Association study of TNFSF15 polymorphisms in Japanese patients with inflammatory bowel disease

Tumour necrosis factor superfamily (TNFSF) 15 is a novel member of the TNFSF and its mRNA and protein expression is upregulated in inflammatory bowel disease (IBD), particularly in Crohn's disease (CD).¹,² Recently, Yamazaki et al performed a large scale case control study using single nucleotide polymorphism (SNP) markers and reported that polymorphisms in TNFSF15 conferred susceptibility to CD in both Japanese and UK populations.³ They also suggested a potential association between a Caucasian ulcerative colitis (UC) cohort and TNFSF15, but this association was not studied in Japanese patients. To investigate this possible association between TNFSF15 and Japanese UC, and to replicate this association with CD in Japanese, we performed a case control association study in Japanese patients with CD and UC.

We selected six SNPs—tnfsf15_26, tnfsf15_28, tnfsf15_31, tnfsf15_33, tnfsf15_35, tnfsf15_36—which were reported to show a strong association (p<10⁻¹⁰), and genotyped these six SNPs in 286 patients with CD, 263 patients with UC, and 277 healthy controls (HCs) by polymerase chain reaction‐restriction fragment length polymorphism (PCR‐RFLP) analysis. All patients and HCs were Japanese, and none had a family history of IBD.

We replicated all six SNPs that were significantly associated with the Japanese CD cohort (table 1). On the other hand, in contrast with Caucasians,³ none of the SNPs were associated with the Japanese UC cohort. Risk allele frequencies of tnfsf15_35, tnfsf15_36, and tnfsf15_31 were higher in the CD group with anal lesions than in those without, but the associations were statistically weak (p = 0.019, 0.019, and 0.037, respectively). No significant differences were found in allele frequencies between the CD subgroups classified by age at diagnosis, location of disease, existence of fistula, stenosis, need for steroid therapy, and past history of surgical treatment, and also the UC subgroups classified by age at diagnosis, extent of disease, need for intensive intravenous steroid therapy, and need for surgical treatment.

Table 1 Allele frequencies of TNFSF15.

SNP	dbSNP	Location	Allele1	Allele2	Frequency of allele2 (%)/OR/p value
					HCs (n = 277)	CD (n = 286)	UC (n = 263)
tnfsf15_36	rs7848647	5′‐flanking region	G	A	39.0	25.7 (OR = 1.85, p = 1.84×10−6)	36.3 (NS)
tnfsf15_35	rs6478109	5′‐flanking region	C	T	39.0	25.7 (OR = 1.85, p = 1.84×10−6)	36.3 (NS)
tnfsf15_33	rs6478108	Intron 1	A	G	51.8	37.6 (OR = 1.78, p = 1.60×10−6)	52.3 (NS)
tnfsf15_31	rs4979462	Intron 1	A	G	39.5	26.2 (OR = 1.84, p = 1.98×10−6)	36.5 (NS)
tnfsf15_28	–	Intron 3	C	T	53.1	38.8 (OR = 1.78, p = 1.58×10−6)	52.3 (NS)
tnfsf15_26	rs3810936	Exon 4 (Val201Val)	G	A	40.3	26.9 (OR = 1.83, p = 2.15×10−6)	36.3 (NS)

SNP, single nucleotide polymorphism; HCs, healthy controls; CD, Crohn's disease; UC, ulcerative colitis; NS, no significant. OR, odds ratio.

In this study, we confirmed the findings of a previous report concerning a significant association between TNFSF15 and CD. On the other hand, no evidence for an association with Japanese UC was observed, although a potential association with Caucasian UC was reported. It is generally accepted that UC and CD may share some susceptibility genes. Ethnic differences in genetic susceptibility may be explained by differences in the haplotypic background. Some TNFSF15 polymorphisms identified in the Japanese were monomorphic or nearly monomorphic in the Caucasian population.³ Thus it seems likely that population specific patterns of haplotypes may contribute to differences in UC susceptibility.

Although a previous report described that tnfsf15_28 showed the lowest p value and highest odds ratio among the SNPs in TNFSF15,³ our results showed that p values and odds ratios of tnfsf15_36 and tnfsf15_35 were similar to those of tnfsf15_28. Thus to identify the pathogenic SNP, a functional study is clearly needed. We analysed the transcription factor binding sites in the promoter region of TNFSF15 by TFSEARCH⁴ and found that GATA‐1, 2, and 3 possibly bind the tnfsf15_35‐T allele while the GATA binding cis element is absent in the tnfsf15_35‐C allele (risk associated). It is well known that GATA‐3 promotes a Th2 mediated immunological state and suppresses expression of Th1 mediated cytokines.⁵,⁶ These findings have raised the possibility that GATA‐3 may not bind the promoter region with the tnfsf15_35‐C risk allele resulting in lack of suppression of TNFSF15 expression. Consequently, overexpressed TNFSF15 promotes a Th1 mediated immunological state and initiates or exacerbates the severity of CD. Although we do not have experimental evidence for this hypothesis, we intend to elucidate the functional significance in the future.