Analysis of IL12B Gene Variants in Inflammatory Bowel Disease

Background IL12B encodes the p40 subunit of IL-12, which is also part of IL-23. Recent genome-wide association studies identified IL12B and IL23R as susceptibility genes for inflammatory bowel disease (IBD). However, the phenotypic effects and potential gene-gene interactions of IL12B variants are largely unknown. Methodology/Principal Findings We analyzed IL12B gene variants regarding association with Crohn's disease (CD) and ulcerative colitis (UC). Genomic DNA from 2196 individuals including 913 CD patients, 318 UC patients and 965 healthy, unrelated controls was analyzed for four SNPs in the IL12B gene region (rs3212227, rs17860508, rs10045431, rs6887695). Our analysis revealed an association of the IL12B SNP rs6887695 with susceptibility to IBD (p = 0.035; OR 1.15 [95% CI 1.01–1.31] including a trend for rs6887695 for association with CD (OR 1.41; [0.99–1.31], p = 0.066) and UC (OR 1.18 [0.97–1.43], p = 0.092). CD patients, who were homozygous C/C carriers of this SNP, had significantly more often non-stricturing, non-penetrating disease than carriers of the G allele (p = 6.8×10−5; OR = 2.84, 95% CI 1.66–4.84), while C/C homozygous UC patients had less often extensive colitis than G allele carriers (p = 0.029; OR = 0.36, 95% CI 0.14–0.92). In silico analysis predicted stronger binding of the minor C allele of rs6887695 to the transcription factor RORα which is involved in Th17 differentiation. Differences regarding the binding to the major and minor allele sequence of rs6887695 were also predicted for the transcription factors HSF1, HSF2, MZF1 and Oct-1. Epistasis analysis revealed weak epistasis of the IL12B SNP rs6887695 with several SNPs (rs11889341, rs7574865, rs7568275, rs8179673, rs10181656, rs7582694) in the STAT4 gene which encodes the major IL-12 downstream transcription factor STAT4 (p<0.05) but there was no epistasis between IL23R and IL12B variants. Conclusions/Significance The IL12B SNP rs6887695 modulates the susceptibility and the phenotype of IBD, although the effect on IBD susceptibilty is less pronounced than that of IL23R gene variants.


