ABSTRACT
Our investigations have identified a mechanism by which exogenous production of nitric oxide (NO) induces resistance of Gram-positive and -negative bacteria to aminoglycosides. An NO donor was found to protect Salmonella spp. against structurally diverse classes of aminoglycosides of the 4,6-disubstituted 2-deoxystreptamine group. Likewise, NO generated enzymatically by inducible NO synthase of gamma interferon-primed macrophages protected intracellular Salmonella against the cytotoxicity of gentamicin. NO levels that elicited protection against aminoglycosides repressed Salmonella respiratory activity. NO nitrosylated terminal quinol cytochrome oxidases, without exerting long-lasting inhibition of NADH dehydrogenases of the electron transport chain. The NO-mediated repression of respiratory activity blocked both energy-dependent phases I and II of aminoglycoside uptake but not the initial electrostatic interaction of the drug with the bacterial cell envelope. As seen in Salmonella, the NO-dependent inhibition of the electron transport chain also afforded aminoglycoside resistance to the clinically important pathogens Pseudomonas aeruginosa and Staphylococcus aureus. Together, these findings provide evidence for a model in which repression of aerobic respiration by NO fluxes associated with host inflammatory responses can reduce drug uptake, thus promoting resistance to several members of the aminoglycoside family in phylogenetically diverse bacteria.
INTRODUCTION
Aminoglycosides are often used in the treatment of infections caused by aerobic or facultative Gram-negative bacilli, Gram-positive cocci, or Mycobacterium spp. The bactericidal activity of aminoglycosides depends on the interaction of the drug with the 30S subunit of bacterial ribosomes. Bacteria, nonetheless, can utilize several mechanisms to enhance resistance to aminoglycosides. For example, enzymatic drug modification by N-acetyltransferases, O-nucleotidyltransferases, or O-phosphotransferases are the most common mechanisms of clinical resistance to aminoglycosides (21). Mutations or methylation of 16S rRNA, and mutations of ribosomal proteins have also been shown to decrease the antimicrobial activity of aminoglycosides (11, 12, 32, 46). In addition, constitutive or induced expression of multidrug efflux pumps represent independent strategies to avoid killing by aminoglycosides (1, 16, 22, 37).
Prior to poisoning their intracellular ribosomal targets, however, aminoglycosides must cross the cytoplasmic membrane (41). The uptake of aminoglycosides by bacterial cells is an energy-dependent process that requires both an electrochemical gradient across the membrane and electron flow through the respiratory chain (41). Consequently, drugs that inhibit respiration or uncouple the electron transport chain (ETC) protect bacteria from the microbicidal effects of aminoglycosides (4, 41). In agreement with data generated by pharmacological means, genetic defects that interfere with ETC activity protect bacteria against the cytotoxicity of aminoglycosides. The relationship between a dysfunctional ETC and reduced effectiveness of aminoglycosides has been recognized for some time in clinical settings (33). Small colony variant (SCV) bacteria are frequently associated with persistent infections and are unresponsive to most antibiotic therapy, particularly aminoglycosides (34). The addition of menadione or hemin can restore the sensitivity of some of the SCV bacteria to aminoglycosides, indicating that defects in the ETC are responsible for some of the phenotypes described for SCV bacteria (34). Due to their defective electron transport, both clinical and genetically engineered SCV bacteria take up low amounts of aminoglycosides (4, 5, 29, 34, 45), thereby shielding the ribosomes from the toxicity of this class of antibiotics.
Recently, endogenously generated nitric oxide (NO) was found to preserve the replication of Bacillus in the presence of toxic levels of aminoglycosides (14). NO is also produced as an antimicrobial molecule by eukaryotic cells in response to microbial products or cytokines through activation of inducible NO synthase (30, 40). NO has been shown to be important in the control of a multitude of clinically important bacterial pathogens, including mycobacteria, Salmonella spp., Staphylococcus aureus, and Pseudomonas aeruginosa (13, 20, 26, 38). Given what we know about the role of the ETC in the uptake of aminoglycosides and the inhibitory effects that NO exerts on the ETC (6, 17, 39, 48), we hypothesized that repression of ETC function by NO levels produced in the inflammatory response could interfere with the antibacterial properties of this important class of antibiotics. As anticipated in this model, our investigations show that gross repression of respiration by NO prevents drug uptake, thus protecting Gram-positive and -negative bacteria from the antibiotic activity of aminoglycosides.
