Tobramycin-resistant small colony variant mutant of Salmonella enterica serovar Typhimurium shows collateral sensitivity to nitrofurantoin

ABSTRACT

The increasing antibiotic resistance poses a significant global health challenge, threatening our ability to combat infectious diseases. The phenomenon of collateral sensitivity, whereby resistance to one antibiotic is accompanied by increased sensitivity to another, offers potential avenues for novel therapeutic interventions against infections unresponsive to classical treatments. In this study, we elucidate the emergence of tobramycin (TOB)-resistant small colony variants (SCVs) due to mutations in the hemL gene, which render S. Typhimurium more susceptible to nitrofurantoin (NIT). Mechanistic studies demonstrate that the collateral sensitivity in TOB-resistant S. Typhimurium SCVs primarily stems from disruptions in haem biosynthesis. This leads to dysfunction in the electron transport chain (ETC) and redox imbalance, ultimately inducing lethal accumulation of reactive oxygen species (ROS). Additionally, the upregulation of nfsA/B expressions facilitates the conversion of NIT prodrug into its active form, promoting ROS-mediated bacterial killing and contributing to this collateral sensitivity pattern. Importantly, alternative NIT therapy demonstrates a significant reduction of bacterial load by more than 2.24-log₁₀ cfu/g in the murine thigh infection and colitis models. Our findings corroborate the collateral sensitivity of S. Typhimurium to nitrofurans as a consequence of evolving resistance to aminoglycosides. This provides a promising approach for treating infections due to aminoglycoside-resistant strains.

KEYWORDS:

• S. Typhimurium
• small colony variant (SCV)
• collateral sensitivity
• hemL

Introduction

Non-typhoid Salmonella (NTS) continues to pose a significant global public health threat, accounting for an estimated 93.8 million illnesses and 155,000 deaths annually [1]. As a zoonotic disease, salmonellosis ranks as the second most frequently reported bacterial gastrointestinal disease in the European Union, following campylobacteriosis [2]. Among these cases, Salmonella enterica serovar Typhimurium (S. Typhimurium) stands out as one of the most prevalent NTS serotypes responsible for humans infections [3,4]. The symptoms of S. Typhimurium infection range from diarrhoea and vomiting to potentially fatal dehydration, particularly among children and the elderly [5]. Globally, S. Typhimurium outbreaks have become increasingly prevalent, with a significant proportion associated with multidrug-resistant strains [6–9]. Despite ongoing endeavours to develop novel treatment strategies, antibiotics such as fluoroquinolones, aminoglycosides, and β-lactams remain pivotal in Salmonella Typhimurium therapy [10].

Bacterial small colony variants (SCV) are naturally occurring slow-growing subpopulations of bacteria characterized by smaller colony morphology, typically about 10% the size of wild-type colonies [11]. In addition to the formation of smaller colonies, altered virulence factors and increased antibiotic tolerance or resistance are also the distinctive phenotypic features [12]. At present, the SCV formation has been observed in various bacteria, including Staphylococcus spp., Pseudomonas spp., and Escherichia coli [13]. Moreover, comprehensive phenotypic and mechanistic investigations into SCV formation in S. Typhimurium have also been reported [14–16]. Cano et al. showed that prolonged bacterial residence in the intracellular environment of fibroblasts resulted in the emergence of genetically stable S. Typhimurium SCVs [14]. Mutations in lpd and aroD were associated with SCV intracellular persistence in fibroblasts and reduced susceptibility to aminoglycosides [14]. Previous studies also demonstrated the spontaneous emergence of S. Typhimurium SCVs after exposures to lactoferricin, colistin, and human defensin hNP-1 [17]. In this instances, mutations related to haem biosynthesis and respiration were implicated in SCV formation [17]. In addition, extensive amino acid auxotrophy, as observed in water cultures of S. Typhimurium, can impede bacterial growth and trigger SCV formation characterized by a loss-of-function mutation in the glutamine synthetase gene (glnA) [18]. Such auxotrophic SCVs may lead to increased resistance to host immune responses and antibiotics, contributing to the challenge of treating chronic S. Typhimurium infections.

Antibiotic resistance poses a growing problem that can emerge during antibiotic therapy as a result of gene transfer or the selection of spontaneous mutations [19–21]. The global dissemination of multidrug-resistant S. Typhimurium strains has significantly restricted clinical treatment options, emphasizing the urgent need for alternative therapeutic strategies to improve antibiotic efficacy [9]. The trade-offs associated with antibiotic resistance are increasingly attracting attention for designing antibiotic strategies that suppress the within-host emergence of antibiotic resistance [22]. In this context, collateral sensitivity, where resistance to one antibiotic results in heightened susceptibility to a second antibiotic, has been proposed as a potential strategy for suppressing antibiotic [23].

