Directed Evolution of a Selective and Sensitive Serotonin Biosensor Via Machine Learning

Abstract

Serotonin is involved in numerous critical physiologic and cognitive processes. Drugs that modulate serotonin transport are among the most widely used medications to treat pervasive mental disorders. Both a deeper understanding of serotoninergic circuitry and design of improved treatments will require the ability to measure serotonin release and transport with high spatial and temporal resolution. Here we developed and applied a binding-pocket redesign strategy guided by machine learning to create a high-performance soluble, fluorescent serotonin sensor (iSeroSnFR). This sensor has requisite selectivity and sensitivity needed for direct detection of serotonin transients in various biological systems, especially permits detection of subsecond-scale serotonin release events during fear conditioning and sleep-to-wakefulness transitions. Furthermore, the solubility of iSeroSnFR allowed us to develop a rapid, cytoplasmic sensor-based assay of serotonin transporter activity in the presence of pharmacological manipulations.