Introduction
The identification of the IL-23/Th17 pathway as a key regulator of intestinal immune homeostasis and proinflammatory responses in defense to microbial infection has elucidated new potential therapeutic targets in inflammatory bowel diseases (IBD) [1,2,3,4]. Genome-wide association studies (GWAS) and large cohort studies demonstrated that IL23R [3,5,6,7] and additional genes involved in Th17 differentiation (e.g., IL12B, JAK2, TYK2, STAT3, CCR6, IL2/IL21 and TNFSF15) are associated with the susceptibility to Crohn's disease (CD) and partly also to ulcerative colitis (UC) [5,8,9,10,11]. Moreover, a pathway analysis using data from the Wellcome Trust Case Control Consortium (WTCCC) uncovered significant associations of CD and IL-12/IL-23 pathway components, harbouring 20 genes such as IL12B, JAK2, STAT3 and CCR6 [12]. Functionally, the proinflammatory cytokines IL-12 and IL-23 play critical and unique roles in bridging the innate and adaptive immune systems in IBD [4,13] and are produced primarily by activated dendritic cells (DCs) and macrophages in response to microbial stimulation. IL-12 promotes the differentiation of naive CD4+ T cells into mature interferon-c (IFN-c)-producing Th1 effector cells and is a potent stimulus of natural killer and CD8+ T cells [14,15]. In contrast, IL-23, a heterodimeric cytokine composed of a p19 subunit and a p40 subunit of which the latter is shared with IL-12, is required for the generation of memory T cells and drives differentiation of Th17 cells [16,17]. Th17 cytokines such as IL-17A, IL-17F, IL-21, IL-22, IL-26 and the Th17 chemokine CCL20 have been particularly implicated as key cytokines involved in the intestinal inflammation of CD [1,18,19,20,21,22,23].
On the genetic level, IL12B encodes the IL-12 p40 subunit shared by IL-12 and IL-23 cytokines while IL23R encodes one of the two subunits of the IL-23 receptor [24]. Antibody therapy directed against the IL-12/IL-23 p40 subunit demonstrated some clinical efficacy in CD patients [25]. Since the first identification of IL23R as susceptibility gene in IBD by Duerr et al. [3], various GWAS and cohort studies have confirmed IL23R not only as a major susceptibility gene in IBD but also in the pathogenesis of other autoimmune diseases such as psoriasis [26,27] and ankylosing spondylitis [28], implicating common proinflammatory pathways. Genetic variants in the IL12B region have also been associated with the susceptibility to psoriasis, ankylosing spondylitis, and infectious diseases such as leprosy and tuberculosis [29,30,31,32,33]. In contrast, data on IL12B variants and their role in asthma [34,35], rheumatoid arthritis [36] or multiple sclerosis [37] remain controversial. While recent GWAS metaanalyses by Barrett et al. [5], Franke et al. [9], and Anderson et al. [10] established IL12B as IBD susceptibility gene, smaller studies showed inconsistent results [38,39]. Cohort studies in the Spanish [38] and Japanese [39] population demonstrated an association of IL12B SNPs with the susceptibility to IBD; however with different results, reporting associations with CD susceptibility in the Japanese cohort (rs6887695) [39] and to UC (rs6887695) but not to CD susceptibility in the Spanish cohort [38]. Moreover, the phenotypic effects of IL12B and potential gene-gene interactions contributing to IBD susceptibility are largely unknown.
Since the exact role of IL12B for IBD susceptibility in the German population and IBD phenotype behaviour remains unclear, we aimed to perform a detailed genotype-phenotype analysis in a large IBD cohort and an analysis for potential epistatic interactions with other gene variants involved in IL-12 and IL-23 signaling and implicated in IBD susceptibility such as IL23R and STAT4 [3,40] which encodes the major IL-12-induced downstream transcription factor STAT4.

Ethics statement
The study was approved by the Ethics committee of the Medical Faculty of Ludwig-Maximilians-University Munich and written, informed consent was obtained from all patients prior to the study. Study protocols adhered to the ethical principles for medical research involving human subjects of the Helsinki Declaration (as detailed under: http://www.wma.net/en/30publications/10policies/ b3/index.html).

Study population and IBD phenotype assessment
Overall, the study population (n = 2196) consisted of 1231 IBD patients including 913 patients with CD, 318 patients with UC, and 965 healthy, unrelated controls, all of Caucasian origin. The demographic and clinical data (behaviour and location of IBD, disease-related complications, immunosuppressive therapy and history of previous surgeries) of the patients were recorded by analysis of patient charts and a detailed questionnaire including an interview at time of enrolment. The diagnosis of CD or UC was determined according to established guidelines based on endoscopic, radiological, and histopathological criteria. Patients with CD were assessed following the Montreal classification based on the age at diagnosis (A), location (L), and behaviour (B) of the disease [41]. In patients with UC, anatomic location was also based on the Montreal classification using the criteria ulcerative proctitis (E1), left-sided UC (distal UC; E2), and extensive UC (pancolitis; E3). Patients with indeterminate colitis were excluded from the study. The demographic characteristics of the IBD study population were collected blind to the results of the genotype analyses and are summarized in Table 1.