In the current study, we showed phenotypic trade-offs associated with the use of aminoglycosides that can perturb antibiotic susceptibility profiles in S. Typhimurium SCV mutants. This adaptation leads to collateral sensitivity to nitrofurantoin (NIT) by perturbing nfsA/B expression and haem biosynthesis, resulting in ETC dysfunction, redox imbalance, and ultimately ROS-induced bacterial eradication. Our findings provide proof-of-principle evidence for collateral sensitivity as a potential treatment strategy to address aminoglycoside resistance in S. Typhimurium infections.

Materials and methods

Bacterial culture and growth condition

All S. Typhimurium strains used in this study are detailed in supplemental Table S1. The SCV mutants derived from S. Typhimurium ATCC 14028 were cultured and maintained at 37 °C in Luria-Bertani broth (LB) or agar (LA, Oxoid, Basingstoke, UK). When required, tobramycin (TOB), nitrofurantoin (NIT) or 5-aminolevulinic acid (ALA) was supplemented to the growth media at the specified concentrations.

Plasmids and primers

The primers and plasmids used in this study are listed in Table S2 and S3. All plasmids were extracted using the TIANprep Mini Plasmid Kit (Tiangen, China).

Isolation of SCV mutants and adaptation evolution

TOB was used to select the SCV mutants of S. Typhimurium ATCC 14028 by gradient plate method on LB agar as described [24] with minor modifications. In brief, overnight cultures of ATCC 14028 were streaked onto gradient plate containing 0–512 mg/L of TOB and incubated at 37 °C overnight. Small and extremely small colonies were picked and restreaked on fresh selective media with 64 mg/L TOB to purify. This process was repeated until the SCV phenotype was consistently observed. Notably, the colony sizes of these SCVs were approximately one-tenth of those observed in wild-type bacterial colonies [13]. The SCV mutants of S. Typhimurium ATCC 14028 (designated as TOB-R) were then serially passaged by streaking onto TOB-supplemented plates to assess the phenotypic stability. Stable TOB-R SCV mutants, exhibiting single colonies approximately one-tenth the size of the WT, were picked, purified, and stored in 30% glycerol at −80 °C.

Whole genome sequencing and single nucleotide polymorphism analysis

DNA was extracted from S. Typhimurium ATCC 14028 and SCVs using a commercial genomic DNA purification kit according to the manufacturer’s instructions (Tiangen, Beijing, China). Whole genome sequencing (WGS) was performed with the Illumina HiSeq 2500 system (Novogene, Guangzhou, China) using the paired-end 2 × 150-bp sequencing protocol. The draft genome was de novo assembled using SPAdes version 3.9.0. The complete genome of ATCC 14028 (GenBank accession no. CP034479.1) served as the reference sequence for single nucleotide polymorphism (SNP) analysis of WT and SCVs using Snippy v4.6.0.

Construction of ΔhemL and complemented S. Typhimurium strains

Two plasmid systems consisting of pSGKP_hemL_HR and pCasKP-apr were used to constructed the hemL deletion mutant (ΔhemL) in the ATCC 14028 (WT) background [25]. For pSGKP_hemL, a sgRNA targeting the hemL was constructed in the pSGKp-rif using reverse PCR, and the resulting amplicon was transferred into E. coli DH5α. Subsequently, the plasmid was extracted, confirmed by Sanger sequencing, and designated as pSGKP_hemL. For pSGKP_hemL_HR, the pSGKP_hemL was digested by Sma I, and the ~ 500 bp homologous arm flanking hemL were cloned into the digested pSGKP_hemL. The resulting plasmid was named pSGKP_hemL_HR.

The plasmid pCasKP-apr was electroporated into the WT strain to enable the expression of Cas9 and lambda proteins. To introduce the pSGKP_hemL_HR plasmid into the WT strain harbouring pCasKP-apr, an overnight culture (1 mL) of the pCasKP-apr-harbouring WT strain was diluted into 100 mL of LB broth supplemented with 30 mg/L apramycin and incubated at 30 °C. Upon reaching an OD₆₀₀ of approximately 0.2, 1 mL of 20% L-arabinose was added to induce the lambda Red recombineering operon of pCasKP-apr. After 2 h of induction at 30 °C and reaching an OD₆₀₀ of approximately 0.6, the culture was prepared as electrocompetent cells. For electroporation, 100 μL of electrocompetent cells was thawed on ice for 5 min and mixed with 300–500 ng of pSGKP_hemL_HR. The mixture was then transferred into a 1 mm electroporation cuvette and electroporated. Subsequently, the pulsed cells were recovered in 1 mL of antibiotic-free LB broth and incubated at 30 °C for 4 h before being plated onto LB agar plates supplemented with the 30 mg/L apramycin and 50 mg/L rifampicin for overnight culture at 30 °C. The resulting colonies were verified by PCR and Sanger sequencing. To eliminate the pCasKP-apr and pSGKP_hemL_HR plasmids used in the knockout process, the knockout strains were obtained by three successive passages on LB agar plates containing 5% sucrose at 42 °C.