MATERIALS AND METHODS
Bacterial strains.The bacterial strains and plasmids used in the present study are listed in Table 1. Salmonella enterica serovar Typhimurium strain ATCC 14028s was used as the parental strain for targeted chromosomal mutations according to the λ Red-mediated gene replacement method (8). Primers containing nucleotides homologous to Salmonella genomic DNA, followed by pKD13-encoded sequences, were used for PCR amplification of the FLP recognition target (FRT)-flanked kanamycin resistance cassette encoded within the pKD13 plasmid (Table 2). After DpnI digestion, the PCR product was electroporated into S. Typhimurium strain TT22236 carrying the pTP2223 plasmid that expresses the λ Red recombinase under Ptac control (31). Mutations were moved into S. Typhimurium strain 14028s by bacteriophage P22-mediated transduction, and pseudolysogens were eliminated by streaking on Evans blue uranine agar plates. Chromosomal mutations were verified by PCR. S. aureus strain Newman was obtained from the American Type Culture Collection (ATCC; Manassas, VA), and P. aeruginosa strain ATCC PAO1 was a gift from M. Vasil at the University of Colorado School of Medicine.
Bacterial strains and plasmids
Oligonucleotides
Chemicals.Spermine and spermine NONOate were resuspended in 10 mM Tris-HCl (pH 8.5) and stored at −80°C. Spermine NONOate was used as a source of NO in the experiments conducted in vitro. [3H]gentamicin (American Radiolabeled Chemicals, Inc., St. Louis, MO) was supplied as a 1-mCi/ml solution (specific activity, 200 mCi/mmol) in ethanol. Gentamicin sulfate, tobramycin, amikacin, and streptomycin were dissolved in phosphate-buffered saline (PBS), while the kanamycin used for the selection of mutant bacteria was prepared in water. Unless noted, all reagents were obtained from Sigma-Aldrich.
O2 consumption.Bacteria grown overnight in Luria-Bertani (LB) broth were subcultured 1:100 in EG (1.66 mM MgSO4, 9.5 mM citric acid monohydrate, 57 mM K2HPO4, 16.7 mM NaNH3PO4, 0.4% [wt/vol] glucose) (44) or LB broth at 37°C with shaking to an optical density at 600 nm (OD600) of 0.5. Selected bacterial suspensions were treated with 100 to 750 μM spermine NONOate, which at 37°C and pH 7.4 has a half-life of 39 min. The consumption of O2 by bacteria placed in microcentrifuge tubes was measured amperometrically using an ISO-OXY-2 O2 sensor attached to an APOLLO 4000 free radical analyzer (World Precision Instruments, Inc., Sarasota, FL). The data are expressed as nA.
Difference spectra analysis.Salmonella grown overnight in LB broth were inoculated at 1:100 in EG medium (pH 7.0) for 4 h. To study the nitrosylation of heme d, specimens were suspended at an OD600 of 0.1 in 50 mM Tris-HCl (pH 6.5) and treated with 750 μM spermine NONOate for 10 min with shaking in a 37°C water bath incubator. Bacterial cultures were adjusted to an OD600 of 0.1. Spermine NONOate-treated minus aerated whole-cell difference spectra were recorded in a Varian Cary 50 Bio UV-visible spectrophotometer.
Antimicrobial killing assays.Salmonella or Pseudomonas strains were grown overnight in LB broth and subcultured 1:100 in EG medium at 37°C with shaking to an OD600 of 0.2 or 0.5. S. aureus was grown overnight in LB broth, subcultured 1:100 in LB broth, and grown at 37°C with shaking to an OD600 of 0.5. Where indicated, bacterial cultures were treated before antibiotic administration with spermine NONOate. After exposure to the antibiotics, the viable bacteria were enumerated after overnight incubation at 37°C on LB agar plates. The percent survival was calculated as follows: [CFU (tn)/CFU (t0)] × 100.
