Transcriptional regulation of the MHC II gene DRA in untransformed human thyrocytes.

MHC class II molecules are heterodimeric, polymorphic transmembrane glycoproteins physiologically expressed on cells of the immune system and pathologically expressed on the affected target cells of autoimmunity. Their function is to present processed peptides to antigen-specific CD4(+) T cells. To understand the molecular mechanism of the regulation of class II genes in autoimmune target cell thyrocytes, we investigated the transcriptional regulation of DRA on untransformed, differentiated human thyroid cells following IFN-gamma stimulation, which is potentially relevant to the inappropriate class II expression found in Graves' disease. Data from this study show that IFN-gamma enhances a promoter Y box binding protein and induces an X box binding protein in untransformed thyrocytes, but not in SV-40-transfected thyrocytes. Initial characterization of the proteins has indicated that the Y box binding protein is approximately 132 kDa in size while the X box binding protein binds to the X2 region and is approximately 116 kDa. The X box binding protein may correspond to poly(ADP-ribose) polymerase, a recently described component of the X2 box binding protein, X2BP. In addition, the signal transducer and activator of transcription 1alpha protein (STAT1alpha) is also induced by IFN-gamma in these cells. These results further suggest that there are differences in class II gene regulation between differentiated cells and transformed cell lines.

HLA-F is a predominantly empty, intracellular, TAP-associated MHC class Ib protein with a restricted expression pattern.

HLA-F is currently the most enigmatic of the human MHC-encoded class Ib genes. We have investigated the expression of HLA-F using a specific Ab raised against a synthetic peptide corresponding to amino acids 61-84 in the alpha1 domain of the predicted HLA-F protein. HLA-F is expressed as a beta2-microglobulin-associated, 42-kDa protein that shows a restricted tissue distribution. To date, we have detected this product only in peripheral blood B cells, B cell lines, and tissues containing B cells, in particular adult tonsil and fetal liver, a major site of B cell development. Thermostability assays suggest that HLA-F is expressed as an empty heterodimer devoid of peptide. Consistent with this, studies using endoglycosidase-H and cell surface immunoprecipitations also indicate that the overwhelming majority of HLA-F contains an immature oligosaccharide component and is expressed inside the cell. We have found that IFN-gamma treatment induces expression of HLA-F mRNA and HLA-F protein, but that this does not result in concomitant cell surface expression. HLA-F associates with at least two components of the conventional class I assembly pathway, calreticulin and TAP. The unusual characteristics of the predicted peptide-binding groove together with the predominantly intracellular localization raise the possibility that HLA-F may be capable of binding only a restricted set of peptides.

Osteopontin is an autoantigen of the somatostatin cells in human islets: identification by screening random peptide libraries with sera of patients with insulin-dependent diabetes mellitus.

Random peptide libraries (RPLs) screening with IDDM sera has identified 5 disease-specific 'mimotopes' displayed on phage (phagotopes). We characterised one phagotope (CH1p), by raising a rabbit antibody against the peptide insert on phage, which was employed in immunohistochemistry, Western blotting and cDNA libraries screening. The CH1p mimotope was detected in somatostatin cells of human islets and experimentally raised anti-osteopontin antibodies or human sera positive for the phagotope, detected a similar subpopulation of islet cells. The screening of cDNA library identified a clone corresponding to human osteopontin. In summary, RPLs proved to be successful in the identification of a novel islet-related autoantigen (osteopontin), whose significance in disease remains to be established.

HLA-DMB expression by thyrocytes: indication of the antigen-processing and possible presenting capability of thyroid cells.

Expression of HLA class II molecules on thyrocytes is a characteristic feature of autoimmune thyroid disease and may lead the thyroid cells to present autoantigens to CD4+ T lymphocytes. Since HLA-DM is a critical molecule in class II-restricted antigen processing and presentation, we assessed the expression of HLA-DMB, -invariant chain (Ii), class II transactivator (CIITA) and DRA in an untransformed, pure thyrocyte strain HTV-59A. Here we report that both HLA-DMB mRNA and the protein are expressed in thyrocytes and that CIITA expression is enhanced by interferon-gamma (IFN-gamma) treatment and occurs before DMB, Ii and DRA up-regulation, suggesting CIITA expression is a requirement for antigen processing in thyrocytes. These results indicate that thyrocytes are capable of antigen processing and possibly antigen presentation to T cells.

