Non-synonymous single-nucleotide variations of the human oxytocin receptor gene and autism spectrum disorders: a case–control study in a Japanese population and functional analysis

Background The human oxytocin receptor (hOXTR) is implicated in the etiology of autism spectrum disorders (ASDs) and is a potential target for therapeutic intervention. Several studies have reported single-nucleotide polymorphisms (SNPs) of the OXTR gene associated with ASDs. These SNPs, however, reside outside the protein-coding region. Not much is known about genetic variations that cause amino acid substitutions that alter receptor functions. Methods Variations in the OXTR gene were analyzed in 132 ASD patients at Kanazawa University Hospital in Japan and 248 unrelated healthy Japanese volunteers by re-sequencing and real-time polymerase chain reaction-based genotyping. Functional changes in variant OXTRs were assessed by radioligand binding assay and measurements of intracellular free calcium concentrations ([Ca2+]i) and inositol 1,4,5-trisphosphate (IP3) levels. Results Six subjects (4.5%) in the ASD group and two in the control group (0.8%) were identified as heterozygotes carrying the R376G variation (rs35062132; c.1126C>G); one individual from the ASD group (0.8%) and three members of the control group (1.2%) were found to be carrying R376C (c.1126C>T). The C/G genotype significantly correlated with an increased risk of ASDs (odds ratio (OR) = 5.83; 95% confidence interval (CI) = 1.16 to 29.33; P = 0.024, Fisher’s exact test). Consistently, the G allele showed a correlation with an increased likelihood of ASDs (OR = 5.73; 95% CI = 1.15 to 28.61; P = 0.024, Fisher’s exact test). The frequencies of the C/T genotype and the T allele in the ASD and control groups did not differ significantly. We also examined changes in agonist-induced cellular responses mediated by the variant receptors hOXTR-376G and hOXTR-376C. OXT-induced receptor internalization and recycling were faster in hOXTR-376G-expressing HEK-293 cells than in cells expressing hOXTR-376R or hOXTR-376C. In addition, the elevation in [Ca2+]i and IP3 formation decreased in the cells expressing hOXTR-376G and hOXTR-376C tagged with enhanced green fluorescent protein (EGFP), in comparison with the cells expressing the common-type hOXTR-376R tagged with EGFP. Conclusions These results suggest that the rare genetic variation rs35062132 might contribute to the pathogenesis of ASDs, and could provide a molecular basis of individual differences in OXTR-mediated modulation of social behavior.

Previous studies have shown that impaired OXTRmediated signaling causes behavioral abnormalities [2,6,14]. Mice deficient in the OXTR gene show diminished social discrimination, increased aggressive behavior, reduced cognitive flexibility, and increased susceptibility to seizures [15,16]. Using homozygous mutant mice lacking CD38, a transmembrane protein with ADP-ribosyl cyclase activity, we have shown that a decrease in the formation of cyclic ADP-ribose (cADPR) results in dysfunction of Ca 2+ -induced Ca 2+ release for OXT secretion in hypothalamic neurons. The reduction in cADPR levels also causes marked defects in maternal nurturing and social behaviors similar to those observed in OXT-and OXTR-knockout mice [17,18]. We also demonstrated that ADP-ribosyl cyclase activity increases after OXTR stimulation, regulating OXT release in the hypothalamus and pituitary in adult male mice [19].
In humans, OXTR has been implicated in the pathogenesis of neuropsychiatric disorders and considered a potential target for therapeutic intervention [20]. Gregory et al. [21] have reported a heterozygous deletion of the OXTR gene, located on chromosome 3p25, in a patient with autism whose mother had a putative obsessive-compulsive disorder. In addition, many studies have demonstrated that single-nucleotide polymorphisms (SNPs) of the OXTR gene are associated with autism spectrum disorders (ASDs), social anxiety disorders, and schizophrenia [22][23][24][25][26][27][28][29]. However, all SNPs reported so far reside outside the protein-coding region, and their functional importance is unclear. If allelic variations leading to amino acid substitutions in the OXTR gene were associated with these neuropsychiatric disorders, then such variations would help to predict the etiological relevance based on protein structure-function relationship and to introduce individualized OXT-based therapy.
Therefore, we analyzed nucleotide variations in the protein-coding regions of the human OXTR gene in ASD patients and unrelated healthy controls. Furthermore, we focused on a triallelic variation (rs35062132; c.1126C> G/T; the nomenclature of variation recommended by the Human Genome Variation Society [30]) and investigated whether the amino acid substitution R376G/C causes any changes in OXTR-mediated cellular responses.

