A shared mechanism of multidrug resistance in laboratory-evolved uropathogenic Escherichia coli

ABSTRACT

The emergence of multidrug-resistant bacteria poses a significant threat to human health, necessitating a comprehensive understanding of their underlying mechanisms. Uropathogenic Escherichia coli (UPEC), the primary causative agent of urinary tract infections, is frequently associated with multidrug resistance and recurrent infections. To elucidate the mechanism of resistance of UPEC to beta-lactam antibiotics, we generated ampicillin-resistant UPEC strains through continuous exposure to low and high levels of ampicillin in the laboratory, referred to as Low Amp^R and High Amp^R, respectively. Whole-genome sequencing revealed that both Low and High Amp^R strains contained mutations in the marR, acrR, and envZ genes. The High Amp^R strain exhibited a single additional mutation in the nlpD gene. Using protein modeling and qRT-PCR analyses, we validated the contributions of each mutation in the identified genes to antibiotic resistance in the Amp^R strains, including a decrease in membrane permeability, increased expression of multidrug efflux pump, and inhibition of cell lysis. Furthermore, the Amp^R strain does not decrease the bacterial burden in the mouse bladder even after continuous antibiotic treatment in vivo, implicating the increasing difficulty in treating host infections caused by the Amp^R strain. Interestingly, ampicillin-induced mutations also result in multidrug resistance in UPEC, suggesting a common mechanism by which bacteria acquire cross-resistance to other classes of antibiotics.

KEYWORDS:

• Antibiotic resistance
• uropathogenic Escherichia coli
• efflux pump
• adhesion
• type 1 fimbriae

Introduction

Overuse or misuse of antibiotics has led to the development of antibiotic resistance in bacteria, which has become a global crisis. Therefore, understanding the underlying resistance mechanisms is critical for addressing the current challenges in antibiotic resistance. Antibiotic resistance can arise through genetic mutations or the acquisition of resistance genes from other bacteria [1]. Bacteria become resistant to antibiotics through various mechanisms, including limiting access to antibiotics, modifying antibiotic target sites, degrading antibiotics, or promoting antibiotic efflux [2]. When bacteria develop resistance to one type of antibiotic, they often exhibit resistance to other types of antibiotics. Thus, there is a need to understand the underlying mechanisms that drive multidrug antibiotic resistance.

Uropathogenic Escherichia coli (UPEC) is a bacterium that commonly causes urinary tract infections (UTIs). UTIs are among the most common bacterial infections in humans and affect millions of people worldwide [3,4]. In the host, UPEC has developed specific adaptations that enable it to colonize and infect the urinary tract, leading to symptoms such as pain, frequent urination, and fever. During colonization of epithelial cells, UPEC employs type 1 fimbriae to establish infection in the urinary tract. A key player in this process is FimH, an adhesin protein located at the tip of type 1 fimbriae that enables UPEC to recognize and attach to host cells [5]. UTIs are classified according to their symptoms, ranging from urosepsis syndrome (the most severe form) to pyelonephritis (kidney infection) and cystitis (bladder infection) [6]. E. coli CFT073, isolated from a patient with pyelonephritis [7], has been widely used as a model strain to study the virulence mechanisms and pathogenesis of UPEC infections.

A significant challenge in the treatment of UPEC infections is the increasing prevalence of antibiotic resistance. UPEC strains have the ability to acquire resistance to multiple antibiotics, making treatment options more limited and challenging. The treatment of UTIs relies on the use of antibiotics such as β-lactams, fluoroquinolones, and trimethoprim-sulfamethoxazole (TMP-SMZ), which are selected based on the resistance profiles of UPEC [8]. The prevalence of antibiotic resistance in UPEC strains has been increasing globally over the years, and these antibiotic-resistant strains are a significant concern for human health [9]. Recently, UPEC strains isolated from pregnant women with a history of recurrent UTIs were found to be resistant to ampicillin, tetracycline, amikacin, ciprofloxacin, and gentamicin [10]. This highlights the urgent need for effective strategies to address antibiotic resistance in UTIs.

Here, we generated beta-lactam antibiotic-resistant UPEC strains isolated in 25 µg/ml or 50 µg/ml ampicillin (Low Amp^R and High Amp^R respectively) in the laboratory to investigate the underlying mechanisms of antibiotic resistance of UPEC. Using whole-genome sequencing (WGS), we identified various point mutations, including missense and nonsense mutations, within the Amp^R UPEC strains. The ampicillin-resistant UPEC strains harbor mutations in the marR, acrR, envZ, and nlpD genes, all of which lead to ampicillin resistance. By investigating the molecular genetic processes underlying UPEC resistance to ampicillin, we aimed to understand how these mechanisms contribute to multidrug resistance.

Materials and methods

Bacteria strains, plasmids, oligodeoxynucleotides, and growth conditions

The bacterial strains and plasmids used in this study are listed in Table S2. All uropathogenic Escherichia coli (UPEC) CFT073 strains were derived from the wild-type strain CFT073 [7]. Bacteria were grown at 37°C in Luria-Bertani (LB) broth and MacConkey agar. For ampicillin-resistant bacteria, 25 µg/ml ampicillin was used in LB broth and 50 µg/ml ampicillin was used on LB solid medium unless otherwise indicated. For measuring antibiotic induced mRNA expression, ampicillin was used at 4 or 8 µg/ml, and ciprofloxacin was used at 0.06 or 0.12 µg/ml.

