Written by 
A new antibiotic termed Halicin has been discovered through the revolutionary use of artificial intelligence. The discovery was led by multiple departments including the Massachusetts Institute of Technology and Harvard MIT. The drug has been shown to be effective against both antibiotic sensitive and tolerant strains of bacteria.
Traditionally, antibiotics were discovered through screening of microbes for growth inhibitory molecules. However, recently developed algorithms in neural network-based molecular representations have been able to hugely advance the rate of discovery of new antibiotics. The network learnt to predict the inhibition of E. coli growth then applied this to a database of known molecules to identify which molecules would be potentially be suitable for further investigation. 120 compounds were discovered that could inhibit growth of bacteria by at least 80%, with 99 molecules common to the Drug Repurposing Hub, a database of known drugs that could have additional functions to their current use). From this, 51 molecules were isolated and screened for human toxicity using the ClinTox database. This finally led to the discovery of the antibiotic halicin as being the only molecule to comply to all the criteria set by the AI models.
The nitrothiazole structure of halicin makes it unique to other antibiotics in use, which would suggest it can have a divergent function from other drugs. Excitingly, halicin was shown to be efficient in killing both actively growing bacteria and cells halted in their growth. Even more interestingly, halicin was able to kill cells that persisted after treatment with ampicillin, a common antibiotic that is becoming subject to resistance.
The World Health Organisation has recommended M. tuberculosis, A. baumannii, P. aeruginosa and Enterobacteriaceae as being the pathogens that most urgently need new treatments. Halicin was able to rapidly kill M. tuberculosis and greatly slow the growth of the other pathogens excluding P. aeruginosa, as salicin could prevent these bacteria from being able to cross the cell membrane. This data is extremely promising for the prospect of halicin becoming available to cure dangerous infections.
Halicin functions by preventing the bacterium from being able to generate ATP, a universal molecule absolutely essential for the release of energy in the cell, and hence its survival. Scientists have been able to detect which genes in bacteria were expressed differently when they were treated with halicin. In addition to loss of ATP, genes involved in the cells ability to move in its environment were downregulated, while genes for iron regulation were upregulated. It has been suggested that halicin prevents its essential iron uptake by the bacterium and therefore, stops bacterial growth.
The fact that halicin showed significant ability to kill bacteria in usually antibiotic-intolerant cells is very promising in the current climate of progressive antibiotic resistance. Halicin was able to kill extremely resistant bacteria that have resistance genes against multiple common antibiotics. The possibility of this new drug being able to combat antibiotic resistance has huge implications in healthcare, with the WHO estimating approximately 700,000 people die every year from resistant infections and predicting this statistic is likely to continue rising due to the lack of funding for new antibiotic research. Halicin could present hope for treating resistant infections, showing a potential new generation of antibiotic and also demonstrating how effective the use of AI in drug discovery can be.
antibiotic resistance antibiotics Artificial Intelligence e.coli
Last modified: 8th March 2020
