Schizophrenia (SCZ) is a complex psychiatric disorder with unclear biological mechanisms. Spectrins, cytoskeletal proteins linked to neurodevelopmental disorders, are regulated by the AKT/GSK3 pathway, which is implicated in SCZ. However, the impact of SCZ-related dysregulation of this pathway on spectrin expression and distribution remains unexplored. Here, we show that βIV spectrin protein levels were reduced in neurons of the dorsolateral prefrontal cortex in SCZ postmortem samples compared to healthy control (HC) from the Human Brain Collection Core (HBCC). To investigate potential links between βIV spectrin and the AKT/GSK3 pathway, we analyzed the PsychEncode dataset, revealing elevated SPTBN4 and AKT2 mRNA levels with correlated gene transcription in both HCs and individuals with SCZ. Next, computational tools were employed to identify potential AKT and GSK3 phosphorylation sites on βIV spectrin, and two GSK3 sites were validated through in vitro assays. To assess whether βIV spectrin distribution and sensitivity to AKT/GSK3 are altered in SCZ, we used iPSC-derived neurons from two independent cohorts of patients with significantly increased familial genetic risk for the disorder. Alteration in βIV spectrin levels and sensitivity to AKT/GSK3 inhibitors were consistently observed across both cohorts. Importantly, a Random Forest classifier applied to βIV spectrin imaging achieved up to 98% accuracy in classifying cells by diagnosis in postmortem samples, and by diagnosis or diagnosis × perturbation in iPSC samples. These findings reveal altered βIV spectrin levels and AKT/GSK3 sensitivity in SCZ, identifying βIV spectrin image-based endophenotypes as robust, generalizable predictive biomarkers of SCZ, with the potential for scalable clinical applications.
Read full abstract