Isolation of ampicillin-resistant strains

Escherichia coli (UPEC) CFT073 was cultured in LB broth medium. After 18 h of culture, cells were spread on LB agar medium containing 12.5 µg/ml ampicillin. Plates were incubated for 2–7 days until resistant colonies started to grow on the ampicillin-containing plate. Resistant colonies were purified on LB agar medium containing 12.5 µg/ml ampicillin and inoculated in LB broth medium with the same concentration of ampicillin. This process was repeated with increasing ampicillin concentrations up to 25 µg/ml (Low Amp^R) or 50 µg/ml (High Amp^R).

Determination of minimal inhibitory concentration (MIC)

In vitro antimicrobial susceptibility testing was performed by measuring the minimal inhibitory concentration (MIC) using the broth microdilution method [11]. Briefly, the strains were cultured in Mueller-Hinton (MH) broth. Overnight-grown bacteria were diluted in broth so that each tube contains approximately 5 × 10⁵ colony-forming units per ml (CFU/ml) (OD₆₀₀ = 0.005) and dispensed at 100 µl per well in a 96-well plate. Ampicillin was dissolved and then diluted in MH medium to a 2 × top concentration desired in the test. Using a multi-pipette, ampicillin was mixed into the wells of column 1 by sucking up and down 6–8 times. Fifty microliter of sample from column 1 was withdrawn and added to column 2. We repeated the procedure down to column 10. After incubating the plate overnight at 37°C, resazurin (0.015%) [12] was added to all wells (30 μl per well) to monitor bacterial growth and further incubated for 2–4 h for the observation of color change. Pink color indicates bacterial growth and blue color indicates inhibition of bacterial growth.

Real-time quantitative reverse transcription PCR (qRT-PCR)

Total RNA was isolated using the RNeasy kit (Qiagen) according to the manufacturer’s instructions. The purified RNA was quantified using a NanoDrop spectrophotometer (Thermo Fisher). Five hundred nanogram of cDNA was synthesized using a PrimeScript RT reagent kit (TaKaRa). Transcripts were quantified by real-time PCR using the SYBR green PCR master mix (Applied Biosystems) on a StepOnePlus real-time PCR system (Applied Biosystems) with the following PCR protocol: 95°C for 20 sec followed by 40 cycles of 95°C for 2 sec and 60°C for 30 sec. The primers used for the detection of transcripts of each gene are listed in Table S2. The mRNA levels of each target gene were calculated using a standard curve of UPEC CFT073 genomic DNA at known concentrations, and the data were normalized to 16S rRNA levels of the rrsH gene. All real-time qRT-PCR results were derived from three biological replicates.

Whole genome sequencing

Genomic DNA from each strain was extracted using a G-spin^TM Genomic DNA extraction Kit (iNtRON). Sixty μg/μl of genomic DNA was processed to construct a sequencing library using a TruSeq Nano DNA Sample Prep Kit (#20015964, Illumina, USA). Sequencing was performed using the NovaSeq 6000 platform (Illumina, USA) on a 151 bp paired-end platform. We obtained an average of 5.62 Gb of data (BioProject accession number PRJNA987582).

The genome of the reference strain, uropathogenic E. coli CFT073 was assembled using SPAdes v3.14.1 [13] and annotated using Prokka [14]. The generated raw reads were adapter- and quality-trimmed using Trimmomatic v.0.36 [15]. Cleaned reads were mapped to the reference genome using CLC genomics workbench 21 (Qiagen) and showed an average 99.45% mapping rate, which can cover approximately over 1,000 × depth per sample. Variant calling and comparison were performed using the Basic Variant Detection tool in the CLC genomics workbench 21 (Qiagen).

Transmission electron microscopy (TEM)

Bacterial samples for TEM were prepared as described previously with some modifications [16,17]. Briefly, bacteria were cultured in LB broth supplemented with 0.1% glucose to promote type 1 fimbriae expression overnight. The bacterial samples were loaded onto a carbon-coated grid for 2 min and washed with distilled water. The remaining water was removed using filter paper. Then, the type 1 fimbriae were negatively stained with 3% of uranyl acetate (UA) or 1% of phosphotungstic acid (PTA) for 10 and 1–2 sec, respectively. After staining, samples were rinsed twice with distilled water and dried at room temperature. Finally, the samples were visualized by TEM (80 kV, JEOL, JEM1010).

Determination of UPEC colony-forming units (CFUs) in the mouse bladder

To assess bacterial loads in the mouse bladder, this study employed the C57BL/6J mouse strain, which is used to investigate bacterial cystitis [18,19]. C57BL/6J female mice aged eight weeks were used for in vivo experiments. Mice were anesthetized with an intraperitoneal injection of pentobarbital sodium (Hanlim Pharmaceuticals). To investigate the persistence of UPEC during antibiotic treatment, the anesthetized mice were transurethrally inoculated with 1 × 10⁸ UPEC using the designated UPEC strains. Subsequently, the mice were administered ampicillin (200 mg/kg) through oral gavage and ciprofloxacin (10 mg/kg) through subcutaneous injection twice per day at 8-hr intervals over a period of three days. One day after the last treatment, the bladders were harvested and homogenized in 0.1% Triton X-100 PBS solution using a bead-beating homogenizer. The UPEC CFUs in the homogenized bladders were determined by plating the appropriately diluted homogenate on MacConkey agar plates that were incubated overnight at 37°C.

