Structure of class I major histocompatibility antigens

Major histocompatibility complex class II deficiency: a clinical review

Major histocompatibility complex Class II deficiency or bare lymphocyte syndrome is a rare combined immunodeficiency that accounts for 5% of all cases of severe combined immunodeficiency. The syndrome is characterized by a lack of human leucocyte antigen Class II gene expression, absence of cellular and humoral T-cell immune response to foreign antigens, and impaired antibody productions, resulting in extreme susceptibility to viral, bacterial and fungal infections. In some patients, there is a reduced cell surface expression of human leucocyte antigen Class I molecules also. Major histocompatibility complex Class II deficiency is an autosomal recessive disease, most frequent in the Mediterranean area. The disease is caused by impaired gene regulation involving trans-acting proteins. Somatic cell genetics using cell fusion experiments identified four complementation groups, all resulting in the same clinical manifestation. Two regulatory genes have been identified so far: Class II trans activator and regulatory factor X5. Supportive treatment includes intravenous gammaglobulin and prophylaxis against Pneumocystis carinii. The only curative treatment is bone-marrow transplantation.

Bioengineered soluble HLA-B7. Genesis, characterization, and occurrence of dimerization

A soluble, secreted form of HLA-B7 was engineered by replacing the exons encoding the transmembrane and cytoplasmic domains of the B7 gene with a CI. The modified gene, gsB7, transfected into J27.2 or C1R cell lines, produced a secreted protein, sB7, serologically recognized as B7. Size fractionation showed one species of sB7 at the approximately 55 kD expected for an sB7 alpha-chain-beta 2m heteroduplex, and another at approximately 120 kD which had the same constituent chains and was a dimer of the 55-kD species. Dimer formation appeared to be related to protein concentration but not to disulfide bridging. The sB7 heavy chain on SDS-PAGE showed a doublet at approximately 39 and approximately 42 kD; enzyme analysis indicated that the two bands differed only by a carboxyl terminal polypeptide. Analysis of gsB7 transfectants' mRNA by Northern blots and PCR revealed message fully spliced or with retained CI, accounting for the 39- and 42-kD bands, respectively, and apparently untranslated message with I3 retained. sB7 was not detectable on the surface of gsB7 transfectants by CTLs, nor did it inhibit those CTLs. Production of the sB7 protein provides a ready, consistent source of soluble class I antigen for further study, including test materials for tolerogenicity studies in animal models.

Soluble form of an HLA-B7 class I antigen specifically suppresses humoral alloimmunization

A soluble HLA-B7 molecule, designated sB7 and generated by genetically engineering the B7 gene to remove the transmembrane and cytoplasmic domains, was tested as a tolerogen. Supernatants from cultures of C1R cells transfected with the gene for sB7 were harvested and concentrated, as were control supernatants. From days -17 to -1, C57Bl/6 mice were pretreated with a total of 11 intraperitoneal doses of 1.0 microgram each of sB7 or appropriate control supernatant, and then were challenged intraperitoneally on each of days 0, 7, and 14 with 10(6) C1R-B7 cells (expressing surface HLA-B7). Antibody kinetics revealed (1) anti-B7 was not induced after sB7 pretreatment; (2) the anti-B7 response of sB7-pretreated mice was marginal and of apparent low avidity compared with the brisk anti-B7 response of control mice; (3) none of the mice made antibody to a control HLA antigen, A24; (4) all mice made strong antibody responses to the non-B7 surface antigens of C1R; (5) free sB7 did not appear in the blood of the treated mice; and (6) all mice appeared to be generally healthy. These data show soluble B7 antigen is not immunogenic and appears to specifically block humoral immune response to cell membrane-bound HLA-B7 in a nontoxic manner.

Participation of a novel 88-kD protein in the biogenesis of murine class I histocompatibility molecules

Chemical cross-linking and gel permeation chromatography were used to examine early events in the biogenesis of class I histocompatibility molecules. We show that newly synthesized class I heavy chains associate rapidly and quantitatively with an 88-kD protein in three murine tumor cell lines. This protein (p88) does not appear to possess Asn-linked glycans and it is not the abundant ER protein, GRP94. The class I-p88 complex exists transiently (t1/2 = 20-45 min depending on the specific class I heavy chain) and several lines of evidence suggest that p88 dissociates from the complex while still in the ER. Dissociation is not triggered upon binding of beta 2-microglobulin to the heavy chain (t1/2 = 2-5 min). However, the rate of dissociation does correlate with the characteristic rate of ER to Golgi transport for the particular class I molecule studied. Consequently, dissociation of p88 may be rate limiting for ER to Golgi transport. Class I molecules bind antigenic peptides, apparently in the ER, for subsequent presentation to cytotoxic T lymphocytes at the cell surface. p88 could promote peptide binding or it may retain class I molecules in the ER during formation of the ternary complex of heavy chain, beta 2-microglobulin, and peptide.

Comparative behavior of membrane protein-antibody complexes on motile fibroblasts: implications for a mechanism of capping

A characteristic feature of fibroblast locomotory activity is the rearward transport across the leading lamella of various materials used to mark the cell surface. The two processes most frequently invoked as explanations for this transport phenomenon, called capping, are (a) retrograde membrane flow arising from directed membrane insertion and (b) rearward cortical cytoskeletal flow arising from cytoskeletal assembly and contraction. The retrograde lipid flow hypothesis, the most current form of the membrane flow scheme, makes explicit predictions about the movement of membrane proteins subjected to the postulated rearward lipid flow. Several of these predictions were tested by comparing the behavior of four membrane proteins, Pgp-1, Thy-1, H-2, and influenza HA0, identified by fluorescent antibodies. With the exception of Pgp-1, these proteins were uniformly distributed under nonaggregated conditions but were capped when aggregated into patches. In contrast, Pgp-1 was capped in similar time frames in both nonaggregated and aggregated states where the lateral diffusion coefficients were very different. Furthermore, the capping behavior of two tagged membrane proteins was markedly different yet both had similar diffusion coefficients. The results from these tests disprove the bulk membrane flow hypothesis and are at odds with explicit predictions of the retrograde lipid flow hypothesis for the mechanism of capping. This work, therefore, supports the alternative cytoskeletal-based mechanism for driving capping. Requirements for coupling cytoskeletal movement to membrane components are discussed.

