Metabolic Compensation of Fitness Costs Associated with Overexpression of the Multidrug Efflux Pump MexEF-OprN in Pseudomonas aeruginosa

Abstract

The acquisition of antibiotic resistance has been associated with a possible nonspecific, metabolic burden that is reflected in decreased fitness among resistant bacteria. We have recently demonstrated that overexpression of the MexEF-OprN multidrug efflux pump does not produce a metabolic burden when measured by classical competitions tests but rather leads to a number of changes in the organism's physiology. One of these changes is the untimely activation of the nitrate respiratory chain under aerobic conditions. MexEF-OprN is a proton/substrate antiporter. Overexpression of this element should result in a constant influx of protons, which may lead to cytoplasmic acidification. Acidification was not observed in aerobiosis, a situation in which the MexEF-overproducing mutant increases oxygen consumption. This enhanced oxygen uptake serves to eliminate intracellular proton accumulation, preventing the cytoplasmic acidification that was observed exclusively under anaerobic conditions, a situation in which the fitness of the MexEF-OprN-overproducing mutant decreases. Finally, we determined that the early activation of the nitrate respiratory chain under aerobic conditions plays a role in preventing a deleterious effect associated with the overexpression of MexEF-OprN. Our results show that metabolic rewiring may assist in overcoming the potential fitness cost associated with the acquisition of antibiotic resistance. Furthermore, the capability to metabolically compensate for this effect is habitat dependent, as demonstrated by our results under anaerobic conditions. The development of drugs that prevent metabolic compensation of fitness costs may help to reduce the persistence and dissemination of antibiotic resistance.

INTRODUCTION

Bacterial pathogens are capable of becoming resistant to antibiotics either through the acquisition of resistance genes by horizontal gene transfer (1) or through mutations (2). The latter frequently occur in genes encoding antibiotic targets or transporters that are fundamental to bacterial physiology. Such mutations could impair bacterial metabolism, and consequently, it has been widely assumed that acquisition of antibiotic resistance should result in a metabolic burden (3). This burden would decrease the ability of the resistant mutants to compete with their susceptible, wild-type counterparts (fitness cost). The implementation of antibiotic cycling strategies arose in light of this assumption in an attempt to reduce the occurrence of antibiotic resistance. Unfortunately, different studies have shown that antibiotic cycling has limited efficacy for preventing and diminishing antibiotic resistance (4). Antibiotic-resistant microorganisms are also found in habitats without previous antibiotic exposure (5); altogether, these findings indicate that once resistance is achieved, resistant microorganisms may not be as easily outcompeted by susceptible ones, as would be expected if resistance always imposed a fitness cost (6, 7).

There are different explanations for this situation, the first one being that fitness costs can be compensated for by secondary mutations (8) and the second one being that the effect of acquired resistance on bacterial physiology may be more specific than previously thought, with some mutations being highly detrimental and others lacking a substantial impact on bacterial fitness (9–11). In this article, we explore yet another explanation: the possibility that resistant microorganisms adjust their metabolism to compensate, without further mutations, for the fitness cost associated with the acquisition of resistance.

In order to explore this possibility, we studied in more detail the metabolic changes associated with the overexpression of the Pseudomonas aeruginosa MexEF-OprN multidrug (MDR) efflux pump. We chose this model because we have recently shown that overexpression of this pump alters the quorum-sensing response without affecting bacterial fitness when measured by classical pairwise competition tests in rich medium under aerobic conditions (12).

MDR efflux pumps are relevant elements that contribute to antibiotic resistance in many bacterial pathogens. In addition to antimicrobial extrusion, efflux pumps are capable of extruding a wide variety of different compounds, host-derived antimicrobials, quorum-sensing signals, solvents, plant exudates, and bacterial metabolites, among others (13–15). As a consequence of this activity, different works have shown that some efflux pumps are relevant elements in modulating both the bacterial quorum-sensing response and virulence (13–21).

