GABAA receptor subunit gene polymorphisms predict symptom-based and developmental deficits in Chinese Han children and adolescents with autistic spectrum disorders

GABAA receptor subunit genes GABRB3, GABRA5, and GABRG3 located on chromosome 15q11-q13 have been implicated in the etiology of autistic spectrum disorders (ASD). This study intended to investigate the possible role of single-nucleotide polymorphisms (SNPs) present in GABRB3 (rs2081648 and rs1426217), GABRA5 (rs35586628), and GABRG3 (rs208129) genes in ASD susceptibility and symptom-based and developmental phenotypes of ASD in Chinese Han children and adolescents. 99 ASD patients and 231 age- and gender- frequency-matched typical developing (TD) controls were tested by TaqMan® genotyping assay. Symptom-based phenotypes were evaluated by Childhood Autism Rating Scale (CARS) and Autism Behavior Checklist (ABC), and developmental phenotypes were assessed by Early Childhood Development Questionnaire (ECDQ) in ASD patients. Three haplotypes and global χ 2 test of all SNPs demonstrated significant associations between ASD and TD groups. Besides, GABRB3 rs2081648, GABRA5 rs35586628, and GABRG3 rs208129 polymorphisms were associated with symptom-based deficits in social interaction, sensorimotor and somatosensory coordination, visual response, imitation, activity level, language expression and adaptability. Developmental abnormalities in late emergences of social interaction and fine motor were detected in GABRB3 rs2081648 polymorphism. Overall results indicated that gene synergy may participate in ASD pathogenesis, and GABAA receptor gene polymorphisms can predict symptom-based and developmental deficits in ASD individuals.

GABRB3, GABRA5, and GABRG3 SNPs and ASD symptom-based phenotypes. For the CARS total scores deviated from a normal distribution (p < 0.05), we conducted a Kruskal-Wallis H test to analyze the relationships between CARS total scores and the genotypes of four SNPs, and found no evidence for any associations. Nevertheless, an ordinal polytomous logistic regression analysis showed some significant associations between ratings in items of CARS and genotypes of GABRB3 SNP rs2081648, GABRA5 SNP rs35586628, and GABRG3 SNP rs208129 in ASD patients (while adjusting for sex, age, and IQ). ORs (95% CI) for the '4 = severely abnormal' among different genotypes were shown in Table 3. ASD patients with a TT genotype of GABRB3 SNP rs2081648 trended to suffer from severe abnormality of 'visual response' compared to those with a TC genotype. Moreover, ASD participants with CC and TT genotypes of GABRA5 SNP rs35586628 were likely to perform worse in 'verbal communication' than those with a CT genotype, and ASD individuals with TA and TT genotypes of GABRG3 SNP rs208129 showed worse performance in 'imitative behavior' and 'activity level' than those with an AA genotype.
Scores of each ABC category and ABC total scores conformed to a normal distribution and homogeneity of variance (p > 0.05), but no significant associations between genotypes of four SNPs and all ABC scores were observed after the one-way ANCOVA with sex, age, and IQ as covariates (p > 0.05). However, a binary logistic regression analysis found some significant associations between ratings in items of ABC and genotypes of GABRB3 SNP rs2081648, GABRA5 SNP rs35586628, and GABRG3 SNP rs208129 in ASD patients (while adjusting for sex, age, and IQ). ORs (95% CI) for '1 = yes' of existing relevant symptoms among different genotypes were shown in Table 4.

