MDR1polymorphisms are associated with inflammatory bowel disease in a cohort of Croatian IBD patients

Inflammatory bowel disease (IBD) encompasses Crohn's disease (CD) and ulcerative colitis (UC). Both CD and UC are chronic inflammatory disorders of the gastrointestinal tract with a combined incidence of 2–20 per 100,000 in the developed world [1]. In recent years, considerable progress in the field of IBD genetics has led to identification of a number of genetic factors involved in the pathogenesis of the disease. However, the underlying pathogenesis of IBD remains unclear. It is presumed that both disorders result as a consequence of defective mucosal barrier and a dysregulated immune response to a host's microbiota in genetically susceptible individual. In that respect, the multidrug resistance 1 (MDR1) gene, also known as adenosine triphosphate-binding cassette superfamily member 1 (ABCB1) is an interesting candidate gene for IBD for several reasons. MDR1 encodes a transmembrane protein, P-glycoprotein 170 (Pgp) which functions as an ATP-dependent efflux transporter pump and is highly expressed in the epithelial surfaces of the intestine, biliary ductules, proximal tubules of kidneys and central nervous system [2–4]. Although its physiological substrate remains unknown, numerous, structurally different compounds, have been identified as a substrates of Pgp [5]. In the gut, Pgp is constitutively expressed on the apical surfaces of the superficial columnar epithelial cells in the intestine with levels of expression gradually rising from the duodenum to the distal parts of the intestine with the highest levels of expression in the distal small bowel and colon [6, 7]. Although it physiological function in the gut remains unclear, such high levels of expression suggest not only a role in protection against xenobiotics but possibly a role in modulating host-bacterial interactions.

First clue to a possible role od MDR1 in IBD pathogenesis comes from a mouse model in which mdr1a knock-out mice develop colitis resembling ulcerative colitis in a specific, pathogen free enviroment, that can be alleviated by a course of antibiotics [8]. Furthermore, MDR1 is located within a region of suggestive IBD linkage on chromosome 7q [9]. Finally, MDR1 gene expression is significantly reduced in the colonic tissue of UC patients [10]. These findings combined make MDR1 an excellent both positional and functional IBD candidate gene.

In summary, MDR1 is an excellent positional and functional candidate gene for IBD. However, inconsistent results of previous studies make its role in IBD controversial. It has to be noted that a vast majority of association studies performed in Caucasians were done in North European populations known to have a higher incidence of IBD compared to regions of the Southern Europe suggesting possible genetic differences in these populations. Therefore, we aimed to investigate the association between MDR1 G2677T/A and C3435T SNP and susceptibillity for UC and CD in an independent cohort of Croatian IBD patients and healthy controls.

In this study, we investigated the association of MDR1 polymorphisms and IBD in a cohort of well charaterized patients. We detected a significant association of C3435T and G2677T polymorphism and UC with higher frequency of mutant allele and/or genotype carriers in the UC patient group. In that respect, our results are in accordance with previously published results reporting the association of UC with G2677T and C3435T SNP [15, 17, 20, 21, 23]. Contrary to the previously published results, we detected no association of C3435T variant with pancolitis although a large proportion of our UC patient group had pancolitis (69.4%) [21]. Furthermore, due to the high LD between the two studied SNP's, we were able to investigate the association with two-locus haplotypes. Previous studies reported both risk and protective haplotypes in the same cohort suggesting a complex influence of MDR1 in pathogenesis of disease [20, 21]. We detected a significantly higher frequency of 3435T/2677T haplotype carriers in the UC patient group. The same haplotype is also a part of two haplotypes associated with UC in the study by Potocnik et al. performed on Slovenian IBD patients [20].