Delineation of antigen contact residues on an MHC class II molecule

This report describes a detailed mutational analysis of a major histocompatibility complex class II molecule--the alpha chain of the Ak complex. Each residue from 50-79 was replaced by an alanine, and the effects on recognition of Ak by panels of antibodies and T cells determined. The results provide the strongest existing experimental evidence that the antigen binding site on a class II molecule can be modelled on the crystal structure of a class I molecule. The data have also permitted the delineation of residues that actually contact antigenic peptides.

Soluble HLA class I antigen secretion by normal lymphocytes: relationship with cell activation and effect of interferon-gamma

HLA class I antigens are thought to be integral membrane proteins. However, soluble forms of these molecules have been detected. Our laboratory has recently shown that the predominant form of these soluble proteins present in human serum, spleen tissue and culture supernatant of activated lymphocytes exhibits molecular weight and structure similar to classical HLA class I antigens, but lacks HLA A or B polymorphic determinants. In the present study, the secretion of such soluble proteins by lymphocytes has been further explored. Phytohaemagglutinin-stimulated normal lymphocytes secrete considerable quantities of soluble HLA (sHLA) class I proteins. This secretion seems to be a general property of lymphocytes, since activation of T as well as B cells by appropriate mitogens equally induce sHLA I secretion. Lymphocytes require RNA and protein synthesis, but not DNA synthesis, for the secretion to occur. Kinetic studies reveal that maximal sHLA I secretion precedes the peak of DNA synthesis by 24 h. In vitro stimulation with antigens or alloantigens also provokes sHLA I secretion. Moreover, this phenomenon has also been detected for in vivo-activated lymphocytes, as enhanced spontaneous sHLA I secretion was observed in cultures of low-density blastic B and T cells, and of blood lymphocytes obtained from normal subjects who had received a booster immunization 5 days earlier. Interferon-gamma (IFN-gamma) increases the expression of membrane-bound class I antigens but does not induce any sHLA I secretion, suggesting that both molecules are under different regulatory mechanisms. Our results indicate that human lymphocytes, upon stimulation, actively secrete considerable amounts of a soluble form of these biologically relevant proteins.

Prediction of the secondary structure and functional sites of major histocompatibility complex molecules

A method for the theoretical prediction of the antigenic determinants and the antigen-interactive receptor sites of immunological proteins from their primary structure would constitute a useful tool for their study. Such a method developed in this laboratory uses hydrophilicity, accessibility, flexibility, and recognition profiles, together with the predicted secondary structure (alpha-helices, beta-sheets, and turns). The secondary structure is determined by a modification of the method of Lim (1974), as described below. A study of human and mouse class I and class II major histocompatibility complex (MHC) antigens, central to the regulation of immune responses and to the phenomenon of graft rejection, was carried out using the above method. Comparison of the predictions with some of the available experimental and theoretical information supports the validity and usefulness of the approach.

Biochemical analysis of the rat MHC class I antigens RT1.Aa, RT1.Fa and Pa

In DA strain rats, there are two other MHC class I loci (Pa and RT1.Fa) in the vicinity of the classical class I locus RT1.Aa. The Pa antigen is the pregnancy-associated antigen, and it was detected by antibodies elicited in WF females pregnant by DA males without any other immunization. The Fa antigen was detected by a monoclonal antibody raised by alloimmunization. In the present work, the Aa, Fa and the Pa antigens have been compared by HPLC peptide mapping and by isoelectric focusing after their isolation by appropriate monoclonal antibodies. All the three antigens are identical in primary structure with respect to lysine, methionine, asparagine and the aromatic amino acid residues, but they differ from one another with respect to glutamic acid and/or aspartic acid residues. The pI values of the antigens differ slightly. All three antigens have two identical N-linked glycans, but the Fa antigen has an additional N-linked glycan. Based on the available amino acid sequence of the Pa antigen, it can be concluded that both Aa and Pa antigens are devoid of glycosylation in the second domain. This lack of glycosylation of the classical antigen Aa is unique for the rat, since classical class I antigens of the mouse show glycosylation in the first and second, and sometimes in the third domain, and those in the human, in the first domain only. The high degree of similarity among the Aa, Fa, and Pa molecules that this study indicates is also unique for the rat, since antigens encoded by different class I genes of the same haplotype are quite disparate in the mouse and human.

Isolation of a specific membrane protein by immunoaffinity chromatography with biotinylated antibodies immobilized on avidin-coated glass beads

Avidin-coated, solid glass beads have been used as an immobilization support for attaching biotinylated antibodies. These beads have been packed into analytical, semi-preparative and preparative columns and used to isolate the B27 histocompatibility anigen (HLA) from human lymphocytes. The beads provided a suitable column material for all three chromatographic procedures and, depending on the size of the immunoaffinity column, B27 antigen could be isolated in nanogram to microgram quantities. Polyacrylamide gel electrophoresis demonstrated the presence of only a single band in the immunoaffinity peaks isolated by all three procedures. Enzyme-linked immunosorbent assay analysis of these immunoaffinity-isolated materials revealed that they were biologically active and could be used to determine the levels of anti-B27 antibodies in clinical studies.

Secretion of genetically engineered human/mouse class I antigens

Two soluble, secreted forms of HLA-B7 were engineered by the creation of hybrid human/mouse molecules containing the polymorphic 5' region of the HLA-B7 gene and the secretory 3' region of the mouse Q10d gene. The hybrid, designated F1, is the first construct with only human extracytoplasmic domains, consisting of exons for the leader peptide and the three extracellular domains (alpha 1, alpha 2, alpha 3) of B7 spliced to the exons for the Q10d truncated transmembrane and 3' untranslated (3'UT) sequences. The second construct, designated C2, is similar but has the human alpha 3 replaced by the Q10 alpha 3 domain. Protein product from each construct was best demonstrated after gene transfection into the J27.2 cell line. In particular, secretion of the F1 product proves that the Q10 alpha 3 domain is not necessary for secretion of class I/Q10 hybrids. Moreover, the two soluble B7 forms, which differ only in their alpha 3 domain, are similarly recognized by monoclonal antibodies W6/32 (anti-HLA-ABC), BBM.1 (anti-human beta 2 microglobulin), and allo-B7-antibody, but differentially recognized by monoclonal antibody Q1/28 (anti-HLA class I heavy chain). Production of such soluble hybrid class I molecules in large amounts should allow critical structural and functional studies of these proteins.

