An autoimmune disease with multiple molecular targets abrogated by the transgenic expression of a single autoantigen in the thymus.

Many autoimmune diseases are characterized by autoantibody reactivities to multiple cellular antigens. Autoantigens are commonly defined as targets of the autoimmune B cell response, but the role, if any, of these autoantigens in T cell-mediated autoimmune diseases is generally unknown. Murine experimental autoimmune gastritis is a CD4+ T cell-mediated organ-specific autoimmune disease induced by neonatal thymectomy of BALB/c mice. The murine disease is similar to human autoimmune gastritis and pernicious anemia, and is characterized by parietal and chief cell loss, submucosal mononuclear cell infiltrates, and autoantibodies to the alpha and beta subunits of the gastric H/K ATPase. However, the specificity of T cells that cause the disease is not known. To examine the role of the H/K ATPase in this T cell-mediated disease, transgenic mice were generated that express the beta subunit of the H/K ATPase under the control of the major histocompatibility complex class II I-Ek alpha promoter. We show that transgenic expression of the gastric H/K ATPase beta subunit specifically prevents the onset of autoimmune gastritis after neonatal thymectomy. In addition, thymocyte transfer experiments suggest that tolerance of pathogenic autoreactive T cells is induced within the thymus of the transgenic mice. We conclude that the beta subunit of the gastric H/K ATPase is a major T cell target in autoimmune gastritis and that thymic expression of a single autoantigen can abrogate an autoimmune response to multiple autoantigens.

From coiled tubules to a secretory canaliculus: a new model for membrane transformation and acid secretion by gastric parietal cells.

The acid-secreting gastric parietal cell has a unique secretory membrane system. This membrane system exists in an inactive (non-secreting) and an active (secreting) form. The current accepted model to explain the transformation events associated with the conversion of the non-secreting membrane to the secreting membrane, and vice versa, invokes membrane recycling of elongated vesicle structures. However, recent studies employing cryopreparation have shown that the non-secreting membrane in these cells is actually a complex network of helically coiled tubules. Here, we present an alternative model to explain how the membrane in parietal cells is activated to secrete HCl.

Homing and adhesion molecules in autoimmune gastritis.

The pathogenesis of autoimmune gastritis is the result of lymphocyte infiltration of the gastric mucosa, however, the events leading to the selective extravasation of autoreactive lymphocytes are unclear. Here we have examined the expression of adhesion molecules in the gastric mucosa of BALB/c mice with neonatal thymectomy-induced gastritis. The overall area of vascular endothelium was not significantly different between gastritic and non-gastritic mice. However, a significant increase in the area of mucosal endothelium expressing MAdCAM-1 in gastritic mice was observed. Treatment of neonatally thymectomized BALB/c mice with a MAdCAM-1 specific monoclonal antibody (MECA 367) reduced the incidence of autoimmune gastritis from 80 to 26%. Treatment with a monoclonal antibody (R1-2) directed to the MAdCAM-1 ligand, alpha4beta7, also resulted in a reduction in the incidence of gastritis to 40%. These findings identify the alpha4beta7/MAdCAM-I interaction as a pivotal event in the initiation of autoimmune gastritis.

Autoimmune gastritis: tolerance and autoimmunity to the gastric H+/K+ ATPase (proton pump).

The alpha and beta subunits of the gastric H+/K(+)-ATPase (proton pump) have been identified as the major molecular targets of parietal cell autoantibodies associated with pernicious anaemia and with murine experimental autoimmune gastritis (EAG) induced by neonatal thymectomy. Recent studies with EAG suggest that the mechanisms of peripheral tolerance and autoimmunity to extrathymic autoantigens are mediated by subsets of "regulator" and "effector" CD4+ T cells, respectively. The persistence of "effector" CD4+ autoreactive T cells in the periphery may be a direct consequence of the delayed developmental expression of the target autoantigen. We hypothesize that cytokines produced by the "regulator" T cells prevent the clonal expansion of the "effector" autoreactive T cells, and that neonatal thymectomy induces organ-specific autoimmunity in genetically susceptible individuals by the reduction of the "regulator" T cell population.