Although some efflux pumps present detectable levels of expression and hence contribute to intrinsic antibiotic resistance (22–26), expression of MDR efflux pumps is subject to tight regulation at the local level, and as a result, it is usually repressed (27). However, increased expression levels and, hence, enhanced resistance to antibiotics can be achieved in the presence of an effector (transient resistance) (28, 29) or as a consequence of mutations in the elements regulating their expression (acquired resistance) (30, 31).

The fact that MDR efflux pumps are expressed at high levels only when needed suggests that constitutive overexpression might impose severe fitness costs. Such an effect might conceivably arise from the overexpression of these elements and the resulting metabolic burden derived from both the energetic requirements for their functioning and the constant, nonregulated, extrusion of metabolic intermediates (14, 15, 32). However, as stated above, we found that overexpression of MexEF-OprN does not confer a nonspecific fitness cost but rather causes specific changes in bacterial physiology (12), including alterations in the quorum-sensing response and in type III secretion, which may be relevant for P. aeruginosa behavior during infection (17, 33). It has been suggested that the overexpression of some efflux pumps can be compensated for by concomitant reductions in the expression levels of others (34). However, our previous transcriptomic analysis showed that the levels of expression of the already known P. aeruginosa efflux pumps did not change when the MexEF-OprN mutant grew exponentially (12).

Among the different changes associated with the overexpression of MexEF-OprN, we noticed that elements from the nitrate (NO₃⁻) respiratory chain were unexpectedly overexpressed during aerobic growth. P. aeruginosa is a facultative anaerobe capable of utilizing nitrate as an electron acceptor under anaerobic conditions and reducing it to molecular nitrogen via nitrite, nitric oxide (NO), and nitrous oxide (35). Transcription of the genes encoding elements of the nitrate respiratory chain is carefully regulated to enable expression exclusively when the pathway is needed the most, that is, in response to low oxygen and to the presence of nitrate and/or nitrite (36, 37).

Since the expression of the nitrate respiratory chain constitutes a significant metabolic shift, we wondered whether the potential fitness costs associated with MexEF-OprN overexpression could be metabolically compensated for during growth under optimal laboratory conditions (rich medium in aerobiosis). Our results showed that a mutant overexpressing MexEF-OprN consumes more oxygen than does a wild-type P. aeruginosa strain. MexEF-OprN belongs to the resistance/nodulation/division (RND) family of efflux pumps (38). Given that RND-type efflux pumps are proton antiporters with many similarities to the ATP synthase responsible for ATP synthesis (39), it is conceivable that the observed increase in oxygen consumption is a compensatory mechanism to balance the constitutive influx of protons resulting from MexEF-OprN overexpression and the possible impact that this may have on cellular energetics. In support of this view, we found that under anaerobic conditions, overexpression of MexEF-OprN results in a decrease in the intracellular pH. Coincidentally, we observed that the overexpression of this system imposes a severe fitness cost under these conditions.

When growing under nonrestrictive-resource conditions, bacteria do not need to exploit all of their metabolic capabilities. The ability to utilize different metabolic configurations may explain why the overexpression of certain efflux pumps does not necessarily lead to a fitness cost under said conditions: the resulting metabolic burden is immediately compensated for by a metabolic shift. In this scenario, a metabolic burden can be detected exclusively under conditions where a metabolic shift cannot take place (anaerobiosis, in our case). This might be particularly relevant in the case of metabolically versatile organisms such as P. aeruginosa.

MATERIALS AND METHODS

Bacterial strains and growth media.

Bacterial strains used in this study are listed in Table 1. Cells were routinely grown in LB medium (Pronadisa). Unless otherwise noted, bacteria were grown in 25 ml of medium in 100-ml flasks, with agitation at 250 rpm at 37°C. Doubling times were calculated from cultures in duplicate on three different days (six samples in total).

TABLE 1.