Different class I antigen oligosaccharides on a murine tumor and a lectin-resistant variant are not responsible for the differential recognition of the tumors by CTL

Previous studies have shown that whereas a highly malignant mouse cell line termed MDAY-D2 (d haplotype) does not elicit a detectable response by cytotoxic T lymphocytes (CTL) in DBA/2 mice, strong anti-tumor CTL are generated against a wheat-germ-agglutinin-resistant variant, designated MDW3. Additional evidence suggests these anti-MDW3 CTL may not be a consequence of a unique antigenic determinant on the variant cells. Because MDW3 cells are expected to differ from MDAY-D2 cells in their surface carbohydrate structures (due to their lectin resistance) and Class I major histocompatibility molecules play a crucial role in CTL-mediated responses, we speculated that the Asn-linked oligosaccharides present on Class I molecules of MDAY-D2 and MDW3 might be different and could potentially influence recognition analyses and Con A-Sepharose affinity chromatography clearly demonstrated that the oligosaccharides isolated from the H-2Dd molecule of MDAY-D2 cells are larger and more highly branched than those of the MDW3 variant. Taken together with the finding that anti-MDW3 CTL are restricted by H-2Dd, these results suggested that the larger H-2Dd oligosaccharides on MDAY-D2 cells could potentially mask or perturb determinants required for recognition by these CTL. To test this postulate, the surface Class I oligosaccharides of both MDAY-D2 and MDW3 cells were converted to simpler hybrid structures by treatment with the oligosaccharide processing inhibitor, swainsonine. However, no effect was observed on the lysis or binding of either MDAY-D2 or MDW3 cells by anti-MDW3 CTL. Thus, the results do not support the possibility that the larger H-2Dd oligosaccharides on MDAY-D2 cells are, in themselves, responsible for the poor recognition of the parent tumor by anti-MDW3 CTL. Our data do indicate, however, that CTL target binding and effector functions are not dependent on the fine structure of complex Asn-linked carbohydrates present on Class I molecules and possibly on other, accessory molecules at the target cell surface, since MDW3 cells maintained their sensitivity to lysis by CTL following swainsonine treatment.

The relationship between MHC antigen expression and metastasis

From the studies summarized here a complex picture of the role played by MHC products in determining tumorigenicity and metastasis is emerging. In order to be able to understand this relationship better, it is necessary to consider several factors. 1. Each tumor system or neoplastic tissue is unique, and its behavior reflects the influence of cell-specific characteristics, as well as its ability to modulate other cells and tissues--including cells belonging to the immune system--and also to be modulated by other cells and soluble factors. 2. Since metastasis formation is a multistep process in which only small subpopulations of tumor cells with complex and defined phenotypes are able to colonize secondary tissues, elimination of even one single phenotypic component of this structured process can easily reverse the metastatic capacity of the cells. Acquisition of metastatic ability, on the other hand, would be a more difficult task, since any new characteristic expressed by the cells or induced experimentally, such as gene transfection or results of IFN treatment, must be expressed in a temporal manner and in concert with other cellular characteristics. Therefore, an experimental protocol measuring a specific element in determining metastasis can easily produce conflicting results, depending on the type of cells and genetic background of the host studied. 3. The level of specific MHC products on tumor cells is one among many other cell characteristics that may determine the metastatic potential of cells. Moreover, each of the class 1 MHC products, and the relationship among them, including other than the classical K, L, or D products (Brickell et al., 1983), should be regarded as independent entities, with possible different regulatory roles in cell-cell recognition, in a general sense, which may be involved in determining invasiveness and homing as well as recognition by the immune system. 4. Both specific T-cell and nonspecific natural mediated immunity (which is much less understood) are involved in the selection of the metastatic cell population. 5. Immunogenicity of tumors is not necessarily determined by high levels of MHC antigen expression; it is also dependent on the level of TSA. Thus, immunoselection mediated by T lymphocytes during metastasis formation could be directed against both MHC and TSA antigens. Therefore, low expression of MHC antigens by metastatic cells as a result of immunoselection is not always observed.(ABSTRACT TRUNCATED AT 400 WORDS)

Microrecombinations generate sequence diversity in the murine major histocompatibility complex: analysis of the Kbm3, Kbm4, Kbm10, and Kbm11 mutants

The mechanism that generates spontaneous mutants of the Kb histocompatibility gene was analyzed. Nucleotide sequence analysis of four mutant genes (Kbm3, Kbm4, Kbm10, and Kbm11) revealed that each mutant K gene contains clustered, multiple nucleotide substitutions. Hybridization analyses of parental B6 genomic DNA and cloned class I genes with mutant-specific oligonucleotide probes, followed by sequence analyses, have identified major histocompatibility complex class I genes in the K, D, and Tla regions (K1, Db, and T5, respectively) that contain the exact sequences as substituted into mutant Kb genes. These data provide evidence for the hypothesis that the mutant Kb genes are generated by a microrecombination (gene conversion) mechanism that results in the transfer of small DNA segments from class I genes of all four regions of the major histocompatibility complex (K, D, Qa, and Tla) to Kb. Many of the nucleotides substituted into the mutant Kb genes were identical to those found in other naturally occurring K alleles such as Kd. Thus, we propose that the accumulation of microrecombination products within the K genes of a mouse population is responsible for the high sequence diversity among H-2 alleles.

Microrecombinations generate sequence diversity in the murine major histocompatibility complex: analysis of the Kbm3, Kbm4, Kbm10, and Kbm11 mutants

The mechanism that generates spontaneous mutants of the Kb histocompatibility gene was analyzed. Nucleotide sequence analysis of four mutant genes (Kbm3, Kbm4, Kbm10, and Kbm11) revealed that each mutant K gene contains clustered, multiple nucleotide substitutions. Hybridization analyses of parental B6 genomic DNA and cloned class I genes with mutant-specific oligonucleotide probes, followed by sequence analyses, have identified major histocompatibility complex class I genes in the K, D, and Tla regions (K1, Db, and T5, respectively) that contain the exact sequences as substituted into mutant Kb genes. These data provide evidence for the hypothesis that the mutant Kb genes are generated by a microrecombination (gene conversion) mechanism that results in the transfer of small DNA segments from class I genes of all four regions of the major histocompatibility complex (K, D, Qa, and Tla) to Kb. Many of the nucleotides substituted into the mutant Kb genes were identical to those found in other naturally occurring K alleles such as Kd. Thus, we propose that the accumulation of microrecombination products within the K genes of a mouse population is responsible for the high sequence diversity among H-2 alleles.

