Association of polymorphic variants in serotonin re-uptake transporter gene with Crohn's disease: a retrospective case-control study

Aim To analyze the distribution of SLC6A4 gene polymorphisms in Crohn’s disease (CD) patients and their association with the disease. Methods We evaluated the presence/absence of promoter (5-HTTLPR, rs25531) and intron 2 (STin2 VNTR) polymorphic variants of SLC6A4 gene in a retrospective case-control study including 192 CD patients and 157 healthy controls (HC). Genotyping was performed by polymerase chain reaction. The association of polymorphisms with CD and its clinical subtypes was analyzed using χ2 and Fisher exact test, binary logistic regression, and haplotype analysis. Results CD patients and healthy controls had similar sex (88 [45.8%] vs 84 [53.5%] women, respectively; P = 0.154) and age (41.3 ± 12.8 years vs 41.7 ± 8.8 years, respectively, P = 0.091) distribution. Significant differences were observed in the STin2 genotype and allele distribution between CD patients and healthy controls (P = 0.003 and P = 0.002, respectively) and between the corresponding female subgroups (P = 0.004 and P = 0.007, respectively), with a significant negative association of biallelic ss (STin2.9 and Stin2.10) STin2 genotype with CD (P = 0.013, age- and sex-adjusted odds ratio [OR] 0.5, 95% confidence interval [CI] 0.29-0.86; women: P = 0.006, age-adjusted OR 0.32, 95% CI 0.14-0.72) and a significantly higher S-STin2.12 (5-HTTLPR/rs25531: S-STin2: STin2.12) haplotype distribution in CD patients (P = 0.004, OR 1.62, 95% CI 1.16-2.26). There was no significant association between 5-HTTLRP and rs25531 genotype or allele frequencies and CD and between any SLC6A4 polymorphic loci with clinical CD subtypes. Conclusion STin2 VNTR polymorphism of SLC6A4 gene may contribute to CD pathogenesis.

Neuroimmunological interactions are also important because various pro-and anti-inflammatory cytokines may affect neuronal activity and the release of neurotransmitters influencing the activity of immuno-effector cells in the GI tract. The best characterized among them is serotonin (5-HT), 90% of which is produced and secreted by intestinal enterochromaffin cells (5)(6)(7)(8). Alterations in 5-HT biosynthesis, quantity, release, or clearance are important for both sensory signal transduction in GI motility and the development of visceral hypersensitivity (5)(6)(7)(8). 5-HT is also a chemotactic molecule and may promote lymphocyte activation and secretion of pro-inflammatory cytokines (5)(6)(7)(8). Therefore, upon its release and binding to targeted receptors, 5-HT action must be rapidly terminated. This is maintained by the action of serotonin re-uptake transport-er (SERT), expressed by serotonergic neurons and the mucosal enterocytes (5)(6)(7)(8).
Animal models and studies on human cell lines and tissue indicate increased 5-HT availability and reduced SERT expression in the inflamed colon, accompanied with an increased expression of inflammatory genes, thus supporting the idea that the loss of SERT gene (SLC6A4) transcription can either cause intestinal inflammation or result from it (8,9). SLC6A4 gene polymorphisms have also been linked with its translation and expression (10,11).
Sikander et al (20) has reported a significant association of 5-HTTLPR polymorphism with UC and microscopic colitis (MC). However, the association between 5-HTTLPR and CD has never been determined. The same was true for other SLC6A4 polymorphic regions in any form of IBD (11,20). Therefore, the aim of this study was to compare genotype and allele frequencies of 5-HTTLPR, rs25531and STin2 VNTR FiGuRe 1. Location of SLC6A4 gene polymorphic regions and their genotype combinations examined in this study. SLC6A4 gene is affected by the insertion/deletion variation of the promoter region 5-HTTLPR, resulting in two alleles, long L and small S. This promoter region also functionally couples with the common rs25531 (A/G) single nucleotide polymorphism (SNP), resulting in alleles L A and L G . L A allele is associated with the higher transcriptional activity, whereas the L G allele exhibits lower serotonin uptake and the transcriptional activity equivalent to S allele of the 5-HTTLPR polymorphic region. Transcriptional activity of SLC6A4 gene is further modulated by the enhancer activity of STin2 polymorphic region, 17-18 bp long variable number of tandem repeats (VNTR), found in intron 2. The STin2 allelic variants were identified as 10-repeat and 12-repeat alleles that have been identified in all ethnicities, and the less common 9-repeat allele found only in individuals of european or African descent. Thus, promoter polymorphic region may be presented in biallelic, triallelic, and clustered biallelic genotype model while STin2 VNTR region may be presented in triallelic and biallelic genotype forms. Rare allele forms (eg, XL or STin2.7) not encountered in this study, are not presented. polymorphic regions in CD patients and healthy control participants and investigate their association with CD. Given the crucial role of SERT protein in the regulation of intestinal 5-HT availability, we hypothesized that the SLC6A4 polymorphism may contribute to the genetic predisposition for the development of CD.