Strains and plasmid used in this study

Strain or plasmid	Relevant genotype and/or phenotype	Source and/or reference
Strains
P. aeruginosa PT5	PAO1; wild type	Köhler laboratory collection, 17
P. aeruginosa PT149	PT5 nfxC; overproduces MexEF-OprN	Köhler laboratory collection, 38
P. aeruginosa PT637	PT149 mexE::ΩHg; does not produce MexEF-OprN	Köhler laboratory collection, 17
P. aeruginosa PT149 ΔnirS	PT149 ΔnirS; nitrate respiratory chain inactivated	This study
E. coli TG1	K-12 supE thi-1 Δ(lac proAB) Δ(mcrB hsdSM) (rK− mK−)	Martínez laboratory collection, 73
E. coli S17	recA pro hdsR RP4–2-Tc::MuKm::Tn7	Martínez laboratory collection, 74
Plasmid pEX18Ap	Apr; oriT+ sacB+; gene replacement vector with multiple-cloning site from pUC18	44

Spent medium was obtained after the PT149 mutant was grown for 24 h in LB medium. The culture was initially centrifuged for 1 h at 8,000 × g, followed by a second centrifugation round at 10,000 × g for 1 h at 4°C. The pH was adjusted to 7.0 with a 2 M NaOH solution, and the supernatant was filtered through 0.45-μm and 0.22-μm filters. Sterility tests were performed in all cases.

Growth under anaerobic conditions.

For anaerobic nitrate respiration experiments, 16- by 125-mm Hungate tubes capped with septum stoppers and open-top screw caps were used (Bellco Glass, Inc.). LB medium was sparged with N₂ gas for approximately 30 min and subsequently placed into an anaerobic chamber, where the Hungate tubes were filled and capped. A 1 M KNO₃ solution was prepared similarly. The filled tubes were autoclaved for sterilization. Nitrate was injected into each tube prior to inoculation at a final concentration of 10 mM.

The inoculum was prepared as follows for all experiments: 2 to 3 colonies were used to inoculate 2.5 ml LB medium and grown under regular aerobic conditions for approximately 8 h at 37°C at 250 rpm in test tubes. In order to obtain an inoculum with cells accustomed to nitrate respiration under anaerobic conditions, Hungate tubes were inoculated with 100 μl of the 8-h culture by using a syringe, and the cultures were allowed to grow for 15 h at 37°C without agitation. Hungate tubes were injected with the inoculum at an optical density at 600 nm (OD₆₀₀) of approximately 0.01. The cultures were incubated at 37°C without agitation. Sampling/inoculation was performed with a syringe, the tubes were carefully mixed to obtain a homogenous culture, and no more than 200 μl was extracted at a time.

Growth in a fermentor.

A 2.0-liter fermentor (Biostat MD) operated at a 1.5-liter volume was used. A preinoculum grown overnight in a flask was used to achieve an OD₆₀₀ of 0.01, and the culture was stirred at 650 rpm. For the “aerated culture” conditions, the culture was constantly bubbled with an air stream (∼20% oxygen) with an airflow rate set at 300 ml · min⁻¹. “Nonaerated” cultures were simply stirred without additional oxygen being added.

Real-time reverse transcription-PCR assay.

Cultures were grown in LB medium and harvested at mid-logarithmic phase at an OD₆₀₀ of 0.5 to 0.6 as described previously (12), unless noted otherwise. Fermentor-grown cultures were harvested at an OD₆₀₀ of 0.8 for logarithmic phase. RNA was isolated with RNeasy columns (Qiagen), as previously described (40). A high-capacity cDNA reverse transcriptase kit (Applied Biosystems) was used to synthesize cDNA. Fifty nanograms of initially isolated RNA was loaded per reaction mixture in triplicate inter- and intra-assays. Relative expression was calculated according to the 2^−ΔΔCT method (41), using the rpsL gene to normalize the expression level results. Primers used in this study are listed in Table 2.

TABLE 2.

