Autoimmune gastritis is a CD4+ T cell-mediated disease induced in genetically susceptible mice by thymectomy on the third day after birth. Previous linkage analysis indicated that Gasa1 and Gasa2, the major susceptibility loci for gastritis, are located on mouse chromosome 4. Here we verified these linkage data by showing that BALB.B6 congenic mice, in which the distal ∼40 Mb of chromosome 4 was replaced by C57BL/6 DNA, were resistant to autoimmune gastritis. Analysis of further BALB.B6 congenic strains demonstrated that Gasa1 and Gasa2 can act independently to cause full expression of susceptibility to autoimmune disease. Gasa1 and Gasa2 are located between D4Mit352-D4Mit204 and D4Mit343-telomere, respectively. Numerical differences in Foxp3+ regulatory T cells were apparent between the BALB/c and congenic strains, but it is unlikely that this phenotype accounted for differences in autoimmune susceptibility. The positions of Gasa1 and Gasa2 correspond closely to the positions of Idd11 and Idd9, two autoimmune diabetes susceptibility loci in nonobese diabetic (NOD), mice and this prompted us to examine autoimmune gastritis in NOD mice. After neonatal thymectomy, NOD mice developed autoimmune gastritis, albeit at a slightly lower incidence and severity of disease than in BALB/c mice. Diabetes-resistant congenic NOD.B6 mice, harbouring a B6-derived interval encompassing the Gasa1/2-Idd9/11 loci, demonstrated a slight reduction in the incidence of autoimmune gastritis. This reduction was not significant compared with the reduction observed in BALB.B6 congenic mice, suggesting a difference in the genetic aetiology of autoimmune gastritis in NOD and BALB mice.
Chronic autoimmune gastritis, in which adaptive immune responses target the gastric mucosa, is one of the most common autoimmune conditions in humans (1). While this condition may persist in an asymptomatic form for many years, a proportion of affected individuals will progress to pernicious anaemia. Pernicious anaemia is the major cause of vitamin B12 deficiency and affects ∼1.9% of the population over the age of 60 (1). Despite this relatively high prevalence of these diseases, little is known concerning the underlying genetic cause of autoimmune gastritis in humans.
In order to identify genes conferring susceptibility to autoimmune gastritis, we analysed BALB/c mouse strains that are highly susceptible to the disease (2–4). In a linkage analysis of (BALB/cCrSlc × C57BL/6)F2 segregating progeny, we identified two major genes on chromosome 4 that confer susceptibility to autoimmune gastritis, termed Gasa1 and Gasa2 (4). Subsequent analyses identified Gasa3 on chromosome 6 and Gasa4 on chromosome 17 as minor gastritis susceptibility genes (3).
It has been observed that many genetic loci are linked to more than one autoimmune disease (5, 6). It is reasonable to hypothesize that at least some of the underlying genes for these loci may contribute to the development of multiple autoimmune disorders (7) for at least three reasons. First, some autoimmune conditions occur concurrently. For example ∼20% of humans with autoimmune diabetes also have indications of gastric autoimmunity (8, 9). Second, there are several similarities in the pathogenesis of organ-specific autoimmune diseases, such as their dependence on activation of autoreactive CD4+ T cells. Third, linkage analyses have revealed genes conferring susceptibility to different autoimmune diseases tend to cluster to particular genomic segments, a finding most readily explained by postulating the existence of loci that control susceptibility to autoreactive immune responses (5, 10). For example, the PTPN22 and CTLA4 genes have been linked to several autoimmune diseases including autoimmune diabetes, thyroiditis and systemic lupus erythematosus (11–16). Autoimmune gastritis and autoimmune (type 1A) diabetes are excellent diseases in which to test whether some loci contribute to the development of multiple autoimmune disorders because they frequently co-occur in individuals and families (9) and several susceptibility genes for the two diseases have been co-located. For example, all four gastritis susceptibility loci mapped in BALB/c mice by us (2–4) are located in chromosomal regions bearing loci conferring susceptibility to autoimmune diabetes loci mapped in nonobese diabetic (NOD) mice by ourselves and others. Idd9 has been subdivided into three subregions Idd9.1, Idd9.2 and Idd9.3 by the production and phenotypic analyses of subcongenic strains (17). The position of Gasa1 overlaps the Idd9.1 congenic interval and the Idd11 linkage peak (Fig. 1) (17, 18, 19). The linkage peak of Gasa2 overlaps with the predicted Idd9.2 and Idd9.3 congenic intervals (Fig. 1). Gasa3 on chromosome 6 is predicted to be in close proximity to Idd6, Idd19 and Idd20 (20, 21). Finally, the H2 locus on chromosome 17 is associated with both diabetes (Idd1) and autoimmune gastritis (Gasa4) (3, 22, 23).
