Type 1 diabetes mellitus, and autoimmune thyroid disease (AITD) (Graves' disease and Hashimoto's thyroiditis) are autoimmune disorders with a complex genetic control that are believed to develop through a process mediated by T lymphocytes.1 In addition to a strong contribution to disease susceptibility by the HLA locus2,3,4 each of these diseases have been reported to be associated and/or linked to the CTLA4 locus, and the co-localised IDDM12 type 1 diabetes mellitus susceptibility locus on chromosome 2q33.5,6,7,8 Whether the same allele, different alleles in the same locus, or different loci in strong linkage disequilibrium mediate this increase in risk, remains unclear. To address this question, it is informative to study large pedigrees, in which the diseases are concurrent and that contain members that are affected with one of these autoimmune diseases.

Materials and methods

The identification of a large 5-generation family from northern Sweden with multiple members affected by type 1 diabetes mellitus or AITD (Figure 1) allowed us to investigate the common etiology hypothesis. Family members were coded with a binary phenotype representing the presence or absence of autoimmune disease; blood from 72 family members was collected and analysed. Diagnosis of type 1 diabetes mellitus was according to the criteria of the National Diabetes Data Group.9 Within the family there are four patients with type 1 diabetes mellitus, seven patients with HT and four with GD. None of the patients were affected with both type 1 diabetes mellitus and AITD. Age of onset was 39.0±4.5 years (mean±SD) for type 1 diabetes mellitus, 43.6±9.0 for GD and 49.8±9.4 for HT. Sera from the patients were tested for anti-GAD autoantibodies, but were found to be negative. It should be noted, however, that all patients were tested several years after onset of disease. The study was carried out with the approval of the regional ethical review board and after informed consent of the individuals included in the study.

Figure 1
figure 1

Pedigree of a five-generation family from northern Sweden. Cases of type 1 diabetes mellitus (filled symbols), GD (filled upper right quadrant), or HT (filled lower left quadrant). Alleles corresponding to markers D2S2214, D2S307, D2S72, D2S2189 and D2S2237 are shown from top to bottom. Recombination events in individuals IV:9 and IV:16 restrict the disease critical region to a 2.4 Mb interval flanked by D2S2214 and D2S1384. A star beneath the individual indicates presence of the disease associated HLA haplotype. Arrow indicates proband. Only children to patients with the disease haplotype are shown.

Genomic DNA was prepared from leucocytes using standard methods.10 Using a candidate region approach the family members were genotyped for marker loci in the HLA region on chromosome 6p21 and in the CTLA4 region on chromosome 2q33. Two-point parametric linkage analysis was carried out using a model of dominant inheritance with 90% penetrance. An alternative disease model was also used assuming autosomal dominant inheritance with an age-dependent penetrance. Using this model, however, did not alter the maximum lodscores obtained more than marginally. Although previous epidemiological studies have reported disease prevalence for type 1 diabetes mellitus of 0.4% and of 1% for AITD,13,14 the familial form of type 1 diabetes mellitus /AITD studied here is likely to be very rare and the disease allele frequency was therefore specified as 0.001.


The marker loci used included known polymorphisms in CTLA4 as well as several microsatellite markers in the region (see Figure 2B). Using this approach, a positive maximum two-point lodscore of 0.72 at θ=0.25, 0.89 at θ=0.30 and 0.74 at θ=0.16 was obtained over the marker loci D2S2392, D2S309, and D2S72, respectively. These data indicated linkage to a susceptibility locus in this region (Figure 2C).

Figure 2
figure 2

Map of the chromosome 2q33 region. (A) CTLA4. Polymorhpisms identified in the disease associated allele of CTLA4. Alleles represented in the disease-associated haplotype are highlighted. (B) Critical markers used for linkage analysis and the disease critical region is indicated. Table shows maximum two-point and multipoint lodscore for markers in the 2q33 region.