Primers used in this study

Primer	Sequence	Annealing temp (°C)
nirQF	5′-GCGGTATCTGCTACCTGGAC-3′	58
nirQR	5′-TAGGACACCACCAGCATGAA-3′	58
norBF	5′-CTACAACCCGGAAAACCTCA-3′	58
norBR	5′-AGCCACTTCTCGATCACCTC-3′	58
nosZF	5′-ACGACGGCAAGTACCTGTTC-3′	58
nosZR	5′-AGAACACGTAGCGGGTATGC-3′	58
fhpF	5′-ATTTCTACCGCACCATGCTC-3′	58
fhpR	5′-AGTTCCTGCAACTGGTCGAT-3′	58
anrF	5′-TGAAGAAAGGCGAATTCCTG-3′	58
anrR	5′-CGGATAGGTCTCGGTATCCA-3′	58
dnrF	5′-GACGAGATCGAGACGCTTTC-3′	58
dnrR	5′-CATGATTCGCGAGAAGGTTT-3′	58
ccoN2F	5′-TCTACCACCTGATCCCGAAG-3′	58
ccoN2R	5′-CGACGAAGGAGTAGGTCAGG-3′	58
rpsLF	5′-GCAAGCGCATGGTCGACAAGA-3′	58
rpsLR	5′-CGCTGTGCTCTTGCAGGTTGTGA-3′	58
nirSFup	5′-CCCAAGCTTTCTCGGCCATCCGCCGATAGG-3′	67
nirSRup	5′-TAGACCATGCCATTTGGCACGCGGGTCTCAGTACAC-3′	74
nirSFdow	5′-GTGTACTGAGACCCGCGTGCCAAATGGCATGGTCTA-3′	76
nirSRdow	5′-CCCAAGCTTGTCAGGGCCAGGAACAGCAGG-3′	73

Measurement of oxygen consumption.

Oxygen uptake studies were performed by using an Oxygraph (Hansatech, United Kingdom) fitted with a Clark-type oxygen electrode, according to the manufacturer's instructions. PT149 and the wild type were grown in 25 ml, and the cultures were allowed to reach an OD₆₀₀ of 0.6 (equivalent to ∼10⁶ CFU/ml) or an OD₆₀₀ of 2.0 (equivalent to ∼10⁹ CFU/ml). A total of 2.0 ml of these cultures was used to measure respiration rates. Samples were lyophilized in order to obtain the dry weight. The consumption rate is expressed as nmol O₂ consumed · min⁻¹ · mg⁻¹ (dry weight) of cell culture.

Construction of PT5 ΔnirS and PT149 ΔnirS mutants.

The primers used to construct the nirS in-frame deletion by overlap extension PCR (42, 43) are shown in Table 2. FailSafe PCR mix (Epicentre Biotechnologies, Madison, WI) was used for all PCRs. The resulting 2-kb PCR deletion product was cloned into plasmid pEX18Ap (44) and mobilized into either PT5 or PT149 from Escherichia coli S17-1 by conjugation. Recombinants were obtained as previously described (44).

NO production and NO₃⁻ consumption measurements.

A Nitric Oxide Colorimetric Assay kit (Abcam, Cambridge, United Kingdom) was used to measure nitric oxide and nitrate. Cultures grown in LB medium under aerobic conditions were sampled every 2 h; cells were pelleted by centrifugation at 6,000 rpm for 30 min, and the medium was filtered with a 0.22-μm filter. An 85-μl volume of medium was used for the assay.

Competition experiments.

Competition experiments were performed as previously described (12). Briefly, the mutant strain (either PT149 or PT149 ΔnirS) and the wild type were grown individually overnight, and equivalent amounts of each strain were added to antibiotic-free medium. Mixed cultures were incubated at 37°C, and every 24 h, a 1:1,000 dilution was inoculated into fresh medium. Viable-cell counts were obtained by plating serial dilutions onto LB agar and onto LB agar supplemented with 15 μg/ml of chloramphenicol in order to differentiate resistant mutant cells from wild-type cells. The number of wild-type cells was calculated as the total number of bacterial cells minus the number of antibiotic-resistant cells, and the results were expressed as a percentage of drug-resistant cells with respect to the wild-type strain.

Measurement of internal pH.