In this study, we specifically studied the major gastritis loci Gasa1 and Gasa2 by producing BALB/c congenic strains bearing C57BL/6-derived resistance loci to confirm our previous linkage data and improve the precision of their localization. In addition, the susceptibility of NOD mice to autoimmune gastritis was determined and compared with the disease that occurs in BALB mice. The susceptibility to gastritis of NOD.B6 congenic mice bearing C57BL-derived loci at Gasa1 and Gasa2 was characterized in an attempt to identify a relationship between gastritis and diabetes susceptibility, as this strain bears diabetes-resistance alleles at Idd11 and Idd9.
BALB/cCrSlc and BALB.B6 congenic mice were housed at The University of Melbourne, Department of Biochemistry and Molecular Biology, under conventional conditions. BALB.B6 congenic mice were generated by mating BALB/cCrSlc and C57BL/6 mice. The F1 progeny from this mating were backcrossed to BALB/cCrSlc mice to produce the first backcross generation. The appropriate progeny for subsequent backcrosses were selected based on presence of C57BL/6-derived alleles at microsatellites spanning from D4Mit31 to D4mit344 and backcrossed for 10 or 11 generations. A background screen of 128 markers distributed evenly across all chromosomes was performed by the Australian Genome Research Facility, Melbourne, Australia, and all markers except those in the congenic interval were found to be homozygous for the BALB/cCrSlc parental alleles (data not shown). Names of congenic mouse strains have been abbreviated in the manuscript as follows: BALB.B6-D4Mit122-D4Mit256 (BALB.B6-GasaA), BALB.B6-D4Mit204-D4Mit256 (BALB.B6-GasaB), BALB.B6-D4Mit284-D4Mit256 (BALB.B6-GasaC), BALB.B6-D4Mit122-D4Mit343 (BALB.B6-GasaD).
A23 TCR transgenic mice express a TCR directed to an epitope of the gastric H+/K+ ATPase α subunit as described (24) and were housed at The Department of Biochemistry and Molecular Biology, The University of Melbourne, under conventional conditions. NOD.B6-Chr4 congenic mice, produced as previously described (25), carry diabetes-resistance alleles at Idd9 and Idd11 on chromosome 4 and demonstrate a significant decrease in diabetes incidence compared with NOD/Lt mice when housed in clean conditions (25). All BALB/cJ, NOD/Lt and NOD.B6-Chr4 mice were housed at the Walter and Eliza Hall Institute under specific pathogen-free conditions.
Induction and assessment of gastric autoimmunity
Autoimmune gastritis was induced by thymectomy on the third day after birth as previously described (26). Thymectomized mice were killed at 12 weeks of age. Autoimmune gastritis was assessed by histological examination of stomachs and by ELISA for H+/K+ ATPase-specific autoantibodies as previously described (26). Gastritis was scored using a scale of 1 to 6 as follows. Score 0, normal gastric mucosa. Score 1 has very mild sub-mucosal mononuclear cell infiltration throughout the glandular mucosa or more substantial infiltration restricted to the glandular mucosa adjacent to the forestomach. Score 2 is characterized by mild sub-mucosal mononuclear cell infiltration throughout the glandular mucosa accompanied by focal aggregates of mononuclear cells that impinge into the mucosal area but no widespread depletion of differentiated cell types. Score 3 is characterized by sub-mucosal infiltration and mild, disseminated mononuclear cell infiltration of the mucosa accompanied by marked depletion of zymogenic cells in some areas of the oxyntic mucosa although many areas retain zymogenic cells but parietal cells still abound. Score 4 as for score 3 except that depletion of zymogenic cells is nearly complete and low-level hyperplasia is often seen. Score 5, sub-mucosal and mucosal mononuclear cell infiltrates, almost complete depletion of zymogenic cells, severe but not complete depletion of parietal cells and a very marked increase in immature cell types or mucous-rich cells usually resulting in substantial hyperplasia. Score 6 as for score 5 except near complete loss of zymogenic and parietal cell types.