The hemL genes from the WT and TOB-R strains were recombined into a low-copy plasmid pGEN by seamless cloning. Subsequently, they were electroporated into the ΔhemL mutant to generate the complemented strains, designated as ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R), respectively.

Antimicrobial susceptibility testing and IC₉₀ determination

The minimal inhibitory concentrations (MICs) of tobramycin and nitrofurantoin with or without ALA (200 µM) against the parental ATCC 14028, TOB-R, ΔhemL mutant, and its complemented strains was determined using the broth microdilution method in accordance with the CLSI M100-S25 guideline [26]. Escherichia coli ATCC 25,922 served as the quality control strain. To map the collateral sensitivity profiles of antibiotic-resistant S. Typhimurium, overnight cultures of the tested strains were grown at 37°C with shaking in 96-well microtiter plates containing various gradient concentrations of antibiotics (amikacin, kanamycin, gentamycin, tobramycin, cefotaxime, cefquinome, ciprofloxacin, tetracycline, tigecycline, polymyxin E, polymyxin B, chloramphenicol, florfenicol, azithromycin, nitrofurantoin and rifampicin). Endpoint absorbance measurements (OD_600 nm) were taken after 18 h of incubation using a plate reader (Perkin Elmer, USA) and background-subtracted. The inhibition percentage (%) was calculated as follows: [(OD _{positive control} – OD _{negative control}) – (OD _antibiotic - OD _{negative control})]/(OD _{positive control} – OD _{negative control}) × 100%. The inhibitory concentration was defined as the lowest antibiotic concentration that inhibited 90% of the growth of the tested strain (IC₉₀).

Determination of bacterial growth in the presence of ALA, TOB and NIT

The WT, TOB-R, ΔhemL mutant, and its complemented strains were cultured to exponential phase and adjusted to an OD_600 nm to 0.5 (equivalent to ~ 10⁸ cfu/mL). The cultures were then diluted to ~ 10⁶ cfu/mL in fresh broth, with or without the addition of 200 µM ALA. Bacterial growth curves were monitored by measuring the OD_600 nm every 20 minutes at 37 °C to evaluate the impact of ALA on S. Typhimurium growth. Bacterial suspensions were then serially diluted to ~ 10³ cfu/mL, and 100 μL of the suspensions were spread evenly on agar plates with or without 200 µM ALA to observe colony size and identify the SCV phenotype. In addition, serial dilutions of bacterial cultures were spotted onto agar plates containing TOB (4 mg/L) or NIT (8 mg/L), respectively, to validate the collateral sensitivity of these strains. All plates were cultured at 37 °C overnight and subsequently photographed.

Heme concentration determination

The haem concentration was measured using the porphyrin fluorescence assay [27]. This assay involves generating the fluorescent product protoporphyrin from haem by heating samples in oxalic acid. Briefly, overnight cultures of bacterial strains were diluted 1:100 and then incubated at 37 °C and 180 rpm for 4 h. Following incubation, bacterial cells were washed, collected and mixed with 500 µL of 20 mM oxalic acid at 4 °C for 16 h. The samples were then divided into two 500 μL aliquots. One set of samples were heated at 95 °C for 30 min to detect the total fluorescence value of porphyrin and haem, while the second set served as the control group stored at room temperature to measure the fluorescence value of porphyrin concentration. After cooling, the supernatant was centrifuged, and fluorescence was measured with excitation at 400 nm and emission at 620 nm using a Perkin Elmer fluorescence spectrometer. The absolute difference between the fluorescence values of the two groups represented the haem concentration.

Membrane depolarization assay

The membrane depolarization assay was performed following a previously established protocol [28]. Briefly, an overnight bacterial culture was diluted 1:100 in fresh LB broth and incubated for 4 h at 37 °C with agitation at 180 rpm. Bacterial cells were then harvested, washed twice, and resuspended to obtain an OD_600 nm of 0.5 with 5 mM HEPES buffer (containing 5 mM glucose, pH 7.0). After incubation for 30 min, a final concentration of 5 μM 3,3-dipropylthiadicarbocyanine iodide [DiSC₃(5)] was added at 37 °C in the dark. Subsequently, the dissipated membrane potential of each tested S. Typhimurium strain was determined by measuring the fluorescence intensity using an EnSight Multimode plate reader (Perkin Elmer, USA), with excitation and emission wavelengths set at 622 nm and 670 nm, respectively.

Hydrogen peroxide killing assay

The resistance to hydrogen peroxide (H₂O₂) killing was determined as described previously [29]. Bacterial cells in exponential phase were diluted to ~ 10 ⁶ cfu/mL in fresh LB broth, and H₂O₂ was added at a final concentration of 2 mM for 1 h. After H₂O₂ treatment, the bacterial cultures were serially diluted 10-fold in PBS, and 100 μL aliquots from each dilution were plated on LB plates and then incubated at 37 °C for 16 h to quantify colony counts.

