Defining the contribution of different classes of de novo mutation to autism. I. Iossifov1, B. J. O'Roak2, S. J. Sanders3,4, N. Krumm5, M. Ronemus1, D. Levy1, J. Shendure5, E. E. Eichler5, M. W. State3,4, M. Wigler1 1) CSHL, Cold Spring Harbor, NY; 2) Molecular & Medical Genetics Department, Oregon Health & Science University, Portland, OR; 3) Department of Psychiatry, University of California, San Francisco, CA; 4) Department of Genetics, Yale School of Medicine, New Haven, CT; 5) Department of Genome Sciences, University of Washington, Seattle, WA.
Autism spectrum disorders (ASD) are the most prevalent form of neurodevelopmental disorders affecting ~1% of the human population, and are characterized by impairments in social interactions and communication as well as restricted interests and repetitive behavior. We have completed the exome sequencing of 2,519 simplex ASD families from the Simons Simplex Collection. For all families we have sequenced the affected child and both parents and, for 1,913 of the families, we have also sequenced an unaffected child.
We observe 392 'Likely Gene-Disrupting' (LGD) de novo mutations, which include nonsense, frame-shift and canonical splice-site mutations that are likely to result in a truncated gene product, and we estimate that 45% of them are contributory to the disease. There are 27 genes that are recurrently hit by de novo LGD mutations, 90% of which are likely to contribute to the disorder. CHD8 is hit by 9 de novo LGDs (in nine affected children), DYRK1A by 4, and each of ANK2, GRIN2B, DSCAM, and CHD2 is hit by 3 de novo LGDs. 147 genes are affected by more than one of the observed 1,688 missense mutations, and we expect that 30% of these genes are contributory. By measuring the recurrence against the 163 LGD mutations predicted to contribute to ASD risk and correcting for the distribution of gene sizes, we estimate a target size for LGD mutations between 400 and 1,200 genes, with a smaller range if we exclude affected males with higher IQ. Recurrence estimates of missense mutations show a similar target size and there is evidence of strong overlap between these two gene target classes. Genes affected by de novo mutation (LGD and missense) overlap strongly with sets of genes encoding transcripts bound by the Fragile X mental retardation protein (FMRP), genes encoding chromatin and transcription modulators, and genes expressed in early embryonic development. We find that affected children with de novo LGD mutations in recurrently hit genes or in FMRP associated genes have significantly lower non-verbal IQ.
We estimate that 9% of simplex autism can be attributed to de novo LGD mutations, and that de novo missense mutations contribute an additional 12%. Together with the established 6% contribution from de novo copy number variants, the total burden from observable de novo variants is nearly 30%. Moreover, de novo variants explain 45% of low IQ simplex autism.
You may contact the first author (during and after the meeting) at