Next, HLA-allele determination was performed for all collected subjects. We noted that all of the affected members of the family were found to carry at least one copy of the previously identified, major HLA susceptibility haplotype contributing to type 1 diabetes mellitus in northern Sweden (DR4-DQA1*0301-DQB1*0302).15 This haplotype has a high frequency in this population and 23% of the unaffected family members had one or more copies of it. It is therefore relatively uninformative and, indeed, no significant positive lodscore was obtained when the HLA locus was analysed independently for linkage to the disease (data not shown). In contrast, however, when the family members were analysed for linkage to markers in the CTLA4 region, using the same model parameters as specified for the two point analysis discussed above but conditioning with respect to HLA DR4-DQA1*0301-DQB1*0302 haplotype, a maximum lodscore of 2.69 at θ=0.0 was obtained over the D2S2392 marker locus (Figure 2C). Stratifying on the basis of HLA haplotype, all family members not carrying one or more copies of the HLA DR4-DQA1*0301-DQB1*0302 haplotype were set as having an ‘unknown’ disease status. This method of conditioning for HLA haplotype has proven useful in other studies of complex diseases allowing the identification of additional susceptibility loci.16

Multipoint analysis using the markers CTLA4(AT)n, D2S72 and D2S297, and using LINKMAP (part of the LINKAGE package)11 and VITESSE12 provided a maximum lodscore of 3.27 close to the marker D2S72 when conditioning for HLA.

To identify the disease critical region on chromosome 2q33, affected family members were submitted to haplotype analysis. Minimising the number of recombination events, a 2.4 Mb region restricted by the markers D2S2214 and D2S1384 could be defined as the region of interest (Figure 2B). Interestingly, one of the recombinations identified allows us to exclude a cluster of apoptosis-related genes previously considered as possible candidates for the IDDM12 susceptibility locus including CASP8, CASP10 and FLIP. Using additional marker loci previously identified in this region we constructed extended haplotypes between these markers and used these for reanalysis of linkage. Using the model parameters specified above and conditioning on HLA haplotype, this approach provided a maximum two-point lodscore of 4.20 at θ=0.0. Without conditioning for HLA, the corresponding maximum lodscore obtained was 1.45 at θ=0.23. Together these analyses provide strong evidence in favour of linkage.

Scrutiny of the disease critical region revealed the presence of 46 known or predicted genes including CTLA4, CD28 and ICOS. Particular interest has been paid to CTLA4 as a candidate for mediating susceptibility to autoimmune disease due to its role as a regulator of T lymphocyte activity.17,18 To search for etiological mutations in the CTLA4 gene, we sequenced genomic DNA from affected and non-affected members of the family (Figure 2A). The disease-associated haplotype identified in the family, was found to carry a combination of three previously identified disease-associated alleles (the G allele in the A/G polymorphism at position 49 in exon 1, the T allele in the C/T polymorphism in position 1822 in the intron between exon 1 and 2, and the 106 bp allele in the CTLA4 (AT)n polymorphism in the 3′ UT region).5,6,7,8,19,20,21,22


The observed linkage of susceptibility factors for type 1 diabetes mellitus/AITD to HLA and to the CTLA4 region confirms previous reports of linkage/association of type 1 diabetes mellitus, GD and HT to these loci. Moreover, it provides evidence that susceptibility to each of these diseases can be mediated by the same etiologic factor. Thus, it is highly unlikely that more than one allelic form of the same gene or genes, in strong linkage disequilibrium, is mediating susceptibility to each of the different diseases in this family.

Two of the polymorphisms found in the CTLA4 gene, the 49A/G in exon 1 and the AT repeat region located in the 3′ UT region, have been suggested to contribute functionally to disease.5,6,23 We find it unlikely, however, that these polymorphisms would constitute the etiological mutations explaining the dominant inheritance pattern and high penetrance observed in this family. On the other hand, accumulation of risk factors, that each contribute only a limited risk increase, as well as extended SNP haplotypes affecting gene expression have previously been suggested to contribute susceptibility to other complex diseases.24 Additional polymorphisms of this type could, in principle, be present also in several different candidate genes in strong linkage disequilibrium. Recent findings in the NOD mouse, where the susceptibility loci Idd5 and Idd9 have been shown to be constituted by several, closely linked susceptibility genes, illustrate such a situation.25,26 Thus, susceptibility factors in familial forms of complex genetic diseases such as type 1 diabetes mellitus and AITD may be constituted by the combined effect of minor risk factors accumulating in a risk haplotype of the kind identified here. It should be noted, however, that additional genetic factors located in the disease critical region have not been excluded and further analysis of additional candidate genes in the region are ongoing to address this possibility.