Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Reporter: Adina Hazan, PhD

Elizabeth Unger from the Tian group at UC Davis, Jacob Keller from the Looger lab from HHMI, Michael Altermatt from the Gradinaru group at California Institute of Technology, and colleagues did just this, by redesigned the binding pocket of periplasmic binding proteins (PBPs) using artificial intelligence, such that it became a fluorescent sensor specific for serotonin. Not only this, the group showed that it could express and use this molecule to detect serotonin on the cell, tissue, and whole animal level.

By starting with a microbial PBP and early version of an acetyl choline sensor (iAChSnFR), the scientists used machine learning and modeling to redesign the binding site to exhibit a higher affinity and specificity to serotonin. After three repeats of mutagenesis, modeling, and library readouts, they produced iSeroSnFR. This version harbors 19 mutations compared to iAChSnFR0.6 and a K_d of 310 µM. This results in an increase in fluorescence in HEK293T cells expressing the serotonin receptor of 800%. Of over 40 neurotransmitters, amino acids, and small molecules screened, only two endogenous molecules evoked some fluorescence, but at significantly higher concentrations.

To acutely test the ability of the sensor to detect rapid changes of serotonin in the environment, the researchers used caged serotonin, a technique in which the serotonin is rapidly released into the environment with light pulses, and showed that iSeroSnFR accurately and robustly produced a signal with each flash of light. With this tool, it was then possible to move to ex-vivo mouse brain slices and detect endogenous serotonin release patterns across the brain. Three weeks after targeted injection of iSeroSnFR to specifically deliver the receptor into the prefrontal cortex and dorsal striatum, strong fluorescent signal could be detected during perfusion of serotonin or electrical stimulation.

Most significantly, this molecule was also shown to be detected in freely moving mice, a tool which could offer critical insight into the acute role of serotonin regulation during important functions such as mood and alertness. Through optical fiber placements in the basolateral amygdala and prefrontal cortex, the team measured dynamic and real-time changes in serotonin release in fear-trained mice, social interactions, and sleep wake cycles. For example, while both areas of the brain have been established as relevant to the fear response, they reliably tracked that the PFC response was immediate, while the BSA displayed a delayed response. This additional temporal resolution of neuromodulation may have important implications in neurotransmitter pharmacology of the central nervous system.

This study provided the scientific community with several insights and tools. The serotonin sensor itself will be a critical tool in the study of the central nervous system and possibly beyond. Additionally, an AI approach to mutagenesis in order to redesign a binding pocket of a receptor opens new avenues to the development of pharmacological tools and may lead to many new designs in therapeutics and research.

SOURCE: