Article Text


CTLA-4/CD28 gene region is associated with genetic susceptibility to coeliac disease in UK families
  1. A L King1,
  2. S J Moodie1,
  3. J S Fraser1,
  4. D Curtis2,
  5. E Reid3,
  6. A M Dearlove4,
  7. H J Ellis1,
  8. P J Ciclitira1
  1. 1Gastroenterology Unit, GKT, The Rayne Institute, St Thomas's Hospital, London SE1 7EH, UK
  2. 2Academic Department of Psychological Medicine, St Bartholomew's and the Royal London School of Medicine and Dentistry, 3rd Floor Alexandra Wing, Turner Street, London E1, UK
  3. 3Department of Medical Genetics, University of Cambridge, Addenbrooke's Hospital, Cambridge CB2 2QQ, UK
  4. 4UK HGMP Resource Centre, Hinxton, Cambridge CB10 1SB, UK
  1. Correspondence to:
 Professor P J Ciclitira, Department of Gastroenterology (GKT), The Rayne Institute, St Thomas's Hospital, London SE1 7EH, UK;

Statistics from

Coeliac disease (CD) is a malabsorption disorder characterised by a small intestinal enteropathy that reverts to normal on removal of dietary gluten. Susceptibility to disease has a strong genetic component. Ninety percent of patients in northern Europe have the HLA class II alleles DQA1*0501 and DQB1*0201, which encode the cell surface molecule HLA-DQ2.1 However, haplotype sharing probabilities across the HLA region in affected sib pairs suggest that genes within the MHC complex contribute no more than 40% of the sib familial risk of CD, making the non-HLA linked gene (or genes) the stronger determinant.2

Attempts have been made to identify these loci using genome wide linkage studies. Zhong et al3 performed an autosomal screen in 45 affected sib pairs from the west coast of Ireland, using 328 microsatellite markers. They found evidence of linkage with lod scores of greater than 2.0 in five areas: 6p23 (separate from HLA), 7q31.3, 11p11, 15q26, and 22cen. A larger genome wide search involving 110 affected Italian sib pairs using 281 markers found no evidence of linkage in these five areas.4 It did, however, propose a novel susceptibility locus at 5qter, important in both symptomatic and silent CD, and another at 11qter, which appeared to differentiate the two forms. In UK families an initial genome wide search,5 followed by a study of 17 candidate regions6 identified five areas with lod scores of greater than 2.0: 6p12, 11p11, 17q12, 18q23, and 22q13. Of these, 11p11 replicates one of the loci identified by Zhong et al3 and it is likely that this area contains an important non-HLA susceptibility locus. However, in general the results of these studies are disappointingly inconsistent.

A number of candidate genes have been investigated in linkage and association studies. Of these, the only region with repeatedly positive results is the locus on chromosome 2q33 containing the cytotoxic T lymphocyte associated (CTLA-4) gene and the CD28 gene. CD28 and CTLA-4 molecules are expressed by T lymphocytes and interact with their ligands B7-1 (CD80) and B7-2 (CD86) during antigenic stimulation of T cells via the T cell receptor. CD28 provides a co-stimulatory signal to T cell activation, while CTLA-4 provides a negative signal and thus is thought to be an important regulator of autoimmunity.7CTLA-4 was investigated in a French study of coeliac patients versus controls and showed allelic association of disease with the A allele of the position +49 A/G dimorphism (+49*A/G).8 The association has recently been replicated in a study of Swedish families,9 which also showed some evidence of linkage and association with neighbouring microsatellite markers. A study of Finnish families also showed linkage and association in this region10; however, this was maximal at the marker locus D2S116, and association was not detected at +49*A/G. A study of CTLA-4/CD28 in Italian and Tunisian families,11 however, showed no evidence of linkage or association.

The CTLA-4/CD28 gene region has shown linkage and/or association with a number of chronic inflammatory disorders, including type 1 diabetes.12,13 These studies show linkage and association of CTLA-4 polymorphisms with type 1 diabetes in Italian, Spanish, and French populations, but not in UK, Sardinian, or Chinese data sets.13 We have studied this region in our own sample of UK coeliac families using the transmission/disequilibrium test (TDT). In contrast to conventional case-control studies, the TDT is not liable to produce false positive results owing to unrecognised population stratifications.


