A novel type 1 diabetes locus was mapped to the interleukin-2 receptor α gene (IL2RA) on chromosome 10p15.1, encoding an important modulator of immunity. The aim of the current study was to confirm the association of IL2RA with type 1 diabetes and to attempt further mapping of the genetic effect with a new set of 12 single nucleotide polymorphisms (SNPs). We genotyped 949 nuclear family trios with one type 1 diabetes–affected offspring and two parents (2,847 individuals). Two of the 12 IL2RA SNPs genotyped (rs706778 and rs3118470) had statistically significant type 1 diabetes association (P = 6.96 × 10−4 and 8.63 × 10−4, respectively). Both SNPs are located in the 5′ end of the long intron 1 within 3 kb of each other and are in high linkage disequilibrium (D′ = 0.997, r2 = 0.613). The A-C haplotype (frequency = 0.331) was associated with increased type 1 diabetes risk (P = 3.02 × 10−4). Our study identifies two markers in the IL2RA gene that are significantly associated with type 1 diabetes, supporting IL2RA as a promising candidate gene for type 1 diabetes and suggesting a potential role of IL2Rα in the pathogenesis of type 1 diabetes, likely involving regulatory T-cells.
The interleukin (IL)-2 receptor-α gene (IL2RA) encodes the α chain of the IL-2 receptor complex (also known as CD25), an important modulator of immunity (1). IL-2 is a powerful growth factor for both T- and B-cells (2). It acts through a quaternary receptor signaling complex containing α, β, and a common γ (γc, shared by the family of IL receptors) chain (3). Recently, Vella et al. (4) identified a novel type 1 diabetes locus, mapping to IL2RA, whose eight exons extend over 48 kb on chromosome 10p15.1 (supplementary Fig. 1 [available in an online appendix at http://dx.doi.org/10.2337/db06-1555]). Multilocus analysis of data from 20 tag single nucleotide polymorphisms (SNPs) from the 3′ and 5′ ends of the gene was used in case-control and family-based population samples (4). The large sample sizes (7,457 case and control subjects and 725 multiplex families) substantiated the association with high statistical significance, but further work is needed on this locus. Even though odds ratios (ORs) were calculated for individual SNPs, Vella et al. (4) report the significance of the association with IL2RA only in terms of a multilocus analysis. In addition, analysis of the family-based data was not confined to heterozygous parents, which compromised the protection against population stratification (5).
The aim of the work reported here was to confirm this observation in a new set of 12 SNPs (including 5 tested in ref. 4) and attempt further fine mapping of the genetic effect.
RESEARCH DESIGN AND METHODS
Genomic DNA was obtained after informed consent was given, which was approved by the research ethics board of the Montréal Children's Hospital and other participating centers. Ethnic backgrounds were of mixed European descent, with the largest single group being of Québec French-Canadian origin (40% of the total collection). All patients were diagnosed under the age of 18 years and required continuous insulin treatment from the time of diagnosis. The sample consisted of 949 nuclear family trios with one type 1 diabetes–affected offspring and two parents (2,847 individuals).
SNP selection.
In a search for the functional variant, or a better marker for it, we selected 10 tag SNPs from the linkage disequilibrium (LD) blocks in which Vella et al. (4) found multilocus evidence of association. The data of HapMap public release no. 20 was used. Two additional SNPs (rs28360490 and rs706778) were generously suggested to us by C.E. Lowe and J.A. Todd (personal communication) based on their unpublished data. All 12 SNPs are contained in two LD blocks, which are in strong LD with each other (supplementary Fig. 1).
SNP genotyping.
We genotyped 10 SNPs by the Sequenom MassARRAY system and 2 (rs28360490 and rs706778) by an AcycloPrime-FP SNP detection kit (Perkin Elmer, Boston, MA). Rates for genotyping success and Mendelian error are shown in supplementary Table 1. In the 60 European HapMap samples (the parents in the CEU set) genotyped by both us and J.A. Todd's group for rs28360490 and rs706778, concordance was 100%. Our genotyping of rs706778 is also completely concordant with the HapMap data.
Statistics.