Patients
This retrospective case-control study included all 192 consecutive CD patients diagnosed between December 2012 and 2016 at the University Hospital Center Zagreb (Table  1). CD diagnosis was established according to standard endoscopic, radiological, and histopathological criteria (21). The healthy control group consisted of 157 participants with no family history of IBD or irritable bowel disease. They were selected from 217 completely healthy voluntary participants (hospital employees or their acquain-tances) and age-and sex-matched to CD patient group. Genetically related voluntary control participants and participants who were incompatible according to age (outside the age range of CD group members) were excluded from further analysis. All participants signed the informed consent for study participation and data publication. The study protocol was approved by the Ethics Committee of the University Hospital Center Zagreb (Approval number 04/31-JG; contract number 108-1081874-1917, start date January 1, 2007).

DNA extraction and genotyping
Blood samples were collected by venipuncture in EDTA BD Vacutainer® blood collection tubes (Becton, Dickinson and Company, Franklin Lakes, NJ, USA), and genomic DNA was isolated from peripheral blood leukocytes using the salting out procedure (22). Study participants were genotyped for both promoter and intron 2 polymorphisms of the SLC6A4 as previously described (22)(23)(24). Biallelic 5-HTTL-RP variants were discriminated with some modifications as previously described (22)(23)(24). The following primers were used: forward 5′-ATG CCA GCA CCT AAC CCC TAA TGT-3' , reverse 5'-GGA CCG CAA GGT GGG CGG GA-3' . The 5-HT-TLPR polymorphic region was subdivided into functional variants (S, LA, LG) by rs25531 (A/G) SNP using a TaqMan qPCR procedure according to Hu et al (25). Amplification was performed in a 7500 qPCR System under the following conditions: at 50°C for 2 minutes, at 95°C for 15 seconds, and at 62.5°C for 1.5 minutes. Genotype sequencing/ capillary electrophoresis (ABI PRISM 3100 Genetic Analyzer, TermoFisher Scientific, Waltham, MA, USA) was performed to acquire controls. The STin2 VNTR polymorphism was discriminated using the method described by Ito et al (26). For the PCR reaction, the following primer pair was used: forward 5'-GTCAGTATCACAGGCTGCGAG-3' , reverse 5'-TGT-TCCTAGTCTTACGCCAGTG-3' .

Statistical analysis
The normality of data distribution was analyzed by Shapiro-Wilk test. Data are presented as frequencies or means with standard deviations (±SD). The groups were compared using t test or Mann-Whitney test for parametric and non-parametric continuous variables, respectively, and Pearson χ 2 test for categorical variables. Conformity of genotype distributions to Hardy-Weinberg equilibrium was assessed using goodness-of-fit χ 2 test. The difference in allele frequency was determined by using either χ 2 or Fisher exact test as appropriate. Crude and adjust- ed odds ratios (OR) (controlling for age and sex) and the corresponding 95% confidence intervals (CI) determined by binary logistic regression were also used to test the association between CD and genotype variants of both promoter and intron 2 SLC6A4 gene polymorphisms. Association analysis based on patient sex, or clinical CD subtypes such as age at diagnosis, localization, and disease behavior were also performed. Considering that LG allele has the same transcriptional activity as the S allele, the triallelic 5-HTTLPR-rs25531 genotypes were also coded in a clustered biallelic form (L'L' , L'S' and S'S'; Figure 1). Likewise, the triallelic STin2 VNTR polymorphism was also coded in a biallelic (ll -ls -ss; Figure 1) genotype form. All biallelic genotype forms were analyzed under the codominant, dominant, and recessive model. Construction of haplotypes, linkage disequilibrium (LD), and haplotype association analysis were calculated by permutation test using the online SHEsisPlus software (27).
