|Home | About | Journals | Submit | Contact Us | Français|
The inflammatory disorders type 1 diabetes (T1D) and celiac disease co-segregate in populations, suggesting a common genetic origin. Both are associated with the HLA class II genes on chromosome 6p21, and the present paper tested whether non-HLA loci are shared.
We evaluated eight celiac disease risk loci in T1D by genotyping and statistical analyses of 8,064 T1D cases, 9,339 controls and 2,519 families. We also investigated 18 T1D loci in 2,560 celiac disease cases and 9,339 controls.
Three celiac disease loci, listed as chromosome/candidate gene: 1q31/RGS1, 2q12/IL18RAP and 6q25/TAGAP, were associated with T1D (P < 10−4). The 3p21/CCR5 32 base pair insertion/deletion variant was newly identified as a T1D locus (P = 1.81 × 10−8), and was also associated with celiac disease, as were 18p11/PTPN2 and 2q33/CTLA4, bringing the total loci shared to seven, including 12q24/SH2B3. The 2q12/IL18RAP and 6q25/TAGAP allele associations were in the opposite direction in T1D as compared to celiac disease. Distinct effects included 11p15/INS, 10p15/IL2RA and 1q13/PTPN22 in T1D and 3q25/IL12A and 3q28/LPP in celiac disease.
Genetic susceptibility to T1D and celiac disease shares common alleles. These data suggest that common biological mechanisms, such as autoimmunity related tissue damage and intolerance to dietary antigens may be a feature of T1D.
Type 1 diabetes (T1D) is caused by autoimmune destruction of the insulin-producing β cells in the pancreatic islets, affecting approximately 0.4% of European populations and strongly clustered in families. The major susceptibility genes, the HLA class II loci, HLA-DQB1 and HLA-DRB1 on chromosome 6p21, act in combination with many other non-HLA loci across the genome,1, 2 with unknown environmental factors playing a major role.3-6 Celiac disease, which results from an immune, inflammatory reaction in the small intestine to ingested barley, wheat and rye gluten proteins, occurs in 0.1% of northern European-origin populations, an estimate based on clinically-diagnosed symptoms. However, within that population, there may be as much as 1% prevalence, if the highly sensitive and specific test for autoantibodies to tissue transglutaminase is used.7, 8 The major susceptibility gene is also HLA-DQB1.9, 10
Celiac disease and anti-tissue transglutaminase antibodies occur more frequently in T1D cases than in the general population, depending on the age of the patients; at most 10% of children and 2% of adults with T1D are positive.11 Increasing incidence of celiac disease over recent decades has also been reported.8 It has been suggested that gluten consumption, and gut permeability and inflammation, are also factors in the development of T1D.6, 12 These results suggest that the etiologies of T1D and celiac disease may share some genetic and environmental factors.
Eight chromosome regions outside the HLA region have recently been associated with celiac disease using the genome wide association (GWA) approach and achieving genome-wide statistical support at P < 5 × 10−7, probably providing a representative view of the major genetic effects in the northern European population for this disorder 10 (Methods and Glossary). In T1D, 15 non-HLA regions have been established to date 1, 13-15 (Methods and Glossary) and two other loci, 5p13/IL7R and 18q22/CD226, have been implicated in T1D and multiple sclerosis.1, 16, 17 It has already been reported that the 12q24/SH2B3 locus is shared between T1D and celiac disease, and possibly loci 4q27/IL2-IL21 and 3p21/CCR3.9, 10 In addition, there is some evidence for association of the established T1D loci, 2q33/CTLA49, 18 and 1p13/PTPN22, 19 in celiac disease. In the present report, we evaluated the association of all these loci in T1D and celiac disease, including the CCR5 32 base pair insertion/deletion variant that we report here as a T1D locus, in order to assess the genetic similarities and differences between these two inflammatory disorders.
