Association of Serotonin2c Receptor Polymorphisms With Antipsychotic Drug Response in Schizophrenia

There is conflicting evidence for the association between genetic polymorphisms in the serotonin (5-HT)2C receptor (HTR2C) and response to antipsychotic drugs (APD) in schizophrenic patients. We tested the association between the HTR2C polymorphisms, Cys23Ser, −759C/T, and −697G/C, and response to APDs (mainly clozapine) in a 6 month prospective study in 171 patients with schizophrenia. Ser23 was significantly associated with treatment response (positive symptoms, X2 = 7.540, p = 0.01; negative symptoms, X2 = 4.796, p = 0.03) in male patients only. A −759C-Ser23 haplotype was similar associated with positive (X2 = 6.648, p = 0.01) and negative (X2 = 6.702, p = 0.01) symptom improvement. Logistic regression, after controlling for covariates, also showed significant haplotypic associations. A meta-analysis of six studies for Ser23 and treatment response showed an overall odds ratio of 2.00 (95%CI, 1.38–2.91, p = 0.0003) or 1.94 (95%CI, 1.27–2.99, p = 0.0024) under fixed or random effect models. These results provide additional evidence that HTR2C polymorphisms are associated with treatment response to APD with HTR2C antagonism or inverse agonism, in male schizophrenic patients.


INTRODUCTION
The serotonin (5-HT) 2C receptor (HTR2C), located at Xq24, belongs to the seven-transmembranespanning G protein-coupled receptor superfamily. It is widely distributed in brain regions which are relevant to schizophrenia. HTR2C receptors exert a tonic inhibitory effect on dorsal and ventral striatal, limbic, hippocampal, and cortical dopamine (DA) release (1,2), modulate serotonergic activity in the dorsal raphe (3), and regulate 5-HT and glutamate efflux in rat cortex (4).
HTR2C is involved in the neurobiology of schizophrenia and the efficacy and side effects of some APDs. It is one of the key regulators of dopaminergic activity in the limbic system. Stimulation of DA D2 receptors in the ventral and dorsal striatum can lead to delusions and hallucinations (5). Activation of HTR2C receptors might lead to decreases in DA release in key brain regions for schizophrenia (6). HTR2C are expressed on principal neurons and GABAergic interneurons in the prefrontal cortex (7) and, thus, may be relevant to the hypoglutamatergic basis for various components of the schizophrenia syndrome (8). The ability of 5-HT-stimulated and constitutively active HTR2C receptors to inhibit DA release in limbic brain areas has been postulated to cause psychosis and to modulate the efficacy of APDs that act by blocking DA D2 receptors (9)(10)(11)(12). Blockade of the constitutive activity of HTR2C receptors enhances cortical and limbic DA release by some APDs (12). The ability of HTR2C agonists to reduce DA release from terminals of VTA neurons in mesolimbic areas is consistent with the antipsychotic effect of the HTR2C agonist, vabicaserin (13,14). Vabicaserin is effective in reversal of phencyclidine and amphetamine-induced hyperactivity (15).
Evidence from genetic association studies also implicates HTR2C in a variety of neuropsychiatric diseases. The HTR2C has a well-characterized promoter region harboring multiple polymorphisms ( Figure 1A), suggesting their potential impact on CpG methylation and putative transcription factor binding, resulting in alteration of HTR2C expression. −759C/T and −697G/C are the most widely investigated promoter polymorphisms. −759C/T polymorphism is associated with antipsychotic induced weight gain (17). −759C/T or −697G/C, has also been linked to therapeutic response to APDs (18)(19)(20). However, these results are contradictory to each other with regard to gender and risk allele.
Cys23Ser, is a non-synonymous SNP which results in an amino acid substitution of cysteine to serine at position 23. This substitution can disrupt a disulfide bridge and potentially alter the structure or stability of the HTR2C protein (21). Although this functional polymorphism has been found to be associated with numerous neuropsychiatric diseases, including anorexia nervosa (22), unipolar, and bipolar depression (16,23), psychotic symptoms in late-onset Alzheimer's disease (24), vulnerability to cocaine cue reactivity (25), migraine with aura (26), and stressrelated cortisol levels (27), it's association with schizophrenia is less clear (28). An association between the Cys23Ser and visual hallucinations and depression in schizophrenia patients has been reported (24), but has not been replicated by others (29,30). Cys23Ser is associated with chronic hospitalization in schizophrenia patients (31) and APD-induce extrapyramidal side effects (32) suggests it is more related to the impact of treatment on the disease process, possibly through effects on dopaminergic activity. Most importantly for this study, Cys23Ser has also been linked in some (33,34) but not all studies (35)(36)(37) to the extent of response to clozapine. These studies will be the subject of a metaanalysis included in the Results section. Clozapine response in treatment-resistant schizophrenia patients does not occur within the conventional 6 week clinical trial period in many patients (38,39). Due to the inconsistent relationship between HTR2C polymorphisms and the psychopathology of schizophrenia, the response to clozapine, and functional activity assays, we examine all three widely-investigated HTR2C SNPs as possible predictors of response of positive and negative symptoms to APD treatment in schizophrenia. Male and female subjects were analyzed separately in order to determine a possible association with gender. Finally, a meta-analysis was performed to determine the overall association between HTR2C polymorphisms and response to clozapine and other APDs.