[3H]gentamicin uptake.Salmonella strains were grown overnight with shaking in LB broth at 37°C, subcultured at 1:100 in EG medium, and grown to an OD600 of 0.5. Spermine NONOate or its parent compound spermine was added to the Salmonella cultures immediately prior to the addition of [3H]gentamicin. A 10-μCi/ml stock solution of [3H]gentamicin prepared in EG medium containing 1.25 mg of unlabeled gentamicin/ml was diluted 1:100 into the Salmonella cultures. After the indicated time points, 500 μl of the bacterial cultures was run through a vacuum manifold (Promega, Madison, WI) equipped with 0.45-μm-pore-size HA nitrocellulose filters (Millipore, Billerica, MA) that had been prewashed with 400 μl of EG medium containing 12.5 μg of gentamicin/ml. The cells on the filters were washed with 400 μl of EG medium containing 12.5 μg of gentamicin/ml, and the specimens were placed in scintillation fluid. The beta emissions were counted with a Beckman LS-6000 scintillation counter, averaged over 1 min, and corrected for background. The data are presented as counts per minute.
NADH dehydrogenase enzymatic activity.Salmonella strains grown to an OD600 of 0.4 in EG medium were treated for 30 min with 750 μM the NO donor spermine NONOate or the parent compound spermine. Cells resuspended in 1 ml of ice-cold buffer A (20 mM KH2PO4, 5 mM MgSO4 buffer [pH 7.5]) were lysed five times for 10 s with a sonic dismembrator (Fisher). Cellular debris was removed by centrifugation at 15,000 × g at 4°C for 10 min. Supernatants containing the membrane fractions were treated with 1 mM dithiothreitol, and the samples were stored on ice. The concentration of protein in cellular extracts was determined spectrophotometrically at 660 nm using Pierce reagent according to the manufacturer's instructions (Pierce, Rockford, IL). Membrane fractions containing 25 μg of protein were treated with 500 μM NADH (Sigma) in a final volume of 200 μl of buffer A in a 96-well plate. NADH consumption was assessed spectrophotometrically at 340 nm over 5 min by using a VERSAmax microplate reader (Molecular Devices, Sunnyvale, CA).
Macrophages.Gp91phox−/− (30a) and doubly deficient gp91phox−/− iNOS−/− (38a) mice in a C57BL/6 background were bred in our animal facility according to Institutional Animal Care and Use Committee guidelines. Peritoneal macrophages were harvested from mice 4 days after intraperitoneal inoculation of 1 mg of sodium periodate/ml as described previously (43). The peritoneal exudate cells were resuspended in RPMI 1640 medium (Cellgro; Mediatech, Inc.) supplemented with 10% heat-inactivated fetal bovine serum (BioWhittaker Inc), 15 mM HEPES, 2 mM l-glutamine, 1 mM sodium pyruvate (Sigma), 100 U of penicillin/ml, and 100 mg of streptomycin/ml (Cellgro). The peritoneal exudate cells were seeded in 96-well tissue culture plates at a density of 105 cells/well, selected by adherence after 48 h of culture at 37°C in a 5% CO2 incubator, and treated with 200 U of gamma interferon (IFN-γ; Life Technologies)/ml during the last 20 h of culture before Salmonella infection.
Intracellular gentamicin killing.Macrophages were challenged at a multiplicity of infection of 2 with wild-type Salmonella. The bacteria were opsonized with 10% normal mouse serum for 20 min before infection. Extracellular bacteria were removed from the monolayers 25 min after challenge by washing with prewarmed RPMI 1640 medium containing 6 μg of gentamicin (Sigma)/ml (27). To assess the ability of gentamicin to kill intracellular Salmonella, increasing concentrations of gentamicin were added to the monolayers 1 h postinfection, and the numbers of viable intracellular salmonellae were determined 7 h later. The Salmonella-infected macrophages were lysed with 0.2% deoxycholate in PBS, and the surviving bacteria were enumerated on LB agar plates. The results are expressed as number of CFU/well 8 h postinfection.
Statistical analysis.Statistical significance was determined by two-way analyses of variance of raw or transformed data, followed by Bonferroni's post-test. Differences were considered statistically significant when P ≤ 0.05.
