Characterization of baboon class I major histocompatibility molecules. Implications for baboon-to-human xenotransplantation.

Increasingly strong medical and political pressures are stimulating consideration of the transplantation of baboon organs and cells into humans. Critical to the success of these xenotransplants is management of the immune system such that graft rejection and, in the case of bone marrow transplantation, graft-versus-host disease do not result in transplant failure. The polymorphic products of the major histocompatibility complex (MHC) are the primary barrier to successful allotransplantation, and here we describe class I MHC molecules from baboon (Papio anubis) to gain an understanding of how similarities and differences between baboon and human MHC molecules might affect xenograft survival and function. Comparative analyses of our five novel baboon class I molecules with defined HLA class I molecules demonstrate that the baboon class I molecule are up to 90% identical. Disparity between baboon class I proteins and their human homologues lies predominately at positions in the antigen-binding groove, while C-terminal portions of the class I heavy chain are more conserved between the two species. Such concentration of cross-species differences within the alpha1 and alpha2 domains involves a majority of substitutions at positions demonstrating polymorphism in human alleles; the location of substitutions distinguishing baboon and human molecules thus resembles the positioning of human class I allopolymorphisms. Because this preliminary characterization indicates that both baboon and human T cells with be restricted by xenogeneic class I molecules, immune responses triggered during baboon-to-human transplantation should mimic those arising during MHC mismatched human allotransplantation.

C-terminal epitope tagging facilitates comparative ligand mapping from MHC class I positive cells.

Purification of specific class I molecules prior to peptide ligand characterization is complicated by the presence of multiple class I proteins in most cell lines. Immortalized B, T, and tumor cell lines typically express endogenous HLA-A, -B, and -C; and most individuals from which the cell lines are derived are heterozygous at these loci. Antibodies specific for a particular HLA molecule may be used for purification, but allele-specific antibodies can be biased by ligands occupying the peptide-binding groove. Through the use of C-terminal tagging, we have developed a method of soluble HLA production such that downstream purification does not skew the peptide analysis of the examined molecule. Comparison of peptides eluted from HLA class I molecules with and without C-terminal tags demonstrates that addition of a tag does not abrogate the peptide binding specificity of the original molecule. Both pooled Edman sequencing and mass spectrometric sequencing identified no substantial differences in peptides bound by untailed, 6-HIS-tailed, and FLAG-tailed class I molecules, demonstrating that the peptide specificity of a given molecule is not distorted by either tag. This production methodology bypasses problems with isolation of specific molecules and permits ligand mapping and epitope discovery in a variety of pathogen-infected and tumor cell lines.

Allelic diversity within the high frequency Mamu-A2*05/Mane-A2*05 (Mane-A*06)/Mafa-A2*05 family of macaque MHC-A loci.

Macaque species serve as important animal models of human infection and immunity. To more fully scrutinize their potential in both the analysis of disease pathogenesis and vaccine development, it is necessary to characterize the major histocompatibility complex (MHC) class I loci of Macaca mulatta (Mamu), Macaca nemestrina (Mane), and Macaca fascicularis (Mafa) at the genomic level. The oligomorphic Mamu-A2*05/Mane-A2*05 (previously known as Mane-A*06) family of macaque MHC-A alleles has recently been shown to be present at high frequency in both Indian rhesus and pig-tailed macaque populations. Using a locus-specific amplification and direct DNA typing methodology, we have additionally found that the locus encoding this family is very prevalent (75%) among a sampling of 182 Chinese rhesus macaques and has a high prevalence (80%) within a larger, independent cohort of 309 pig-tailed macaques. Interestingly, among the Chinese rhesus macaques, only six alleles previously identified in Indian-origin animals were observed, while three recently identified in Chinese-origin animals and 25 new alleles were characterized. Among the pig-tailed macaques, we observed 1 previously known (Mane-A*06) and 19 new alleles. Examination of the orthologous locus in a preliminary sampling of 30 cynomolgus macaques showed an even higher presence (87%) of Mafa-A2*05 family alleles, with 5 previously identified and 15 new alleles characterized. The continued discovery of novel alleles and thus further diversity within the Mamu-A2*05/Mane-A2*05/Mafa-A2*05 family indicates that this MHC-A locus, although highly conserved across the three species of macaques, has remained a dynamic entity during evolution.

