Cutting Edge: The Minor Histocompatibility Antigen H60 Peptide Interacts with Both H-2K^b and NKG2D¹

Rejection of transplanted tissues and organs within a species is caused by recognition of proteins encoded by genes that demonstrate allelic polymorphism. Differences in the MHC Ags between a transplant donor and recipient lead to rapid elimination of the graft; however, polymorphisms in other genes can also cause rejection of tissues in MHC-matched individuals. These minor histocompatibility (minor H)⁴ Ags have been found to be peptide fragments derived from polymorphic proteins that are presented by class I MHC molecules, initiating a CD8⁺ CTL response (1). Minor H Ags can cause graft-vs-host disease and chronic rejection of solid tissues (2). Peptides derived from proteins encoded on the Y chromosome function as minor H Ags when male tissues are transplanted into females (3, 4, 5). Other polymorphic autosomal genes have been shown to encode proteins expressed in the mitochondria or nucleus, or on the cell surface (6, 7, 8, 9, 10, 11).

The minor H60 Ag is an 8-aa peptide (LTFNYRNL, designated LYL8) presented by H-2K^b to CTL of C57BL/6 mice immunized with BALB.B splenocytes (12). The intact polypeptide encoded by the H60 gene is a cell surface glycoprotein that has an α₁ and an α₂ domain with homology to MHC class I. H60 is expressed in BALB.B mice but is not transcribed in C57BL/6 mice (12). While C57BL/6 and BALB.B mice differ by many minor H Ags, H60 appears to dominate the allogeneic CTL response (13). The H60 glycoprotein is a ligand for NKG2D, an activating NK cell receptor expressed by NK cells, CD8⁺ T cells, and activated macrophages (14, 15). Thus, the response of CD8⁺ T cells to H60 may be caused by TCR recognition of the H-2K^b-LYL8 complex and costimulation provided by interactions between NKG2D on the T cell and H60 on APCs. In this study we have examined the role of the LYL8 peptide in the function of the H60 protein.

NK cells were used in ⁵¹Cr release cytotoxicity assays, as described (16). Ba/F3 and RMA cells were transduced with retroviruses generated by using the pMX-pie vector (17). Site-specific mutants of H60 were generated by standard methods using PCR. Ba/F3 cells stably transduced with mouse NKG2D and DAP10 (18) and CTL cell lines (13) recognizing the LYL8-H-2K^b complex were described.

NKG2D, H60, and retinoic acid early inducible-1 (RAE-1) proteins were expressed in Escherichia coli (19). Tetrameric complexes of H-2K^b loaded with LYL8 (13) and a fusion protein containing the extracellular domain of mouse NKG2D and the Fc region of human IgG1 (14) were produced as described. LYL8 and control peptides were synthesized by using F-moc chemistry and purified by HPLC. RMA-S cells were cultured with LYL8 peptide and stabilization of H-2K^b was evaluated by staining with anti-H-2K^b mAb (BD PharMingen, San Diego, CA) (20). Flow cytometry was performed by using a FACSCalibur (BD Biosciences, San Jose, CA) or a small desktop Guava Personal Cytometer with Guava ViaCount and Guava Express software (Guava, Burlingame, CA).

Surface plasmon resonance measurements were performed by using a BiaCore 2000 (BIAcore, Uppsala, Sweden) (19). Carboxymethylated dextran matrix CM5 Research Grade chips were used with filtered and degassed HBS-EP (10 mM HEPES (pH 7.4), 150 mM NaCl, 3 mM EDTA, 0.005% polysorbate 20 (v/v)) buffer. Streptavidin (Sigma-Aldrich, St. Louis, MO) was covalently coupled to chips via primary amines using the Amine Coupling kit (BIAcore). Biotinylated proteins were immobilized at the required levels by injection over the streptavidin-coupled surfaces. Typically, immobilization levels were around 400 response units and the concentration of soluble protein was at the known K_d of 19 nM for NKG2D and H60 (19). In all experiments, the signal from flow cells coated with the protein of interest was compared with the signal from mock-coupled cells or cells coupled with an irrelevant protein to control for nonspecific effects.

