Firstly, the mutation screening research was conducted on 192 children who were diagnosed as autism and 192 heathy controls subjects called as the primary sample set (the detailed information see Table 3). Then, we further expanded the sample size called as the second set (comprising the primary sample set, 192 cases and 192 controls) (the detailed information see Table 3) for an association study. The patients were fulfilling the criteria for the diagnosis of autism according to the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). They were recruited from the Shanghai Mental Health Center, Shanghai Jiao Tong University School of Medicine. Patients were excluded if they had known mental and physical illness or chromosomal abnormalities. The healthy controls subjects were volunteers recruited by advertisement from Shanghai Mental Health Center and local community, an brief unstructured interview were evaluated by senior psychiatrists to exclude individuals who suffer from any severe physical diseases, as well as personal and family history of mental diseases. All subjects are Chinese Han descendant. Ethical approval was obtained from the ethics committees of Shanghai Mental Health Center and an informed written consent of participation in the study was signed by the parents or the legal guardians of the studied subjects.
Five milliliters of the whole blood was taken and ethylenediaminetetraacetic acid (EDTA) was used to be anticoagulant. DNA was isolated according to the established laboratory protocols.
Sequencing data were aligned to the annotated human genome sequence (hg19) using the BLAT tool of the UCSC Genome Browser (http://genome.ucsc.edu/). The UBE3A gene sequence was searched from the UCSC Genome Browser database, which showed that there were three spliceosomes (NM_130838, NM_000462 and NM_130839) of UBE3A gene. The sequence information of spliceosome 2 includes spliceosome 1 (NM_130838) and 3 (NM_130839). Therefore, the sequence information of spliceosome 2 (NM_000462) was selected as the candidate sequence. UBE3A gene has 14 exons (Fig. 1), exon 1, 2, 3, initiation 36 bp of exon 4 and the terminal 1888 bp of exon 14 are non-coding regions (untranslated regions, UTR), the terminal 29 bp of the exon 4, exon 5, 6, 7, 8, 9, 10, 11, 12, 13 and initiation 121 bp of exon 14 are coding sequence (CDS) (Fig. 1). In this study, we performed mutation screening for all coding regions and their adjacent non-coding regions named by CDS1-CDS11, the length of CDS1-CDS11 are 29bp, 42bp, 299bp, 1247bp, 145bp, 206bp, 165bp, 156bp, 74bp, 144bp, 121bp, respectively. The total sequenced length after amplifying was 6273bp (see Table1,2).
The two methods high resolution melting (HRM) and Sanger sequencing were performed in our study. The HRM method was used to detect CDS 3, 8, 9, 10 and 11, and the CDS 1,2,4,5,6,7 were directly sequenced by Sanger method according to the amplicon length of every CDS after a series of preliminary experiments. The found variants were further detected in the expanded samples using the HRM method and performed follow-up analyses.
The CDS 3, 8, 9, 10 and 11 were amplified using polymerase chain reaction (PCR) and the primers information summarized in Table 1. The primers were designed to generate amplicons of 200–400bp. Then HRM were performed using LightCycler® 96 Instrument (Roche Diagnostics, Roche Instrument Center AG, Rotkreuz, Switzerland). The amplifications test were performed in 10 μL volumes containing 10 ng of genomic DNA, 2 μmol/L primers, 2.5 mmol/L MgCl2 and 5 μL 2X LightCycler® 96 High Resolution Melting Master (Roche Diagnostics) buffer. PCR cycling included an initial preincubation at 95℃ for 10 min, followed by 45 cycles of 10s at 95 ℃, from 65℃ to a “touchdown” at 55℃, and 30s at 72℃. The melting program included three steps: denaturation at 95℃ for 10s, renaturation at 65℃ for 1 min, and a subsequent melting cycle consists of a continuous fluorescent reading from 60℃ to 90℃ at a rate of 25 acquisitions per ℃. The detected mutations were confirmed with an independent PCR and bidirectional sequencing.
CDS 1, 2, 4, 5, 6 and 7 were directly sequenced using Sanger method, which could detect about 800bp sequence length. The length of CDS4 was 1247bp, which was sequenced forward and backward. The PCR primers information were listed in Table 2. Each PCR master mix included 2 μM of each primer, 1 μl mix dNTP, 1.5 μl MgCl2, 5 μl of 10× PCR buffer, (5 U/μl) AmpliTaq Gold DNA polymerase, 3 μl of DNA sample, and double-distilled water to reach the total volume of 50 μl. PCR condition contained initial denaturation at 95 °C for 5min, 30 cycles of 95 °C for 30 s and 62°C-65 °C for 30 s and 72 °C for 45 s, then 72 °C for 5 min as final extension was performed. After purification of PCR products using TaKaRa® Shrimp Alkaline Phosphatase (SAP) and ExonucleaseI (ExonI), sequencing was performed by BigDye Terminator Kit (Applied Biosystems, USA) according to the method by Sanger F, et al . The sequence information were read using Applied Biosystems (ABI) 3130 Genetic Analyzer (Applied Biosystems, USA). As well, the detected mutations were confirmed with an independent PCR and bidirectional sequencing.
For each detected variant, we further expanded the sample size of case and control groups, the case-control genotyping studies and association analysis were performed.
The statistic power was calculated using QUANTO software (version 1.2.4.) for allele of the detected variants. The parameters were set to the prevalence of autism to be 1% and a moderate odds ratio (OR) to be 1.3-1.5, the inheritance model was set as log-additive and a type I error rate of 0.05 (two-sided).
Each detected variant were genotyped using HRM method in the expanding samples.
For the association analysis, the online software SHEsis  (http://analysis2.bio-x.cn/myAnalysis.php) was used to compare the allelic and genotypic frequencies between the case and control groups. Another online software SNPStats  (http://bioinfo.iconcologia.net/snpstats/start.htm) was used to calculate the association between the detected variants and the risk of autism under 5 inheritance models, including codominant, dominant, recessive, overdominant and log-additive models. All of the statistical tests were two-sided, and p<0.05 was defined as statistically significant.