RESULTS
Nitrosylation of terminal cytochromes of the ETC inhibits the respiration of Salmonella.The NO donor spermine NONOate inhibited the respiration of log-phase Salmonella in a concentration-dependent manner (Fig. 1A). As little as 100 μM spermine NONOate, which initially generate ∼3.4 μM NO/min, noticeably affected the respiratory activity of Salmonella. The addition of 750 μM spermine NONOate, which generates ∼25.5 μM NO/min, shut down Salmonella respiration. The respiratory activity of cells grown on glucose relies heavily on the enzymatic activity of NADH dehydrogenases and quinol cytochrome oxidases. Spermine NONOate-treated minus aerated whole-cell difference spectra showed a peak at 630 nm typical of nitrosylated heme d (Fig. 1B). The 660-nm peak that corresponds to the O2-bound ferrous heme d was lost in cells treated with 750 μM spermine NONOate. These findings are consistent with the idea that NO outcompetes O2 for binding to the reduced heme d, which together with heme b595, forms the catalytically active site of the cytochrome oxidase. NO does not exert direct effects on NADH dehydrogenases. However, peroxynitrite (ONOO−), which arises through the reaction of NO with endogenous superoxide (O2−), exerts long-lasting inhibition of NADH dehydrogenases of the ETC (18). We therefore assessed the effects of spermine NONOate on the enzymatic activity of NADH dehydrogenases. As expected, membranes from a Δnuo Δndh strain lacking both NADH dehydrogenases failed to consume NADH (Fig. 1C). NO did not exert long-lasting inhibition of NADH dehydrogenases since NADH consumption from cells treated with 750 μM spermine NONOate was similar to that observed from the wild type (Fig. 1C). Collectively, these findings suggest that the nitrosylation of terminal cytochromes of the electron transport chain is responsible for the gross repression in respiratory activity seen in NO-treated Salmonella.
Effect of NO on O2 consumption. (A) O2 consumption was estimated polarographically. Salmonella grown to an OD600 of 0.5 in EG medium were treated with the NO donor spermine NONOate (sNO) (n = 3 to 10). (B) Difference spectrum of 750 μM sNO-treated minus aerated whole-cell spectra of Salmonella. Bacteria grown in EG medium (pH 7.0) to OD600 of 0.5 were treated with the NO donor for 10 min before the spectrum was recorded. The arrow at 630 nm indicates nitrosylated heme d of the quinol cytochrome bd, while the arrow at 660 nm indicates the trough left in heme d devoid of O2. The trace is representative of three independent experiments. (C) NADH dehydrogenase activity of the ETC was measured in membranes of wild-type (WT) Salmonella. Some of the WT bacteria were treated with 750 mM sNO prior to membrane isolation. A strain deficient in both nuo and ndh NADH dehydrogenases (CI) was included as control.
NO-mediated inhibition of respiration induces resistance to aminoglycosides.The uptake of aminoglycosides is an energy-dependent process that requires a threshold membrane potential and a functional ETC (41). Because NO inhibits respiration by reversibly nitrosylating metal cofactors in terminal quinol cytochrome oxidases of the ETC (Fig. 1) (6, 17, 18, 39), we undertook investigations to evaluate whether NO can promote bacterial resistance to aminoglycosides. Time-kill curves were carried out to assess the effects of NO on the anti-Salmonella activity of aminoglycosides (Fig. 2A). The viability of log-phase Salmonella decreased by 1,000-fold 2 h after exposure to 12.5 μg of gentamicin/ml. Remarkably, Salmonella cotreated with 750 μM spermine NONOate survived gentamicin treatment. The induction of resistance appears to be due to the generation of NO, because the spermine base did not abrogate the killing activity of gentamicin (Fig. 2B). We tested whether the protective effects of NO are generalizable to structurally diverse members of the aminoglycoside family. The four major families of aminoglycosides are classified based on an aminocyclitol ring of 2-deoxystreptamine or streptidine linked by glycosidic bonds to amino- or non-amino-containing sugars. Members of the gentamicin and kanamycin families contain 2-deoxystreptamine disubstituted at positions 4 and 6, while members of the neomycin family are substituted at positions 5 and 6. Streptomycin (the only member of the streptomycin family) contains streptidine. The viability of Salmonella decreased by 30- to 100-fold after exposure to tobramycin (Fig. 2C) and amikacin (Fig. 2D), respectively. Similar to gentamicin, the anti-Salmonella activity of tobramycin and amikacin was lost after the addition of 750 μM spermine NONOate. The effects of NO on streptomycin could not be tested, because this aminoglycoside failed to kill Salmonella, even at 200 μg/ml. Collectively, these findings suggest that NO protects Salmonella against most clinically important members of the aminoglycosides.