The T1D cases (www.childhood-diabetes.org.uk/grid) were under 16 years of age at the time of sample collection (mean age at diagnosis = 7.5 years, range 0.5 – 16 years).1 The control samples (n = 6,164) are from the British 1958 Birth Cohort (www.b58cgene.sgul.ac.uk); and from a collection of blood donors (n = 3,175), established by the Wellcome Trust Case Consortium.13 The family collection consisted of 455 Diabetes UK Warren 1 families; 250 Northern Irish families; 243 USA families from the Human Biological Data Interchange; 411 Romanian families; 360 Norwegian families and 800 Finnish families.1 The 2,560 celiac disease patients were recruited throughout England, Scotland and Wales. DNA was extracted from peripheral blood for 1,175 persons recruited from hospital outpatient clinics, and from saliva DNA for 1,385 persons recruited via Coeliac UK advertisement. Diagnosis of celiac disease was based on clinical symptoms, current gluten free diet, serology, small intestinal biopsy and response to treatment. Mean age at diagnosis was 41.0 years (range 3 months - 84 years), and 75.1% were female. The Irish collection consisted of 416 celiac cases and 957 controls, and the Dutch collection of 507 celiac cases and 888 controls.10 All cases (T1D and celiac disease) controls and families self-reported as white ethnicity. The relevant research ethics committees approved the study, and written informed consent was obtained from the participants, or their parents/guardian for those too young to consent.
SNPs were genotyped from eight celiac disease loci: chromosome 1q31/candidate gene RGS1, 2q12/IL18RAP, 3p21/ CCR3, 3q25/IL12A, 3q28/LPP, 4q27/IL2-IL21, 6q25/TAGAP and 12q24/SH2B3 and 15 T1D loci: 1p13/ PTPN22; 2q24/IFIH1; 2q33/CTLA4; 4q27/IL2-IL21; 6q15/BACH2; 10p15/IL2RA/CD25; 10p15/PRKCQ; 11p15/INS; 12q13/ERBB3; 12q24/SH2B3; 15q24/CTSH; 16p13/CLEC16A, 18p11/PTPN2; 21q22/UBASH3A and 22q13/C1QTNF6 [see Glossary for full gene names]. SNPs from 5p13/IL7R, 18q22/CD22 and 3p21/CCR5 were also genotyped.
In our study we claim “significance” when P<10−4; this approach is conservative because the two disorders being studied must have established evidence that they have a familial (co-segregation) and/or clinical-epidemiological association (i.e., more or less cases of one occur in the patient population of the other disease), and the two diseases share some clinical and biological phenotypes 20, 21. We also required that the evidence for the locus association with the first disease is robust and convincing, i.e., P < 5 × 10−7 in multiple populations, and there be robust marker scoring and statistical analyses (Supplemental Appendix).
We genotyped in 8,064 T1D cases and 9,339 controls, and where appropriate, in 2,519 parent-child trio families, the nine SNPs with the highest disease association from the eight non-HLA celiac disease regions 10(Table 1; Supplementary Table 1). Three of these newly analyzed regions showed strong evidence of association (P < 10−4; Methods) with T1D in case-control and family sample sets, 1q31/RGS1, 2q12/IL18RAP and 6q25/TAGAP (Table 1; Supplementary Table 1). Therefore, along with the 12q24/SH2B3 sharing reported previously,10 four of these eight celiac loci are shared with T1D (Fig. 1, A). The celiac disease-associated SNPs/variants of 3p21/CCR3/rs6441961 and 4q27/IL2-IL21/rs6822844 did not reach our statistical threshold in T1D (P > 10−4; Table 1). The regions 3q25/IL12A and 3q28/LPP showed no evidence of association with T1D (P > 0.147; Table 1; Fig 1, D).