Total ROS determination

Total ROS levels in S. Typhimurium strains were determined by the fluorescence probe 2,’7’-dichlorofluorescein diacetate (DCFH-DA) [30]. Exponential phase bacteria were washed twice with PBS and adjusted to an OD_600 nm of 0.5. The DCFH-DA was then added at a final concentration of 10 µM, and the mixture was further incubated at 37 °C for 30 min. After washing with PBS for three times, fluorescence intensity was measured at an excitation wavelength of 488 nm and an emission wavelength of 525 nm using an EnSight Multimode plate reader (Perkin Elmer, USA).

RNA isolation and RT-qPCR

The mRNA levels of soxS, marA, rob, nfsA and nfsB in TOB-R, ΔhemL mutant and its complemented strains relative to the WT were quantified by RT-qPCR analysis using the SYBR Green Master Mix (Vazyme, China). Bacterial cultures were grown in antibiotic-free LB broth until reaching the exponential phase (i.e. OD₆₀₀ = 0.5). Total RNAs were extracted using the OMEGA Total RNA Kit I (Omega, China), followed by reverse transcription of 500 ng total RNA using the HiScript III All-in-one RT SuperMix Perfect (Vazyme, China). The primers used for RT-qRCR were listed in Table S3. Relative gene expression levels were determined using the 2^−ΔΔCt method [31].

Nitroreductase overexpression

The open reading frames of nfsA and nfsB were separately amplified from the WT chromosome using primers containing EcoR I restriction sites. The plasmid pBAD24 was then digested by EcoR I, and the amplified nfsA and nfsB sequences were individually cloned into the digested pBAD24. The resulting plasmids were designated as pBAD:nfsA and pBAD:nfsB, respectively. These plasmids were then electroporated into the WT to generate nfsA and nfsB overexpression strains.

Murine infection models

Six-week-old specific-pathogen-free ICR mice (Guangdong Medical Lab Animal Center, Guangzhou, China) were used for the in vivo studies. All animal experiments were approved by the Animal Research Committees of the South China Agricultural University (SCAU-IACUC). Murine thigh infection and colitis models were established to assess the differential efficacy of NIT against the WT, TOB-R, ΔhemL mutant and its complemented strains ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R).

For the murine thigh infection model, mice were rendered neutropenic by the injection of cyclophosphamide intraperitoneally 4 days (150 mg/kg) and 1 day (100 mg/kg) prior to infection [32]. Thigh infections were induced by the intramuscular injection of 0.10 mL of bacterial suspension (10⁴ cfu/mL) into the thighs of neutropenic mice. Control groups received only the same volume of PBS inoculation. At 2 h post-infection, mice were randomized to receive (i) control treatment with PBS vehicle, intragastrically (i.g.) twice a day (bid), or (ii) NIT at 25 mg/kg i.g. bid. Control and NIT-treated mice were euthanized by CO₂ inhalation at 24 h after start of therapy. Thighs were aseptically removed, homogenized, and quantitatively cultured to determine the mean log₁₀ cfu/g tissue (± standard deviation; n = 6).

For the murine colitis model, mice were administered 20 mg streptomycin in 100 μL of sterile water by gavage 24 h prior to S. Typhimurium infection [33]. Each mouse was then orally gavaged with 0.10 mL of bacterial suspension (~10 ⁸ cfu/mL). Control groups received only the same volume of PBS inoculation. At 24 h post-infection, mice were randomized to receive (i) control treatment with PBS i.g. once a day (qd), or (ii) NIT at 12.5 mg/kg i.g. qd. The treatments last for 2 days. Mice were euthanized by CO₂ inhalation and their liver and spleen were harvested, homogenized, serially diluted, and quantitatively cultured to determine the mean log₁₀ cfu/g tissue (± standard deviation; n = 4).

Ethics statement

Animal experiments were conducted in strict compliance with the ARRIVE guidelines 2.0. Animals were maintained in accordance with the National Standards for Laboratory Animals of China. The Animal Research Committee (IACUC) of the South China Agricultural University (SCAU) approved these studies (#2022B121 and #2022C014).

Statistical analysis

All experiments were performed with three biological replicates. Statistical analysis was performed using SPSS software (IBM, USA). Unless stated otherwise, the statistical significance of comparison was assessed using the Student’s t-test, one-way or two-way analysis of variance. The MIC variables were analysed using the chi-square test, as appropriate (p < 0.05, p < 0.01, p < 0.001, ns: not significant).

