In the present study, E. coli K12 BW25113 [genotype, Δ(araD-araB)567, ΔlacZ4787(:rrnB-3), lambda-, rph-1, Δ(rhaD-rhaB)568, hsdR514] was taken from the KEIO collection. A single colony of E. coli K12 BW25113 was picked from the Luria-Bertani (LB) agar plate and cultured in LB medium for 16 h at 37°C. The overnight cultures were diluted 1:100 in fresh LB medium and grew to phase OD₆₀₀ of 0.5 at 37°C. These bacteria were passaged in LB medium with and without gentamicin for gentamicin-resistant strains and control, respectively. The three strains, ancestor strain (K12), gentamicin-resistant strain (K12-R_GEN), and control strain (K12-S), were collected to determine the minimum inhibitory concentration (MIC) by antimicrobial susceptibility testing.

Measurement of MIC was performed as previously described (Kuang et al., 2021). In brief, 160 μg/ml of gentamicin sulfate [Sangon Biotech (Shanghai) Co., Ltd.] was serially double diluted by row in a 96-microwell plate. The overnight culture was diluted at 1:100 into 5 ml LB medium and grew to a phase of 0.5 at OD₆₀₀. A bacterial sample of 5 × 10⁴ CFU was then added to each well. After incubating for 16 h at 37°C, the bacteria growth in each well was recorded. The antibiotic concentration of a well without bacteria growth is the MIC of the tested strain. Data were obtained from three biological replicates.

The overnight cultures were diluted 1:100 in LB medium and grew at 200 rpm at 37°C. Then, the optical density of the culture at OD₆₀₀ was measured every 2 h. The growth curve was drawn using GraphPad Prism version 8.0. At least three biological replicates were performed.

The overnight culture was diluted 1:1000 into tubes with 5 ml LB medium. The tube was added to gentamicin with different concentrations. After growing for 6 h at 200 rpm at 37°C, the optical density of the culture at OD₆₀₀ was measured. The survival percentage was calculated as follows: optical density of the culture with different concentrations of antibiotic divided by that without antibiotic.

Overnight bacterial cultures were collected by centrifugation at 8,000 rpm for 3 min and washed three times with sterile saline. To confirm the drug resistance of K12-R_GEN, precipitates were adjusted to OD₆₀₀ of 0.2 and then diluted 100-fold using a fresh LB medium. K12-S was used as a control. To investigate whether glutamine improved the sensitivity of bacteria to antibiotics, precipitates were adjusted to OD₆₀₀ of 0.2 and then diluted 100-fold using M9 minimal medium with 10 mM NaAc, 2 mM MgSO₄, and 0.1 mM CaCl₂. Each tube was added to 5 ml of diluted bacterial solution in the presence and/or absence of gentamicin and glutamine. After growing for 6 h with 200 rpm at 37°C, 100 μl of cultures were serially 10-fold diluted and 5 μl of cultures were plated onto LB agar. Only the clearly visible colonies were counted and multiplied by the dilution. Percent survival was determined by dividing the CFU obtained from a treated sample by those from the control.

K12 was passaged in LB medium with or without 1/2 minimum inhibitory concentration (MIC) and became K12-R_GEN and K12-S, respectively. MIC of the three strains was measured using a microplate method. The passage led to 32 MIC (40 μg gentamicin) of K12-R_GEN and 1 MIC (1.25 μg gentamicin) of K12-S compared with their parent strain (Figure 1A), suggesting that K12-R_GEN was a gentamicin-resistant strain. To further demonstrate the resistance, survival capability and bactericide assays were performed. The survival capability of the two strains was reduced with increasing gentamicin concentration, but higher survival was detected in K12-R_GEN than in K12-S (Figure 1B). Equally, higher viability was found in K12-R_GEN than in K12-S in the bactericide assay (Figure 1C). Finally, the growth curve showed slower growth in K12-R_GEN than in K12-S (Figure 1D). These results indicate that K12-R_GEN is a gentamicin-resistant strain with a differential-resistant phenotype.

To understand metabolic alterations related to the resistance, a GC-MS-based metabolomics approach was used to characterize the metabolic profile of K12-R_GEN compared with K12-S. Four biological samples with two technical repeats in each group yielded 16 data sets. The correlation coefficient between technical replicates varied between 0.9946 and 0.9995, demonstrating the reproducibility of the data (Figure 2A). A total of 240 aligned individual peaks were obtained from each sample. After the removal of internal standard ribitol and any known artificial peaks, 56 metabolites were identified as shown in Figure 2B. Among them, 33.93%, 26.78%, 17.85%, 12.50%, and 8.93% were categorized as carbohydrates, amino acids, fatty acids, nucleotides, and others, respectively (Figure 2C). These results indicate that K12-R_GEN has a metabolome that is different from that of K12-S.

