Discovery of antibiotics historically relied on screening microbes in the soil to find secondary metabolites that could prevent pathogenic bacterial growth. This process revealed the usefulness of B-lactams, aminoglycosides, polymyxins, and glycopeptides. Then, semisynthetic derivatives of many antimicrobials provided improved choices and efficacy from greater potency with less toxicity. Fully synthetic antibiotics include quinolone and sulfa drugs.
Very few new antibiotics have been developed in recent decades, however, and the ones that have are variations on the chemistry of existing drugs. With growing concerns about antimicrobial resistance, researchers at MIT turned to artificial intelligence (AI) in hopes the deep learning model could predict potentially useful next-generation therapeutics. Their research used algorithms to predict molecular properties and interactions with pathogens.
Artificial Intelligence Identifies Antibiotic Halicin
Using artificial intelligence, researchers identified the antibiotic halicin from the Broad Institute’s Drug Repurposing Hub. Halicin is structurally divergent from conventional antibiotics. It is a potent inhibitor of E. coli growth, disrupting a microbe’s ability to maintain an electrochemical gradient across its cell membranes. Loss of the electrochemical gradient interferes with microbial production of ATP and causes microbial cells to die. This particular therapeutic pathway sidesteps bacterial resistance because complex mutations are necessary to change the chemistry of the outer membrane.
Study Analyzing Halicin
Halicin is a broad-spectrum bactericidal antibiotic that performs far better than ampicillin against E. coli. In a study analyzing the efficacy of a 30-day treatment with halicin, E. coli did not develop resistance to the antibiotic. E. coli did become resistant to ciprofloxacin within one to three days. By Day 30, E. coli was 200 times more resistant to ciprofloxacin.
Halicin also inhibits the growth of and provides good efficacy against Mycobacterium tuberculosis, carbapenem-resistant Enterobacteriaceae, Clostridium difficile, and Acinetobacter baumannii, which to date had been resistant to all known antibiotics. A. baumannii infection cleared within 24 hours of treatment with halicin. Researchers also determined that halicin clears Clostridium difficile infection at a greater rate than metronidazole, which has been the first-line treatment for this infection. In 2022, halicin was found to be effective against Staphylococcus aureus biofilms in vitro. However, halicin is not effective against Pseudomonas aeruginosa.
Final Thoughts
In summary, halicin can control many problematic diseases that also include strains resistant to all antibiotics. Its pharmacokinetics are yet to be fully tested to the satisfaction of the FDA. Using AI, the researchers identified eight other antibacterial compounds structurally different from conventional antibiotics, with two of these showing broad-spectrum activity, meriting more testing. This type of ongoing research might yield many next-generation antibiotics, including those that have a targeted narrow spectrum that outpaces antimicrobial resistance genes.
Reference
Stokes JM, Yang K, Swanson K, et al. A Deep Learning Approach to Antibiotic Discovery. Cell 2020, vol. 180; pp. 688-702; DOI: 10.1015//j.cell.2020.01.021
Related Reading
- Disease Du Jour: Artificial Intelligence Use in Veterinary Medicine
- Interpreting Antimicrobial Sensitivity: More Than Sensitivity/Resistance
- Microrobots for Targeted Treatment with Drugs
Stay in the know! Sign up for EquiManagement’s FREE weekly newsletters to get the latest equine research, disease alerts, and vet practice updates delivered straight to your inbox.