Defining T cell receptors which recognise the immunodominant epitope of the gastric autoantigen, the H/K ATPase beta-subunit.

We have previously shown that autoimmune gastritis can be elicited in mice by immunisation with the gastric parietal cell H/K ATPase alphabeta heterodimer, and, furthermore, have identified the H/K ATPase beta-subunit epitope, H/Kbeta253-277 as the dominant epitope of the gastric H/K ATPase. Using gastric H/K ATPase-immunised mice, here we have generated two T cell hybridomas specific for the H/Kbeta253-277 peptide, namely 4B11.F4.5 and 1E4.C1. Hybridoma 4B11.F4.5 uses Valpha8 and Vbeta8.2 TCR chains and 1E4.C1 uses Valpha9 and V1beta8.3 chains. Although both hybridomas are specific for H/Kbeta253-277, T cell assays using overlapping 14-mers of the 25-mer epitope showed that the two autoreactive TCRs recognise different regions of the 25-mer. The TCR from 1E4.C1 has been used to generate a TCR beta-chain transgenic mouse. >80% of peripheral CD4+ T cells utilise the Vbeta8.3 transgene. As expected, 1E4-TCR beta-chain transgenic mice are susceptible to neonatal thymectomy induced autoimmune gastritis. While none of the 1E4-TCR beta chain transgenic mice spontaneously developed a destructive gastritis, a minority (20%) of the transgenic mice developed a non-invasive and non-destructive gastritis. This suggests that the pathogenic T cells are maintained in a tolerant state in the periphery of the transgenic mice.

The role of natural killer cells in the induction of autoimmune gastritis.

A number of experimental models of organ-specific autoimmunity involve a period of peripheral T cell lymphopenia prior to disease onset. In particular, experimental autoimmune gastritis, induced in susceptible mouse strains by neonatal thymectomy, is a CD4+ T cell mediated autoimmune disease. We have previously demonstrated that this disease displays the hallmarks of a Th1-mediated DTH inflammatory response with an essential role for IFN-gamma very early in the pathogenesis of disease. Given the interplay between the innate and adaptive immune responses, a potential source of early IFN-gamma production in these lymphopenic mice is the innate immune response. Here we have assessed the contribution of innate immunity to the induction of experimental autoimmune gastritis, in particular, the role of natural killer (NK) cells in production of IFN-gamma. Analysis of NK cells and macrophages revealed no difference in either the number or activation status between euthymic and neonatally thymectomised mice. Furthermore, in vivo depletion of NK cells immediately after neonatal thymectomy of (BALB/cCrSlcxC57BL/6) F1 mice demonstrated no reduction in disease incidence compared to control groups of neonatally thymectomised mice. Therefore, we conclude that NK cells are not the primary source of IFN-gamma required for the pathogenesis of autoimmune gastritis following neonatal thymectomy but rather the small cohort of T cells in the periphery of lymphopenic mice are likely to be responsible for the IFN-gamma production.

A novel method for isolating mononuclear cells from the stomachs of mice with experimental autoimmune gastritis.

Autoimmune gastritis induced in BALB/c mice by neonatal thymectomy is a CD4+ T cell-mediated disease. The disease is characterised by mononuclear cell infiltrates in the gastric mucosa, loss of gastric parietal and chief cells and autoantibodies to the gastric H/K ATPase. Here we describe a simple non-enzymatic method for isolating cellular infiltrates from stomachs of gastric mice by injection of medium directly into stomach walls, causing swelling and rupture. Using this method, large numbers of viable lymphocytes were released from stomachs for analysis by flow cytometry. An 8.3 fold increase in the total number of lymphocytes from diseased stomachs compared to normal controls was observed. Total cell numbers of CD4+ and B cells were increased 4.8 fold and 39.5 fold respectively, in diseased stomachs compared with controls. No change was observed in the CD8+ T cell population. This method will allow detailed quantitative analysis of cellular infiltrates during the development of the gastric lesion and enrichment of pathogenic T cells for analysis and cloning. This procedure may have general application for the isolation of cellular infiltrates from lesion sites of other organs.