GABRB3, GABRA5, and GABRG3 SNPs and developmental phenotypes of ASD. An ordinal
polytomous logistic regression analysis detected significant associations between ratings in two items of ECDQ and genotypes of GABRB3 rs2081648 in ASD patients (while adjusting for sex, age, and IQ) (n = 9 TT, n = 31 TC, and n = 27 CC). The OR for '3 = late' of 'smiling to his/her mother' in those with a TT genotype compared to those with a CC genotype was 4.39 (95% CI = 1.43-13.52, p = 0.010), those with a TC genotype compared to those with a CC genotype was 0.21 (95% CI = 0.09-0.48, p = 0.000), and those with a TT genotype compared to those with a TC genotype was 20.86 (95% CI = 3.39-128.50, p = 0.001). We found that ASD patients with a TT genotype trended to show social interaction in older ages. In addition, the OR for '3 = late' of 'grasping things by himself/herself ' in those carrying a TT genotype compared with those carrying a CC genotype was 2.26 (95% CI = 0.86-5.91, p = 0.098), those carrying a TC genotype compared with those carrying a CC genotype was 0.28 (95% CI = 0.13-0.63, p = 0.002), and those carrying a TT genotype compared with those carrying a TC genotype was 8.02 (95% CI = 1.59-40.56, p = 0.012). As a result, ASD participants with a TC genotype developed earlier in fine motor.  Table 2. Haplotype analysis for the genetic association of GABRB3, GABRA5, and GABRG3 between ASD patients and TD controls. Haplotypes with a frequency of < 0.03 in both groups have been dropped. Haplotypes were constructed with rs2081648-rs1426217-rs35586628-rs208129 in proper order. *p < 0.05; **p < 0.01. ASD, autistic spectrum disorders; CI, confidence interval; ORs, odds ratios; TD, typical developing.

Discussion
To our knowledge, this is the first age-and gender-frequency-matched case-control study designed to investigate the association between the three GABA A receptor genes, GABRB3, GABRA5, and GABRG3, and ASD in Chinese Han children and adolescents, as well as to detect relevant symptom-based and developmental deficits associated with the three GABA A receptor gene polymorphisms in Chinese Han children and adolescents with ASD.
Previous studies have investigated the association of the three GABA A receptor genes with ASD, but the contradictory findings demand subsequent replications in different populations. In this study, we found no positive evidence of the associations between any single SNP of the three GABA A receptor genes and ASD. Similar to our result, McCauley et al. 27 demonstrated the negative evidence for the associations between three SNPs (rs2081648, rs1426217, and rs208129) and ASD, as well as Kim et al. 46 reported no significant supports for four SNPs (rs2081648, rs1426217, rs35586628, and rs208129) of ASD mainly in Caucasian population with family trios. However, Kim et al. 29 found an allele at SNP rs2081648 exhibited prior transmission in Korean trios. These differences may originate from the multiple demographic characteristics of the ASD populations (age, gender, ethnicity, sample size, diagnostic criteria, etc.) and different investigative methods (family-based association,  case-control association, etc). The statistical power of our case-control association study is approximately 0.65 (α = 0.05) under the assumptions of 0.28% prevalence of ASD in Tianjin of China 49 and genotypic relative risk of 1.8 34 . Thus small effect might not be detected because of the sample size. Moreover, we detected the TD controls had lower D′ than ASD patients, which predicted that TD individuals might have more chance to restructure randomly in accordance with genetic equilibrium than ASD individuals. Further studies can apply family-based association test (FBAT) in Chinese Han family trios of ASD and TD children and adolescents with the three GABA A receptor genes to discover more significant evidence on this difference. In addition, our study revealed that individuals with a CATT haplotype and a TGCA haplotype were more likely to have ASD, but individuals with a TACA haplotype had more potential to be TD controls. Our findings supported that the etiology of ASD may include gene-gene interactions rather than the effect of one single gene 50 .
For the limited evidence on the association between GABA A receptor genes and characteristic phenotypes of ASD, we further observed some significant associations between three SNPs of the three GABA A receptor genes and phenotypic features of ASD without GABRB3 SNP rs1426217. In the symptom-based phenotypes, as a supplement for the detections of associations between GABRB3 region and high degree of insistence-on-sameness 21 , as well as savant skills 41 in previous studies, the ASD patients in our study with a TT genotype of GABRB3 SNP rs2081648 trended to perform worse in 'visual response' of CARS, and those with a CC genotype of GABRB3 SNP rs2081648 were more likely to show a deficit in social interaction according to the 'relating' part of ABC (38 = (R) Has not developed any friendships). Moreover, ASD patients with a CC genotype of GABRA5 SNP rs35586628 exhibited a deficit in language expression or cognitive functioning basing on the 'verbal communication' of CARS and the 'language' part of ABC (37 = (L) Cannot point to more than five named objects), and the others with a CT genotype showed sensorimotor and somatosensory abnormalities (13 = (R) Does not reach out when reached for; 29 = (S) Sometimes shows no 'startle response' to loud noise), and rigid language patterns (48 = (L) Repeats sounds or words over and over) according to the 'relating' , 'sensory' and 'language' parts of ABC. Our finding provided more supports for the relationships between genotypes