The mechanism for the association of type 1 diabetes (T1D) with IL2RA remains to be clarified. Neither of the two distinct, transmission-disequilibrium confirmed loci mapping to this gene can be explained by a coding variant. An effect on the levels of the soluble protein product sIL-2RA has been reported but its cause and relationship to disease risk is not clear. To look for an allelic effect on IL2RA transcription in cis, we examined RNA from 48 heterozygous lymphocyte samples for differential allele expression. Of the 48 samples, 32 showed statistically significant allelic imbalance. No known single nucleotide polymorphism (SNP) had perfect correlation with this transcriptional effect but the one that showed the most significant (p = 1.6 × 10−5) linkage disequilibrium with it was the SNP rs3118470. We had previously shown rs3118470 to confer T1D susceptibility in a Canadian dataset, independently of rs41295061 as the major reported locus (p = 5 × 10−3, after accounting for rs41295061 by conditional regression). Lower IL2RA levels consistently originated from the T1D predisposing allele. We conclude that an as yet unidentified variant or haplotype, best marked by rs3118470, is responsible for this independent effect and increases T1D risk through diminished expression of the IL-2R, likely by interfering with the proper development of regulatory T cells.

One of the strongest known genetic associations of type 1 diabetes (T1D)3 maps to the IL-2R α-chain subunit gene IL2RA, encoding CD25 (1, 2). Adult CD25-deficient mice are characterized by impaired T cell death after activation and autoimmune disorders (2). In humans, genetic association of noncoding polymorphisms at IL2RA with T1D has been reported (3, 4) and replicated independently (5). Genetic variation of IL2RA has also been reported to be associated with other autoimmune diseases, e.g., multiple sclerosis (6, 7, 8) and juvenile idiopathic arthritis (9).

The IL2RA gene contains no common nonsynonymous single nucleotide polymorphism (SNP) of frequency (in cases or in controls) sufficient to explain the association (10). Fine-mapping studies (4, 7) have suggested as many as three independent effects at that locus, in one of which the same allele has opposite effects in T1D vs multiple sclerosis (7). However, this last effect was not replicated in stratification-proof family studies (see Ref. 7 plus an additional 5,003 cases and their parents from the T1D Genetics Consortium, Qu et al., accepted for publication) and likely represents a stratification artifact. The predisposing allele at both the remaining two loci is associated with lower circulating soluble IL-2RA concentrations (4, 7), a biochemical phenotype that may be upstream or downstream of T1D autoimmunity (i.e., cause or consequence). The latter is suggested by the heterogeneity between T1D cases and controls at rs41295061 (7): only T1D cases are significant, a discrepancy not due to lack of statistical power, as joint analysis of cases and controls loses significance (p > 0.05) despite the larger sample size. Furthermore, for both the C/C and C/A genotypes, the T1D cases have higher soluble IL-2RA than the controls (p < 10−6). In addition, little is known about the biological function of soluble IL-2RA or how well its level reflects that of the functional membrane receptor. A study of allelic effects at the transcriptional level will help clarify these issues.

Recently we showed differential allelic expression (AE) of IL2RA in human lymphoblastoid cell lines (LCLs) (11). In this study, we fine-mapped the cis-regulatory effect, and found that it accounts for at least one of the two distinct genetic effects.

To look for a transcriptional effect in cis, we determined the relative expression of each allele at gene transcription levels in 55 LCLs of the CEU set (Caucasians of European-descent individuals used in HapMap, www.hapmap.org). Because of the lack of exonic SNPs common enough to permit statistical analysis, we selected four intronic markers (rs2104286 (C/T genotype), rs3134883 (G/A), rs11598648 (A/G), and rs12722574 (A/G)) for AE measurements of unspliced heteronuclear RNA. A total of 40 of the 55 parental LCLs were informative (heterozygous) at one or more sites and yielded AE data. The ratio of the peaks corresponding to each allele, after correction for the amplitude of adjacent peaks, was compared with the corresponding ratio in genomic DNA (taken as the 1:1 control). The normalized heterozygote ratio was log10 transformed. More details of the assay can be found in our previous reports (11, 12).