Alpha and beta subunits of the gastric H+/K(+)-ATPase are concordantly targeted by parietal cell autoantibodies associated with autoimmune gastritis.

We have previously shown that parietal cell autoantibodies predominantly react with a 60-90 kDa gastric autoantigen, subsequently identified as the beta subunit of the gastric H+/K(+)-ATPase (EC 3.6.1.3) (proton pump) whereas Karlsson et al showed that these autoantibodies primarily target the 95 kDa alpha subunit of the pump. In view of these discordant results, we have reassessed the reactivity of parietal cell autoantibodies with the two subunits of the gastric H+/K(+)-ATPase. We show here that all 26 parietal cell autoantibody-positive sera immunoblot both subunits under appropriate, but mutually exclusive, conditions. Thus, reactivity of anti-parietal cell autoantibodies with the 95 kDa alpha subunit is optimal when the SDS-PAGE is carried out with samples which are reduced but not boiled. Whereas reactivity with the 60-90 kDa beta subunit is optimal with samples which are boiled but not reduced. Autoantibody reactivity with the beta subunit is critically dependent on the presence of a full complement of N-linked glycans since partially deglycosylated protein, and recombinant beta subunit expressed in COS cells, bearing high mannose N-glycans, failed to bind to the autoantibody. These studies also suggest that B cell auto-epitopes are located on the lumenal domain of the beta subunit. Reactivity of parietal cell autoantibodies with a bacterial fusion protein incorporating the catalytic cytoplasmic domain of the alpha subunit suggests the presence of auto-epitopes in this region of the molecule.

Expression of the gastric H/K-ATPase alpha-subunit in the thymus may explain the dominant role of the beta-subunit in the pathogenesis of autoimmune gastritis.

The two subunits of the gastric H/K ATPase, namely the catalytic alpha-subunit and the glycoprotein beta-subunit, are the major targets of parietal cell autoantibodies associated with human and murine autoimmune gastritis. The murine disease induced by neonatal thymectomy is T cell-mediated. We have previously shown that transgenic expression of the H/K ATPase beta-subunit gene in the thymus prevented the development of autoimmune gastritis induced by thymectomy. However, little is known of the contribution of the H/K ATPase alpha-subunit in disease development. Here, we show that (1) in contrast to the gastric H/K ATPase beta-subunit, the alpha-subunit gene is expressed in normal BALB/c thymus. (2) transgenic expression of the gastric H/K ATPase alpha-subunit gene in the thymus failed to prevent the development of autoimmune gastritis and (3) normal BALB/c and transgenic mice expressing the alpha-subunit in the thymus develop autoimmune gastritis following immunisation with purified murine gastric H/K ATPase, whereas transgenic mice expressing the beta-subunit in the thymus do not. We propose that the expression of the H/K ATPase alpha-subunit in the normal thymus may account for the predominant role of the beta-subunit in the development of autoimmune gastritis induced either by thymectomy or by immunisation with the ATPase.

Reciprocal changes in trefoil 1 and 2 expression in stomachs of mice with gastric unit hypertrophy and inflammation.

