Optimizing the phenotyping of rodent ASD models: enrichment analysis of mouse and human neurobiological phenotypes associated with high-risk autism genes identifies morphological, electrophysiological, neurological, and behavioral features

Background: There is interest in defining mouse neurobiological phenotypes useful for studying autism spectrum disorders (ASD) in both forward and reverse genetic approaches. A recurrent focus has been on high-order behavioral analyses, including learning and memory paradigms and social paradigms. However, well-studied mouse models, including for example Fmr1 knockout mice, do not show dramatic deficits in such high-order phenotypes, raising a question as to what constitutes useful phenotypes in ASD models. Methods: To address this, we made use of a list of 112 disease genes etiologically involved in ASD to survey, on a large scale and with unbiased methods as well as expert review, phenotypes associated with a targeted disruption of these genes in mice, using the Mammalian Phenotype Ontology database. In addition, we compared the results with similar analyses for human phenotypes. Findings: We observed four classes of neurobiological phenotypes associated with disruption of a large proportion of ASD genes, including: (1) Changes in brain and neuronal morphology; (2) electrophysiological changes; (3) neurological changes; and (4) higher-order behavioral changes. Alterations in brain and neuronal morphology represent quantitative measures that can be more widely adopted in models of ASD to understand cellular and network changes. Interestingly, the electrophysiological changes differed across different genes, indicating that excitation/inhibition imbalance hypotheses for ASD would either have to be so non-specific as to be not falsifiable, or, if specific, would not be supported by the data. Finally, it was significant that in analyses of both mouse and human databases, many of the behavioral alterations were neurological changes, encompassing sensory alterations, motor abnormalities, and seizures, as opposed to higher-order behavioral changes in learning and memory and social behavior paradigms. Conclusions: The results indicated that mutations in ASD genes result in defined groups of changes in mouse models and support a broad neurobiological approach to phenotyping rodent models for ASD, with a focus on biochemistry and molecular biology, brain and neuronal morphology, and electrophysiology, as well as both neurological and additional behavioral analyses. Analysis of human phenotypes associated with these genes reinforced these conclusions, supporting face validity for these approaches to phenotyping of ASD models. Such phenotyping is consistent with the successes in Fmr1 knockout mice, in which morphological changes recapitulated human findings and electrophysiological deficits resulted in molecular insights that have since led to clinical trials. We propose both broad domains and, based on expert review of more than 50 publications in each of the four neurobiological domains, specific tests to be applied to rodent models of ASD.