Search

In type 1 diabetes, insulin-producing beta cells in the islets of the pancreas are destroyed by autoreactive T cells. Rotavirus (RV) has been implicated in the pathogenesis of type 1 diabetes. Peptides in VP7, a major immunogenic protein of RV, have high sequence similarity to T cell epitope peptides in the islet autoantigens tyrosine phosphatase-like insulinoma Ag 2 (IA2) and glutamic acid decarboxylase 65 (GAD65). We aimed to educe evidence for the hypothesis that molecular mimicry with RV promotes autoimmunity to islet autoantigens. Peptides in RV and their sequence-similar counterparts in IA2 and GAD65 were assayed for binding to HLA molecules associated with type 1 diabetes and for the ability to elicit T cell proliferative responses in HLA-typed individuals. T cells expanded or cloned to epitopes in IA2 or RV were then tested for cross-reactivity with these epitopes. Peptides in RV-VP7, similar to T cell epitopes in IA2 and GAD65, bound strongly to HLA-DRB1*04 molecules that confer susceptibility to type 1 diabetes and were also T cell epitopes in humans at risk for type 1 diabetes. The proliferative responses of T cells to the similar peptides in RV and islet autoantigens were significantly correlated. T cells expanded to the IA2 epitope could be restimulated to express IFN-gamma by the similar peptide in RV-VP7, and T cell clones generated to this RV-VP7 peptide cross-reacted with the IA2 epitope. Our findings are consistent with the hypothesis that molecular mimicry with RV could promote autoimmunity to islet Ags.

Plasticity of TCR interactions during CD4(+) T cell activation by an MHC-peptide complex accommodates variation in the peptide or MHC contact sites in which recognition of an altered ligand by the T cell can modify the T cell response. To explore the contribution of this form of TCR cross-recognition in the context of T cell selection on disease-associated HLA molecules, we have analyzed the relationship between TCR recognition of the DRB1*0401- and DRB1*0404-encoded HLA class II molecules associated with rheumatoid arthritis. Thymic reaggregation cultures demonstrated that CD4(+) T cells selected on either DRB1*0401 or DRB1*0404 could be subsequently activated by the other MHC molecule. Using HLA tetramer technology we identify hemagglutinin residue 307-319-specific T cells restricted by DRB1*0401, but activated by hemagglutinin residues 307-319, in the context of DRB1*0404. One such clone exhibits an altered cytokine profile upon activation with the alternative MHC ligand. This altered phenotype persists when both class II molecules are present. These findings directly demonstrate that T cells selected on an MHC class II molecule carry the potential for activation on altered self ligands when encountering Ags presented on a related class II molecule. In individuals heterozygous for these alleles the possibility of TCR cross-recognition could lead to an aberrant immune response.

T cell responses to Ags involve recognition of selected peptide epitopes contained within the antigenic protein. In this report, we describe a new approach for direct identification of CD4+ T cell epitopes of complex Ags that uses human class II tetramers to identify reactive cells. With a panel of 60 overlapping peptides covering the entire sequence of the VP16 protein, a major Ag for HSV-2, we generated a panel of class II MHC tetramers loaded with peptide pools that were used to stain peripheral lymphocytes of an HSV-2 infected individual. With this approach, we identified four new DRA1*0101/DRB1*0401- and two DRA1*0101/DRB1*0404-restricted, VP16-specific epitopes. By using tetramers to sort individual cells, we easily obtained a large number of clones specific to these epitopes. Although DRA1*0101/DRB1*0401 and DRA1*0101/DRB1*0404 are structurally very similar, nonoverlapping VP16 epitopes were identified, illustrating high selectivity of individual allele polymorphisms within common MHC variants. This rapid approach to detecting CD4+ T cell epitopes from complex Ags can be applied to any known Ag that gives a T cell response.