Since the 3p21/CCR3 association in the T1D case-control at P = 3.40 × 10−4 (Table 1) just failed to reach our threshold for statistical significance, and CCR3 is one of several chemokine receptor genes on chromosome 3p21, we hypothesized that a stronger T1D association might exist due to a polymorphism in one of the other CCR genes in this region, all of which are functional candidates for both diseases. We, therefore, tested the association of two established functional variants, one in the CCR2 gene (rs1799864/ Ile64Val) and the other in CCR5 (rs333, the 32 base pair insertion/deletion), which have been reported to be associated with susceptibility to HIV infection, its outcome and treatment.22 Moreover, polymorphisms of CCR5 and its ligand, CCL3L1, have also been associated with susceptibility to rheumatoid arthritis,23 and for CCR5 with T1D in several smaller studies.24-28 We did not find any evidence for an association between CCR2/rs1799864 and T1D (P = 0.506; Table 2). Homozygosity of the CCR5/rs333 deletion allele, which encodes a non-functional receptor, in contrast, was associated with decreased T1D risk, odds ratio (OR) = 0.54, (95% confidence interval (CI) = 0.40-0.72), P = 1.88 × 10−6 (2-df)(Table 2). We validated the association in the family collection relative risk (RR) = 0.53 (95% CI = 0.34-0.82), P = 9.10 × 10−3, and overall the combined results gave P = 1.81 × 10−8 (Table 2). The CCR5 insertion/deletion rs333 is located 62 kb away from the CCR3/rs6441961 SNP, with D′= 0.98 and r2 = 0.05, and logistic regression analysis indicated that the potential T1D association with CCR3/rs6441961 is not due to linkage disequilibrium with CCR5/rs333: CCR5/rs333 added to CCR3/rs6441961, P = 3.39 × 10−5, and in the reverse analysis, adding CCR3/rs6441961 to CCR5/rs333 gave P = 7.6 × 10−3.
We analysed the 18 loci that have been associated with T1D, including the CCR5 deletion/rs333 variant (Table 2) and the 12q24/SH2B3 locus previously recognised to be shared between the two diseases, in celiac disease by genotyping 19 SNPs and the CCR5/rs333 variant in 2,560 celiac cases and comparing the results with those from the 9,339 controls (Table 3; Supplementary Table 4; Fig 1). The most significantly associated loci were CTLA4/rs3087243 and CCR5/rs333 (P = 1.26 × 10−6 and 9.18 × 10−6, respectively), indicating that these two regions are likely to be true effects, a conclusion supported by previous reports that these loci have been associated with both disorders and other immune-mediated diseases.9, 18, 23, 26 Markers in the CCR5 and CCR3 genes were independently associated with celiac disease: in a logistic regression analysis CCR5/rs333 added to CCR3/rs6441961, P = 1.00 × 10−3, and in the reverse analysis, adding CCR3/rs6441961 to CCR5/rs333 gave P = 0.0127. These results indicate two or more causal variants or genes in this chemokine gene rich region of chromosome 3p21.
We previously reported two independent T1D associations within the PTPN2 region marked by the SNPs rs1893217 and rs478582.1 Resequencing of the PTPN2 gene, genotyping and analyses identified a SNP rs45450798 in high linkage disequilibrium with rs1893217 (r2 = 0.97) that replaces rs1893217 as the most associated SNP in the PTPN2 region. Logistic forward regression analysis revealed that rs45450798 explained the association at rs1893217 and combined with rs478582 explains the T1D association of the PTPN2 chromosome region (Supplementary Table 3). In the celiac disease cases, the PTPN2 SNP rs45450798 just failed to pass our threshold of P ≤ 10−4 (P = 2.61 × 10−4; Table 1). Therefore, we analysed the available, but unpublished, data for the two independent case-control sample sets from Ireland and the Netherlands, obtaining consistent support for the association of PTPN2 rs1893217 with celiac disease (P = 0.045; Supplementary Table 5). Combined with the fact that PTPN2 has also been associated with the inflammatory bowel disease, Crohn's disease, 29 it is highly likely that PTPN2 is also a celiac disease locus, bringing the total of non-HLA celiac disease loci from eight to 11.
