Resistance characterization and transcriptomic analysis of imipenem-induced drug resistance in Escherichia coli

Induction of imipenem-resistant E. coli

The WT strain K12 MG1655 was subjected to induction with sub-inhibitory concentration drugs, as illustrated in Fig. 1. Subsequently, a strain with a MIC of 16 µg/mL was produced (Fig. 2). Following the bacterial resistance assessment guidelines outlined in the CLSI 2022 version, a strain exhibiting an inhibition zone diameter greater than or equal to 23 mm is classified as drug-resistant and denoted as IPM-R (Table 2 and Fig. S1). The resistance level of the strain increased by over 500 times compared to its pre-induction state. The quality control strain used was E. coli ATCC 25922. However, it is worth noticing that the drug-resistant bacteria induced in this study did not have stable resistance. When cultured in a medium without imipenem, the MIC broke off at the third passage and returned to the same level as wild bacteria.

Comparison of growth curves between wild and induced strains

This study utilized the 1/2 MIC induction method to generate strains with varying minimum inhibitory concentrations, followed by an analysis of their growth curves. The findings indicated that the growth rate of the induced strain was slower compared to the parent strain. Additionally, a correlation was observed between the drug concentration and the inhibitory effect on bacterial growth, as depicted in Fig. 3.

Cross-resistance in imipenem-resistant bacteria

Table 3 displays the drug susceptibility test outcomes for WT and drug-resistant strains. IPM-R strains exhibit cross-resistance to multiple antibiotics, such as β-lactams, lactones, tetracyclines, quinolones, and carbapenems, when compared to WT E. coli. Although antibacterial drugs may not belong to the same class, they can exhibit commonalities in their resistance mechanisms. Certain multidrug efflux systems and specialized barrier structures can facilitate the efflux of multiple antibiotics, resulting in bacterial cross-resistance. Transcriptomic analyses revealed that, in response to the selective pressure exerted by imipenem, E. coli increased the expression of seven transporters by more than 1.5 times. Notably, the ABC-type multidrug transport system AcrB has been confirmed as an active efflux pump contributing to bacterial multidrug resistance, which serves as a defense mechanism for bacteria (Tam et al., 2021). Furthermore, specific strains or communities possess unique barrier structures, such as biofilms and outer membrane proteins, which provide additional protection for bacteria. Structural and functional alterations in the cell wall and cell membrane can also lead to reduced permeability. Consequently, these non-specific resistance mechanisms are significant contributors to cross-resistance, as discussed in ‘Transcriptional Changes in Cell Movement and Communication’.

Swimming and swarming ability of wild and resistant strains

The swimming and swarming abilities of bacteria were assessed by measuring the lawn diameter after 24 h of culturing on semi-solid mediums. Figure 4 displays the culture results of both the strains before and after induction, while Fig. 5 presents the lawn diameter measurements. Comparing the induced drug-resistant bacteria with the parent strains, it was observed that the diameter of the bacterial lawn significantly decreased post-imipenem induction, indicating a reduction in bacterial motility. This is also consistent with the findings presented in ‘Observation of Strain Morphology by Field Emission Scanning Electron Microscopy’. Once bacteria form a biofilm, their mobility is significantly reduced (Tu et al., 2019).

Biofilm forming ability of wild and drug-resistant strains

When bacteria form biofilms, they secrete more extracellular matrix and aggregate. The staining intensity is typically deeper in biofilm-forming strains compared to those that do not form biofilms. Crystal violet staining was used to detect the biofilm formation of different strains over time. Results indicated that the biofilm formation ability of IPM-R was significantly enhanced compared to the wild strain at 24 h, and this ability increased over time (Fig. 6), suggesting that drug-resistant bacteria induced by imipenem form more biofilms to survive in adverse environments.

Observation of strain morphology by field emission scanning electron microscopy

The scanning electron microscopy results depicted in Fig. 7 reveal the cell morphology. The WT strains show cells with a smooth surface, short rod shape, and uniform size (Figs. 7A and 7B). Upon drug induction (Figs. 7C and 7D), the cells exhibited significant swelling, presenting spherical or irregular shapes, some with damaged surfaces displaying wrinkles and defects. The shapes varied in size and irregularity, with many cells aggregating and an increase in the extracellular matrix. It was evident that a majority of the bacteria induced by imipenem were in a state of biofilm formation.

Antibiotic action site mutations and carbapenem hydrolase gene detection in resistant strains

This study investigated the mutations in the PBP1-7 genes and the presence of carbapenem hydrolase genes, only the penicillin adapter protein PBP6b was found to be mutated (Figs. 8–9). PBP6b encodes a protein with carboxypeptidase activity, which is crucial for maintaining the shape of E. coli. Deletion of carboxypeptidase can result in morphological defects (Peters et al., 2016). The functional domain PBP5-C at amino acids 285-376 (i.e., 855–1,128 bp) of PBP6b is important for penicillin affinity. Amino acid substitutions in PBP5-C have been shown to impact β-lactam antibiotic resistance in vitro (Zhou et al., 2015). The sequencing results indicated that the 124th amino acid of PBP6b was mutated from glycine to valine. Additionally, the 350th amino acid was mutated from glutamic acid to aspartic acid, the 351st amino acid was mutated from isoleucine to serine, the 355th amino acid also underwent a mutation from isoleucine to serine, and the 364th amino acid was mutated from proline to serine, these mutations are hypothesized to contribute to resistance to imipenem.