Results

TOB-resistant S. Typhimurium SCV with a point mutation in hemL showed collateral sensitivity to NIT

Three evolutionary-derived S. Typhimurium ATCC 14028 mutants (TOB-R) exhibiting stable resistance to TOB were selected under the gradient selective pressure of TOB. Notably, these TOB-R mutant strains presented a tip-like morphology, with smaller and more diffuse colonies on both LB and blood agar plates compared to the WT (Figure 1a and S1), indicative of the SCV phenotype [13]. WGS and Snippy-SNP spectrum analysis of the WT and three TOB-R mutants consistently showed a SNP mutation (T:A>C:G) in the hemL gene (Table S4). This mutation led to a change from cysteine (Cys) to arginine (Arg) at amino acid position 212 (HemL-C212R).

Figure 1. TOB-resistant S. Typhimurium SCV shows collateral sensitivity to NIT.

(a) Comparison of colony morphologies between the WT and TOB-R SCV strains grown on LB agar and sheep blood agar. (b) Fold changes in IC₉₀ values of 16 tested antibiotics for the TOB-R mutant relative to the WT. Antibiotics include: amikacin (AMI), kanamycin (KAN), gentamicin (GEN), tobramycin (TOB), cefotaxime (CTX), cefquinome (CFQ), ciprofloxacin (CIP), tetracycline (TET), tigecycline (TGC), Colistin (COL), polymyxin B (POL), chloramphenicol (CHL), florfenicol (FFC), azithromycin (AZI), nitrofurantoin (NIT), and rifampicin (RIF).

For the TOB-R SCV, we further performed dose-response curves to determine the IC₉₀ changes for 16 antibiotics (Figure 1b and S2). The S. Typhimurium SCV mutant that evolved resistance to TOB, a commonly used treatment for multidrug-resistant Gram-negative bacteria, exhibited increased sensitivity to four unrelated antibiotics from diverse chemical classes, including NIT, RIF, FFC and AZI (Figure 1b). While most observed collateral sensitivity resulted in a less than twofold reduction in IC₉₀ relative to WT, a fourfold reduction of IC₉₀ was observed for the TOB-R SCV against NIT. Expectedly, all three evolutionarily derived ATCC 14028 mutants (TOB-R) of S. typhimurium showed resistance to TOB and susceptibility to NIT (Table S5). Importantly, the enduring collateral sensitivity between TOB and NIT persisted even after 28 days of consecutive passaging (data not shown). This result indicated that NIT could be a promising sequential-treatment approach against such resistant pathogens when S. Typhimurium SCV mutants develop resistance to aminoglycosides. Of concern, apart from antibiotics belonging to the aminoglycosides (AMI, KAN and GEN), we also observed collateral resistance with a > 16-fold increase in the IC₉₀ for the TOB-R SCV mutant compared to the WT (Figure 1b). In particular, a significant level of collateral resistance to β-lactams (CTX and CFQ; Figure 1b) was noted for the TOB-R SCV mutant. Combinations of β-lactam and aminoglycoside classes are commonly used in clinical practice to broaden the antimicrobial spectrum [34]. Hence, this collateral resistance may lead to treatment failure due to the selection of resistant S. Typhimurium SCV mutants during TOB treatment.

The ΔhemL induces SCV formation and NIT collateral sensitivity

To explore the impacts of the hemL gene on SCV formation and collateral sensitivity to NIT, we used CRISPR/Cas-9 to construct a hemL knockout strain of S. Typhimurium ATCC 14028 (ΔhemL). We then complemented this strain with plasmids containing either the WT or mutated hemL, generating the ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R) strains, respectively. Similar to the TOB-R SCV strain (Figure 1a), both the ΔhemL and ΔhemL (hemL^TOB-R) colonies exhibited irregular and smaller morphologies on LB agar plates compared to the WT (Figure 2a). However, complementation of the ΔhemL strain with hemL^WT restored colony morphology similar to that of the WT (Figure 2a). Consistent with ΔhemL, the ΔhemL (hemL^TOB-R) strain exhibited resistance to TOB (Figure 2b). In contrast, an increased sensitivity to NIT was observed for both ΔhemL and ΔhemL (hemL^TOB-R) strains, with a 3.78 to 4.35-fold reduction in the IC₉₀ value of NIT (Figure 2c; p < 0.05). This suggests that the intact hemL gene and its involvement in synthesizing ALA and haem are crucial for the collateral sensitivity between TOB and NIT.

Figure 2. S. Typhimurium ΔhemL mutant confers resistance to TOB and sensitivity to NIT.

(a) Dilutions of the WT, ΔhemL and its complemented strains [ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R)] were spotted on indicated plates with or without 4 mg/L TOB and 8 mg/L NIT, then incubated overnight at 37 °C. (b,c) Inhibitory rate curves of the WT, ΔhemL, ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R) in the presence of TOB (0.0625-64 mg/L; panel B) and NIT (0.125-128 mg/L; panel C).