Intracellular pH was measured with the fluorescent probe 5 (and 6)-carboxyfluorescein diacetate succinimidyl ester (cFDASE) (Invitrogen), as previously described (45). Cells were collected at an OD₆₀₀ of 0.6 for aerobic and anaerobic cultures by centrifugation at 7,000 rpm for 10 min at room temperature. The resulting pellet was resuspended to a cell density of 10⁵ cells/ml in HEPES buffer (50 mM; pH 9.0) with 1 mM EDTA and incubated for 10 min at 30°C in the presence of 2 μM cFDASE. This was followed by a washing step and resuspension in 50 mM phosphate-buffered saline (PBS) buffer (pH 7.0) with 1 mM MgCl₂. To eliminate excess nonconjugated 5 (and 6)-carboxyfluorescein succinimidyl ester, cells were treated with 10 mM glucose in order to maintain optimal regulation of the internal pH. Cells containing the fluorescent probe were washed twice and resuspended in a buffer composed of 50 mM Tris, 50 mM morpholineethanesulfonic acid (MES), 140 mM choline chloride, 1 mM MgCl₂, 10 mM KCl, 10 mM NaHCO₃, and 0.5 mM CaCl₂ and adjusted with 6 M HCl at pH 7.0. Samples were placed into a black 96-well plate (Nunclon Delta Surface) and incubated at 30°C in a spectrophotometer (Infinite 200; Tecan). Anaerobic samples were incubated in an anaerobic jar, and the oxygen was eliminated by using the Atmosphere Generation system (AnaeroGen; Oxoid). All experiments were performed in triplicate.

Statistical analysis.

Student's t test was performed to determine the statistical significance of the results.

RESULTS

Expression of the nitrate respiratory chain increases under aerobic conditions upon MexEF-OprN overexpression.

A previous transcriptomic study showed that several genes encoding elements of the nitrate respiratory chain were expressed at higher levels in P. aeruginosa MexEF-OprN-overproducing mutant strain PT149 (12). To study in more detail whether or not MexEF-OprN overexpression was associated with increased expression levels of the nitrate respiratory chain under aerobic conditions, the expression levels of genes belonging to the nor, nos, nir, and nar operons were measured in the wild-type and mutant strains (Fig. 1). Increased expression levels were observed for all genes in the strain overexpressing MexEF-OprN (P < 0.05 in all cases), with the exception of the nar operon, for which no change was observed. The expression level of fhp, encoding a flavohemoprotein involved in nitric oxide detoxification under aerobic conditions, was also measured (46) and was found to be increased with respect to the wild type (P < 0.05). Expression of MexEF-OprN is regulated by the MexT transcriptional activator, which also regulates the expressions of additional genes (47, 48). To distinguish between the effects of the activity of the efflux pump and that of the regulator itself, we measured the expression levels of these genes in strain PT637, which has an active form of MexT but presents a truncated, nonactive form of the efflux pump. As shown in Fig. 1, no relevant changes in expression levels were observed for PT637, indicating that the overexpression of the nitrate respiratory chain under aerobic conditions is due to the increased activity of MexEF-OprN, independently of MexT.

FIG 1.

Expression of nitrate respiratory chain genes in mutant strains PT149 and PT637. The expression levels of genes coding for different components of the nitrate respiration chain were measured by real-time reverse transcription-PCR. Expression levels were increased exclusively in PT149 (black bars) and not in PT637 (white bars), indicating that it is an effect caused by the overexpression of the pump, independently of MexT.

PT149 consumes more oxygen than does its wild-type parental strain.

Expression of the P. aeruginosa nitrate respiratory chain is induced under anaerobic or low-oxygen conditions in the presence of nitrate or nitrite (36, 37). Expression can also be induced under low-oxygen conditions in the absence of added nitrate or nitrite in the medium (49). We explored the possibility that PT149 consumes more oxygen and thus creates a low-oxygen environment that cues the expression of the nitrate respiratory chain. To this end, we measured oxygen consumption in exponential and stationary phases by using a respirometer and found that PT149 consumes 286 nmol O₂/mg · min, whereas the wild-type strain consumes only 178 nmol O₂/mg · min in exponential phase (Table 3). In stationary phase, PT149 consumes 152 nmol O₂/mg · min, and the wild type consumes 98 nmol O₂/mg · min. Considering both growth phases, PT149 consumes oxygen approximately 60% faster than the wild-type strain (P < 0.05 in all cases). The increased respiration rate does not affect growth, as both strains exhibited similar doubling times (12).

TABLE 3.