The polyclonal nature of T cells expanding in an ongoing immune response results in a range of disparate affinities and activation potential. Recently developed human class II tetramers provide a means to analyze this diversity by direct characterization of the trimolecular TCR-peptide-MHC interaction in live cells. Two HSV-2 VP16(369-379)-specific, DQA1*0102/DQB1*0602 (DQ0602)-restricted T cell clones were compared by means of T cell proliferation assay and HLA-DQ0602 tetramer staining. These two clones were obtained from the same subject, but show different TCR gene usage. Clone 48 was 10-fold more sensitive to VP16(369-379) peptide stimulation than clone 5 as assayed by proliferation assays, correlating with differences in MHC tetramer binding. Clone 48 gave positive staining with the DQ0602/VP16(369-379) tetramer at either 23 or 37 degrees C. Weak staining was also observed at 4 degrees C. Clone 5 showed weaker staining compared with clone 48 at 37 degrees C, and no staining was observed at 23 degrees C or on ice. Receptor internalization was not required for positive staining. Competitive binding indicates that the cell surface TCR of clone 48 has higher affinity for the DQ0602/VP16(369-379) complex than clone 5. The higher binding affinity of clone 48 for the peptide-MHC complex also correlates with a slower dissociation rate compared with clone 5.

Studies of the stability of HLA-DQ have revealed a correlation between SDS stability of MHC class II alphabeta dimers and insulin-dependent diabetes mellitus (IDDM) susceptibility. The MHC class II alphabeta dimer encoded by HLA-DQA1*0102/DQB1*0602 (DQ0602), which is a dominant protective allele in IDDM, exhibits the greatest SDS stability among HLA-DQ molecules in EBV-transformed B-lymphoblastoid cells and PBLs. DQ0602 is also uniquely SDS stable in the HLA-DM-deficient cell line, BLS-1. We addressed the molecular mechanism of the stability of DQ0602 in BLS-1. A panel of mutants based on the polymorphic differences between HLA-DQA1*0102/DQB1*0602 and HLA-DQA1*0102/DQB1*0604 were generated and expressed in BLS-1. An Asp at beta57 was found to be critical for SDS stability, whereas Tyr at beta30, Gly at beta70, and Ala at beta86 played secondary roles. Furthermore, the level of class II-associated invariant chain peptide bound to HLA-DQ did not correlate with SDS stability, suggesting that class II-associated invariant chain peptide does not play a direct role in the unique SDS stability of DQ0602. These results support a role for DQB1 codon 57 in HLA-DQ alphabeta dimer stability and IDDM susceptibility.

Ag-specific CD4+ T cells are present in peripheral blood in low frequency, where they undergo recruitment and expansion during immune responses and in the pathogenesis of numerous autoimmune diseases. MHC tetramers, which constitute a labeled MHC-peptide ligand suitable for binding to the Ag-specific receptor on T cells, provide a novel approach for the detection and characterization of such rare cells. In this study, we utilized this technology to identify HLA DQ-restricted Ag-specific T cells in the peripheral blood of human subjects and to identify immunodominant epitopes associated with viral infection. Peptides representing potential epitope regions of the VP16 protein from HSV-2 were loaded onto recombinant DQ0602 molecules to generate a panel of Ag-specific DQ0602 tetramers. VP16 Ag-specific DQ-restricted T cells were identified and expanded from the peripheral blood of HSV-2-infected individuals, representing two predominant epitope specificities. Although the VP16 369-380 peptide has a lower binding affinity for DQ0602 molecules than the VP16 33-52 peptide, T cells that recognized the VP16 369-380 peptide occurred at a much higher frequency than those that were specific for the VP16 33-52 peptide.

Glutamic acid decarboxylase 65 (GAD65) is one of the major autoantigens in type 1 diabetes. We investigated whether there is variation in the processing of GAD65 epitopes between individuals with similar HLA backgrounds and whether the processing characteristics of certain immunogenic epitopes are different in distinct APC subpopulations. Using DR401-restricted T cell hybridomas specific for two immunogenic GAD65 epitopes (115-127 and 274-286), we demonstrate an epitope-specific presentation pattern in human B-lymphoblastoid cell lines (B-LCL). When pulsed with the GAD protein, some DRB1*0401-positive B-LCL, which presented GAD65 274-286 epitope efficiently, were unable to present the GAD65 115-127 epitope. However, all B-LCL presented synthetic peptides corresponding to either GAD epitope. In addition, when pulsed with human serum albumin, all cell lines gave equal stimulation of a DR4-restricted human serum albumin-specific T hybridoma. GAD65-transfected cell lines displayed the same presentation phenotype, showing that lack of the presentation of the 115-127 epitope was not due to inefficient uptake of the protein. Blood mononuclear adherent cells, B cells, or dendritic cells derived from the same individual displayed the same presentation pattern as observed in B cell lines, suggesting that the defect most likely is genetically determined. Therefore, individual differences in Ag processing may result in the presentation of distinct set of peptides derived from an autoantigen such as GAD65. This may be an important mechanism for the deviation of the immune response either into a regulatory pathway or into an inflammatory autoimmune reactivity.