Subjects
The 171 (male/female, 115/56) patients with schizophrenia or schizoaffective disorder who participated in this study were part of an NIMH-sponsored extramural clinical research center at Case Western Reserve University School of Medicine and Vanderbilt University School of Medicine. Details about recruitment and assessment of subjects have been reported previously (40). Categorical treatment response was evaluated at 6 week and 6 months, using the criteria based upon Kane et al. (41). Subjects with a reduction of >20% in the Brief Psychiatric Rating Scale (BPRS) total score or subcategories, BPRS Positive and Negative, was defined as a responder. Patients were treated with standard doses of the following atypical antipsychotic drugs: clozapine, 550 mg(400-900 mg), 78%; melperone, 250 mg(100-400 mg), 7.0%; risperidone, 6 mg(4-8 mg), 3.8%; or olanzapine, 20 mg(15-40 mg), 2.1%, or typical antipsychotic drugs, mainly haloperidol (10 mg, 9.0%). Antidepressants (14%) and mood stabilizers (5%) were used sparingly.

Statistical Analysis (see Supplementary Information for Details)
We analyzed the males and females separately. The relationship between genotypes and demographic variables was analyzed using chi-square (χ 2 ) or ANOVA. Genotype or haplotype associated differential response to APD was initially evaluated by χ 2 test and then ANCOVA (SPSS), adjusted for race, drugs, age of onset, and the corresponding baseline psychopathology or status of early response. Statistical significance was defined as p < 0.05. As all results were considered exploratory, there was no adjustment for multiple testing. Mapping cis eQTL or methylation QTL was performed using Braincloud data (42,43). In order to review and elucidate the general relationship between HTR2C polymorphisms and drug response to APDs, a meta-analysis of six studies, including ours, with accessible genotyping data for Cys23Ser and binary outcome for symptom improvement, were conducted by R "meta" package. Heterogeneity among the studies was assessed by means of the I 2 inconsistency test and Cochran's Q statistics under a null hypothesis test in which p < 0.05. FIGURE 1 | Haploview plot of cis-eQTLs identified by Braincloud for HTR2C in EUR. A. Multiple custom annotation tracks were created at UCSC Genome Browser to show the cis eQTLs for HTR2C. "expression/CpG Probe" track represents the probe location for gene expression (blue) and CpG island (red); (A) Single expression probe selected for HTR2C is located at the 3 ′ end of mRNA which represents the major form of HTR2C cloned from multiple brain regions according to NCBI Aceview. "HTR2C CisAssoc" track represents identified cis-eQTLs (p < 0.05) by Braincloud data (purple) and all other SNPs in LD (r 2 > 0.8) with cis-eQTLs (red). We successfully identified the tag SNPs for the candidate SNPs (r 2 > 0.85) by SNAP, based on earlier Hapmap data (16) in EUR. rs12846241 (r 2 = 0.96), rs498207 (r 2 = 0.98), and rs5987834 (r 2 = 0.86) are the tag SNPs, genotyped in Braincloud, for −759C/T, −697G/C, and Cys23Ser, respectively, in EUR. Most of cis-eQTLs are aggregated at the promoter or 5 ′ UTR regions as expected. −759C/T, tagged by rs12846241, has been confirmed as cis-eQTL for HTR2C. The T genotype has a lower mRNA level than C genotype (p = 0.006). −697G/C, tagged by rs498207, also suggests a potential impact on gene expression (p = 0.043), but not as strong as −759C/T. Cys23Ser and its tag SNP rs5987834 were negative controls (blue); "CpG Islands" track represents putative CpG Islands for HTR2C. We also include putative cis regulatory elements identified by DNase I hypersensitivity analysis on several cell lines with potential neuronal lineage. (B) A haploview plot of those SNPs in EUR from 1,000 Genome data. Grayscale from black to white is correlated with r 2 from high to low in each box. (C) Haploview plot of genotyped HTR2C SNPs in our sample separated by ethnicity. The LD and haplotype pattern of genotyped SNPs from our EUR samples matched those from 1,000 Genome EUR sample.

