The swine major histocompatibility complex (MHC), also referred to as swine leukocyte antigen (SLA), has been associated with the porcine immune response to various infections and vaccinations (Lunney and Murrell, 1988; Lumsden et al., 1993; Lunney et al., 2009). SLA has been mapped to pig chromosome 7, and comprises highly polymorphic classical class I, class II, and conservative class III gene clusters (Renard et al., 2006). The SLA class I gene cluster contains three constitutively expressed classical genes, SLA-1, SLA-2, and SLA-3, which are expressed by most cells. The SLA-1 gene has the highest expression level whereas SLA-3 has the lowest as a result of different promoter activity (Tennant et al., 2007). The SLA-1, SLA-2, and SLA-3 genes are highly polymorphic, and 192 alleles (69 SLA-1, 87 SLA-2, 36 SLA-3) have been designated by the SLA Nomenclature Committee of the International Society for Animal Genetics in the Immuno Polymorphism Database (IPD)-MHC SLA sequence database to date¹ [Release 2.2.0.0 (2017-06-01)] (Maccari et al., 2017). The extreme polymorphisms of SLA class I genes are concentrated in the α1 and α2 domains, which resemble each other structurally, and together form the class I heavy chain protein peptide-binding groove (PBG). Three-dimensional (3D) crystal structure analysis has indicated that the PBG contains six pockets (A–F), and epitope peptides fixed in the PBG by these pockets. The different allelic forms of SLA class I genes were confirmed to bind different classes of peptide, determined by the fit between the pockets in the PBG of the SLA complex, and the anchor residues in the peptides (Zhang N. et al., 2011; Fan et al., 2016). SLA class I heavy chain, epitope peptide, and β2-microglobulin (β2m) have also been shown to form a ternary complex, with the protein complexes being expressed constitutively on the surface of virtually all nucleated cells. These are crucially important for the normal growth of CD8+ cytotoxic T lymphocytes (CTLs), and play a pivotal role in the cell-mediated immune response against viral infections and cancer. The protein complexes are mainly involved in the adaptive immune response through their presentation of endogenous epitope peptide antigens to circulating MHC I-restricted CTLs (Neefjes et al., 2011). CTLs kill the specific target cells directly, and also induce host cell-mediated specific immune responses by simultaneously identifying the epitope peptides specifically bound to SLA class I molecules and the SLA class I molecules (MHC restriction). This represents one of the important host-mediated immune-defense responses for controlling virus infection. Furthermore, the protein complexes also interact with natural killer (NK) cells to prevent NK-mediated cytotoxicity (Kwiatkowski et al., 1999).

Pigs are important experimental animals for veterinary and medical studies, including studies of virus infections and pathogenesis, host immunological responses, vaccine evaluation, the identification of T-cell epitopes, and so on. An understanding of peptide binding to SLA class I proteins is therefore of major practical interest. In addition, pigs have an evolutionary resemblance to humans, and share anatomical, physiological, immunological, metabolic, and nutritional similarities, making them promising organ donors for xenotransplantation. The structures of human and mouse MHC class I molecules and their interaction with bound antigenic peptides along with the roles of peptide-interacting pockets in the PBG have been intensively studied. Nevertheless, only a very limited set of the SLA class I molecules have been well documented and reported (Zhang N. et al., 2011; Fan et al., 2016), and the characteristics of peptide presentation for SLA class I molecules and cellular immune mechanisms have remained elusive until now. The structure and function of SLA class I complexes constructed in vitro are currently used to simulate the functions of SLA class I molecules in vivo and numerous in vitro SLA class I complexes have been constructed and different peptide–SLA-I binding assays have been suggested (Oleksiewicz et al., 2002; Sylvester-Hvid et al., 2002; Gao et al., 2006; Pedersen et al., 2011; Zhang N. et al., 2011; Gao et al., 2012; Fan et al., 2016). A relative simple and rapid, in vitro refolding enzyme-linked immunosorbent assay (ELISA)-based method was able to discriminate between peptide-occupied and peptide-free SLA-I complexes based on monoclonal antibody PT85A binding (Oleksiewicz et al., 2002), because the PT85A monoclonal antibody can recognize all SLA class I molecules, from outbred as well as inbred pigs, and the conformational epitope was recognized by PT85A, required the presence of the ‘correct’ peptide as well as the ‘correct’ SLA class I molecule sequence (Lunney, 1994; Davis et al., 2000; Mosaad et al., 2006). We therefore selected seven SLA-1 molecules identified in Chinese Bama miniature pigs, and one CTL epitope peptide of classical swine fever virus (CSFV) and four epitope peptides of porcine reproductive and respiratory syndrome virus (PRRSV), previously identified by bioinformatics and immunological tests, for the current study. We aimed to construct SLA class I complexes consisting of viral epitope peptides, the extracellular region of the SLA-1 molecules, and β2m using splicing overlap extension polymerase chain reaction (SOE-PCR). The constructed protein complexes were then induced and expressed using an Escherichia coli prokaryotic expression system in vitro, and the obtained proteins were purified and refolded. Specific binding of SLA-1 proteins to the viral CTL epitope peptides was detected using an ELISA-based method and the MHC restrictions of the five epitope peptides were identified. The SLA-1 molecules selectively binding to an appropriate epitope peptide during antigen presentation were thus characterized in vitro. The results of this study will lay the foundations for further studies of cellular immune mechanisms and for the development of effective polypeptide vaccines.