Conjugated mAbs were purchased from BD PharMingen (San Diego, CA, USA), except for the anti-Foxp3 Staining Set, which was obtained from eBioscience (San Diego, CA, USA). Flow cytometric analyses were performed on a FACSort (BD Biosciences, San Jose, CA, USA) using CellQuest Pro software (BD Biosciences). Cells were analysed for expression of cell surface markers using a combination of the following antibodies: anti-CD3ϵ-FITC (145-2C11), anti-CD4-PerCP (RM4-5), anti-CD4 (L3T4), anti-mouse CD8α-PE (53-6.7), anti-CD90.1-PerCP (Thy1.1, encoded by Thy1a, OX-7) and anti-Foxp3-allophycocyanin (FJK-16s). Staining of cells and flow cytometric analysis was as previously described (27).
Indirect immunofluorescent staining of paraffin sections
Indirect immunofluorescence was performed on paraffin stomach sections prepared from BALB/cCrSlc mice. Sections were dewaxed in two 5-min histolene washes. Sections were then washed twice for 1 min in ethanol followed by rinses in water for 1 min and PBS for 1 min. Slides were treated in methanol for 10 min at −20°C followed by one rinse in PBS for 5 min. Sections were incubated in PBS/1% BSA for 20 min followed by one rinse in PBS/0.05% Tween-20 for 5 min. Mouse sera were diluted 1:20 in PBS/1% BSA and applied for 45 min followed by six 5-min washes in PBS/0.05% Tween-20. FITC-conjugated anti-mouse Ig secondary antibody was then applied for 30 min followed by six 5-min washes in PBS/0.05% Tween-20. Sections were mounted under a coverslip using Mowiol mounting fluid (13.3% w/v Mowiol 5-88, 33.3% w/v glycerol, 0.15 M Tris–HCL pH 8.5) and viewed using a fluorescence microscope (Zeiss, Göttingen, Germany).
Immunofluorescent staining of transfected HEK293 cells
Coverslips were incubated with 0.01% poly-L-lysine (w/v in water) for 15 min, washed twice with ultra pure water, dried and transferred to 24-well culture plates. HEK293 cells were seeded into the wells and allowed to grow to ∼40–60% confluence in complete DMEM. Cells were then transfected with plasmids encoding either mCherry-H+/K+ ATPase α subunit or yellow fluorescent protein (YFP)-H+/K+ ATPase β subunit (28) using Fugene (Invitrogen) according to manufacturer's instructions. After 48 h, coverslips were washed twice in PBS and fixed in 4% paraformaldehyde for 15 min, washed in 1% Triton-X100 for 4 min, incubated in 5% FCS/PBS for 30 min, incubated with sera samples (diluted 1:20 in PBS) for 30 min, washed in PBS and incubated with either FITC-conjugated anti-mouse Ig in the case of cells transfected with mCherry-H+/K+ ATPase α subunit plasmid or Alexa568-conjugated anti-mouse Ig, in the case of cells transfected with YFP-H+/K+ ATPase β subunit plasmid, for 30 min. Coverslips were washed in PBS followed by a final wash in water before mounting on microscope slides using Mowiol mounting fluid. Cells were visualized with a TCS SL Confocal Laser Microscope (Leica Microsystems, Wetzlar, Germany).
In vivo proliferation assays
Unfractionated spleen and lymph node cells from CD90.1+ A23 TCR transgenic mice were labelled with carboxy fluoroscein succinimidyl ester (CFSE) as described (29) and injected intraperitoneally into mice (1 × 107 cells per mouse, 3 mice per group). After 72 h, paragastric and inguinal lymph node cells from the recipient mice were harvested and single-cell suspensions were prepared and stained with CD90.1 and CD4 antibodies and analysed by flow cytometry as described above. Donor TCR transgenic A23 T cells were identified by expression of CD90.1 and CD4.
Two-tailed Fisher's exact test was used to compare the incidence of autoimmune gastritis or frequency of autoantibodies between different mouse strains. Comparisons giving P values equal to or below 0.05 were considered significant. Cell population data were analysed using a two-tailed Mann–Whitney U test.
Autoimmune gastritis in BALB.B6 congenic mice
Two genes conferring susceptibility to autoimmune gastritis were previously mapped in a linkage analysis of (BALB/cCrSlc × C57BL/6)F2 mice to the distal ∼40 Mb of mouse chromosome 4 (4). Here we verified our previous linkage data and narrowed the intervals containing Gasa1 and Gasa2 by the production and analysis of BALB.B6 congenic mice.