Demographic Data Grouped by Genotypes of Three HTR2C SNPs
Cys23Ser, −759C/T, and −697G/C were genotyped for schizophrenic patients with European (EUR, n = 118) and African (AFR, n =53) ancestry Table 1. In the male group, the age at onset for the Ser23 carriers was significantly older than that for non-carriers (p = 0.009; Table 1A). This difference was not observed in the females (p = 0.899; Table 1B). There was no significant difference in the proportions of patients who were treatment resistant or unmedicated at baseline between the genotypes for each SNP. Duration of illness and number of previous hospitalization also did not differ. Although Ser23 carriers had a higher total BRPS score in the male patients (p = 0.04), there was no significant difference with regard to the SNP ID  The relationship between genotypes and demographic variables was analyzed using chi-square (χ 2 ) or ANOVA. p-values reported are two tailed whenever applicable. Statistical significance was defined as p < 0.05. a Self-described ethnicity.
Frontiers in Psychiatry | www.frontiersin.org subcategories of psychopathology including positive, negative, and anxiety/depression subscales. Race, drug, age of onset, and baseline psychopathology were included as covariates in the following ANCOVA.

Genotype/Haplotype Associated Differential Response to APDs
We successfully identified the tag SNPs for the candidate SNPs (r 2 > 0.85) by SNAP and these tag SNPs are in LD with the three candidates by a haploview analysis of the genotype data from 1,000 Genome  Table 2). Cys23Ser, tagged by rs5987834, has no impact on gene expression and % methylation. Based on the previous studies (genetic and functional) and our cis eQTL findings, −759C/T, −697G/C, Cys23Ser, and the combinations of two or all three, were targets for the subsequent genotype-phenotype association study. Since in vitro functional assays indicated that −759C/T and Cys23Ser have a significant impact on HTR2C activity through distinctive mechanisms, we further explored if −759C-Ser, "a super combination, " we propose it produces a greater expression of the constitutively more active form of HTR2C, may demonstrate an even stronger association with dichotomous symptom improvement in an additive mode, after treatment with the APDs, which are HTR2C inverse agonists or antagonists studied here. A significant association between Cys23Ser and dichotomous treatment response was observed only in the male group for both positive symptoms, X 2 = 7.540, p = 0.006; and negative symptoms, X 2 = 4.796, p = 0.029, at 6 month (Table 2A). Haplotype analysis showed that −759C-Ser23 maintained the same level of significant association with positive symptom improvement (X 2 = 6.648, p = 0.010) and negative symptom improvement (X 2 = 6.702, p = 0.010) at 6 month ( Table 2A). All of the above significant findings were only observed in the male patients, except for a borderline significance for Cys23Ser associated with negative symptom improvement in female (X 2 = 3.9, p = 0.048) at 6 month (Supplementary Table 3).
ANCOVA test on absolute change (Table 2B) or % change in BPRS (data not shown) in symptom improvement after controlling for race, drugs, age of onset, and the corresponding baseline psychopathology indicated that male Ser23 carriers had a significant improvement in positive and negative symptoms (p = 0.025 and 0.019, respectively) after 6 months treatment (Table 2B). Neither −759C/T nor −697G/C alone were significantly associated with symptom improvement.
A similar significant association was observed between Cys23Ser and positive/negative symptom improvement in male subjects treated with clozapine only (Supplementary Table 4). Female Ser23 carriers also showed an association with negative symptom improvement at 6 week.

