Gene Expression-Genotype Analysis Implicates GSDMA, GSDMB, and LRRC3C as Contributors to Inflammatory Bowel Disease Susceptibility

To investigate the biological foundation of the inflammatory bowel disease (IBD), ulcerative colitis and Crohn's disease, susceptibility locus rs2872507, we have investigated the expression of 13 genes using ileal and colonic biopsies from patients with IBD (inflamed and noninflamed mucosa) or from individuals without IBD (noninflamed mucosa). The susceptibility allele was consistently associated with reduced expression of GSDMB (P = 4.1 × 10−3–7.2 × 10−10). The susceptibility allele was also associated with the increased expression of GSDMA (P = 1.6 × 10−4) and LRRC3C (P = 7.8 × 10−6) in colon tissue from individuals without IBD and with the reduced expression of PGAP3 (IBD; P = 2.0 × 10−3) and ZPBP2 (Crohn's disease; P = 7.7 × 10−4) in noninflamed ileum. Inflammation resulted in the reduced colonic expression of ERBB2, GRB7, MIEN1, and PGAP3 (P = 1.0 × 10−4–1.0 × 10−9) and the increased colonic expression of IKZF3 and CSF3 (P = 2.4 × 10−7–3.5 × 10−8). Based on our results and published findings on GSDMA, GSDMB, LRRC3C, and related proteins, we propose that this locus in part affects IBD susceptibility via effects on apoptosis and cell proliferation and believe this hypothesis warrants further experimental investigation.


Introduction
Inflammatory bowel disease (IBD) is a group of intestinal disorders, subdivided into Crohn's disease (CD), ulcerative colitis (UC), and, in the absence of a confident diagnosis, unclassified colitis (IBD-U) [1,2]. Susceptibility to IBD is to a large extent determined by genetic predisposition [3]. Consistent with that, genetic association studies have uncovered a large number of susceptibility loci in relation to IBD. Recently, a total of 163 IBD loci were identified in a genomewide analysis, and 110 of those susceptibility loci were shared between CD and UC [4]. Within each susceptibility locus, candidate risk genes have been prioritized, based largely on bioinformatic evaluations of the relationships among genes, the presence of coding single nucleotide polymorphisms (SNPs), or the investigation of gene expression-genotype correlations [4][5][6].
A linkage disequilibrium block encompassing IKZF3, ZPBP2, GSDMB, and ORMDL3, which was previously associated with childhood-onset asthma [7], has been identified as an IBD susceptibility locus [6,8]. Barrett et al. [6] and McGovern et al. [8] highlighted ORMDL3 in relation to CD and IBD, respectively, based either on a correlation between genotype and gene expression in lymphoblastoid cell lines [7] or on the biological role and previous implication of ORMDL3 in diseases involving dysregulated immune responses. More recent studies of UC and CD have emphasized several genes (GSDMA, GSDMB, IKZF3, ORMDL3, PNMT, and ZPBP2) within the same region, using information based on bioinformatics, the presence of coding SNPs, or correlations between genotype and gene expression [4,5,9].
The aim of this study was to investigate disease relevant intestinal tissue samples in order to further pinpoint the genes involved in IBD susceptibility. The ileal and colorectal  Figure 1: Overview of the genes present within the investigated region. Location of genes in the vicinity of the IBD susceptibility marker rs2872507 (position marked with an upward-pointing arrow) according to the human genome assembly GRCh37/hg19 viewed using the UCSC Genome Browser (http://genome.ucsc.edu/). expression of 13 genes within the genetic region of the shared (CD and UC) susceptibility locus rs2872507 (PGAP3-MED24; Figure 1) was analyzed in relation to genotype, inflammation, and sampling location.

Study Samples.
Colorectal and ileal mucosal biopsy specimens ( = 183) were collected during routine endoscopies of 85 adult patients who were being investigated for a known IBD diagnosis or were in the work-up for suspected gastrointestinal disorders (Table 1). Thirty-three patients not afflicted with IBD and without intestinal inflammation were included as noninflamed, non-IBD controls. Study biopsies were collected in parallel to and from the same areas as the biopsies collected for histopathologic assessment. Each biopsy was categorized as "inflamed" or "noninflamed" based on a compound evaluation of endoscopic findings as assessed by one experienced endoscopist (Sven Almer) and on a routine histopathologic assessment for inflammation.
The biopsy specimens used for RNA purification were immersed in RNAlater RNA stabilization reagent (Qiagen, Hilden, Germany) and stored at 4 ∘ C overnight and at −20 ∘ C thereafter, awaiting RNA purification.