Three constitutively expressed and highly polymorphic classical SLA class I genes map to the SLA complex, SLA-1, SLA-2, and SLA-3 (Renard et al., 2006). SLA class I molecules play a crucial role in cellular immune antigen presentation in pigs and in xenotransplantation of pig organs into humans. The SLA-1 gene has the highest expression level (Tennant et al., 2007), implying that SLA-1 molecules might play a dominant role in the immune process, including presentation of CTL epitopes. Sixty-nine SLA-1 alleles, identified in various pig breeds and cell lines (Smith et al., 2005; Lee et al., 2005, 2008; Ho et al., 2006, 2009, 2010; Gao et al., 2014, 2017), have been deposited in the IPD-MHC SLA sequence database to date, and this number is increasing as more efficient SLA-typing techniques are employed and more experimental pig breeds are developed worldwide. Polymorphism of SLA class I molecules is concentrated in the region of the PBG, and determines the distinct structure of the PBG. The structural basis of swine CTL epitope presentation has been illustrated by determination of the crystal structure of SLA-1^∗04:01:01 with a peptide from swine influenza A virus and Ebola virus (Zhang N. et al., 2011). The PBG was shown to be classified into six pockets, A–F, which determined the peptide-binding specificity. The binding of different SLA molecules to different classes of peptides was determined by the fit between these pockets and the anchor residues in the peptides (Zhang N. et al., 2011; Fan et al., 2016). Viral peptides interact with the pockets to form a heterotrimeric complex, including SLA class I heavy chain, the epitope peptide, and β2m. In the initial stage of virus infection, antigenic proteins are processed in a proteasome-dependent or -independent manner, and the resultant short peptides are then transported to the endoplasmic reticulum and loaded onto the PBG of SLA class I molecules. The peptide-loaded SLA class I complexes are then translocated to the cell surface and recognized by CTLs with specific T-cell receptors (Shastri et al., 2002; Yewdell and Haeryfar, 2005). This immune recognition induces an MHC-restricted CD8+ T-cell response characterized by the proliferation of CTLs, prevention of pathogen release, lysis of the virally infected cells, and killing and elimination of the infected cells by host effector T cells (Neefjes et al., 2011). Identification of epitope peptide binding to SLA class I molecules represents a critical step in this process. Moreover, viral epitopes bind to the PBG with different specificities, in an MHC-restricted manner. It is therefore essential to characterize the binding of SLA class I molecules to viral CTL epitope peptides because of the significance of this process for monitoring CD8+ T cell immune responses and understanding the mechanisms of cellular immunity.

Historically, the peptide-binding specificity of SLA class I molecules has been characterized by various methods, each with particular advantages and drawbacks (Oleksiewicz et al., 2002; Shastri et al., 2002; Sylvester-Hvid et al., 2002; Gao et al., 2006; Pedersen et al., 2011; Zhang N. et al., 2011; Pedersen et al., 2014, 2016). The refolding of the MHC class I complex is known to be influenced by the presence or absence of a peptide (Garboczi et al., 1992). The refolding and conformation of the SLA class I complex depends on whether the epitope peptide can bind to the SLA class I molecules. Epitope has generally been reported to be required for in vitro refolding of SLA class I molecules (Garboczi et al., 1992). The refolding and conformational differences can be assessed using the monoclonal antibody PT85A, and differences in binding reactivities between epitope peptides and SLA-1 molecules can then be reflected by the relative OD values following ELISA. In this study, we used a rapid and simple in vitro refolding assay (Oleksiewicz et al., 2002) to examine the specific binding of seven SLA-1 proteins to five viral CTL epitope peptides, previously identified using bioinformatics and immunological tests (Pauly et al., 1995; Vashisht et al., 2008; Díaz et al., 2009), but with unknown MHC restriction. Although the CSFV NS4A epitope peptide (ENALLVALF) has been described as a SLA d/d haplotype (H4/H4 haplotype: SLA-1^∗04:01:01-SLA-2^∗04:01-SLA-3^∗04:01)-restricted CD8+ T cell epitope (Pauly et al., 1995), it was unknown to which of these three class I genes it was restricted. The current results showed that SLA-1^∗13:01, SLA-1^∗11:10, and SLA-1^∗11:01:02 proteins were able to bind specifically to different CTL epitopes of CSFV and PRRSV. This suggests that the compatibilities of the pockets in the PBG of SLA class I molecules are distinct, and that a broad range of protective CD8+ T cell responses could potentially be elicited, associated with the various peptide conformations. The distinct peptide conformations mainly result from the flexible side chains of the pockets, and do not involve the α-helixes and β-sheets that form the PBG. The fact that different SLA-1 can bind to the same CTL epitope indicates that the same peptide could be presented by different SLA class I molecules, with chemical specificity determined by different preferences for certain anchor residues. Many SLA-1 suballeles could also show similar anchor residue preference due to the presence of identical key residues in pockets (Stewart-Jones et al., 2005). Many studies have also shown that water molecules, platform adjustment, or the presence of only one flexible pocket can contribute to the accommodation of different peptides (Smith et al., 1996; Stewart-Jones et al., 2005; Koch et al., 2007). Furthermore, although the conformations of one SLA-1 pockets are unique, similar pockets were could found in other SLA-1 molecule structures. These pockets could accommodate the same residues of one peptide in similar manners (Fan et al., 2016).