HLA-DQ alleles are closely associated with susceptibility and resistance to insulin-dependent diabetes mellitus (IDDM) but the immunologic mechanisms involved are not understood. Structural studies of the IDDM-susceptible allele, HLA-DQA1*0301/DQB1*0302, have classified it as a relatively unstable dimer, particularly at neutral pH. This is reminiscent of studies in the nonobese diabetic mouse, in which I-A(g7) is relatively unstable, in contrast to other murine I-A alleles, suggesting a correlation between unstable MHC class II molecules and IDDM susceptibility. We have addressed this question by analysis of dimer stability patterns among various HLA-DQ molecules. In EBV-transformed B-lymphoblastoid cell lines and PBL, the protein encoded by the IDDM-protective allele HLA-DQA1*0102/DQB1*0602 was the most SDS stable when compared with other HLA-DQ molecules, including HLA-DQA1*0102/DQB1*0604, a closely related allele that is not associated with protection from IDDM. Expression of six different HLA-DQ allelic proteins and three different HLA-DR allelic proteins in the bare lymphocyte syndrome cell line, BLS-1, revealed that HLA-DQA1*0102/DQB1*0602 is SDS stable even in the absence of HLA-DM, while other HLA class II molecules are not. These results suggest that the molecular property of HLA-DQ measured by resistance to denaturation of the alphabeta dimer in SDS may play a role in IDDM protection.

TCR engagement of peptide-MHC class II ligands involves specific contacts between the TCR and residues on both the MHC and peptide molecules. We have used molecular modeling and assays of peptide binding and T cell function to characterize these interactions for a CD4+ Th1 cell clone, ESL4.34, which recognizes a peptide epitope of the herpes simplex type 2 virus virion protein, VP16 393-405, in the context of several HLA-DR alleles. This clone responded to VP16 393-405 in proliferation and cytotoxicity assays when presented by DRB1*0402, DRB1*1102, and DRB1*1301, which share a common amino acid sequence, ILEDE, at residues 67-71 in the alpha-helical portion of the DRbeta polypeptide, but not when presented by other DR4, DR11, and DR13 alleles that are negative for this sequence. Using a panel of APCs expressing DR4 molecules that were mutagenized in vitro at individual residues within this shared epitope and using peptide analogues with single amino acid substitutions of predicted MHC and TCR contact residues, a unit of recognition was identified dependent on DRbeta residues 67-71 and relative position 4 (P4) of the VP16 393-405 peptide. The interactions of this portion of the peptide-DR ligand with the ESL4.34 TCR support a structural model for MHC-biased recognition in some Ag-specific and alloreactive T cell responses and suggest a possible mechanism for autoreactive T cell selection in rheumatoid arthritis.

Vaccination with synthetic TCR peptides from the BV5S2 complementarity-determining region 2 (CDR2) can boost significantly the frequency of circulating CD4+ peptide-specific Th2 cells in multiple sclerosis (MS) patients, with an associated decrease in the frequency of myelin basic protein (MBP)-reactive Th1 cells and possible clinical benefit. To evaluate the immunogenicity of CDR2 vs other regions of the TCR, we vaccinated seven MS patients with overlapping BV5S2 peptides spanning amino acids 1-94. Six patients responded to at least one of three overlapping or substituted CDR2 peptides possessing a core epitope of residues 44-52, and one patient also responded to a CDR1 peptide. Of the CDR2 peptides, the substituted (Y49T)BV5S2-38-58 peptide was the most immunogenic but cross-reacted with the native sequence and had the strongest binding affinity for MS-associated HLA-DR2 alleles, suggesting that position 49 is an MHC rather than a TCR contact residue. Two MS patients who did not respond to BV5S2 peptides were immunized successfully with CDR2 peptides from different BV gene families overexpressed by their MBP-specific T cells. Taken together, these results suggest that a widely active vaccine for MS might well involve a limited set of slightly modified CDR2 peptides from BV genes involved in T cell recognition of MBP.