Drug efflux gene knockout and imipenem susceptibility assays

Transcriptome data reveals an up-regulation of putative drug transporters, as shown in Table 4, We constructed gene-deleted strains of E. coli for each of these genes, the electrophoretic patterns of agarose gel for the target gene deletion strains are illustrated in Figs. S2–S5, and the primers used for PCR are shown in Table S6. Sequencing results confirmed that all constructs were successful (Tables S4–S5). The acrB, ybhR, and ybhS gene deletion strains were derived from Feng et al. (2020). Subsequently, we assessed the changes in drug sensitivity of each strain to imipenem (Table 5). The results indicated that the deletion of various drug efflux genes, including acrB, did not significantly affect the minimum inhibitory concentration, suggesting that imipenem resistance is not primarily attributed to drug efflux mechanisms in sexual formation.

Statistics of differentially expressed genes

Using q value ≤ 0.05 and |log2Fold Change | ≥ 1 as screening criteria, a total of 1,260 differentially expressed genes were identified. Among them, 880 genes showed up-regulated expression in drug-resistant strain B compared to the original strain A, while 380 genes displayed down-regulated expression (Fig. 10).

GO, COG, and KEGG classification of differentially expressed genes

To elucidate the functions of differentially expressed genes (DEGs), all unigenes were aligned with entries in the Gene Ontology (GO) database (Fig. 11). The classification outcomes indicate that drug-resistant bacteria exhibit significant differences from WT strains in terms of cell structure, substance metabolism, and transport (Figs. S6–S9). Figure 12 displays the top 20 most significantly enriched COG categories, with a focus on amino acid transport and metabolism, defense mechanisms, inorganic ion transport and metabolism, energy production, and conversion, translation, and ribosome structure and biosynthesis (Figs. S10–S11). Figure 13 displays only the most significantly enriched genes. The analysis of the top 30 KEGG pathways revealed a high concentration of genes in the ABC transport system (ko02010), amino acid biosynthesis (ko01230), and ribosome (ko03010) pathways, which were found to be significantly enriched (Fig. S12).

Transcriptional changes to nutrient uptake and metabolism

Under antibiotic stress, distinct transcriptional responses were observed in the nutrient assimilation pathways for carbon, sulfur, and nitrogen sources. In carbon metabolism, both conventional and unconventional carbohydrates (galactose, fructose, mannose, starch, and sucrose) are upregulated to varying extents (Figs. S13–S14). These findings collectively support the upregulation of metabolic pathways. In sulfur metabolism, the assimilation of sulfate reduced to hydrogen sulfide is upregulated (Fig. S15), in line with the upregulation of genes encoding ABC sulfate and thiosulfate transporters (Fig. S34), indicating global assimilation of sulfur that was upregulated. Furthermore, the sulfur transfer system was also upregulated (Fig. S16). Nitrogen metabolism, including nitrate and nitrite, exhibited an upregulation (Fig. S17).

The results align with changes in gene expression of the KEGG pathway ABC transporter (Eco02010) and phosphotransferase system (Pts) transporter (Eco02060), along with genes associated with transporter activity gene ontology term (GO 0005215). Overall, it is suggested that upon induction, there is an increase in the intake of nitrogen, sulfur, and phosphorus sources, a decrease in sugar intake, and an increase in the intake of amino acids and other carbon sources to supplement carbon intake. The degree of upregulation of efflux proteins following the ingestion of peptides (>5.0) was significantly higher than that observed for other nutrients (1.0–2.0).

Transcriptomic changes in energy-producing metabolic pathways

Polysaccharides, particularly glycogen, can function as energy reserves for E. coli. In response to antibiotic stress, the metabolic pathways of glycogen are transcriptionally upregulated, similar to other carbohydrate pathways (Fig. S18). Furthermore, the PTS system, a multi-protein phosphorylation system located in the cell membrane, integrates carbohydrate transport and phosphorylation. Genes encoding transporters for β-glucoside, N-acetylmuramic acid, galactose, and fructose were found to be up-regulated in the PTS system, while no down-regulation of genes was observed (Fig. S21).

In the glycolytic pathway from glucose to pyruvate, the enzymes ptkA (6-phosphofructokinase) and pykF (pyruvate kinase) were down-regulated about 1.0 to 1.5 times, both being rate-limiting enzymes. Subsequently, the TCA cycle showed a general up-regulation (Fig. S19). Within the pentose phosphate pathway, the key genes zwf, pgl, and gnd in the first stage from glucose-6 phosphate to ribose-5 phosphate remained unchanged. In the second stage, the genes encoding transaldolase (talA, talB) and transketolase were unchanged, while tktA and tktB were up-regulated (Fig. S20). Beta-oxidation and the acyl-CoA cycle were significantly upregulated more than 2.0 times during fatty acid degradation (Fig. S22), leading to the production of a large amount of acetyl-CoA for entry into the tricarboxylic acid cycle, generating substantial energy. Both sugars and non-sugar carbohydrates ultimately enter the TCA cycle for energy production. The TCA cycle exhibited significant up-regulation (2.0–4.0) after antibiotic induction (Fig. S23), suggesting that post-induction, the bacteria generated more energy, possibly shifting from glucose-derived energy to utilizing other carbon sources for energy production.