In the haem synthesis pathway of S. Typhimurium, HemL catalyzes the conversion of glutamate-1-semialdehyde into ALA. We therefore speculated that dysfunction in the hemL function leads to intracellular deficiencies of ALA and haem, potentially prompting the formation of TOB-resistant SCV [14]. After exogenous addition of 200 µM ALA, all hemL-deficient strains including TOB-R, ΔhemL and ΔhemL (hemL^TOB-R), exhibited growth restoration comparable to the WT level (Figure 3a–e). However, ALA supplementation had no discernible effect on the growth of the functionally intact WT or ΔhemL (hemL^WT) strain (Figure 3a–e). Similarly, the provision of ALA in agar plates inhibited SCV formation and substantially restored colony morphology for these hemL-deficient strains (Figure 3a–e; p < 0.001). In addition, the resistance to TOB and collateral sensitivity to NIT, as evidenced by changes in MIC, could be reversed by ALA supplementation (Figure 3f,g; p < 0.001, p < 0.05). In particular, the MICs of TOB-R and ΔhemL strains reverted nearly to the WT level in the presence of 200 µM ALA (Figure 3f,g).

Figure 3. ALA deficiency induces S. Typhimurium SCV phenotype and collateral sensitivity to NIT.

(a–e) Growth curves and corresponding colony morphology of the WT, TOB-R, ΔhemL and its complemented strains [ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R)] in the absence or presence of 200 µM ALA, respectively. (f,g) The effects of exogenous addition of 200 µM ALA on the MIC changes of TOB (f) and NIT (g) against these tested S. Typhimurium strains. MIC variables were analysed using the chi-square test, with P<0.05 considered statistically significant. ns, not significant.

Reduced PMF generated by the ETC leads to TOB resistance

The ETC protein complexes rely on haem for their function and thus depend on porphyrin-haem biosynthesis [35]. The hemL encodes glutamate-1-semialdehyde aminomutase, which catalyzes the final step in ALA synthesis, a crucial precursor in the haem biosynthetic pathway [36]. As expected, intracellular haem levels were decreased significantly in TOB-R, ΔhemL and ΔhemL (hemL^TOB-R) strains compared to WT (Figure 4a; p < 0.001). Since TOB internalization usually requires a PMF generated by the ETC [37,38], we therefore measured the cytoplasmic membrane potential (ΔΨ) of the tested S. Typhimurium strains using the fluorescent probe DiSC₃(5). Compared to the WT, the TOB-R and ΔhemL strains exhibited a 1.31 to 1.50-fold increase in fluorescence (Figure 4b; p < 0.01), indicating a loss of electric potential in the hemL-deficient S. Typhimurium strains. The membrane potential depolarization is related to PMF production, which facilitates aminoglycoside uptake through its transporter [39], potentially contributing to increased TOB resistance in hemL-deficient strains.

Figure 4. Mechanism of collateral sensitivity to NIT in TOB-resistant S. Typhimurium SCV mutant strain.

(a) Relative haem production in TOB-R SCV, ΔhemL mutant, ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R) strains compared to the WT. (b) Dissipation of membrane potential in TOB-resistant S. Typhimurium SCV strain due to hemL mutation or deletion. Fluorescence probe DiSC₃(5) was added at 30 min to assess membrane potential changes of tested strains. (c) Comparison of bacterial survival after 1 h of exposure to 2 mM H₂O₂ treatment among tested strains, expressed as a percentage of the initial inoculum. (d) Quantification of total ROS production in tested strains using fluorescence dye DCFH-DA. (e) Relative expression of genes (soxS, marA, rob, nfsA and nfsB) associated with NIT sensitivity.

Intracellular redox imbalance and increased nfsA/B expression contribute to NIT collateral sensitivity

The dissipation of membrane potential generally leads to an imbalance in redox homoeostasis, culminating in bacterial death [40]. Thus, we compared the sensitivity to H₂O₂ for the WT, TOB-R, ΔhemL and their complemented S. Typhimurium strains. Under the exogenous oxidative stress, the survival rates of TOB-R and ΔhemL strains were significantly lower than that of the WT (Figure 4c; p < 0.001). Similar results were observed between the ΔhemL (hemL^TOB-R) and ΔhemL (hemL^WT) strains (p < 0.001). Further analysis with the DCFH-DA probe demonstrated that the hemL-deficient strains, including TOB-R, ΔhemL and ΔhemL (hemL^TOB-R), accumulated higher intracellular ROS levels compared to the WT and ΔhemL (hemL^WT) strains (Figure 4d; p < 0.05).