Engagement of CD28 induces a major costimulatory pathway required by many CD4+ T cells in addition to activation via the TCR. In the absence of signals provided by CD28, ligation of the TCR alone can induce anergy or apoptosis in CD28+ cells. However, we report here characterization of a distinct subset of CD4+ T cells that are CD28-. Three autoreactive CD4+ human T cell clones that could be activated to produce IL-2 and proliferate by anti-CD3 alone were found to lack expression of CD28. CD28- clones that were activated with anti-CD3 alone were not anergic to restimulation via CD3. The presence of CD28-CD4+ T cells was verified in peripheral blood, and their frequency ranged from 0% to >22% of CD4+ T cells in different individuals. The percentage of CD28-CD4+ T cells in the peripheral blood of 57 individuals was significantly correlated with specific class II MHC alleles. Persons with HLA-DRB1*0401 and DR1 alleles had significantly higher numbers of CD28- T cells, while individuals with HLA-DR2(15) had significantly fewer CD28-CD4+ T cells than the mean. Like the CD28- clones, CD28-CD4+ T cells isolated from peripheral blood proliferated upon CD3 cross-linking in the absence of costimulation. The finding that CD28-CD4+ T cells resist induction of anergy following engagement of the TCR in the absence of conventional costimulation demonstrates one mechanism by which autoreactive T cells can escape processes that censor self-reactivity. The MHC associations observed suggest a relationship with autoimmunity and loss of self-tolerance.

Intermolecular interactions between peptides and class II molecules intracellularly are likely to be influenced by different pH conditions in different endosomal compartments. We compared the pH-dependent binding of peptides to HLA-DQ alleles 3.1 and 3.2, two well-studied human class II molecules with limited structural variability but different genetic associations with autoimmune diseases. Binding of the "promiscuous" binding peptide 34P3A as well as that of several allele-specific peptides were optimal for DQ3.2 at relatively acidic pH, compared with DQ3.1, with a pH optimum for DQ3.2 at pH 5.5 and for DQ3.1 at pH 6.5. A specific polymorphism at residue 57 of the DQ molecule accounted for some, but not all, of this difference. When the intrinsic stability of the class II dimer was assessed using partially denaturing gel electrophoresis after incubation at different pH, the DQ3.2 molecule demonstrated relative instability at neutral pH compared with DQ3.1, again suggesting a preference for peptide binding at relatively acidic conditions for HLA-DQ3.2. Interestingly, this preference for binding at acidic pH was not found for a class II-associated invariant chain peptide. Intrinsic alphabeta dimer stability and the pH optima for peptide binding studied in this report are likely to be significant determinants for allelic differences in DQ binding profiles in vivo.

Polymorphic residues of HLA class II molecules influence immune activation in part by determining specific structural constraints for binding antigenic peptides. We identified a peptide from glutamic acid decarboxylase, a diabetes-associated autoantigen that preferentially bound to HLA-DQ3.2 molecules, one of the HLA determinants highly associated with insulin-dependent diabetes. We analyzed interactions of specific HLA-DQ residues with modified peptide analogues and found a pattern of permissive site-specific amino acids that accommodated allele-specific binding. Four anchor residues constrain binding to different DQ alleles; limited variation at two of these sites, residues 4 and 9, accounts for the unique pattern of peptide binding to HLA-DQ3.1 or HLA-DQ3.2.

