An Autism-Associated Signaling Network Differentiates Glutamate Receptor Inputs at Cortical Glutamatergic Synapses

Background: The genomic revolution has revealed hundreds of genetic risk alleles for autism. Clustering these genes into functional groups using a variety of methods- gene ontology terms, mRNA co-expression networks, protein-protein interaction databases- has revealed that many risk genes are expressed at the neural synapse. Other genes participate in coupling synaptic activity to downstream homeostatic responses. Together, these genetic data suggest that ASD-linked genes constitute a molecular pathway that translates synaptic input into homeostatic responses.

Objectives: To test the hypothesis that ASD-linked genes are involved in the translation of extracellular inputs into cellular responses.

Methods: We used Quantitative Multiplex co-Immunoprecipitation (QMI), a flow-cytometry-based method that allows the simultaneous monitoring of hundreds of dynamic protein-protein interactions. We developed an experimental system to monitor the interactions among a network of synaptic proteins that have been genetically linked to autism, including NMDA receptors, mGluRs, Shanks, SynGAP, PSD95, and others. We stimulate cultured neurons with various excitatory chemicals (glutamate, high potassium, NMDA, DHPG, ect), and monitor the acute changes in our selected protein interaction network.

Results: We find acute, dynamic rearrangement of the targeted protein interaction network in response to activity-inducing stimuli. The majority of changes were dissociative: previously described reductions in co-associated Homer-mGluR5 were detected, as well as many previously undescribed changes largely centered on the scaffolding molecule Homer. Using weighted correlation analysis, we identified two correlated modules of co-regulated interactions. Using agonist and antagonist stimulation, we found that one of the modules responded to NMDA stimulation, while the other responded to mGluR stimulation. Glutamate stimulation resulted in simultaneous activation of both modules.

Conclusions: ASD-linked genes form a network of co-associated proteins at the synapse that responds to synaptic activity by changing its pattern of interactions. Information about the nature of the synaptic input seems to be encoded in the protein interaction network, with a set of interactions responding to mGluR stimulation and another set responding to NMDA. We hypothesize that many different autism risk alleles could convergently disrupt network-scale translation of synaptic input into homeostatic cellular responses, which has implications for LTP vs. LTD coordination, E/I imbalance and sensory hyperactivity phenotypes.