The conversion of nitrofurantoin from its prodrug to active form necessitates intracellular processing by bacterial nitroreductases NfsA and NfsB [41]. As expected, the overexpression of nfsA and nfsB resulted in increased sensitivity to nitrofurantoin (Figure S3 and Table S6). We also observed higher expression levels of nfsA and nfsB genes in the TOB-R and ΔhemL strains compared to the WT (Figure 4e). Consistently, complementation of the ΔhemL strain with hemL^TOB-R significantly elevated the expressions of nfsA and nfsB compared to the hemL^WT-complemented ΔhemL strain (Figure 4e; p < 0.001). In addition, increased expression levels of key genes (soxS, marA and rob) that positively regulate nitroreductases were observed in all hemL-deficient S. Typhimurium strains (Figure 4e; p < 0.05). Based on these results, a schematic model illustrating the possible mechanisms underlying NIT collateral sensitivity is shown in Figure 5. In TOB-resistant S. Typhimurium SCV, the hemL mutation decreases haem biosynthesis, leading to the ETC dysfunction, redox imbalance, and intracellular ROS accumulation. Upregulated expression of nfsA/B facilitates the conversion of NIT into its active form, thereby promoting ROS-induced bacterial killing. These factors collectively contributed to the collateral sensitivity of S. Typhimurium to NIT.

Figure 5. Graphic illustration of the possible mechanism of collateral sensitivity to NIT in TOB-resistant S. Typhimurium SCV induced by HemL deficiency.

Importantly, in vivo experiments further demonstrated the enhanced efficacy of NIT against TOB-R SCV infections (Figure 6). In the murine thigh infection model, NIT treatment had no significant effects on WT and ΔhemL (hemL^WT) infections, but significantly decreased bacterial loads in the thighs of TOB-R, ΔhemL, and ΔhemL (hemL^TOB-R) infected mice by 1.54 to 2.06-log₁₀ cfu/g (Figure 6a,b, p < 0.001). Similar results were observed in the murine colitis model, where NIT treatment markedly reduced bacterial loads in the liver and spleen of TOB-R, ΔhemL and ΔhemL (hemL^TOB-R) infected mice by 0.99 to 1.42-log₁₀ cfu/g (Figure 6c–e, p < 0.01). Therefore, NIT may represent a promising strategy for combating aminoglycoside-resistant S. Typhimurium SCV infections.

Figure 6. Comparison of NIT efficacy in murine models of thigh infection and colitis due to the WT, TOB-R, ΔhemL and its complemented strains.

(a) Experimental protocols for the murine thigh infection model. (b) Bacterial loads in the thigh from mice infected with the WT, TOB-R, ΔhemL, ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R) strains after intragastrical (i.g.) administration of NIT at 25 mg/kg twice a day (bid). (c) Experimental protocols for the murine colitis model. (d–e) Bacterial loads in the liver (d) and spleen (e) from mice infected with the WT, TOB-R, ΔhemL, ΔhemL (hemL^WT) and ΔhemL (hemL^TOB-R) strains after i.g. administration of NIT at 12.5 mg/kg once a day (qd). Statistical analysis was performed by the unpaired t-test. ns, not significant.

Discussion

Collectively, this study delineates an experimentally isolated haem-auxotrophic S. Typhimurium SCV under the selective pressure of TOB, revealing its pronounced collateral sensitivity to NIT. Since the initial discovery of bacterial SCV from Eberthella typhosa (now known as S. Typhimurium) in 1910, SCV formation has been observed across various bacteria from clinical specimens [13,42]. Proctor et al. highlighted a persistent and relapsing infection associated with S. aureus SCVs, immediately recognizing the clinical importance of SCV infection in considering therapeutic strategies [43]. With respect to the underlying mechanisms, previous studies have showed that ALA deficiency due to the hemL mutation induces S. Typhimurium SCV formation in fibroblasts [14]. This is in line with our findings in the TOB-resistant S. Typhimurium SCVs, as we also identified a point mutation in the hemL gene in this study. Importantly, our study verified that supplementation with exogenous ALA can restore the small colony morphology of S. Typhimurium SCV mutants. Furthermore, we elucidated the molecular mechanisms by which ALA deficiency-induced S. Typhimurium SCVs lead to tobramycin resistance and nitrofurantoin sensitivity.

Intracellular bacterial adaptation and alteration in energy metabolism are associated with phenotype switching to SCV [44]. In particular, the ETC dysfunction is frequently correlated with SCV formation and persistent infections [45]. Previous studies have demonstrated that SCV isolates of E. coli, P. aeruginosa, and S. aureus with ETC deficiencies exhibit lower growth rates under aerobic conditions [46–48]. These deficiencies lead to a rapid reduction in membrane potential, diminish the transmembrane H⁺ gradient, and impair energy-dependent drug transport processes. Since aminoglycoside uptake relies on both ATP and pH [49], even minor ETC dysfunction can impact ATP production and aminoglycoside susceptibility during SCV formation. This inference was supported by results presented herein. The haem-deficient S. Typhimurium strains showed reduced PMF and increased resistance to TOB. Complementation of the ΔhemL strain with the intact hemL^WT gene restored growth and all other key SCV phenotypic features, including colony size, haem levels, and PMF. These findings parallel previous studies showing reduced aminoglycoside susceptibility in clinically isolated S. aureus and P. aeruginosa SCV strains from cystic fibrosis patients [50].