Large-scale production of class I bound peptides: assigning a signature to HLA-B*1501.

A peptide-based vaccine must be bound and presented by major histocompatibility complex class I molecules to elicit a CD8(+) T-cell response. Because class I HLA molecules are highly polymorphic, it has yet to be established how well a vaccine peptide that stimulates one individual's CD8(+) cytotoxic T lymphocytes will be presented by a second individual's different class I molecules. Therefore, to facilitate precise comparisons of class I peptide binding overlaps, we uniquely combined hollow-fiber bioreactors and mass spectrometry to assign precise peptide binding signatures to individual class I HLA molecules. In applying this strategy to HLA-B*1501, we isolated milligram quantities of B*1501-bound peptides and mapped them using mass spectrometry. Repeated analyses consistently assign the same peptide binding signature to B*1501; the degree of peptide binding overlap between any two class I molecules can thus be determined through comparison of their peptide signatures.

Cross-priming is under control of the relB gene.

Cross-priming is an important mechanism of intercell transfer of antigenic material leading to the specific activation of cytotoxic T lymphocytes. Dendritic cells (DCs) are considered the central antigen-presenting cell in cross-priming. Here we decided to probe the role of the relB gene, a regulator of DC differentiation, in the in vivo cross-priming of a model tumour antigen, TAP(-/-) murine embryo cells (MEC), expressing human adenovirus type 5 early region 1. To this end, we used relB(-/-) mutant mice to generate bone marrow (BM) chimeras as these possess few residual DC but are capable of initiating CD4+ and CD8+ T-cell responses in vivo. Our results show that relB(-/-) BM chimeras are unable to cross-prime CD8+ T cells, suggesting that the relB gene regulates cross-priming.

HLA-B polymorphism affects interactions with multiple endoplasmic reticulum proteins.

To explore the nature of amino acid substitutions that influence association with TAP, we compared a site-directed mutant of HLA-B*0702 (Y116D) to unmutated HLA-B7 in regard to TAP interaction. We found that the mutant had stronger association with TAP, and, in addition, with tapasin and calreticulin. These data confirm the importance of position 116 for TAP association, and indicate that (1) an aspartic acid at the 116 position can facilitate the interaction, and (2) association with tapasin and calreticulin is affected along with TAP. Furthermore, we tested three natural subtypes of HLA-B15, and found that a B15 subtype with a tyrosine at position 116 (B*1510) was strongly associated not only with TAP, but also with tapasin and calreticulin. In contrast, two B15 subtypes with a serine at position 116 (B*1518 and B*1501) exhibited very little or no association with any of these proteins. Thus, very closely related HLA-B subtypes can differ in regard to interaction with the entire assembly complex. Interestingly, when their surface expression was tested by flow cytometry, the HLA-B15 subtypes with little to no detectable intracellular assembly complex association had a slightly, yet consistently, higher level of the open heavy chain form than did the B15 subtype with intracellular assembly complex association. These data suggest that the relatively low strength or short length of interaction between endoplasmic reticulum proteins and natural HLA class I molecules can decrease their surface stability.

Novel alleles HLA-B*7802 and B*51022: evidence for convergency in the HLA-B5 family.