Six other regions showed nominal evidence at P < 0.05 of association with celiac disease. The currently most associated SNP for T1D in the 4q27/IL2-IL21 region, rs2069763, (a synonymous SNP in exon 1 of IL2) is weakly associated with celiac disease (P = 0.018) indicating that while this region is linked to both diseases the genetic variants are different. The remaining regions are: 5p13/IL7R (P = 7.23 × 10−3), 6q15/BACH2 (P = 2.78 × 10−3), 10p15/PRKCQ (P = 0.0178), 18q22/CD226 (P = 0.0133) and 21q22/UBASH3A (P = 8.88 × 10−3). Figure 1 illustrates the combined results of Tables Tables11 and and3,3, with 14 loci showing some evidence for co-localisation, seven of which are convincing: 1q31/RGS1, 2q12/IL18RAP, 2q33/CTLA4, 3p21/CCR5, 6q25/TAGAP, 12q24/SH2B3, and 18p11/PTPN2. At least five showed distinct differences (Figure 1, C and D), namely a strong association in one disease, and no or little evidence for association in the other (11p15/INS, 1p13/PTPN22, 10p15/IL2RA, 3q28/LPP and 3q25/IL12A).
The results presented here and reported recently elsewhere1, 10, 14, 15 indicate that one may be confident in 21 non-HLA loci in T1D and 11 in celiac disease (Fig. 1), of which four are newly identified loci in T1D (1q31/RGS1, 2q12/IL18RAP, 3p21/CCR5 and 6q25/TAGAP; Tables Tables11 and and2)2) and of which two are new for celiac disease, 3p21/CCR5 and 18p11/PTPN2. Further, the results provide convincing confirmation of the importance of the 2q33/CTLA4 region (Table 3). Seven of these chromosome regions are shared between the two diseases, suggesting that for an investigation of shared loci in two diseases that are known to co-segregate, the prior odds of 1000:1 against there being a true association at any locus tested 20 is too conservative.
Four alleles, 1q31/RGS1, 2q33/CTLA4, 12q24/SH2B3 and 18p11/PTPN2, show the same direction of association in the two diseases, constituting evidence for shared causal variants. We know that this is not due to bias in ascertainment of the cases (Supplementary Appendix); nor is the use of a common set of controls a problem since we have consistent results from families (Supplementary Appendix).
The minor alleles of the SNPs 2q12/IL18RAP rs917997 and 6q25/TAGAP rs1738074 were negatively associated with T1D (Table 1),with the effects in the opposite direction to the previous celiac disease findings (Table 1).10 These results may be interpreted in two ways: the causal variants in these two regions have opposite biological effects in T1D and celiac disease, or that there are different causal variants for each disease in each region with the typed marker SNPs tagging these causal variants. For the 2q12/IL18RAP and 6q25/TAGAP regions we have found no evidence so far in T1D GWA studies13, 15 for a second loci within these regions (data not shown). Moreover, there is precedent for a causal variant having opposing effects in different diseases. For example, the minor allele of 1p13/PTPN22 variant Arg620Trp predisposes a person to many immune-mediated diseases but is protective for Crohn's disease.30 Hence, we favor the possibility that the causal variants have opposite effects in the T1D and celiac disease patients. In contrast, for 4q27 our current data indicate that different causal variants are involved, perhaps affecting different genes, in T1D and celiac disease. The important immune response genes, IL-2 31 and IL-21 are strong functional candidates. Before we can draw further conclusions, all the regions discussed here must be thoroughly resequenced from multiple persons to ascertain a complete catalogue of polymorphisms, followed by further genotyping in order to identify all of the most associated variants.
Nevertheless, the 32 base pair insertion/deletion in CCR5 (rs333), which causes loss of expression of the receptor,22 could well be the actual functional, causal variant involved. The disease associations of the two chemokine receptor genes, CCR3 (Table 3) and CCR5 (Table 2), suggest the central importance of lymphocyte trafficking in these organ specific diseases. The development and anatomy of the small intestine and pancreas are close, and the gut immune system shares close connections with the pancreatic lymph nodes, which have been linked to insulitis and β-cell destruction.32 In recent-onset T1D patients alterations in the levels of CCR5 ligands, CCL3 (MIP-1α) and CCL4 (MIP-1β) have been reported.33 In the NOD mouse model of T1D, CCR5 and its ligand CCL4 have multiple reported significant roles in the disease development. 34
There are, however, distinct differences in genetic susceptibility between the two diseases, including at 1p13/PTPN22, 10p15/IL2RA and 11p15/INS (Table 1), and although there are shared T1D and celiac disease predisposing alleles at the HLA-DQB1 gene, there are distinguishing HLA-DQB1 genotype differences (Supplementary Appendix). One possibility is that there is a common autoimmunity-inflammatory genetic background, and that further combinations of more disease-specific variation at HLA and non-HLA genes, in interaction with epigenetic and environmental factors, determine the final clinical outcomes.