Congenic strains, in which BALB DNA was replaced by C57BL/6 DNA on distal chromosome 4, were constructed by intercrossing BALB/cCrSlc and C57BL/6J mice and performing 10 or 11 serial backcrosses to BALB/cCrSlc, while selecting breeders bearing C57BL-derived alleles at microsatellite loci linked to Gasa1 and Gasa2. At N11 or N12, heterozygous congenic mice were intercrossed and homozygous progeny selected to found inbred congenic strains. The genotypes of the resulting strains are illustrated in Fig. 2. BALB/cCrSlc and BALB.B6 congenic mice were thymectomized on the third day of life and autoimmune gastritis was analysed 12 weeks later by ELISA of serum for anti-H+/K+ ATPase autoantibodies and examination of haematoxylin- and eosin-stained sections of stomach. As in our previous studies (3, 4), only mice with gastric inflammation and serum autoantibodies were scored as autoimmune gastritic (Fig. 3). In the BALB.B6-GasaA strain, the entire BALB/cCrSlc region predicted to contain Gasa1 and Gasa2 is replaced by C57BL/6 DNA. The prevalence of autoimmune gastritis was significantly reduced in BALB.B6-GasaA mice relative to BALB/cCrSlc mice [5/34 (15%) versus 13/27 (48%), P < 0.01, two-tailed Fisher's exact test]. These data confirm the presence of autoimmune gastritis susceptibility loci on distal chromosome 4 of BALB/cCrSlc mice.
The BALB.B6-GasaB and BALB.B6-GasaC strains bear “proximally” truncated congenic intervals (Fig. 2). The incidence of autoimmune gastritis in these two strains was similar to the incidence in BALB/cCrSlc mice [not significant (NS), P > 0.5 for both] and significantly greater than the incidence in BALB.B6-GasaA mice (Fig. 3). The BALB.B6-GasaD strain bears a “distally” truncated congenic interval (Fig. 2). Again the incidence of autoimmune gastritis in this strain was similar to the incidence in BALB/cCrSlc (NS, P > 0.4) mice and significantly greater than the incidence in BALB.B6-GasaA mice (Fig. 3). The low incidence of autoimmune gastritis in BALB.B6-GasaA mice indicates that BALB/cCrSlc genomic region distal to D4Mit352 contains genes that confer susceptibility to this disease. In the region from D4Mit352 to the telomere, the BALB/cCrSlc-derived DNA in the BALB.B6-GasaB and BALB.B6-GasaC strains do not overlap with the BALB/cCrSlc-derived DNA in the BALB.B6-GasaD strain. As all of BALB.B6-GasaB, BALB.B6-GasaC and BALB.B6-GasaD strains have high incidences of autoimmune gastritis, this demonstrates that there are at least two BALB/cCrSlc genes that confer susceptibility to autoimmune gastritis on distal chromosome 4, which is in accordance with our previous data identifying Gasa1 and Gasa2 (4). Based on the genomic intervals in BALB.B6-GasaB and BALB.B6-GasaD strains, we conclude that the Gasa1 and Gasa2 genes lie between D4Mit352-D4Mit204 and D4Mit343-telomere, respectively (Fig. 2). These data also provide evidence of relative independence in conferring susceptibility to disease but dependence on C57BL/6 alleles at both loci for conferring resistance in BALB mice.
Presentation of a gastritogenic epitope is similar in BALB/cCrSlc and gastritis-resistant BALB.B6-GasaA congenic mice
The availability of a congenic strain that is resistant to autoimmune gastritis affords an opportunity to test whether changes to specific immune parameters and cells may be the basis for differences in incidence of disease. Here we examined presentation of an epitope targeted by gastritis-inducing T cells. T cells that cause autoimmune gastritis recognize the gastric H+/K+ ATPase (2, 26, 30–33). Presentation of the H+/K+ ATPase epitopes takes place in the lymph node that drains the stomach, the paragastric lymph node (27, 34). We examined the proliferation of transferred transgenic H+/K+ ATPase-specific T cells in the paragastric lymph node of the gastritis-resistant BALB.B6-GasaA mice as a measure of antigen presentation (Fig. 4). CD4+ T cells specific for the H+/K+ ATPase from A23 TCR transgenic mice (24) were labelled with CFSE and injected into BALB.B6-GasaA congenic mice or the parental BALB/cCrSlc mice. Three days later, CFSE levels in the transferred CD4+ cells present in the paragastric lymph nodes were analysed. Representative analyses and data from cohorts of mice are shown in Fig. 4. The percentage of A23 T cells that divided in the paragastric nodes of the BALB.B6-GasaA strain was comparable to that observed in BALB/cCrSlc mice. Furthermore, we found that proportions of cells that had undergone a given number of rounds of cell division were very similar in the two mouse strains (data not shown) indicating that the rate of mitosis of T cells was indistinguishable. These data suggest that the level of antigen presentation of the epitope recognized by the A23 T cells was similar in the two mouse strains.