The human amnion is a site of MHC class Ib expression: evidence for the expression of HLA-E and HLA-G.

The expression of HLA class I Ag by term human amnion epithelial cells was investigated. In immunostaining and FACS analysis, mAb to monomorphic class I Ag reacted extensively with amnion cells, whereas polymorphic mAb reactivity was more limited and variable. Further studies were conducted on amnion cell preparations containing negligible contaminants. Northern analysis with use of locus-specific probes demonstrated that amnion expresses two class Ib genes, HLA-E and HLA-G. Radio-immunoprecipitation with use of monomorphic mAb identified two fully glycosylated cell surface class I H chains of 44 and 41 kDa; polymorphic mAbs failed to immunoprecipitate the 41-kDa product, although 44-kDa products, typical of class Ia Ag, were identified in some preparations. Class I H chains were isolated from amnion by affinity chromatography. Microsequencing revealed that the first nine residues of the N-terminus of the 41-kDa product aligned perfectly only with HLA-E. Overall, amnion at term appears to express class Ib Ag with limited class Ia Ag. HLA-G is therefore expressed in two extrafetal epithelia: amnion and trophoblast. Identification of the class Ib protein HLA-E in amnion epithelium may have implications for preterm labor that can be associated with infection of the placental membranes.

A gene in the HLA class I region contributes to susceptibility to IDDM in the Finnish population. Childhood Diabetes in Finland (DiMe) Study Group.

In Finland the haplotype A2, Cw1, B56, DR4, DQ8 is the third most common haplotype in insulin-dependent diabetic (IDDM) patients and has the highest haplotype-specific absolute risk for IDDM. Cw1, B56, DR4, DQ8 haplotypes containing HLA-A alleles other than A2 are infrequent in the population and are not associated with IDDM. Comparison of the A2 and non-A2 haplotypes at the DNA level showed that they were identical at HLA-B, -DR, and -DQ loci. Evidence that class I alleles confer susceptibility to IDDM was obtained from the two HLA-C, -B, -DR and -DQ haplotypes most frequently found in IDDM patients in Finland. A24, A3 and A2 on the Cw3, B62, DR4, DQ8 haplotype, and A28, A2 and A1 on the Cw7, B8, DR3, DQ2 were all found to be associated with IDDM. In Finland these seven haplotypes, including A2, Cw1, B56, DR4, DQ8, account for 33% of diabetic haplotypes and 10.3% of non-diabetic haplotypes (p < 0.00001). The contribution of the class I region to IDDM susceptibility was also apparent in those IDDM patients lacking the disease-predisposing class II alleles. Significantly more non-DR3/non-DR4 IDDM patients (47 of 55) possessed two of the IDDM-associated HLA-A alleles compared to non-DR3/non-DR4 control subjects (40 of 58; p = 0.038). Moreover, IDDM patients confirmed by oligotyping as unable to form a 'diabetes-susceptibility' DQ heterodimer, tended to possess two diabetes-associated HLA-A alleles (12 of 13) compared to control subjects (12 of 20; p = 0.056).

Sequence-specific oligonucleotide typing in Shona patients with rheumatoid arthritis and healthy controls from Zimbabwe.

Seventy-two patients with rheumatoid arthritis (RA) and 82 controls have been typed with the XI Histocompatibility Workshop DRB1 and DQB1 sequence-specific oligonucleotide probes. The increase of DRB1*04 corresponds to an increase of the serologically defined DR4, previously found in a small group of Zimbabwean RA patients and we now show that this increase is due to the subtype DRB1*0405 in association with DQB1*0302. In addition there is a clearcut increase of DRB1*1001 equivalent to the serologically defined DR10. There was no increase amongst RA patients of DRB1*0102 which was the predominant DR1 sub-type amongst controls. In the course of our investigation, we observed a DRB1*04 variant which corresponds to DRB1*0412, newly defined in the XIth Histocompatibility Workshop.

HLA-DP polymorphism in Sudanese controls and patients with insulin-dependent diabetes mellitus.

