Abstract
The neurodevelopmental period is highly sensitive; insults during this period impair neural network connectivity, causing lasting cognitive deficits associated with many neuropsychiatric disorders. Medial prefrontal cortex (mPFC) networks subserve flexible behavior, but the mechanisms underlying their disruption after developmental insults remain unclear. We used an early-life seizure (ELS) model to investigate how mPFC networks become impaired and tested whether adrenocorticotropic hormone (ACTH), a clinically relevant neuroprotective peptide, could restore network function. Using in-vivo single-unit recordings during baseline and fear extinction learning, we found ELS-induced dysfunction was characterized by reduced neuronal firing, rigid spike-timing, and weakened functional connectivity, all predicting impaired extinction learning. ACTH treatment prevented these deficits, preserving dynamic spike-timing, flexible connectivity, and network organization. Advanced graph neural network modeling identified neuronal features predictive of cognitive outcomes, revealing potential biomarkers broadly relevant to other developmental disorders. These findings highlight fundamental mechanisms of mPFC network dysfunction and emphasize the translational potential of targeting network dynamics to restore cognition in neurodevelopmental disorders.