of GABRA5 and core features of ASD than Kim et al. 46 , who only reported a significant association between one GABRA5 SNP rs2075716 and 'relative failure to initiate or sustain conversational interchange' . Furthermore, the ASD patients with a TA genotype of GABRG3 SNP rs208129 showed the highest potential to bear the most severe deficits in 'imitative behavior' and 'activity level' of CARS, as well as to display insistence-on-sameness or worse adaptability (14 = (s′) Strong reactions to changes in routine/environment), somatosensory disability (26 = (S) Sometimes painful stimuli such as bruises, cuts, and injections evoke no reaction) and difficulties in social communication (56 = (L) Uses at least 15 but less than 30 spontaneous phrases daily to communicate) in the 'social and self-help' , 'sensory' and 'language' parts of ABC. Above all, the heterozygotes of GABRA5 SNP rs35586628 and GABRG3 SNP rs208129 showed the primary potential to increase the probability for ASD patients to undertake relevant core deficits, learning disabilities, and especially the decreased sensory sensitivity in voice and pain conformed abnormal sensory processing as one of the core features of individuals with ASD 1 . In the developmental phenotypes, ASD patients with a TT genotype of GABRB3 SNP rs2081648 were likely to show negative effects on initial social interaction and fine motor with the late emergences of 'smiling to his/her mother' and 'grasping things by himself/herself ' . As far as we know, there has been no previous studies intending to distinguish different phenotypes of early childhood development based on SNPs in children and adolescents with ASD. However, during the fetus period, the GABA A receptor subunit cluster (containing GABRB3, GABRA5, and GABRG3) on chromosome 15q11-q13 region acts as a developmental role in GABAergic signaling for building neuronal connectivity, and plays a pivotal role in the maintenance of inhibitory tone in the adult brain 41 . Therefore, any abnormalities in the three genes are possible to cause developmental deficits in the whole life. Subsequent researches should pay more attention to the associations between polymorphisms of GABA A receptor genes and early childhood development in both ASD patients and TD controls to obtain better and precise clues of early diagnosis for ASD patients, so as to intervene at an earlier time.
In conclusion, our findings are supportive of the fact that different subunits of GABA A receptor subtypes may work together in the pathogenesis of ASD, as well as play differential roles in the symptom-based and developmental deficits in Chinese Han children and adolescents with ASD. To date, very few studies have explored the role of these genetic variants of GABRB3, GABRA5, and GABRG3 genes in both ASD susceptibility and phenotypes, replication studies with a larger sample size are required to confirm our findings in different races.

Materials and Methods
Participants. The study protocol was approved by the Tianjin Medical University Institutional Review Board (Ethics Committee of Tianjin Medical University) and written informed consent was obtained from school principals, parents and/or caregivers to allow the collection and genetic analysis of DNA samples after a complete and extensive description of the study, in accordance with the Declaration of Helsinki. All participants enrolled into this study were Chinese Han children and adolescents from Tianjin, China. We recruited 99 ASD patients (79 males and 20 females) from five local special education schools and 231 age-and gender-frequency-matched TD controls (185 males and 46 females) from four local mainstream schools with a male to female ratio of 4:1 between 2 and 18 years of age. The diagnosis of ASD was made by qualified and experienced psychiatrists basing on the criteria of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) 51 and was further identified using the Childhood Autism Rating Scales (CARS) 52 . A CARS total score of ≥ 30 is the cutoff for distinguishing children 'at risk' for autism from pervasive developmental disorder-not otherwise specified (PDD-NOS) 53,54 , and a CARS total score of ≥ 25.5 is indicative of ASD 54 . Meanwhile, the Autism Behavior Checklist (ABC) 55 was also administered for the assessment of their symptom-based phenotypes and Early Childhood Development Questionnaire (ECDQ) was used to evaluate their developmental phenotypes. In addition, intelligent quotient (IQ) was determined using the Chinese Binet Scale (Binet) 56 by clinical pediatric psychologists. ASD subjects were included with a full-scale IQ of ≥ 30 and TD controls were included with an IQ in the average range. Participant characteristics are presented in Table 5. In the present study, none of the participants in any group were taking drugs. Exclusion criteria for both groups included medical history of epileptic seizures, head trauma, and known psychiatric, neurological or genetic disorders, and TD controls were with a score of < 7 on the Clancy Autism Behavior Scale (CABS) 57 55 . The CARS has been proved to have a high degree of internal consistency, inter-rater and test-retest reliability, good discriminant validity, and high criterion-related validity 58 . It is widely used by psychiatrists to identify children with autism and as a further measurement of the severity of this disease. Total scores can range from a low of 15 to a high of 60; scores of less than 30 indicate that the individual is in the non-autistic range, scores between 30 and 36.5 indicate mild/moderate autism, and scores from 37 to 60 indicate severe autism 53 . In this study, all ASD patients got this assessment.
