Autism-Linked Gene Products Form an Activity-Dependent Signaling Network at the Synapse

Background: Among the hundreds of genes that contribute to autism risk, there is significant enrichment of genes expressed at the synapse. These genes include neurotransmitter receptors, scaffolds, and signal transduction molecules that bind to each other, often using protein-protein interaction motifs that can be modified by activity-inducible kinase activty. However, it is not clear how or if the protein products of these genes act together to perform a coordinated cellular function.

Objectives: To define activity-dependent changes among a protein interaction network composed of autism-linked, synaptic gene products.

Methods: We used quantitative multiplex immunoprecipitation and computational modeling of dynamic protein-protein interaction networks to define a "protein interaction network signature" associated with various experiemental manipulations, including neuronal stimulation (with KCl, glutamate, or specific agonists) or autism-linked genetic insults.

Results: We developed a synaptic multiplex assay based on the protein products of autism candidate genes, measuring ~400 binary interactions. Genes were selected based on genetic linkage with autism, synaptic localization, known interactions with other autism risk factors, and antibody availablilty. Using this technology, we first stimulated wild-type neurons with KCl or glutamate, and found activity-dependent changes in 20+ protein-protein interactions (PPIs) among this network of autism-linked proteins. We then performed a screen of 7 different mouse models of autism, and found altered protein-protein interactions in each ASD model (compared to wild-type littermates). Some altered PPIs were shared among models, some were unique to certain models. We found significant enrichment of activity-dependent PPIs among those that were identified in the ASD models. Focusing on 3 ASD models (Shank3, FMR1, & Ube3a overexpressing mice) we find genotype-specific differences in the PPI network response to activity. Principle component analysis shows seperation of experimental groups by both genotype and stimulation status, while our statistical analysis identified specific interactions driving PCA seperation. Finally, we treated FMR1 neurons with mGluR antagonists known to correct morphological and behavioral deficits, and observed corresponding changes in the PPI signiture, suggesting a mechanism of drug action.

Conclusions: The protein products of autism-linked genes form a dynamic, activity-dependent protein interaction network at the glutamate synapse. Disruption of information flow through a protein interaction network comprised of synaptic ASD-linked genes may contribute to autism pathogenesis.