingly, the distantly related rat Ly-49i2 receptor, which, like Ly-49W and -G^BALB/c, recognizes RT1-A1^c (18), also bears the DCGK sequence at its predicted β4–β5 loop. We demonstrate that transfer of specificity for an RT1-A^c molecule can be conferred to a mutant Ly-49P receptor by simply substituting the β4–β5 loop residues common to Ly-49W, -G^BALB/c, and -i2 for those in wild-type Ly-49P. This result suggests that these residues can confer RT1-A1^c specificity, perhaps by strengthening interactions with class I residues at site 1 or 2, or alternative site(s), exceeding a functional threshold. However, the DCGK sequence is not solely responsible for RT1-A^c specificity, because mutating the Ly-49W receptor to express the β4–β5 residues from Ly-49P does not affect the recognition of RT1-A^c by Ly-49W. These findings indicate that β4–β5 residues, DCGK in particular, can influence the specificity of Ly-49 receptors for xenogeneic class I, but contributions of other Ly-49 residues are also significant to recognition.

We demonstrate that recognition of xenogeneic rat class I molecules by mouse Ly-49W and -G^BALB/c receptors is specific, observing interactions with RT1-A1^c, but not RT1-A^a or RT1-A2^c. An examination of amino acid sequence differences between RT1-A1^c, RT1-A^a, and RT1-A2^c may provide insights into the differential recognition of the rat class I molecules. Only a few residues, modeled to be involved in site 1 or 2 interactions, differ between RT1-A1^c and the other two, rat class I molecules. All site 2 residues on the rat class I molecules are conserved and thus do not offer an obvious explanation for the differential interactions observed. RT1-A2^c and RT1-A^a both differ from RT1-A1^c at site 1, with RT1-A2^c having Q50 and H174 and RT1-A^a having S169 compared with R50, L174 and R169 in RT1-A1^c. The site 1 residues differing in RT1-A2^c and RT1-A^a also differ from those at the same positions in the mouse class I ligands of Ly-49W and -G^BALB/c. These differences at site 1 may account for the differential recognition of rat class I molecules; however, mutagenesis experiments will be necessary to substantiate this possibility. Additionally, outside of the site 1 and site 2 interfaces, RT1-A2^c and RT1-A^a have several additional nonconservative amino acid substitutions that may contribute to the differences in class I specificity.

The absence of amino acid differences at site 2 and limited differences at site 1, between RT1-A1^c and RT1-A^a or RT1-A2^c, suggest additional influences on mouse Ly-49W and -G^BALB/c specificity for xenogeneic class I ligands. Observations in humans and subsequently in other species, indicate that class I molecules can be categorized into a limited number of supertypes based on their preferences for specific anchor residues in peptide binding (54, 55, 56, 57). In humans, although class I alleles exhibit extensive diversity, they can be grouped into only nine supertypes (58) when the variation of anchor residues is considered. Supertype groupings also extend to class I molecules of other, even distantly related, species. For example, mouse and humans share several supertypes between nonorthologous class I genes, suggesting convergent evolution of supertypes (51, 59), presumably due to a need in multiple species to possess class I molecules that collectively have the ability to bind peptides with the same 10 or fewer distinct patterns of anchor residues. The HLA-B7 supertype typically requires peptides to have proline at the second, or P2, position and a hydrophobic C terminus (60). It is worthy of note that most of the rodent class I molecules that are recognized by Ly-49W and/or Ly-49G^BALB/c have this B7 supertype (H-2L^d and RT1-A1^c) (60) or prefer a P2 residue with a small side chain (H-2D^d prefers G at P2 and P at P3) (60). In contrast, all rat or mouse class I molecules not recognized by Ly-49W or -G^BALB/c (RT1-A^a, H-2K^k, -K^d, -K^b, and -D^b), belong to other supertypes with distinct anchor residues (60). One exception to this pattern is H-2D^k, which does not have a B7 supertype (60) but is recognized by Ly-49W and -G^BALB/c. RT1-A2^c is not included in this comparison, because the peptide anchor residues it prefers have not been reported. These relationships raise the possibility that Ly-49W and -G^BALB/c, along with related receptors, may recognize a conformational epitope on class I that is influenced by peptide anchor residues and have evolved to survey cells for class I MHC molecules of, or related to, the HLA-B7 supertype. Potentially, the shared recognition of RT1-A1^c by the rat Ly-49i2 and mouse Ly-49W and -G^BALB/c receptors is due to similar evolutionary pressures and convergent evolution, leading to a common requirement for murine NK cell recognition of the HLA-B7/RT1-A1^c/H-2L^d supertype.