In the AE assay, each sample was quantified twice from two RT-PCR. Replicable unequal AE that exceeded ±2 SD of the allele ratio distribution in genomic DNA in both assays was taken as heterozygous for the cis-effect. Because the HapMap individual data are phased, the allelic imbalance phenotype (increased or decreased expression) could be mapped to specific haplotypes in the region. Based on the known haplotypes phased with AE, the genotypes of individuals with no differential AE were inferred by the hidden Markov Chain-based algorithm using the MACH software (http://www.sph.umich.edu/csg/abecasis/MaCH/index.html). Heterozygote calling of an individual with no differential AE was set as genotype undetermined, which removed potential false negative for the cis-effect. A homozygous genotype (high/high or low/low) was called for 7 of 14 individuals with no differential AE, which was based on our assumption that individuals with no differential AE were homozygous for the cis-effect, i.e., homozygote for either the higher or the lower expressed allele. The average estimated imputation certainty is 0.847 ± 0.158 (x̄ ± SD). The linkage disequilibrium between cis-effect on AE and SNPs in the 500-kb region surrounding the IL2RA gene was tested based on the HapMap release 24 data. Linkage disequilibrium of the variant determining AE with each SNP was evaluated.

To replicate the AE effect, AE was also measured in Con A-stimulated PBMC. PBMCs were sampled from 8 individuals of the Canadian T1D families known to be heterozygous for the two intronic markers rs706778 and rs3134883. Freshly isolated PBMCs in 5% FCS culture medium were treated with Con A at 5 μg/ml at 37°C for 72 h before being harvested for RNA isolation. RT-PCRs were performed in triplicates. The AE ratio was determined as for the LCLs.

The SNP rs41295061 (ss52580101) was genotyped by the AcycloPrime-FP SNP Detection kit (PerkinElmer) in the same dataset of 949 European-descent case-parent trios from Canada as reported before (5). Call rate was 100% and the Mendelian error rate was 0.1%. The T1D association of rs41295061 was tested by the transmission disequilibrium test. To test whether rs41295061 and the previous SNPs rs706778 and rs3118470 have independent effects on T1D susceptibility, we used the CETDT (conditional extended transmission disequilibrium test) method (13), implemented in the UNPHASED software package (http://www.mrc-bsu.cam.ac.uk/personal/frank/software/unphased/) (14). The CETDT method is based on the haplotype transmission test. If there is an effect at a second marker conditional on the first marker in case-parent trios, a significant difference in transmission of haplotypes identical at the first marker, but different at the second marker locus will be identified (13). The Research Ethics Board of the Montreal Children’s Hospital approved the study, and written informed consent was obtained from all subjects.

Of the 40 HapMap LCLs that were heterozygous for at least one of the intronic SNPs, 26 had unequal AE that exceeded ±2 SD of the allele ratio distribution in genomic DNA (Table I). For the individuals with unequal AE, the average log10(AE) is 0.259 ± 0.205 (x̄ ± SD) for the c1 higher individuals (corresponding to allele ratio = 1.815), which corresponds to an allele ratio of 64.5%/35.5%; and −0.347 ± 0.268 for the c1 lower individuals (corresponding to allele ratio of 31.0%/69.0%). The origin of this cis-effect was fine-mapped by linkage disequilibrium analysis of the 500-kb region surrounding the IL2RA gene. Outside this region, no linkage disequilibrium signal with the AE was detected. Peak linkage disequilibrium with the transcriptional effect mapped to two SNPs in proximity to each other and in tight linkage disequilibrium (D′ = 1, r2 = 0.921), rs3134883 (r2 = 0.305 with the transcriptional effect), and rs3118470 (r2 = 0.232) (Fig. 1). The two values for the correlation coefficient r2 have no statistical difference (p = 0.619). Linkage disequilibrium with AE was also detected from the other T1D-associated SNPs as we reported previously (5). The predisposing allele of each of the T1D-associated SNPs is in linkage disequilibrium with the underexpression phenotype of the IL2RA heteronuclear RNA. By contrast, the SNPs correlated with circulating soluble IL-2RA concentrations (4, 7), i.e., rs41295061, rs11594656, and rs2104286, showed no significant linkage disequilibrium with AE (Table II). Thus, this biochemical phenotype remains unexplained.

Table I.

