Graph Theory Identifies Autistic Patterns in the Prefrontal Circuit of a Mouse Model of Autism

As a well-developed branch of mathematics, graph theory provides unique tools to quantifiably assess various properties of complex networks. Applied to brain circuits, network-level analyses can illustrate disruptions to brain organization that yield both mechanistic and diagnostic insights. Previously, graph theory has been used with functional magnetic resonance imaging datasets to quantify connections among different brain regions, readily capturing the macroscopic-scaled differences in brain networks between healthy and Alzheimer's subjects. Here, we applied graph theory on the microscopic scale, using miniscope-based calcium imaging from the freely behaving wild type (WT) and Shank3 ^fx mice (a mouse model of autism), and compared functional connections among individual neurons in the prefrontal microcircuits during social behavior. We demonstrated that Shank3 ^fx mice displayed reduced neural activity, less-integrated network, and fewer network changes in the prefrontal microcircuits between the presence and absence of social targets. Furthermore, we employed machine learning to test whether graph-theoretic metrics extracted from the prefrontal microcircuits could be predictive of genotype and genotype-associated social behavior difference between Shank3 ^fx and WT mice. Our results indicate a strong link between altered prefrontal microcircuits and social behavior deficits in an autism mouse model, highlighting prefrontal microcircuitry as a potential diagnostic and therapeutic target for autism.