Differences in regulatory T cells in BALB.B6-GasaA congenic mice are unlikely to be responsible for resistance to autoimmune gastritis
Autoimmune gastritis following day 3 thymectomy is considered to result from a depletion of CD4+CD25+Foxp3+ regulatory T cells (Treg) during the neonatal period (31, 35, 36). Hence, we considered the possibility that the BALB.B6-GasaA congenic mice may be resistant to autoimmune gastritis because the ontogeny of Treg differed in the congenic mice from that in BALB mice. We analysed the numbers of Treg in 3-day-old unmanipulated mice and 6-day-old mice that had been thymectomized at day 3 by staining cells derived from the thymus and spleen with anti-Foxp3 antibodies (Fig. 5a). A population of Foxp3+CD4+ cells was readily identified in both the thymus and spleen of BALB/cCrSlc and BALB.B6-GasaA mice at these ages. The proportions and number of Treg in the BALB/cCrSlc mice were slightly higher than in BALB.B6-GasaA mice in thymus and spleens of the 3-day-old mice, as well as in the spleens of the 6-day-old mice (Fig. 5b). Therefore, the congenic interval in the BALB.B6-GasaA mice contains genes that may influence the census of Treg in the thymus and periphery. However, the gastritis-resistant BALB.B6-GasaA mice had fewer Treg than BALB/cCrSlc mice, a difference that is unlikely to explain the resistance of the congenic strain to autoimmunity.
Autoimmune gastritis in NOD mice
The susceptibility of NOD/Lt mice to a range of organ-specific autoimmune diseases, and the proximity of the Gasa1 and Gasa2 genes to the Idd11 and Idd9 linkage regions, raised the possibility that the Gasa1 and Gasa2 allelic variants were the same as those encoding Idd11 and Idd9, respectively. In order to directly test this hypothesis, susceptibility to gastritis was determined for NOD/Lt mice and NOD.B6-Chr4, a NOD congenic strain bearing diabetes-resistance alleles of Idd9 and Idd11 (25). Twenty diabetic NOD/Lt mice between 99 and 200 days old, as well as age-matched non-diabetic mice, were analysed for the occurrence of spontaneous anti-gastric H+/K+ ATPase autoantibodies and gastric inflammation. No signs of gastric autoimmunity were present in either cohort (data not shown).
Susceptibility to gastritis was then determined in 12-week-old NOD/Lt mice following thymectomy on the third day after birth. NOD/Lt mice developed gastric inflammation and gastric autoantibodies after day 3 thymectomy. The pathological features of the inflammation observed in the gastric mucosa of NOD/Lt mice were similar to those seen in BALB/cJ mice (not shown). Inflammation varied from very mild mononuclear cell infiltrates close to the lamina propria, to more substantial mononuclear infiltration often localized to pockets in the mucosa or to very widespread mononuclear cell inflammation accompanied by depletion of differentiated epithelial cells and an expansion of immature epithelial cells.
