Posted on Feb 28, 2020, 5 p.m.
A powerful new antibiotic compound has been identified using a machine deep learning algorithm that in MIT lab testing was successful at killing off many of the world’s most problematic disease causing bacteria which included some strains that have become resistant to all currently known antibiotics and clearing infections in two different mouse models.
In MIT research the computer model screened over one hundred million chemical compounds within a matter of a few days, and it was designed to identify potential antibiotics that can kill bacteria using different mechanisms than those of currently existing drugs.
"We wanted to develop a platform that would allow us to harness the power of artificial intelligence to usher in a new age of antibiotic drug discovery," says James Collins, the Termeer Professor of Medical Engineering and Science in MIT's Institute for Medical Engineering and Science (IMES) and Department of Biological Engineering. "Our approach revealed this amazing molecule which is arguably one of the more powerful antibiotics that has been discovered."
Several other promising candidates were also identified in the process which will also be tested further. The model may also be useful to design new drugs based on what the deep learning algorithm has learned about chemical structures that enable drugs to kill off bacteria.
"The machine learning model can explore, in silico, large chemical spaces that can be prohibitively expensive for traditional experimental approaches," says Regina Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science in MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL).
"We're facing a growing crisis around antibiotic resistance, and this situation is being generated by both an increasing number of pathogens becoming resistant to existing antibiotics, and an anemic pipeline in the biotech and pharmaceutical industries for new antibiotics," Collins says.
Using predictive computer models for screening is not a new idea, but until recently these models were not accurate enough to transform drug discovery. New neural networks can learn representation of chemical groups automatically then map the molecules into continuous vectors which are subsequently used to predict properties.
This model was designed to look for chemical features that make molecules effective at killing E.coli in a process that involved training on 2,500 molecules including 1,700 FDA approved drugs and a set of 800 natural products with diverse structures and a range of bioactivities. Once trained it was tested on a library of 6,000 compounds, and the model picked out one molecule predicted to have strong antibacterial activity and chemical structure different from any existing antibiotics. Then using a different machine deep learning algorithm model the newly identified Halicin molecule was shown to likely have low toxicity to human cells.
The Halicin molecule was tested against dozens of bacterial strains which were both isolated from patients and grown in lab; it was found to kill off many that are resistant to current treatments including C. diff, acinetobacter baumannii, and mycobacterium tuberculosis. The newly identified drug worked against every species it was tested against with exception to the Pseudomonas aeruginosa lung pathogen.
It was successfully used to treat live animals infected with A. baumannii that is known to be resistant to all known antibiotics and found to be able to completely clear infections within 24 hours of application of a halicin containing ointment.
Preliminary findings suggest that the newly AI identified halicin is able to kill bacteria by disrupting their ability to maintain an electrochemical gradient across cell membranes in a process which could be difficult for bacteria to develop resistance to according to the researchers.
"When you're dealing with a molecule that likely associates with membrane components, a cell can't necessarily acquire a single mutation or a couple of mutations to change the chemistry of the outer membrane. Mutations like that tend to be far more complex to acquire evolutionarily," Stokes says.
During testing E. coli was not able to develop resistance to halicin over a 30 day treatment period, while bacterial resistance developed against the antibiotic ciprofloxacin within 1-3 days and after 30 days the bacteria was 200 times more resistant than it was at the beginning of the study.
The model was also used to screen over 100 million molecules selected from the ZINC15 database in a process that only took 3 days; 23 candidates were identified that were structurally different from existing antibiotics and were predicted to be non toxic to human cells. In lab testing against 5 species of bacteria 8 of the molecules displayed antibacterial activity, 2 of which were especially potent. There are plans to further test these molecules and screen more from the remaining 1.5 billion chemical compounds in the ZINC15 library.
The new model may also be used to design new antibiotics as well as optimize existing models, for example adding features to an antibiotic to enable it to only target certain bacteria while preventing it from harming beneficial bacteria within the digestive tract.
Materials provided by:
Note: Content may be edited for style and length.
This article is not intended to provide medical diagnosis, advice, treatment, or endorsement.