H+/K+-ATPase beta-subunit-deficient mice (129/Sv background) display numerous pathologies in the stomach. Expression of the mutation in BALB/cCrSlc mice results in the development of an aberrant 'mucus-rich' cell population. 'Mucus-rich' cells have been described in stomachs of mice with autoimmune gastritis, a disease mediated by CD4+ T cells. Other pathological features of autoimmune gastritis are similar to those in H+/K+ beta-deficient mice and include a mononuclear cell infiltrate in the gastric mucosa, non-functional or absent parietal cells, depletion of zymogenic cells, hypergastrinaemia, and gastric unit hypertrophy caused by immature cell hyperplasia. The present study investigates further the aberrant gastric 'mucus-rich' cell lineage and analyses the mRNA expression of mucus cell products TFF1 and TFF2. 'Mucus-rich' cells stained for both acidic and neutral mucins, and with a TFF2-specific antibody. Stomachs from both models expressed decreased TFF1 mRNA and reciprocally increased TFF2 mRNA. The involvement of gastrin in regulating trefoil mRNA expression was also investigated using gastrin-deficient mice. In contrast to previous findings, gastrin did not positively regulate TFF1 mRNA expression, but there was possible augmentation of TFF2. Additionally, a clear role for inflammation was established involving both polymorphonuclear and mononuclear cells in these models, and a link was found between mucosal hypertrophy and increased interleukin-11 (IL-11) expression.

Heterogeneous glycosylation of murine transferrin receptor subunits.

The N-linked glycosylation of the murine receptor for transferrin has been investigated. Previously we have found that purified receptors appear as two bands after sodium dodecyl sulfate/polyacrylamide gel electrophoresis and Coomassie blue staining [van Driel, I.R., Stearne, P.A., Grego, B., Simpson, R.J. and Goding, J.W. (1984) J. Immunol. 133, 3220-3224]. In the current report we show that the two bands are due to different glycosylation of individual receptor molecules. The receptors have three asparagines to which N-linked glycans can be added, but only two sites are glycosylated in all receptors. The level of glycosylation of the third site varies depending on cell line or tissue. Limited endoglycosidase digestion of mature receptors indicates that differential glycosylation probably occurs at only one particular asparagine residue. Possible mechanisms that could result in such a glycosylation pattern are discussed.

Plasma cell membrane glycoprotein PC-1. Primary structure deduced from cDNA clones.

The PC-1 protein is a membrane glycoprotein that is selectively expressed on the surface of antibody-secreting cells. Previous work has shown that it consists of two apparently identical disulfide-bonded polypeptides, each of molecular weight approximately 120,000. We now describe the sequence of PC-1 mRNA and protein. The PC-1 protein is shown to consist of 905 amino acids and to have an uncommon transmembrane orientation. The NH2-terminal 58 residues are intracellular and the COOH-terminal 826 residues are extracellular. A cysteine-rich region of 85 amino acids lies adjacent to the extracellular surface of the membrane and appears to have arisen by exon duplication. In common with other membrane glycoproteins with this orientation, there is no obvious signal sequence other than the transmembrane segment. The PC-1 protein therefore has an overall structure and membrane orientation that resembles those of the transferrin receptor, the asialoglycoprotein receptor, and the Ia invariant chain.

Hijacking a chaperone: manipulation of the MHC class II presentation pathway.

Novel antigen delivery systems are currently being developed by genetic manipulation of the MHC class II trafficking pathway. Specific targeting of endogenously synthesized antigens to the class II loading compartment can result in massively enhanced presentation of peptide epitopes. This emerging technology holds promise for a variety of clinical applications including vaccine development, cancer therapies and control of autoimmune diseases.

Proton and potassium transport by H+/K(+)-ATPases.

1. H+/K(+)-ATPases are members of the P-type ATPase multigene family. The prototypical H+/K(+)-ATPase is the protein that acidifies gastric luminal contents. The physiological and pharmacological significance of this pump has led to a detailed investigation of its biochemistry and molecular and cell biology. 2. Recently, a number of closely related H+/K(+)-ATPase isoforms have been discovered. These isoforms are present in organs other than the stomach, including the colon and kidney, where they contribute to acid-base and potassium homeostasis. The structure, expression and physiological roles of the gastric H+/K(+)-ATPase and other isoforms are reviewed.