We examined the specificity of antibodies present in sera of the thymectomized NOD/Lt and BALB/cJ mice. Immunofluorescent staining of sections of stomach tissue (Fig. 6a) demonstrated that the autoantibodies bound only to parietal cells in the gastric mucosa. The subcellular distribution of staining of the antibodies in NOD sera was very similar to that of the autoantibodies from BALB/c gastritic mice, which have previously been shown to bind to the secretory membrane structures of parietal cells (37). This pattern of staining, along with the fact that the antibodies reacted with purified gastric H+/K+ ATPase in ELISAs, is consistent with the antibodies binding exclusively to the gastric H+/K+ ATPase, which is the major protein constituent of the parietal cell secretory membranes. The H+/K+ ATPase is composed of two subunits, α and β. To determine which of these two subunits were bound by the antibodies in the sera, we performed immunofluorescent staining of cells expressing the subunits of the H+/K+ ATPase that were tagged with fluorescent proteins (Fig. 6b). Cells expressing either the tagged α or β subunits of the H+/K+ ATPase stained with the sera from the BALB/cJ and NOD mice with gastric autoimmunity. The autoantibody staining and the localization of both H+/K+ ATPase α and β subunit–fluorescent protein fusions overlapped almost perfectly, thus demonstrating that the autoimmune sera contained antibodies to both subunits. These histological and immunochemical data indicate that the immunopathological features of autoimmune gastritis in NOD/Lt and BALB/cJ mice are very similar.
The incidence of autoimmune gastritis, as defined by gastric inflammation accompanied by anti-H+/K+ ATPase antibodies, was not significantly different between BALB/cJ and NOD/Lt mice thymectomized contemporaneously [BALB/cJ, 18/41 (44%); NOD/Lt, 13/40 (33%); NS, P = 0.36; two-tailed Fisher's exact test]. However, comparisons of the semi-quantitative scores for degree of gastric inflammation revealed a significant difference between the strains (P < 0.005; two-tailed Mann–Whitney U test; Fig. 7). Of the BALB/cJ mice that developed autoimmune gastritis, 72% had the most severe forms of the disease (score 5 or 6, Fig. 7) that featured sub-mucosal and mucosal mononuclear infiltration, severe depletion of differentiated cell types and substantial hyperplasia due to the amplification of immature cell types. On the other hand, 62% (8/13) of NOD mice with gastritis were assigned a score of 1 or 2 with only sub-mucosal mononuclear cell infiltration apparent (Fig. 7). A score >4 was assigned to only one NOD/Lt mouse. Despite these differences in pathology scores, neither the frequency of anti-H+/K+ ATPase autoantibodies nor the titres of those antibodies, if present, was significantly lower in thymectomized NOD/Lt mice than in thymectomized BALB/cJ mice (data not shown).
Autoimmune gastritis in NOD.B6-Chr4 congenic mice
We also analysed autoimmune disease following day 3 thymectomy in NOD.B6-Chr4 mice that contain a B6 congenic interval spanning Idd9.1, Idd9.2, Idd9.3 and Idd11, and thus also the Gasa1 and Gasa2 loci (Fig. 1). In unmanipulated NOD.B6-Chr4 mice, a significant decrease in the incidence of diabetes was observed relative to NOD/Lt mice [72 versus 32% at 300 days; (25)]. After thymectomy at day 3, gastritis and gastric autoantibodies were observed in the NOD.B6-Chr4 mice. The histopathological features of gastritis in the NOD.B6-Chr4 mice was similar to that seen in the BALB/cJ and NOD/Lt mice (not shown) and furthermore the autoantibodies appeared to be directed to both subunits of the gastric H+/K+ ATPase (Fig. 7). The incidence of autoimmune gastritis in the NOD.B6-Chr4 mice was 9/46 (20%) compared with 13/40 (32%) NOD/Lt mice (NS, P = 0.22; two-tailed Fisher's exact test) (Fig. 8). Thus, while a C57BL/6-derived congenic segment spanning the Gasa1 and Gasa2 loci resulted in significant suppression of autoimmune gastritis on the BALB/cCrSlc genetic background, significant suppression was not observed with a similar segment on the NOD/Lt background, suggesting that the genetic basis for gastritis susceptibility may differ between these two strains.
We have chosen to utilize the well-characterized mouse model of autoimmune gastritis to identify the genes that predispose to this disease. Our previous work identified two regions on distal chromosome 4 of BALB/cCrSlc mice that are strongly linked to autoimmune gastritis (4). Here, we have confirmed these data by the production and analysis of BALB.B6-GasaA congenic mice in which the telomeric ∼40 Mb of the BALB/cCrSlc chromosome 4 was replaced with the corresponding region from the C57BL/6 genome. These mice were then thymectomized on the third day after birth and the prevalence and severity of autoimmune gastritis assessed 12 weeks later. The incidence of disease in this strain was 15%, significantly below the incidence in the parental BALB strain (48%). These data provide strong confirmatory evidence that the regions identified in our previous linkage studies contain genes that confer susceptibility to autoimmune gastritis.
