Antimicrobial activity of antibiotic-induced resistant lactobacilli against carbapenemase-producing Enterobacterales

In an in vitro study, lactobacilli made less susceptible to several antibiotics through repeated exposure to sublethal drug concentrations retained their ability to reduce viable carbapenemase-producing Enterobacterales. Wild-type and antibiotic-induced variants of three lactobacillus strains reduced all 17 clinical isolates tested below the culture assay’s limit of detection by 48 hours, while Enterobacterales grown without lactobacilli remained at high viable counts. The findings are a preliminary microbiological proof of concept and do not show that antibiotic-resistant probiotics are ready for clinical use.

The researchers evaluated Lacticaseibacillus paracasei LYC1154, Lactiplantibacillus plantarum LYC1322 and Lacticaseibacillus rhamnosus LYC1511. Each strain was successively exposed to sublethal concentrations of ampicillin, minocycline and erythromycin. Antibiotic concentrations were adjusted as minimum inhibitory concentrations increased, and passage-matched wild-type strains served as controls. The resulting variants, identified with the suffix “ame,” showed increased MICs for several drugs, including increases exceeding 1,000-fold for erythromycin and azithromycin in two strains. The study did not test whether resistance persisted after drug withdrawal or examine genomes, so it could not distinguish transient adaptive resistance from stable acquired resistance.

The antimicrobial assay used a co-culture design intended to favor growth of the bacterial pathogens. Each clinical Enterobacterales isolate was added at 10^6 CFU/mL, while lactobacilli were added at 10^7 CFU/mL, producing a 10:1 lactobacilli-to-Enterobacterales ratio. Cultures were incubated at 37 °C under aerobic conditions and sampled after 24 and 48 hours. Parallel cultures containing the pathogen alone established the baseline growth. Recoverable Enterobacterales were measured by plating on MacConkey agar, with cultures containing fewer than 10 CFU/mL classified as below the assay’s limit of detection.

The panel comprised 17 clinical isolates from six species: five Escherichia coli, five Klebsiella pneumoniae, two Citrobacter freundii, one Citrobacter koseri, three Enterobacter cloacae and one Serratia marcescens. It represented five carbapenemase types—KPC-2, NDM, OXA-181, IMP-8 and VIM-1—and included isolates collected from sputum, rectal swabs, blood and urine. At 24 hours, the magnitude of reduction depended on the bacterial species, isolate, lactobacillus strain and whether the strain had undergone resistance induction. Some co-cultures produced counts below 10 CFU/mL, while others showed reductions ranging from more than two to more than four logarithmic units relative to the corresponding pathogen-only controls.

By 48 hours, every combination involving one of the six lactobacilli—three parental strains and three antibiotic-induced variants—had reduced all 17 isolates below 10 CFU/mL. In contrast, the Enterobacterales-only controls maintained viable counts between 10^6 and 10^9 CFU/mL. All three Citrobacter freundii and Citrobacter koseri isolates, all three Enterobacter cloacae isolates and the Serratia marcescens isolate fell below detection at both 24 and 48 hours with every lactobacillus tested. The K. pneumoniae isolates generally reached undetectable counts by 24 hours, while results for E. coli showed greater strain-to-strain variation at that point. These observations establish activity only in the tested laboratory mixture; they do not show that the strains can clear CPE from an infected person or prevent transmission.

The investigators also examined two laboratory proxies for probiotic characteristics: survival under gastrointestinal stress and adhesion to intestinal epithelial cells. Under simulated gastric conditions—PBS at pH 3.0 containing 3 mg/mL pepsin—all six strains remained viable over four hours, with reductions of less than 1.5 log10 relative to their starting counts. The induced variants appeared similar to their parental strains at both the two- and four-hour sampling points. In a separate assay, strains were exposed to 0.3% oxgall to represent physiological bile-salt conditions. Bile tolerance varied substantially by strain rather than showing a consistent advantage or disadvantage after induction.