Signals controlling alternative splicing of major histocompatibility complex H-2 class I pre-mRNA

The use of alternative splice acceptor sites during the removal of intron 7 in pre-mRNA splicing produces two forms of H-2Kb protein: the predominant form, derived from a transcript that has spliced at the upstream splice acceptor site for exon 8 (long exon 8), and a Kb molecule derived from a transcript that has spliced at the downstream acceptor site for exon 8 (short exon 8). We have identified a potential lariat branch point adenosine for the upstream acceptor splice site. This adenosine is found 28 bp from the splice junction and is contained in the sequence AGTGATGG. D-region genes, which use only the downstream splice site, have the sequence AGTGGTGG. We have used in vitro mutagenesis to change this A of the H-2Kb gene to G and have made the reciprocal change in H-2Dd. Elimination of this adenosine in H-2Kb alters the pattern of pre-mRNA splicing and results in a predominance of the Kb molecules with short exon 8 encoded sequences. However, the addition of an adenosine in H-2Dd is not sufficient to direct splicing to the upstream site.

Evidence that multiple residues on both the alpha-helices of the class I MHC molecule are simultaneously recognized by the T cell receptor

Single amino acid substitutions at nine different positions on the H-2Kb molecules from in vitro-mutagenized, immunologically altered, somatic cell variants were correlated with their patterns of recognition by monoclonal antibodies (MAbs) and allogeneic cytotoxic T lymphocyte (CTL) clones. While MAbs were found to detect spatially discrete, domain-specific sites, CTLs interacted simultaneously with multiple residues on the alpha 1 and alpha 2 domains of the Kb molecule. The computer graphic three-dimensional Kb model structure showed that, of the seven CTL-specific residues analyzed, six residues were located on the alpha-helical regions of the two domains. Every CTL clone was found to interact with a distinct pattern of residues composed of a specific subset of the CTL-specific residues.

H-2K molecules have two different C-termini, one of which is K-region specific

Amino acid sequences encoded by exon 8 of the H-2K and H-2D/L genes appear to be locus specific. The majority of H-2Kb molecules contain 10 amino acids that are derived from exon 8. In contrast, the H-2Db, -Dd and -Ld molecules contain only one amino acid which is encoded by exon 8, even though the genetic information exists to encode 10 amino acids analogous to those encoded by the majority of H-2Kb transcripts. We have produced a rabbit anti-peptide serum reactive with the exon 8 encoded sequence of H-2Kb (alpha K-C) that specifically immunoprecipitates a molecule of 45 K mol. wt from spleen cell lysates of b, d, p, q and k haplotype mice. Further analysis by Western blots indicated that virtually all mouse strains express a 45 K protein reactive with alpha K-C. In sequential immunoprecipitations of spleen cell lysates from b, q, p and d haplotype mice using alpha K-C followed by H-2K or H-2D private specificity alloantisera, the anti-peptide serum removed nearly all of the molecules reactive with the anti-H-2K alloantisera (except in the k haplotype) and none of the molecules reactive with the anti-H-2D serum. In addition, no D-region molecules possessing a long C-terminal sequence were detected with an antiserum directed against a representative D-region long C-terminal peptide. We conclude that even though the genetic information for an extended exon 8 exists in K, D and L locus genes, apparently only K-region molecules are expressed with such a C-terminus. Furthermore, in most haplotypes the great majority of H-2K molecules are produced using long exon 8; however, H-2Kk is produced mostly from short exon 8. The absence or presence of key adenosine residues is predicted to be responsible for the variability in class I exon 8 splicing.

The H-2Kkml mutation: a single nucleotide substitution is responsible for multiple functional differences in a class I MHC molecule

Nucleotide sequence analysis of mRNA from the H-2K locus of the CBA.M523 mouse, which has the class I murine MHC mutation H-2Kkml, has established the only alteration to be at the codon for amino acid position 152 as compared to the sequence of standard Kk from both the AKR and CBA inbred mouse lines. Complete sequence information for the nucleotides coding for amino acids 1-292, which includes all of the extracellular protein domains, demonstrated an A----C alteration in the codon for amino acid 152 as compared to the standard Kk sequence, changing Asp (GAT) in Kkml. The GCT codon occurring in Kkml may be the result of a gene conversion in Kkml. The GCT codon occurring in Kkml may be the result of a gene conversion event because a potential donor gene, the pH-2III pseudogene of H-2k, is transcribed in the CBA.M523 mouse and has a GCT codon at amino acid position 152. This sequence information obtained for Kkml also demonstrates that Kk gene transcripts from two genetically distinct inbred mouse lines, CBA and AKR, are completely identical. Finally, several other murine and human class I MHC variants have similar alterations at amino acid position 152 which result in altered biological functions. This information suggests that amino acid 152 is an important part of a T-cell-recognized antigenic determinant on MHC class I antigens.