Autism Behavior Checklist (ABC). The Autism Behavior Checklist (ABC) is a behavior checklist that consists of 57 items in 5 categories: 'sensory (S) (9 items)' , 'relating (R) (12 items)' , 'body and object use (B) (12 items)' , 'language (L) (13 items)' , and 'social and self-help (s′) (11 items)' . Each item corresponds to a single score referring to a single symptomatological area 55 . For each item, we set the ratings as a two-point scale (0 = no; 1 = yes). The scale utilizes an observer's rating of a series of typical autistic behaviors in a certain subject and provides advice for educational intervention. In this study, all ASD patients got this assessment.
Early Childhood Development Questionnaire (ECDQ). The initial ages of early childhood development were assessed by parents or caregivers according to the normal range of age in different developmental parts of the ECDQ. It contains 9 items, including 'smiling to his/her mother' , 'grasping things by himself/herself ' , 'sitting by himself/herself ' , 'walking by himself/herself ' , 'calling daddy or mummy' , 'speaking phrases' , 'controlling defecate and urinate' , 'stopping wetting the bed' , 'wearing by himself/herself ' . The developmental stage was evaluated on the ratings of a three-point scale (1: early, 2: normal, 3: late) for each participant. 67 parents or caregivers of ASD patients completed this assessment. Sample Collection. Participants were previously informed to fast in the morning before sampling to avoid the influence of food and/or drink. DNA samples were obtained by an experienced technician by taking a sterile swab and rubbing it against the inside of participants' cheeks for 1 min and placing it into a labeled sterile 2-ml Falcon tube. This was done for both cheeks.  Statistical Analysis. In both ASD patients and TD controls, Haploview version 4.2 59, 60 was conducted to assess the Hardy-Weinberg equilibrium for genotypic frequency distributions by χ 2 test and the linkage disequilibrium (LD) analysis among SNPs by D′. A value of D′ > 0.5 was defined as a positive linkage and D′ > 0.75 were considered as a strong linkage. Power and Sample Size Calculation (http://biostat.mc.vanderbilt.edu/wiki/ Main/PowerSampleSize) was employed to perform a power analysis. Haplotype counts and frequencies in both ASD and TD groups were evaluated using SHEsis (http://analysis.bio-x.cn/SHEsisMain.htm) 61 , a frequency of < 0.03 in both groups was ignored in the association analysis. The other analyses were conducted by SPSS version 17.0 (IBM Corporation New York, USA) and SAS version 9.3.2 (SAS Institute Inc, Cary, NC). Comparisons of genotypic and allelic frequencies between ASD patients and TD controls were performed with the χ 2 test. The Kolmogorov-Smirnov test was performed to check the distribution of our data. The Levene's test was utilized to determine the homogeneity of variance. Kruskal-Wallis H tests were utilized for comparisons of CARS total scores among genotypes of four SNPs. A one-way analysis of covariance (ANCOVA) with sex, age, and IQ as covariates was applied for comparisons of ABC scores among genotypes of four SNPs. While adjusting for sex, age, and IQ, an ordinal polytomous logistic regression analysis was performed to discover the associations between ratings in items of CARS as well as ECDQ and genotypes of four SNPs, and a binary logistic regression analysis was utilized to find the associations between ratings in items of ABC and genotypes of four SNPs in ASD patients. Odds ratios (ORs) were expressed with 95% confidence intervals (95% CI). Post-hoc Bonferroni correction for multiple comparisons was performed. A p value of < 0.05 was described as significant.
Data Availability. The datasets generated and analyzed during the current study are not publicly available due to the public availability would violate the privacy of participants, especially for the special individuals with ASD. Data are from the Department of Maternal, Child and Adolescent Health, School of Public Health, Tianjin Medical University and are available from the corresponding author (email: zhangxin@tmu.edu.cn) on a reasonable request.