DNA and RNA Purification.
The biopsies were homogenized using TissueRuptor and disposable probes (Qiagen). DNA and RNA were purified using the AllPrep DNA/RNA mini kit (Qiagen) according to the manufacturer's instructions, either manually or using the automated QIAcube system (Qiagen). Concentration and purity were spectrophotometrically measured using a Nanodrop ND-1000 values were normalized to the average of the selected reference genes (S1 Table) using the GenEx software version 5.4.2. The resulting delta-values were further processed to obtain the relative expression in relation to the sample with the lowest expression of each gene.

2.5.
Genotyping. The genotype of the SNP susceptibility marker rs2872507 [5,9] on chromosome 17q12 (http://www .ncbi.nlm.nih.gov/snp) was assessed using 5 ng genomic DNA per sample, a TaqMan SNP genotyping assay (assay ID C 11630970 20), and the TaqMan genotyping master mix (Life Technologies). All genotyping was conducted on the 7500 Fast real-time PCR system using the standard run mode, and the genotypes were generated using the 7500 Fast system SDS software version 2.0.6 (Life Technologies).

Data Analysis.
Reference genes were evaluated for low sample-to-sample variation using the GeNorm [10] and NormFinder [11] algorithms in the GenEx software version 5.4.2.
In order to reduce potential confounding effects on the analysis of genotype and gene expression, the samples were stratified based on inflammatory status (inflamed versus noninflamed) and sampling location (ileum versus colon). Multiple colorectal biopsies within inflamed or noninflamed areas from single individuals were treated as biological replicates (see Effects of Sampling Location on Gene Expression in Results). Additionally, the samples were grouped based on disease status (CD, UC, overall IBD, and non-IBD).
The gene expression was investigated in relation to the genotypes using Spearman's rank correlation test. For group comparisons, the Mann-Whitney test or the Kruskal-Wallis ANOVA were used, as appropriate. A Bonferronicorrected value < 0.00385 was considered significant. Statistical analyses were performed using Statistica version 10.0 (StatSoft Inc., Tulsa, OK, USA).

Results
Only biopsies with concordant results between the macroscopic appearance at endoscopy and the histopathological findings (i.e., "inflamed" or "noninflamed") were used for further analyses.

Correlation between Genotype and Gene Expression.
The susceptibility allele (A) of the rs2872507 locus ( Figure 1) was associated with the reduced expression of Gasdermin B (GSDMB) in intestinal biopsies from patients with IBD; the associations for noninflamed ileum and noninflamed colon were significant and that for inflamed colon was borderline significant (Table 2, Figure 2). The negative correlation between the number of susceptibility alleles and GSDMB expression was also noted separately for CD and UC, although the correlation only reached nominal significance ( < 0.05) in noninflamed ileal biopsies and inflamed colonic biopsies from patients with UC (Table 2). Among the patients with IBD, the expression of GSDMB was at least 2.17fold lower in individuals homozygous for the susceptibility allele compared with that in individuals homozygous for the opposite allele ( Figure 2). The susceptibility allele was further associated with the reduced expression of PGAP3 in noninflamed ileum biopsies from the overall IBD group and with the reduced expression of ZPBP2 in noninflamed ileum biopsies from the CD group (Table 2, Figure 3), with the expression of PGAP3 reduced 1.81-fold (2 versus 0 susceptibility alleles) and that of ZPBP2 reduced 2.27-fold (1 versus 0 susceptibility alleles). A correlation between the number of susceptibility alleles and PGAP3 expression was also identified with nominal significance ( < 0.05) separately for CD and UC.
A negative correlation between GSDMB expression and the number of susceptibility alleles was also present in the noninflamed, non-IBD ileal and colonic biopsy samples (Table 2, Figure 4), and the expression was lower at least 1.84fold in individuals homozygous for the IBD susceptibility allele compared with that in individuals homozygous for the opposite allele. Furthermore, the IBD susceptibility allele was associated with the increased expression of GSDMA and LLRC3C in noninflamed, non-IBD colonic samples (Table 2, Figure 5) and with expression levels increased 5.53-fold and 6.78-fold, respectively, in individuals homozygous for the susceptibility allele compared with those in individuals homozygous for the opposite allele.

