(a) A t(3;18)(q13.32;q21.2) translocation was sequenced using our custom jumping library protocol in monozygotic twin boys with multiple developmental abnormalities, directly disrupting TCF4 in intron 8, a gene previously associated with Pitt-Hopkins syndrome (PHS) by mutation analysis and with schizophrenia by GWAS. (b) Analysis of CNVs revealed disruption of coding and noncoding exons in 14 independent NDD among 19,556 NDD cases compared to no CNVs observed in 13,991 controls (p = 0.0006), implicating hemizygosity of TCF4 as a highly penetrant locus in NDD, PHS, and ASD (see also Table S2 and Rosenfeld et al., 2009b). (c) Reduced mRNA expression of TCF4 was detected in both cases from EBV-transformed lymphoblastoid cell lines compared to two gender-matched controls using two independent primers spanning exons 16–17 and exons 18–19.

Gene-specific view of sequenced BCAs and CNVs implicating individual genes involved in transcriptional or epigenetic regulation and localized to previously described microdeletion syndromes, each of which implicates a single gene by the combined BCA and CNV analyses. Haploinsufficiency of mRNA was confirmed for each gene (Figure S1). A breakpoint disrupting a two gene locus (SNURF and SNRPN) was detected in a fourth microdeletion syndrome, 15q11-13 (not shown). Red bars indicate copy loss, blue bars are copy gains, and striped bars are gene-specific alterations. Green bars represent “Other” alterations (balanced or unbalanced translocation, inversion, or complex alteration involving both deletion and duplication). The bar at the top provides the count of gains and losses that extend beyond the window and is color coded to reflect the relative proportion of each variant type. Confidence intervals provide the distance to the next probe without a detected dosage imbalance. A single RefSeq transcript is provided. Breakpoint sequence from each derivative is provided with text color coded by originating chromosome or non-templated inserted sequence in gray. A subset of the SATB2 patients included here were previously reported by Rosenfeld et al. (Rosenfeld et al., 2009), and all cases reported here for MBD5, as well as 44 additional structural variants, are reported in a follow-up consortium study (Talkowski et al., 2011b).

A large network of genes disrupted by BCAs in this study are connected by first, second, or higher order interactions. No networks were significantly enriched for genes disrupted by BCAs after correction for multiple comparisons, though a number of loci have limited functional annotation available or remain of unknown function.

All genes disrupted by a BCA and analyzed in the CNV analyses are shown. While all genes are implicated in autism spectrum disorder (ASD) or other neurodevelopmental disorder (NDD) by BCA disruption in this study, some loci also represented single gene contributors to previously recognized genomic disorder (GD) regions (three microdeletion syndromes, two terminal deletion syndromes, and one duplication syndrome), and genes discovered in ASD or NDD in this study that had been previously linked to adolescent or adult onset neuropsychiatric disorders (NPD) by common variation association studies. Genes in italics are novel loci implicated for the first time in ASD or NDD by this study (Category 3), bold italics represents novel loci not previously implicated in any ASD, NDD, or NPD.

Provided is an example of genes illustrated by this study that appear to be sensitive to a spectrum of mutational mechanisms reciprocally altering their gene dosage. In Figures 1 and 2 are examples of genes such as TCF4 and SATB2 which are disrupted exclusively by deletions. (a) Genes sensitive to dosage dysregulation from which both deletion and duplication appear to confer risk for similar phenotypes (AUTS2, CDKL5, C18orf1). (b) Three loci, FOXP1, GRIN2B, and CHD8, are recognized for duplication of the locus as the predominant mutational mechanism, yet all three loci are implicated by heterozygous inactivation from BCA disruption in this study. Two of these genes, FOXP1 and GRIN2B are also supported by de novo mutation in ASD families (O’Roak et al., 2011). CHD8 is a completely novel candidate not previously implicated in ASD or any human abnormality. These data suggest annotation of the human genome is insufficient to reliably predict the impact of various mutational mechanisms of specific loci involved in neurodevelopment.