Uncovering novel therapeutics for schizophrenia: a multitarget approach using the CANDO platform

Abstract

Schizophrenia is a complex and debilitating neuropsychiatric disorder characterized by positive, negative, and cognitive symptoms, many remaining insufficiently addressed by current treatments. High rates of treatment resistance and heterogeneous pathophysiology pose significant challenges to traditional single-target drug discovery. To address this, we applied the CANDO platform to identify repurposable drugs for schizophrenia using a multitarget strategy. The platform evaluates how compounds interact with the entire human proteome, generating interaction signatures that capture a compounds effect across all targets. By comparing these signatures, CANDO computes compound similarity scores and enables consensus prediction of novel therapeutics. Across all indications and for schizophrenia in particular, CANDO outperformed controls by orders of magnitude across several benchmarking metrics, accurately recovering known drug-indication relationships. We applied our CANDO pipeline, benchmarked by its ability to recover approved drugs for their indications using similarity- and consensus-based metrics, to generate a ranked list of repurposable compounds. Our comprehensive literature review confirmed clinical or biochemical evidence supporting 25 high-corroboration drug candidates, including phenothiazine and benzamide antipsychotics, tricyclic antidepressants, benzodiazepines, and monoamine oxidase inhibitors. We identified protein targets with the highest likelihood of interaction with these top drugs and assessed prediction quality through overlap analysis with gold standards, showing significantly better concordance with established schizophrenia-related biology than controls from bottom-ranked and randomly selected drugs. Among these proteins are corroborated targets such as canonical neurotransmitter receptors of the dopamine, serotonin, and adrenergic classes, as well as monoamine transporters, tyrosine aminotransferase, and lysine-specific histone demethylase 1A. Enrichment of calcium-binding proteins among the top predicted targets for the phenothiazine antipsychotic thiethylperazine highlights a potential role for dysregulated calcium signaling, including calmodulin-CaMKK2 pathways, in schizophrenia pathology and treatment response. Another phenothiazine antipsychotic, triflupromazine, binds the serotonin transporter in addition to its canonical dopamine D2 receptor interaction, highlighting its potential relevance to depressive symptom modulation in schizophrenia. These findings demonstrate the utility and accuracy of the CANDO platform in elucidating multitarget pharmacological mechanisms and accelerating the identification of effective repurposable treatments for schizophrenia.