The power analysis (to compute a priori sample size and the post-hoc power of the study) was performed using G*Power software (version 3.1.9.2 for Windows) (28). For the analysis of 5-HTTLRP genotype frequencies with a χ 2 goodness-of-fit test with the small effective size (w = 0.2, df = 2); power (1-β) = 0.80, and α = 0.017 (Bonferroni corrected P value), the required total sample size was 313 and actual sample size was 349. Post-hoc computed achieved power (1-β) of analysis was 0.085 (w = 0.2), 0.997 (w = 0.3; medium effective size), 1.00 (w = 0.5; large effective size) for genotype distribution, and 0.80 for allele frequency analysis (w = 0.2 -0.5; df = 1). For the a priori analysis of STin2 VNTR genotype frequencies with a χ 2 goodness-offit test with the small effective size (w = 0.23, df = 4); power (1-β) = 0.80 and α = 0.008 (Bonferroni corrected P value), the required total sample size was 329, and the post-hoc achieved power (1-β) of analysis was 0.83-1.00 for genotype distribution and 0.99-1.00 (w = 0.2 -0.5; df = 2) for allele frequency analysis. Furthermore, with a power of 0.80 we were able to detect an effect size of 0.223 for a significant difference in STin2 VNTR genotype frequencies. The two-tailed P < 0.05 was considered significant and corrected according to Bonferroni procedure (the corrected level of significance is: Pc = 0.05/N; N -number of independent tests). All reported P values were uncorrected unless stated otherwise. Statistical analysis was performed using SPSS Statistics software trial version (IBM Corp., Armonk, NY, USA), unless stated otherwise.

Population characteristics
There were 88 (45.8%) women and 104 (54.2%) men in the CD group, and 84 (53.5%) women and 73 (46.5%) men in the control group (Pearson χ2 = 2.033, P = 0.154). The groups were also comparable according to age (mean±SD: 41.3 ± 12.8 years vs 41.7 ± 8.7 years, respectively; Mann-Whitney test, P = 0.091).  CD group and control group showed a different distribution of STin2 VNTR allele frequencies, which remained signif-icant after Bonferroni correction (Table 2). This was also the case for women with CD (χ2 = 9.850, df = 2, P = 0.007) and their healthy female controls (data not shown). In addition, STin2.12 allele showed a positive and STin2.9 allele a negative association with CD both in the overall ( Table 2) patient sample and in the subgroup of female patients (data not shown). However, the association remained significant only for the rare STin2.9 allele in the overall study sample and the subgroup of female patients after Bonferroni correction (Table 2). No significant association following Bonferroni correction was found between STin2 VNTR allele frequency and clinical CD subtypes (data not shown).
There was no significant association in 5-HTTLRP and rs25531 allele frequency between the CD and control group or between the corresponding sex subgroups (data not shown). The results remained the same after the analysis for clinical CD subtypes (data not shown). Nevertheless, we did observe a positive, although non-significant, association of S allele of 5-HTTLRP and rs25531 polymorphism with CD (Table 2).
Pearson χ 2 test showed significant differences in STin2 genotype distribution (Table 3) between CD patients and healthy controls (P = 0.003) and between the corresponding female (χ2 = 15.326, df = 4, P = 0.004) subgroups (data not shown). This was even more pronounced when carriers of the rare STin2 genotype forms (STin2 12/9 + STin2 10/9; Fisher exact test for CD vs controls; P = 0.002) were grouped together (overall CD: χ2 = 15.720, df = 3, P = 0.001; female subgroup: χ2 = 15.326, df = 3, P = 0.002). Furthermore, STin2 12/12 and Stin2 12/10 genotypes were more common, while STin2 10/10 genotype was less common among CD patients and controls. The same was true when we analyzed the corresponding female subgroups only (data not shown). However, even at very liberal correction for multiple comparisons of Pc = 0.008 (Pc = 0.05/N -number of genotypes analyzed), we did not find any significant difference in the distribution of individual STin2 VNTR genotypes between CD and controls ( Table 3).