Ethic statement

The experimental procedures involving animals were conducted in accordance with the guidelines set forth by the Korea University Institutional Animal Care & Use Committee (KUIACUC-2022–0100), receiving approval on 26 December 2022. In this study, we complied with the ARRIVE 2.0 guidelines for animal experiments.

Determination of UPEC CFUs in the 5637 human bladder cell line

5 × 10⁴ cells per well of human 5637 bladder epithelial cells (ATCC, HTB-9) were seeded in a 24 well plate. After overnight incubation, the cells were treated with media containing 1 × 10⁷ CFU of wild-type or High Amp^R CFT073 bacteria (corresponding to a multiplicity of infection (MOI) of 200). The whole plate was centrifuged for 5 min at 1200 rpm. After 1 h of incubation, the medium was changed to RPMI-1640 containing 200 µg/ml gentamicin to kill extracellular bacteria (control), 100 µg/ml ampicillin, or 10 µg/ml ciprofloxacin. After 30 min, the medium was changed to RPMI-1640 containing 50 µg/ml gentamicin or the same amount of ampicillin or ciprofloxacin. The cells were then incubated until 24 hr of initial infection. Then, media were removed, and 500 µl of 0.1% triton X-100 in PBS was added into each well. Using a pipette tip, wells were scraped. The extracted solution was plated on MacConkey agar, and the CFU was calculated accordingly.

Quantification and statistical analysis

The data underwent analysis, and several graphs were generated utilizing Prism software (GraphPad Software). Statistical significance was determined by considering p values less than 0.05. In the graphs representing in vitro experiments, each dot represents samples obtained from two to three independent experiments. In the graphs depicting mouse experiments, the number of mice utilized to obtain bladder samples is indicated by the dots. Statistical significance was assessed using either the unpaired Student’s t-test (Figure 6d,e,f) or the one-way ANOVA with Tukey’s multiple comparisons test (Figure 6b,c). Each dot on the graphs representing mouse experiments indicates the number of mice used to obtain bladder samples.

Results

Generating mutations in UPEC strain to confer resistance to ampicillin

To study antibiotic resistance generated by continuous exposure to ampicillin, we incubated E. coli CFT073 with increasing concentrations of ampicillin and repeatedly isolated colonies that had developed resistance to ampicillin (Figure 1a). Using this process, we isolated a strain named Low Amp^R, which displayed resistance to 25 μg/ml ampicillin. Taking this a step further, we treated Low Amp^R to 50 μg/ml ampicillin and isolated a strain with even higher resistance, which we named High Amp^R.

Figure 1. The ampicillin-resistant UPEC strains generated in the laboratory.

(a) Schematic diagram of the generation of resistant strains. (b) MIC assay was performed for ampicillin using 96-well plates with the wild-type UPEC, Low Amp^R, and High Amp^R strains. After incubation, 0.015% resazurin dye was added. MIC assays were performed duplicate for each strain. (c) The wild-type UPEC, Low Amp^R, and High Amp^R strains streaked on LB plate containing 50 μg/ml ampicillin. (d-i) The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB liquid media with different concentrations of ampicillin. (j) Location of mutations in the Low and High ampicillin-resistant UPEC strains. These mutations were found by whole-genome sequencing and filtered for nucleotide changes that resulted in amino acid changes. Specifically, mutations with a reference position value >1000 and forward/reverse balance >0.4. * is the stop codon.

To evaluate the levels of resistance in the generated strains, we performed a minimal inhibitory concentration (MIC) assay for ampicillin on the wild-type UPEC, Low Amp^R, and High Amp^R strains, which were visualized by adding 0.015% resazurin. The MIC of the wild-type strains was 8 μg/ml, whereas the Low Amp^R and High Amp^R strains exhibited 4-fold and 8-fold increases in resistance, respectively (32 and 64 μg/ml) (Figure 1b). We further examined their resistance by culturing the wild-type and Amp^R strains on LB ampicillin agar plates (50 μg/ml). The wild-type strain did not exhibit any growth at this concentration, whereas the Low Amp^R strain formed pale colonies. In contrast, the High Amp^R strain formed colonies of similar size and shape as the wild-type strain grown on LB plates without ampicillin (Figure 1c).

We compared the growth curves of Amp^R strains with those of the wild-type strain in liquid media. In LB medium without ampicillin, the mutants displayed slightly lower growth than the wild-type (Figure 1d). However, in the presence of 4 and 8 μg/ml ampicillin, the Amp^R strains exhibited robust growth, similar to that observed in the absence of ampicillin. Under these conditions, the wild-type strain did not grow, similar to what was observed on LB solid medium (Figure 1e,f). Additionally, the High Amp^R strain showed greater resistance than the Low Amp^R strain at 16 μg/ml ampicillin in LB media (Figure 1g). Above 32 μg/ml, none of the strains exhibited significant growth (Figure 1h,i).