Plasma cell antigen PC-1 and the transferrin receptor in mouse, rat, and hamster: serologic and biochemical analysis.

The plasma cell membrane antigen PC-1 and the receptor for the iron transport protein transferrin are high m.w., developmentally regulated proteins consisting of two similar or identical disulfide-bonded subunits. In this paper, we report the results of a serologic and biochemical analysis of these proteins in various strains of inbred mice, and in rats and hamsters. A monoclonal antibody against the PC-1a allelic product is shown to detect an antigenic determinant on the PC-1 molecule that has the same strain distribution as the antigen previously detected with polyclonal alloantisera. The mouse PC-1 protein was purified from plasma cells of the PC-1a genotype and was used to generate polyclonal rabbit anti-PC-1 antibodies. These antibodies precipitated a homologous protein from plasmacytoma cells derived from PC-1- congenic mice, demonstrating that PC-1b is not a "null" allele. The PC-1b allelic product had a slightly lower apparent m.w. than the PC-1a product, and had a slightly more basic isoelectric point. Rabbit anti-mouse PC-1 antibodies also precipitated a homologous protein from immunoglobulin-secreting cells of rat and hamster origin, but did not show detectable cross-reaction with the transferrin receptor. Disulfide bonding between chains was conserved in both PC-1 and the transferrin receptor in all species examined, but transferrin receptors from mouse cells had a significantly higher apparent m.w. than those of rat, hamster, or human cells.

Self-association of the low density lipoprotein receptor mediated by the cytoplasmic domain.

When the low density lipoprotein (LDL) receptor was solubilized from bovine adrenal cortex membranes and subjected to electrophoresis in the absence of reducing agents, a disulfide-bonded dimeric species was demonstrated. Formation of these covalent bonds was blocked when the tissue was homogenized in the presence of sulfhydryl alkylating agents, indicating that the native receptor was self-associated noncovalently and that the disulfide bond formation occurred only after homogenization. The disulfide-linked dimers were disrupted and the receptor was restored to a monomeric form when inside-out adrenal vesicles were treated with trypsin, suggesting that the disulfide bond formation involved the 50-amino acid cytoplasmic domain of the receptor. When the receptor was solubilized from bovine adrenal cortex membranes and then purified by ion exchange and affinity chromatography, it could be covalently coupled into dimers and trimers in the presence of bivalent cross-linking agents. Receptor dimers could also be demonstrated by chemical cross-linking of intact cells that were transfected with an expressible cDNA encoding the normal human LDL receptor. Dimer formation was markedly reduced in transfected cells expressing mutated cDNAs that had premature termination codons at positions 792, 807, and 812, which produced shortened receptors that retained 2, 17, and 22 of the original 50 amino acids of the cytoplasmic domain, respectively. The first two mutant receptors, which did not form oligomers, did not enter coated pits and were not rapidly internalized by cells. However, the mutant receptor that terminates at position 812 was internalized normally even though oligomer formation was greatly reduced. Moreover, a mutant receptor with a cysteine substituted for a tyrosine at position 807, which internalized very slowly, showed a normal susceptibility to chemical cross-linking. Deletion of external domains of the LDL receptor, including the epidermal growth factor homology region and the O-linked sugar domain, did not alter susceptibility to chemical cross-linking. We conclude that the cytoplasmic domain of the LDL receptor is responsible both for self-association into oligomers and for clustering in coated pits, but the available data do not establish a causal connection between these two events.

Identification and characterization of a soluble form of the plasma cell membrane glycoprotein PC-1 (5'-nucleotide phosphodiesterase).