Human leukocyte antigen (HLA) genes are candidates for susceptibility to insulin-dependent diabetes mellitus (IDDM). The association of IDDM with particular DR and DQ alleles has been reported in all populations studied, but its association with HLA-DP alleles has been controversial. To address this question we analyzed 19 DPB1 and 2 DPA1 alleles and their associations in well-characterized Sudanese (an admixture of Arab and Black) IDDM patients (n = 71) and ethnically matched controls (n = 86) using polymerase chain reaction-sequence specific oligonucleotide (PCR-SSO) typing. There were no significant differences between the patient and control groups in the DPB1 frequencies. DPB1*0201, *0401 and DPA1*01 were the most frequent alleles in both IDDM patients and control subjects. Significant positive and negative associations between DPB1 and DPA1 alleles were detected in both groups. A novel DPB1 allele included in DPB1*1701 was identified.

Evidence for the expression of non-HLA-A,-B,-C class I genes in the human fetal liver.

HLA class I Ag expression was investigated in the human fetal liver. By immunostaining, mAb to monomorphic class I Ag showed widespread reactivity with both epithelial and hemopoietic cells. By contrast, mAb to polymorphic determinants showed more restricted reactivity that was confined to a proportion of hemopoietic cells: the hepatic epithelium was essentially unreactive. This suggested that the developing liver might express nonclassical HLA class I. Class I Ag were examined in membrane and cytosol fractions of mid-trimester fetal liver. Because of its broad reactivity with HLA-A,-B, mAb Q1/28 was selected to identify classical class I Ag in these studies. Immunoprecipitations were carried out against radiolabeled glycoprotein extracts of fetal liver membranes. W6/32 detected a 40-kDa product characteristic of nonclassical class I proteins, as well as a 43-kDa product, in lysates immunodepleted with Q1/28. By immunoblotting, an anti-H chain antiserum (HC) identified a Q1/28- 40-kDa component and a 43-kDa Q1/28+ component in fetal liver membrane glycoproteins. The fetal liver cytosol fraction was found to contain a 42- to 43-kDa product by W6/32-chromatography. This component partitioned to the aqueous phase upon condensation in TritonX-114 detergent and by immunoblotting was reactive with monomorphic mAb HC10 but not with Q1/28. Total RNA and polymerase chain reaction amplified class I transcripts of fetal liver were probed using oligonucleotides specific for HLA-E, -F, and -G. HLA-F, was readily detected in total RNAs by Northern analysis. HLA-E, HLA-F, and HLA-G were all detected in fetal liver by polymerase chain reaction. Differential expression of these genes may occur between the first and second trimester of liver development. Overall therefore, the human fetal liver expresses multiple class I protein products and contains transcripts for non-classical class I genes; particularly HLA-F.

Molecular characterization of serologic recognition sites in the human HLA-A2 molecule.

The localization of the amino acid residues involved in the serologic specificity of the HLA-A2 molecule has been investigated using a combination of site-directed mutagenesis, DNA-mediated gene transfer, indirect immunofluorescence and flow cytometry techniques. Synthetic oligonucleotides were designed to introduce individual and combined amino acid substitutions in both the alpha 1 (positions 9, 43, and the highly polymorphic cluster of residues from aa 62 to 83) and alpha 2 (positions 107, 152, and 156) domains to investigate the effect of the specific mutation on the recognition of the molecule at the surface of transfected human and mouse cell lines by a panel of mAb that recognize monomorphic or polymorphic determinants in MHC class I molecules. At least three non-overlapping serologic epitopes were identified. Mutations in the highly polymorphic region at aa 62 to 66 completely eliminated binding of mAb MA2.1 (A2/B17 cross-reactive). Mutation at position 107 resulted in complete loss of binding of the A2/Aw69-specific mAb PA2.1 and MA2.2 and partial loss of mAb BB7.2 binding. The recognition by other allotypic mAbs was not affected by these mutations and they therefore represent at least a third serologic epitope. Mutations at positions 152 and 156, known to be important for T cell recognition, did not affect serologic recognition. Introduction of residues of HLA-B7 origin in the polymorphic segment spanning aa 70 to 80 created a molecule carrying the -Bw6 supertypic determinant as demonstrated by mAb SFR8-B6 binding.

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.

Amino acid sequences in the alpha 1 domain and not glycosylation are important in HLA-A2/beta 2-microglobulin association and cell surface expression.