Identification of mutations leading to ampicillin resistance in the low and high Amp^R strains

To identify the genetic mutations responsible for ampicillin resistance, we performed whole-genome sequencing (WGS) of Amp^R strains. The analysis revealed specific mutations in the acrR, marR, and envZ genes in the Low Amp^R strain (Figure 1j). In acrR, a G-to-T transversion mutation led to a nonsense mutation, causing a change from glutamate to a stop codon (GAA-TAA) at position 122. This results in the production of truncated AcrR proteins. Similarly, the marR gene exhibited a point mutation, with a change from valine to phenylalanine (GTC-TTC) at position 66. The envZ mutation involved a change from threonine to methionine (ACG-ATG) at position 402.

Given that the High Amp^R strain was derived from the Low Amp^R strain, it retained all the mutations found in the Low Amp^R strain. However, it also harbors an additional C to T transition nonsense mutation in the nlpD gene. This mutation resulted in a change from glutamine to a stop codon (CAA-TAA) at position 187 of NlpD. These results suggest that the identified mutations in the Amp^R strains could be responsible for conferring resistance to ampicillin.

Mutation in MarR of ampicillin-resistant UPEC leads to constitutive expression of MarA and MarB

MarR is a helix-turn-helix (HTH) type repressor of the marRAB operon that is involved in resistance to multiple antibiotics [20]. Among the marRAB operon, MarA acts as a global regulatory protein that controls various genes associated with antibiotic resistance, including the acrAB operon [21]. Additionally, MarA serves as an autoregulator that activates the marRAB operon [22]. MarB is a periplasmic protein with unknown function.

The mutation site observed in MarR (V66F) in the Amp^R strains is located within the DNA binding domain. This substitution introduced an aromatic ring structure of phenylalanine, which was expected to reduce the affinity of MarR for DNA, rendering it unable to repress the marRAB operon (Figure 2a–c). To test this hypothesis, we performed qRT-PCR to measure the expression levels of the MarR-regulated genes marA and marB. In the wild-type strain, the mRNA levels of marA and marB increased in a dose-dependent manner with increasing ampicillin concentrations. However, in the Amp^R strains, the expression of marA and marB genes was already significantly elevated even without ampicillin treatment, and the addition of ampicillin did not further increase their expression levels (Figure 2d,e). This suggests that the MarR V66F mutation abolished repression of the marRAB operon, leading to constitutive expression of MarA and MarB (Figure 2c).

Figure 2. The transcription of the marA and marB genes is up-regulated by the mutation of MarR in the ampicillin-resistant UPEC strains.

(a) The mutation of MarR has an amino acid change, from valine to phenylalanine, at the 66^th amino acid. (b) The modeled structure of MarR and its DNA binding domain region. The region including the mutation site is considered essential for DNA binding, and the substitution of valine with phenylalanine can weaken the DNA binding by adding an aromatic ring structure. The 3D modeled structure of the mutated proteins was generated by Alphafold 2.0. (c) Diagram illustrating the potential regulatory mechanism in the marRAB operon comparing the wild-type MarR and MarR V66F variant. (d and e) The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB with ampicillin (+; 4 μg/ml, ++; 8 μg/ml). The levels of marA and marB mRNA levels were determined by qRT-PCR. The mean values and standard deviations of three independent experiments are shown.

Mutation in AcrR of ampicillin-resistant UPEC leads to constitutive expression of AcrA and AcrB

AcrR is an additional global transcriptional regulator that acts as a repressor of the acrAB operon [23] and is controlled by MarA [21]. Together with TolC, AcrA and AcrB constitute the AcrAB-TolC multidrug efflux pump, which plays a crucial role in multidrug resistance [24–26]. The Amp^R strains harbored a nonsense mutation in the acrR gene, leading to the translation of a truncated AcrR protein (AcrR^1–121) lacking dimerization and ligand-binding domain (Figure 3a,b). As a result, truncated AcrR^1–121 was unable to repress the acrAB operon. This led to the expression of AcrA and AcrB, which form the AcrAB-TolC multidrug efflux pump (Figure 3c).

Figure 3. The transcription of the acrA and acrB genes is up-regulated due to the mutation of AcrR in the ampicillin-resistant UPEC strains.

(a) The mutation of AcrR is truncated at the 122^nd amino acid site, which lacks dimerization domain. (b) The 3D modeled structure of AcrR dimer. The truncated dimerization domain colored in light gray. (c) Diagram illustrating the potential regulatory mechanism in the acrR and acrAB operon comparing the wild-type AcrR and AcrR 1–121 variants. (d and e) The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB medium with ampicillin (+; 4 μg/ml, ++; 8 μg/ml). The mRNA levels of acrA and acrB were determined by qRT-PCR. The mean values and standard deviations of three independent experiments are shown.