Transmission disequilibrium test and LD analysis were performed by Haploview software (www.broad.mit.edu/personal/jcbarret/haploview) (6). Haplotypic association was tested by family-based association test software (http://www.biostat.harvard.edu/∼fbat/fbat.htm) (7). To maintain a study-wide α level of 0.05 after testing 12 SNPs, we set the significance threshold for individual SNPs to 0.004, an overly conservative adjustment because, in the presence of LD, SNP genotypes are not independent. The OR was estimated by logistic regression, based on the methods described by Lohmueller et al. (8).
RESULTS
Two of the 12 IL2RA SNPs genotyped (rs706778 and rs3118470) had statistically significant type 1 diabetes association (Table 1). Both are located in the 5′ end of the long intron 1 within 3 kb of each other, are in high LD (D′ = 0.997, r2 = 0.613), and confer similar ORs (Table 1). They map to the second LD block, which encompasses the 5′ flanking region, the first exon, and the 5′ end of intron 1 of IL2RA.
Of the four possible haplotypes that the two significant SNPs can form, three accounted for 99% of chromosomes in our sample. The G-T haplotype (frequency = 0.551) was associated with a decreased (OR 0.81 [95% CI 0.71–0.92], P = 5.66 × 10−4) and the A-C haplotype (frequency = 0.331) with an increased (1.26 [1.10–1.46], P = 3.02 × 10−4) type 1 diabetes risk. There was no significant type 1 diabetes association with the A-T haplotype (frequency 0.116, P = 0.824), suggesting that the A-C haplotype is the marker for the genetic effect. However, this will need to be confirmed in additional studies because the difference in transmission frequencies between A-T and A-C did not reach statistical significance (P = 0.098).
DISCUSSION
Our study confirms the type 1 diabetes association of the IL2RA locus and identifies two markers for the effect. The biological mechanism of this association is still not apparent. Extensive resequencing of both case and control subjects by Vella et al. (4) revealed no nonsynonymous substitutions occurring at any frequency that could explain the association through protein sequence changes, and there are no known alternative splicing isoforms in which abundance could be affected by the intronic SNPs. In cultured lymphoblastoid cells lines, we found no association between the type 1 diabetes–predisposing haplotype and IL2RA mRNA expression levels (data not shown).
IL2RA maps within the chromosome 10p14-q11 linkage peak identified in the largest study to date (1,435 families) with the second highest logarithm of odds score outside the HLA region (9). It is unlikely that this peak is accounted for by the relatively weak effect we detected, and it may be due to aggregate effects from additional associated loci. However, the intriguing possibility that the IL2RA SNPs are merely markers in weak LD with a much stronger remote effect must be further pursued. As suggested by Vella et al. (4), IL15RA, which maps to a neighboring block in weak LD, is a distinct possibility.
What makes IL2RA a particularly interesting candidate for type 1 diabetes is the recent increase in understanding of the importance of regulatory T-cells in self-tolerance. CD25 is upregulated upon engagement of the T-cell receptor, but a small subset of CD4+ T-cells express it constitutively upon their maturation in the thymus. They constitute the population of CD4+CD25+ regulatory T-cells that are anergic to T-cell receptor signals and potently suppress activated T-cells in a contact-dependent and cytokine-independent fashion (10). This active immune suppression mechanism plays an important role in maintaining immune homeostasis and inhibiting autoimmune disease (11). As direct evidence of the importance of CD4+CD25+ T-cells in type 1 diabetes, the transfer of CD4+CD25+ T-cells can prevent diabetes in recipient NOD mice (12). CD4+CD25+ T-cells depend on IL-2 for their growth and survival, and the IL-2 signaling effect is mainly mediated by IL2Rα (1,13). The two SNPs in the IL2RA gene are significantly associated with type 1 diabetes, supporting IL2RA as a promising candidate gene for type 1 diabetes and suggesting a potential role of IL2Rα in the pathogenesis of type 1 diabetes, likely involving regulatory T-cells. In addition, considering the important role of IL-2 receptor-α in immune regulation, the association of IL2RA with other autoimmune diseases must also be examined.
