Compensation of the metabolic costs of antibiotic resistance by physiological adaptation in Escherichia coli

Abstract

Antibiotic resistance is often associated with metabolic costs. To investigate the metabolic consequences of antibiotic resistance, the genomic and transcriptomic profiles of an amoxicillin-resistant Escherichia coli strain and the wild type it was derived from were compared. A total of 125 amino acid substitutions and 7 mutations that were located <1,000 bp upstream of differentially expressed genes were found in resistant cells. However, broad induction and suppression of genes were observed when comparing the expression profiles of resistant and wild-type cells. Expression of genes involved in cell wall maintenance, DNA metabolic processes, cellular stress response, and respiration was most affected in resistant cells regardless of the absence or presence of amoxicillin. The SOS response was downregulated in resistant cells. The physiological effect of the acquisition of amoxicillin resistance in cells grown in chemostat cultures consisted of an initial increase in glucose consumption that was followed by an adaptation process. Furthermore, no difference in maintenance energy was observed between resistant and sensitive cells. In accordance with the transcriptomic profile, exposure of resistant cells to amoxicillin resulted in reduced salt and pH tolerance. Taken together, the results demonstrate that the acquisition of antibiotic resistance in E. coli is accompanied by specifically reorganized metabolic networks in order to circumvent metabolic costs. The overall effect of the acquisition of resistance consists not so much of an extra energy requirement, but more a reduced ecological range.

Fig 1

Total number of up- and downregulated genes in amoxicillin-resistant cells compared to the wild type in the absence (−AMX) or presence (+AMX; wild type, 1 μg/ml; resistant cells, 150 μg/ml) of amoxicillin. Selected differentially expressed genes for each condition examined are presented, with the fold change in parentheses. Genes are considered to be differentially expressed when the expression ratio exceeds a factor of 2 and shows a significantly different (95% confidence) log expression ratio greater than or equal to ±0.5.

Fig 2

Overlap of upregulated genes categorized into the cellular respiration, transport, and membrane groups of amoxicillin-resistant E. coli cells in the presence of amoxicillin compared to the wild type (0.25× MIC; wild type, 1 μg/ml; resistant cells, 150 μg/ml amoxicillin). Genes are considered to be differentially expressed when the expression ratio exceeds a factor of 2 and shows a significantly different (95% confidence) log expression ratio greater than or equal to ±0.5.

Fig 3

Specific glucose consumption (q_gluc) of WT and AR E. coli strains in continuous culture at dilution rates of 0.2 h⁻¹ (a) and 0.4 h⁻¹ (b). After cells reached steady state (t = 0), WT E. coli and the AR strain were exposed to 1 and 150 μg/ml amoxicillin, respectively.

Fig 4

Maintenance energies of WT and AR E. coli (a); methicillin-sensitive S. aureus (MSSA0027 and MSSA0029) and methicillin-resistant S. aureus (MRSA0026 and MRSA0029) (b); and E. faecium E1039 (ampicillin and vancomycin sensitive), E1162 (ampicillin resistant, vancomycin sensitive), and E155 (ampicillin and vancomycin resistant) (c). By measuring the q_gluc value as a function of the dilution rate (D) in steady-state chemostat cultures and extrapolating by linear regression to a D value of 0 h⁻¹, the maintenance energy was estimated.

Fig 5

Maximal specific growth rates (μ_max) of WT and AR E. coli cultured at different pH values with or without subinhibitory concentrations of amoxicillin of 2 and 256 μg/ml, respectively. The results were obtained by calculating the μ_max based on averaged OD₆₀₀ values of 2 independent replicates. The significance (P ≤ 0.05) of the difference in μ_max was determined by Student's t test. The error bars indicate standard deviations.

Fig 6

μ_max values of WT and AR E. coli with increasing sodium chloride concentrations with or without subinhibitory concentrations of amoxicillin of 2 and 256 μg/ml, respectively. These were the highest concentrations that allowed growth, though at reduced rates. The results were obtained by calculating μ_max based on averaged OD₆₀₀ values of 2 independent replicates. The significance (P ≤ 0.05) of the difference in μ_max was determined by Student's t test. The error bars indicate standard deviations.

Fig 7

Intracellular levels of reactive oxygen species in WT and AR E. coli cells as measured with 100 μM H₂DCFDA. Bacteria grown for 3 h to an OD₆₀₀ of approximately 0.3 were incubated for 1 h with or without amoxicillin (untreated) or 10 μM H₂O₂ (pos. control). Sub-MIC, 1 and 150 μg/ml amoxicillin for WT and AR, respectively. These were the highest concentrations at which dying cells were not observed in the culture. MICs, 4 and 512 μg/ml amoxicillin for WT and AR. The results are presented as the means and standard deviations from 2 independent measurements.

Fig 8

Schematic model summarizing the main metabolic consequences of amoxicillin resistance in E. coli. In drug-exposed resistant cells, gene expression of alternative electron acceptors (frdABCD, narGHJI, and dmsABC) is induced, indicating a partial switch in metabolism from aerobic to anaerobic. Depletion of NADH may counter the elevated NADH-dependent superoxide production via the electron transport chain that was proposed by Kohanski and coworkers as a common mechanism of cell death induced by bactericidal antibiotics (20). Metabolic changes in amoxicillin-resistant cells include a suppressed SOS response compared to sensitive cells, regardless of the presence or absence of amoxicillin. Resistance is further enhanced by a mutation in the promoter region of ampC, resulting in increased expression of the β-lactamase.