Common variants of HTR3 genes are associated with obsessive-compulsive disorder and its phenotypic expression

Evidence from literature supports the existence of associations between serotonin-related genetic variants and obsessive-compulsive disorder (OCD), but few studies have explored the involvement of serotonin receptor type 3 genes (HTR3) in OCD. To identify whether HTR3 variability affects an individual’s susceptibility to OCD, we examined 10 HTR3 variants in 596 individuals with OCD and 599 controls. A significant difference existed in the genotypic distribution of the HTR3B variant rs1176744 between individuals with OCD and controls (odds ratio [OR] = 0.74, 95% confidence interval [CI] = 0.60–0.91, P = 0.0043). A protective haplotype in HTR3B was also associated with OCD (OR = 0.77, CI = 0.63–0.95, permutated P = 0.0179). Analyses of OCD sub-phenotypes demonstrated significant associations between rs3758987 and early onset OCD in male subjects (OR = 0.49, CI = 0.31–0.79, P = 0.0031) and among rs6766410, rs6443930, and the cleaning dimension in female subjects (OR = 0.36, CI = 0.18–0.69, P = 0.0016 and OR = 0.47, CI = 0.29–0.79, P = 0.0030, respectively). Additionally, rs6766410 was related to contamination-based disgust in OCD (P = 0.0044). These results support that common HTR3 variants are involved in OCD and some of its clinical phenotypes.

Scientific RepoRts | 6:32564 | DOI: 10.1038/srep32564 To date, five distinct HTR3 genes have been cloned for humans: HTR3A and B are located on chromosome 11q23.1-2 16 while HTR3C, D, and E are located on chromosome 3q27.1 17 . A large genome-wide linkage study for OCD provided evidence that OCD is linked to markers on chromosome 3q27-28, although the findings did not reach the accepted level of statistical significance 18 . Considering that the markers are 2.5 Mb downstream of HTR3C-E, these genes may be positional candidates in OCD. Additionally, several authors have indicated that the single nucleotide polymorphism (SNP) rs1062613 in HTR3A is associated with the personality trait of harm avoidance 19 and the modulation of amygdala activation 20 in healthy subjects, both of which are suggested to have particular relevance for OCD 21 . HTR3 genes may therefore be plausible candidates with regard to their involvement in OCD. To the best of our knowledge, however, only two association studies have investigated the involvement of HTR3 in OCD. In a study with 75 trio samples, no significant association was found between the HTR3A variant rs1062613 and early onset OCD 22,23 . The other study utilized case-control samples and demonstrated that the HTR3E variant rs7627615 was related to the washing dimension and visual organization scores in OCD 24 .
Given this paucity of data and the promising clinical outcomes that are being achieved in subjects with OCD following the use of 5-HT 3 antagonists, we aimed to perform a case-control association study with common HTR3 variants in a larger sample of adult OCD probands and controls. Clinical characteristics such as the onset age and symptom dimensions were included in the analyses in terms of their relationship to the genetic variants, as these phenotypes have been proposed as a means of determining subgroups that are more genetically valid 25 . Furthermore, we sought to establish whether variability in HTR3 contributes to disgust sensitivity, a psychological trait closely associated with OCD 6 . We hypothesized that variations within HTR3 may confer genetic vulnerability to OCD and its associated clinical characteristics and psychological traits.

Results
Genotyping quality control. The threshold for the genotyping call rate was set at 95% for each SNP, with an average call rate of 99.2%. None of the SNPs in controls, individuals with OCD, or the entire sample significantly deviated from the Hardy-Weinberg equilibrium at a Bonferroni-corrected significance level of α = 0.005. The minor allele frequencies were > 0.05 for all SNPs. Table 1 provides a detailed description of each SNP. Table 2, no significant differences were found regarding the sex distribution or years of education between the two groups, but individuals with OCD were significantly older than were controls. Of the individuals with OCD, 103 (17.3%) were drug-naïve at enrolment. As for the disgust sensitivity trait, both groups demonstrated similar scores for core disgust, whereas individuals with OCD scored significantly higher on animal reminder disgust and contamination-based disgust.

Subjects. As shown in
Single SNP association analysis. Regarding the genotype distributions, four SNPs were nominally significantly different between individuals with OCD and controls: rs1062613, rs3758987, rs1176744, and rs3782025. However, only rs1176744 remained significantly different after Bonferroni correction under an additive model ( Table 3).
