The Digitization of Biotech: AI Designs a Serotonin Sensor

Researchers design new neuroscience tool with machine learning.

Posted Jan 02, 2021 | Reviewed by Gary Drevitch

GDJ/Pixabay
Source: GDJ/Pixabay

Artificial intelligence (AI) machine learning is transformative technology that is accelerating scientific research. The immediacy of the gains is rapidly unfolding in biotechnology, pharmaceutical, and life sciences. In a recent study published in Cell, scientists affiliated with the NIH Brain Research through Advancing Innovative Neurotechnologies® (BRAIN) Initiative created a sensor using machine learning that is capable of detecting real-time changes in the brain serotonin levels, a neuroscience tool to enable research on how the neurotransmitter impacts sleep and mental health.  

Serotonin (also known as 5-hydroxytryptamine or 5-HT) is a neurotransmitter that the body uses to send messages between nerve cells. It is the precursor for melatonin, and therefore plays a role in the body’s sleep cycles. Serotonin is linked to influencing digestion, appetite, emotions, and mood. Low levels of serotonin may play a role in insomnia and depression. Serotonin is used to treat anxiety, migraines, depression, panic disorders, PTSD, obsessive-compulsive disorder, aggressive behavior, social phobia, bulimia nervosa, premenstrual dysphoric disorder, and many other psychiatric and neurological disorders.

The market opportunity for sleep and mental health treatment is increasing, with an anticipated surge of demand triggered by the COVID-19 pandemic. By 2030, the global market for sleeping aids is projected to reach USD 162.5 billion, according to a Research and Markets’ July 2020 report. The worldwide market for anxiety disorder and depression treatment is estimated to reach USD 19.21 billion by 2027, and selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) are among the most frequently prescribed antidepressants, according to July 2020 figures from Reports and Data. To put the market opportunity in context, the entire global medical glove market value by comparison is projected to reach USD 20.4 billion during 2020-2030, according to November 2020 figures from Research and Markets.

Scientists at the University of California Davis School of Medicine led the research under the principal investigator Lin Tian, who leads a lab focused on optical neurophysiology, in collaboration with scientists from Loren Looger’s lab at the Howard Hughes Medical Institute Janelia Research Campus, and Viviana Gradinaru at Caltech.

To design an effective drug for serotonergic circuitry, it is critical to have the capability to observe the serotonin transport and release in the brain in real-time and in high resolution. The research team applied AI machine learning to the task of predicting designs of such a serotonin sensor.

“We developed and applied a binding-pocket redesign strategy, guided by machine learning, to create a high-performance, soluble, fluorescent serotonin sensor (iSeroSnFR), enabling optical detection of millisecond-scale serotonin transients,” the researchers reported. “We demonstrate that iSeroSnFR can be used to detect serotonin release in freely behaving mice during fear conditioning, social interaction, and sleep/wake transitions. We also developed a robust assay of serotonin transporter function and modulation by drugs.”

“Our combined Rosetta and machine-learning-guided directed evolution approach was quite effective,” the researchers wrote. “After just one round of each, we screened fewer than 2,600 variants, but made dramatic improvements to the sensor’s affinity, specificity, and fluorescence response. After just two more rounds of machine-learning-guided directed evolution, we had screened a total of ~16,000 variants, interrogated more than 60 different protein scaffold positions, and introduced 19 mutations into our final sensor, increasing its 5-HT affinity by >5,000-fold, abolishing choline/ACh binding, and increasing fluorescence response by 3-fold compared to the starting scaffold, iAChSnFR0.6.”

With this new proof-of-concept, not only can this approach be used to create other neuromodulator sensors, but also more broadly for complex protein-engineering—a new way to accelerate novel drug discovery and treatments in efforts to help improve the human condition in the future.

Copyright © 2021 Cami Rosso All rights reserved.