IL2RA AE measurements in LCL samplesa

Sample ID No.bHaplotypecAE RatiodAErs3118470rs41295061
c1c2rs2104286rs3134883rs11598648rs12722574Averagec1c2c1c2c1c2
GM12156 TGGG TAGG  0.279 (0.655/0.345)   0.279 (0.655/0.345) − 
GM12145 TGGG TAAG  0.801 (0.863/0.137) 0.308 (0.670/0.330)  0.555 (0.782/0.218) − 
GM12239 CGAG CGGA   0.208 (0.618/0.382) 0.310 (0.671/0.329) 0.259 (0.645/0.355) − 
GM11840 TGGG TAAG  0.613 (0.804/0.196) 0.960 (0.901/0.099)  0.787 (0.86/0.14) − 
GM11832 CGGA TGGG 0.116 (0.566/0.434)   −0.027 (0.485/0.515) 0.045 (0.526/0.474) − 
GM12043 TGAG TAAG  0.266 (0.649/0.351)   0.266 (0.649/0.351) − 
GM11992 TGAA TAAG  0.073 (0.542/0.458)  0.321 (0.677/0.323) 0.197 (0.611/0.389) − 
GM11993 TGGG TAAA  0.220 (0.624/0.376) 0.180 (0.602/0.398) −0.301 (0.333/0.667) 0.033 (0.519/0.481) − 
GM12264 TGAA TAGG  0.162 (0.592/0.408) 0.131 (0.575/0.425) 0.163 (0.593/0.407) 0.152 (0.587/0.413) − 
GM12234 TGGG TAAG  0.110 (0.563/0.437) 0.225 (0.627/0.373)  0.167 (0.595/0.405) − 
GM12751 TGGG CGGA 0.126 (0.572/0.428)    0.126 (0.572/0.428) − 
GM12763 TGGG TAAG  0.140 (0.580/0.420) 0.076 (0.544/0.456)  0.108 (0.562/0.438) − 
GM12872 CGAG CGGA   0.217 (0.622/0.378) 0.264 (0.648/0.352) 0.240 (0.635/0.365) − 
GM12873 TAGG TAAG   0.516 (0.766/0.234)  0.516 (0.766/0.234) − 
GM07055 TGGG TAAG  0.215 (0.622/0.378) −0.045 (0.474/0.526)  0.085 (0.549/0.451) − 
GM12891 TAAG CGGA 0.299 (0.666/0.334) 0.264 (0.647/0.353) 0.182 (0.603/0.397) 0.561 (0.785/0.215) 0.327 (0.68/0.32) − 
GM07022 TAAG TGGG  −0.207 (0.383/0.617) −0.264 (0.352/0.648)  −0.236 (0.368/0.632) − 
GM07000 TAGG TGGG  −0.696 (0.168/0.832)   −0.696 (0.168/0.832) − 
GM07034 TGAA CGAG −0.189 (0.393/0.607)   −1.523 (0.029/0.971) −0.856 (0.122/0.878) − 
GM12155 TGGG CGGA −0.178 (0.399/0.601)   −0.336 (0.316/0.684) −0.257 (0.356/0.644) − 
GM12006 CGGA TGAA −1.122 (0.070/0.930)  −0.065 (0.463/0.537)  −0.594 (0.203/0.797) − 
GM11831 TAGA TGAA  −0.095 (0.446/0.554) −0.136 (0.422/0.578)  −0.116 (0.434/0.566) − 
GM11881 TGGG CGGA −0.051 (0.471/0.529)   −0.242 (0.364/0.636) −0.146 (0.417/0.583) − 
GM11994 TAAG TGGG  −0.108 (0.438/0.562) −0.199 (0.387/0.613)  −0.153 (0.413/0.587) − 
GM12760 TAAA CGGA −0.038 (0.478/0.522) −0.116 (0.434/0.566) −0.200 (0.387/0.613)  −0.118 (0.433/0.567) − 
GM12813 TAAG CGGA −0.312 (0.328/0.672) −0.284 (0.342/0.658) −0.047 (0.473/0.527) −0.536 (0.225/0.775) −0.295 (0.337/0.663) − 
GM06993 TGGG TAGG     
GM06985 TGAG TGGG     
GM07056 CGAG TGGG    
GM12249 TGGG CGGA    
GM12004 TGGG TAAG    
GM12144 TAGG CGGA   
GM12057 TGGG TGAG     
GM11829 TGGG TAGG     
GM11830 TGGG TAAG    
GM11995 TGAG TGGG     
GM12750 TGGG CGAG    
GM12762 CGGA TGGG    
GM12815 TAGG TGGG     
GM12892 TGGG CGAG    
gDNA ratio x ± se 0.000 ± 0.003 (0.500/0.500) 0.000 ± 0.020 (0.500/0.500) −0.003 ± 0.045 (0.498/0.502) −0.002 ± 0.052 (0.499/0.501)        
Sample ID No.bHaplotypecAE RatiodAErs3118470rs41295061
c1c2rs2104286rs3134883rs11598648rs12722574Averagec1c2c1c2c1c2
GM12156 TGGG TAGG  0.279 (0.655/0.345)   0.279 (0.655/0.345) − 
GM12145 TGGG TAAG  0.801 (0.863/0.137) 0.308 (0.670/0.330)  0.555 (0.782/0.218) − 
GM12239 CGAG CGGA   0.208 (0.618/0.382) 0.310 (0.671/0.329) 0.259 (0.645/0.355) − 
GM11840 TGGG TAAG  0.613 (0.804/0.196) 0.960 (0.901/0.099)  0.787 (0.86/0.14) − 
GM11832 CGGA TGGG 0.116 (0.566/0.434)   −0.027 (0.485/0.515) 0.045 (0.526/0.474) − 
GM12043 TGAG TAAG  0.266 (0.649/0.351)   0.266 (0.649/0.351) − 
GM11992 TGAA TAAG  0.073 (0.542/0.458)  0.321 (0.677/0.323) 0.197 (0.611/0.389) − 
GM11993 TGGG TAAA  0.220 (0.624/0.376) 0.180 (0.602/0.398) −0.301 (0.333/0.667) 0.033 (0.519/0.481) − 
GM12264 TGAA TAGG  0.162 (0.592/0.408) 0.131 (0.575/0.425) 0.163 (0.593/0.407) 0.152 (0.587/0.413) − 
GM12234 TGGG TAAG  0.110 (0.563/0.437) 0.225 (0.627/0.373)  0.167 (0.595/0.405) − 
GM12751 TGGG CGGA 0.126 (0.572/0.428)    0.126 (0.572/0.428) − 
GM12763 TGGG TAAG  0.140 (0.580/0.420) 0.076 (0.544/0.456)  0.108 (0.562/0.438) − 
GM12872 CGAG CGGA   0.217 (0.622/0.378) 0.264 (0.648/0.352) 0.240 (0.635/0.365) − 
GM12873 TAGG TAAG   0.516 (0.766/0.234)  0.516 (0.766/0.234) − 
GM07055 TGGG TAAG  0.215 (0.622/0.378) −0.045 (0.474/0.526)  0.085 (0.549/0.451) − 
GM12891 TAAG CGGA 0.299 (0.666/0.334) 0.264 (0.647/0.353) 0.182 (0.603/0.397) 0.561 (0.785/0.215) 0.327 (0.68/0.32) − 
GM07022 TAAG TGGG  −0.207 (0.383/0.617) −0.264 (0.352/0.648)  −0.236 (0.368/0.632) − 
GM07000 TAGG TGGG  −0.696 (0.168/0.832)   −0.696 (0.168/0.832) − 
GM07034 TGAA CGAG −0.189 (0.393/0.607)   −1.523 (0.029/0.971) −0.856 (0.122/0.878) − 
GM12155 TGGG CGGA −0.178 (0.399/0.601)   −0.336 (0.316/0.684) −0.257 (0.356/0.644) − 
GM12006 CGGA TGAA −1.122 (0.070/0.930)  −0.065 (0.463/0.537)  −0.594 (0.203/0.797) − 
GM11831 TAGA TGAA  −0.095 (0.446/0.554) −0.136 (0.422/0.578)  −0.116 (0.434/0.566) − 
GM11881 TGGG CGGA −0.051 (0.471/0.529)   −0.242 (0.364/0.636) −0.146 (0.417/0.583) − 
GM11994 TAAG TGGG  −0.108 (0.438/0.562) −0.199 (0.387/0.613)  −0.153 (0.413/0.587) − 
GM12760 TAAA CGGA −0.038 (0.478/0.522) −0.116 (0.434/0.566) −0.200 (0.387/0.613)  −0.118 (0.433/0.567) − 
GM12813 TAAG CGGA −0.312 (0.328/0.672) −0.284 (0.342/0.658) −0.047 (0.473/0.527) −0.536 (0.225/0.775) −0.295 (0.337/0.663) − 
GM06993 TGGG TAGG     
GM06985 TGAG TGGG     
GM07056 CGAG TGGG    
GM12249 TGGG CGGA    
GM12004 TGGG TAAG    
GM12144 TAGG CGGA   
GM12057 TGGG TGAG     
GM11829 TGGG TAGG     
GM11830 TGGG TAAG    
GM11995 TGAG TGGG     
GM12750 TGGG CGAG    
GM12762 CGGA TGGG    
GM12815 TAGG TGGG     
GM12892 TGGG CGAG    
gDNA ratio x ± se 0.000 ± 0.003 (0.500/0.500) 0.000 ± 0.020 (0.500/0.500) −0.003 ± 0.045 (0.498/0.502) −0.002 ± 0.052 (0.499/0.501)        
a

Relative expression of each allele at gene transcription levels in LCL samples from the HapMap European descent set. The “x” represents the sample that was not called of unequal AE, i.e. replicable unequal AE that exceeded ±2 SD of the allele ratio distribution in genomic DNA in two AE assays.

b

Each cell line corresponds to a unique HapMap individual.

c

The haplotype of the four SNPs, rs2104286, rs3134883, rs11598648, and rs12722574. Each individual has two haplotypes, and each haplotype is comprised of a specific allele of each SNP.

d

The average is log10(c1/c2).

e

Genomic DNA (gDNA) was used as control for equal representation of alleles.

FIGURE 1.

Fine-mapping of linkage disequilibrium with AE. The genotypes of the HapMap individuals are based on the HapMap release 24 data. The AE fine-mapping (top). Results indicate r2 values (red). The linkage disequilibrium peak with AE is from rs3134883 (r2 = 0.305) and rs3118470 (r2 = 0.232) at the IL2RA 5′ end. In the 500-kb region surrounding the IL2RA gene, no other r2 value was found that approached that peak. Results show the linkage disequilibrium map (bottom) with the standard linkage disequilibrium color scheme made by Haploview (v.4.0 software; www.broad.mit.edu/personal/jcbarret/haploview). Low to high D′ values are shown as a percentage represented by the depth of color (red); high D′ with low confidence (blue) also represented.

FIGURE 1.

Fine-mapping of linkage disequilibrium with AE. The genotypes of the HapMap individuals are based on the HapMap release 24 data. The AE fine-mapping (top). Results indicate r2 values (red). The linkage disequilibrium peak with AE is from rs3134883 (r2 = 0.305) and rs3118470 (r2 = 0.232) at the IL2RA 5′ end. In the 500-kb region surrounding the IL2RA gene, no other r2 value was found that approached that peak. Results show the linkage disequilibrium map (bottom) with the standard linkage disequilibrium color scheme made by Haploview (v.4.0 software; www.broad.mit.edu/personal/jcbarret/haploview). Low to high D′ values are shown as a percentage represented by the depth of color (red); high D′ with low confidence (blue) also represented.

Close modal
Table II.

The linkage disequilibrium with AE of the 5′ and upstream region of IL2RA

MarkerPositionAllele (Frequency)R2 Valuep Value
rs706778 6,138,954 T (0.450) 0.074 0.012 
rs2104286 6,139,050 G (0.233) 0.014 0.337 
rs3134883 6,140,730 T (0.275) 0.305 1.56 × 10−5 
rs3118470 6,141,718 C (0.292) 0.232 2.39 × 10−4 
rs12722486 6,143,767 A (0.100) 0.021 0.603 
rs4147359 6,148,444 A (0.333) 0.148 2.39 × 10−4 
rs41295061 (ss52580101) 6,154,665 A (0.067) 0.068 0.234 
rs11594656 6,162,014 A (0.225) 0.038 0.124 
MarkerPositionAllele (Frequency)R2 Valuep Value
rs706778 6,138,954 T (0.450) 0.074 0.012 
rs2104286 6,139,050 G (0.233) 0.014 0.337 
rs3134883 6,140,730 T (0.275) 0.305 1.56 × 10−5 
rs3118470 6,141,718 C (0.292) 0.232 2.39 × 10−4 
rs12722486 6,143,767 A (0.100) 0.021 0.603 
rs4147359 6,148,444 A (0.333) 0.148 2.39 × 10−4 
rs41295061 (ss52580101) 6,154,665 A (0.067) 0.068 0.234 
rs11594656 6,162,014 A (0.225) 0.038 0.124 