Oxygen consumption

Strain	Mean oxygen consumption (nmol/g · min) ± SD in log phase	% of oxygen consumption of wild type in log phase	Mean oxygen consumption (nmol/g · min) ± SD in stationary phase	% of oxygen consumption of wild type in stationary phase
PT5	178 ± 8	100	98 ± 12	100
PT149	286 ± 17	160.6	152 ± 15	159.1
PT637	189 ± 19	106.1	108 ± 21	110.2
PT149 ΔnirS	294 ± 13	165.1	158 ± 17	161.2

Decreased environmental oxygen levels affect expression of the nitrate respiratory chain in PT149.

Cultures were grown in LB medium in a 2.0-liter fermentor (Biostat MD) in order to control the oxygen flow. Two conditions were used: aerated cultures, constantly bubbled with a stream of air (20% oxygen), and nonaerated cultures. We reasoned that the effect on the expression of the nitrate respiratory chain should be more pronounced when PT149 was grown with lower oxygen concentrations (nonaerated cultures). PT149, PT637, and the wild-type strain exhibited similar doubling times under both conditions tested (Table 4). All strains grew 58% faster in aerated cultures than in nonaerated cultures (P < 0.05 in all cases). The doubling time in nonaerated cultures was similar to that observed for cultures grown in flasks under constant agitation (Table 4).

TABLE 4.

Growth rates under different culture conditions

Culture condition(s)	Mean doubling time (min) ± SD
	PT5	PT149	PT637	PT149 ΔnirS
LB medium	38 ± 2	38 ± 3	38 ± 2	39 ± 4
Aerated LB medium (fermentor)	23 ± 2	25 ± 4	25 ± 5	24 ± 3
Nonaerated LB medium (fermentor)	41 ± 4	40 ± 3	42 ± 5	43 ± 6
Spent LB medium	255 ± 27	308 ± 24	252 ± 28	264 ± 31
Spent LB medium + 20 μM KNO3	252 ± 29	258 ± 31	259 ± 27	271 ± 29

As previously observed for flask cultures, expression levels of the nitrate respiratory chain genes increased in nonaerated cultures of PT149 in exponential phase compared to the wild type; however, the expression level was significantly higher under these conditions (Fig. 2A). The nir operon was upregulated 100-fold with respect to the wild type. Expression of the nor and nos operons increased 2,500-fold and 600-fold, respectively, while fhp expression increased 250-fold compared to the wild type (P < 0.05 in all cases). No changes were observed for PT637, further confirming that the observed effect is due solely to the activity of MexEF-OprN and not to a direct effect of the MexT regulator.

FIG 2.

Differential expression of nitrate respiration genes and genes involved in microaerobic functions under nonaerated conditions in a fermentor. Expression levels of the nitrate respiration genes (A) and genes involved in microaerobiosis (B) were increased under nonaerated conditions in PT149 (black bars) and did not change in PT637 (white bars) during exponential phase. PT637 fold changes in panel A are in the range of 1 in all cases.

Expression levels of genes involved in microaerobic growth are increased in PT149 cells growing under aerobic conditions.

The untimely expression of the nitrate respiratory chain suggests that the MexEF-OprN-overexpressing mutant perceives its environment as being microaerobic. Increased expression levels of additional genes normally expressed under low-oxygen conditions would be expected for PT149. One such gene is the global transcriptional regulator ANR, which is required for the expression of the nitrate respiratory chain. The ANR expression level increases with low oxygen levels (50); a 3-fold-increased expression level (P < 0.05) was observed for PT149 (Fig. 2B). To further verify that PT149 cells were subject to a self-created microaerobic environment, we measured the expression levels of the high-oxygen-affinity terminal oxidase Cbb₃-2. Expression of the cbb₃-2 genes (PA1557 to PA1552) is regulated by ANR under low-oxygen conditions (51). We found that the expression level of PA1557, encoding the ccoN2 subunit, was increased by 3-fold (P < 0.05) compared to the wild type (Fig. 2B). No changes in anr and ccoN2 expression levels were observed for the aerated PT149 culture.

MexEF-OprN overproduction impairs P. aeruginosa fitness in anaerobiosis.