To test this hypothesis, we performed qRT-PCR to measure the mRNA levels of acrA and acrB, which are regulated by AcrR. Unlike marA and marB, there was no significant change in the mRNA levels of acrA and acrB observed in the wild-type, even when treated with ampicillin. However, the Low Amp^R strain exhibited upregulated acrA and acrB genes regardless of the presence of ampicillin. Although the High Amp^R strain showed slightly lower mRNA levels of acrA and acrB genes compared to the Low Amp^R strain, they still maintained higher expression levels irrespective of the presence of ampicillin compared to the wild-type strain (Figure 3d,e). Therefore, the truncated AcrR protein was unable to repress the acrAB operon, leading to the constitutive expression of AcrA and AcrB. Elevated levels of these two proteins form the AcrAB-TolC multidrug efflux pump, which is likely to confer resistance to ampicillin.

Mutation in EnvZ of ampicillin-resistant UPEC decreases OmpF expression

EnvZ-OmpR is a two-component system responsible for regulating osmotic stress by controlling the expression of the porin proteins OmpF and OmpC [27]. EnvZ is a histidine kinase that comprises the HAMP, dimerization and histidine phosphotransfer (DHp), and histidine kinase domains. Under osmotic stress conditions, EnvZ autophosphorylates at a histidine residue located at position 243 within the DHp domain, and subsequently phosphorylates the transcriptional regulator OmpR [28]. Phosphorylated OmpR (OmpR-P) inhibits the expression of the large-pore protein OmpF while promoting the expression of the small-pore protein OmpC [29]. OmpF, known for its ability to transport various substrates including sugars, nutrients, and antibiotics, exhibits higher permeability than OmpC [30]. On the other hand, OmpC has lower permeability, thus preventing the entry of antibiotics [31–33].

The mutation site of EnvZ T402M in the Amp^R strains resides in the histidine kinase domain. This mutation was suspected to affect the phosphorylation of the histidine residue due to the increased length of the side chain, which includes a sulfur atom (Figure 4a–c). To investigate the impact of this mutation on EnvZ function, we performed qRT-PCR to measure the mRNA levels of ompF, ompC, envZ and ompR. In the wild-type strain, the mRNA levels of ompF and ompC decreased upon ampicillin treatment (Figure 4d,e), while no significant difference was observed in the transcription of envZ and ompR (Figure 4f,g). In contrast, the Amp^R strains showed complete abolition of OmpF expression, irrespective of the presence or absence of ampicillin (Figure 4d), while OmpC expression remained consistently upregulated (Figure 4e). The expression of envZ was slightly increased or nearly unchanged, whereas the expression of ompR was slightly elevated and reduced upon ampicillin treatment (Figure 4f,g). These qRT-PCR results suggest that the mutated EnvZ in the Amp^R strains is active as a kinase regardless of the presence of ampicillin. Consequently, only the small pore protein OmpC was produced, resulting in decreased membrane permeability and prevention of external antibiotics from entering the cell (Figure 4c).

Figure 4. The transcription of the ompF gene was down-regulated, while ompC, envZ, and ompR were up-regulated due to the mutation of EnvZ in the ampicillin-resistant UPEC strains.

(a) The mutation of EnvZ has an amino acid change, from threonine to methionine, at the 402^nd amino acid (highlighted in red). (b) The 3D modeled structure of EnvZ and Histidine kinase domain region. The mutation site is located at the ATP binding domain (highlighted in blue), and the autophosphorylation histidine site (highlighted in yellow) is located at the other EnvZ monomer. (c) Diagram illustrating the potential regulatory mechanism in the EnvZ-OmpR two-component system and the ompF and ompC genes operon comparing the wild-type EnvZ and EnvZ T402M variant. (d–g) The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB medium with ampicillin (+; 4 μg/ml, ++; 8 μg/ml). The mRNA levels of ompF, ompC, envZ, and ompR were determined by qRT-PCR. The mean values and standard deviations of three independent experiments are shown.

NlpD mutation impacts membrane permeability and ampicillin resistance in the high Amp^R strain

NlpD is a murein hydrolase activator, which is required for septal peptidoglycan cleavage and cell division in the cytokinesis process and participates in cell lysis induced by β-lactam antibiotics [34]. NlpD consists of two functional domains, LysM and degenerate LytM (dLytM) (Figure 5a). The LysM domain is responsible for localization, whereas the dLytM domain activates murein hydrolase AmiC (Figure 5B) [35]. The High Amp^R strain exclusively contains NlpD^1–186, lacking the dLytM domain required for interaction with AmiC, likely resulting in inefficient murein hydrolysis. This rigid peptidoglycan structure is expected to decrease membrane permeability and inhibit ampicillin-induced cell lysis. In the wild-type strain, the mRNA level of nlpD increased slightly in response to varying ampicillin concentrations, whereas no significant changes were observed in the Amp^R strains. The Low Amp^R strain exhibited higher expression levels of nlpD compared to the wild-type, while the High Amp^R strain showed similar levels to the wild-type (Figure 5c).

Figure 5. The mutation of NlpD was found only in the High Amp^R strain, leading to the loss of murein hydrolysis function.