Transcriptional changes in biosynthesis

Under antibiotic stress, significant changes have been observed in the biosynthetic pathway of amino acids, leading to an up-regulation of amino acid biosynthesis (Figs. S24–S29). Among the 21 common amino acids, 13 species exhibit clear metabolic/biosynthetic regulation patterns. Notably, the biosynthetic pathways of histidine, tryptophan, valine, leucine, and isoleucine were significantly up-regulated. Additionally, alanine, aspartic acid, glutamic acid, arginine, proline, glycine, serine, threonine, histidine, cysteine, and methionine showed varying degrees of up-regulation. This upregulation aligns with the increased expression of genes encoding amino acid transporters in the ABC system (Fig. S34).

In protein biosynthesis, ribosome-encoding genes are typically down-regulated about 1.5 times (Fig. S30). Imipenem treatment leads to the down-regulation of nucleoside triphosphate (NTP) and their precursors biosynthesis, while deoxynucleoside triphosphates (dNTPs) show no significant effect. Notably, the dnaE gene encoding DNA polymerase is up-regulated 1.37 times, whereas rpoA encoding the α subunit of RNA polymerase is down-regulated 1.1 times (Fig. S31). This suggests that drug-resistant bacteria down-regulate intracellular transcription reactions and protein synthesis processes while increasing amino acid synthesis to meet energy needs or cell wall biosynthesis and other growth essentials.

Transcriptional changes in cell membrane components

In the context of cell wall biosynthesis, lipopolysaccharide synthesis remained relatively stable (Fig. S32). Following exposure to the cell wall inhibitor imipenem, there was a notable increase in DAP-type epidermal polysaccharide synthesis within peptidoglycan synthesis, with an overall significant up-regulation observed. Imipenem specifically targets the penicillin-binding protein PBP2 which is upregulated 1.12 times (Fig. S33). The peptidoglycan molecule serves as a key constituent of the bacterial cell wall, characterized by a network of N-acetylglucosamine and N-acetylmuramic acid linked by β-1,4 glycosidic bonds and short peptide tails. The presence of E. coli PBP2 plays a crucial role in maintaining the rod-shaped structure of the bacteria, as its reduction or absence can lead to spherical transformation, ultimately resulting in bacterial dissolution and death. Consequently, it is postulated that there is an elevation in the synthesis of essential cell wall materials required for bacterial survival.

Transcriptional changes in membrane transporter proteins

Genes within the KEGG pathway ABC transporter (Eco02010) and phosphotransferase system (Pts) transporter (Eco02060), along with genes associated with transporter activity gene ontology term (GO 0005215), were examined.

In the ABC transport system (Fig. S34), the uptake of sugars such as maltose, sorbitol, mannitol, ribose, and xylose are down-regulated (1.0–1.5). Conversely, the uptake of other carbon sources like L-arabinose, methyl galactopyranoside, and glycerophosphate is upregulated (>2.0). Additionally, the uptake of sulfur sources such as sulfates, sulfonates, and thiamine is upregulated (>2.0), while the uptake of nitrogen sources like putrescine and phosphorus sources like phosphate is up-regulated (>2.0). Moreover, the uptake of amino acids such as arginine, glutamic acid, aspartic acid, cysteine, methionine, and branched-chain amino acids is up-regulated (1.0–1.5), and the intake of peptides like oligopeptides, secondborn, cationic peptides, and glutathione is upregulated (>2.0). Furthermore, the intake of trace elements like nickel ions is increased (>2.0), while zinc ions are decreased by 1.56 times. In the phosphatase transfer system (Fig. S35), the transfer of β-glucoside, N-acetylmuramic acid, galactose, fructose, and nitrogen is up-regulated less than 2.0 times, with β-glucoside and N-acetylmuramic acid being important substances for synthetic peptide polypeptides. Additionally, the cell membrane contains numerous transporters that aid in removing harmful substances from the cell.

Transcriptional changes in cell movement and communication

Cell motility, communication, flagellar assembly, biofilm formation, quorum sensing, and bacterial chemotaxis are common stress responses in bacteria. Flagellum assembly remained unchanged (Fig. S36). In the quorum sensing pathway, the upregulation of luxS leads to an upregulation of the AI-2 transport system (Figs. S37–S38). Among them, the luxS gene was upregulated 1.23 times, while the AI-2 transport system exhibited an upregulation of more than 2.0 times. During bacterial chemotaxis (Fig. S39), Tsr and Trg, which are methyl-accepting chemotactic proteins, were up-regulated about 1.0 times. Along with the substrate-binding proteins MglB (for the methylgalactoside transport system) and DppA (for the dipeptide transport system), were up-regulated 3.0 times.