New drugs to treat circadian rhythm disturbances like jet lag, insomnia and shift work-associated sleep disorders may be on the horizon thanks to an unprecedented breakthrough made possible by ultra-large-scale “virtual pharmacology,” a cutting-edge technique that allows scientists to rapidly test hundreds of millions of molecules in computer simulations to identify those with the greatest pharmaceutical potential.
Using this approach, a research team led by scientists at UC San Francisco, the University of North Carolina at Chapel Hill and the University at Buffalo screened a massive library of over 150 million virtual molecules – chemical compounds that don’t exist, but which can be easily and inexpensively synthesized – and discovered the first drugs that selectively target one of two mammalian melatonin receptors. These receptors modulate sleep-wake cycles in response to melatonin, a hormone produced by the brain during daily light-to-dark transitions.
The new study, published Feb. 10, 2020, in the journal Nature, is the first to show how combining ultra-large-scale virtual pharmacology with the latest findings from structural biology can lead to the rapid identification of new drugs that produce strong, targeted responses in animal models of human disorders – in this case, circadian rhythm imbalances that disrupt sleep and lead to conditions like jet lag.
“Starting from the atomic structure of the melatonin receptor, we were able to find potent molecules that had unexpected effects on circadian rhythms in animal models of jet lag. This was only possible thanks to the chemical novelty afforded by an ultra-large virtual library of synthesizable compounds,” said Brian Shoichet, PhD, professor of pharmaceutical chemistry in UCSF’s School of Pharmacy and co-senior author of the new study.
Scientists estimate that there are over 1063 (1 followed by 63 zeros) drug-like molecules that can be chemically synthesized. However, only a tiny fraction of those are likely to have a desirable effect in living systems, which makes it extremely difficult for scientists to find new drugs. Virtual pharmacology and ultra-large molecular libraries are making it possible for scientists to rapidly probe a much larger portion of this immense chemical landscape, leading to new drug discoveries that may have been impossible just a few years ago.
This revolutionary approach was first described in a paper published last year – a collaboration between Shoichet’s lab and the lab of Bryan L. Roth, MD, PhD, the Michael Hooker Distinguished Professor of Pharmacology at the UNC School of Medicine and co-senior author of the new study. Though both papers demonstrate how virtual pharmacology has the potential to rapidly accelerate the pace of drug discovery, the new one is the first to take drugs discovered via virtual pharmacology beyond the petri dish and into animal models for testing. And the new study shows that this can lead to significant advances in other fields as well.
Scientists who study circadian rhythms have spent the past 15 years searching for drugs capable of distinguishing between MT1 and MT2, the two melatonin receptors found in humans. “Ever since we demonstrated that melatonin’s effect of resetting biological clocks in in vivo circadian models occurs through actions at the MT1 receptors, we have focused through various collaborations on searching for ligands that would better fit this human melatonin receptor,” said Margarita L. Dubocovich, PhD, SUNY Distinguished Professor in the Department of Pharmacology and Toxicology in the Jacobs School of Medicine and Biomedical Sciences at UB and co-senior author of the new study.
Despite these efforts, the search for MT1-specific drugs had proven fruitless. Every chemical tested thus far, including the melatonin that our own bodies produce, was equally good at binding to MT1 and MT2. And the recently published structures of the two receptors helped explain why.
“Both receptors are almost identical at their binding site, so it’s very difficult to find MT1-selective molecules,” explained Reed Stein, a doctoral candidate in the UCSF Pharmaceutical Sciences and Pharmacogenomics Graduate Program and lead author of the new study.
To surmount this daunting hurdle, Stein and Shoichet performed what are known as “docking” experiments – computer simulations in which virtual molecules are rotated and adjusted in order to find those that can to bind to a biological target of interest. Each virtual compound in the 150-million-molecule library was tested an average of 1.6 million times, with every test representing a slightly different interaction between the molecule and the two melatonin receptors. In total, over 72 trillion different interactions were tested.
The researchers identified 40 drug candidates from these docking experiments. They were able to have 38 synthesized from scratch by their collaborators at Enamine, a Ukraine-based company that pioneered methods to efficiently produce over a billion chemical compounds on demand. Of the 38 compounds synthesized, 15 induced a response from one or both human melatonin receptors, MT1 and MT2, with subtle pharmacological properties measured in test tubes and lab-cultured cells in the Roth lab.