Differential expression of MHC class I antigens on the placenta of the rat. A mechanism for the survival of the fetal allograft

In some mating combinations in rats, there is a maternal antibody response to the maternal antigenic components of the placenta without any previous immunization of the mother. The highest response occurs in the WF (u) female mated to the DA (a) male, and it is against a unique MHC-encoded class I antigen, the Pa antigen, and not against the major allele-specific transplantation antigen of the DA strain, RT1.Aa. The development of mAbs to the Pa and Aa antigens allowed us to localize these antigens on the placenta and to explore the reason for the differential antibody response to them using immunohistochemical and biochemical techniques. Both antibodies reacted with the WF X DA placenta and stained the endovascular and interstitial trophoblast of the decidua, the basal trophoblast, Reichert's membrane, and the yolk sac epithelium, but they did not stain the labyrinthine trophoblast. Blocking studies showed that each antibody reacted with a separate molecule in the placenta. Anti-class II mAbs reactive with the a or u haplotype did not stain the WF X DA, DA X DA, or WF X WF placenta; hence, there are no class II antigens in the placenta. Electron microscopic studies of the semiallogeneic WF X DA placenta using the immunogold technique with both single- and double-labeling showed that only the Pa antigen was expressed on the surface of the basal trophoblast, but that both the Pa and Aa antigens were in the cytoplasm of these cells; neither antigen was found in the labyrinthine trophoblast. By contrast, the placenta from the syngeneic DA X DA mating expressed both the Pa and Aa antigens on the surface of the basal trophoblast as well as in the cytoplasm; neither antigen was found in the labyrinthine trophoblast. These observations were quantified morphometrically using electron photomicrographs of single-labeled tissues. Both the Pa and Aa antigens isolated from the plasma membrane of lymphocytes have heavy chains of 46 kD, but those antigens isolated from the plasma membrane of basal trophoblast cells have heavy chains of 43 kD. Based on densitometric measurements of autoradiographs, the Pa/Aa ratio in the basal trophoblast membrane is 23.5, whereas it is 0.46 in lymphocyte membranes. These studies show that there is differential regulation of the expression of class I antigens on basal trophoblast cells in semiallogeneic pregnancies, but not in syngeneic pregnancies, such that the major allele-specific transplantation antigen is scarcely expressed on the surface of the basal trophoblast.(ABSTRACT TRUNCATED AT 400 WORDS)

Site-directed mutagenesis of class I HLA genes. Role of glycosylation in surface expression and functional recognition

We have investigated the role of the carbohydrate moiety on the HLA-B7 molecule in mAb and CTL recognition using oligonucleotide-directed mutagenesis and gene transfer techniques. A conservative substitution of asparagine to glutamine at amino acid 86 in HLA-B7 was created to abolish the unique glycosylation site present on all HLA molecules. A second mutant B7 molecule was made by substituting asparagine-aspartic acid-threonine for the resident lysine-aspartic acid/lysine tripeptide at amino acids 176-178, thus creating an N-linked glycan at amino acid 176, which is additionally present on all known murine H-2 class I antigens. Upon gene transfer into mouse and human cell recipients, the HLA-B7M176+ mutant and normal HLA-B7 expressed identical levels of surface protein. However, the binding of two mAbs (MB40.2 and MB40.3) thought to recognize different epitopes of the HLA-B7 molecule was completely eliminated. In contrast, the HLA-B7M86- mutant displayed no surface expression (mouse L cells) or minimal surface expression (human RD cells or mouse L cells coexpressing human beta 2 microglobulin [beta 2m]) after indirect immunofluorescence (IIF) and flow cytometric analysis with a panel of 12 HLA-B7 mAb reactive with monomorphic and polymorphic determinants. Immunoprecipitation analysis demonstrated that intracellular denatured mutant protein was present. Tunicamycin treatment did not rescue the expression of HLA-B7M86- antigens to the cell surface; while interferon did induce higher levels of surface expression. Tunicamycin treatment also did not allow binding of the mAbs MB40.2 or MB40.3 to HLA-B7M176+ mutant antigens, suggesting that the carbohydrate moiety itself was not directly involved in the recognition or conformation of these mAb epitopes. Further mutation of the B7M86- molecule to create a glycan moiety at amino acid position 176 (B7M86-/176+) did not rescue normal levels of surface expression. Finally, neither mutation was seen to affect recognition by a panel of 12 allospecific CTL clones. The low expression of HLA-B7M86- on the surface of human cell transfectants was sufficient to achieve lysis, albeit at a reduced efficiency, and lysis could be increased by interferon induction of higher levels of expression. Thus, the carbohydrate moiety on HLA antigens plays a minimal or nonexistent role in recognition by available mAb and allospecific CTL clones.

Two forms of HLA class I molecules in human plasma

Soluble HLA-A,-B antigens have previously been detected in human plasma. More recently, these molecules have been demonstrated to be secreted in water soluble form by cell lines and peripheral blood lymphocytes in vitro due to RNA splicing events which delete exon five from a fraction of class I antigen transcripts. In order to determine whether plasma HLA-A,-B molecules arise by this mechanism, their biochemical properties are analyzed. Based upon molecular weight, detergent binding, and lipid binding properties, it was demonstrated that two forms of HLA-A,-B molecules are present in the plasma. A 43 kd form that binds both lipids and detergents is presumed to be a shed version of the membrane form. However, a 39 kd form that fails to bind lipids and detergents has properties identical to those molecules secreted in vitro. It is suggested that these molecules arise by alternative splicing events in vivo that are identical to those characterized in vitro. Although all tested individuals display shed and/or secreted plasma class I molecules, it is striking that HLA-A24 occurs in both a shed and a secreted form at levels higher than all other alleles examined.

Serologic and biochemical analysis of HLA B15 and B5 complexes

Charge heterogeneity of HLA-B15 and HLA-B5 complexes was analyzed by one-dimensional isoelectric focusing (1D-IEF). Frozen peripheral blood lymphocytes were metabolically labeled with 35S methionine. The class I antigens were immunoprecipitated with monoclonal antibody 4E, which detects a determinant shared by HLA-B locus and Aw19-complex antigens. The desialated 1D-IEF banding patterns were correlated to microcytotoxicity data of a panel of donors from a variety of racial groups. Serologic analysis indicated the presence of specific variants: Te76, Te78, and Te79. 1D-IEF analysis clearly showed polymorphism in the B15 and B5 complexes. The Bw62 associated variant Te79 exhibited bands distinct from Bw62. One Bw62 typed donor produced a band that was different from other Bw62 typed cells. A migration pattern difference was discovered between blacks and Caucasians that were typed Bw57. Investigated antigens included HLA-B35, w46, 51, w52, w53, w57, w58, w62, w63, w70, Te76, Te78, and Te79.