Autoimmune gastritis still occurred in the BALB.B6-GasaA mice, albeit at a low incidence, indicating that other loci in the BALB genome also confer susceptibility to this disease, or that the C57BL/6 alleles are not completely protective. This is in accord with our previous findings. In our original linkage study, we found suggestive linkage to gastritis at regions on chromosomes 6, 9 and 15 (4). Subsequent analysis of these data using a partitioned χ2 analysis revealed significant linkage on chromosome 6 and this region was designated Gasa3 (3). The characterization of gastritis susceptibility in C57BL/6.D-H2 and BALB.B-H2, parental strains and F1 intercross mice identified the existence of an H2-linked gastritis susceptibility gene, which was designated Gasa4 (3). However, as heterozygosity at H2 and Gasa3 was found to confer the greatest susceptibility and BALB.B6-GasaA mice are homozygous at these loci, it is therefore possible that further minor gastritis susceptibility loci exist and have so far eluded detection.
To further localize and analyse the chromosome 4 Gasa genes, we constructed congenic mice that dissect the distal chromosome 4 Gasa region. We found that BALB congenic mouse strains containing C57BL/6-derived resistance alleles at either Gasa1 or Gasa2, but not both loci, had similar susceptibility to autoimmune gastritis as the BALB/cCrSlc parental strain. In other words, resistance was conferred only when the C57BL/6 alleles were present at both positions simultaneously. The positions of the congenic intervals in these strains allow us to draw the following conclusions. First, these data confirm the presence of at least two gastritis susceptibility genes in distal chromosome 4. Second, it also appears that Gasa1 and Gasa2 can act independently to confer gastric autoimmunity because BALB.B6-GasaB strain, harbouring only the BALB-derived Gasa1 susceptibility allele, and BALB.B6-GasaD, harbouring only the BALB-derived Gasa2 susceptibility allele, each had similar gastritis incidences compared with the parental BALB strain. Third, comparison of the intervals in these congenic strains allows us to narrow the regions that contain Gasa1 to D4Mit352 to D4Mit204 (∼18 Mb) and Gasa2 to D4Mit343 to the distal end of chromosome 4 (∼13 Mb; Fig. 2).
Redundancy in susceptibility alleles is frequently observed in complex genetic traits [discussed in (38)], including type 1 diabetes in NOD mice. For example, the individual contributions of linked diabetes loci in the Idd21.1 region (39) could only be identified in the presence of resistance alleles at other loci in the region. As a consequence, fine congenic mapping required systematic truncation of a large congenic interval and microcongenic strains bearing only a single resistance locus only weakly expressed a resistant phenotype (T. Merriman, personal communication).
It has been postulated that one of the reasons autoimmune diseases occur after neonatal thymectomy of mice is because this procedure restricts the thymic emigration of Treg (35). Our previous work and that of others has shown that the presence of Treg in the neonatal period is particularly important in preventing autoimmune disease (36, 40). In day 3 thymectomized adult mice, a deficit in Treg proportions is not apparent (41) presumably because Treg rapidly expand or are generated in the lymphopenic environment after thymectomy. It has also been shown that the number of Treg in the thymus varies between inbred strains (42) and that the number of thymic Treg in C57BL/6 mice is lower than other strains including BALB/c mice (42, 43). Therefore, we considered the possibility that the Gasa genes may influence the neonatal development of Treg. We found that the numbers and proportions of Treg in the thymi and spleens of 3-day-old BALB.B6-GasaA mice and spleens 6-day-old BALB.B6-GasaA mice that were thymectomized on day 3 were slightly lower than those of BALB/cCrSlc mice. The proportions of CD4+ cells that were Treg in thymi of 7-week-old mice were similar in both these strains (data not shown). These data suggest that genes in the B6 congenic interval of BALB.B6-GasaA mice may have a minor influence on proportions of Treg. It seems unlikely that this phenomenon is responsible for the relative resistance of the BALB.B6-GasaA mice to gastritis as this would likely involve a relative increase in number of Treg, not a decrease as was observed. It is possible that the activity of the Treg is altered in congenic strains, but due to the small number of T cells in these neonatal mice it is very difficult to test this possibility.