In addition, we verified this cis-regulatory effect in Con A-stimulated PBMCs from eight Canadian T1D cases and parents. Two intronic SNPs (rs706778 and rs3134883) were examined. Consistent with AE data observed in the LCLs, a small but consistent underexpression of the predisposing T-A haplotype (rs706778-rs3134883) was detected in 6 of 8 samples tested. As seen in Table III, 6 of the 8 samples have a 95% confidence interval that does not overlap 0 (log10(1), i.e., normalized allele ratio in DNA is 1), showing the relative higher C-G (rs706778-rs3134883) overexpression compared with the T-A haplotype. Because the strong linkage disequilibrium with genetic variation, the AE in LCLs cannot be due to random monoallelic expression in LCLs that happen to be clonal (15) or other nongenetic confounders (16). Instead, the lower level of AE in Con A-stimulated PBMCs is probably due to the fact that PBMCs are a mixture of cell types in which different T cell subtypes have highly varied IL2RA expression. CD25+ T cells accounts only for a minor fraction of T cells.

Table III.

IL2RA AE measurements in the PBMC samples

Samplers706778rs3134883Haplotype
log10 (C/T)log10 (G/A)log10 (C-G/T-A)a
M0052_3 0.053 0.063 0.058 
 (0.530/0.470) (0.536/0.464) (0.533/0.467) 
M0225_1 −0.011 0.036 0.012 
 (0.493/0.507) (0.520/0.480) (0.507/0.493) 
M0369_1 −0.017 0.023 0.003 
 (0.490/0.510) (0.513/0.487) (0.501/0.499) 
M0434_1 0.024 0.064 0.044 
 (0.514/0.486) (0.537/0.463) (0.525/0.475) 
M0540_2 0.034 −0.003 0.016 
 (0.52/0.48) (0.498/0.502) (0.509/0.491) 
M0892_2 0.020 0.063 0.042 
 (0.512/0.488) (0.536/0.464) (0.524/0.476) 
M1047_1 0.045 0.049 0.047 
 (0.526/0.474) (0.528/0.472) (0.527/0.473) 
M1047_3 0.019 −0.006 0.007 
 (0.511/0.489) (0.497/0.503) (0.504/0.496) 
Samplers706778rs3134883Haplotype
log10 (C/T)log10 (G/A)log10 (C-G/T-A)a
M0052_3 0.053 0.063 0.058 
 (0.530/0.470) (0.536/0.464) (0.533/0.467) 
M0225_1 −0.011 0.036 0.012 
 (0.493/0.507) (0.520/0.480) (0.507/0.493) 
M0369_1 −0.017 0.023 0.003 
 (0.490/0.510) (0.513/0.487) (0.501/0.499) 
M0434_1 0.024 0.064 0.044 
 (0.514/0.486) (0.537/0.463) (0.525/0.475) 
M0540_2 0.034 −0.003 0.016 
 (0.52/0.48) (0.498/0.502) (0.509/0.491) 
M0892_2 0.020 0.063 0.042 
 (0.512/0.488) (0.536/0.464) (0.524/0.476) 
M1047_1 0.045 0.049 0.047 
 (0.526/0.474) (0.528/0.472) (0.527/0.473) 
M1047_3 0.019 −0.006 0.007 
 (0.511/0.489) (0.497/0.503) (0.504/0.496) 
a

Overexpression of the C-G haplotype (rs706778-rs3134883) (or underexpression of the T-A haplotype) is the same as the one observed in the LCLs.

In our 949 Canadian case-parent trios, rs41295061 was associated with T1D (p = 0.053), with the minor A allele being protective as reported by Lowe et al. (4). The minor allele frequency was 0.063 and the odds ratio (95% CI) was 0.728 (range, 0.558–0.950). However, as stated, this SNP had no apparent effect on allelic IL2RA expression and the AE phenotype was best tagged by two intronic SNPs, rs706778 and rs3118470, which we had previously found to be T1D-associated (5). This association was independent of rs41295061, as the transmission disequilibrium of both SNPs remains significant after conditioning with rs41295061 (p = 7.15 × 10−4 and 5.00 × 10−3, respectively). Therefore, there are at least two independent effects in this locus, only one of which affects transcription in lymphocytes and is best (though not perfectly) tagged by rs706778 and rs3118470. The causal variant remains to be identified.