Anaerobic LB medium was amended with 10 mM nitrate, and bacterial growth was monitored for 10 h. No differences in growth between PT149 and the wild type were observed during the first 4 h of incubation. However, a decline in the growth ability of PT149 was observed after 6 h (Fig. 3). The expression levels of genes of the nitrate respiratory chain along with the fhp gene were assessed under these conditions. Unlike the results obtained during aerobic respiration, no change in expression was observed for PT149 compared to the wild type, most likely because efficient induction of the genes is achieved under the proper conditions for both strains. Overexpression of MexEF-OprN in PT149 was confirmed under these conditions.

FIG 3.

Growth of wild-type (wt) strain PT5 and the PT149 mutant during anaerobic nitrate respiration. Strains PT5 and PT149 were grown in LB medium supplemented with 10 mM nitrate. Growth was estimated by viable-bacterial-cell counts by extracting no more than 200 μl of culture every 2 h and plating dilutions onto LB plates. The graph was built by using data from an average of two independent experiments comprising two biological replicates each.

The growth defect observed during anaerobic nitrate respiration suggests that MexEF-OprN overexpression affects the fitness of PT149 under these conditions. Competition experiments between PT149 and the wild type were conducted as previously described (12). Unlike the results obtained during aerobic growth in LB medium, we found that by day 3, the wild type had outcompeted PT149 (Fig. 4).

FIG 4.

Competition experiment between wild-type strain PT5 and the PT149 mutant during anaerobic nitrate respiration in nitrate-containing LB medium. The medium was supplemented with 10 mM nitrate. Serial dilutions were plated onto LB plates to estimate the total population (PT5 plus PT149) and on LB plates with chloramphenicol (15 μg/ml) to estimate the PT149 population. The graph was built by using data from an average of two independent experiments comprising two biological replicates each. After 4 days of coculture, the mutant was displaced by the wild-type strain, confirming that oxygen is important to avoid the fitness cost caused by overexpression of MexEF-OprN.

PT149 consumes more nitrate and produces higher levels of nitric oxide than the wild type.

Aerobic denitrification has been reported for P. aeruginosa under specific conditions (52). The use of this respiratory chain in a medium with low nitrate levels, such as LB medium, may deplete this compound and convert it to nitric oxide. Nitrate consumption and nitric oxide (NO) production were hence measured in LB medium under aerobic conditions in order to determine whether, in addition to presenting increased levels of their corresponding mRNAs, the nitrate respiratory chain was indeed more active in PT149 than in its parental wild-type strain. To the best of our knowledge, the amount of nitrate present in LB medium has not been reported to date. Therefore, we started out the experiment by measuring the amount of nitrate present in LB medium (Pronadisa) and determined an average concentration of 14 ± 2 μM. Nitrate consumption was determined by measuring the amounts of nitrate present in PT149 and wild-type culture supernatants at different time points. After 24 h, PT149 consumed all the available nitrate, while the wild type consumed approximately 50% of the initial concentration (Fig. 5A). These differences were statistically significant (P < 0.05). In agreement with these results, we found that PT149 produces at least twice (P < 0.05) the amount of nitric oxide as the wild type (Fig. 5B). As a control, we deleted the nirS gene, encoding the NirS nitrite reductase (53), in PT149 and assessed nitric oxide production in the resulting strain. No nitric oxide was detected in the PT149 ΔnirS mutant. It was previously stated that nitrosative stress positively regulates MexEF-OprN expression (54). Whether or not the enhanced production of nitric oxide by PT149 might produce positive feedback to further increase MexEF-OprN expression remains to be established.

FIG 5.

NO₃ consumption and NO production by different mutants used in this study. NO₃ and NO concentrations were measured with a Nitric Oxide Colorimetric Assay kit (Abcam, Cambridge, United Kingdom). (A) After 24 h, the PT149 mutant consumes the nitrate present in LB medium entirely, while the PT637 mutant and wild-type strain PT5 consume approximately one-half of the nitrate available. The PT5 ΔnirS and PT149 ΔnirS mutants do not consume nitrate. (B) In logarithmic phase, PT149 produces more nitric oxide than the mutants that do not overexpress MexEF-OprN. These results indirectly indicate that the denitrification pathway is indeed active in PT149.