SNP . | Minor allele (frequency) . | Hardy-Weinberg P . | Allele transmission counts (overtransmitted allele) . | χ2 (P) . | OR (95% CI) . |
---|---|---|---|---|---|
rs12722588 (A/G) | A (0.174) | 0.650 | 264:249 (G) | 0.4 (0.51) | 1.06 (0.89–1.26) |
rs28360490 (G/T) | G (0.103) | 0.451 | 158:143 (T) | 0.7 (0.39) | 1.10 (0.88–1.39) |
rs2274037 (A/G) | A (0.046) | 1.000 | 80:80 (—) | 0.0 (1.00) | 1.00 (0.73–1.36) |
rs2076846 (A/G) | G (0.337) | 0.139 | 364:363 (G) | 0.0 (0.97) | 1.00 (0.87–1.16) |
rs2025345 (A/G) | G (0.359) | 0.722 | 406:387 (A) | 0.5 (0.50) | 1.05 (0.91–1.21) |
rs12722563 (C/T) | T (0.083) | 0.073 | 141:115 (C) | 2.6 (0.10) | 1.23 (0.96–1.57) |
rs12722561 (A/G) | A (0.140) | 0.633 | 227:194 (G) | 2.6 (0.11) | 1.17 (0.97–1.42) |
rs706778 (A/G) | A (0.448) | 0.621 | 456:359 (A) | 11.5 (6.96 × 10−4) | 1.27 (1.11–1.46) |
rs3134883 (C/T) | T (0.308) | 0.912 | 410:347 (T) | 5.2 (0.02) | 1.18 (1.02–1.36) |
rs3118470 (C/T) | C (0.337) | 0.192 | 426:334 (C) | 11.1 (8.63 × 10−4) | 1.28 (1.10–1.47) |
rs12722486 (A/G) | A (0.062) | 0.698 | 103:100 (G) | 0.0 (0.83) | 1.03 (0.78–1.36) |
rs4147359 (A/G) | A (0.368) | 1.000 | 445:373 (A) | 6.3 (0.01) | 1.19 (1.04–1.37) |
SNP . | Minor allele (frequency) . | Hardy-Weinberg P . | Allele transmission counts (overtransmitted allele) . | χ2 (P) . | OR (95% CI) . |
---|---|---|---|---|---|
rs12722588 (A/G) | A (0.174) | 0.650 | 264:249 (G) | 0.4 (0.51) | 1.06 (0.89–1.26) |
rs28360490 (G/T) | G (0.103) | 0.451 | 158:143 (T) | 0.7 (0.39) | 1.10 (0.88–1.39) |
rs2274037 (A/G) | A (0.046) | 1.000 | 80:80 (—) | 0.0 (1.00) | 1.00 (0.73–1.36) |
rs2076846 (A/G) | G (0.337) | 0.139 | 364:363 (G) | 0.0 (0.97) | 1.00 (0.87–1.16) |
rs2025345 (A/G) | G (0.359) | 0.722 | 406:387 (A) | 0.5 (0.50) | 1.05 (0.91–1.21) |
rs12722563 (C/T) | T (0.083) | 0.073 | 141:115 (C) | 2.6 (0.10) | 1.23 (0.96–1.57) |
rs12722561 (A/G) | A (0.140) | 0.633 | 227:194 (G) | 2.6 (0.11) | 1.17 (0.97–1.42) |
rs706778 (A/G) | A (0.448) | 0.621 | 456:359 (A) | 11.5 (6.96 × 10−4) | 1.27 (1.11–1.46) |
rs3134883 (C/T) | T (0.308) | 0.912 | 410:347 (T) | 5.2 (0.02) | 1.18 (1.02–1.36) |
rs3118470 (C/T) | C (0.337) | 0.192 | 426:334 (C) | 11.1 (8.63 × 10−4) | 1.28 (1.10–1.47) |
rs12722486 (A/G) | A (0.062) | 0.698 | 103:100 (G) | 0.0 (0.83) | 1.03 (0.78–1.36) |
rs4147359 (A/G) | A (0.368) | 1.000 | 445:373 (A) | 6.3 (0.01) | 1.19 (1.04–1.37) |
Additional information for this article can be found in an online appendix at http://dx.doi.org/10.2337/db06-1555.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Article Information
This work was funded by the Juvenile Diabetes Research Foundation and Genome Canada through the Ontario Genomics Institute. H.-Q.Q. is supported by a fellowship from the Canadian Institutes of Health Research. T.J.H. is supported by a clinician-scientist award in translational research by the Burroughs Wellcome Fund and an investigator award from the Canadian Institutes of Health Research.
We thank all families who participated in this study and Diane Laforte and her team for their recruiting efforts.