AI-powered antibiotic discovery is proving effective in identifying new chemical structures and targets in the ongoing fight against antibiotic-resistant gonorrhea.
Written by Benjamin Bottner
(BOSTON) — Gonorrhea is the second most commonly reported sexually transmitted infection (STI), with tens of millions of cases annually. In the United States alone, more than 600,000 cases are reported each year. If gonorrhea is left untreated, it can lead to severe reproductive health problems such as infertility and pelvic inflammatory disease in both women and men. Infection also increases the risk of HIV infection, and if the pathogen spreads from the genitals and throat to other parts of the body, it can damage the heart and cause meningitis and sepsis. The main challenge in controlling the disease more effectively lies in the capacity of the causative pathogen. gonorrheadue to the rapid development of resistance to newly available antibiotics.
“Two new oral antibiotics, zoliflodacin and gepotidacin, were recently approved for the treatment of uncomplicated urogenital gonorrhea. These are the first completely new classes of antibiotics developed to fight infections in more than 30 years,” said physician-scientist Melis Anatar, MD, associate director of the clinical microbiology laboratory at Massachusetts General Hospital (MGH). “However, if these two antibiotics are widely used, it is almost certain that eventually pathogens will develop significant resistance to them. We have seen cycles of acquired resistance occur within just five to 10 years after the deployment of first-line drugs, and they have repeated themselves time and time again. To win in this continuing arms race, we will need new antibiotics to fill the pipeline.”
We have seen cycles of resistance development occur within just 5 to 10 years after the deployment of first-line drugs, and it has repeated itself over and over again. To win this continuing arms race, we will fill our pipeline with new antibiotics.
Now, a new study has been published scientific translational medicine A research team from the Wyss Institute at Massachusetts Institute of Technology and Harvard University, led by Wyss Institute core faculty member Dr. James Collins, and the Broad Institute of Massachusetts Institute of Technology and Harvard University, led by Anatal, Jacqueline Valeri, and Majid Modareshi, is proposing an exciting new strategy that could identify new compounds that can be further developed with high selectivity for antibiotic therapy. gonorrhea. The researchers initially hypothesized that entirely new chemical structures with antimicrobial activity could dramatically reduce the likelihood that antimicrobial resistance would develop because they could also target unusual cellular pathways in pathogens, and that deep learning-based antimicrobial discovery approaches could take the lead in identifying these structures.
“We have reached a critical juncture in which a vast chemical space has opened up in which billions of compounds with well-defined structures can be synthesized. This, combined with the rapidly evolving capabilities of machine learning, will allow us to explore that space with very specific biological activities in mind, such as much-needed new antimicrobial activities,” said senior author Collins. “This research builds on a body of research in our lab that focuses on leveraging artificial intelligence to fight infectious diseases. gonorrhea To address the growing crisis of antimicrobial resistance against this rapidly evolving pathogen. ” Collins also said, Termeer Professor of Medical Engineering and Science from MIT and is a member of the Broad Institute at MIT and Harvard University.
We have reached a critical point in time, when a vast chemical space has opened up in which billions of compounds with well-defined structures can be synthesized. This is fused with rapidly evolving machine learning capabilities, which allow us to explore the area with very specific biological activities in mind, such as much-needed new antimicrobial activities.
Building a machine learning pipeline
To lay the foundation for their approach, the team first tested the ability of 38,650 small molecules to inhibit proliferation. gonorrhea We analyzed data from laboratory assays and used this dataset to train a predictive deep learning model. They validated that the model could identify potential antimicrobial-like molecules with chemical structures that differ from common antibiotics.
After gaining confidence in the model’s ability to find “hidden gems” with anti-gonococcal activity, they used the AI model to virtually screen a much larger library of approximately 6 million compounds. This resulted in 213 candidates that were further validated. After a series of growth inhibition and antimicrobial resistance assays, as well as cell biological assays to exclude compounds with undesired toxicity, we were able to identify two compounds with promising selectivity and potent potency against multidrug resistance. gonorrhea The strain itself causes resistance at very low frequencies.