The analyses based on the subjects' clinical characteristics, including their onset age and symptom dimensions, yielded no significant results for the entire OCD sample. However, significant associations were observed in a further analysis stratified by sex. Considering the onset age, the genotype distribution of rs3758987 differed significantly between male subjects with early onset OCD and male subjects with late onset OCD under a dominant model (Table 4). For the analysis of symptom dimensions, two SNPs, rs6766410 and rs6443930, were significantly associated with the cleaning dimension in female subjects under an overdominant model and additive model, respectively (Table 5). To determine whether these associations were independent, we performed a conditional analysis between these two SNPs. The results of conditional analysis remained significant, which were consistent with the low linkage disequilibrium (LD) between these SNPs (D′ = 0.05 and r 2 = 0). The results of this conditional analysis are provided in Supplementary Table S1. Concerning the relationship between disgust sensitivity and HTR3 variants in OCD, we found a nominally significant effect of rs6766410 on the combined disgust scale-revised (DS-R) scores under an overdominant model (F [3,252] = 3.472, P = 0.0167, Wilk's λ = 0.960, partial η 2 = 0.040). A follow-up univariate analysis of variance revealed that the contamination-based disgust scores were significantly lower in subjects with the AC genotype than they were in subjects with the AA/CC genotypes (8.42 ± 3.98 and 9.85 ± 3.98, respectively; F [1,254] = 8.251, P = 0.0044, partial η 2 = 0.031). No significant associations were found between the other HTR3 variants and disgust sensitivity scores in OCD.
Haplotype association analysis. We identified four LD blocks, three of which contained two markers from each gene and one of which contained three markers from HTR3C and HTR3E (Supplementary Figure S1). For the haplotypes estimated in HTR3B, a significant difference was observed for the distribution of haplotypes between individuals with OCD and controls. As shown in Table 6, a specific haplotype C-C was significantly associated with a lower risk of being affected by OCD. For the haplotypes in the other HTR3 genes, no evidence of a relationship with OCD was found.

Discussion
Here, we explored whether HTR3 genetic variants confer risk for OCD and/or for certain clinical characteristics of the disorder. Our results support the involvement of HTR3 in OCD, both in the onset age and in the manifestation of specific symptom dimensions. We found a global relationship between the HTR3B variant rs1176744 and OCD under an additive model, suggesting that the odds of being affected by OCD were reduced by 0.74 times with a one-copy increase of the variant C allele. Similarly, the HTR3B haplotype with the C allele was associated with OCD subjects in a protective manner. These results imply that the variant C allele of rs1176744 may decrease an individual's susceptibility to OCD. On the protein level, this functional Tyr129Ser variant results in slow deactivation and desensitization kinetics in variant (p.129S) 5-HT 3 AB receptors compared to in wild-type ones 26 [27][28][29] , one possible explanation for our result is that alterations in receptor responsiveness might play a role in fear conditioning and extinction, which could in turn contribute to OCD susceptibility 30 . Moreover, it has been suggested that the AA genotype of this SNP might increase the risk of developing nausea during paroxetine treatment 31 . As nausea is a common side effect of SSRIs, ascertaining those individuals who are likely to develop nausea may facilitate more tailored SSRI treatment strategies. Collectively, the functional variant rs1176744 may underlie the genetic aetiology of OCD and could serve as a predictor of an individual's drug response.
We also found nominally significant associations between OCD and rs1062613, rs3758987, and rs3782025. Among these, the odds ratio of rs1062613 was 3.51 under a recessive model, which seems to be a large value for a common variant in OCD. Considering that there were only four TT genotypes in controls, our result may have been overfitted 32 . A larger sample of minor allele homozygotes would be needed to obtain a more reliable statistic.
Regarding the onset age of obsessive-compulsive symptoms, we found a significant association between early onset OCD and rs3758987 in male subjects. Although this 5′ upstream variant does not directly affect the amino  Table 4. Association between rs3758987 and the onset age. OCD, obsessive-compulsive disorder; OR, odds ratio; CI, confidence interval.
acid sequence of the encoded protein, this variant may be in LD with a nearby functionally important, but unexplored, polymorphism. On the other hand, this SNP might influence regulatory processes related to HTR3B expression. As individuals with early onset OCD may represent a genetically more valid subgroup 25 , further research on the physiological relevance of rs3758987, as well as replication of this association in different populations, is needed. With regard to the symptom dimensions, the cleaning dimension was significantly associated with two non-synonymous SNPs, rs6766410 and rs6443930, in female subjects. When analysing their putative effects with PolyPhen-2 33 , we found that neither variant was predicted to be damaging. However, in terms of rs6807362 and rs1000952, which were in high LD with rs6766410 and rs6443930, respectively, the analysis revealed that these variants were predicted to be possibly or probably damaging (PolyPhen-2 scores [HumDiv] 0.647 and 0.998, respectively). Thus, the observed association with rs6766410 and rs6443930 may be attributed to other tightly linked functional variants. Notably, rs6766410 was also related to the contamination-based disgust sensitivity. Interestingly, the relationships among rs6766410, the cleaning dimension, and disgust sensitivity were the most robust under an overdominant model in the same direction, in which the heterozygote genotype AC was significantly associated with a reduced risk of the cleaning dimension and with lower contamination-based disgust scores. These results suggest that molecular heterosis may underlie the relationships among this HTR3C variant, the cleaning dimension, and disgust sensitivity. According to Comings and MacMurray 34 , our results are likely related to interaction effects between the wild-type and variant 5-HT 3 receptor subunits.