We identified a functional allelic effect that can explain the T1D association, at least in part, in one of the two loci that have been confirmed by the stratification-free transmission disequilibrium test. One obvious mechanism suggested is that lower IL2RA expression may decrease IL2 signaling, crucial for the suppression of autoimmunity by regulatory T cells, consistent with a similar effect of lower IL-2 production determined by the Idd3 locus in the NOD mouse (17). CD4+CD25+ T cells, naturally generated in the thymus, is a population of regulatory T cells that plays an important role in maintaining immune homeostasis and inhibiting autoimmune disease by suppressing activated T cells (18). The crucial role of CD4+CD25+ T cells in a T1D mouse model has been suggested by multiple studies and reviewed by Piccirillo et al. (19) IL-2 is an absolute requirement for the expansion of CD4+CD25+ T cells in NOD mice and has been suggested as a potential therapeutic agent in the prevention of T1D (20). Currently, the IL-2 treatment for T1D is being tested in a clinical trial (Identifier no. NCT00525889, www.ClinicalTrials.gov). Alternatively, the effect could be in other types of immune cells or on the target tissue (pancreatic β cells). The tissue distribution of the allelic effect on transcription will be a subject for future studies.

The cis-regulation we observed is a different effect from the previous reported genetic regulation of soluble IL-2RA (4, 7). The SNPs reported to be associated with soluble IL-2RA are not correlated with the cis-regulation identified by our study. It is still possible that the soluble IL-2RA effect is due to a transcriptional effect that we were not able to detect in this study because of low allele frequency or linkage disequilibrium pattern with the intronic SNPs used to assess AE. Nevertheless, these two independent effects both point to T1D susceptibility being associated with lower IL2RA expression.

This study demonstrates the successful application of fine-mapping of a disease locus using a functional approach. Given the relatively low maximal r2 of the effect with any HapMap SNP, we can state that none of the SNPs examined corresponds to the causal variant for the transcriptional effect (or tags it perfectly). Further identification of the unknown DNA variation in the IL2RA region that can explain the cis-regulatory effect will require thorough resequencing.

The authors have no financial conflict of interest.

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.

1

This work was supported by funding by the GRiD Project from Genome Canada, Génome Québec and by the Juvenile Diabetes Research Foundation International. H.Q.Q. holds a fellowship from the Canadian Institutes of Health Research. T.P. holds a Canada Research Chair (tier 2).

3

Abbreviations used in this paper: T1D, type 1 diabetes; SNP, single nucleotide polymorphism; LCL, lymphoblastoid cell line; AE, allelic expression.