Participants
We recruited 132 ASD subjects (102 males, 30 females; 15.9 ± 0.7 years) from the outpatient psychiatry department of the Kanazawa University Hospital. All subjects fulfilled the DSM-IV criteria for pervasive developmental disorder. The diagnoses were made by two experienced child psychiatrists through interviews and clinical record reviews as described previously [31], and the subjects had no apparent physical anomalies. The two experienced child psychiatrists independently confirmed the diagnosis of ASD for all patients by semi-structured behavior observation and interviews with the subjects and their parents. During each of these interviews with parents, which were helpful in the evaluation of autism-specific behaviors and symptoms, the examiners used one of the following methods: the Asperger Syndrome Diagnostic Interview [32], the Autism Diagnostic Interview-Revised (ADI-R) [33], the Pervasive Developmental Disorders Autism Society Japan Rating Scale [34], the Diagnostic Interview for Social and Communication Disorders [35], or the Tokyo Autistic Behavior Scale [36]. Based on these evaluation methods, 97 patients were classified as autistic disorder (autism), 10 as Asperger's disorder, and 25 as pervasive developmental disorder not otherwise specified (PDD-NOS). The 248 controls (143 males, 105 females; 31.3 ± 0.6 years) were unrelated healthy Japanese volunteers. All patients and controls were Japanese with no non-Japanese parents or grandparents. This study was approved by the ethics committees of Kanazawa University School of Medicine. All examinations were performed after informed consent according to the Declaration of Helsinki.

Re-sequencing
Genomic DNA was extracted from venous blood samples using the Wizard Genomic DNA Purification kit (Promega, Madison, WI, USA), or from nails using the ISOHAIR DNA extraction kit (Nippon Gene, Tokyo, Japan). DNA fragments containing exons 3 and 4 of the hOXTR gene were amplified by PCR in a 20-μl reaction mixture containing 2.5 ng genomic DNA, 200 μM dNTPs, 200 nM of each primer, 4 μl Ampdirect G/C (Shimadzu, Kyoto, Japan), 4 μl Ampdirect-4 (Shimadzu), and 1 unit Ex Taq DNA polymerase (Takara Shuzo, Otsu, Japan). The amplification procedure consisted of denaturation at 96°C for 1 min, followed by 35 cycles of 96°C for 30 s, 65°C for 1 min, and 72°C for 1 min. The primers used were FWD1 or FWD4 and REV1 for exon 3, and FWD5 and REV3 for exon 4 (sequences are given in Table 1). After amplification, PCR products were purified with the High Pure PCR Cleanup Micro kit (Roche, Mannheim, Germany), subjected to cycle sequencing reaction (BigDye version 1.1; Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol, and analyzed on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems). The primers used were FWD1-FWD5 and REV1-REV6 (Table 1).

Real-time PCR
Two primers, FWD8 and REV7 (Table 1), were designed to amplify a 138-bp product encompassing rs35602132 at the mRNA position 1748 (g.8734707; c.1126). Three custom-designed TaqMan minor groove binder (MGB) probes were obtained from Applied Biosystems: 376G-FAM and 376C-FAM targeted to the variant alleles, and 376R-VIC to the common allele (Table 1). PCR was performed in 20-μl mixtures containing 10 μl TaqMan universal master mix (Applied Biosystems), 200 nM of each primer, 100 nM 376R-FAM, 100 nM 376G-VIC or 376C-VIC, and 10 ng of sample DNA. Thermocycling was performed on the ViiA 7 Real-Time PCR System (Applied Biosystems). The amplification was conducted at 60°C for 30 s, 95°C for 5 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. Data were analyzed with SDS2.3 software (Applied Biosystems).
Site-directed mutagenesis cDNAs encoding hOXTR variants at amino acid residue 376 were generated by site-directed mutagenesis as described previously [37]. pCHOXTR [38] harboring hOXTR-376R, a common-type receptor, was used as a template for PCR; the primers used were R376G-FWD and R376G-REV (Table 1) for the R376G substitution, and R376C-FWD and R376C-REV (Table 1) for R376C. The amplified products were treated with EX Taq DNA polymerase (Takara Shuzo) at 72°C for 10 min and cloned into a pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA, USA) to yield pCHOXTR-376G and pCHOXTR-376C. The cDNA inserts were verified by nucleotide sequencing.