Meta-analysis and Power Test
Six studies from Table 3, including ours, with accessible genotyping data for Cys23Ser and binary outcome for positive symptom improvement in EUR, were included in a metaanalysis. We reported the overall odds ratio is 2.00 (95%CI, 1.38-2.91, p = 0.0003) or 1.94 (95%CI, 1.27-2.99, p = 0.0024) under the fixed or random effect models, respectively (Figure 2). The heterogeneity between the studies was insignificant (Cochran's Q = 6.03, p = 0.30; I 2 = 0.17 (95% CI, 0.00 to 0.62). QUANTO 1.2 was used to calculate the power of the test. The Ser23 carriers were found to have a frequency of 0.15 to 0.45 according to Table 3. As Ser23 carriers increased the odds of having treatment response by 2.0, population risk (Kp) = 0.30, dominant mode of inheritance, and 216 responders/738 non-responders were genotyped, the power (chance) to detect an association with significance p < 0.01 was over 90%.

DISCUSSION
We tested the association between the HTR2C polymorphisms, Cys23Ser, −759C/T, and −697G/C, and treatment response in 171 schizophrenic patients after treatment with APDs, mainly clozapine, for 6 months. One of the strengths of this study was that the majority of the patients were unmedicated at the time of initial assessment. Both Ser23 and −759C-Ser23 haplotype were significantly associated with positive and negative symptom improvement in male patients and Cys23Ser, not the promoter polymorphisms, is the major genetic contributor of the HTR2C in modulating symptom improvement to clozapine. The previously published association studies ( Table 3) showed no consistent results. Attempted replication studies (37, 46) for Sodhi et al.' (33) reported negative results, but the individual pvalues as well as the p-value from a meta-analysis (47) were suggestive of a trend for association (Masellis, p = 0.18, Vehof, p = 0.13, and Gressier, p = 0.12). Our meta-analysis of six original studies (33,(35)(36)(37)46) suggests that HTR2C Cys23Ser is associated with symptom improvement after treatment with clozapine. This is consistent with a previous meta-analysis that included several APDs (48). Some atypical APDs, e.g., clozapine, olanzapine, risperidone, and sertindole, are potent inverse agonists of both HTR2C and HTR2A receptors (49,50). On the other hand, some typical APDs, e.g., chlorpromazine, thioridazine, spiperone, and thiothixene, are HTR2C neutral antagonists, which would preclude their affecting the constitutive activity of HTR2Cs, although the combination of a neutral antagonist and inverse agonist, could lead to blockade of the neutral antagonist (51). The Ki (nM) for D2, HTR2A, and HTR2C are provided in Supplementary Table 5 for each APD (51). Studies based mainly on a single APD are more likely to generate positive results than those based on diverse drug treatments because APDs have variable effects on non-5-HT2C, receptors which can affect their actions as antipsychotics and cognitive enhancers (49,50). However, the meta-analysis reported here suggests our findings may generalize to atypical APDs which are HTR2C antagonists or inverse agonists at clinically effective doses.