Deficiencies in the bacterial ETC can also affect antibiotic susceptibility through disrupting redox balance and energy homoeostasis [51]. A recent study demonstrated that E. coli mutants with impaired ubiquinone biosynthesis grow poorly under aerobic conditions due to oxidative stress [52]. Similarly, in this study, S. Typhimurium strains with hemL mutations were notably sensitive to H₂O₂ treatment. In addition, the ETC defects exacerbate the leakage of electrons, impair cellular redox-buffering capacity, and ultimately result in uncontrolled ROS-induced bacterial killing [53]. This is consistent with our observation of elevated ROS levels in the hemL-deficient S. Typhimurium strains. Subsequent ROS accumulation may promote bacterial killing through oxidative damage, thus heightening the bactericidal activity of NIT.

In principle, drug switches can restore control of evolving pathogen populations and may be pre-planned according to the collateral sensitivity profiles. This approach could prolong the efficacy of existing antibiotics and help to predict the evolutionary directions of emerging resistance genes [54]. To our knowledge, this study provides the first demonstration of an intriguing collateral sensitivity strategy involving nitrofurans exploited by a stable haem-auxotrophic S. Typhimurium SCV isolate under constant exposure to aminoglycosides. In fact, this collateral sensitivity to nitrofurans herein could potentially represent a functional evolutionary trade-off to aminoglycoside resistance. While the precise mechanism of nitrofurans remains to be elusive, their bactericidal activity is believed to primarily involve ROS-mediated oxidative DNA damage [41]. However, the efficacy of NIT depends on the intracellular conversion of the prodrug to its active form by bacterial nitroreductases NfsA/B [55,56]. In S. Typhimurium, the expressions of nfsA and nfsB genes are regulated by three activators, marA, rob, and soxS [57]. Interestingly, our study showed significant upregulation of multiple regulons (marA/rob/soxS) in hemL-deficient S. Typhimurium strains, which further promoted the expressions of nfsA and nfsB. Consequently, more NIT prodrug could be converted to its active component, thereby boosting NIT activity against S. Typhimurium mutants. Furthermore, the nitroreduction process generates the highly reactive byproducts of aerobic metabolism, including ROS from the oxidation of amines [58]. This process could synergistically interact with haem deficiency-induced redox imbalance. Since excessive ROS production and redox imbalance can result in DNA damage, it is plausible that heterogeneity in the expression of DNA repair or ROS detoxification enzymes contribute to collateral sensitivity [51].

Conclusions

In summary, the present study demonstrated the formation of TOB-resistant S. Typhimurium SCV and their collateral sensitivity to NIT caused by mutations in the hemL gene. Mechanistic studies revealed that hemL mutation decreases haem biosynthesis, leading to the ETC dysfunction, redox imbalance and the accumulation of lethal ROS. The upregulated nfsA/B expression further facilitates the production of activated NIT, promoting ROS-induced bacterial killing, and ultimately contributing to the collateral sensitivity of S. Typhimurium SCVs to NIT. This approach may provide a promising avenue for combination therapy to combat resistance evolution. Nevertheless, future in vivo and clinical studies are warranted to determine whether this antibiotic collateral sensitivity strategy could help address antibiotic resistance in clinical settings.

Author contributions

Yu-Feng Zhou and Xiao-Ping Liao completed the conception and design of this study. Chang-Zhen Wang, Liang-Xing Fang and Jian Sun analysed the data. Chang-Zhen Wang and Yu-Feng Zhou drafted the manuscript. Chang-Zhen Wang, Yue-Jun Zhang, Yue-Fei Chu, Long-Gen Zhong, Jin-Peng Xu, Liu-Yan Liang and Teng-Fei Long carried out the experiments. Xiao-Ping Liao revised it critically for intellectual content. Yu-Feng Zhou final approval of the version to be published. All authors agreed to be accountable for all aspects of the work.

Disclosure statement

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

Data availability statement

Whole genome sequencing data of the WT, TOB-R, TOB-R1 and TOB-R2 strains have been submitted to the NCBI under the BioSample accession number SAMN 40,697,886, SAMN 39,637,111, SAMN40697887, and SAMN 40,697,888. Derived data supporting the findings of this study are available at https://doi.org/10.6084/m9.figshare.25650747.v1.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2024.2356692

Additional information

Funding

This work was supported by the National Key Research and Development Program of China [2023YFD1800100 and 2022YFD1802100], Guangzhou Science and Technology Plan Project [2024A04J3412], the National Natural Science Foundation of China [32273066 and 32373063], the Foundation for Innovative Research Groups of National Natural Science Foundation of China [32121004], the Local Innovative and Research Teams Project of Guangdong Pearl River Talents Program [2019BT02N054], Laboratory of Lingnan Modern Agriculture Project [NT2021006], Double First-Class Discipline Promotion Project [2023B10564003] and the 111 Center [D20008].