Cytotoxic T lymphocyte recognition of a xenogeneic major histocompatibility complex antigen expressed in transgenic mice

Introduction of a porcine major histocompatability complex (MHC) class I gene (PD1) into the genome of a C57BL/10 (B10) mouse has been shown to lead to cell surface expression of the porcine MHC antigen, SLAPD1 in a transgenic mouse. The PD1 product expressed on spleen cells from the transgenic mice stimulated B10 spleen cells in a mixed lymphocyte culture to generate PD1-specific cytotoxic T lymphocytes (CTL). The CTL were PD1 specific since they lysed transgenic splenic blast cells and PD1-transfected L cells, but not B10 blasts or control L cells. The CTL were L3T4-, Lyt-2+ and their activity was partially inhibited by either anti-Lyt-2 antibody or by anti-swine MHC alloantibodies. The repertoire of responding B10 anti-transgenic CTL was assessed by examining their cross-reactivity on a series of murine allogeneic targets. The B10 anti-transgenic CTL showed some cross-reactivity on conventional allogeneic targets, but reacted strongly on a series of mutant H-2Kbm blast cells. In addition, B10 anti-B6.cH-2bm6 CTL cross-reacted extensively on the transgenic target cells. These results demonstrated that normal B10 CTL possess a repertoire specific for the products of the xenogeneic class I gene PD1, that this repertoire is cross-reactive with the conventional alloreactive CTL repertoire, and that there exists an unanticipated relationship between PD1-specific CTL and CTL specific for Kb mutant determinants.

Site-directed mutagenesis of an HLA-A3 gene identifies amino acid 152 as crucial for major-histocompatibility-complex-restricted and alloreactive cytotoxic-T-lymphocyte recognition

Major histocompatibility complex-restricted and alloreactive cytotoxic T lymphocytes (CTL) can discriminate between the HLA-A3.1 and HLA-A3.2 antigens. HLA-A3.1 and the rare variant HLA-A3.2 have been shown to differ by two amino acids in the alpha 2 domain at positions 152 (A3.1, glutamic acid; A3.2, valine) and 156 (A3.1, leucine; A3.2, glutamine). To determine the structural basis for the ability of CTL to differentiate A3.1 from A3.2, two site-directed mutants of the HLA-A3.2 gene were produced, 152A3.1-156A3.2 and 152A3.2-156A3.1, that have the indicated codons for positions 152 and 156. These mutated HLA-A3 genes, as well as the nonmutated HLA-A3.1 and HLA-A3.2 genes, were then transfected into the murine cell line P815-HTR and used as targets for human CTL. Influenza virus-specific HLA-A3.1-restricted CTL lysed virus-infected P815 cells transformed with the HLA-A3.1 and 152A3.1-156A3.2 genes, but not P815 cells transformed with the HLA-A3.2 and 152A3.2-156A3.1 genes. HLA-A3.2-allospecific CTL lysed the P815 cells transformed with the HLA-A3.2 and 152A3.2-156A3.1 genes but did not lyse P815 cells transformed with the HLA-A3.1 or 152A3.1-156A3.2 genes. Thus, a single amino acid change at position 152, substituting valine for glutamic acid and thereby introducing a charge difference, produces major structural changes in the epitopes recognized by major histocompatibility complex-restricted and alloreactive CTL.

Unglycosylated Mtaa expresses an Mtab-like determinant

Antigenic polymorphism of the class I-like maternally transmitted antigen (Mta) is controlled by a maternally transmitted factor (Mtf) thought to reside in mitochondria. However, the mechanisms by which Mtf generates antigenic polymorphism are not known. To address this issue, we investigated a possible role of posttranslational oligosaccharide addition in the formation of Mta determinants. We examined the expression of Mta on cytotoxic T lymphocyte (CTL) target cells cultured in tunicamycin (TM), a known inhibitor of asparagine(N)-linked glycosylation. Of 18 Mtab-specific CTL lines, 8 lysed TM-treated Mtaa targets. Furthermore, a subclone of one of these eight lines, 17D5.G2, lysed TM-treated targets from all Mtaa strains tested, regardless of H-2K/D haplotype. On the other hand, this CTL clone did not lyse TM-treated target cells from the Mta null, but H-2 expressing strain B10.CAS2. Therefore expression of this Mtab-like determinant is concordant with the expression of Mtaa and seems unlikely to represent a cross-reactive H-2K/D epitope. Our data suggest that an Mtab-like determinant is expressed on unglycosylated Mtaa molecules. Thus, N-linked oligosaccharides probably prevent the expression of an Mtab-like determinant on the Mtaa molecule. We discuss how Mtf may contribute to Mta polymorphism through glycosylation.

DNA sequence of HLA-A11: remarkable homology with HLA-A3 allows identification of residues involved in epitopes recognized by antibodies and T cells

The human class I alleles HLA-A11 and HLA-A3 have a well-documented history of serological cross-reactivity. This cross-reactivity suggests that they are closely related, a suggestion which is supported by the fact that the HLA-A11 and HLA-A3 genes are distinguished from all other A-locus genes by a restriction fragment length polymorphism observed in Bam HI digests. To examine the extent of sequence homology between HLA-A11 and HLA-A3, we have cloned the HLA-A11 gene and sequenced the coding regions (exons). The results reveal that HLA-A11 and HLA-A3 display the highest degree of homology reported for any pair of serologically defined class I alleles. Only nine base differences resulting in six amino acid differences were observed in exons 2-8. One of the amino acid substitutions is in the alpha 1 domain and the other five are in the alpha 2 domain. comparison of this sequence with that of other human class I molecules implicates Gln62 as a critical residue involved in HLA-A11 - HLA-A3 serological cross-reactivity. In addition, the amino acid sequence allowed us to successfully predict cross-reactive recognition of HLA-A11 by cytotoxic T lymphocytes specific for a rare subtype of HLA-A3, HLA-A3.2. This result provides further support for the importance of the alpha 2 domain residues 152 and 156 in forming determinants on class I molecules that are recognized by cytotoxic T lymphocytes.

