Rational pharmacological treatment of autism spectrum disorders can only occur when the genes and the molecular pathways disrupted in this disease are well-understood. Today, three papers in Nature by Matthew State and colleagues, Evan Eichler and colleagues and Mark Daly and colleagues report the largest exome sequencing efforts in autism to date, involving nearly 600 trios and 935 further cases with the disease. Altogether, the papers provide strong evidence for three new autism genes (CHD8, KATNAL2 and SCN2A) and support the idea that autism is an extremely heterogeneous disease, meaning that many genes can confer high-risk for the disease. Previously, scientists estimated that the number of high-risk autism genes was ~200. However, the new data suggests that there are likely ~1000 high-risk autism genes, which can and should be identified with further sequencing.
Last week, the CDC announced new data that shows that autism spectrum disorders currently affect 1 in 88 children in the United States. While this complex set of debilitating developmental diseases are thought to have a significant genetic component, the genetic causes are mostly unknown. Currently, treatment for the disorder relies on intensive behavioral therapy.
However, there are efforts to develop drugs for autism, particularly syndromic disorders in which the genetic etiology is well-defined. A clinical trial for IGF-1 in 22q13 Deletion Syndrome (Phelan-McDermid Syndrome), which is caused by deficiency at SHANK3 just recently started at Mt. Sinai School of Medicine. (In 2006, this journal published the first report of SHANK3 mutations in autism. This journal subsequently published the first mutations in SHANK2 in autism in 2010.) In addition, Seaside Therapeutics has four different clinical trials for STX209 (arbaclofen) in Fragile X syndrome as well as in children with autism spectrum disorders.
However, the best hope for developing new drugs for autism will be realized when the genetic basis of the disease is better understood. To that end, efforts such as the Autism Sequencing Consortium and the Simons Simplex Collection are planning to sequence the exomes and/or genomes of thousands of individuals with autism within the next few years.
Last year, this journal published the first autism exome trios, mostly from the Simons Simplex Collection. In that study, Evan Eichler and colleagues specifically chose to sequence the exomes of 20 families with sporadic autism (i.e. no other familial history), reasoning that these families would be more likely to harbor de novo mutations of large effect. That study identified possibly causative de novo mutations in 4 out of the 20 families in the genes FOXP1, GRIN2B, SCN1A and LAMC3. The study showed that trio-based exome sequencing might be an effective approach for identifying new autism genes.
Today, three papers deliver on that approach and report the largest set of autism exomes published to date.
Matthew State and colleagues sequenced whole exomes of 238 families in the Simons Simplex Collection (SSC). These pedigrees consist of 2 unaffected parents, an affected individual and in 200 families, an unaffected sibling. Within these 200 quartets, the authors identified significantly more non-synonymous de novo mutations in autistic individuals compared to their unaffected siblings (125 vs. 87 respectively). The authors hypothesized that observing de novo mutations in the same gene in multiple, unrelated individuals would be a statistically robust approach to identifying high-risk autism genes. Using simulations and modeling, the authors determined that two or more nonsense and/or splice site de novo mutations would be highly unlikely to occur by chance. Only a single gene in the entire cohort met these criteria; two autistic individuals were found to carry nonsense mutations in SCN2A. The authors then performed an analysis with the sequencing data generated by Evan Eichler and colleagues. In these 414 individuals with autism, the authors found two other genes that met this strict set of criteria (KATNAL2 and CHD8).
Evan Eichler and colleagues sequenced the exomes of 189 trios and 50 unaffected siblings. This study finds that the majority of de novo mutations are of paternal origin (4:1 bias) and that this is positively correlated with paternal age. Altogether, the study validated ~120 severely disruptive de novo mutations, 39% of which occur on a highly interconnected b-catenin/chromatin remodeling protein network. The authors also identified recurrent disruptive mutations in NTNG1 and CHD8. The authors note that this data is consistent with a multi-genic model of autism “where both de novo and extremely rare inherited mutations contribute” to the overall genetic component of the disease in any one affected individual.
Mark Daly and colleagues sequenced the exomes of 175 ASD trios and found that about half of the ASD cases harbored a de novo missense or nonsense variant. Consistent with the other two papers, genetic modeling suggests that most of these de novo events are not related to ASD. The authors note that the de novo events that do confer a high-risk for autism (i.e. those that are nonsense or occur at splice-sites in two or more unrelated individuals) likely occur in several hundred (most of which are, as yet, unidentified) genes and are incompletely penetrant. In combination with data from the two other papers, the authors conclude that the de novo nonsense, splice and frameshift mutations that occur in SCN2A, KATNAL2 and CHD8 are unlikely to occur by chance, and are therefore strong evidence that they are bona fide autism genes. Daly and colleagues further evaluated these three genes in 935 autism exomes and 870 controls. They found 3 additional loss-of-function mutations in CHD8 and KATNAL2 as well as a splice site de novo mutation in SCN2A. The authors conclude that despite the significant challenges to conclusively identifying genes that confer high-risk for autism, analysis of larger datasets and deeper integration with inherited variants should pave the way to a more complete genetic understanding of this debilitating set of neurological diseases.
Finally, the involvement of CHD8, an ATP-dependent chromatin remodeling factor is intriguing. In the current issue of this journal, there are several papers that report mutations in the SWI/SNF complex and ARID1B in Coffin-Siris syndrome and mutations in SMARCA2 in Nicolaides-Baraitser syndrome. These three papers identify germline mutations in chromatin remodeling factors in syndromes that include intellectual disability (some with marked language impairment) and epilepsy. The molecular links between chromatin remodeling and these phenotypes, which overlap features of autism spectrum disorders, remain to be established.