Inflammatory Effects on Gene Expression.
Five of the genes displayed either modestly decreased (MIEN1, PGAP3, GRB7, and ERBB2; 1.24-2.01-fold) or increased expression (IKZF3; 2.41-fold) in biopsies from inflamed, compared with noninflamed, colonic IBD mucosa (Table 3). Similar effects were seen separately for CD and UC, although only nominal significance ( < 0.05) was reached for IKZF3 in CD and for MIEN1 in both UC and CD. CSF3 expression was increased at least 116-fold in inflamed, compared with noninflamed, biopsies from the overall IBD group and from the separate disease subentities (Table 3).

Drug-Mediated Effects on Gene Expression.
Drugmediated effects on gene expression were investigated to elucidate a potential explanation for the observed differences in the relationships between genotype and gene expression among patients with and without IBD, respectively. Colonic gene expression levels were compared among patients with IBD receiving a particular treatment versus those not receiving that treatment. No drug-mediated effects were observed in relation to treatment with thiopurines, aminosalicylates, or corticosteroids (S2 Table). There were too few individuals on treatment with anti-TNF--antibodies for a meaningful statistical analysis.

Effects of Sampling Location on Gene Expression.
No differences in gene expression were observed among four different segments of colon (ascending, transverse, descending, and sigmoid) from the noninflamed intestinal mucosa of individuals without IBD (S3 Table; caecum and rectum were excluded because of a low number of samples). Four genes were differentially expressed between the ileum and the colon, however, with either lower (PSMD3 and LRRC3C, 1.34and 3.85-fold, resp.) or higher (ORMDL3 and IKZF3, 1.34fold and 2.23-fold, resp.) expression in the ileum compared with that in the colon (S4 Table).

