Introduction

Inflammatory bowel diseases (IBDs) that are usually classified into two clinical entities, Crohn’s disease (CD; MIM 266600) and ulcerative colitis (UC; MIM 191390), are chronic conditions characterised by remitting and relapsing inflammation of the small and/or large intestines. The combined prevalence of the two diseases in the West and in Asia has been increasing significantly over the last decade; the prevalence rate of UC and of CD in the Japanese population was estimated to be 6.31 and 0.88 per 100,000 in 1985, but the rate increased to 18.12 and 5.85 per 100,000 in 2006 (Hilmi et al. 2006). A similar increasing trend in the annual incidence was observed in Korea, Singapore and China (Law et al. 1998; Yang et al. 2000; Leong et al. 2004). Although these values in Asian countries were still relatively low when compared to Western countries (Loftus 2004), IBDs are now thought to be common diseases in Asia.

In spite of a significant number of studies to identify the fundamental pathophysiologic processes, the etiology of IBDs still remains largely unknown. IBDs were thought to be multifactorial diseases. In fact, aggregate effects of genetic, environmental and other processes are found to induce an abnormal response of the mucosal immune system (Podolsky 2002). The role of genetic factors in the etiology of IBDs was suggested by familial aggregation through twin studies (Vermeire and Rutgeerts 2005).

Linkage analyses followed by fine-mapping or genome-wide association (GWA) analyses have identified susceptible variants to CD in several genes, including CARD15 (NOD2) (Hugot et al. 2001; Ogura et al. 2001), DLG5 (Stoll et al. 2004), SLC22A4 and SLC22A5 (Peltekova et al. 2004), CARD4 (also known as NOD1) (McGovern et al. 2006) and TNFSF15 (Yamazaki et al. 2005). Recently, three independent groups have newly reported susceptible genes by way of genome-wide linkage disequilibrium-based association studies. The first group determined ten single nucleotide polymorphisms (SNPs) in the IL23R (interleukin 23 receptor) gene as significant markers (Duerr et al. 2006). The IL23R gene is located on chromosome 1p31 and forms a receptor for IL23 (interleukin 23), together with the beta 1 subunit of IL12 (IL12RB1) (Parham et al. 2002). The allele A of rs11209026 (c.1142G>A, p.Arg381Gln) is found to be protective against CD development in the two ethnic cohorts, European and Jewish. The second gene identified was the ATG16L1 (autophagy-related 16-like 1) (Hampe et al. 2007) located on chromosome 2. Although a recent study has revealed that a mouse orthologue ATG16l is localised to the autophagic isolation membrane during autophagosome formation (Mizushima et al. 2003), the function of human ATG16L1 remains uncertain. The allele G of rs2241880 (c.898G>A, p.Thr300Ala) conferred susceptibility to CD and was found to interact with the CARD15 risk genotype. The last group identified a susceptible 250-kb region on chromosome 5p13.1 (Libioulle et al. 2007). Several genetic variants were found to influence the expression of the PGER4 (prostaglandin E receptor 4) gene, which resides closest to the associated region.

To investigate a possible role of these candidate genes, IL23R and ATG16L1 and 5p13.1 loci, in the pathogenesis of CD development of Japanese patients, we assessed the distribution of 29 selected markers and examined the genotype–phenotype analysis. In addition, we performed haplotype analysis and found an ethnical divergence between the European and Japanese populations.

Methods

Human subjects and phenotypic analysis

Japanese blood samples were obtained with written informed consent from 484 CD patients at the Social Insurance Chuo General Hospital and from 439 unaffected control individuals belonging to the Osaka-Midosuji Rotary Club. The study protocol was approved by the local ethics board. All CD cases were diagnosed at the Inflammatory Bowel Unit of the Social Insurance Hospital by clinical, radiological, endoscopic and histological findings according to the Lennard-Jones criteria (Lennard-Jones 1989) and patients with indeterminate colitis were excluded. Extensive clinical characterisation was available in 482 CD patients. The clinical characteristics of CD patients were assessed at the time of diagnosis and were categorised using the Vienna classification (Gasche et al. 2000). In addition to that, the past medical history of surgical operation was obtained from the clinical records of 418 patients. The demographics of the CD patients and controls are shown in Table 1.