Our results support further evaluation of the hypothesis that cereal and gluten consumption might be an environmental factor in T1D leading to the alteration of the function of the gut immune system and its relationship with the pancreatic immune system.6, 12, 32, 35 Furthermore, insulin and its precursors are major targets of the T and B lymphocyte autoreactive response in T1D; thus, one might speculate that bovine insulin in infant feeds could enhance anti-insulin responses,3 particularly if there are genetically-determined defects in oral tolerance predisposing to T1D. Conversely, genes classified as autoimmunity genes, because they are associated with T1D, contribute to celiac disease.
This work was funded by Juvenile Diabetes Research Foundation International; the Wellcome Trust; the National Institute for Health Research Cambridge Biomedical Research Centre, Coeliac UK, the Coeliac Disease Consortium (an innovative cluster approved by the Netherlands Genomics Initiative and partly funded by the Dutch Government), the European Union, the Netherlands Organization for Scientific Research, the Science Foundation Ireland and the Irish Health Research Board.
The Cambridge Institute for Medical Research (CIMR) is in receipt of a Wellcome Trust Strategic Award (079895).
We gratefully acknowledge the participation of all the patients and control subjects. We acknowledge use of the DNA from the British 1958 Birth Cohort collection, funded by the Medical Research Council and Wellcome Trust. We thank the Avon Longitudinal Study of Parents and Children laboratory in Bristol and the British 1958 Birth Cohort team, including S. Ring, R. Jones, M. Pembrey, W. McArdle, D. Strachan and P. Burton for preparing and providing the control DNA samples. We thank the Human Biological Data Interchange and Diabetes UK for the USA and UK multiplex families, respectively, the Norwegian Study Group for Childhood Diabetes for the collection of the Norwegian families (D. Undlien and K. Ronningen); D. Savage, C. Patterson, D. Carson and P. Maxwell for the Northern Irish samples. Genetics of Type 1 Diabetes in Finland (GET1FIN) J. Tuomilehto, L. Kinnunen, E. Tuomilehto-Wolf, V. Harjutsalo and T. Valle for the Finnish family and C. Guja and C. Ionescu-Tirgoviste for the Romanian family samples. Irish Control DNA was supplied by Irish Blood Transfusion Service/ Trinity College Dublin Biobank.
We thank the Barts and The London Genome Centre for genotyping support; J. Swift, R. Crimmins, P. Kumar, D.P. Jewell, L. Dinesen, S.P.L. Travis, K. Moriarty, P. Howdle, D.S. Sanders, G.K.T. Holmes, S. Sleet and Coeliac UK for collection of British celiac samples, A. Monsuur, C.J. Mulder, M.L. Mearin, W.H.M. Verbeek, for patient recruitment, G. Meijer and J. Meijer for histology review, K. Duran for DNA extraction, H. van Someren and F. Mulder for clinical database management, L. Franke, A. Zhernakova, M. Plateel, the Genotyping facilities at UMCG and UMC Utrecht for technical assistance (the Netherlands), F.M. Stevens, C. O'Morain, N.P. Kennedy, M. Abuzakouk, R. McLoughlin, K. Brophy, C. Feighery, J. McPartlin, D. Kelleher, A.W. Ryan, G. Turner for sample collection and genotyping. We thank L. Wicker for comments on the manuscript.
Dr McManus reports having received Grant support from Hitachi Europe, Ltd.
Prof van Heel reports holding stock options and consulting fees from NexPep Pty Ltd
No potential conflict of interest to this article was reported.