Neurons derived from induced pluripotent stem cells (iPSCs) from individuals with autism offer the singular avenue to directly test novel therapeutic compounds for efficacy in treating autism spectrum disorder (ASD). A major question, however, regards the translatability of findings from in vitro neuronal culture models derived from iPSCs to clinically relevant cognitive, social, and/or motor skill impairments experienced by ASD patients. The synaptic molecular system is one of the most likely loci for the etiology of distinct ASD-associated genes. The complexity, heterogeneity and compartmentalization of this system requires measurements that integrate synaptic signaling activity and localization and synaptic translation of different proteins, simultaneously, at single-synapse resolution, to fully characterize the synaptic effects of these genes.

The Mark Bathe lab has recently developed Multimodal PRISM, a high-throughput method that outputs a high-dimensional distribution of millions of synapses across treatment groups with information on 10–20 protein levels, local mRNA translation of select genes, and glutamate spiking or calcium flux for each synapse. Using the Multimodal PRISM Bathe was able to extract potentially relevant autism-associated phenotypes in the structure of the network of causal synaptic composition-activity relationships, not accessible with bulk or single-variable measurements. In this pilot study, Bathe will apply Multimodal PRISM to iPSC-derived neurons with targeted CRISPRi knockdown of four autism-associated genes and in neurons derived from ASD patients carrying loss-of-function mutations in the same genes. Using these cells, they will perform deep, high-dimensional profiling of synapse populations and in parallel they will validate causal relationships between genetic variation and synaptic phenotypes. The overarching goal is to identify and delineate a disease-predictive in vitro model of ASD based on high-dimensional changes in the synaptic molecular network that is (1) robust and validated in patient cell lines and across autism-associated genotypes and (2) can be used to profile therapeutics in vitro as upstream identification of potential efficacy in restoring wildtype neuronal function. Bathe and colleagues will search for overall and conditional changes to measured synaptic variables and dependency network structure that are similar across the four tested genes, test for the appearance of previously identified convergent network phenotypes, and test binary classifiers of synapse distributions into case/control or treated/untreated labels for generalizability of training across genes.