Table 1 Demographic and clinical features of the Crohn’s disease and the control groups
Full size table

SNP analysis and genotyping

We selected a total of 29 SNPs for genotyping; 12 in ATG16L1, which was reported to constitute susceptible haplotypes, 10 in IL23R and 7 in 5p13.1, with which positive association was shown (Duerr et al. 2006; Libioulle et al. 2007; Hampe et al. 2007). TaqMan assay (Applied Biosystems) was applied for genotyping 18 SNPs. Nine SNPs, four of ATG16L1 and five of 5p13.1 locus, were genotyped by the Invader assay (Ohnishi et al. 2001). The genotyping of rs1441090 was performed by sequencing and rs11465804 was by polymerase chain reaction restriction fragment length polymorphisms (PCR-RFLP) analysis. The primers and typing methods used are described in Table 2.

Table 2 List of genotyping methods for each single nucleotide polymorphism (SNP) and primer sequences
Full size table

Statistical analysis

The markers were evaluated for deviation from the Hardy-Weinberg equilibrium in the controls before inclusion in the analysis (P>0.05). Allele frequencies were analysed by χ 2 tests and the p-values were corrected by 1,000 random permutations for each SNP. Haplotype frequencies and linkage disequilibrium (LD) were estimated and visualised using Haploview 3.32 (available at http://www.broad.mit.edu/mpg/haploview/) (Barrett et al. 2005). Plots of the relative D′ levels between each locus in the Japanese population were calculated from the genotype data of the case and control populations and those in the Caucasian population were from the HapMap data. LD structure of the ATG16L1 locus in the Caucasian population was constructed with 11 SNPs instead of 12, as rs2241879 has been eliminated in the HapMap data. p values for haplotype association analysis were obtained after 1,000 permutation tests.

Results

Case–control analysis of IL23R, ATG16L1 and 5p13.1 loci with Japanese CD patients

As shown in Table 3, all 29 SNPs located in the three candidate loci did not show any positive association with Japanese CD cases. Among the ten SNPs of IL23R examined, two SNPs, rs11465804 and rs11209026, were absent both in the Japanese CD cases and controls; the latter was identified during the GWA study and both SNPs were in strong LD. Another SNP named rs4613763 in the 5p13.1 locus was also absent in Japanese population; only allele T existed.

Table 3 Genotype distribution for IL23R, ATG16L1 and 5p13.1 loci markers
Full size table

Genotype–phenotypic analysis of three candidate loci with Japanese CD patients

Duerr et al. 2006 focussed on ileal CD to minimise pathogenic heterogeneity and identified IL23R as a susceptible gene. That prompted us to analyse ten SNPs of IL23R with ileal CD as cases and we found that one SNP, rs11209032, was marginally associated; however, the significance was not confirmed after permutation tests (Table 4). Other sub-phenotypes of CD (ileocolitis and colitis) did not demonstrate any significant effects (data not shown). Likewise, all 12 SNPs of ATG16L1, including rs2241880 and seven SNPs on 5p13.1, were not significant when any sub-phenotype was used as cases (data not shown).

Table 4 Genotype distribution for IL23R between ileal CD patients and controls in Japanese
Full size table

Haplotype analysis of ATG16L1, IL23R and 5p13.1 loci

The haplotype frequency of each locus was estimated with the genotyping data of 29 markers. As shown in Tables 5, 6 and 7, haplotype association analysis of the three loci to CD susceptibility did not show any positive result. In the IL23R locus, a slight increase of major haplotype (TTACCAAG) frequency in Japanese CD patients to controls (CD 41.3% vs. controls 36.1%) was observed but was not deemed significant. Furthermore, the frequency of the haplotype of three loci also has a significantly different structure between the Japanese and Caucasian populations when compared to Caucasian frequency data. In order to recognise the difference between genetic and racial background in these regions visually, we characterised the LD structure of the three loci in both populations. Great divergence was clearly demonstrated (Fig. 1) and this might partly explain the ethnic difference of CD susceptibility.