(a) The mutation of NlpD produced a truncated protein terminating at the 186^th amino acid, resulting in the loss of murein hydrolysis function. (b) Schematic diagram of NlpD-AmiC murein hydrolysis model comparing the wild-type NlpD and NlpD 1–186 variants (c) The wild-type UPEC, Low Amp^R, and High Amp^R strains grown in LB medium with ampicillin (+; 4 μg/ml, ++; 8 μg/ml). The mRNA level of nlpD was determined by qRT-PCR. The mean values and standard deviations of three independent experiments are shown. (d, e) The wild-type UPEC, Low Amp^R, and High Amp^R strains were normalized for cell density (OD₆₀₀ = 1.0) in LB media. Following serial dilution (10⁻¹ to 10⁻⁶), 5 μl of each dilution was spotted on LB plates containing SDS and EDTA in the absence (d) or presence (e) 10 μg/ml ampicillin.

To investigate the impact of NlpD^1–186 on membrane permeability, we treated the strains with the detergent sodium dodecyl sulfate (SDS) and the metal chelating agent ethylenediaminetetraacetic acid (EDTA). In the absence of ampicillin, no significant differences were observed between strains (Figure 5d). However, at an ampicillin concentration of 10 μg/ml, the wild-type strain exhibited poor growth, while the Amp^R strains showed robust growth. Notably, the High Amp^R strain demonstrated significantly greater survival in the presence of ampicillin than the Low Amp^R strain (Figure 5e). Therefore, the NlpD mutation likely contributes to reduced membrane permeability and increased resistance to ampicillin by inhibiting the lysis of the murein layer (Figure 5b).

Maintenance of the Amp^R UPEC strain during antibiotic treatment in urinary tract infection

To assess the pathogenicity of the Amp^R strains, we transurethrally inoculated C57BL/6J mice with the wild-type and High Amp^R strains. Following an intravesical infection with 1 × 10⁸ CFU/bladder in C57BL/6J mice, oral administration of ampicillin was given twice daily for three days. On the fourth day post-inoculation, the bladders were harvested, homogenized, serially diluted, and then cultured on MacConkey agar medium. The CFU of the wild-type strain within the bladder showed a significant decrease after ampicillin treatment compared to no antibiotic treatment (Figure 6a). However, in the case of the High Amp^R strain, despite exhibiting a lower level of infection within the bladder compared to the wild-type, there was no significant reduction even after ampicillin treatment (Figure 6b). Moreover, when we determined the CFUs of UPEC strains within the 5637 bladder carcinoma cell line, CFU of the High Amp^R strain was slightly lower than that of the wild-type in the absence of ampicillin but more than 3-log higher than that of the wild-type in the presence of ampicillin (Figure 6d,e).

Figure 6. The High Amp^R strain maintains bacterial burden in the bladder during antibiotic treatment despite decreasing type 1 fimbriae expression.

(a) Schematic representation of mouse infection experiment using wild-type UPEC and the High Amp^R strains in the presence or absence of ampicillin or ciprofloxacin treatment. (b-c) In vivo experiment with C57BL/6J mice. Intrabladder CFU (-; saline, +; 200 mg/kg ampicillin). (d-f) CFUs of wild-type and the High Amp^R UPEC strains in the 5637 urothelial cell lines in the absence (d) or the presence of ampicillin (e) or ciprofloxacin (f) (p value *; <0.05, **; <0.01, and ****; <0.0001). (g) Diagram illustrating the potential regulatory mechanism of fimAICDFGH operon comparing the wild-type and High Amp^R UPEC strains. (H-J) Relative mRNA levels of fimB (h), fimE (i), and fimA (j) from the wild-type UPEC, Low Amp^R, and High Amp^R strains were determined by qRT-PCR. The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB with ampicillin (-; control, +; 5 μg/ml). (K) Transmission electron microscopy images of wild-type, Low Amp^R, and High Amp^R UPEC strains stained with 1% phosphotungstic acid (PTA) (Scale bars, 500 nM).

In the High Amp^R strain, the decrease in adhesion to the mouse bladder could be due to the lower expression of type 1 fimbriae, which is responsible for the attachment of UPEC to the host cell [36]. To test this hypothesis, we measured mRNA levels of the type 1 fimbriae operon (fimAICDFGH) and its regulators (fimB and fimE) (Figure 6g). The mRNA levels of fimB, encoding a regulatory protein for type 1 fimbriae, slightly decreased in the wild-type strain after ampicillin treatment but were still maintained at a significant level. However, in the Amp^R strains, fimB mRNA levels were much lower than those in the wild-type regardless of the presence of ampicillin (Figure 6h). In contrast, the mRNA levels of fimE, which encodes another regulatory protein for type 1 fimbriae, showed a slight decrease after ampicillin treatment, but there was no apparent difference between the wild-type and Amp^R strains (Figure 6i). Given that the expression of the type 1 fimbrial structural protein, FimA, is determined by the FimB/FimE ratio (Figure 6g) [37], fimA mRNA was highly expressed in the wild-type but nearly undetectable in the Amp^R strains (Figure 6j). Similarly, other genes in the type 1 fimbriae operon (fimC, fimD, fimF, and fimG genes) were downregulated in the Amp^R strains (Fig. S1). When we detected fimbriae by negative staining, the Amp^R strains produced fewer fimbriae than the wild-type (Figure 6k and S2). This is in agreement with previous reports that both MarA and OmpR-P decrease type 1 fimbriae expression [38,39], given that the levels of both MarA and OmpR-P are likely elevated in the Amp^R strains (Figure 2 and 4).

