Copy number variations (CNVs) account for a substantial proportion of human genomic variation, and have been shown to cause neurodevelopmental disorders. We sought to determine the relevance of CNVs to the aetiology of schizophrenia (SZ). Whole-genome, high-resolution, tiling path BAC array comparative genomic hybridization (array CGH) was employed to test DNA from 93 individuals with DSM-IV SZ. Common DNA copy number changes that are unlikely to be directly pathogenic in SZ were filtered out by comparison to a reference dataset of 372 control individuals analyzed in our laboratory, and a screen against the Database of Genomic Variants. The remaining aberrations were validated with Affymetrix 250K SNP arrays or 244K Agilent oligo-arrays and tested for inheritance from the parents. A total of 13 aberrations satisfied our criteria. Two of them are very likely to be pathogenic. The first one is a deletion at 2p16.3 that was present in an affected sibling and disrupts NRXN1. The second one is a de novo duplication at 15q13.1 spanning APBA2. The proteins of these two genes interact directly and play a role in synaptic development and function. Both genes have been affected by CNVs in patients with autism and mental retardation, but neither has been previously implicated in SZ.

Autism and mental retardation (MR) are often associated, suggesting that these conditions are etiologically related. Recently, array-based comparative genomic hybridization (array CGH) has identified submicroscopic deletions and duplications as a common cause of MR, prompting us to search for such genomic imbalances in autism. Here we describe a 1.5-Mb duplication on chromosome 16p13.1 that was found by high-resolution array CGH in four severe autistic male patients from three unrelated families. The same duplication was identified in several variably affected and unaffected relatives. A deletion of the same interval was detected in three unrelated patients with MR and other clinical abnormalities. In one patient we revealed a further rearrangement of the 16p13 imbalance that was not present in his unaffected mother. Duplications and deletions of this 1.5-Mb interval have not been described as copy number variants in the Database of Genomic Variants and have not been identified in >600 individuals from other cohorts examined by high-resolution array CGH in our laboratory. Thus we conclude that these aberrations represent recurrent genomic imbalances which predispose to autism and/or MR.

Our data show that submicroscopic deletions and duplications play an important role in the aetiology of this condition, either as direct causes or as genetic risk factors for CHD. These findings have immediate consequences for genetic counselling and should pave the way for the elucidation of the pathogenetic mechanisms underlying CHD.

We report on three unrelated mentally disabled patients, each carrying a de novo balanced translocation that truncates the autism susceptibility candidate 2 (AUTS2) gene at 7q11.2. One of our patients shows relatively mild mental retardation; the other two display more profound disorders. One patient is also physically disabled, exhibiting urogenital and limb malformations in addition to severe mental retardation. The function of AUTS2 is presently unknown, but it has been shown to be disrupted in monozygotic twins with autism and mental retardation, both carrying a translocation t(7;20)(q11.2;p11.2) (de la Barra et al. in Rev Chil Pediatr 57:549-554, 1986; Sultana et al. in Genomics 80:129-134, 2002). Given the overlap of this autism/mental retardation (MR) phenotype and the MR-associated disorders in our patients, together with the fact that mapping of the additional autosomal breakpoints involved did not disclose obvious candidate disease genes, we ascertain with this study that AUTS2 mutations are clearly linked to autosomal dominant mental retardation.