The researchers further narrowed the field by testing the compounds in lab mice. Of those that were able to pass through the blood–brain barrier, two compounds in particular, dubbed UCSF7447 and UCSF3384, proved highly selective for MT1 and produced a potent response at low doses in mice. But unlike melatonin, which ramps up the activity of MT1, these compounds had the opposite effect and reduced MT1 activity.
“My UNC lab spent more than a year characterizing the pharmacology and drug-like properties of the molecules before we could hand them off for animal testing in the Dubocovich lab at UB,” said Roth. “We were all excited to see that the new compounds Brian and I had discovered had interesting properties in mice."
When researchers in the Dubocovich lab, which specializes in circadian biology, administered the compounds to mice that were placed in an environment where “night” arrives six hours earlier than normal – an animal model that mimics the jet lag experienced when flying from New York to Paris – the drugs made it harder for the mice to adjust to the earlier-than-expected onset of dark. By contrast, melatonin has the exact opposite effect, making it easier for mice (and humans) to adjust to the shift.
But the most surprising finding came when the researchers moved mice from an environment where day and night arrived at regular intervals into one that was perpetually dark. Under these conditions, if melatonin is administered when the animal’s internal clock thinks that it’s nighttime, the internal clock will shift forward, making the mouse active about an hour later than it would be under normal conditions. The researchers expected that the new drugs would have the opposite effect, as they had in all prior experiments. However, when the drugs were administered under these same conditions, the effects were identical to melatonin’s.
“The Dubocovich lab didn’t share the data with us for several months because they didn’t believe they were real. The molecule behaves like a melatonin receptor antagonist at dawn, but behaves like melatonin at dusk. We still don’t know what to make of this, but it shows that circadian clocks are stranger than we ever imagined,” said lead author Reed Stein.
For Dubocovich, these drugs open up new and unexpected avenues of circadian rhythm research. Meanwhile, Shoichet and Roth are eager to apply virtual pharmacology to other “orphan” receptors, signaling proteins that scientists cannot yet selectively target with any existing molecules.
“MT1 was one of about 100 orphans we’re planning to tackle. With no selective molecules with which to hit them, understanding the biology of orphans has been really hard,” Shoichet said.
With the latest updates to their vast virtual libraries, the researchers now have around 10 billion molecules at their disposal. Shoichet and Roth fully expect to be able to find more orphan receptor–targeting drugs. However, they’re also turning their attention to pain. “We’re looking for new molecules to treat pain without addiction, and the chemical novelty offered by our libraries has opened a whole new universe of candidates for drug discovery,” Shoichet said.
Authors: Additional authors include Hye Jin Kang, John D. McCorvy, Tao Che, Samuel Slocum, Xi-Ping Huang and Terry Kenakin of the University of North Carolina at Chapel Hill; Grant C. Glatfelter and Anthony J. Jones of the University at Buffalo; Olena Savych of Enamine Ltd; Yurii S. Moroz of National Taras Shevchenko University of Kyiv and Chemspace; Benjamin Stauch, Linda C. Johansson and Vadim Cherezov of the University of Southern California; and John J. Irwin of UCSF.
Funding: This work was funded by US NIH awards U24DK1169195 (Illuminating the Druggable Genome), R35GM122481, the NIMH Psychoactive Drug Screening Contract, GM133836, ES023684, UL1TR001412 and KL2TR001413, PhRMA Foundation Fellowship, Jacobs School of Medicine and Biomedical Sciences unrestricted funds, R35GM127086, EMBO ALTF 677-2014, HFSP long-term fellowship LT000046/2014-L, postdoctoral fellowship from the Swedish Research Council, and the NSF BioXFEL Science and Technology Center 1231306.
Disclosures: Shoichet and Irwin are founders of a company, BlueDolphin LLC, that works in the area of molecular docking. All other authors declare no competing interests.
The University of California, San Francisco (UCSF) is exclusively focused on the health sciences and is dedicated to promoting health worldwide through advanced biomedical research, graduate-level education in the life sciences and health professions, and excellence in patient care. UCSF Health, which serves as UCSF’s primary academic medical center, includes top-ranked specialty hospitals and other clinical programs, and has affiliations throughout the Bay Area.