We also examined the activation and proliferation of T cells specific for the gastric H+/K+ ATPase, which is the autoantigen targeted in autoimmune gastritis (31). We found that there was no discernable difference in the behaviour of the T cells whether they were transferred to BALB/cCrSlc or BALB.B6-GasaA mice. This suggests that the ability of APCs in the two strains of mice to present gastric autoantigenic epitopes is not significantly different and therefore differences in this parameter are unlikely to explain the difference in susceptibility to autoimmunity.
Autoimmune diabetes in NOD mice is one of the most intensely studied animal models of human autoimmune disease. A great deal is known about the cellular and molecular events that underlie this disease. Diabetes in NOD mice and autoimmune gastritis following neonatal thymectomy are both tissue-specific inflammatory autoimmune diseases in which CD4+ T cells play a critical role. NOD mice are also susceptible to other autoimmune diseases such as haemolytic anaemia (44), sialiaditis (45), thyroiditis (46), experimental allergic encephalomyelitis (47) and lupus (48, 49). In recent work, NOD mice deficient in the aire protein were shown to develop gastritis (50). We found no evidence for anti-gastric autoantibodies or significant gastric inflammation in NOD/Lt mice up to 200 days of age, regardless of their diabetic status. In contrast, by 12 weeks of age, 33% of day 3 thymectomized NOD mice had developed significant gastric autoantibodies and gastric inflammation. Immunofluorescent staining of stomach sections and transfected cells indicated that autoantibodies in thymectomized NOD mice were directed to the gastric parietal cell proton pump, the H+/K+ ATPase. We found no other evidence for other autoantibodies reactive with other gastric proteins. This strongly suggests that in NOD mice as in BALB strains, the major gastric autoantigen is the H+/K+ ATPase. In addition, the histopathological features of the gastritis in NOD mice were very similar to those seen in BALB/c mice. These data indicate that autoimmune gastritis in NOD mice is very similar to that of BALB mice.
We have used NOD congenic strains to investigate the genetics of gastritis susceptibility in NOD mice. We day-3 thymectomized NOD.B6-Chr4 congenic mice that contained a C57BL/6-derived interval that spans distal chromosome 4 encompassing Gasa1 and Gasa2. The incidence of autoimmune gastritis in NOD.B6-Chr4 congenic mice was lower than observed in wild-type NOD/Lt mice (19.6 versus 32.5%). Unlike the situation with the BALB strains, this difference was not statistically significant even though the incidence of disease in the NOD.B6-Chr4 mice was comparable to that of the BALB.B6-GasaA mice (19.6 versus 14.7%). Significance was not reached because incidence of disease in the NOD/Lt mice was somewhat lower than in BALB/c mice. These data indicate that NOD Gasa1 and Gasa2 alleles are less effective, if at all, at conferring susceptibility to autoimmune gastritis suggesting that genetic aetiology of gastritis in NOD mice is different from that in BALB mice.
We have hypothesized that the genes on distal chromosome 4 that confer susceptibility to gastritis and diabetes may be identical. Our data here confirmed the presence of powerful gastritis susceptibility genes on distal chromosome 4 of BALB/c mice and that C57BL/6 alleles in this genomic region confer resistance to diabetes in NOD mice and autoimmune gastritis in BALB mice. However, data on the role of chromosome 4 alleles on the incidence of gastritis in NOD mice were inconclusive. While the analysis of larger cohorts of mice may help to clarify this issue, identification of the coding regions of the Gasa and Idd genes on distal chromosome 4 will be required to absolutely settle whether the gastritis and diabetes susceptibility genes on distal chromosome 4 are identical. In the case of the Gasa genes, the lower incidence of gastritis in the NOD strains will make the pursuit of these genes impractical in this strain background. Future efforts will concentrate on narrowing the location of Gasa1 and Gasa2 genes in the BALB.B6 congenic strains.
National Health and Medical Research Council of Australia (I.R.V.D. and A.G.B.), Juvenile Diabetes Research Foundation (T.C.B.), and US National Institutes of Health (T.C.B.). T.C.B. is supported by the Syme Fellowship.
CD4+CD25+Foxp3+ regulatory T cells
yellow fluorescent protein
The authors wish to thank Max Walker, Shiralee Whitehead, Fiona Quirk and Maya Kesar for technical assistance and animal care and Olga Vagin, Department of Physiology and Medicine, UCLA, Los Angeles, CA, USA, for providing the YFP-H+/K+ ATPase β subunit plasmid.