PC-1 is a membrane glycoprotein, found on the surface of plasma cells and a few types of nonlymphoid cells, which has recently been found to have 5'-nucleotide phosphodiesterase activity. In this paper, we demonstrate the existence of enzymically active water-soluble forms of PC-1 in ascites from plasmacytoma-bearing mice, normal mouse serum, and in supernatants of cultured mouse plasmacytoma cells and mouse L cells transfected with a cDNA encoding the membrane form of PC-1. The water-soluble enzyme activity can be specifically immunoprecipitated by a monoclonal antibody to an allotypic determinant on the membrane form of PC-1, and resides on a slightly smaller polypeptide than membrane PC-1. Biosynthetic studies revealed a single, monomeric, endoglycosidase-H-sensitive membrane PC-1 precursor, which was gradually converted to a disulphide-bonded, endoglycosidase-H-resistant form over a period of about 2 h. Soluble PC-1 was first detectable in the supernatant after about 2 h. A distinct intracellular form of soluble PC-1 was not seen. The soluble form of PC-1 does not appear to arise by proteolytic cleavage from the cell surface, although cleavage inside the cell remains a possibility. When taken together with the structure of the relevant portions of PC-1 gene exons, the data suggest that the most likely site of cleavage is between Pro152 and Ala153.

Stoichiometric binding of low density lipoprotein (LDL) and monoclonal antibodies to LDL receptors in a solid phase assay.

The current paper describes a solid phase ligand binding assay for the low density lipoprotein (LDL) receptor that takes advantage of the domain structure of the protein. An antibody directed against one domain, e.g. the cytoplasmic tail, is adsorbed to a microtiter well. A detergent solution containing the LDL receptor is added, and the receptor is allowed to bind to the antibody. The wells are then washed, and one of the following radioiodinated ligands is added: 125I-LDL or an 125I-labeled monoclonal antibody directed against a different domain than the antibody adsorbed to the well. Under these conditions, the human LDL receptor shows high affinity for 125I-LDL and for 125I-IgG-HL1, a monoclonal antipeptide antibody directed against a 10-amino-acid "linker" between repeats 4 and 5 in the ligand binding domain. The binding affinity is the same at 4 degrees C and 37 degrees C. The binding of 125I-LDL and 125I-IgG-HL1 occurs with 1:1 molar stoichiometry, suggesting that the human LDL receptor binds 1 mol of LDL per mol of receptor. The acid-dependent dissociation of 125I-LDL and 125I-labeled monoclonal antibody from LDL receptors that is observed in intact cells was also shown to occur in the solid phase binding assay. We used the solid phase assay to demonstrate the secretion of LDL receptors from monkey cells that have been transfected with a cDNA encoding a truncated form of the human receptor that lacks the membrane-spanning domain. This assay may be useful in measuring the relative amounts of the intact LDL receptor in tissue extracts and the secreted receptor in transfected cells.

First cysteine-rich repeat in ligand-binding domain of low density lipoprotein receptor binds Ca2+ and monoclonal antibodies, but not lipoproteins.

The ligand binding domain of the low density lipoprotein receptor consists of seven cysteine-rich repeats of approximately 40 amino acids each. These repeats, which are located at the NH2 terminus of the protein, are homologous to sequences in complement components C8 and C9. To determine the role of the first repeat (amino acids 2-42), we prepared two plasmids containing expressible low density lipoprotein receptor cDNAs. The first plasmid, p delta R1, lacks only the nucleotides encoding the first repeat. It produced a receptor that bound and internalized lipoproteins and recycled to the cell surface with the same efficiency as the normal receptor. This deleted receptor failed to bind two monoclonal antibodies, IgG-C7 and IgG-15C8, which were shown previously to react with the ligand-binding domain. The second plasmid, pR1, encodes a markedly truncated protein whose extracellular domain consists of the first repeat joined to the transmembrane and cytoplasmic domains. This protein bound the two monoclonal antibodies with the same affinity as the normal receptor, but failed to bind lipoproteins. Binding of IgG-15C8 to the normal receptor and the pR1-encoded protein was Ca2+-dependent, indicating that the first repeat binds Ca2+. We conclude that repeats 2-6 in the ligand-binding domain are sufficient for binding lipoproteins and that the first repeat is highly immunogenic, but is not required for lipoprotein binding.