To gain a differential abundance of metabolites between K12-R_GEN and K12-S, a two-sided Mann–Whitney U-test coupled with a permutation test was utilized. Using the analysis, a total of 43 differential abundance of metabolites were identified in K12-R_GEN (Figure 3A). The Z-value showed the dispersion of data with 18 upregulation and 25 downregulation (Figure 3B). These differential abundances of metabolites were classified into five categories. Among them, 37.21%, 23.26%, 18.60%, 16.28%, and 4.65% belonged to carbohydrates, amino acids, nucleotides, lipids, and others, respectively (Figure 3C). Therefore, a metabolic shift was determined in K12-R_GEN.

A metabolic pathway is a set of biochemical reactions that the cells need to carry out their function. Thus, it is especially important to know the metabolic pathways enriched by these differential abundances of metabolites for understanding gentamicin-mediated metabolic alteration. Metabolic pathway enrichment analysis showed that eight metabolic pathways were enriched. According to the impact, they were ranked from high to low as follows: glycine, serine, and threonine metabolism > alanine, aspartate, and glutamate metabolism > TCA cycle > aminoacyl-tRNA biosynthesis > butanoate metabolism > cyanoamino acid metabolism > biosynthesis of unsaturated fatty acids > nitrogen metabolism (Figure 4A). Integrative analysis showed that among the eight enriched metabolic pathways, all metabolites of alanine, aspartate, and glutamate metabolism and TCA cycle were decreased (Figure 4B). These findings with the above more depressed metabolites than elevated metabolites in the gentamicin-mediated metabolome together suggest that the depressed metabolic pathway plays a key role in the resistance. Meanwhile, alanine, aspartate, and glutamate metabolism fuel the TCA cycle. Thus, alanine, aspartate, and glutamate metabolism can be identified as the most important metabolic pathways.

Biomarker(s) may provide a differential metabolome value and thereby identification of biomarkers is a key step in the analysis of metabolomics. Thus, orthogonal partial least square-discriminate analysis (OPLS-DA) was conducted to recognize the sample pattern. Component t [1] differentiated K12-R_GEN from K12-S and Component t [2] discriminated variation within the two groups (Figure 5A). Discriminating variables were displayed with an S-plot when we set cutoff values as greater or equal to 0.05 and 0.5 for the absolute value of covariance p and correlation p(corr), respectively. Among these metabolites used for the analysis, 11 played more roles than the others in the differentiation and were identified as biomarkers (Figure 5B). The scatter plot showed their differential abundances between K12-R_GEN and K12-S, where only glutamine was depressed in K12-R_GEN (Figure 5C). Glutamine belongs to alanine, aspartate, and glutamate metabolism. Reports have shown that the complementation of crucially depressed metabolites may restore antibiotic-mediated killing efficacy (Peng et al., 2015b; Zhao et al., 2021). Therefore, glutamine as the crucial biomarker may revert the resistance.

To test whether glutamine reverted gentamicin resistance to increase bacterial sensitivity to gentamicin, gentamicin and glutamine were synergistically used to kill K12-R_GEN. Glutamine promoted gentamicin-mediated killing in a dose-dependent manner (Figure 6A). When 20 mM glutamine was used, the killing efficacy was elevated with increasing gentamicin dose (Figure 6B). The killing efficacy was also incubation period-dependent (Figure 6C). Therefore, glutamine-potentiated gentamicin-mediated killing is effective for lab-evolved gentamicin-resistant E. coli. On the other hand, four clinically isolated multidrug-resistant E. coli strains and three clinically isolated multidrug-resistant bacteria were used to test the glutamine-induced potentiation (Figure 6D). Lower survival was detected in the synergistic use of gentamicin and glutamine than in gentamicin alone (Figure 6E). Therefore, the glutamine-potentiated gentamicin-mediated killing is effective for both lab-evolved gentamicin-resistant and clinically isolated multidrug-resistant E. coli. Furthermore, lower survival was also detected in the synergistic use of glutamine and other antibiotics, such as cefoperazone-sulbactam, ofloxacin, and tobramycin than antibiotic alone (Figure 6F).