“Using proteomics methods, we succeeded in identifying the most promising compound target called A1, a so-called aminothiazole compound with previously undescribed anti-gonococcal activity. It specifically binds to and inhibits the key enzyme alanine racemase.” gonorrhea “We used genetic tools to validate A1’s alanine racemase specificity and are now investigating how exactly A1 inhibits its enzymatic activity,” Anatal said. Multiple existing antibiotics inhibit the cell wall biosynthesis process of pathogenic bacteria, but specifically targeting alanine racemase with small molecules is a novel mechanism uncovered by the research team.
from in silico to in vivo
In the next translational step, the research team investigated whether their compounds could exert antigonococcal activity in the physiological tissue environment of the vagina. gonorrhea It is done frequently. Working with the group of Wyss founding director and co-author Donald Ingber, MD, who had previously developed a microfluidic organ-chip model of the human vagina, they demonstrated that their first compound, MP20, significantly reduced pathogen titers after being introduced into the device and interacting with vaginal epithelial cells. In addition, in a mouse vaginal infection model inoculated intravaginally, gonorrhea For bacteria, treatment with the second compound A1 five times over 24 hours significantly reduced pathogen concentrations compared to the antibiotic-free control.
“While our observations with A1 are promising, further validation and hit-to-lead optimization through medicinal chemistry and other efforts are required before it becomes a clinically relevant antibiotic for the treatment of gonorrhea,” Anatal said. “However, our deep learning-enabled discovery pipeline has the potential to screen much broader, ultra-large make-on-demand chemical libraries to identify unexpected compounds as a new starting point for a gonorrhea-focused antibiotic development program.”
Our deep learning-enabled discovery pipeline has the potential to screen broader, ultra-large make-on-demand chemical libraries to identify unexpected compounds as new starting points for gonorrhea-focused antibiotic development programs.
“This work by Jim Collins and his team once again demonstrates the enormous power of AI, combined with high-quality biological datasets, in discovering potential therapeutic compounds that are otherwise completely out of reach. It also shows how the Wyss Institute is seamlessly integrating important advances in AI with human-relevant models, in this case the human vaginal chip,” said co-author Ingber, MD. who is too Judah Folkman Professor of Vascular Biology Harvard Medical School and Boston Children’s Hospital, and Hansjörg Wyss Professor of Bioinspired Engineering Completed Harvard University John A. Paulson School of Engineering and Applied Sciences.
Other authors of this study include Aarti Krishnan, Nina Donghia, Samantha Palace, Erica Zheng, Aakanksha Gulati, Alicia Jorgenson, Abidemi Junaid, Parijat Bandyopadhyay, Andreas Luttens, Krishna Suresh, Paige Edwards, Felix Wong, Yu Zhang, Danilo Ritz, Margaux Gaboleau, Edmund Loh, Massimiliano Gaetani, Marie-Stéphanie Astgen, Amir Ata Saei, and Yonatan Grad.
This research was supported by the Wyss Institute at Harvard University, the Massachusetts Institute of Technology and the Broad Institute at Harvard University, the Defense Threat Reduction Agency (grant number HDTRA-12210032), and the National Institutes of Health (grant numbers R01AI146194, K08AI182474, T32CA921641, R01AI132606, R01AI153521). K25AI168451), Siebel Scholars Foundation, MIT–Novo Nordisk Artificial Intelligence Postdoctoral Fellows Program, Swiss National Science Foundation (Grant # SNSF_203071), Knut and Alice Wallenberg Foundation Grant (Grant # KAW2022.0347), Swedish Research Council (Grant # 2023-02692), the Bill and Melinda Gates Foundation, which is part of the Antibiotic AI Project led by James Collins and supported by the Audacious Project, Full Lab, LLC, the Sea Grape Foundation, Rosamund Sander and Hansjorg Wyss of the Wyss Foundation, and an anonymous donor.