DNA extraction and genotyping of the IL12B variants
From all study participants, blood samples were taken and genomic DNA was isolated from peripheral blood leukocytes using the DNA blood mini kit from Qiagen (Hilden, Germany) according to the manufacturer's guidelines. Four IL12B SNPs (rs3212227, rs17860508, rs10045431 and rs6887695) were genotyped by PCR and melting curve analysis using a pair of fluorescence resonance energy transfer (FRET) probes in a LightCyclerH 480 Instrument (Roche Diagnostics, Mannheim, Germany) as described in previous studies [6,40,42,43]. These IL12B SNPs were selected based on previous studies showing associations with CD or other autoimmune diseases such as psoriasis. Specifically, the SNPs rs3212227 and rs6887695 were selected from the study of Cargill and co-workers [29], while rs6887695 was also investigated in the study of Parkes et al. [8]. The SNP rs10045431 was selected from the study of Barrett and co-workers [5]. Additionally, the SNP rs17860508 located within the promoter of IL12B was included because of its potential functional relevance regarding gene expression [44]. The total volume of the PCR was 5 ml containing 25 ng of genomic DNA, 1| Light Cycler 480 Genotyping Master Mix (Roche Diagnostics), 2.5 pmol of each primer and 0.75 pmol of each FRET probe (TIB MOLBIOL, Berlin, Germany). In the case of rs3212227 and rs10045431, the concentrations of the forward primers and in the case of rs17860508 the concentration of the reverse primer was reduced to 0.5 pmol. In the case of rs6887695 the concentration of the reverse primer was reduced to 1.25 pmol and in the case of rs744166 the concentration of the forward primer was reduced to 0.75 pmol, respectively. Two SNPs were analyzed in a multiplex reaction, the combinations were rs3212227+rs17860508 and rs10045431+rs6887695. The PCR comprised an initial denaturation step (95uC for 10 min) and 45 cycles (95uC for 10uC sec, 60 for 10 sec, 72uC for 15 sec). The melting curve analysis comprised an initial denaturation step (95uC for 1 min), a step rapidly lowering the temperature to 40uC and holding for 2 min, and a heating step slowly (1 acquisition/uC) increasing the temperature up to 95uC and continuously measuring the fluorescence intensity.
The results of the melting curve analysis were confirmed by analyzing two patient samples for each possible genotype using sequence analysis. For sequencing, the total volume of the PCR was 100 ml containing 250 ng of genomic DNA, 1| PCR-buffer (Qiagen, Hilden, Germany), a final MgCl 2 concentration of 2 mM, 0.5 mM of a dNTP mix (Sigma, Steinheim, Germany), 2.5 units of HotStar Plus Taq TM DNA polymerase (Qiagen) and 10 pmol of each primer (TIB MOLBIOL). The PCR comprised an initial denaturation step (95uC for 5 min), 35 cycles (denaturation at 94uC for 30 sec, primer annealing at 65uC for 30 sec, extension at 72uC for 30 sec) and a final extension step (72uC for 10 min). The PCR products were purified using the QIAquick PCR Purification Kit (Qiagen) and sequenced by a commercial sequencing company (Sequiserve, Vaterstetten, Germany). All sequences of primers and FRET probes and primer annealing temperatures used for genotyping and for sequence analysis, respectively, are given in Supplementary Tables S1 and S2.

In silico analysis of transcription factor binding sites
SNPs rs3212227, rs17860508, rs10045431 and rs6887695 were analyzed for potential transcription factor binding sites applying the online tool TFSEARCH (http://www.cbrc.jp/research/db/ TFSEARCH.html). It is based on the TRANSFAC database developed at GBF Braunschweig, Germany [45]. The threshold score for binding sites was set to 75.0 (score = 100.0 * ('weighted sum'2min)/(max2min); max. score = 100). For each SNP, major and minor alleles including the flanking sequences 15 bp upstream and downstream were analyzed. Only human transcription factors were included in the analysis.

Statistical analyses
Power calculation was performed using the Genetic Power Calculator (http://pngu.mgh.harvard.edu/,purcell/gpc/). Each genetic marker was tested for Hardy-Weinberg equilibrium in the control population. For data evaluation, we used PLINK (http:// pngu.mgh.harvard.edu/,purcell/plink/) and R-2.13.1. (http:// cran.r-project.org). Haplotypes tests were performed in a slidingwindow approach including up to 4 markers. Epistasis between different SNPs was tested using the -epistasis option on PLINK. Genotype-phenotype associations were assessed using logistic regression.