We have characterized two novel HLA-B alleles, B*7802 and B*51022. The Caucasian-derived variant B*7802 most resembles the African-derived variant B*7801, from which B*7802 differs by two nucleotides. Only one of these modifications, however, is translated: a tyrosine for aspartate substitution occurs at residue 74 in B*7802, while the second nucleotide difference reflects a proximal synonymous substitution in codon 23. A second variant, B*51022, differs synonymously only at codon 23 from B*51021. Comparative analysis of the B5 CREG demonstrates that other pairs of B5 alleles differ synonymously only at codon 23 or synonymously at codon 23 and non-synonymously at a second more distal location. Contrary to the genesis of like pairs of B5 alleles via introduction of coordinate yet distant mutagenic events onto a single B5 progenitor, we postulate that synonymously different B5 progenitor molecules, B5ATT and B5ATC, are evolving in convergence to generate homologous B5 allele pairs differing silently at codon 23. Our finding that B*7802 is a single amino acid away from complete convergence with B*7801 and that B*51022 and B*51021 are in complete convergence is exemplary of such evolution.

Complexity among constituents of the HLA-B*1501 peptide motif.

Analysis of peptides derived from HLA class I molecules indicates that thousands of unique peptides are bound by a single molecular type, and sequence examination of the pooled constituents yields a motif which collectively defines the peptides bound by a given class I molecule. Motifs resulting from pooled sequencing are then used to infer whether particular viral and tumor protein fragments might serve as class I-presented peptide therapeutics. Still undetermined from a pooled motif is the breadth or range of peptides in the population which are brought together to form the pooled motif, and it is therefore not yet known how representative of the population a pooled motif is. By employing hollow fiber bioreactors for large-scale production of HLA class I molecules, sufficient peptides are produced to investigate individual subsets of peptides comprising a motif. Edman sequencing and mass spectrometric analysis of peptides eluted from HLA-B*1501 reveal that many peptide sequences fail to align with either the N- or C-terminal anchors predicted for the B*1501 peptide motif through whole pool sequencing. These analyses further reveal auxiliary anchors not previously detected and peptides significantly larger and smaller than the predicted nonamer, ranging from 6 to 12 amino acids in length. These results demonstrate that constituents of the B*1501 peptide pool vary markedly in comparison with one another and therefore in comparison with previously established B*1501 motifs, and such complexity indicates that many of the peptide ligands presented to CTL cannot be predicted using class I consensus motifs as search criteria.

HLA-B15 peptide ligands are preferentially anchored at their C termini.

Therapies to elicit protective CTL require the selection of pathogen- and tumor-derived peptide ligands for presentation by MHC class I molecules. Edman sequencing of class I peptide pools generates "motifs" that indicate that nonameric ligands bearing conserved position 2 (P2) and P9 anchors provide the optimal search parameters for selecting immunogenic epitopes. To determine how well a motif represents its individual constituents, we used a hollow-fiber peptide production scheme followed by the mapping of endogenously processed class I peptide ligands through reverse-phase HPLC and mass spectrometry. Systematically mapping and characterizing ligands from B*1508, B*1501, B*1503, and B*1510 demonstrate that the peptides bound by these B15 allotypes i) vary in length from 7 to 12 residues, and ii) are more conserved at their C termini than their N-proximal P2 anchors. Comparative peptide mapping of these B15 allotypes further pinpoints endogenously processed ligands that bind to the allotypes B*1508, B*1501, and B*1503, but not B*1510. Overlapping peptide ligands are successful in binding to B*1501, B*1503, and B*1508 because these B15 allotypes share identical C-terminal anchoring pockets whereas B*1510 is divergent in the C-terminal pocket. Therefore, endogenous peptide loading into the B15 allotypes requires that a conserved C terminus be anchored in the appropriate specificity pocket while N-proximal anchors are more flexible in their location and sequence. Queries for overlapping and allele-specific peptide ligands may thus be contingent on a conserved C-terminal anchor.

Alpha-2 domain polymorphism and HLA class I peptide loading.