The gender-specific associations found here are consistent with the findings of previous studies, which revealed sexually dimorphic effects of the genetic variants on OCD [35][36][37] . Although several of our results did not reach the experiment-wise significance of α = 0.0017 (0.05/30 ≈ 0.00167) after Bonferroni correction for three different strata, the P values obtained from these stratified analyses were the three most significant ones in the present study. Interestingly, these associations were consistent with the gender differences observed for OCD symptoms, including an earlier onset in men 38 and more contamination-related symptoms in women 39 . The gender differences in clinical manifestations might thus reflect underlying genetic heterogeneity.
Previous studies have shown that 5-HT 3 antagonists may be beneficial as OCD treatments [11][12][13][14][15] . Hence, the results of this study may have implications for pharmacogenetic studies utilizing 5-HT 3 antagonists in OCD. It is plausible that genetic variations and the subsequent alterations in receptor function might elicit different responses to 5-HT 3 antagonists; clarifications regarding the effects of such genetic variations on an individual's treatment responses may aid in selecting the appropriate treatment options.
To our knowledge, our study on the involvement of HTR3 in OCD analysed the most individuals. Nonetheless, our study has several limitations. First is the potential for population stratification. As we did not have information on the migration histories of the subjects nor did we include a panel of ancestry-information markers, we could not control the potential effects of an undetected population substructure. Although the considerable degree of genetic homogeneity among the Korean population 40,41 might make bias less likely, the possibility of false-positive associations stemming from population stratification could not be completely excluded. Second, as controls were significantly younger than were individuals with OCD, control subjects may develop  Table 6. Estimated haplotype frequencies in individuals with OCD and controls. OCD, obsessivecompulsive disorder; OR, odds ratio; CI, confidence interval.
obsessive-compulsive symptoms later in life. However, this inevitable factor may have exerted only a trivial effect on the power, because the lifetime prevalence of OCD is ~1-2% and controls had largely passed the mean age of OCD onset 42 . Third, the DS-R scores were obtained from a subset of subjects, which may have reduced the statistical power. Fourth, as the onset age information was collected retrospectively, the potential for recall bias cannot be disregarded. Finally, we investigated only 10 of the HTR3 polymorphisms, thus associations with other variants may have been missed. In summary, we found that HTR3 variants influenced the affected status of individuals with OCD and several of its phenotypes. These findings support that 5-HT 3 receptors are involved in the pathophysiology and clinical manifestations of OCD. Future studies focusing on the relationships among these HTR3 variants and the treatment response to 5-HT 3 antagonists may elucidate whether genetic variations in the 5-HT 3 receptor also influence the medication response in individuals with OCD.

Methods
Subjects. The study sample consisted of 596 individuals with OCD and 599 healthy control subjects.
Unrelated individuals with OCD were consecutively recruited from the outpatient department of psychiatry at Severance Hospital, Yonsei University Health System, and diagnosed with the Korean version of the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) Axis I disorders 43 by a trained psychiatrist. Exclusion criteria were as follows: age < 19 or > 65 years, a lifetime history of psychotic symptoms, history of substance abuse or dependence in the preceding 6 months, or severe organic or neurologic disorders. Subjects with comorbid DSM-IV Axis I disorders were not excluded as long as the obsessive-compulsive symptoms were the main reason for seeking treatment. Gender-matched, unrelated controls were recruited from the local community through advertisements. Controls with a lifetime history of DSM-IV Axis I disorders or neurological disorders were not included in the study. Ethnicity was ascertained through self-reports, and only those subjects who identified themselves as ethnically Korean were enrolled.
The onset age of OCD was defined as the age at which the obsessive-compulsive symptoms first occurred, as recalled by the subject or family members. The threshold for early onset OCD was considered 17 years of age 44 . The severity of the obsessive-compulsive and depressive symptoms was evaluated with the Yale-Brown Obsessive-Compulsive Scale 45 and Montgomery-Åsberg Depression Rating Scale 46 , respectively. The Yale-Brown Obsessive-Compulsive Scale symptom checklist was employed to identify the following four previously reported symptom dimensions in the meta-analysis 47 : (1) symmetry-symmetry obsessions and repeating, ordering, and counting compulsions; (2) forbidden thoughts-aggressive, sexual, religious, and somatic obsessions and checking compulsions; (3) cleaning-contamination obsessions and cleaning compulsions; and (4) hoarding-hoarding obsessions and compulsions. The presence of a dimension was determined based on a lifetime history of one or more symptoms in the respective category.