A shared mechanism of multidrug resistance in the Amp^R UPEC

In the previous section, we demonstrated that ampicillin resistance in the Amp^R strains originated from the higher expression of components in the multidrug efflux pump (via marR and acrR mutations), limiting the access of ampicillin by decreasing membrane permeability (via envZ mutation), or inhibiting cell lysis (via nlpD mutation). This underlying mechanism of ampicillin resistance led us to examine whether the identified Amp^R strains would exhibit resistance to other classes of antibiotics.

Previously, it was established that mutations in the acrR gene are associated with resistance to fluoroquinolone antibiotics [40]. Many multidrug-resistant clinical isolates exhibit mutations in the mar operon and genes involved in outer membrane permeability [41]. Additionally, several genes such as marR, acrR, and envZ have been implicated in antibiotic and organic solvent resistance in the previous studies [40,42,43]. Therefore, we performed MIC assays to investigate whether the Amp^R strains displayed resistance to other antibiotics with different mechanisms of action, including ciprofloxacin, kanamycin, chloramphenicol, tetracycline, trimethoprim, and nitrofurantoin. As expected, the Amp^R strains exhibited higher MIC values for the tested antibiotics than the wild-type strain. However, the Amp^R strains showed a decreased MIC for kanamycin compared to the wild-type strain, although the High Amp^R strain showed higher resistance than the Low Amp^R strain. The MIC of nitrofurantoin was unaffected by the Amp^R mutations (Table S1).

Using ciprofloxacin as a representative example, we performed further experiments. The MIC of ciprofloxacin for the wild-type strain was 0.015 μg/ml. In contrast, both the Low and High Amp^R strains exhibited an MIC of 0.12 μg/ml, with no significant difference between the two Amp^R strains (Figure 7a). Using qRT-PCR analysis, we measured the changes in mRNA levels of the nine aforementioned genes (excluding fimbrial genes) in response to ciprofloxacin. The marA, marB, acrA, and acrB genes exhibited lower expression levels in the wild-type strain regardless of the ciprofloxacin concentration, whereas the Amp^R strains maintained high expression levels of these genes (Figure 7b–e). Conversely, the mRNA levels of the ompF and ompC genes decreased in the wild-type strain upon exposure to ciprofloxacin. However, the Amp^R strains exhibited undetectable levels of ompF and constitutively high expression levels of ompC regardless of the presence of ciprofloxacin (Figure 7f,g). The envZ and ompR genes in the wild-type strain were transcribed at similar levels regardless of ciprofloxacin treatment, whereas in the Amp^R strains, their transcripts were slightly higher than those in the wild-type strain (Figure 7h,i). Lastly, ciprofloxacin reduced the mRNA levels of nlpD, but the Amp^R strains maintained higher expression levels than the wild-type strain (Figure 7j). Overall, these data indicate that the expression behaviors of the Amp^R strains in response to ciprofloxacin are similar to those in response to ampicillin.

Figure 7. Amp^R UPEC strains gain resistance to other classes of antibiotics including ciprofloxacin.

(a) MIC assay was performed for ciprofloxacin using 96-well plates with the wild-type UPEC, Low Amp^R, and High Amp^R strains. After incubation, 0.015% resazurin dye was added. MIC assays were performed duplicate for each strain. (b-j) The wild-type UPEC, Low Amp^R, and High Amp^R strains were grown in LB liquid media with different concentrations of ciprofloxacin (+; 0.06 µg/ml and ++; 0.12 µg/ml). The mRNA levels of marA, marB, acrA, acrB, ompF, ompC, envZ, ompR, and nlpD were determined by qRT-PCR.

Similar to what we observed with ampicillin treatment, the High Amp^R strain maintained bacterial CFU even after ciprofloxacin treatment in the mouse bladder (Figure 6c) and increased bacterial burden in the 5637 urothelial cell line (Figure 6f). These data suggest that acquired mutations during ampicillin exposure render the Amp^R strain resistant to other antibiotics as well and pose a challenge to antibiotic treatment.

Discussion

In this study, we generated strains that exhibited resistance to ampicillin by continuous exposure to ampicillin in the laboratory. Instead of acquiring a plasmid-associated expected-spectrum beta-lactamase (ESBL) gene [44], we identified four chromosomal mutations in the marR, acrR, envZ, and nlpD genes. These mutations indirectly confer resistance to β-lactam antibiotics by altering bacterial physiology, such as reducing membrane permeability or increasing the expression of efflux pumps. Previous studies have shown that E. coli strains evolved by ampicillin or biocide treatment often harbor mutations in the marR, acrR, envZ, and ompR genes [45,46]. In both cases, the isolated strains exhibited cross-resistance to a variety of other antibiotics [45,46], supporting the notion that reducing membrane permeability or increasing the expression of efflux pumps are common regulatory mechanisms of multidrug resistance. Our findings provide additional evidence that nlpD mutation significantly enhances ampicillin resistance. Using the MIC assay, we determined that the four mutations induced by ampicillin also conferred resistance to several other antibiotics with different mechanisms of action. This suggests that these genes are also involved in the acquisition of multidrug resistance as well.