Results
The IL12B polymorphisms rs6887695 is associated with increased IBD susceptibility In all three subgroups (CD, UC, and controls), the allele frequencies of the IL12B SNPs (rs3212227, rs17860508, rs10045431, rs6887695) were in accordance with the predicted Hardy-Weinberg equilibrium (

IL12B haplotype analysis
We next performed a detailed haplotype analysis in our IBD cohort. As shown in Tables 3 and 4, we could not demonstrate significant associations of IL12B haplotypes with CD and UC susceptibility. A haplotype of all four investigated SNPs (rs3212227-rs17860508-rs10045431-rs6887695) showed a trend for association with CD (p = 0.053); however, this potential association signal did not reach statistical significance and was uncorrected for multiple testing.
Genotype-phenotype analysis: C/C homozygosity for the SNP rs6887695 is associated with non-stricturing, non-penetrating Crohn's disease The majority of previous GWAS meta-analyses showing an association with IL12B were performed in CD patients with predominant ileal involvement [5,9]. To exclude that only a certain IBD subphenotype such as ileal CD is associated with IL12B, we further investigated potential associations of IL12B SNPs with the anatomic location in IBD patients Genotypephenotype analysis showed a weak association of rs3212227 with colonic CD (p = 0.04) but not with ileal involvement. In addition, we analyzed the IL12B SNP rs6887695, for which we found a trend for association with CD and UC, regarding potential phenotypic consequences in CD and UC. CD patients, who were homozygous C/C carriers of this SNP, had significantly more often a non-stricturing, non-penetrating disease phenotype than carriers of the G allele ( = combined group of heterozygous C/G and wildtype G/G carriers; p = 6.8610 25 (Table 5). This strong association remained significant following Bonferroni correction for multiple testing. C/C homozygous carriers of the IL12B SNP rs6887695 had also less stenoses than G carriers (p = 0.038) and there was a trend towards less penetrating disease (p = 0.062), less fistulas (p = 0.085), and less surgery (p = 0.063; Table 5). In UC, homozygous C/C carriers of the IL12B SNP rs6887695 had significantly more often left-sided UC (p = 0.006) but less often extensive UC (p = 0.029) than the combined group of heterozygous C/G and wildtype G/G carriers (Table 6).
Analysis for epistasis of IL12B with IL23R and STAT4 gene variants regarding susceptibility to Crohn's disease Next, we analyzed potential evidence for epistasis of IL12B variants with other CD susceptibility genes involved in IL-12 and Th17 signaling, including CD-associated variants in the IL23R and STAT4 gene [3,6,40]. Analysis for gene-gene interaction revealed weak epistasis of IL12B SNP rs6887695 with 7 out of the 8 analyzed STAT4 gene variants (p,0.05) regarding CD susceptibility (Table 7). However, following Bonferroni correction, none of these associations remained significant. In addition, there was no evidence for epistasis between IL23R and IL12B variants ( Table 7).
In silico analysis of IL12B SNPs identifies potential differences in transcription factor binding to major and minor alleles To analyze if SNPs in the IL12B region modify the binding of transcription factors to DNA and thereby modulating gene expression, we analyzed SNPs rs3212227, rs17860508, rs10045431 and rs6887695 including the surrounding sequences for potential binding sites as described in the methods section. The results are shown in table 8. Interestingly, the highest differences in binding scores for the major and the minor alleles was found for rs6887695, the only SNP that was associated with overall IBD susceptibility in our study cohort. Moreover, in homozygous carriers of the minor allele of rs6887695, there were significant associations with specific CD and UC phenotypes. While the transcription factors HSF1, HSF2, MZF1 and Oct-1 were predicted to bind with very high probability to the sequence comprising the major G allele, predicted  binding to the minor C allele was substantially lower (Table 8). On the other hand, the binding score for the transcription factor RORa was higher for the minor C allele (Table 8). Therefore, differential DNA binding of transcription factors in this genomic region followed by differential gene transcription might be one reason for the diseasemodifying abilities of this SNP that we observed in our genotypephenotype analysis.

