Deep learning identifies promising antibiotic candidates against drug-resistant gonorrhea

Deep learning identifies novel compounds against antibiotic-resistant gonorrhea

Wyss Institute for Biologically Inspired Engineering at HarvardJun 17 2026 With tens of millions of annual cases, gonorrhea is the second most frequently reported sexually transmitted infection (STI). Alone in the U.S., over 600,000 cases are reported each year. If left untreated, gonorrhea can result in severe reproductive health issues, including infertility in both women and men and pelvic inflammatory disease. The infection also increases the risk of HIV transmission and, if the pathogen spreads from the genitals or throat to other parts of the body, it can damage the heart and cause meningitis and sepsis. The major challenge in more effectively controlling the disease lies in the ability of the responsible pathogen, Neisseria gonorrhoeae, to rapidly develop resistance against newly available antibiotics. "With zoliflodacin and gepotidacin, two new oral antibiotics have recently been approved to treat uncomplicated urogenital gonorrhea. These are the first entirely new classes of antibiotics developed to fight the infection in over thirty years," said Melis Anahtar, M.D., Ph.D., a physician-scientist who serves as Assistant Director of the Clinical Microbiology Laboratory at Massachusetts General Hospital (MGH). "But if these two antibiotics get used broadly, it's nearly guaranteed that the pathogen will develop significant resistance against them eventually. We've seen the cycle of resistance development occur within just five to 10 years after first-line roll-out, it has happened over and again. To be able to prevail in this continuous arms race, we will new antibiotics to fill the pipeline." Now, a new study published in Science Translational Medicine led by Wyss Institute Core Faculty member James Collins, Ph.D. at the Wyss Institute at Harvard University, MIT, and Broad Institute of MIT and Harvard, which was spearheaded by Anahtar, Jacqueline Valeri, and Majed Modaresi, offers an exciting new strategy capable of identifying new chemical compounds that could be further developed into antibiotic therapies with high selectivity for N. gonorrhoeae. At its start, the researchers hypothesized that entirely new chemical structures with antimicrobial activity could dramatically lower the chances of antimicrobial resistance from occurring because they might also target uncommon cellular pathways in the pathogen, and that to identify those structures, deep learning-guided antimicrobial discovery approaches could lead the way. "We have arrived at an incredibly important point in time in which a vast chemical space has opened up in which billions of chemical compounds with clearly defined structures can be synthesized. This converges with the rapidly evolving capabilities of machine learning that allow us to explore that space with very specific biological activities, such as much-needed new antimicrobial activities, in mind," said senior author Collins. "This study builds on a body of work in our lab that leverages artificial intelligence to combat infectious diseases and brings that focus to N. gonorrhoeae to help address the growing crisis of antimicrobial resistance for this fast-evolving pathogen." Collins is also the Termeer Professor of Medical Engineering & Science at MITand an Institute member of the Broad Institute of MIT and Harvard. Building a machine learning pipeline To build the foundation for their approach, the team first tested 38,650 small molecules for their ability to inhibit the growth of N. gonorrhoeae in laboratory assays and then used this data set to train a predictive deep learning model. They validated that the model was able to identify potential antibacterial, drug-like molecules with chemical structures that differed from those of common antibiotics. After gaining confidence in the model's ability to find "hidden gems" with anti-gonococcal activity, they used their AI model to virtually screen a much larger library of about 6 million compounds. This yielded 213 candidates that they validated further. Following a series of growth inhibitory and antimicrobial resistance assays, as well as cell biological assays to exclude compounds with unwanted toxicities, they were able to pinpoint two compounds with promising selectivity for and strong potency against multi-drug resistant N. gonorrhoeae strains that themselves caused resistance at very low frequencies. "Using proteomic methods, we succeeded in identifying the target for our most promising compound called A1, a so-called aminothiazole compound with previously undescribed anti-gonococcal activity. It specifically binds and inhibits the critical enzyme alanine racemase, which N. gonorrhoeae needs to build its cell wall," said Anahtar, adding "We validated the alanine racemase-specificity of A1 using genetic tools and are now in the process of investigating how exactly A1 inhibits its enzyme activity." Multiple existing antibiotics inhibit the cell wall biosynthesis process of pathogenic bacteria, however, specifically targeting alanine racemase with a small molecule is a novel mechanism revealed by the team. From in silico to in vivo In a next translational step, the team investigated whether their compounds could exhibit anti-gonococcal activity in the physiological tissue environment of the vagina where infection with N. gonorrhoeae frequently takes place. Collaborating with the group of Wyss Founding Director and co-author Donald Ingber, M.D., Ph.D., which had previously developed a microfluidic Organ Chip model of the human vagina, they demonstrated that their first compound, MP20, significantly lowered the titers of the pathogen after it had been introduced into the device and interacted with vaginal epithelial cells. Also, in a mouse vaginal infection model where they intravaginally inoculated N. gonorrhoeae bacteria, five treatments with their second compound, A1, over a period of 24 hours significantly lowered the pathogen concentration relative to the no antibiotic control. "While our observations on A1 are promising, it requires further validation and hit-to-lead optimization through medicinal chemistry and other efforts in order to become a clinically relevant antimicrobial drug for treating gonorrhea," said Anahtar. "However, our deep learning-enabled discovery pipeline has potential for screening much more extensive, ultra-large, make-on-demand chemical libraries to identify unexpected chemical compounds as new starting points in gonorrhea-focused antibiotic development programs." "This study by Jim Collins and his team showcases once again the enormous power of AI combined with high quality biological data sets in the discovery of potentially therapeutic compounds that otherwise would be entirely out of reach. It also shows how, at the Wyss Institute, we seamlessly integrate critical advancements in AI with human-relevant models, in this case a human Vagina Chip," said co-corresponding author Ingber, M.D., Ph.D. who is also the Judah Folkman Professor of Vascular Biology at Harvard Medical School and Boston Children's Hospital, and the Hansjörg Wyss Professor of Biologically Inspired Engineering at Harvard John A. Paulson School of Engineering and Applied Sciences. Other authors on the study are 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 Gaborieau, Edmund Loh, Massimiliano Gaetani, Marie-Stephanie Aschtgen, Amir Ata Saei, and Yonatan Grad. Source: Wyss Institute for Biologically Inspired Engineering at Harvard Journal reference: Anahtar, M. N., et al. (2026). Deep learning-enabled discovery of antibiotics effective against Neisseria gonorrhoeae. Science Translational Medicine. DOI: 10.1126/scitranslmed.ads4699. https://www.science.org/doi/10.1126/scitranslmed.ads4699