Cell culture and transfection
Human embryonic kidney HEK-293 cells and COS-7 cells were maintained in DMEM supplemented with 10% FBS at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . NG108-15 neuroblastoma × glioma hybrid cells were cultured as described previously [39]. Cells were grown in culture dishes to 80 to 90% confluence and transfected with expression plasmids using FuGENE HD Transfection Reagent (Roche) following the manufacturer's instruction.

Measurement of receptor recycling
HEK-293 cells were plated on poly-D-ornithine-coated glass coverslips and cultured overnight. After transfection with pcDNAHOXTR-376R, pcDNAHOXTR-376G, or pcDNAHOXTR-376C, the cells were incubated at 37°C in serum-free DMEM for different times, in the absence or presence of OXT (100 nM). Reactions were stopped by removing the medium and fixing the cells with 4% paraformaldehyde at room temperature for 10 min.

Statistical analyses
Data obtained from genetic studies were analyzed using a contingency table and the Fisher's exact test (GraphPad Prism 5; GraphPad Software Inc.). All data from the in vitro studies are shown as mean ± standard error of the mean. Statistical significance was determined by the Student's t test or two-way analysis of variance (ANOVA) using an interactive fitting program (GraphPad Prism 5; GraphPad Software Inc.); P values smaller than 0.05 were considered to be statistically significant.

Genetic analysis
As the entire amino acid sequence of the hOXTR is encoded by exons 3 and 4, as illustrated in Figure 1A, we initially re-sequenced both exons for genetic variations. Of 27 single-nucleotide variations listed in the NCI dbSNP database, we detected minor alleles of rs4686302, rs151257822, and rs35062132 in 59 ASD patients and 30 unrelated healthy Japanese volunteers ( Figure 1A, Table 2). As c.1126 G>C (rs35062132; R376G) was detected in two ASD patients but not in controls ( Table 2, Figure 1B), we selected rs35062132 for further analysis.
We genotyped an additional 73 ASD patients and 218 healthy volunteers for rs35062132 variation using the TaqMan-based real-time PCR method. In total, we obtained data from 132 ASD patients and 248 controls. Six individualsin the ASD group (4.5%) and two in the control group (0.8%) were identified as heterozygotes carrying the R376G variation (rs35062132; c.1126C>G). One person in the ASD group (0.8%) and three in the control group (1.2%) carried R376C (c.1126C>T; Table 3). No homozygous carriers for the minor alleles were observed. The χ 2 goodness-of-fit test showed that the genotype frequency distribution of rs35062132 did not deviate from Hardy-Weinberg equilibrium in the patient group (χ 2 = 0.098, P = 0.99) or the control group (χ 2 = 0.026, P = 1.00). The genotype and allele frequencies of rs35062132 obtained from all participants are summarized in Table 3. The C/G genotype was significantly correlated with an increased risk of ASDs (odds ratio (OR) = 5.83; 95% confidence interval (CI) = 1.16 to 29.3; P = 0.024, Fisher's exact test; Table 3). Consistently, the G allele also showed a correlation with an increased likelihood of ASDs (OR = 5.73, 95% CI = 1.15 to 28.6; P = 0.024, Fisher's exact test; Table 3). The frequencies of the C/T genotype and the T allele did not significantly differ between the ASD and control groups (Table 3).

Radioligand binding properties of the common and variant hOXTRs
To find out whether the amino acid changes at the residue 376 affect the receptor functions, we compared ligand-binding properties of the common and variant hOXTRs. In saturation binding and Scatchard analysis of 3 [H]-OXT binding to membranes prepared from COS-7 cells transfected with hOXTR-expression plasmids, the variants hOXTR-376G and hOXTR-R376C, as well as the common-type hOXTR-376R, exhibited single highaffinity binding sites. Values of K d were 0.90 ± 0.28, 1.1 ± 0.12, and 1.2 ± 0.19 nM (n = 15 in each group), and values of B max were 3153 ± 531, 3202 ± 234, and 3120 ± 339 fmol/mg of protein (n = 15 in each group) for hOXTR-376R, hOXTR-376G, and hOXTR-376C, respectively. Competition binding experiments confirmed that the variant and common receptors did not differ significantly: values of IC 50 were 1.05 ± 0.47, 0.49 ± 0.23, and 0.71 ± 0.30 nM (n = 15 in each group) for hOXTR-376R, hOXTR-376G, and hOXTR-376C, respectively. These data indicate that both variations have no or little effect on OXT binding.