This study was approved by the Institutional Review Board of Severance Hospital. The methods were performed in accordance with the approved guidelines. Written informed consent was obtained from each subject at the beginning of the study.
Disgust scale-revised. Information on individual differences in the sensitivity to disgust was obtained with the Korean version of the DS-R 48 in subsets of the individuals with OCD (n = 256) and controls (n = 478). The DS-R is comprised of the following three subscales: core disgust scale, animal reminder disgust scale, and contamination-based disgust scale. Core disgust reflects the avoidance or rejection response to disgusting stimuli, including bodily waste products, small animals, and rotting foods. Animal reminder disgust indicates the aversion to stimuli that reminds the individual of the animal origins of humans, including body envelope violations and death. Finally, contamination-based disgust is associated with the perceived risk of disease contagion 6 . SNP selection. We selected 10 SNPs from across all of the HTR3 genes according to either of the following criteria: (1) functional variants annotated in dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP/) that reside within the regulatory region or alter the amino acid sequence of a protein, or (2) variants previously reported to be related to OCD or other psychiatric disorders. All selected variants had a verified minor allele frequency > 0.05 in Asians, as ascertained via the HapMap project database (http://hapmap.ncbi.nlm.nih.gov/; Data Release 28, phase II + III August 10, on NCBI B36 assembly, dbSNP b126).
Genotyping. Genomic DNA was prepared from blood samples with the QuickGene-mini80 (FUJIFILM, Tokyo, Japan). In a subset of controls (n = 160), DNA was extracted from saliva using the Oragene DNA collection kit (DNA Genotek, Kanata, Ontario, Canada). Genotyping of rs3782025 was performed with the ABI PRISM SNaPShot Multiplex kit (ABI, Foster City, CA, USA) according to the manufacturer's recommendations. Analyses were conducted using the GeneMapper software (version 4.0; Applied Biosystems, USA). Genotyping of the remaining nine SNPs (rs1062613, rs1176713, rs3758987, rs1176744, rs6766410, rs6807362, rs6443930, rs1000952, and rs7627615) was performed with the TaqMan fluorogenic 5′ nuclease assay (ABI, Foster City, CA, USA) according to the manufacturer's instructions. Primer sequences and assay IDs are shown in Supplementary  Table S2.
Sample power calculation. Statistical power was evaluated under a dominant genetic model using the Quanto software (version 1.2.4; http://biostats.usc.edu/software); statistical significance was set at P < 0.05. Given the available sample size, the statistical power for detecting a risk allele with an effect size of 1.5 ranged from 0.77 to 0.88, depending on the minor allele frequency.
Scientific RepoRts | 6:32564 | DOI: 10.1038/srep32564 Statistical analysis. Continuous variables are shown as the mean ± the standard deviation. Group differences in the demographic data were evaluated with Pearson's χ 2 tests and independent-samples t-tests for categorical variables and continuous variables, respectively. Deviation from Hardy-Weinberg equilibrium was tested using an exact test. The strength of the associations between HTR3 SNPs and the risk for OCD and its sub-phenotypes (early onset OCD, symmetry, forbidden thoughts, cleaning and hoarding) was examined with binominal logistic regression under dominant, recessive, overdominant, and additive models of inheritance. The analyses were adjusted for age and sex, and the model with the lowest Akaike information criterion was selected as the best-fitting model. The influence of the genetic variants on the disgust sensitivity was evaluated with a one-way multivariate analysis of variance and post-hoc univariate analysis of variance, with the genotype as an independent variable and the DS-R subscales as dependent variables. Analyses were conducted using the R software (version 3.2.1; http://www.r-project.org) and the R package SNPassoc 49 . The overall statistical significance was set at α = 0.005 after Bonferroni correction for the 10 independent SNPs examined. An association was regarded as significant for P < 0.005 and nominally significant for 0.005 ≤ P < 0.05.
Haploview software (version 4.2; http://www.broad.mit.edu/mpg/haploview) was used to estimate the pairwise LD patterns of the examined SNPs. Haplotype blocks were defined by the solid spine of LD method with a D′ threshold of 0.8 50 . Haplotype-based associations were analysed using the R package haplo.stats 51 , which estimates haplotype frequencies with an expectation-maximization algorithm. Haplotype-specific score statistics were computed to test for associations between the haplotype distributions and OCD under an additive model with the haplo.score function. Odds ratios and 95% confidence intervals were calculated using the haplo.cc function. A permutation procedure (n = 100,000) was performed in order to estimate the corrected significance of the best results.