Discussion
This study presents a detailed genotype-phenotype analysis investigating IL12B SNPs as potential susceptibility gene variants in a large Caucasian IBD cohort. A major focus of this study was the analysis of potential epistatic interactions with key gene variants of the IL-12 and IL-23/Th17 pathway such as IL23R and STAT4. We demonstrated that the IL12B variant rs6887695 is weakly associated with overall IBD susceptibility (p = 0.035), with trends for association with both CD and UC susceptibility. Moreover, a haplotype block formed by the four investigated IL12B SNPs showed a trend for association with CD (p = 0.05). Considering our findings and the strong association signal for this gene in the recent GWAS [9,10], IL12B can be regarded as established IBD susceptibility gene. In contrast to the very large IBD GWAS, our sample size (n = 2196) was smaller, resulting in limited power, particularly for the UC cohort. For example, a power calculation, which assumed an OR of 1.2 and an allele frequency of 0.20 for the rarest IL12B variant rs3212227, demonstrated that our study had 63.1% power to detect a nominal significant finding (alpha = 0.05) in the CD cohort but had only a power of 38.2% in the UC cohort. However, based on the results of this study, the strength of a potential association of IL12B with IBD is several log-fold weaker than that shown for Table 5. Genotype-phenotype-analysis of SNP rs6887695 in patients with Crohn's disease (CD).

IL12B
(1) CC (2)  P CC: P-value for testing for differences between homozygous carriers of the C allele (C/C) and heterozygous and non-carriers of the C allele. OR CC : corresponding odds ratios and 95% confidence intervals (95% CI). Significant P-values (,0.05) are depicted in bold, P-values showing a trend towards significance are depicted in Italic fonts. P-values marked with an asterisk * remained significant after Bonferroni correction. 1 Disease behaviour was defined according to the Montreal classification. A stricturing disease phenotype was defined as presence of stenosis without penetrating disease. The diagnosis of stenosis was made surgically, endoscopically, or radiologically (using MRI enteroclysis). 2 Immunosuppressive agents included azathioprine, 6-mercaptopurine, 6-thioguanin, methotrexate, infliximab and/or adalimumab. 3 Only surgery related to CD-specific problems (e.g. fistulectomy, colectomy, ileostomy) was included. doi:10.1371/journal.pone.0034349.t005 IL23R in our cohort [6] and other major CD susceptibility genes such as NOD2 and ATG16L1 (Supplemental Table S3), suggesting a more important role for IL-23R than IL-12/23 p40 in the genetic susceptibility to IBD. This is in agreement with the recent IBD meta-analyses [9,10], which showed much stronger IBD association signals for IL23R than for IL12B. Similarly, smaller studies failed to show an association of IL12B SNPs with IBD or showed only weak associations with CD or UC [46,47,48] (see Supplemental Table S4 for an overview on published studies on IL12B in IBD). Even large studies such as the study by Festen et al., which included 1,455 UC patients and 1,902 controls, was unable to show an association of IL12B with UC [49], further demonstrating that very large patient cohorts are necessary to show convincing associations with this gene locus. Therefore, IL23R is a more important genetic modifier of IBD susceptibility than IL12B, suggesting a more important pathogenic role of Th17 cells, which express IL-23R, than Th1 cells, which develop under the control of IL-12.
Currently, there are limited data on how IL12B may functionally influence IBD susceptibility in humans. A very recent study in mice demonstrated that a polymorphism in the coding region of murine Il12b promotes IL-12p70 and IL-23 heterodimer formation [50]. The authors hypothesized that the high synthesis rate of IL-12/23 cytokines resulting from efficient binding may lead to rapid proinflammatory skewing of immune responses and disturbance of the homeostatic balance resulting in higher susceptibility for IBD [50]. Other gene variants may also modulate IL-12 expression as recently demonstrated by our group showing increased basal levels of IL-12p40 in CD patients with two mutated NOD2 alleles [51]. Currently, it is unclear which IL12B SNP is the ''true'' disease-causing variant. We therefore performed a detailed in silico analysis of IL12B regarding potential transcription factor binding sites and demonstrated for Table 6. Genotype-phenotype-analysis of SNP rs6887695 in patients with ulcerative colitis (UC) for which detailed phenotypic data based on the Montreal classification was available.