Interaction between Kb and Q4 gene sequences generates the Kbm6 mutation

Genetic interaction as a mechanism for the generation of mutations is suggested by recurrent, multiple nucleotide substitutions that are identical to nucleotide sequences elsewhere in the genome. We have sequenced the mutant K gene from the bm6 mouse, which is one of a series of eight closely related, yet independently occurring mutants known collectively as the "bg series." Two changes from the Kb gene are found, positioned 15 nucleotides apart: an A-to-T change and a T-to-C change in the codons corresponding to amino acids 116 and 121, resulting in Tyr-to-Phe and Cys-to-Arg substitutions, respectively. Hybridization analysis with an oligonucleotide specific for the altered Kbm6 sequence identifies one donor gene, Q4, located in the Qa region of the H-2 complex. The two altered nucleotides that differentiate Kbm6 and Kb are present in Q4 in a region where Kb and Q4 are otherwise identical for 95 nucleotides, delineating the maximum genetic transfer between the two genes. Because the Kbm6 mutation arose in an homozygous mouse these data indicate that the Q4 gene contains the only donor sequence and demonstrates that Q-region gene sequences can interact with the Kb gene to generate variant K molecules.

Negative regulation of the major histocompatibility class I gene in undifferentiated embryonal carcinoma cells

Murine embryonal carcinoma F9 cells, which do not express appreciable levels of major histocompatibility complex (MHC) class I mRNA, start to express the mRNA and proteins upon differentiation induced by retinoic acid (RA). To investigate the molecular mechanism of this regulation, we examined in F9 cells transient expression of the chloramphenicol acetyltransferase (CAT) gene directed by the 5' flanking region of a MHC class I gene, H-2Ld. The native 1.4-kilobase H-2Ld 5' upstream region gave very low CAT activity in undifferentiated F9 cells. Deletion between positions -210 and -135 relative to the cap site resulted in a 4- to 5-fold increase in CAT activity as compared with constructs containing the region. However, all of these constructs, regardless of the deletion, expressed comparable CAT activity in differentiated F9 cells. These data suggest the presence of a negative cis-acting element that is under developmental control. Further analysis revealed that the sequence conferring the negative regulation resides between positions -195 and -161. This region, highly conserved among the MHC class I genes, is found to be capable of increasing CAT activity in NIH 3T3 cells that express the class I genes constitutively. Further, this regulatory sequence, when connected to the simian virus 40 promoter, produced repressive and enhancing effects in F9 and NIH 3T3 cells, respectively. Based on these results, we suggest that the expression of MHC class I genes during development involves switching from negative to positive regulation dictated by the class I regulatory element located between positions -195 and -161.

Structural analyses define an additional H-2D region class I antigen that is expressed by a variant mouse strain

The genetic complexity of the H-2D region includes haplotype disparities in apparent gene and product number. To probe the genetic basis of this complexity, the products of two independently derived mouse strains (STU and B10.SAA48) that express Dw3 antigens were compared. Serologic, fluorometric and peptide map comparisons were made using monoclonal antibodies. Although both STU and B10.SAA48 mice were found to express indistinguishable Dw3 molecules, only B10.SAA48 mice were found to express an additional antigen designated Lw3. Several lines of evidence are presented that suggest the gene encoding Lw3 maps to the D region. Furthermore peptide map comparisons of Dw3 with Lw3 molecules implied that they are products of separate genes; but Dw3 and Lw3 molecules were found to be more homologous to each other than Dd and Ld molecules are to each other. Inter-haplotype comparisons of Dw3 and Lw3 molecules with other D region molecules showed no striking homologies to Dd, Ld or eight other molecules compared. However, both Dw3 and Lw3 molecules were found to be unexpectedly homologous to the Ddx and Dw25 molecules, thus defining another family of structurally related D region antigens. This so called Dw3-family was found to be quite distinct from the previously defined Ld-family of molecules, since no joint members were found. The results of these studies of Dw3 encoded antigens are discussed as evidence for intra-D region recombination or mutation.

The preparation and characterization of anti-peptide heteroantisera recognizing subregions of the intracytoplasmic domain of class I H-2 antigens

Peptides corresponding to each of the three intracytoplasmic exons (i.e. exons 6, 7 and 8) of the murine class I H-2Kb gene were synthesized, coupled to bovine serum albumin and used as immunogens in rabbits. In each case the antisera were found to react with the immunizing peptide coupled to a heterologous carrier, and recognized class I heavy chains electrophoretically transferred from SDS-polyacrylamide gels to nitrocellulose. Immunoprecipitation of class I antigens from Nonidet P-40 (NP-40) solubilized EL-4 (H-2b) tumour cells by each of the antisera reflected their ability to recognize the corresponding determinants in non-denatured class I molecules. The same sera were also able to immunoprecipitate class I molecules from NP-40 solubilized RDM-4 (H-2k) and P815 (H-2d) tumour cells, indicating the cross-reactive nature of these antisera for different class I alleles. In addition to reacting with the class I heavy chain in its conventional form as a dimer with beta 2-microglobulin, the antiserum specific for the exon 8 peptide was able to react with "free" (i.e. non-beta 2-microglobulin-associated) class I heavy chains. Thus, a unique set of immunological reagents has been prepared which offer a new approach to studying the structural and functional features of the cytoplasmic domain of class I H-2 antigens.

Structural and functional dissection of an MHC class I antigen-binding adenovirus glycoprotein

The early transmembrane glycoprotein E19 of adenovirus-2 binds to class I antigens of the major histocompatibility complex (MHC). The association is initiated in the endoplasmic reticulum of infected cells and abrogates the intracellular transport of the class I molecules. To examine which parts of the E19 molecule are responsible for the association with the class I antigens and which parts confine the protein to the endoplasmic reticulum we have constructed a series of mutated E19 genes, which have been expressed in an improved mammalian expression vector. By various manipulations the membrane anchoring and the cytoplasmic domains were removed from the protein. The biosynthesis of the mutant protein was examined. All mutant proteins were secreted from the cells suggesting that the transmembrane and/or cytoplasmic portions of the E19 molecule are responsible for its confinement to the endoplasmic reticulum. The ability to associate with class I antigens was retained by the lumenal domain of the E19 protein.

