Copy number variation in Han Chinese individuals with autism spectrum disorder

ASD-affected individuals and their families were referred to the Children Development and Behavior Research Center (CDBRC) at Harbin Medical University, China, by their community physician between January 2007 and June 2011. Proband diagnosis and study inclusion criteria were completed as previously described [8]. The Autism Behavior Checklist (ABC) and Childhood Autism Rating Scale (CARS) were used for diagnosis. We report on 104 consecutive cases with an ASD diagnosis made by two psychiatrists at the CDBRC. Subsequently, DNA was also obtained from the parents of these ASD individuals. Proband participants consisted of 91 males (87.5%) and 13 females (12.5%). The mean age of the probands at enrolment was 4.31 ± 1.80 years. The study was approved by the Ethics Committee at Harbin Medical University and written consent was obtained from parents.

Genotyping was performed on the Affymetrix CytoScan HD platform (Santa Clara, CA, USA) according to the manufacturer's specifications. PLINK software was used to confirm the Han Chinese ethnicity of all individuals in the study from extracted SNP genotypes (Additional file 1: Figure S1). Samples from 100 probands and 200 parents passed our quality control metrics including 93 complete trios. CNVs were called via our CNV detection pipeline (Additional file 1: Figure S2). Population control datasets used to distinguish rare variants included three Han Chinese-specific cohorts and one microarray-specific set of primarily European individuals (see Additional file 1). For all case and control samples, genotyping and CNV calling were performed using identical procedures [10] and CNV data was compared against the Database of Genomic Variants [11]. Validation methods confirmed 96% (27/28) of the CNVs tested.

The SuperScript III First-Strand Synthesis SuperMix for qRT-PCR kit (Life Technologies, Carlsbad, CA, USA) was used to generate cDNA from RNA extracted from 11 tissues and a whole brain sample. Expression analysis and tissue distribution of CASKIN1 and PKD1 were illustrated using quantitative RT-PCR. The housekeeping genes MED13 and ACTB were used to normalize the expression level.

Nine de novo CNV events were detected in eight probands (Table 1). This represents a de novo CNV rate of 8.6% (8/93), similar to that which has been seen previously in other non-Chinese CNV studies [2‐4, 12]. In one female proband (683-3), we discovered a 22-kb deletion at 2q37.1 overlapping GIGYF2, which lies in a susceptibility locus for familial Parkinson's disease [13]. No ASD phenotype has been previously associated with variants affecting this gene.

A 982-kb de novo duplication was uncovered in male proband 527-3. This duplication overlaps SPRY1, a gene that regulates fibroblast growth factor signaling which plays an important role in the patterning and propagation of cells in the developing brain [14]. There is no prior evidence of any link between duplications overlapping this gene or SPATA5 and ASD.

A pair of adjacent de novo duplications at 16p13.3 separated by nearly 1 Mb of two-copy intervening sequence was uncovered in a 6-year-old female (517-3). To better assess which genes might be contributing to the phenotype, we referred to a recent finding showing that highly brain-expressed exons have a lower burden of rare missense variants than more ubiquitously expressed exons and are targets for penetrant mutations in ASD [15]. We then checked each of the exons in the genes residing within the duplications to determine if any of these were characterized as a ‘brain-critical exon.’ This was the case for at least one exon in RAB26, PKD1, E4F1, ABCA3, and CASKIN1 (Additional file 1: Figure S3). We focused on CASKIN1 and PKD1 due to previous studies indicating that they may play some role in synaptic scaffolding and neurodevelopment [16, 17]. In these two genes, we confirmed the presence of a brain-expressed isoform whose dosage could be impacted by the duplication (Additional file 1: Figure S3). The 16p13.3 duplication affects a different region of that chromosomal band than was recently described to be involved in obsessive-compulsive disorder [18].

A male proband (503-3) was found to carry a 232-kb de novo microdeletion at 16p11.2. This deletion lies upstream of the 600-kb ASD-implicated risk locus [1]. Similarly sized deletions have been previously noted in individuals with obesity and developmental delay [12, 19]. Unlike many of the individuals with similar deletions, this male has a BMI of 18.26, putting him in the normal range.

A 994-kb duplication of 17p13.3-17p13.2 was uncovered in a 5-year-old autistic male (692-3). Microduplications have been previously noted at this locus and are usually associated with developmental delay and frequently with growth issues [20]. This proband has typical autistic features and no evidence of any growth problems.

Two unrelated male probands harbor de novo exonic deletions of the Duchenne Muscular Dystrophy (DMD) gene. The first case (611-3), a 5-year-old autistic male presenting with ASD, hypotonia, and progressive motor impairments including difficulty walking, has a 55-kb deletion that overlaps exons 14–17. The second proband (567-3), an autistic 6-year-old male with abnormal muscular development, has a deletion of 154 kb which overlaps exons 46–50. In both cases, the deletion is predicted to cause a frameshift leading to a premature stop and loss of dystrophin. In males, such mutations are predicted to result in DMD. Studies have shown a higher incidence of ASD in boys with DMD, possibly because of a secondary synaptic role for the protein [21].

We identified a 27.8-kb loss of two exons of GRID2 in a 4-year-old male (694-3). GRID2 encodes a glutamate receptor channel subunit and mutations within GRID2 have been associated with ASD [2]. Our proband inherits the mutation from his mother who has a diagnosis of intellectual disability. His maternal grandfather and a maternal cousin also have intellectual disability, but no DNA was available to test the segregation of this variant in these individuals.

A 73-kb deletion affecting SLC39A12 was found in a 6-year-old male proband (686-3). All but the last exon of the gene was deleted in the proband and his unaffected father. A recent study suggests that this zinc transporter stimulates neurite outgrowth during neurodevelopment [22].

Finally, we identified a male proband (511-3) with a maternally inherited deletion of one exon of LINGO2. The gene has been previously associated with adult-onset neurodegenerative disorders and has been implicated as an ASD risk gene [23]. A second male proband (694-3) in our study also harbors an intronic CNV within this gene, but its potential effect on gene expression and any possible contribution to phenotype were not possible to ascertain.

We have also identified a population-specific CNV polymorphism in YWHAE, a gene previously speculated to have a role in ASD [20]. Here, we identified a 24-kb duplication (chr17: 1,235,975-1,259,833) overlapping the last exon of this gene in one proband and his mother. No microduplications of YWHAE were found in Caucasians in our Ontario population controls. However, we identified similar duplications in two unrelated parents of other probands in this Han Chinese cohort (Figure 1). We attempted fine-mapping of the breakpoints in the four different samples carrying the rearrangement and found that both the 3′ and 5′ ends consistently map to the same regions, suggesting that the CNVs likely represent the same ancestral event. We were unable to determine the precise sites of the breakpoints due to complex sequence elements located in the region. Subsequently, we assessed the frequency of this event in the Chinese population and found that 11/1,235 (0.9%) of the Han Chinese population controls carried a microduplication at this locus. Using a TaqMan Copy Number Assay for a probe located within the breakpoints of the microduplication, we found duplications in 3/260 additional Han Chinese controls. In all, approximately 1% of Han Chinese individuals have this duplication, regardless of ASD status. We notice no statistically significant difference between the frequency of this variant in cases versus controls (p = 1.000 using a two-tailed Fisher's exact test).