Although the role of bound peptide is unknown in xenogeneic rat class I recognition by Ly-49 receptors, we considered the potential influence of transporter associated with Ag processing (TAP)2 polymorphism in the differential recognition of rat class I molecules by mouse Ly-49W and -G^BALB/c, and deemed it unlikely. Two versions of the heterodimeric TAP, TAP-A and TAP-B, are expressed in the rat due to the existence of two alleles of TAP2 in this species (61, 62). The specificity for peptide transport differs between TAP-A and TAP-B, and only one or the other TAP2 allele is expressed in each inbred rat strain (63). TAP-A, like human TAP, exhibits broad specificity, transporting peptides with either hydrophobic or positively charged C-terminal residues for assembly with class I. In contrast, TAP-B displays a narrower specificity and strong bias for peptides with hydrophobic or aromatic C termini, similar to mouse TAP (64). RT1-A^a has a strong preference for binding peptides with a C-terminal arginine and is typically coexpressed with TAP-A, which efficiently delivers such peptides (65). In contrast, RT1-A1^c prefers peptides with a C-terminal hydrophobic or aromatic residue and is typically coexpressed with TAP-B, which offers compatible peptides (65). Our studies indicate that Ly-49W and -G^BALB/c recognize the TAP-B-dependent RTI-A1^c, but not TAP-A-dependent RT1-A^a when expressed by transfection in YB2/0 cells. Because YB2/0 (RT1^u) cells express TAP-B, it might be argued that the lack of RT1-A^a recognition could be due to the absence of a coexpressed TAP-A. However, we also do not detect lysis of DA strain Con A blasts, which coexpress RT1-A^a and TAP-A. Thus, the differential recognition of rat class I molecules by the mouse Ly-49 molecules cannot be accounted for by a diminished supply of relevant peptides for RT1-A^a binding. Whether RT1-A1^c specifically requires TAP-B coexpression for Ly-49 recognition, cannot be discerned from the assortment of inbred strains available for this study. It is worth noting that the differential selectivity of peptide transport by TAP-A and TAP-B is limited to differences at the peptide C termini, and there is no additional bias with respect to transport of peptides with a proline at P2, as is preferred by RT1-A1^c.

The induction of tolerance to donor Ags in adult transplantation would prevent transplanted tissue rejection. One potentially promising approach to achieve this goal under study in animal models is the generation of mixed chimeras (44). In mixed chimeras, adult animals are treated to eliminate mature host T cells, yet preserve other host hemopoietic cells, and are then reconstituted with donor bone marrow cells. Donor and host hemopoietic cells in mixed chimeras repopulate the thymus and delete both donor-reactive and host-reactive T cells, resulting in peripheral tolerance to donor and host tissue (44). Xenogeneic mixed chimeras can be generated in mice with rat hemopoietic donor cells (66), and this system serves as a model for induction of tolerance to xenogeneic transplants in adult animals (44). These rodent xenotransplantation models may aid in establishing principles relevant to successful clinical xenotransplantation such as pig tissues into humans. Mouse NK cells play a significant role in resisting engraftment of rat hemopoietic cells and the establishment of xenogeneic mixed chimeras (35). It is suggested that this might occur because mouse NK cell inhibitory receptors may not cross-react with xenogeneic rat class I MHC proteins and thus do not offer protection from NK cell elimination of rat donor cells (35). In this report, we demonstrate for the first time that a mouse inhibitory receptor (Ly-49G) expressed by a common mouse strain (BALB/c) can recognize a xenogeneic rat class I MHC molecule (RT1-A1^c) resulting in functional inhibition. Identification of a rat class I ligand for Ly-49G and the activating Ly-49W receptor may begin to facilitate the rational design of studies to elucidate the role of specific NK cell receptors and NK cell subsets in xenogeneic bone marrow graft tolerance and rejection.

We thank Dong-Er Gong for excellent technical assistance. We thank Dr. Bart Hazes for critical reading of the manuscript.