We identified the MHC restriction of five CTL epitopes. CD8+ T cell epitopes are known to be strictly restricted to SLA class I molecules, suggesting that identification of MHC-I-restricted CTL epitopes could aid the rational development and modification of peptide-based vaccines. Current vaccines are usually ineffective against newly emerged virus strains because of the rapid mutation of viral proteins through both antigenic drift and shift, for example influenza virus. New vaccines therefore need to be developed against their mutant virus, and new-vaccine strategies increasingly are directed at conserved viral CTL epitope-based vaccines. Few polypeptide vaccines have been utilized in swine to date because of the MHC restriction of peptides and the high polymorphism of SLA class I genes. In this study, we constructed SLA class I complexes consisting of viral epitope peptides, the extracellular region of SLA-1 molecules, and β2m, and expressed and refolded them in vitro, and used an ELISA-based method to determine the MHC restriction of five CTL epitopes. It is hoped to develop the more effective polypeptide vaccines for these CTL epitope peptides. Many T-cell epitopes derived from various swine-origin viruses have been identified using in vivo cytotoxicity assays, enzyme-linked immunospot assays, intracellular cytokine staining, flow cytometry, and so on (Pauly et al., 1995; Gerner et al., 2006; Vashisht et al., 2008; Díaz et al., 2009; Wang et al., 2011; Zhang W. et al., 2011; Chen et al., 2013), but most of these methods are relatively labor intensive and technically demanding, and importantly do not clarify the MHC restriction of the epitopes. Gao et al. reconstructed a SLA-2-(G4S)3-b2m protein complex in vitro, which could be used to identify nonameric viral peptides in swine, in conjunction with mass spectrometry (Gao et al., 2006). Similarly, the current study presented a method for not only generating recombinant SLA-1 molecules and mapping their specificities, but also for identifying nonameric MHC-I-restricted epitopes. This method could avoid the one-sidedness of predicting epitopes only depending on protein sequences and thus improve the accuracy of epitope identification.