Agonist-induced receptor internalization and recycling in HEK-293 cells
We then examined the agonist-induced internalization and recycling of hOXTRs in HEK-293 cells, which have been most extensively used for investigation of GPCR internalization and trafficking. HEK-293 cells transfected with an expression plasmid for hOXTR-376R, hOXTR-376G, or hOXTR-376C were stimulated with 100 nM OXT for 5, 10, 15, 30, 45, 60, 90, and 120 min. At each time point, the cells were fixed and stained with antibody pecific for the amino (N)-terminus, located extracellularly, of the OXTR under impermeable conditions. The specificity of the staining was confirmed by the comparison with the staining of mock-transfected and non-transfected HEK-293 cells (Additional file 1: Figure S1).

Discussion
The main findings of this study are as follows. (i) Triallelic non-synonymous variation (rs35062132, c.1126C>G/T; R376G/C) is observed in both ASD patients and healthy controls in a Japanese population; the frequencies of the G allele are significantly higher in the ASD patients than in healthy controls. (ii) In HEK-293 cells expressing hOXTR-376G, the agonist-induced internalization and recycling of the OXTR are faster than that in the cells expressing hOXTR-376C or hOXTR-376R. (iii) In both HEK-293 cells and NG108-15 neuronal cells, the agonist-induced increase in [Ca 2+ ] i mediated by hOXTR-376G-EGFP or hOXTR-376C-EGFP is smaller than the increase associated with hOXTR-376R-EGFP. These results provide new insights into the genetic architecture and therapeutic aspects of ASDs.
The present study is unique in associating nonsynonymous allelic variations of the OXTR gene with ASDs. These variations at rs35062132 have not been reported in other disorders. To date, several studies have found ASD-associated SNPs in the OXTR gene, which reside in introns (rs4564970 [41], rs237897 [25], rs53576 [23,24], rs2254298 [23][24][25], rs2268493 [42], rs7632287 [42], rs11720238 [41]), 3'-untranslated region (rs1042778) [42], and intergenic regions (rs7632287 and rs11720238) [41]. The functional consequences of these SNPs remain unclear. Importantly, all minor alleles of these SNPs have a frequency more than 5%, and thus can be classified as common variations. In contrast, the frequencies of the G and T alleles of rs35062132 are 0.4% and 0.6%, respectively, in controls; these minor alleles can be categorized as rare [43][44][45]. Therefore, most previous studies favor the common disease-common variant model, in which most of the risk is caused by common genetic variations (>5% allele frequency), each allele conferring a slight risk [43][44][45][46][47]. By contrast, our results support the common disease-rare variant model, in which the risk is mostly attributable to rare variations (<5% allele frequency), each variant conferring a moderate but readily detectable increase in the relative risk [43][44][45][46][47]. In agreement with this model, we observed that the minor alleles clearly affected OXT-induced cellular responses. Future studies will be conducted to test whether behavioral changes caused by these variations are moderate and reasonably deleterious. Also it will be interesting to examine whether the rs35062132 variations cosegregate with the risk alleles in the non-coding region [22][23][24]41,42].
The limitation of this study is that sample size is not sufficient to analyze the rare variant alleles at rs35062132. Future studies with larger sample size or family-based association testing are necessary to conclude that the G allele of the rs35062132 is a genetic risk factor for ASDs. Values are mean ± standard error of the mean (n = 24 to 30 cells from three or four independent cultures; six to ten cells were analyzed in each culture). Statistical significance was evaluated by the Student's t test (* P <0.05, **P <0.01; compared with hOXTR-376R-EGFP).
The result reported here should be replicated in independent populations with various ethnic backgrounds.
