AI turns old diabetes drug Halicin into a potent weapon against superbugs

Can an AI-repurposed diabetes drug tackle our toughest superbugs? Researchers reveal Halicin’s powerful action against deadly multidrug-resistant bacteria, except for one elusive foe.

Study: Halicin: A New Approach to Antibacterial Therapy, a Promising Avenue for the Post-Antibiotic Era. Image Credit: Kateryna Kon / Shutterstock

In a recent study in the journal Antibiotics, researchers demonstrate how artificial intelligence (AI)-driven drug discovery can repurpose now-replaced pharmaceuticals and biomolecules with novel, currently relevant therapeutic purposes. Specifically, they report the results of an antibacterial activity assay wherein AI-predicted Halicin efficacy was tested against 18 multidrug-resistant (MDR) bacterial strains.

Minimum inhibitory concentration (MIC) assays revealed that Halicin significantly inhibited the growth of 17 of the 18 clinical bacterial isolates tested. The study also confirmed Halicin’s efficacy against two standard reference strains, Staphylococcus aureus ATCC® 29213™ and Escherichia coli ATCC® 25922™. These findings support future investigation into Halicin’s potential as a broad-spectrum antibiotic against MDRs and highlight the remarkable way in which AI is transforming medicine and drug discovery.

Background

Colloquially termed ‘superbugs’, multidrug-resistant (MDR) bacteria pose an escalating threat to global health. Among them, ESKAPE bacterial strains (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) are consistently recognized by the World Health Organization (WHO) as the greatest threat due to their extraordinary ability to evade most conventional antibiotic courses.

Unfortunately, these threats emerge at a time when traditional antibiotic pipelines are reaching the limits of their innovative potential, primarily due to the time-intensive nature of their discovery processes and the parallel evolution of bacterial defenses. Thankfully, modern innovations in machine learning (ML) and artificial intelligence (AI) technologies are increasingly enabling the rapid screening and simulation of existing pharmaceutical compounds, identifying hidden antibacterial properties that are invisible to traditional drug discovery approaches.

A remarkable success of this approach is Halicin. Originally created as a c-Jun N-terminal kinase (JNK) inhibitor to target diabetes-associated pathways, the drug was identified by deep learning algorithms at the Massachusetts Institute of Technology (MIT) for its unusual antibacterial ability to disrupt the bacterial proton-motive force, a mechanism distinct from conventional antibiotics, thereby suggesting its efficacy against multidrug-resistant (MDR) bacteria. Unfortunately, while promising, detailed investigations into its activity against clinical MDR isolates remain limited, and potential minimum inhibitory concentrations (MICs) against many priority pathogens require further study.

About the study

The present study, described as the first of its kind in Morocco, aims to address this knowledge gap by estimating the MIC of Halicin across a spectrum of 18 clinically validated MDR bacterial isolates. Isolate samples were collected from Moroccan hospitals, and agar disk diffusion assays were used first to confirm their MDR status against 22 commonly used antibiotics. In addition to these clinical isolates, standard reference strains S. aureus ATCC® 29213™ and E. coli ATCC® 25922™ were included as quality controls.

The study methodology adhered to the guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the Clinical and Laboratory Standards Institute (CLSI). Following isolate validation, broth microdilution and Halicin MIC assays were performed to determine the lowest drug concentration (in μg/mL) that prevents visible growth of each isolate strain.

MIC data were used to generate dose-response curves, thereby elucidating the dynamics of bacterial growth across various concentrations. Simultaneously, scanning electron microscopy (SEM) imaging was conducted to visualize the physiological impacts of Halicin treatment on the E. coli reference strain. Differences between MIC distributions across concentration and species results were estimated using the Kruskal-Wallis non-parametric test.

Study findings

Halicin was observed to demonstrate commendable antibacterial activity, producing MICs of 16 μg/mL and 32 μg/mL against reference E. coli ATCC® 25922™ and S. aureus ATCC® 29213™ strains, respectively. Dose-dependent outcomes against the clinically MDR-validated bacterial isolates from the ESKAPE group ranged from 32 to 64 μg/mL, confirming Halicin’s broad-spectrum potential.

Surprisingly, however, P. aeruginosa was found to be completely intrinsically impervious to Halicin, with no growth inhibition observed irrespective of treatment concentration. Researchers attributed this observation to the bacteria’s robust outer membrane, which limits Halicin penetration, effectively restricting its efficacy.

Despite this exception, the hitherto anti-diabetic drug’s ability to kill several multidrug-resistant strains presents a promising step forward in the global search for novel antibacterial agents. Its unique mode of action, disrupting bacterial energy metabolism rather than targeting cell walls or protein synthesis, bypasses the MDR mechanisms of most of today’s most dangerous bacteria, and may make it harder for future bacteria to develop resistance quickly.

Conclusions

The present study validates the antibacterial efficacy of Halicin, a largely discontinued anti-diabetic relic, in significantly inhibiting the growth of 17 of the 18 (94%) clinical MDR bacterial isolates tested. The study also confirmed Halicin’s activity against reference strains of S. aureus and E. coli. The findings indicate that Halicin is effective against bacteria that have already developed resistance to many conventional antibiotics, promoting future research into its safety and optimal dosage.

This study also highlights the ability of novel AI and ML innovations to surpass conventional drug discovery limitations, repurposing existing compounds for new therapeutic uses. Future work should examine pharmacokinetics, toxicity, and in vivo efficacy, and explore combination therapies that might overcome barriers posed by certain bacterial defences. The paper’s authors also stress the importance of establishing bacterial resistance monitoring programs to track Halicin’s long-term efficacy, noting that while no resistance has yet been observed due to its limited use, vigilance will be crucial as development proceeds.