A genomic analysis combining machine learning with laboratory testing identified previously underrecognized genes associated with antibiotic resistance that could eventually serve as targets for new therapies, according to findings to be presented at IDWeek 2026.
Researchers analyzed data from VAMPr, or VAriant Mapping and Prediction of antibiotic resistance, a machine-learning model used to identify associations between bacterial genetic variants and antibiotic resistance. The dataset included 3,393 bacterial genomes with corresponding antibiotic susceptibility data across nine bacterial species and 29 antibiotics.
Presenting author Drew A. Beckman, BS, an MD/MPH candidate at UT Southwestern Medical School and the Peter O’Donnell Jr. School of Public Health, said in an interview with the IDSA Reporter that the broader goal was to show that important antibacterial drug targets may still remain undiscovered.
“This work attempts to illustrate that new antibacterial drug targets remain to be found,” Beckman said. “We were able to use a large dataset that found mutations in genes that were statistically associated with resistance to identify genes that either were underappreciated or not previously known to be associated with resistance to a particular antibiotic.”
From thousands of genetic associations to potential targets
The researchers started with 219,578 combinations of genetic variants, bacterial species and antibiotics. They then focused on variants found exclusively in resistant strains, narrowing the analysis to 732 combinations representing 444 variants across 123 genes or proteins.
Among the candidates were variants involving several transporter systems, as well as a methyltransferase and an oxidoreductase.
To determine whether selected genes might actually contribute to resistance, the researchers used peptide-conjugated antisense phosphorodiamidate morpholino oligomers, or PPMOs. These molecules were designed to block specific bacterial genes and were tested alongside established antibiotics in laboratory synergy assays.
Beckman said the antisense experiments were intended to validate examples identified through the broader genomic analysis.
PPMOs targeting lapB and lapE in Enterobacter species enhanced the activity of existing antibiotics in laboratory testing. One lapB-targeting PPMO showed synergy with both meropenem and tobramycin across several strains of Enterobacter cloacae and Enterobacter hormaechei.
Why the LapEBC system stood out
Beckman said lapB, lapC and lapE appeared more often than any other proteins across the 732 resistance-associated models, regardless of the specific genetic variant, antibiotic or bacterial species being examined.
LapB, LapC and LapE form an ABC transporter system that helps move the adhesion protein LapA outside the bacterial cell. Beckman said the proteins play different roles within that system: LapB is an ATPase, LapC is a membrane fusion protein and LapE is an outer-membrane porin. He also noted that LapB binds to another protein that disrupts synthesis of lipopolysaccharide, an important component of the outer membrane of Enterobacterales.
“This is really quite fascinating because these three genes have not been clearly identified as major players in antimicrobial resistance in prior research studies,” he said.
The researchers hypothesize that the transporter system may also act on antibiotics, pumping them out of bacterial cells before they can successfully inhibit bacterial growth.
Potential for drug development
Beckman described the study as proof of concept that large quantities of nationally available bacterial genomic data can be analyzed to identify potential new targets for antibiotic development, and said the implications extend beyond the PPMOs used in this study.
“While we believe that antisense molecules could be an eventual new type of antibiotic, the broader identification of new drug targets could serve as a platform for new drug development in general,” he explained.
In other words, the genomic analysis could identify useful drug targets even if future treatments ultimately use approaches other than PPMOs.
The work has so far been limited to laboratory testing. Beckman said the group is now developing PPMOs against essential bacterial targets in major human pathogens, including Pseudomonas aeruginosa.
“Our hope is that this antisense chemistry, which is currently FDA approved for treating certain diseases like Duchenne muscular dystrophy, could one day enter human trials as antibiotics,” he said.
Beckman, Jiwoong Kim, MS, and KannaNagamatsu, PhD, reported no relevant financial relationships or conflicts of interest. David E. Greenberg, MD, reported service as a data and safety monitoring board member for Johnson & Johnson and holds multiple patents related to antisense compounds.