The H60 LYL8 peptide conforms to the consensus H-2K^b binding motif xxxx(F,Y)xx(I,V,L,M) (21). This peptide represents a dominant epitope for CTL recognition in C57BL/6 mice immunized with BALB.B splenocytes (12). LYL8 corresponds to amino acids 39–46 in the predicted H60 polypeptide within the MHC class I-like α₁ domain. To distinguish between αβ-TCR-mediated recognition of the H-2K^b-LYL8 complex and recognition of H60 by NKG2D, site-directed mutagenesis was performed at amino acid 43 in the H60 polypeptide to change the anchor residue tyrosine (Y) at position 5 in LYL8 to an alanine (A), an amino acid predicted not to bind H-2K^b. A cDNA encoding the mutant Y43A H60 molecule was transduced into Ba/F3, a MHC class I-negative pro-B cell line. In contrast to Ba/F3 cells transduced with wild-type H60, mouse NKG2D-Ig fusion protein failed to bind the cells transduced with the Y43A H60 mutant (Fig. 1). Similar results were obtained when Ba/F3 cells were transduced with another H60 mutant construct in which the Y anchor residue at position 5 was changed to S (Y43S), in theory another nonpermissive amino acid for H-2K^b binding. A mutant with a conservative Y to F change (Y43F), which is predicted to preserve association with H-2K^b, demonstrated minimal binding to NKG2D. The abolition of NKG2D binding was not observed when other residues in the peptide were mutated; this included mutants replacing T with S at position 2 (T40S H60), L with A at position 1 (L39A), and L to A (L46A) or L to S (L46S) at position 8. To verify that the mutant H60 proteins were expressed on the surface of the transduced Ba/F3 cells, the mutant H60 cDNA were inserted into a plasmid containing a FLAG epitope on the N terminus. Ba/F3 cells transduced with the FLAG-Y43A, FLAG-Y43S, and FLAG-Y43F H60 stained brightly with anti-FLAG mAb but were substantially unreactive with NKG2D-Ig fusion protein. To exclude the possibility that the FLAG epitope interfered with NKG2D binding, FLAG was inserted onto the N terminus of wild-type H60. FLAG-wild-type H60 was detected with both anti-FLAG mAb and NKG2D-Ig. All of the transduced cells expressed high levels of the wild-type or mutant H60 glycoproteins, as demonstrated by staining with anti-FLAG mAb.

The observed differences in NKG2D-Ig fusion protein binding to Ba/F3 cells transduced with wild-type or mutant H60 molecules corresponded with results from cytotoxicity assays using IL-2 stimulated mouse NK cells. Consistent with prior findings (14), NK cells demonstrated enhanced killing of Ba/F3 cells transduced with H60 and efficiently lysed Ba/F3 cells transduced with wild-type H60, L39A, T40S, L46S, and L46A (Fig. 2). In contrast, NK cell lysis of cells transduced with the position 5 mutants Y43A, Y43S, and Y43F was not substantially different from the low levels of killing observed against the untransfected Ba/F3 targets. Ba/F3 cells expressing H60 molecules with mutations at positions 39 and 46 did show slightly diminished susceptibility to NK cell-mediated lysis, consistent with the slightly lower levels of staining with NKG2D-Ig (Fig. 1). Therefore, these mutations may have a minor influence on NKG2D binding; however, they were marginal compared with the dramatic effects observed when residue 43 was altered. The FLAG epitope on the H60 proteins did not prevent NK cell-mediated cytotoxicity, because similar levels of killing were observed when comparing H60 to H60 with an N-terminal FLAG sequence (data not shown). Based on the amount of FLAG epitope detected on the cell surface, all of the transduced cells expressed the mutant H60 proteins at levels comparable to or higher than the wild-type H60 glycoprotein, indicating that failure to efficiently kill the cells bearing mutations at position 5 in LYL8 cannot be attributed to inadequate expression of the protein.

NKG2D binds to the H60 and RAE-1 ligands with relatively high affinity (NKG2D-H60 K_d = ∼20 nM; NKG2D-RAE-1 K_d = ∼500 nM) (19). Evidence for an interaction between NKG2D and the synthetic LYL8 peptide was obtained by binding inhibition studies. Aliquots of soluble NKG2D, at a concentration close to the K_d of the interaction, were pre-equilibrated with different peptide dilutions and injected over the BiaCore chip surfaces coated with H60. A range of peptides was tested for inhibition, including the LYL8 peptide LTFNYRNL, the H60 peptide with the Y-F mutation at position 5 LTFNFRNL, or irrelevant control peptides GGKKKYKL, FLRGRAYGL, and IVKEPVHGD. As shown in Fig. 3, a clear and titratable partial inhibition of binding was seen with the LYL8 peptide but not with the other peptides. Inability of the H60 mutant peptide with the Y-F mutation at position 5 confirms the specificity of the binding inhibition seen with LYL8 and is in accordance with the inability of NKG2D-Ig to bind Ba/F3 cells expressing the H60 Y43F mutant protein. These results are consistent with a direct interaction between NKG2D and LYL8 peptide but suggest a low binding affinity of the linear synthetic peptide. Given the low affinity of the linear synthetic peptide, it was unable to block NK cell cytotoxicity (data not shown); however, multimeric or cyclic LYL8 peptides might have a higher avidity.

Because LYL8 can bind both H-2K^b and NKG2D, this raised the question of whether peptide bound within the class I groove functions as a ligand for NKG2D. We examined whether a tetrameric H-2K^b complex loaded with LYL8 interacts with NKG2D. This was addressed by using this tetramer either to stain Ba/F3 cells stably transduced with mouse NKG2D and DAP10 (18) or mouse NK cells. Neither the NKG2D-transduced Ba/F3 cells nor mouse NK cells bound the tetramer, whereas the tetramer did bind to CTL cell lines expressing a TCR recognizing LYL8 peptide to H-2K^b (13) (data not shown). RMA-S (20), a TAP-deficient cell line, was incubated with LYL8 peptide and this stabilized H-2K^b on the surface of these cells, as determined by using flow cytometry. The LYL8 peptide-loaded RMA-S cells failed to stain with a mouse NKG2D-Ig fusion protein (data not shown). Presumably, when LYL8 is bound within the class I groove it is unable to interact with NKG2D, possibly because critical residues are inaccessible. Therefore, NKG2D reacts with H60 but apparently not with LYL8 peptide presented by H-2K^b.