The role of the single carbohydrate moiety present on the HLA-A2 molecule was studied by introducing several amino acid substitutions (by site-directed mutagenesis of the HLA-A2 gene) in the consensus glycosylation sequence Asn-X-Ser. Two different amino acid substitutions of the asparagine residue at position 86 (glutamine and aspartic acid) resulted in the synthesis of ca. 39,000-molecular-weight nonglycosylated heavy chains that were detected in the cytoplasm but not on the surface of mouse L-cell transfectants. However, a low level of surface expression was detected following transfection of human (rhabdomyosarcoma) cells or mouse L cells containing human beta 2-microglobulin. The defect in surface expression was not due to the absence of the glycan moiety, since the substitution of a glycine for a serine at amino acid 88 did not have the same drastic effect in the presence of human beta 2-microglobulin. These and other data suggest that the asparagine residue may play a critical role in the conformation of the HLA heavy chain and its interaction with beta 2-microglobulin. Immunofluorescence microscopy following permeabilization of the transfectants demonstrated that the unglycosylated HLA heavy chains are sequestered in an unidentified cellular compartment that is different from the Golgi structure.

The expression of murine Qa region gene product(s) in L cell transformants.

The cosmid H3.5, containing genes mapping to the murine H-2 Qa region, was used to transfect L cells by the calcium phosphate co-precipitation method. The resultant transfected cells expressed a Qa-like determinant as detected by an immune serum raised against the transfectant cells and Qa specific monoclonal antibodies. Two-dimensional gel analysis revealed the expression of a class I-like heavy chain with a similar molecular mass to the Qa2 antigens of the positive strain B10 and B10.A but with a different isoelectric point. The cosmid H3.5 spans 40 kb of DNA and contains at least one complete Qa region gene which encodes the Qa-like determinant detected in this study.

Species-restricted recognition of transfected HLA-A2 and HLA-B7 by human CTL clones.

Human cytolytic T lymphocytes (CTL) clones and HLA-A2- and HLA-B7-transfected human, monkey, and mouse cell lines were used to investigate the basis for species-restricted antigen recognition. Most allospecific CTL clones obtained after stimulation with the human JY cell line (source of HLA-A2 and HLA-B7 genomic clones) recognized HLA antigens expressed in human and monkey cell lines but did not recognize HLA expressed in murine cells. By initially stimulating the responder cells with HLA-transfected mouse cells, two CTL clones were obtained that recognized HLA expressed in murine cells. Functional inhibition of these CTL clones with anti-class I monoclonal antibodies (MAb) indicated that clones reactive with HLA+ murine cells were of higher avidity than clones that did not recognize HLA+ murine target cells. MAb inhibition of accessory molecule interactions demonstrated that the LFA-1 and T8 surface molecules were involved in CTL-target cell interactions in all three species. In contrast, the LFA-2/CD2 molecule, previously shown to participate in a distinct activation pathway, was involved in the cytolysis of transfected human and monkey target cells, but not in the lysis of HLA+ murine cells. Thus transfection of HLA genes into different recipient species cell lines provides us with the ability to additionally delineate the functional requirements for allospecific CTL recognition and lysis.

Gene mapping and somatic cell hybrid analysis of the role of human lymphocyte function-associated antigen-3 (LFA-3) in CTL-target cell interactions.

LFA-3 is expressed on a wide variety of human cell lines, including those which have been used as recipients for gene transfer of human class I gene products, whereas a murine counterpart is either absent or significantly different such that the anti-LFA-3 monoclonal antibody (MAb) does not bind. By using a somatic cell genetic approach, we demonstrate that LFA-3 is not a major histocompatibility complex-encoded molecule, and that its gene locus maps to human chromosome 1. When LFA-3 and HLA-A2 are coexpressed on the mouse cell surface, anti-LFA-3 MAb interfered with specific recognition and lysis of these target cells by human CTL capable of lysing HLA-A2-expressing mouse transfectants. A significant contribution of the LFA-3 molecule to CTL reactivity was not observed, however, because the presence of LFA-3 did not restore recognition by CTL clones previously found incapable of lysing HLA-A2-expressing mouse transfectants, nor was it required by those human CTL that could lyse mouse cell transfectants. Thus, we have used genetic techniques to demonstrate that LFA-3 may serve a role in CTL-target cell interactions at the target cell level, but is not a molecule absolutely required for human allospecific CTL recognition of HLA antigens expressed on mouse cells. We suggest that LFA-3 may not participate directly in CTL function under normal circumstances, but delivers a more general inhibitory signal only when provoked by bound MAb.

Recognition of interspecies hybrid class I histocompatibility antigens by antigen-specific cytolytic T lymphocytes.