The impact of HLA-DQ allelic polymorphism on peptide binding interactions was investigated. MHC class II binding peptides from lambda repressor (lambda R), thyroid peroxidase (TPO), and a modified poly-alanine peptide (AKY) were used in two different binding assays, one with purified DQ molecules and another with cells that expressed DQ cell surface dimers. Class II heterodimers of different combinations of DQ alpha-chain and the DQ beta-chain independently influenced specific peptide interactions. Each of the peptides tested bound differentially to DQ alleles: TPO632-645 bound test to DQ2 molecules (DQA1*0501/DQB1*0201), lambda R12-24 bound best to DQ8 molecules (DQ3.2 or DQA1*0301/DQB1*0302), and AYK bound best to DQ1 (DQA*0102/DQB*0602) and DQ7 (DQ3.1 OR DQA1*0301/DQB1*0301). However, in each case in which either the DQ alpha- or DQ beta-chain was exchanged, major alterations or reversals of this pattern of interaction were observed. The interaction of lambda R with DQ8 was highly dependent on specific sequence polymorphisms at residue 57; an Ala-->Asp substitution on the DQ beta-chain at the polymorphic codon 57 prevented peptide binding. However, Ala 57 alone was not sufficient to permit binding in other DQ allelic contexts. These studies demonstrate the critical role of specific DQ polymorphisms in establishing the nature of bound Ag and thereby influencing the potential immune repertoire. Analysis of the ability of DQ molecules to bind and present antigenic peptides should aid in understanding their role in immunity and in development of HLA-DQ-associated autoimmune disease.

HLA-DQ class II molecules are complex heterodimers in which polymorphic alpha- and beta-polypeptides form noncovalent associations leading to cell surface expression. Using gene transfer with retroviral vectors, we demonstrate that surface expression of heterodimers among various DQ alpha and -beta polypeptides is not universal. Two types of allele-specific constraints were found: DQ beta chains that are associated with the DQw1 specificities do not form stable cell surface heterodimers with DQ alpha chains encoded on haplotypes that are associated with DQw2, DQw3, or DQw4. Similarly, surface expression of heterodimers composed of DQw2- or DQw3-associated DQ beta chains and DQw1 DQ alpha chains were not detected. The formation of mixed isotype heterodimers also showed allelic preference: DR alpha/DQw1 beta dimers were detected on the cell surface, but DR alpha/DQw3 beta dimers were not. Failure to express class II heterodimers on the cell surface was apparently due to either inefficient assembly of the alpha- and beta-chains or retention of the mismatched dimers in an intracellular compartment. DQ beta cDNA encoding DQw1- and DQw3-associated beta-polypeptides were modified to create chimeric expression vectors in which portions of the DQ beta polypeptides were exchanged between different alleles. The resulting chimeric DQ beta polypeptides were tested for their ability to form stable heterodimers with different DQ alpha polypeptides. A cluster of polymorphic amino acid residues encoded by the 3' end of the DQB1 second exon, corresponding to residues 60 to 91 in the DQ beta polypeptide, were found to be critical determinants constraining stable cell surface heterodimer formation among preferential alpha- and beta-combinations.

HLA class II genes comprise a large multigene family with intra- and interlocus variation in structure and expression. Within this family of related genes, the HLA-DX alpha and beta loci (HLA DQA2 and DQB2) are highly homologous to functional HLA-DQ loci (HLA DQA1 and DQB1) but are frequently termed pseudogenes because DX gene transcription has not been observed, even in cells expressing HLA-DQ. Analysis of upstream transcriptional regulatory elements for the DX beta and DQ beta genes identified a high degree of nucleotide homology, consistent with their derivation from a common ancestral class II gene. However, transient expression assays with plasmids utilizing promoter elements from the DX beta gene had no activity in transfected human B cells, in contrast to homologous DQ beta sequences. Reciprocal exchange of specific sequences from the DQ beta gene with those of the DX beta gene restored expression to wild-type DQ beta levels, as did mutagenesis of only three DX-specific nucleotides in the upstream regulatory region. These three nucleotides mark two binding sites for distinct nuclear DNA-binding proteins that differentially recognize DQ beta and DX beta sequences. Transcription of these genes is critically dependent on interactions between these two upstream regulatory region sites which distinguish DX beta from its closely related homologue, DQ beta.