IL12B
(1) CC (2)  For each variable, the number of patients included is given. P CC P-value for testing for differences between homozygous carriers of the C allele and heterozygous/noncarriers of the C allele. OR CC : corresponding odds ratios and 95% confidence intervals (95% CI). Significant P-values (,0.05) are depicted in bold fonts. P-values marked with an asterisk remained significant following Bonferroni correction. doi:10.1371/journal.pone.0034349.t006 Table 7. Analysis for gene-gene interaction of IL12B with IL23R and STAT4 variants, respectively, regarding susceptibility to Crohn's disease (CD).  Binding scores differing more than 5 points between major and minor alleles are depicted in bold. Scores differing more than 10 points are depicted in bold italic. The binding score threshold for each allele was set to 75.0. {Different consensus sequences for the same transcription factor are caused by the deduction of the sequences from different matrices in the TRANSFAC database [45]. *For rs17860508, more than two alleles exist. For the major allele, the score comprises the highest score from all four non-minor alleles (-, G, GC, TTAGA). & = score for the GC allele; # = score for the (-) allele; 1 = score for the TTAGA allele. UTR = untranslated region; Nucleotide codes: K = G or T, M = A or C, R = A or G, S = C or G, W = A or T, N = A, G, C or T. doi:10.1371/journal.pone.0034349.t008 rs6887695 (out of the four analyzed IL12B SNPs) the highest likelihood of changes in transcription factor binding. This is consistent with the results of our genotype analysis in which only rs6887695 was associated with overall IBD susceptibility and potential epistasis with STAT4 SNPs. For rs6887695, the in silico analysis predicted changes in the binding of the transcription factors HSF1, HSF2, MZF1, Oct-1 and RORa. HSF1 plays a role in protecting against DSSinduced colitis [52]. MZF1 and Oct-1 are transcriptional regulators of SERPINA3 [53,54] which has been implicated as one of the epithelium-derived genes involved in the antibacterial defense [55]. In addition, our in silico analysis predicted for the transcription factor RORa stronger binding to the minor C allele. Interestingly, RORa plays together with RORct a key role in the development of Th17 cells which are involved in the pathogenesis of CD and UC [1] and may explain the association of rs6887695 with increased IBD susceptibility found in our study. Although Th17 cells are recognized as proinflammatory T cell population, their major effector cytokines IL-17A and IL-22 may have also protective functions under certain circumstances [19,56,57,58,59]. Similarly, studies indicated for RORa not only proinflammatory effects but also a role as negative regulator of inflammatory responses [60]. For example, RORa inhibits TNFa-induced IL-6, IL-8 and COX-2 expression in human primary smooth-muscle cells by negatively interfering with the NF-kB signaling pathway and reducing p65 translocation [60]. This may correlate with our genotype-phenotype analysis which demonstrated a less severe disease phenotype for homozygous C/C carriers of the IL12B SNP rs6887695. CD patients, who were homozygous C/C carriers of this SNP, had significantly more often non-stricturing, non-penetrating disease than carriers of the G allele (p = 6.8610 25 ), while C/C homozygous UC patients had less often extensive colitis than G allele carriers (p = 0.029).
IL12B is another example for a common susceptibility gene for CD and UC which is supported by the results of our study showing a trend for rs6887695 for association with both CD and UC. Out of 99 currently known IBD susceptibility gene loci (71 in CD and 47 in UC), at least 28 susceptibility loci are shared between CD and UC [9,10]. Remarkably, the strongest cluster of common CD and UC susceptibility genes is formed by genes related to the IL-23/Th17 pathway including IL12B [61]. Similar to IL23R, IL12B is also a shared susceptibility gene with other IBD-associated diseases such as psoriasis [29] and ankylosing spondylitis [28], providing an explanation of the increased incidence of these extraintestinal manifestations in IBD patients.