Many factors may contribute to the inconsistent results in pharmacogenetic studies of APDs response. These include the heterogeneity in patient populations, utilization of different rating scales, definition of response, frequency of genetic variants in distinct ethnic groups, APDs which differ with regard to A. Haplotype association analysis of HTR2C SNPs with treatment response at 6mon by Chi-square. Positive 4-item includes suspiciousness, hallucinatory behavior, unusual thought content, and conceptual disorganization. Positive 3-item = Positive 4-item without "concept disorganization." Negative 3 items includes emotional withdrawal, motor retardation, and blunted affect; Male and female subjects were analyzed separately. B. Ser23 carriers have a better symptom improvement in male subjects. All data was presented as Mean ± SD for change (absolute change), which is calculated by (6 week or 6 month-Baseline); & represents F statistic and p-value calculated from ANCOVA on change at 6 week or 6 month after controlling for race, drugs, age of onset, and the corresponding baseline psychopathology. # represents the result from Levene's test of equality of error variances with p < 0.05, which against the null hypothesis that the error variance of the dependent variable is equal across groups, suggesting ANCOVA assuming homogeneity of variance is rejected.
Frontiers in Psychiatry | www.frontiersin.org HTR2C pharmacology and actions on other receptors which impact response, e.g., D2, 5-HT1A, and alpha 2 adrenoreceptors, duration of clinical assessment, proposed mode of inheritance, and statistical methods.
In vitro functional studies provide a partial explanation of why Cys23Ser has a main effect on response to APD treatment. Ser23 receptor displayed greater constitutive activity to mobilize calcium than Cys23 receptor (52). Ser23 receptor had greater cell surface expression and more rapid resensitization following exposure to SB206553, a mixed HTR2B antagonists and HTR2C inverse agonist (53). It may be concluded that prolonged exposure of both HTR2C isoreceptors to an inverse agonist increases receptor responsiveness to endogenous 5-HT or other HTR2C agonists, and cells or presumably individuals carrying Ser23 have prompter response to the stimuli than Cys23 carriers. Dopaminergic circuitry is more sensitive to pain stress in Ser23 carriers (54). Greater dopamine release in the nucleus accumbens, caudate nucleus, and putamen was observed in the Ser23 carriers during pain, suggesting mesoaccumbal stress sensitivity may mediate the effects of HTR2C variation on the risk of neuropsychiatric disorders. Significant differences in regional cerebral blood flow between Ser23 and Cys23 male carriers after treatment with serotonin agonist meta-Chlorophenylpiperazine suggests that this polymorphism does have distinct functional consequences (55).
In conclusion, these results provide additional evidence that HTR2C polymorphisms, particularly Cys23Ser, are associated with response to APD treatment, mainly clozapine with HTR2C antagonism or partial agonism, in male schizophrenic patients.

ETHICS STATEMENT
This study was carried out in accordance with the recommendations of Internal Review Board of Vanderbilt University and Case Western Reserve University with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Internal Review Board of Vanderbilt University and Case Western Reserve University.

AUTHOR CONTRIBUTIONS
JL mostly contributed to data analysis, interpretation of results, and manuscript writing. HH contributed to initial data analysis and manuscript writing. HM has designed the study and written the manuscript in close collaboration with JL. All authors have approved the final manuscript.

FUNDING
This research was supported by donations from the Weisman Family, Mr. Michael Burke, and Mr. Michael Shmerling. Neither had any role in study design, data collection, analysis and interpretation, writing of the report, or the decision to submit the paper for publication.

ACKNOWLEDGMENTS
This study was first presented at the Annual Meeting of American Society of Human Genetics in 2014 as a poster presentation and the abstract was collected at www.ashg.org/2014meeting/ abstracts/fulltext/f140120372.htm.