High NO fluxes protect Salmonella against diverse members of the aminoglycoside family. The survival of Salmonella grown to an OD600 of 0.2 in EG medium after treatment with 12.5 μg of gentamicin (GM)/ml (A), 25 μg of tobramycin (TB)/ml (C), or 50 μg of amikacin (AK)/ml (D) was evaluated. Where indicated, the bacteria were treated with 750 μM spermine NONOate (sNO). The effects of 750 μM spermine or 750 μM sNO on the killing activity of 25 μg of gentamicin/ml are shown in panel B. The experiments in panels B and D were carried out for 2 h. The data are representative of three independent experiments.
The respiratory inhibitor cyanide protects Salmonella against gentamicin.Potassium cyanide was used in order to independently test whether inhibition of respiration can protect Salmonella from gentamicin. According to the phenotypes seen with NO, at 100 μM the electron transport chain antagonist potassium cyanide repressed O2 consumption (Fig. 3A) and protected Salmonella from gentamicin (Fig. 3B). The number of bacteria surviving after 2 h of treatment with gentamicin and potassium cyanide was higher than 100%, likely reflecting the growth of the bacteria.
The respiratory inhibitor cyanide protects Salmonella against gentamicin. Oxygen consumption (A) and gentamicin killing (B) were determined in Salmonella grown in EG medium to an OD600 of 0.5. Selected cultures were treated with 100 μM KCN. Bacterial survival was determined 2 h after the addition of 25 μg of gentamicin/ml. The data are from three independent experiments.
NO produced by macrophages inhibit gentamicin killing.Because pharmacologically generated NO increases the resistance of Salmonella against aminoglycosides, we tested whether NO produced enzymatically by professional phagocytes could avert the cytotoxicity of aminoglycosides against intracellular Salmonella. This possibility was evaluated in gp91phox−/− macrophages. Phagocytic cells lacking the gp91phox subunit of the NADPH oxidase provide two advantages: (i) they are fully capable to generating nitrosative stress (27, 43), and (ii) the intracellular Salmonella burden remains constant in this population of macrophages for ∼10 h (43). Although aminoglycosides penetrate eukaryotic cells poorly, they appear to cross the cytoplasmic membrane in enough concentrations to exert antimicrobial activity against intracellular bacteria (2, 9, 15, 28). Salmonellae were allowed to establish an intracellular infection within IFN-γ-treated, gp91phox-deficient macrophages. The addition of 12.5–200 μg of gentamicin/ml failed to kill intracellular Salmonella in gp91phox-deficient macrophages (Fig. 4). In sharp contrast, gentamicin was highly successful in reducing the intracellular Salmonella burden within congenic gp91phox−/− iNOS−/− macrophages unable to generate NO. Compared to gp91phox−/− controls, 50, 100, and 200 μg of gentamicin/ml reduced by approximately 3-, 5-, and 83-fold, respectively, the intracellular Salmonella burden of IFN-γ-treated gp91phox−/− iNOS−/− macrophages. These results indicate that NO generated by macrophages can protect intracellular Salmonella from the toxicity of aminoglycosides.
NO generated by host macrophages shields intracellular Salmonella from the cytotoxicity of gentamicin. The killing activity of gentamicin against intracellular Salmonella was assessed in macrophages from gp91phox−/− or doubly immunodeficient gp91phox−/− iNOS−/− mice. A total of 200 U of IFN-γ/ml was added to the macrophages 20 h prior to infection. At 1 h after intracellular uptake, the infected cells were treated with increasing concentrations of gentamicin. The intracellular bacteria that survived after 7 h of antibiotic administration were quantified on LB plates. The data represent the number of CFU isolated 8 h postinfection (p.i.) ± the standard deviation from three independent observations.