Diversity within the class I HLA antigen binding groove is positioned to moderate the presentation of peptide ligands. Polymorphism is widely dispersed about the peptide binding groove, and unravelling the functional significance of a given polymorphism requires comparative analysis of peptides presented by class I subtypes differing at the position(s) in question. Previous studies have demonstrated that not all class I polymorphisms act equally, and to determine the impact of substitutions specifically located in the alpha2 domain, peptides purified from B*1501, B*1512, B*1510, and B*1518 were examined by pooled Edman sequencing and comparative mass spectrometric analysis. Molecule B*1512 differs from B*1501 at residues 166 (Glu to Asp) and 167 (Trp to Gly) of the alpha2 domain. The pooled motif and ion mass ligand maps for B*1512 tightly matched those of B*1501, demonstrating that the 166/167 polymorphism between B*1501 and B*1512 has little impact upon ligand presentation. Although the 166/167 polymorphism minimally affects peptide binding preferences, this polymorphism makes B*1512 and B*1501 quite distinct by serology. We then compared the B70 molecules B*1510 and B*1518. The two are almost indistinguishable by serology and differ only by an alpha2 polymorphism at 116. Comparative peptide mapping shows that a Tyr to Ser polymorphism at 116 drastically changes the ligands bound by B*1510 and B*1518; no overlaps could be found. Polymorphisms in alpha2 therefore vary from subtle to extreme in the manner by which they moderate ligand presentation, and serologic crossreactivity did not reflect the ligands presented by these B15 subtypes.

Population of the HLA ligand database.

We have established an HLA ligand database to provide scientists and clinicians with access to Major Histocompatibility Complex (MHC) class I and II motif and ligand data. The HLA Ligand Database is available on the world wide web at http://hlaligand.ouhsc.edu and contains ligands that have been published in peer-reviewed journals. HLA peptide datasets prove useful in several areas: ligands are important as targets for various immune responses while algorithms built upon ligand datasets allow identification of new peptides without time-consuming experimental procedures. A review of the HLA class I ligands in the database identifies strengths and deficiencies in the database and, therefore, the utility of the dataset for identifying new peptides. For instance, 212 HLA-A phenotypes exist of which 23 have a motif determined and 43 have peptides characterized. In terms of number of ligands, HLA-A*0201 has 258 characterized ligands, A*1101 has 25 peptides, while the remaining two-thirds of the HLA-A phenotypes have less than 10 associated peptide sequences. Characterization of ligands and motifs remains roughly the same at the HLA-B locus while the peptides of the HLA-C locus tend to be less characterized. These data show that 74% of HLA class I molecules do not have ligands represented in the database and thus algorithms based on the dataset could not predict ligands for a majority of the US population. Building upon this dataset and knowledge of HLA allelic frequencies, it is possible to plan a systematic expansion of the HLA class I ligand database to better identify ligands useful throughout the population.

Gorillas with spondyloarthropathies express an MHC class I molecule with only limited sequence similarity to HLA-B27 that binds peptides with arginine at P2.

The human MHC class I gene, HLA-B27, is a strong risk factor for susceptibility to a group of disorders termed spondyloarthropathies (SpAs). HLA-B27-transgenic rodents develop SpAs, implicating HLA-B27 in the etiology of these disorders. Several nonhuman primates, including gorillas, develop signs of SpAs indistinguishable from clinical signs of humans with SpAs. To determine whether SpAs in gorillas have a similar HLA-B27-related etiology, we analyzed the MHC class I molecules expressed in four affected gorillas. Gogo-B01, isolated from three of the animals, has only limited similarity to HLA-B27 at the end of the alpha1 domain. It differs by several residues in the B pocket, including differences at positions 45 and 67. However, the molecular model of Gogo-B*0101 is consistent with a requirement for positively charged residues at the second amino acid of peptides bound by the MHC class I molecule. Indeed, the peptide binding motif and sequence of individual ligands eluted from Gogo-B*0101 demonstrate that, like HLA-B27, this gorilla MHC class I molecule binds peptides with arginine at the second amino acid position of peptides bound by the MHC class I molecule. Furthermore, live cell binding assays show that Gogo-B*0101 can bind HLA-B27 ligands. Therefore, although most gorillas that develop SpAs express an MHC class I molecule with striking differences to HLA-B27, this molecule binds peptides similar to those bound by HLA-B27.