1
Randolph, D. A., C. G. Fathman.
2006
. CD4+CD25+ regulatory T cells and their therapeutic potential.
Annu. Rev. Med.
57
:
381
-402.
2
Willerford, D. M., J. Chen, J. A. Ferry, L. Davidson, A. Ma, F. W. Alt.
1995
. Interleukin-2 receptor α chain regulates the size and content of the peripheral lymphoid compartment.
Immunity
3
:
521
3
Vella, A., J. D. Cooper, C. E. Lowe, N. Walker, S. Nutland, B. Widmer, R. Jones, S. M. Ring, W. McArdle, M. E. Pembrey, et al
2005
. Localization of a type 1 diabetes locus in the IL2RA/CD25 region by use of tag single-nucleotide polymorphisms.
Am. J. Hum. Genet.
76
:
773
-779.
4
Lowe, C. E., J. D. Cooper, T. Brusko, N. M. Walker, D. J. Smyth, R. Bailey, K. Bourget, V. Plagnol, S. Field, M. Atkinson, et al
2007
. Large-scale genetic fine mapping and genotype-phenotype associations implicate polymorphism in the IL2RA region in type 1 diabetes.
Nat. Genet.
39
:
1074
-1082.
5
Qu, H.-Q., A. Montpetit, B. Ge, T. J. Hudson, C. Polychronakos.
2007
. Toward further mapping of the association between the IL2RA locus and type 1 diabetes.
Diabetes
56
:
1174
-1176.
6
Weber, F., B. Fontaine, I. Cournu-Rebeix, A. Kroner, M. Knop, S. Lutz, F. Muller-Sarnowski, M. Uhr, T. Bettecken, M. Kohli, et al
2008
. IL2RA and IL7RA genes confer susceptibility for multiple sclerosis in two independent European populations.
Genes Immun.
9
:
259
-263.
7
Maier, L. M., C. E. Lowe, J. Cooper, K. Downes, D. E. Anderson, C. Severson, P. M. Clark, B. Healy, N. Walker, C. Aubin, et al
2009
. IL2RA genetic heterogeneity in multiple sclerosis and type 1 diabetes susceptibility and soluble interleukin-2 receptor production.
PLoS Genetics
5
:
e1000322
8
Alcina, A., M. Fedetz, D. Ndagire, O. Fernandez, L. Leyva, M. Guerrero, M. M. Abad-Grau, C. Arnal, C. Delgado, M. Lucas, et al
2009
. IL2RA/CD25 gene polymorphisms: uneven association with multiple sclerosis (MS) and type 1 diabetes (T1D).
PLoS ONE
4
:
e4137
9
Hinks, A., X. Ke, A. Barton, S. Eyre, J. Bowes, J. Worthington, S. D. Thompson, C. D. Langefeld, D. N. Glass, W. Thomson.
2009
. Association of the IL2RA/CD25 gene with juvenile idiopathic arthritis.
Arthritis Rheum.
60
:
251
-257.
10
Nejentsev, S., N. Walker, D. Riches, M. Egholm, J. A. Todd.
2009
. Rare variants of IFIH1, a gene implicated in antiviral responses, protect against type 1 diabetes.
Science
324
:
387
-389.
11
Verlaan, D. J., B. Ge, E. Grundberg, R. Hoberman, K. C. Lam, V. Koka, J. Dias, S. Gurd, N. W. Martin, H. Mallmin, et al
2009
. Targeted screening of cis-regulatory variation in human haplotypes.
Genome Res.
19
:
118
-127.
12
Pastinen, T., B. Ge, S. Gurd, T. Gaudin, C. Dore, M. Lemire, P. Lepage, E. Harmsen, T. J. Hudson.
2005
. Mapping common regulatory variants to human haplotypes.
Hum. Mol. Genet.
14
:
3963
-3971.
13
Koeleman, B. P. C., F. Dudbridge, H. J. Cordell, J. A. Todd.
2000
. Adaptation of the extended transmission/disequilibrium test to distinguish disease associations of multiple loci: the Conditional Extended Transmission/Disequilibrium Test.
Ann. Hum. Genet.
64
:
207
-213.
14
Dudbridge, F..
2003
. Pedigree disequilibrium tests for multilocus haplotypes.
Genet. Epidemiol.
25
:
115
-121.
15
Gimelbrant, A., J. N. Hutchinson, B. R. Thompson, A. Chess.
2007
. Widespread monoallelic expression on human autosomes.
Science
318
:
1136
-1140.
16
Choy, E., R. Yelensky, S. Bonakdar, R. M. Plenge, R. Saxena, P. L. De Jager, S. Y. Shaw, C. S. Wolfish, J. M. Slavik, C. Cotsapas, et al
2008
. Genetic analysis of human traits in vitro: drug response and gene expression in lymphoblastoid cell lines.
PLoS Genet
4
:
e1000287
17
Yamanouchi, J., D. Rainbow, P. Serra, S. Howlett, K. Hunter, V. E. Garner, A. Gonzalez-Munoz, J. Clark, R. Veijola, R. Cubbon, et al
2007
. Interleukin-2 gene variation impairs regulatory T cell function and causes autoimmunity.
Nat. Genet.
39
:
329
-337.
18
Shevach, E. M..
2001
. Certified professionals: CD4+CD25+ suppressor T cells.
J. Exp. Med.
193
:
F41
-F46.
19
Piccirillo, C. A., M. Tritt, E. Sgouroudis, A. Albanese, M. Pyzik, V. Hay.
2005
. Control of type 1 autoimmune diabetes by naturally occurring CD4+CD25+ regulatory T lymphocytes in neonatal NOD mice.
Ann. NY. Acad. Sci.
1051
:
72
-87.
20
Rabinovitch, A., W. L. Suarez-Pinzon, A. M. Shapiro, R. V. Rajotte, R. Power.
2002
. Combination therapy with sirolimus and interleukin-2 prevents spontaneous and recurrent autoimmune diabetes in NOD mice.
Diabetes
51
:
638
-645.