The evolutionary history of MHC class II genes is characterized by several examples of gene duplication, leading both to the creation of distinct subregions such as DR, DQ, and DP, as well as to duplicated loci within each of these subregions. In the human MHC, a prominent example of this diversification occurs within the HLA-DQ subregion, where the nonpolymorphic and transcriptionally "silent" DQB2 locus is highly homologous to the polymorphic expressed DQB1 locus. In order to gain some insight into the mechanisms constraining polymorphism at the DQB2 locus, the second exons of five nonhuman primate DQB2 alleles were sequenced. Six nonhuman primate DQB2 analogous sequences were obtained, two each from chimpanzee and owl monkey cell lines, and one each from gorilla and gibbon cell lines. Notably, the DQB2 sequences from the gibbon, gorilla, and one of the two chimpanzee sequences, although containing some silent nucleotide changes, encode a predicted DQB2 protein with 100% homology to the human DQB2 sequence. The owl monkey DQB2 allelic sequences and the other chimpanzee DQB2 sequence contain additional polymorphisms, but maintain approximately 95% nucleotide sequence identity with human and the other primate sequences. Identification of the owl monkey DQB2 locus indicates that the ancestral DQ gene duplication event occurred at least 40 million years ago, rather than 10 million years, as previously thought. Remarkably, nucleotide sequences from amplified cDNA indicate that the DQB2 gene, and not the DQB1 gene, may be transcribed in the owl monkey line. Substitutions occur at sites comparable to codons of well-recognized allelic variation in the functional DQB1 genes, implying that variation within the DQB2 locus operates under similar selection constraints to the DQB1 locus, with an extremely high degree of conservation through primate evolution.

We analyzed HLA class II genomic polymorphisms in three families in which bone marrow transplantation was performed between individuals presumed to be HLA identical, but in which unexplained mixed lymphocyte culture reactivity was observed. These families were characterized by classical HLA serology, MLC, and DP typing. In each family, a pair of "HLA-identical" siblings demonstrated a small proliferative response in bidirectional MLC. Southern blotting analysis performed with cDNA probes for DQ alpha, DP alpha, and DP beta identified DP genomic differences in each case. Hybridization of Bgl II-digested genomic DNA with a DP alpha cDNA probe revealed three prominent polymorphic fragments (7.7, 5.8, and 3.7 kb), which discriminated between presumptive identical siblings and indicated crossover events within HLA. Similarly, hybridization of SstI-digested genomic DNA with a DP beta cDNA probe, although resulting in a more complex pattern, identified DP genomic disparity between the presumed HLA identical siblings. Hybridization of SstI-digested DNA from two families with evidence of DP recombination was performed by using an oligonucleotide probe specific for the newly described HLA class II gene DO beta. Two major polymorphic fragments, at 6.2 and 3.3 kb, segregated in these families and localized the crossovers flanking the DO beta gene between the DQ and DP loci. The contribution of the antigenic differences marked by these HLA DP and DO DNA polymorphisms to allorecognition in MLR and in graft-vs-host disease are discussed.

Biochemical diversity among products of class II HLA genes has been observed in individuals who appear to be HLA-D and DR-identical by cellular and serologic typing. We used techniques of restriction enzyme fragment analysis by Southern blotting to analyze this diversity at the level of cellular DNA. A panel of 17 HLA-DR4 homozygous cell lines (HCL) were investigated by using cDNA probes homologous to DQ beta, DQ alpha, and DR beta genes. Each probe was hybridized to cellular DNA digested with a series of different restriction endonucleases. Polymorphisms were observed with the use of the enzymes Pst I, Hind III, and Bam HI: Hybridization of cellular DNA digested with Hind III and Pst I with the DQ beta probe revealed specific polymorphisms, as did hybridization of the Pst I digest with the DQ alpha cDNA probe and the Bam HI digest with the DR beta probe. The observed differences fall into two categories: first, considerable diversity was seen between HLA-DR4 HCL that represent different HLA-D-defined haplotypes; second, diversity was also observed among HCL of the same DR4-associated HLA-D cluster. In contrast to the DQ cDNA probes, hybridization with the DR beta probe revealed relatively limited polymorphism by using a panel of different restriction endonucleases. Thus, although there is a general pattern of polymorphic restriction enzyme fragments homologous to DQ probes within an HLA-D cluster, the pattern seen for any particular cell line was not sufficiently distinct to assign an HLA-D or DR specificity.

