Metabolic phenotyping of acquired ampicillin resistance using microbial volatiles from Escherichia coli cultures

Aims This study sought to assess the volatile organic compound (VOC) profiles of ampicillin-resistant and -susceptible Escherichia coli to evaluate whether VOC analysis may be utilized to identify resistant phenotypes. Methods and Results An E. coli BL21 (DE3) strain and its pET16b plasmid transformed ampicillin-resistant counterpart were cultured for 6 h in drug-free, low- and high-concentrations of ampicillin. Headspace analysis was undertaken using thermal desorption-gas chromatography-mass spectrometry. Results revealed distinct VOC profiles with ampicillin-resistant bacteria distinguishable from their susceptible counterparts using as few as six compounds. A minimum of 30 compounds (fold change >2, p <= 0.05) were differentially expressed between the strains across all set-ups. Furthermore, three compounds (indole, acetoin and 3-methyl-1-butanol) were observed to be significantly more abundant (fold change >2, p <= 0.05) in the resistant strain compared to the susceptible strain both in the presence and in the absence of drug stress. Conclusions Results indicate that E. coli with acquired ampicillin resistance exhibit an altered VOC profile compared to their susceptible counterpart both in the presence and in the absence of antibiotic stress. This suggests that there are fundamental differences between the metabolisms of ampicillin-resistant and -susceptible E. coli which may be detected by means of VOC analysis. Significance and Impact of the Study Our findings suggest that VOC profiles may be utilized to differentiate between resistant and susceptible bacteria using just six compounds. Consequently, the development of machine-learning models using VOC signatures shows considerable diagnostic applicability for the rapid and accurate detection of antimicrobial resistance.

Automated summary

This study aimed to assess the volatile organic compound (VOC) profiles of ampicillin-resistant and -susceptible Escherichia coli and found that resistant strains exhibited distinct VOC profiles compared to susceptible strains, indicating fundamental differences in their metabolisms. The results suggest that machine-learning models using VOC signatures may have diagnostic applicability for the rapid detection of antimicrobial resistance.