NO protects Gram-positive and -negative bacteria against aminoglycosides.A link between reduced respiratory capacity and resistance to aminoglycosides has been documented for the pathogens P. aeruginosa and S. aureus (5). P. aeruginosa and S. aureus were consequently used to test whether NO can protect diverse microorganisms from gentamicin. Our initial attempts at testing the effects of NO on the gentamicin-mediated killing of P. aeruginosa were unsuccessful because the addition of 750 μM spermine NONOate by itself reduced the viability of aerobically grown Pseudomonas by 10-fold (not shown). This observation is in line with the known susceptibility of anaerobic P. aeruginosa to acidified nitrite (24, 47). Lowering the dose of spermine NONOate to 250 μM effectively limited O2 consumption, while preserving the viability of P. aeruginosa from the lethal activity of 25 μg of gentamicin/ml (Fig. 5A). On the other hand, the addition of 750 μM spermine NONOate partially inhibited the respiratory capacity of S. aureus (Fig. 5B, left panel). Nonetheless, the modest inhibition in respiration seen after the addition of 750 μM spermine NONOate was followed by a robust protection of S. aureus against 25 μg of gentamicin/ml (Fig. 5B, right panel). Taken together, these data indicate that NO protects Gram-positive and -negative microorganisms against the cytotoxicity of aminoglycosides through the targeted inhibition of ETC function.
NO protects Gram-positive and Gram-negative bacteria against aminoglycosides. O2 consumption (left panels) and gentamicin killing (right panels) of P. aeruginosa (A) or S. aureus (B) were evaluated. Where indicated, the bacteria were treated with spermine NONOate (sNO). The percent surviving bacteria was recorded 2 h after the addition of 12.5 μg of gentamicin/ml. The data are from three independent experiments.
NO-treated Salmonella displays a small-colony variant-like phenotype.Salmonella exposed to NO displays deficiencies in cellular growth and aerobic respiration, while exhibiting resistance to aminoglycosides. These phenotypes are reminiscent of those of SCV bacteria, which were first noted in Salmonella and have now been documented in S. aureus, P. aeruginosa, Burkholderia cepacia, E. coli, and others (34). The effect of NO on respiratory activity and gentamicin sensitivity was tested in Salmonella strains containing mutations in components of the ETC commonly found in laboratory and clinical SCV isolates. Mutations were constructed in menA and hemL, encoding proteins involved in the synthesis of menadione and hemin, respectively. Isoprenylation of menadione forms menaquinone, while hemin is an intermediate in the biosynthesis of terminal quinol cytochrome oxidases. A complex I-deficient Salmonella strain lacking both nuo and ndh NADH dehydrogenases of the ETC was also tested. The bacteria were cultured overnight on LB agar plates to assess growth. Unlike menadione auxotrophies in S. aureus (3), wild-type and isogenic menA-deficient Salmonella isolates produced colonies of similar size (Fig. 6A). The disparate growth of strains of Salmonella and S. aureus deficient in menadione most likely stems from the alternative use of ubiquinone in the former. Consistent with hemin-deficient SCV S. aureus (3, 45), hemL-deficient Salmonella displayed reduced growth (Fig. 6A). The largest growth defect was, nonetheless, observed for complex I-deficient Salmonella. A relation was noticed between colony size (Fig. 6A) and O2 consumption (Fig. 6B, left panel). Respiratory activity was similar in wild-type and menA-deficient Salmonella isolates, whereas the O2-consuming activities of hemL- and complex I-deficient salmonellae were reduced 73 and 43%, respectively. The remaining respiratory activities of hemL- and complex I-deficient Salmonella isolates indicates that terminal quinol cytochrome oxidases are still active, possibly reflecting a leaky mutation in the former (19) and sources of electrons other than NADH in the latter. Consistently, NO treatment reduced O2 consumption in all Salmonella strains tested (Fig. 6B, right). The susceptibility of respiratory-deficient Salmonella to gentamicin was tested after the addition of 750 μM spermine NONOate (Fig. 6C). Both wild-type and menA-deficient Salmonella isolates were susceptible to 25 μg of gentamicin/ml, whereas strains deficient in hemL or NADH dehydrogenases were resistant (P < 0.001). NO protected both wild-type and menA- and hemL-deficient Salmonella strains from gentamicin. Although NO inhibited its marginal O2-consuming capacity, this treatment did not change the already great innate resistance of complex I-deficient Salmonella to gentamicin. These findings support the hypothesis that, similar to the defects in the ETC of SCV bacteria, blockage of aerobic respiration by high NO fluxes promotes aminoglycoside resistance.