The amino acid variation R376G/C is located in the intracellular carboxy (C)-terminal tail of the receptor protein, which is critical for desensitization, internalization, and recycling of the OXTR, as in many other GPCRs [9][10][11][12][13]. Arg 376 is associated with two structural components involved in these processes: (i) Arg 376 is a part of a PKC consensus motif (Ser 374 -His 375 -Arg 376 ; Figure 1A), which is thought to interact with PKC after OXT stimulation [10]; and (ii) Arg 376 is in conjunction with one of the two Ser triplets in the C-terminus ( Figure 1A) that are primary sites of agonist-induced phosphorylation and β-arrestin-2 binding [11]. When either one of the two Ser triplets was mutated to alanine residues, the stability of the β-arrestin -2-OXTR complex was altered, and the ability of βarrestin-2 to internalize with the receptor was eliminated [11]. In the vasopressin receptor V2, which is highly homologous to OXTR in structure, the equivalent Ser triplets function as a retention motif. GRK phosphorylation of these Ser residues promotes the  formation of stable receptor-β-arrestin complexes, followed by cotrafficking and colocalization of the complexes to endocytotic vesicles and slow recycling to the cell surface [48]. Therefore, it is suggested that the R376G substitution in the OXTR might suppress the phosphorylation of Ser residues in the Ser triplets or reduce the stability of β-arrestin binding to the triplets, resulting in faster receptor recycling.
We demonstrated that, both in HEK-293 cells and in NG108-15 neuronal cells, agonist-induced [Ca 2+ ] i elevation mediated by variant receptors, hOXTR-376G-EGFP and hOXTR-376C-EGFP, is smaller than that by the common receptor hOXTR-376R-EGFP. Despite the marked reduction in the peak [Ca 2+ ] i , the decrease in IP 3 levels was slight. Provided that Arg 376 is not located in the known site for G q binding ( Figure 1A) [49], the G q/11 /PLC β /IP 3independent, unidentified signaling pathway for [Ca 2+ ] i elevation might be affected by the R376G/C substitution.
Ca 2+ signaling pathways in neurons have been implicated in the pathogenesis of ASDs [50]. However, the involvement of OXTR-mediated, PLC/IP 3 -dependent Ca 2+ signaling in ASDs is still largely unknown. Recently, Ninan [51] has demonstrated that U73122, a PLC inhibitor, reduced OXT-induced suppression of glutamatergic synaptic transmission in pyramidal neurons of the medial prefrontal cortex, possibly by inhibiting PLC-dependent increase in postsynaptic calcium. Given that this process is critical for the OXT effects on social cognition, it is conceivable that the OXTR variations may serve as a risk factor for ASDs.
We have previously demonstrated that cADPR, which may act as an intracellular second messenger downstream of the OXTR, activates Ca 2+ release from intracellular stores through the ryanodine receptor Ca 2+ release channel. cADPR also initiates Ca 2+ influx through melastatinrelated transient receptor potential 2 (TRPM2) channels [38]. Our previous SNP analysis has suggested that R140W substitution of CD38 protein, a regulator of ryanodine receptor-mediated Ca 2+ -induced Ca 2+ release for OXT secretion [17], could be a potential risk factor for a subset of Japanese ASD patients [31]. Although this study shares a part of the ASD patients with our previous analysis [31], the 140W allele of CD38 has not been found in the patients carrying the OXTR-376G or OXTR-376C until now (unpublished data). It is therefore conceivable that the underlying signaling process affected by the R376G/C substitution is independent of the CD38-mediated OXT release from hypothalamic neurons [17].
Besides their relevance to the etiology of ASDs, the results presented here might contribute to the development of new pharmacological treatments. Genetic polymorphisms of many drug targets predict responsiveness to drugs [52,53]. It is likely that the alteration in the cellular functions mediated by the variant OXTRs could cause some individual differences in both behavioral and nonbehavioral responses to OXT. Although no side effects specific to intranasal OXT administration have been reported [54], new clinical information regarding functional OXTR variants will be helpful in the determination of individual administration protocols to maximize therapeutic benefit with least adverse effects.

Conclusions
Our analysis of the genetic variation rs35062132 in the human OXTR gene may be important for both etiological and therapeutic aspects of ASDs. However, the results reported here should be reproduced in an independent population. Further studies with a larger sample size using subjects of different ethnicities are required to confirm that the G allele of the rs35062132 is a genetic risk factor for ASDs. Identifying rare functional variants of the OXTR will shed new light on the genetic architecture of ASDs and allow personalized pharmacological intervention using OXT.