Discussion
A large number of genetic susceptibility loci have been uncovered in relation to IBD, with pervasive sharing of genetic susceptibility loci between the two major disease subtypes, UC and CD [4]. In this study, we searched for an association between the risk allele of the shared (UC and CD) susceptibility locus rs2872507 and the expression of genes in the intestinal mucosal. This locus has previously been associated with autoimmune diseases such as type-1 diabetes [12], rheumatoid arthritis [13], primary biliary cirrhosis [14], and childhood-onset asthma [7,15] and has been shown to affect the number of white blood cells [16,17]. We found that an increase in the number of susceptibility alleles was associated with the reduced expression of GSDMB in all of the investigated biopsy locations in noninflamed control individuals and in patients with IBD, irrespective of the presence of active inflammation. Additionally, the susceptibility allele was associated with the increased expression of GSDMA and LRRC3C in noninflamed colon biopsies from individuals without IBD and with the reduced expression of PGAP3 and ZPBP2 in noninflamed ileum biopsies from individuals in the overall IBD cohort and the CD cohort, respectively. Concurrent increase and decrease in the expression of neighboring genes in relation to a susceptibility haplotype for childhood asthma were previously shown for the same genomic region, possibly due to differential associations between the CCCTC-binding factor (CTCF) and different haplotypes [18]. CTCF function as an architectural protein that modifies the topology of the genome, which can result in differential regulation of neighboring genes [19]. A recent analysis of IBD susceptibility loci highlighted ORMDL3 as a candidate gene in the same region based on a correlation between genotype and ORMDL3 expression in lymphoblastoid cell lines, liver tissue, and adipose tissue [4]. Susceptibility alleles might exert tissue-dependent effects on gene expression, which are possibly modulated by disease type and medical treatment [20,21]. It is conceivable that the more extensive alteration of gene expression in the colon (GSDMA, GSDMB, and LRRC3C) compared with that in the CD ileum (ZPBP2) reflects the fact that colonic inflammation     may occur in both CD and UC, whereas ileal engagement is restricted to CD. Recently, our group demonstrated an effect of thiopurine treatment on the expression of genes present in networks involving candidate IBD susceptibility genes [22]. Although it is likely that such therapeutic effects also extend to the actual IBD susceptibility genes, no such effects of ongoing medication were observed in the current study.
GSDMA and GSDMB belong to the Gasdermin family of genes. The expression of GSDMA and GSDMB has been associated with differentiated epithelial cells and with regions 8 BioMed Research International containing proliferating cells or stem cells, respectively, of the esophagus and the gastric mucosa [23,24]. GSDMA, but not GSDMB, expression was capable of inhibiting cell growth in a colony formation assay and could induce apoptosis [24]. Additionally, GSDMA expression was frequently suppressed in esophageal cancer cell lines and gastric cancer cell lines, whereas GSDMB was expressed in all investigated cancer cell lines and also showed evidence of gene amplification and overexpression in some cases of gastric cancer [24].
No functional studies have, to our knowledge, been published regarding LRRC3C and the single-pass membrane protein that it encodes: leucine-rich repeat-(LRR-) containing protein 3C (LRRC3C). Within the human genome, LRRC3C was most similar to the protein encoded by LRRC3B. Based on experiments using transformed carcinoma cell lines, a role in the suppression of cell proliferation has been suggested for LRRC3B [25,26].
ZPBP2 encodes a protein paralogue of the zona pellucida binding protein [27] and harbors a missense SNP associated with primary biliary cirrhosis [14].
The colonic mucosa is a complex tissue, and the biopsies represent a heterogeneous collection of cell types. We did not determine the cellular origin of gene expression. However, all the genes that were differentially expressed with respect to inflammation were of potential interest in relation to either the inflammatory response or the turnover of epithelial cells. Inflamed, compared with noninflamed, colonic biopsies showed strongly increased transcript levels of CSF3 and modestly increased transcript levels of IKZF3, whereas the transcript levels of ERBB2, GRB7, MIEN1, and PGAP3 were modestly decreased. The granulocyte colony-stimulating factor (encoded by CSF3) stimulates the production, survival, and activity of neutrophils [28], and the granulocyte colonystimulating factor produced by nonhematopoietic cells promotes neutrophil trafficking to the synovial tissue during collagen-induced arthritis in a murine model of rheumatoid arthritis [29]. Neutrophil infiltration of the intestinal mucosa is a feature of IBD [30]. The transcription factor zincfinger protein Aiolos (encoded by IKZF3) is important for lymphocyte differentiation and shows higher expression in B cells [31]. The overexpression of IKZF3 inhibited proliferation and suppressed apoptosis in a pre-B acute lymphoblastic leukemia cell line [32]. Ikzf3 knockout mice developed a phenotype similar to human autoimmune systemic lupus erythematosus [33]. Also, Pgap3 knockout mice developed autoimmune-like symptoms, possibly relating to impaired clearance of apoptotic cells by peritoneal macrophages [34]. ERBB2 [35,36], GRB7 [37,38], and MIEN1 [39,40] have all been associated with various carcinomas, affecting cell proliferation and/or cell migration.
Considering the risk genotype-gene expression relationships uncovered in our study, both a dysregulated intestinal epithelium [41][42][43] and aberrant T-cell proliferation and apoptosis [44] have been implicated in the pathogenesis of IBD. However, due to the relatively small sample size of our study the identified relationships need to be cautiously interpreted and confirmed in an independent, preferably larger, study cohort. The exact causes of IBD remain unclear, but the etiology likely includes complex interactions that depend on, at least, the intestinal microbiome and host factors such as genetic susceptibility, intestinal barrier function, and mucosal immune response [45,46]. In this context it is notable that a joint effect of variants within the genetic region of our study and human rhinovirus wheezing illness has been observed on the subsequent risk of developing childhoodonset asthma. [47,48]. Although the mechanism of action is not understood, it is possible that it is pertinent to IBD as well. Microbiome studies in IBD have, so far, mainly focused on alterations in the bacterial flora. However, recent studies have also described alterations in the viral component of the intestinal microbiome of IBD patients [49,50].
Based on our results and published findings on GSDMA, GSDMB, LRRC3C, and related proteins, we propose that the rs2812507 locus in part affects IBD susceptibility via effects on apoptosis and cell proliferation and believe this hypothesis warrants further experimental investigation.

Ethical Approval
The study was approved by the Regional Ethical Review Board in Linköping, Sweden (Dnr 2011/201-31).

Consent
Written and informed consent was obtained from all participants.