Family sample

Affected subjects with both parents available for genotyping were selected from our established collection of multiply affected pedigrees.6 In pedigrees where more than one subject fulfilled this criterion, only one was selected on a random basis. This was to ensure that the study would test for association rather than just linkage. Additional trios of affected subjects with their parents were recruited with the help of an article written for the UK Coeliac Society newsletter. A total of 166 families were studied; however, in 24 of these families only one parent was available for genotyping. All affected subjects were diagnosed according to the revised ESPGAN criteria.14

The study was the approved by the St Thomas's Hospital Ethics Committee and all subjects provided informed consent.

Microsatellite and SNP genotyping

Six finely spaced microsatellite markers and two single nucleotide polymorphisms (SNPs), within and surrounding the CTLA-4 gene were genotyped (fig 1). Markers were chosen in order to allow a direct comparison with previous studies, and at a density sufficient to realistically detect any locus across the region using association methods. Microsatellites were genotyped in all 166 families, but SNPs were genotyped only in the 142 families in which both parents were available for genotyping. Microsatellite markers were selected from the Genethon and CHLC maps: D2S116, D2S2392, D2S2214, D2S2237, D2S1391, and CTLA4(AT) (fig 1). Fluorescently labelled polymerase chain reaction (PCR) products were analysed using the ABI377XL Genetic Analyser (Applied Biosystems). Products were sized using the Genescan version 3.1 program, and scored using the Genotyper version 2.5 program (Applied Biosystems, Foster City, CA USA;

Figure 1

Map of chromosome 2q33-2q34 showing the position of microsatellite markers and SNPs used in the study.

The two known dimorphisms within the CTLA-4 gene, +49*A/G and –318*C/T, were genotyped using the PCR-restriction fragment length polymorphism (PCR-RFLP) method. PCR reactions for +49*A/G were performed using forward and reverse primers previously described by Marron et al.13 The 152 bp product was cleaved for 16 hours at 60°C using 10 units of BstEII per reaction. The digested A allele yields a fragment of 130 bp and the G allele yields an intact 152 bp fragment. The T allele of –318*C/T also creates a restriction site,15 allowing a similar technique to be used for genotyping. Primers were therefore designed around this site: forward primer 5′-TGGACTGGATGGTTAAGGATG-3′ and reverse primer 5′-AGAAGGCACTTGAATAGAAAGC-3′. The 275 bp PCR product was cleaved for 16 hours at 37°C using 4 U MseI per reaction. The C allele produces a 262 bp fragment, whereas the T allele produces a 169 bp and a 95 bp fragment. All fragments were separated on a 2% agarose gel and visualised with ethidium bromide staining under UV fluorescence. Microsatellite and SNP data were checked for genotyping errors using the PEDCHECK program.16

Data analysis

Data for biallelic markers were analysed using the transmission disequilibrium test (TDT).17 In order to avoid biases, only subjects with data available from both parents were included in the analysis. For multiallelic markers such biases do not arise, as long as the affected subject is heterozygous for the marker,18 so all 166 families were analysed. Multiallelic data were analysed using the ETDT program,19 which carries out a logistic regression analysis to determine whether different marker alleles vary in terms of their probability of being transmitted from a heterozygous parent to an affected offspring. When this overall allele-wise test was significantly positive for a marker, we examined the transmission counts for individual alleles, to see which one(s) made the major contribution to the observed effect.


The results of the allele-wise TDT analysis as performed by the ETDT program are shown in table 1. D2S2214 provides evidence for unequal transmission of different alleles to affected offspring (χ2 =19.4, 7 df, p=0.007). Examining the individual allele transmissions (table 2), it seems that most of this effect is the result of preferential transmission of allele *278, which is transmitted from 92 heterozygous parents and not transmitted from 49 (χ2=13.1, 1 df, p=0.0003). A lesser contribution (χ2=9.333, 1 df, p=0.002) to the effect is made by allele *276. The only other somewhat positive result is with D2S1391 (χ2 =13.8, 6 df, p=0.03), but given that eight markers were tested this result could easily have occurred by chance. For both CTLA-4 single nucleotide polymorphisms, alleles were transmitted approximately equally, providing no evidence at all for the direct involvement of these polymorphisms in the susceptibility to CD.