The HLA-DQ region contains highly polymorphic alpha and beta loci, so that a diverse set of cis- and trans-associated class II alpha/beta dimers are potentially generated in heterozygous individuals. To evaluate the extent of this predicted diversity, DQ2 beta or DQ3.2 beta cDNA were introduced into a panel of homozygous B cell lines that expressed different DQ alpha alleles. Restricted patterns of alpha/beta pairing were observed in which DQ2 beta and DQ3 beta molecules were unable to pair efficiently with DQ1 alpha chains. This pairing anomaly may contribute to altered class II phenotypes in heterozygous individuals, and is reflected in the absence of either DQ1 alpha, DQ2 beta or DQ1 alpha, DQ3 beta haplotypes in the known human gene pool.

Bgl II restriction endonuclease digestion of genomic DNA from lymphoblastoid cell lines homozygous for HLA DR and DQ serological specificities, followed by hybridization with a DQ alpha cDNA probe, identified a genomic polymorphism characterized by two reciprocal patterns, one associated with DR 3, 5 and 8 and the other with DR 1, 2, 4, 7, and 9. The former pattern corresponded precisely to the reactivity of monoclonal antibody SFR20-DQ alpha 5, shown by Western blotting to react with isolated alpha-chains, but not with beta-chains. Additional variants of the DQ alpha genes were identified by using a locus-specific oligonucleotide probe for the DQ alpha gene, indicating differences among the DQ alpha 5-negative set of alleles. This analysis defines a set of DQ alpha allelic markers that are distinct from the well-established DQ serologic specificities DQw1, 2, 3 or "blank." Although most DQ alpha 5+ cells carry the DRw52 specificity associated with the DR beta 2 gene, analysis of DQ alpha polymorphisms on DR5, DQw1; DR8, DQw1; and DRw13, DQw1 cells verified that this DQ alpha family of alleles was not invariably linked to the DR beta 2 locus.

Retroviral vectors with an internal cytomegalovirus major immediate-early gene enhancer/promoter regulating HLA class II gene expression were used to transfer HLA cDNA into human EBV-transformed B-lymphoblastoid cell lines. HLA-DQ2 beta and DQ3.2 beta cDNA were transferred into DQ3.2 and DQ2 homozygous lymphoblastoid cell lines, respectively. Serologic analysis of the infected cell lines with allospecific mAb demonstrated surface expression of these exogenous DQ molecules implying that DQ alpha-chains from DR3, DQ2-positive cells can pair with DQ3.2 beta-chains and, similarly, DQ alpha-chains from DR4, DQ3.2-positive cells can pair with DQ2 beta-chains. Immunoprecipitation of the introduced DQ3.2 beta molecule resulted in co-purification of the allotype-mismatched endogenous DQ2 alpha polypeptide. We also show that vectors with a cytomegalovirus major immediate-early gene enhancer/promoter result in higher levels of expression of the transduced gene compared to previously described HLA vectors with either the SV-40 early enhancer/promoter or the lymphotropic papovavirus-enhanced SV-40 promoter. Although deletion of HLA cDNA did sometimes occur in the process of generating virus-producing clones, the HLA cDNA is stably maintained in virus-producing clones, once it is generated. This retroviral expression system is a highly efficient way to study HLA gene function.

The polymorphism of HLA class II molecules in man is particularly evident when comparisons between population groups are made. This study describes a DR3 haplotype commonly present in the American black population. Unlike the Northern European population in which almost all DR3 individuals are DQw2, approximately 50% of DR3-positive American blacks express a serologically undefined DQ allelic product. DNA restriction fragment analysis with the use of several unrelated individuals and an informative family has allowed us to identify unique DQ alpha- and beta-fragments associated with the DR3, DQw- haplotype. Based on fragment size, the DQ alpha genes of the DR3, DQw- and DRw8, DQw- haplotypes are similar as are the DQ beta genes of DR3, DQw-; DRw8, DQw-; and DR4, DQw- haplotypes. In addition, a DX beta gene polymorphism has been identified which is associated with some DR3 haplotypes including the American black DR3, DQw- haplotype. cDNA sequence analysis has revealed a DQw2-like alpha gene and a DQ beta gene which is similar to that previously described for a DR4, DQw- haplotype. It is postulated that recombination between DQ alpha and DQ beta genes and between the DQ and DX subregions has generated the various DR3 haplotypes and has played an important role in creating diversity in the HLA-D region.