New antibiotic candidate emerges from AI screening

Researchers at the Wyss Institute for Biologically Inspired Engineering at Harvard University have identified new antibiotic compounds effective against multi-drug resistant Neisseria gonorrhoeae using deep learning, according to a study published Wednesday in Science Translational Medicine. The development comes amid rising concerns over gonorrhea's resistance to current treatments, with the World Health Organization designating N. gonorrhoeae as a high-priority pathogen. In 2023, Massachusetts reported the first U.S. case of gonorrhea exhibiting reduced response to five classes of antibiotics. The research team utilized directed message-passing neural networks to screen extensive chemical libraries for molecules with antigonococcal activity. They identified candidates distinct from existing antibiotics. This work builds on previous antibiotic discoveries using deep learning methods at the Wyss Institute and MIT, including compounds effective against MRSA published in Nature in 2023 and a generative AI framework yielding other compounds against MRSA and gonorrhea. For the first time, the new study validated its lead compounds using the Wyss Institute's microfluidic Organ Chip technology, specifically a human vagina chip that replicates the vaginal tissue microenvironment, alongside a mouse vaginal infection model. This validation represents a methodological advance in antibiotic preclinical testing by allowing efficacy assessment in a system that more closely mimics human physiology compared to standard cell cultures. The findings contribute to a trend of AI-driven antibiotic candidates moving through preclinical validation stages. Two antibiotics for gonorrhea, gepotidacin and zoliflodacin, received FDA approval in late 2025, marking the first new drugs for the disease in decades. It remains uncertain if the newly identified compounds will progress to clinical trials. The combination of deep learning with organ-on-chip validation offers a quicker pipeline to identify and test possible treatments for resistant pathogens.