We found no significant difference in allele distributions of investigated SNP's between CD patient group and controls. However, we detected a significant association of C3435T heterozygous carriers with CD. Carriers of this genotype were significantly less represented in the CD patient group suggesting a protective effect. How exactly a heterozygous state confers protection is hard to explain at this moment. However, the same genotype was also found to be protective in UC in the study by Huebner et al. [30]. In this study the investigators put forward an interesting hypothesis that heterozygous carriers of certain MDR1 variants are more resistant to certain bacterial infections [30]. This hypothesis is based on the data from other diseases, like cystic fibrosis or sickle cell anemia, where heterozygous advantage exists [30]. However, although appealing, this hypothesis still needs to be tested, especially on a functional level.

The G2677T polymorphism is a coding polymorphism resulting in substitution of alanine by serine in the Pgp amino acid sequence. Studies investigating the influence of this polymorphism on Pgp expression have reached conflicting results with Kim et al. reporting association of 2677T allele with higher levels of Pgp expression [14]. However, Kimchi-Sarfaty et al. reported no influence of this polymorphism on expression level and intracellular localisation of Pgp [37]. However, G2677T polymorphism might exert its effect not by influencing levels of expression of Pgp but by altering Pgp stability or affinity for a certain substrate. On the other hand, C3435T has been associated with low levels of Pgp expression in Caucasians [11]. Interestingly, the opposite was found in Japanese subjects emphasizing the interethnical differences in Pgp expression [12]. The association of C3435T with low levels of Pgp expression in caucasians is in concordance with data from animal models with mdr1^−/− developing colitis. Interestingly, C3435T is a synonymous SNP not leading to change in amino acid sequence. The possible explanation of the association of a synonymous SNP with altered levels of Pgp expression was offered by finding of moderate reduction of Pgp mRNA stability associated with this SNP [38]. Another possible explanation lies in the high levels of linkage disequilibrium within the MDR1 gene with C3435T in linkage with the true causal variant [5].

In general, our results seem to be in accordance with some of the previously published studies [15, 17, 20, 21, 23]. However, in a multitude of studies no association of investigated SNP's and UC was found [16, 18, 19, 22, 25, 27, 29]. The reason for such variability of results in the literature is not clear, however population heterogeneity seems to be the most plausible explanation. In the published studies frequency of the 3435T allele varied from 41%-65.3% in the UC patient group while frequency of 2677T allele varied from 40%-53% [15–30]. Even greater variation of mutant allele frequency can be found in the respective control groups where 3435T allele frequency varied from 37%-63% and 2677T allele frequency varied from 40%-53% [15–30]. The importance of the mutant allele frequency is further emphasized by study by Glas et al. in which the investigators used two separate controls groups. Performing the association analysis with control group 1 that had a 3435T allele frequency of 52.8% no significant association of the 3435T allele and UC was found. However, when control group 2 was used for analysis with 3435T allele frequency of 43.7 a significant association of 3435T allele and UC was found [18]. Due to the fact that the differences in the minor allele frequencies between the IBD group and the control group are quite modest it is quite possible that differences in the minor allele frequencies of the tested SNP's in the controls groups account for the majority of different results in the literature regarding these two SNP's. Importantly, the frequency of 2677T allele in our control group was 35.8% which represents the lowest frequency published to date in a European population. Interestingly, the frequency of 2677T allele, in the ethnically similar, Slovenian control group in the study by Potocnik et al. was also among the lowest reported in the literature (40%) [20]. Our study further emphasizes differences between various European populations and underscores the importance of replication of results of genetic studies in different ethnic groups.

Finally, our study had several weaknesses. IBD patients included in the study represent the IBD population of a tertiary refferal center. As a consequence, majority of patients have extensive disease, especially in the UC group. Therefore, we can not exclude the possibility of bias since our cohort does not include a large number of patients with limited disease of low activity. Furthermore, although we detected a significant association of tested SNP's with UC and CD, the magnitude of the effect is modest. Given our relatively small sample size, our study had only 43% power to detect the effect of this magnitude. Taking this into account, the results of our study, although generally in accordance with previous reports, must be interpreted with caution.

In conclusion, our results confirm the role of MDR1 in both CD and especially in UC pathogenesis although the overall influence of MDR1 variants on disease susceptibility is modest.