In addition, we analyzed the amino acid sequence homologies of the α1 and α2 domains among seven SLA-1 molecules. The SLA-1^∗04:01:01 and SLA-1^∗13:01 had 97.1% homology, implying that they might bind the same classes of peptide. However, these proteins showed distinctly different epitope-binding specificities. We examined the basis for these different binding specificities by alignment of the amino acid residues of the PBG and observed eight discrepant amino acid residues, of which four [66(N/K), 70(T/N), 99(Y/F), 152(E/V)] were involved in the formation of pockets B, C, D, and E, respectively. We mutated each of the four different amino acid residues of one molecule to the corresponding residues of the other molecule and determined the effect of the mutations on the specific binding of the CSFV NS4A epitope peptide. The NS4A epitope peptide could not bind to the SLA-1^∗13:01/152 mutant protein when Val¹⁵² alone was mutated, and could not bind to the SLA-1^∗04:01:01/152 mutant protein. We obtained similar results for the Asn¹⁵¹ residue. However, when the 151(D/N) and 152(E/V) residues of SLA-1^∗13:01 and SLA-1^∗04:01:01 were mutated simultaneously to those of the corresponding molecule, the five epitope peptides could not bind to the SLA-1^∗13:01/151/152 mutant protein, but could bind to the SLA-1^∗04:01:01/151/152 mutant protein. Moreover, the relative level of epitope peptides binding to SLA-1^∗04:01:01/151/152 mutant was obviously increased, indicating the binding strength of peptides to SLA-1^∗04:01:01 was enhanced when simultaneously mutating 151 and 152 amino acid residues. In turn, the binding strength of SLA-1^∗13:01/151/152 mutant to epitope peptides were nearly identical to SLA-1^∗04:01:01, indicating the weak binding strength. This suggests that the fixed combination of Asn¹⁵¹Val¹⁵² residues might be the key residues influencing the binding of viral CTL epitope peptides to SLA-1^∗13:01 and SLA-1^∗04:01:01 proteins. The crystal structure of SLA-1^∗04:01:01 revealed that Arg¹⁵⁶ in pocket D had a ‘one-ballot veto’ function in peptide binding, due to its flexible side chain (Zhang N. et al., 2011). However, the different epitope peptide conformations are caused not only by the flexibility of the side chains of the residues in the PBG, but also by skewing of the α1 and α2 helixes forming the PBG (Fan et al., 2016). Thus although the Asn¹⁵¹ residue of SLA-1^∗13:01 did not constitute part of the pockets of the PBG, it occurred in turn and random coil content and thus increased the flexibility support of pocket E, which may result in skewing of the α helixes forming the PBG. Furthermore, pocket E of SLA-1^∗13:01 was larger than that of SLA-1^∗04:01:01 as a result of the different surfaces of the concave Val¹⁵² and salient Glu¹⁵² residues, suggesting that pocket E in the SLA-1^∗13:01 protein might have fewer steric limitations and be able to accommodate more viral CTL epitope peptides. The flexible pocket E of SLA-1^∗04:01:01/151/152 might also adopt various conformations to accommodate the peptides through mutation of the Asp¹⁵¹ and Glu¹⁵² residues, leading to distinct peptide conformations, and might thus play critical biochemical roles in determining the peptide-binding motif of SLA-1^∗13:01. A strong correlation exists between high-affinity SLA-I binding peptides and the stability of the complexes, with high-affinity SLA-I binding peptides generally forming more stable complexes (Pedersen et al., 2016). Changes in several amino acid residues of SLA class I molecules could thus influence the affinity and stability of complexes, thereby affecting the peptide-binding specificity of SLA class I molecules. Furthermore, the binding specificity also could be affected by changes in amino acid polarity. Interestingly, in contrast to a previous study (Oleksiewicz et al., 2002), we found that the CSFV NS4A epitope peptide could not bind to SLA-1^∗04:01:01 but could bind to SLA-1^∗13:01, with relative OD values of up to 4.7. However, the NS4A epitope peptide could bind to the SLA-1^∗04:01:01/151/152 mutant protein when both the Asp¹⁵¹ and Glu¹⁵² residues were mutated simultaneously. This discrepancy may be because the in vitro refolding method for the protein complexes in the current study differed from that used in the previous study, and the binding specificity of the H4 w/o protein with synthetic (free) NS4A peptide was also not observed in the refold reactions (Oleksiewicz et al., 2002). Further work is therefore needed to determine the crystal structure of the NS4A peptide-SLA-1^∗04:01:01 complex to thoroughly explore this phenomenon. Certainly, it is beneficial if the promiscuity of SLA-1^∗13:01 and SLA-1^∗04:01:01 were investigated in terms of the peptide binding affinity and the stability of the complexes. Therefore, our future study will focused on the peptide binding affinity and stability to SLA class I molecules and stability of the SLA complex by using various methods, for example, luminescent oxygen channeling assay and scintillation proximity assay-based peptide-SLA dissociation. Furthermore, we also will attempt to perform molecular dynamics simulations to investigate the binding difference of SLA class I molecular to epitope peptides by using various software, for example, Discovery Studio and CHARMM.

We successfully constructed SLA class I complexes consisting of viral epitope peptides, the extracellular region of SLA-1 molecules, and β2m in vitro. We also detected the specific binding of seven SLA-1 proteins to five viral CTL epitope peptides through the expression and refolding of the protein complexes. The SLA-1^∗13:01, SLA-1^∗11:10, and SLA-1^∗11:01:02 proteins were able to specifically bind different CTL epitopes of CSFV and PRRSV, and the MHC restrictions of five epitopes were identified. Moreover, the fixed combination of Asn¹⁵¹Val¹⁵² residues might represent the key residues influencing the binding of viral several CTL epitope peptides to SLA-1^∗13:01 and SLA-1^∗04:01:01. The increased flexibility of pocket E in the SLA-1^∗13:01 protein might play a critical biochemical role in determining the peptide-binding motif of SLA-1^∗13:01. Characterization of the peptide-specific binding properties of SLA class I molecules provides essential information for the future identification of novel epitopes, as well as for the overall validation and analysis of currently available or newly developed CTL-based vaccines in swine. This information may also lead to improved understanding of the structural basis of CTL-based immune responses.

CG, HC, and LQ designed the experiments, analyzed the data and wrote the paper. CG, XH, JQ, QJ, and HL performed the experiments.

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.