Among the various mutations in marR, mutations occurring at amino acid positions 69 and 73 within the helix-turn-helix (HTH) motif lead to the separation of DNA-binding domains, resulting in the loss of DNA-binding activity [47]. Similarly, in the case of MarR^V66F mutation, an aromatic ring structure was introduced near the DNA-binding domain. It is expected that this modification will decrease the DNA binding affinity, rendering MarR^V66F unable to act as a repressor, thus allowing the expression of the marRAB operon (Figure 8). The expression of MarA leads to the up-regulation of various multidrug resistance-related genes, including acrA, acrB, and tolC, which are components of the AcrAB-TolC efflux pump [48]. Additionally, AcrR^1–121, which lacks the dimerization domain, was unable to repress the acrAB operon and instead promoted the expression of acrA and acrB (Figure 8). Furthermore, MarA is involved in DNA repair, enabling the repair of DNA damage caused by antibiotics and the regulation of membrane permeability [49]. Although the expression level of OmpF can also be decreased by MarA [50], this does not fully explain the complete repression of OmpF and the increased expression of OmpC. Therefore, it is speculated that EnvZ^T402M promotes the phosphorylation of OmpR; however, further studies are required to elucidate this mechanism.

Figure 8. A proposed model of multidrug resistance mechanisms in the laboratory-evolved Amp^R UPEC strains.

(a) The MarR^V66F mutation impairs its ability to repress the marRAB operon, leading to the expression of marA and marB genes. (b) Expression of the acrAB operon is activated by MarA protein. Moreover, the truncated AcrR is unable to repress the acrAB operon, allowing the expression of acrA and acrB genes. The AcrA, AcrB, and TolC proteins compose the multidrug efflux pump system. AcrAB-TolC complex is capable of exporting intracellular ampicillin and other antibiotics. (c) The EnvZ^T402M possibly leads to constitutive phosphorylation of the OmpR regulator. Phosphorylated OmpR inhibits the expression of ompF but promotes ompC expression. OmpC has a smaller pore size compared to OmpF, making it more difficult for the entry of various external substances, including ampicillin. (d) The truncated NlpD lacks the dLytM domain, which is involved in the interaction with the cell wall remodeling complex, including AmiC. As a result, the cell wall of the resistant strain becomes more rigid compared to the wild type. Also, the lack of the dLytM domain inhibits ampicillin-induced cell lysis.

In terms of the connection between NlpD and antibiotics, previous studies have shown that the dLytM domain of NlpD plays a crucial role in ampicillin-induced cell lysis [34]. In our study, we confirmed that the High Amp^R strain, lacking the dLytM domain in NlpD (NlpD^1–186), exhibited increased resistance to ampicillin. Moreover, the nlpD mutation not only conferred increased resistance to antibiotics but also resistance to the detergent SDS.

In an in vivo experiment using C57Bl/6J mice, the Amp^R strain exhibited reduced infection efficiency compared to the wild-type strain. This is likely due to the decrease in type 1 fimbriae expression under the influence of MarA and OmpR-P [38,39]. However, once the infection was established, the increase in antibiotic resistance of the mutants posed a greater challenge for the treatment of the mouse bladder infection (Figure 6).

The prevalence of multidrug-resistant uropathogenic Escherichia coli (UPEC) is increasing in developing countries, particularly with respect to resistance to beta-lactam, fluoroquinolone, and trimethoprim-sulfamethoxazole (TMP-SMX) antibiotics [9]. In our study, we identified ampicillin-resistant mutants that displayed resistance not only to ampicillin but also to other antibiotics, including ciprofloxacin, trimethoprim, chloramphenicol, and tetracycline. This suggests that multidrug-resistant uropathogenic Escherichia coli (UPEC) may arise due to shared regulatory mechanisms, such as decreased membrane permeability, increased expression of drug efflux pumps, or inhibition of cell lysis. Mutations in envZ and acrB have been reported to result in decreased resistance to kanamycin [43], similar to what we observed in the Amp^R strains (Table S1). Interestingly, susceptibility to nitrofurantoin remained unchanged. This is likely due to the conversion of nitrofurantoin into electrophilic intermediates by nitroreductase, leading to the generation of radical molecules that can target various cellular components [51]. Finally, understanding the mechanisms underlying multidrug resistance and pathogenesis in laboratory-evolved UPEC strains will provide insights into the development of potential therapeutic strategies.

Author contributions

E.-J.L. designed the research, analyzed the data, and wrote the manuscript; N.C. and E.C. performed the experiments and wrote the manuscript; Y.-J.C. analyzed the whole-genome sequencing; H.W.C wrote the manuscript; M.J.K. performed the in vivo experiment.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available in the NCBI SRA database at https://www.ncbi.nlm.nih.gov/sra (reference number PRJNA987582).

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2024.2367648

Additional information

Funding

This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT, and Future Planning [NRF-2022R1A2B5B02002256, NRF-2022R1A4A1025913, and NRF-2020M3A9H5104235 to E.-J.L., NRF-2021R1I1A1A01043879 to E.C., and RS-2023-00260267 to Y.-J.C.], the Korea National Institute of Health [2024ER210600 to Y.-J.C.], and a grant from Korea University [K2404611 to E.-J.L.].