Table 1

Results of allele-wise TDT analysis using ETDT

Table 2

Transmission of individual alleles of D2S2214 to affected offspring from heterozygous parents


This study of UK families provides further evidence that the CTLA-4/CD28 gene region on chromosome 2q33 contains an important non-HLA susceptibility locus for coeliac disease. This evidence comes from a positive TDT result with the D2S2214 microsatellite (p=0.007), indicating that allelic association with the disease is present with this locus. The use of the TDT method rather than a case-control sample means that this positive result is not the result of unrecognised population stratifications. Although association has not previously been shown with this marker, it has been shown with two nearby markers, D2S116 and +49*A/G. D2S116 lies 0.01 cM centromeric of D2S2214 and showed association significant at p=0.0001 in one previous study.10 The +49*A/G SNP lies 0.3 cM telomeric of D2S2214 and has been found to be associated with CD in two previous studies with significance p=0.00018 and p=0.007.9 We have found no evidence for association of +49*A/G or D2S116 in the current study. The –318*C/T SNP lies within the CTLA-4 gene promoter region and is thus a candidate to be an aetiological polymorphism, but it did not show any evidence of association in the current study. The third known polymorphism of the CTLA-4 gene is the CTLA-4 (AT) microsatellite positioned in the 3′ untranslated region of exon 3. Theoretically, this variation could affect gene expression by affecting mRNA stability; however, in keeping with two previous studies,9,10 we failed to detect any evidence for association.

The finding that different polymorphisms within the same region are positive in different samples is consistent with the hypothesis that none of them influences susceptibility to CD directly but that there is another, as yet untested, susceptibility locus within the region. Different patterns of linkage disequilibrium between markers in different populations, along with chance variations, would then account for the different results obtained. Association of the CTLA-4/CD28 region with genetic susceptibility to coeliac disease has now been reported in UK, French, Finnish, and Scandinavian populations. No association has yet been found in Italian, Tunisian, or Dutch20 populations. These differences might result from chance factors, from different frequencies of the susceptibility locus among cases in different populations, or because of differences in the pattern of linkage disequilibrium across the region. It may be of some interest that the pattern observed across Europe seems different from that found in the type 1 diabetes studies, in particular in relation to UK families, in that association is found for coeliac disease but not for diabetes.

We have used a TDT design which provides a robust test for association and which may be more powerful at detecting loci in complex diseases than linkage analysis,21,22 at least when attention can be restricted to a small region. Association based studies can also provide a more precise localisation than linkage studies since they are based on the presence of linkage disequilibrium, which in outbred human populations only exists between markers within a few hundred kb of DNA sequence (probably <300 kb).23–25CTLA-4 and CD28 are both plausible candidate genes and are separated by only 25-150 kb. However, a polymorphism within these genes that predisposes to CD has not yet been identified. Disease predisposition may lie in undiscovered polymorphisms of CTLA-4 or CD28, or alternatively within another gene in very close proximity. A database search of the region using Genemap 99 ( and Ensembl ( did not show any genes with known immunological or gut related functions. Future work will therefore involve detailed mapping of the region with identification of further SNPs within CTLA-4/CD28 and in the surrounding region in order to identify the aetiological polymorphism that confers susceptibility to coeliac disease.

Key points

  • Genetic susceptibility to coeliac disease is not entirely explained by known HLA associations.

  • Association of coeliac disease with the CTLA-4/CD28 gene region on chromosome 2q33 has been reported in some European populations but not in others.

  • This study of UK families did not show significant association with any of the three known polymorphisms of the CTLA-4 gene.

  • The study did, however, show significant association with D2S2214 (p=0.007), a microsatellite marker 0.3 cM centromeric of the CTLA-4 gene.


Microsatellite and SNP genotyping was carried out at the Medical Research Council's UK Human Genome Mapping Project Resource Centre Linkage Hotel. A L King is funded by a British Society of Gastroenterology/Digestive Disorders Foundation two year Research Training Fellowship. S J Moodie is funded by the European Commission. The authors also wish to thank the UK Coeliac Society and Action Research.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.