Table 5 Result of a haplotype analysis of eight SNPs at the IL23R locus in Japanese CD patients and controls
Full size table
Table 6 Result of a haplotype analysis of 12 SNPs at the ATG16L1 locus in Japanese CD patients and controls
Full size table
Table 7 Result of a haplotype analysis of six SNPs on the 5p13.1 locus in Japanese CD patients and controls
Full size table
Fig. 1a–c
figure 1

The structure of linkage equilibrium (D′) among Japanese and Caucasian populations. The D′ scores in the Japanese population were estimated from the genotyping case and control data across each of aIL23R, bATG16L1 and c 5p13.1 loci and in the Caucasian population was as generated by Haploview from the Caucasian HapMap data. Registration of rs2241879 in the ATG16L1 locus was eliminated in the HapMap data

Full size image

Discussion

In this paper, we analysed a total of 29 candidate variants of the three loci, IL23R, ATG16L1 and 5p13.1, recently reported to confer susceptibility with CD in the Japanese population. The three variants described to be responsible for the disease were absent in the Japanese population; two variants in IL23R, only allele G of rs11209026 and allele T of rs11465804, and one variant in 5p13.1 locus, only allele T of rs4613763 existed. All of the remaining 26 genetic markers have failed to show any positive association to Japanese CD. Duerr et al. 2006 performed GWA studies with clinically stratified CD subjects, ileal CD, to minimise pathogenic heterogeneity. That enabled us to analyse in the same way stratified samples employed as cases, however, no significant association were obtained. As shown in Fig.  1, the analysis of LD structure showed us evident ethnic differences in all of the three loci. Our previous finding that most of the susceptible genes identified in cohorts of European and Jewish ancestry were not adaptable for Japanese CD patients (Yamazaki et al. 2002; Yamazaki et al. 2004) but were again reproduced in the present study.

Since CARD15 was identified as the first gene conferring susceptibility to CD in 2001 (Hugot et al. 2001; Ogura et al. 2001), a significant number of studies of its replication and newly identified responsible genes followed. Three major polymorphisms in the CARD15 gene, R702W, G908R and 1007fs, were confirmed to be associated with susceptibility to Caucasian CD patients by independent groups (Ahmad et al. 2002; Lesage et al. 2002), even though the frequencies of these variants were quite different between ethnically divergent populations. Among CD patients of European ancestry, CARD15 variants were more significantly associated in the Central European population (Hugot et al. 2001; Ahmad et al. 2002; Lesage et al. 2002) than in the North European population (Paavola-Sakki et al. 2003; Arnott et al. 2004; Medici et al. 2006). The absence of the variants were widely known in Asia; Japanese (Yamazaki et al. 2002), Korean (Croucher et al. 2003) and Chinese populations (Leong et al. 2003). Susceptible genes identified next were DLG5 (Stoll et al. 2004), SLC22A4 and SLC22A5 (Peltekova et al. 2004). The association has been studied with various ethnic populations and the results shown were also extremely heterogeneous (Yamazaki et al. 2004; Friedrichs and Stoll 2006; Silverberg 2006).

The IL23R, ATG16L1 and 5p13.1 loci were identified by GWA studies and the results were confirmed with large independent Caucasian samples. The biological, technical and statistical foundations have been laid to apply GWA studies as a critical tool for the identification of susceptibility genes for complex diseases and they have produced more robust results. These original comprehensive analyses provided the information of replication in previously reported regions. Libioulle et al. 2007 have identified the 5p13.1 locus to confer CD susceptibility and, at the same time, confirmed previously reported regions, IL23R and ATG16L1. Furthermore, by a group that identified IL23R as a susceptible gene to CD, GWA studies reported replicated positive association with ATG16L1 (Rioux et al. 2007). These findings supported the theory that IL23R and ATG16L1 were common susceptible genes to CD in Caucasian populations. Although it was generally accepted that the clinical profiles of CD are similar between Caucasians and Asians (Hilmi et al. 2006), common susceptible variant(s) to CD has not been reported so far. Great ethnical diversity of susceptible genes to CD between Japanese and European ancestries seems to exist.

In conclusion, we failed to confirm the association between the candidate genetic variations in the IL23R, ATG16L1 genes and the 5p13.1 locus in Japanese CD. Our result suggested that these candidate genes were not common variants to CD among the Japanese and Caucasian populations. In consideration of the increased prevalence of IBDs in Asian, systematic screening should be carried out as GWA studies among various populations with different ethnical backgrounds and it will lead to elucidate the contribution of susceptibility genes to IBD.