In logistic regression analysis, the co-dominant triallelic STin2 genotype model did not show a significant association of individual STin2 genotypes with CD (Table 3). How-  ever, when men and women were analyzed together under the recessive biallelic model (Table 3) or when women only were analyzed (data not shown), carriers of biallelic ss (s = STin2 10 or STin2 9), STin2 genotype form exhibited a significant negative association with CD compared with ll and ls carriers (female patients: Wald = 7.564, df = 1, P = 0.006, age-adjusted OR = 0.32, 95% CI = 0.14-0.72), which remained significant after sex and/or age adjustment (female patients: Wald = 7.564, df = 1, P = 0.006, OR adjusted by age = 0.32, 95% CI = 0.14-0.72). Nevertheless, only the variation of co-dominant STin2 model in which the rare genotypes (STin2 12/9 and 10/9) were grouped together showed significant global P values in crude and ad-justed logistic regression analysis, and only when male and female patients were analyzed together (Table 3).
No significant difference was observed in 5-HTTLRP and rs25531 genotype distribution between CD group and controls (Table 4) or between sex subgroups and clinical CD subtypes (data not shown). The same was found when biallelic 5-HTTLRP and rs25531 genotype forms were analyzed under the codominant, dominant, or recessive genetic model. Nevertheless, we did observe more frequent, although non-significant, distribution of heterozygous (LS, LALG, L'S') and homozygous SS genotype forms in the CD group (Table 4).  Haplotype analysis 5-HTTLPR and rs25531 polymorphic region were in strong linkage disequilibrium, while LDs between them and STin2 VNTR polymorphic region were relatively weak ( Figure 2). After Bonferroni correction, only S-S-STin2.12 haplotype showed a significantly higher frequency in CD group ( Table  5). The higher frequency in CD group was also revealed for L-LG-STin2.10 haplotype, but it was not significant (Table 5).

DiSCuSSiON
Our data revealed a significant difference in STin2 VNTR genotype and allele distribution between the overall CD group and healthy controls and between the female patient group and the corresponding control group. There was also a significant negative association of biallelic ss (STin2. 10/10 and STin2. 10/9 vs STin2. 12/12, 12/10 and 12/9 combined) genotype form and CD. We also found a significantly higher S-STin2.12 (5-HTTLPR/rs25531: S -STin2: STin2.12) haplotype distribution among CD patients. However, we did not find any association between 5-HTTLRP and rs25531 genotype or allele frequencies and CD or between any SLC6A4 polymorphic loci and clinical subtypes of CD. Nevertheless, our results indicate that STin2 VNTR polymorphism of SLC6A4 gene may contribute to CD pathogenesis.
Altered function and down-regulation of SERT protein expression paralleled with abnormal 5-HT concentration, both locally in epithelial layer and in the circulation, has been documented in several human GI inflammatory conditions including CD, UC, and MC, as well as diverticulitis and active celiac disease (32)(33)(34). This was also documented in several animal models of intestinal inflammation (35)(36)(37)(38)(39). Pro-inflammatory mediators and growth factors released during IBD may down-regulate SLC6A4 transcription and decrease SERT protein expression and function (40)(41)(42)(43).
Another possibility is that some individuals with IBD have a genetic predisposition leading to altered SERT expression with consequent changes in gastrointestinal 5-HT levels, which contributes indirectly to pro-inflammatory conditions in the affected intestinal mucosa.
However, despite the recognition that intestinal inflammatory diseases, such as CD, can have a strong genetic component, polymorphisms so far linked with SERT transporter have not been associated with CD.
Furthermore, in previous reports on GI diseases only STin2 VNTR was observed as an attractive candidate for a possible association of the SERT with IBS (10,44). Although IBD and IBS are usually viewed as dichotomous conditions, they exhibit similar alterations in serotonergic-signaling mechanisms (33,(44)(45)(46). Due to the development of IBS symptoms in IBD patients in remission and the clinical overlap between IBD and IBS, some authors even argued that they may represent clinical manifestations of a pathophysiologic spectrum of the same disease (47,48). Several population studies showed that IBS patients have an increased risk of becoming IBD patients than patients without prior IBS history, and this effect may be greater for CD (49,50).
Reports regarding the link between STin2 polymorphism and IBS were controversial and inconclusive, with most of them indicating no association with IBS or its clinical subtypes (47). However, Wang et al (51) found that IBS patients had a greater frequency of STin2.12/10 and a lower frequency of STin2.12/12 genotype compared to controls, but identified no significant difference in STin2 polymorphism among different clinical subtypes of IBS.