Effects of NO on the O2-consuming capacity and gentamicin susceptibility of ETC mutant Salmonella. (A) Colony morphology of wild-type (WT) or respiratory-chain-mutant Salmonella strains after 20 h of culture on LB agar plates. Complex I NADH dehydrogenase mutant (CI) bacteria were used for comparison. (B) O2 consumption of untreated (left panel) or 750 μM spermine NONOate-treated (750 μM sNO, right panel) Salmonella strains grown to an OD600 of 0.5 in LB broth (n = 3 to 11). (C) Susceptibility of Salmonella grown to an OD600 of 0.2 in EG medium to 25 μg of gentamicin/ml. Selected samples were treated with 750 μM sNO. The data represent the percentage of bacteria recovered 2 h after exposure to gentamicin (n = 6 to 34).
High NO fluxes block aminoglycoside uptake.The failure of aminoglycosides to kill SCV bacteria has been linked to limited respiration and the corresponding reduction in drug uptake (34). Thus, intracellular accumulation of [3H]gentamicin was used to gain insights into the mechanism by which high NO fluxes protect Salmonella from aminoglycosides. Low levels of [3H]gentamicin were noted to bind to log-phase Salmonella immediately (t0) upon treatment (Fig. 7A), likely reflecting the energy-independent association of positively charged aminoglycosides with anionic moieties in the bacterial cell envelope (41). Over the following 30 min, [3H]gentamicin gradually accumulated in Salmonella. This slow phase of linear uptake known as energy-dependent phase I (EDPI) requires a functional ETC (41). Subsequently, the rate of [3H]gentamicin accumulation increased rapidly (Fig. 7A), corresponding to energy-dependent phase II (EDPII). Compared to spermine, 750 μM spermine NONOate inhibited [3H]gentamicin uptake over the 60-min period tested (Fig. 7B). The levels of gentamicin in NO-treated cells were similar to those of untreated cells at t0, indicating that NO inhibits both EDPI and EDPII but not the initial electrostatic interaction of the drug with the bacterial cell envelope. These findings are consistent with the idea that blockage of aerobic respiration by high NO fluxes reduces drug uptake, thereby promoting aminoglycoside resistance.
High NO fluxes block the uptake of gentamicin. (A) [3H]gentamicin uptake was estimated from Salmonella grown to an OD600 of 0.5 in EG medium. (B) Gentamicin uptake in 750 μM spermine- or spermine NONOate (sNO)-treated Salmonella was measured 60 min after the addition of [3H]gentamicin. The data are expressed as the mean [3H]gentamicin (cpm) ± the standard deviation from six to nine independent observations. ***, P < 0.001 (compared to spermine-treated controls).
DISCUSSION
Our investigations have identified inhibition of respiration as a mechanism by which exogenously generated NO protects Gram-negative S. enterica serovar Typhimurium and P. aeruginosa bacilli and the Gram-positive S. aureus cocci from gentamicin killing. Figure 8 depicts a model for a mechanism by which NO may protect bacteria from aminoglycosides. Under the conditions used in our investigations, NADH dehydrogenases and terminal quinol cytochrome oxidases are possible targets for the NO-mediated inhibition of respiratory activity. Our biochemical analysis of whole Salmonella membranes has shown that NO nitrosylates heme d, but it does not appear to inhibit NADH dehydrogenase enzymatic activity. These observations are consistent with the higher affinity of NO for terminal cytochromes than NADH dehydrogenases (35). Binding of NO to the catalytically active metal centers of quinol cytochrome oxidases should prevent in a competitive fashion the reduction of O2 to H2O, thereby providing a reasonable explanation for the inhibition of respiration seen in NO-treated Salmonella. The arrest in respiration should in turn block electron transport and the translocation of protons across the bacterial membrane required to sustain a threshold membrane potential, both of which are necessary for the uptake of aminoglycosides (41). The idea that NO-mediated protection against aminoglycosides works through the inhibition of terminal quinol cytochrome oxidases is independently supported by data presented here and elsewhere (4) that the classical respiratory inhibitor cyanide effectively promotes resistance to gentamicin.
Model for the mechanism by which NO protects against gentamicin. NO selectively binds to CuB and heme d in binuclear centers of the bo and bd quinol cytochrome oxidases of the ETC, respectively. Since NO is in competition with O2, the nitrosylation of these metal centers in terminal cytochromes represses respiration, and thus it prevents both the reduction of O2 to H2O and the translocation of protons (H+) across the bacterial cell membrane. The consequent loss in proton motif force limits the uptake of aminoglycosides, thus increasing bacterial resistance against this class of antibiotics.