At 24 hours in the bile assay, viable counts ranged from 5% to 107% of the corresponding bile-free controls for the wild-type strains. L. rhamnosus LYC1511 had the highest value among the parental strains, at 107%, whereas L. plantarum LYC1322 had the lowest, at 5%. The three induced variants ranged from 39% to 68%. Induction increased the relative count for LYC1322ame, from 5% for its parent to 46%, decreased it for LYC1511ame, from 107% to 68%, and produced a similar result for LYC1154ame, at 39% versus 52%. These observations were descriptive rather than inferential, because each assay used only two independent biological replicates.

Adhesion was tested on Caco-2 human colorectal adenocarcinoma cell monolayers. Each wild-type and antibiotic-induced strain was offered bacterial inocula of 10^8 or 10^9 CFU/mL and incubated with the cells for two hours. After non-adherent organisms were washed away and the cells were disrupted, attached lactobacilli were counted. All six strains attached at both inoculum densities, and the induced variants appeared to adhere similarly to their respective parental strains. No obvious loss of epithelial adhesion was therefore detected under these conditions. As with the gastrointestinal assays, the experiment used only two biological replicates, so the findings provide a qualitative comparison rather than a robust estimate of statistical differences.

The results do not establish why the variants continued to suppress CPE. Earlier work by the same research group found that lactic acid was the principal effector of wild-type lactobacilli against carbapenem-resistant Enterobacterales: neutralizing cell-free supernatants to pH 6.5 abolished the activity, whereas catalase and proteinase K treatment did not. The present experiment did not repeat those mechanistic tests on the antibiotic-induced variants. Preserved inhibition therefore does not demonstrate that the same molecular mechanism remained intact; metabolic or transcriptional changes could have altered organic-acid production or introduced other effectors. The possibility that antibiotic induction altered the strains’ physiology was not investigated directly, and no acid production or genomic analysis was reported in the study.

Several design limitations constrain how far the findings can be generalized. The co-cultures used one fixed 10:1 microbial ratio, did not include concurrent antibiotic exposure and did not measure pH. Although the induced variants had been selected with ampicillin, minocycline and erythromycin, these agents are not necessarily those used to treat CPE, and the hypothesis that the variants would remain active during clinical antibiotic therapy therefore remains untested. The isolate panel covered the five major carbapenemase types but included only one or a few isolates for several species and enzyme groups. A single induced lineage was generated for each parental strain, the induced variants were not reconfirmed by molecular species identification, and no animal or human experiments were performed. The work also relied on only two independent biological replicates per assay, limiting the strength of between-strain comparisons and making the results descriptive rather than statistically conclusive.

Safety is an especially important unresolved issue because the study deliberately produced organisms with reduced antibiotic susceptibility. Large MIC increases may reflect stable acquired resistance, yet the absence of reversibility testing, whole-genome sequencing, screening for acquired resistance genes and transfer or conjugation assays leaves the genetic basis and transferability unknown. Those analyses would be essential before the organisms could be considered for administration to patients. The results therefore remain in vitro observations that warrant mechanistic, genomic, animal and clinical investigation. They suggest that reduced susceptibility need not immediately extinguish anti-Enterobacterales activity or the laboratory traits examined, but they do not establish safety, efficacy in people or a clinical strategy for managing CPE colonization or infection.

Subject of Research: In vitro antimicrobial and probiotic characteristics of antibiotic-induced resistant lactobacilli tested against carbapenemase-producing Enterobacterales

Article Title: Antimicrobial activity of antibiotic-induced resistant lactobacilli against carbapenemase-producing Enterobacterales

Article References: Chen, S.-C., Chen, C.-C., Lai, C.-C., & Tang, H.-J. (2026). Antimicrobial activity of antibiotic-induced resistant lactobacilli against carbapenemase-producing Enterobacterales. Frontiers in Microbiology, 17, Article 1931966. https://doi.org/10.3389/fmicb.2026.1931966

Image Credits: AI Generated

DOI: 10.3389/fmicb.2026.1931966

Keywords: antibiotic-induced resistance, lactobacilli, Enterobacterales, carbapenemase, antimicrobial activity, in vitro co-culture, probiotics, gastrointestinal tolerance, Caco-2 adhesion, genomic safety testing, Antimicrobial, activity