Current research in functional implications of the STin2 polymorphism is also inconclusive. It is known that STin2 acts as a transcriptional regulator and has allele-dependent enhancer-like properties that may influence tissue-specifc regulation of SLC6A4 gene (18,52). However, it seems that individual SLC6A4 polymorphisms have a weak influence compared with the combined effect of 5-HTTLPR and STin2 region (14,18,52,53). Therefore, their allelic combinations should be identified before concluding about functional and phenotypic associations. LD between these two loci, ranging from moderate (European) to very strong (Native Americans), was found in most of the studied populations (31). In addition, a partial linkage of STin2.12 allele with S allele of 5-HTTLPR (S12 haplotype combination) has been reported and had stronger enhancer-like properties on SLC6A4 transcription than L10 or S10 haplotype (53,54). We found a significantly higher distribution of S12 haplotype in CD group, showing its positive association with disease occurrence. However, since we did not measure the level of SERT expression or 5-HTT plasma or tissue levels, we were not able to indicate any functional consequences of this finding.
We did not confirm the association of 5-HTTLRP and rs25531 with CD occurrence, but we did observe the tendency toward higher frequencies of heterozygote 5-HTTLRP and rs25531 genotype forms in CD group.
To the best of our knowledge, there are no reports analyzing the relationship between the promoter polymorphic regions of the SLC6A4 gene with CD. Sikander et al (20) demonstrated a potential association between 5-HT-TLPR polymorphism and UC and MC by finding a significantly lower frequency of SS vs non-SS (LL and LS combined) 5-HTTLPR genotypes in MC patients or UC patients in remission and significantly higher serum 5-HT levels (SS>LS>LL) in these two patient groups. This was especially the case in MC LS and SS genotype carriers compared with controls (20). In addition, UC patients with active disease and SS genotype also had increased 5-HT levels compared with control subjects expressing the same genotype (20). Similar to our results in CD patients, they did not observe any significant differences in 5-HTTLPR genotype and allele distributions between UC patients with active disease and healthy controls or between male and female patient subgroups (20). Likewise, Shiotani et al (34) also observed no significant association of 5-HTTLPR variants and UC. Our results were contrary to the findings of Sikander et al (20), as we found a higher frequency of SS genotype in CD group.
The observed discrepancies are possibly due to heterogeneous background of CD and other forms of IBD and/ or ethnical or regional differences between the studied groups. They are also consistent with the previously mentioned hypothesis that 5-HTTLPR might represent only one of the contributing polymorphic factors responsible for the disease occurrence.
Regarding the IBS and 5-HTTLRP polymorphism, different associations were reported, but the results were contradictory and inconclusive even for the same population under the study. The most recent meta-analysis found no significant association between 5-HTTLPR and IBS in overall population, while in the IBS subtype-and ethnic subgroupbased analysis the LL genotype was demonstrated as a risk factor for constipation predominant IBS (58).
With respect to rs25531 polymorphism, we found only two studies examining its association with IBS, and only one reported a positive association with disease occurrence with three times higher odds ratio for the LG allele distribution in IBS patients compared to controls (59,60).
There are several limitations to our study. The sample size was too small for a comprehensive genotype analysis, so the results cannot be generalized and should be interpreted cautiously. Second, we did not determine the 5-HT levels and were unable to evaluate the functional consequences of individual genotype or haplotype variants on SERT expression. Third, association studies of unrelated individuals, such as ours, warrant cautious interpretation as unknown sources of population stratification may affect the results.
In conclusion, we demonstrated significant differences in STin2 genotype and allele distribution between CD patients and healthy controls, and a negative association of biallelic ss STin2 genotype variant and a positive association of S12 haplotype with CD disease. Further large-scale studies in this and other populations aimed to confirm the obtained results and decipher the exact functional role of STin2 VNTR polymorphic region in intestinal inflammatory diseases are warranted.
Oxford Centre for evidence-based Medicine level of evidence 3b/4a.
Funding This work has been funded by Croatian Science Foundation; project name: Inflammatory Bowel Diseases (Crohn's disease and ulcerative colitis); project number 108-1081874-1917.
ethical approval The study protocol was approved by the Ethics Committee of the University Hospital Center Zagreb (Approval number 04/31-JG; contract number 108-1081874-1917, start date January 1, 2007).
Declaration of authorship All authors except BD conceived and designed the study; all authors acquired the data, analyzed and interpreted the data, drafted the manuscript; critically revised the manuscript for important intellectual content, gave approval of the version to be submitted; and agree to be accountable for all aspects of the work.