Two reciprocal interspecies hybrid class I histocompatibility genes have been constructed between genomic clones of human HLA-A2 and murine H-2Kb. The proteins encoded by these genes have been designated A21+2/Kb, where the polymorphic domains, alpha 1 and alpha 2, of HLA-A2 are linked to the carboxyl-terminal domains (alpha 3, transmembrane, and intracytoplasmic domains) of H-2Kb, and Kb1+2/A2, where the alpha 1 and alpha 2 domains of the H-2Kb antigen are linked to the carboxyl-terminal domains of HLA-A2. These genes have been transfected and expressed in recipient mouse L cells and human RD (rhabdomyosarcoma) cells. Both hybrid antigens were found to be serologically intact when tested with a panel of antigen-specific monoclonal antibodies. The monoclonal antibody W6/32, which recognizes a monomorphic determinant on all HLA-A, -B, and -C antigens, recognizes the alpha 1 and/or the alpha 2 domain, rather than the more conserved alpha 3 domain. Human cytolytic T lymphocytes (CTL) specific for the HLA-A2 antigen recognized the A2 and A21+2/Kb proteins only when expressed in human cells and not when expressed in mouse cells, even when surface antigen levels were 10-fold greater on the mouse cells than on the human cells. In contrast, a long-term, murine anti-H-2b CTL line not only lysed mouse L-cell lines that expressed the parental Kb and hybrid Kb1+2/A2 antigens but also lysed the Kb and Kb1+2/A2 human cell RD transformants as well. In both cases, the level of CTL recognition and lysis of the transformants that expressed the native antigen Kb was greater than of those transformants that expressed the hybrid antigen Kb1+2/A2. These data suggest that the carboxyl-terminal domains play some role in CTL allorecognition. The lack of human CTL recognition of HLA molecules expressed in mouse L cells, however, cannot be explained by the presence of a xenogeneic carboxyl terminus. Since murine CTL can recognize their target antigen when expressed on the surface of human cells, the possibility remains either that a ligand necessary for other molecular interactions of human CTL may be absent on mouse target cells or that murine and human CTL differ in affinity of binding to target antigens in the absence of accessory-molecule interactions.

Structural analysis of the functional sites of class I HLA antigens.

Considerable knowledge of the molecular organization of class I HLA antigens has been attained through extensive structural analysis of these proteins and their genes. Particularly, the nature and location of the polymorphic regions has been established, as well as the basic patterns of structural variability. This work has not unveiled the functionally relevant sites of the HLA molecules but has provided the basis to develop new strategies to do so. The molecular analysis of the determinants recognized by specific antibodies and cytolytic T lymphocytes is being approached through the biochemical characterization of mutants induced in vitro and population variants that are selected by their loss of specific serological or CTL allodeterminants. Other approaches include the immunological analysis of sera raised against synthetic peptides whose structure mimics highly variable segments of class I HLA molecules. These studies have already revealed the participation of several regions in specific allorecognition by antibodies or CTLs and their potential is becoming increasingly evident. A new and possibly powerful approach is currently being used for the dissection of functional sites. It makes use of the structural information derived from sequence analysis and involves expression of cloned HLA genes in transfected mouse or human cells in conjunction with site-directed mutagenesis techniques. Although some difficulties still lie ahead in developing a system suitable for functional assays, the possibility of tailoring HLA mutants and studying the modulation of their recognition determinants by predetermined structural alterations open new pathways to the molecular analysis of HLA function.

Molecular organization of the class I genes of human major histocompatibility complex.

In this brief review, our main emphasis has been on the analysis of the sequence diversity among various class I genes and their functional implications. The availability of complete nucleotide sequences of 7 different genes representing different loci allowed us to derive a consensus sequence. One mouse MHC Class I gene was included in these comparisons as a representative of H2 genes Evolutionary patterns can be seen on the basis of divergence of various genes from the derived consensus sequence. At least 1 human gene which has a promoter similar to that of H2 genes and which contains a single initiation codon following this promoter (unlike all other human genes and like all the H2 genes) has been identified. Both variable and homology regions can be identified in the entire length of the gene. While exons show relatively strong conservation of sequences, the introns have many variable regions, introns 6 and 7 being the most heterogeneous. Stretches of conserved nucleotide sequences are noticed at the 3' regions of most introns. Estimation of total number of class I genes is presented on the basis of cloning experiments, and the abundance of 1 particular pseudogene is discussed.