A structural analysis of the carbohydrate side chains on class I and class II histocompatibility antigens of the swine facilitated by heteroantisera specific for the denatured polypeptides

Rabbit heteroantisera specific for the denatured glycoprotein subunits of swine class I and class II major histocompatibility complex (MHC) antigens have been prepared and utilized to monitor changes in the mobilities of these polypeptides on SDS-polyacrylamide gel electrophoresis subsequent to various deglycosylation procedures. This information, in combination with lectin reactivity patterns for the glycoproteins bound to nitrocellulose, has made it possible to define specific structural features of the MHC antigen-associated carbohydrate side chains. Both the class I heavy (alpha) chain and the class II light (beta) chain bear a single, N-linked, complex-type oligosaccharide which reacts with lentil lectin (LcH), but not concanavalin A (Con A); a reactivity pattern suggesting the possibility of a special triantennary structure. In contrast, the class II heavy (alpha) chains appear to possess two carbohydrate units, one an N-linked, LcH-reactive, complex-type side chain, and the other, an N-linked, Con A-reactive, high-mannose-type of oligosaccharide. The data suggest considerable homology between the swine and human MHC antigens with respect to the structure of their carbohydrate side chains. The analysis also serves to illustrate how antibodies specific for the denatured polypeptide backbone of individual glycoproteins, along with lectin reactivity patterns, can be used to extract structural information about the attached carbohydrate moieties using minimal amounts of partially purified glycoproteins.

Alternative protein products with different carboxyl termini from a single class I gene, H-2Kb

In addition to mRNA encoding the canonical form of the murine class I antigen H-2Kb (348 amino acid residues), mRNA that would encode a shortened form of H-2Kb (missing 9 amino acids from the C-terminus) has been identified in C57BL/6 spleen cells by RNase-protection studies. The alternative transcripts of H-2Kb arise through the use of different AG acceptor splice sites for exon VIII. The existence of a shortened H-2Kb protein was demonstrated by sequential immunoprecipitation. Lysates of spleen cells that had been labeled with [35S]methionine and [3H]histidine were precleared with rabbit anti-peptide serum reactive with the C-terminus of the canonical H-2Kb. The shortened form of H-2Kb was immunoprecipitated from this lysate with H-2Kb alloantiserum. Both forms of H-2Kb were isolated by NaDodSO4/PAGE. Tryptic peptide mapping confirmed that these molecules differed only at their C-terminus. The shortened form of H-2Kb is also found in a B-cell line (R8) but not in three cloned T-cell lines or in a T-cell lymphoma (EL4), suggesting that regulatory events are involved in producing the two forms of H-2Kb. Putative lariat branch points involved in these alternative splicing events for the 3' coding region of H-2 class I pre-mRNAs are proposed.

Mitotic recombination in germ cells generated two major histocompatibility complex mutant genes shown to be identical by RNA sequence analysis: Kbm9 and Kbm6

RNA sequencing represents a major procedural simplification for nucleotide sequence analysis of a transcribed gene. Using newly adapted mRNA and cDNA sequencing procedures, we have sequenced 855 nucleotides of Kbm9 mRNA, corresponding to the codons for the aminoterminal 285 amino acids. The inferred DNA sequence of the Kbm9 gene differs from the parental Kb sequence by single nucleotide alterations in each of codons 116 and 121, resulting in Tyr----Phe and Cys----Arg substitutions, respectively. The Kbm9 sequence is identical to that of another independently arising MHC mutant gene, Kbm6. As both the Kbm9 and Kbm6 genes were generated by recombination between the Kb and Q4 genes, our data indicate that the identical genetic interactions have occurred at least twice. The relatively large extent of identity between Q4 and Kb may be responsible for frequent recombination between the two genes. The parents of the original bm9 mutant mice had five identical mutant offspring, which can be explained by mitotic recombination in the germ cells, producing gonadal mosaicism in the C57BL/6 mother. Thus, mitotic recombination, and not meiotic recombination, appears to be responsible for the formation of at least some of the Kb mutants. Such a mechanism probably plays a major role in the generation of diversity in the major histocompatibility complex.

Viral-restricted cytolytic T lymphocyte recognition of hybrid human-murine class I histocompatibility antigens

Hybrid human-murine major histocompatibility antigens have been constructed and expressed on the surface of both human RD and murine L cell lines after DNA mediated gene transfer. These antigens linked the polymorphic domains (alpha 1 and alpha 2) of H-2Kb and the carboxy-terminal domains (alpha 3, transmembrane, and intracellular) of HLA-A2. Previously we demonstrated that these antigens were serologically intact and were recognized by allospecific cytolytic T lymphocytes. However, the cell lines expressing the hybrid antigen were less well lysed than the native H-2Kb expressing cell lines. In this study, we extend these observations and demonstrate that virally restricted cytolytic T lymphocytes specific for vesicular stomatitis virus and for Sendai virus can recognize cell lines expressing the hybrid antigen, whether expressed on murine (L cell) or human (RD cell) lines. Furthermore, the data show a profound influence by the carboxy-terminal domains upon the polymorphic T-cell restricting epitopes.

Interaction between Kb and Q4 gene sequences generates the Kbm6 mutation

Genetic interaction as a mechanism for the generation of mutations is suggested by recurrent, multiple nucleotide substitutions that are identical to nucleotide sequences elsewhere in the genome. We have sequenced the mutant K gene from the bm6 mouse, which is one of a series of eight closely related, yet independently occurring mutants known collectively as the "bg series." Two changes from the Kb gene are found, positioned 15 nucleotides apart: an A-to-T change and a T-to-C change in the codons corresponding to amino acids 116 and 121, resulting in Tyr-to-Phe and Cys-to-Arg substitutions, respectively. Hybridization analysis with an oligonucleotide specific for the altered Kbm6 sequence identifies one donor gene, Q4, located in the Qa region of the H-2 complex. The two altered nucleotides that differentiate Kbm6 and Kb are present in Q4 in a region where Kb and Q4 are otherwise identical for 95 nucleotides, delineating the maximum genetic transfer between the two genes. Because the Kbm6 mutation arose in an homozygous mouse these data indicate that the Q4 gene contains the only donor sequence and demonstrates that Q-region gene sequences can interact with the Kb gene to generate variant K molecules.

Unusual features of the T-cell receptor C domains are revealed by structural comparisons with other members of the immunoglobulin superfamily

The amino acid sequences of the domains of human, mouse and rabbit T-cell antigen receptors have been aligned with those of immunoglobulin domains of known three-dimensional structure. Computerized secondary structure predictions have been performed on the sequences and putative models of the domains have been constructed. The receptor V alpha and V beta domains are closely related to immunoglobulin VH domains. The receptor C alpha domain shows some major divergences from immunoglobulin C domains and the C beta domain displays an unusual feature. The implications for T-cell receptor function are discussed.