Neural Mechanisms Underlying Cognitive Deficits in Rat Models of Autism

Abstract

Autism Spectrum Disorder (ASD) is a developmental disorder which, though previously viewed as rare, is known to impact approximately 1 in every 68 children. ASD is characterized by impairments in cognitive flexibility and inhibitory control, and recent studies have posited a genetic link between mutations in specific genes (known as ASD risk genes) and the occurrence of ASD. However, given the heterogeneity in presentation of ASD, it has been very challenging to quantify which cognitive deficits are associated with which ASD risk genes as well as investigate the underlying neural circuits causing the deficits. In addition, given the emphasis on dysregulation of inhibitory neurons as a mechanism for ASD, recent computational models have predicted that excitation/inhibition (E/I) balance plays a role in the cognitive deficits seen in ASD patients. These models, however, are in disagreement regarding exactly how changes in E/I ratio impact behavior. Thus, in order to bridge these gaps in knowledge, working with my postdoc, I trained ASD rat models with following gene mutations - Fmr1, Nrxn1, Arid1b, Grin2b, and Dyrk1a - on a reversal learning task and a cognitive flexibility task which is comparable to the tests performed on humans. We analyzed the performance of the rats on the respective tasks in order to quantify the cognitive deficits associated with each gene as well as better understand the heterogeneity of presentation, and we computed psychophysical kernels for ASD rats that learned the cognitive flexibility task in order to determine which computational model most accurately accounts for behavior. We found that Fmr1-KO, Nrxn1-KO, Arid1b-KO, and Grin2b-KO ASD rats all showed impairments in task-switching while Dyrk1a-KO ASD rats do not demonstrate this impairment. Moreover, Fmr1-KO and Nrxn1-KO rats illustrated deficits in feature selection but were successful in accumulating evidence. On average, Fmr1-KO and Nrxn1-KO rats performed worse on the cognitive flexibility task than wildtype (WT) rats, but there was still heterogeneity in performance seen across the ASD rats who had the same gene mutation. The psychophysical kernels were computed for the two Fmr1-ASD rats, and these kernels suggested that the rats were engaging in more impulsive decision-making as they were placing more weight on the early evidence, thus supporting the E/I balance computational model posited by Lam et al. Findings, limitations, and suggestions for future research are discussed. The methodology described in this work could be extended by coupling a standardized training protocol with flexible recording probes to characterize neural activity throughout the training of the ASD rats in order to contextualize the timeframe in which cognitive deficits manifest for different ASD risk genes; this will provide a better understanding for when individuals diagnosed with ASD should be treated.