Studies were performed to address TCR recognition of LYL8 and the Y43F mutant. TAP-deficient RMA-S cells (which lack endogenous NKG2D ligands (22)) were loaded with LYL8 or Y43F mutant peptides and used as target cells for NKG2D-bearing H60-specific CTL cell lines established from B6 mice immunized with BALB.B splenocytes (13). Equivalent cytolytic activity was observed when RMA-S cells were loaded with the LYL8 and Y43F peptides over a broad range of concentrations (data not shown), demonstrating TCR recognition of both peptides. RMA cells (lacking endogenous NKG2D ligands) were stably transfected with the wild-type H60 and Y43F mutant H60 cDNA, each containing a FLAG epitope tag on the N terminus. Equivalent levels of expression of the transduced proteins were confirmed by staining with anti-FLAG mAb. RMA cells transfected with either the wild-type H60 or Y43F H60 mutant were efficiently and equivalently killed by two NKG2D-bearing H60-specific CTL cell lines (data not shown), suggesting that NKG2D may not contribute in the effector phase of CTL function. These results do not exclude the possibility that NKG2D may augment the activity of other CTL, possibly with low-affinity TCR.

Choi et al. (13) have shown that H60 is a potent minor alloantigen. Within 7 days after a single immunization of C57BL/6 mice with BALB.B splenocytes, ∼6% of CD8⁺ T cells in the blood were specific for LYL8, compared with <0.02% of T cells reactive with H-Y or H13 peptides, as detected by staining with MHC-peptide tetramers. This dominant response to H60 appeared to be a consequence of efficient expansion of T cells against this Ag. The diverse TCR Vβ usage by T cells recognizing the LYL8-H-2 complex indicated the Ag does not behave like a superantigen. H60 is a high-affinity ligand for NKG2D (19), a costimulatory molecule that has been shown to augment T cell-mediated immunity against viruses (23) and tumors (24). H60 transcripts are present in splenocytes (12), and we have detected them in peritoneal macrophages from BALB/c mice (J. A. Hamerman, unpublished observation). Therefore, if APC express both H60 on their cell surface and the LYL8/H-2K^b complex, this may provide an efficient mechanism whereby TCR recognition of class I induces NKG2D on the CD8⁺ T cell, which in turn costimulates the response. However, while it is possible that H60 on an APC may act as a costimulator of T cell responses through NKG2D, it does not explain the preferential response against H60 compared with other minor histocompatibility Ags also expressed by APC (13). Whether LYL8 peptides are more efficiently processed by the class I pathway and hence are more abundant should be explored. Finally, our studies using H60-specific CTL cell lines indicate that H60 interactions with NKG2D are not essential for the effector phase of CTL activity in vitro but leave open the possibility that H60 may contribute to priming a CD8 T cell response. This is similar to prior studies of CD28, a costimulatory molecule that may contribute to priming CD8 T cell responses but is not necessary for CTL activity in vitro (25).

Our study was designed originally to distinguish between the contribution of NKG2D vs the TCR in immune responses against H60. By mutation of critical anchor residues in LYL8 to prevent binding to H-2K^b, we attempted to abrogate binding to the TCR yet preserve a high-affinity interaction with NKG2D. Surprisingly, mutation of H60 at amino acid 43, corresponding to the anchor residue at position 5 in LYL8, completely prevented binding to NKG2D. This was unexpected because it implied that this site functions as both a TCR epitope and an important binding site for NKG2D. An interaction between LYL8 and NKG2D was supported by demonstrating that the linear synthetic LYL8 peptide partially blocked the binding of NKG2D to H60. Our results predict the existence of a contact site between NKG2D and H60 in the region of the LYL8 peptide. Formal proof that LYL8 in the native H60 protein binds NKG2D awaits a crystal structure of the H60-NKG2D complex.

H60 is similar to the RAE-1 family of molecules (26) that also function as ligands for NKG2D (14, 22). Three RAE-1 genes (α, β, and γ) are present in 129 strain mice, and these are genetically linked on chromosome 10 to H60. Mouse RAE-1 genes are polymorphic (our unpublished observation), suggesting the possibility that the RAE-1 proteins might also function as alloantigens in certain mouse strains. Human orthologs of the RAE-1 genes have been identified, and these encode proteins that bind NKG2D (27, 28). In humans, NKG2D also binds to the MICA and MICB glycoproteins encoded by MHC genes (29). MICA and MICB show extensive allelic polymorphism, and alloantibodies against MICA have been detected in the sera of organ transplant recipients (30). These findings suggest the possibility that polymorphic ligands for NKG2D also may function as minor H transplantation Ags in humans.

We thank Pamela Bjorkman for helpful discussions and Koetsu Ogasawara for providing NK cells.