The initial GWAS focused on CD patients with ileal CD. To exclude that the associations found in recent GWAS meta-analysis are based only on certain IBD subphenotypes, we performed a detailed genotype-phenotype analysis focusing also on the anatomic disease localization. Interestingly, genotype-phenotype analysis showed an association of IL12B rs3212227 with colonic CD but not with ileal CD. It is likely that many of the susceptibility loci, which are shared by CD and UC, may predispose to a common phenotype such as colonic (and not ileal) IBD. This is supported by the association of the IL12B SNP rs3212227 with colonic CD and the trend for an association signal of rs6887695 with both CD and UC susceptibility found in our study. Similarly, in another study from New Zealand, carriers of the minor C allele of the IL12B SNP rs6887695 had a decreased risk of ileal disease [48]. A recent study analyzing the same SNP (rs6887695) suggested additional environmental triggers regarding the risk of this IL12B SNP on CD susceptibility. Analyzing differences in associated genes between smoking and non-smoking CD patients, it implicated a complex gene-environment interaction demonstrating an association of IL12B SNP rs6887695 in non-smoking, but not in smoking patients [62].
In addition, our analysis of potential gene-gene interactions revealed weak epistasis of IL12B SNP rs6887695 with 7 out of 8 analyzed STAT4 gene variants (p,0.05). This is highly interesting, given that STAT4 is the major downstream transcription factor of IL-12. However, considering our limited sample size and the borderline significance of this interaction, which was lost after Bonferroni correction, this potential gene-gene interaction needs further analysis in very large cohorts or GWAS meta-analyses. However, in contrast to our hypothesis-driven study, which analyzed SNPs in the IL12B gene encoding IL-12 p40 and the gene STAT4 encoding the major IL-12 downstream transcription factor STAT4, GWAS are limited by the problem of multiple testing to a much greater extent than our study. Given the large number of SNPs analyzed in GWAS, the correction factor (to correct for multiple testing) is much higher than in our study, which may result in the elimination of potentially true gene-gene interactions. Moreover, in the study by Anderson et al. no formal testing for epistasis has been performed [10]. In addition, in the manuscript by Franke et al. [9], only epistasis between the 71 significantly associated CD loci, which did not include STAT4, was analyzed.
In summary, in contrast to Th17 cell-modifying and strongly IBD-associated IL23R gene variants, IL12B variants have a lesser role in the susceptibility to CD and UC in the German population, suggesting a more important role of IL-23Rexpressing Th17 cells than Th1 cells in the CD pathogenesis. This observation is supported by similar results of large GWAS [5,9,10], in which IL23R showed stronger associations with IBD susceptibility than IL12B, although these GWAS clearly established IL12B as IBD susceptibility gene. Our IL12B in silico analysis and the results of our genotype analysis suggest rs6887695 as likely disease-causing IL12B variant. Homozygous C/C carriers of this SNP were protected against stricturing and penetrating CD and showed less often extensive UC which may be related to the alteration of the binding of certain transcription factors such as RORa as predicted by our in silico analysis. We demonstrated for rs688769 potential epistasis with STAT4, encoding the major downstream transcription factor of IL-12 which would link IL-12 and STAT4 for the first time on a genetic level. Moreover, similar to an association of the STAT4 SNP rs7574865 with colonic CD [40], we found an association of the IL12B SNP rs3212227 with an increased risk for colonic CD. Based on the gene-gene interaction found for IL12B and STAT4 in this study, large meta-analyses (e.g., by the International IBD Genetics Consortium), which may include our data should further investigate potential epistasis between the main IBD susceptibility genes as major pathomechanism in the disease pathogenesis. Further fine mapping and functional studies are required to clarify the ''true'' disease-causing IL12B variant and its pathogenic role in IBD susceptibility.

Supporting Information
Table S1 Primer sequences (F: forward primer, R: reverse Primer), FRET probe sequences, and primer annealing temperatures used for genotyping of IL12B variants. Note: FL:

Table S3
Comparison of the association signals of IL12B with the association signals of the three most strongly CD-associated genes NOD2, IL23R and ATG16L1 in the Munich IBD casecontrol cohort. Minor allele frequencies (MAF), allelic test Pvalues, and odds ratios (OR, shown for the minor allele) with 95% confidence intervals (CI) are depicted for both the CD and UC case-control cohorts. Details on the analyses of NOD2, IL23R and ATG16L1 gene variants were published in previous studies. (DOC)