Can we predict antibiotic resistance before it strikes?
Multidrug-resistant bacteria kill five million people each year, with newly resistant germs emerging faster than scientists can develop treatments. Now, researchers have developed a platform that identifies drug resistance genes already circulating in the environment before they emerge in the clinic and directly couples this information to the design of resistance-evasive antibiotics.
Image by Freepik.
The findings, published in PNAS, use metagenomic surveys of the so-called 'resistome' as an early warning system that can alert scientists to resistance likely to become a problem in the future. With this information, antibiotics in development can be proactively optimised to make them more resilient against our microbial foes.
“We’re predicting the types of resistance likely to be a problem in the future,” said lead author James Peek, a research associate in the laboratory of Sean F Brady at The Rockefeller University, adding: “We hope that our platform will help give antibiotics longer clinical lifespans.”
Untapped potential
Antibiotic development is often an endless cycle of finding new compounds to replace those that have become ineffective. Although scientists try to optimise drugs against resistances predicted in the lab and resistant strains that crop up in the clinic, the current system has proven ill-equipped to accurately anticipate novel threats.
Brady’s Laboratory of Genetically Encoded Small Molecules at Rockefeller suspected that there was a better way forward. They knew that bacteria in nature have spent millennia battling one another with antibiotics and resistance genes, forming a vast reservoir of resistance mechanisms in the environment, which we now know includes many of the same mechanisms that appeared in clinics. For instance, the very same types of resistance genes that dealt a major blow to antibiotic classes such as beta-lactams circulated in populations of soil bacteria long before these drugs entered clinical use.
“There’s now strong evidence that clinical resistance can originate among bacteria fighting in the environment,” said Peek. Tomorrow’s resistance mechanisms may already be present in today’s soil samples. The challenge was finding a way to access that information and use it to improve human health.
Mining the data
For the study, the team focused on albicidin, a promising antibiotic candidate. With 3.5 terabase pairs of microbial DNA extracted from soil, roughly 700,000 bacterial genomes, they built a metagenomic library and introduced it into E coli, a model bacterial host that could be easily screened to identify albicidin resistance genes. Bacteria that survived albicidin exposure were isolated, and their resistance genes were sequenced. The screen revealed eight classes of resistance genes, which were further analysed to identify how each disables the drug. “We found a lot of interesting, unusual mechanisms,” added Peek, continuing: “We were surprised by how well this model lent itself to finding unknown types of resistance.”
To figure out how to evade these resistance mechanisms, the researchers looked at natural structural variants of albicidin with the rationale that these variants may have evolved in the battle between soil microbes to circumvent resistance. Each variant tested had a unique vulnerability profile against the different types of resistance, which revealed chemical features that helped some variants remain effective. With this information, they began prioritising promising drug leads. One variant (congener 10), with several structural differences compared to albicidin, was particularly promising, as it continued to function in the face of the most common resistance types.
Ultimately, the team demonstrated that their method could guide drug design by engineering new versions of albicidin that combined the most protective features into compounds that remained potent in the face of even the most formidable resistance proteins.
Brady, Peek, and colleagues hope that pharmaceutical companies will adopt their technique to test a candidate drug’s susceptibility to pre-existing forms of resistance in the environment as they decide whether to move forward with development. “It’s fast and efficient,” Peek said, adding: “We think it would be easy for drug companies to integrate this method into the standard drug development pipeline.”
In the short term, the team plans to apply their screening platform to other antibiotics developed in the Brady lab. By identifying and addressing environmental vulnerabilities early on, they hope to generate candidates with longer clinical lifespans and fewer chances of being undermined by resistance.
DOI: 10.1073/pnas.2504781122