Members of the aminoglycoside group appear to have different degrees of effectiveness against Salmonella. The uneven anti-Salmonella activity of tobramycin, amikacin, and gentamicin likely reflects differences in the kinetics of drug uptake (42). Regardless of their intrinsic anti-Salmonella activity, NO efficiently protected this Gram-negative pathogen against the cytotoxicity of all aminoglycosides tested. Following their initial electrostatic binding to the bacterial cell envelop, aminoglycosides are transported into the cell in two consecutive energy-dependent phases (41). The immediate electrostatic interaction of gentamicin with the bacterial cell surface appears to be preserved in cells exposed to NO. The subsequent energy-dependent phases of gentamicin uptake are, however, blocked by the nitrosative stress generated by NO. Our studies indicate that the protective effects of NO are not limited to gentamicin but apply to several aminoglycosides containing the 4,6-disubstituted, 2-deoxystreptamine structure. Since all aminoglycosides are transported across the bacterial membrane in a similar energy-dependent fashion, gross repression of respiration may help rationalize the protective effects of NO against structurally diverse aminoglycosides.
Adaptive resistance to aminoglycosides through NO-mediated repression of ETC function is reminiscent of the hyper-resistance to aminoglycosides observed for SCV bacteria (34). SCV bacteria cause recalcitrant infections unresponsive to conventional antibacterial therapy. The remarkable resistance of SCV bacteria to aminoglycosides has been associated with defective ETC function and limited drug uptake. In analogy to the restrictions of ETC function in SCV bacteria, the NO-dependent adaptive response may limit the effectiveness of aminoglycoside intervention. However, contrary to the lasting effects of genetic mutations in SCV bacteria, the nitrosylation of terminal cytochromes of the ETC is transitory (25, 39, 48) and may provide a selective advantage in situations when fast bacterial growth is possible.
Our studies indicate that NO protects phylogenetically diverse microorganisms such as S. enterica, P. aeruginosa, and S. aureus against the cytotoxicity associated with aminoglycosides. NO did not, however, inhibit the respiratory capacity of all of the bacteria tested to the same extent. For instance, NO partially inhibited the respiration of S. aureus. This observation is in keeping with the idea that this Gram-positive bacterium shows remarkable adaptability to the growth restrictions imposed by NO (36). Despite the partial inhibition of respiration, NO afforded S. aureus outstanding protection against gentamicin. Two possibilities might account for this observation. First, partial respiratory inhibition may limit enough gentamicin uptake to protect the ribosomes from the cytotoxicity of the drug. This idea would also help to explain the remarkable resistance of our hemL-deficient Salmonella strain to gentamicin despite measurable respiratory activity. Second, it is possible that, in addition to affording resistance through blocking drug uptake, NO could independently protect against aminoglycosides by a mechanism other than a complete shut down of respiratory activity.
In conclusion, NO-mediated signaling through quinol cytochrome oxidases of the ETC triggers a conserved adaptive response in Gram-positive and -negative pathogens against the bactericidal activity of aminoglycosides. These findings may be of biological relevance since aminoglycosides are used as adjunctive therapy to treat septicemias, endocarditis, and bone and lung infections caused by P. aeruginosa, S. aureus, and Mycobacterium spp. The propensity of these clinically relevant pathogens to elicit host NO synthesis (13, 23, 36, 38) may in turn diminish the effectiveness of aminoglycoside therapy. This notion is supported by the fact that NO generated by IFN-γ-primed macrophages prevented killing of intracellular Salmonella by gentamicin.
ACKNOWLEDGMENTS
We are grateful to Michael Vasil from the University of Colorado School of Medicine for the generous gift of Pseudomonas strain PA01 and to Ben Sattler for technical assistance with the antimicrobial killing assays. We thank Jessica Jones-Carson for helpful discussions during the preparation of the manuscript.
This study was supported by the National Institutes of Health grants U54 AI-065357, AI053213, and AI054959, the Burroughs Welcome Fund, and the Mucosal and Vaccine Research Program of Colorado (MAVRC).
FOOTNOTES
- Received 31 August 2010.
- Returned for modification 29 November 2010.
- Accepted 13 February 2011.
- Accepted manuscript posted online 22 February 2011.
- Copyright © 2011, American Society for Microbiology.