The absence of nitrate impairs the fitness of PT149.

The nitrate concentration present in regular LB medium (14 ± 2 μM) does not support significant growth during anaerobic nitrate respiration (55), and the contribution to energy generation should be minimal under fully aerobic conditions (52). Nevertheless, we decided to assess whether the presence of nitrate in LB medium contributes to the fitness of PT149 when grown under aerobic conditions. Nitrate-depleted LB medium was obtained from PT149 cultures grown for 24 h. The spent medium was properly filtered, and the pH was adjusted to 7.0. The growth rate of PT149 in spent medium was 27% lower (P < 0.05) than that of the wild-type strain (Table 4). PT149 displayed a growth rate similar to that of the wild type when the spent medium was amended with 20 μM nitrate, thus confirming that the observed effects were related strictly to nitrate depletion and not to the absence of additional nutrients or to the presence of secreted metabolites.

Competition experiments were conducted with spent medium under aerobic conditions to explore whether the absence of nitrate affects the fitness of PT149 during growth in coculture. We found that after 24 h, the PT149 population percentage had been reduced to 10%, and it was completely displaced by the wild type on day 4 (Fig. 6A). On the other hand, when the spent medium was supplemented with 20 μM KNO₃, PT149 was not outcompeted by the wild type after 7 days of coculture. However, we observed that the population of PT149 decreased every day, thus indicating that the addition of 20 μM KNO₃ is not enough to fully eliminate the fitness cost caused by the overexpression of MexEF-OprN.

FIG 6.

Competition experiments under conditions unfavorable for denitrification. The relative fitnesses of the PT149 and PT149 ΔnirS strains in comparison with the wild-type strain were estimated by competitive growth in cocultures as a percentage of mutant cells present in the culture at each time point. (A) The absence of nitrate in the medium produces a fitness cost in the mutant that overexpresses MexEF-OprN. (B) In order to test the possibility that the inactivation of the nitrate respiratory chain may cause a metabolic burden, the PT149 ΔnirS and wild-type strains were grown in coculture. After 6 days of coculture, the PT149 ΔnirS strain was completely displaced by the wild type.

The inactivation of the nitrate respiratory chain impairs the fitness of PT149.

As mentioned previously, the nitrate respiratory chain is carefully regulated to enable expression only in response to low oxygen and to the presence of nitrate and/or nitrite (36, 37). Our previous results demonstrated that in nitrate-depleted LB medium, the PT149 mutant grows slower, and it is displaced by the wild-type strain in competition experiments. In order to test the effect that the nitrate respiratory chain has on the compensation of the potential fitness cost associated with MexEF-OprN overexpression in the PT149 mutant, we conducted a competition experiment between the PT149 ΔnirS mutant (obtained as described in Materials and Methods) and the wild-type strain. We found that after 3 days, the PT149 ΔnirS population had been reduced to 20%, and it was completely displaced by the wild-type strain on day 6 (Fig. 6B). This result confirms that the nitrate respiratory chain plays a role in avoiding the fitness cost caused by overexpression of the MexEF-OprN system.

PT149 shows a decrease in the internal pH under anaerobic conditions.

RND-type efflux pumps derive their energy from the proton motive force (56). Constant proton influx in RND-overexpressing mutants might cause acidification of the bacterial cytoplasm. In order to explore this possibility, we measured the internal pH under different culture conditions. Differences were minor, if any, when PT149 was grown in LB medium under aerobic conditions compared with the wild-type strain (Fig. 7A). However, the internal pH of PT149 dropped down to 6.0 under anaerobic conditions (Fig. 7B). These observations indicate that the increase in oxygen consumption might be necessary to maintain internal pH homeostasis upon MexEF-OprN overexpression.

FIG 7.

Internal pH variation under different culture conditions. To measure the internal pH, cells were incubated with the fluorescent probe 5 (and 6)-carboxyfluorescein succinimidyl ester. The internal pH in the wild-type strain and the PT149 mutant was measured under different conditions. (A) LB medium under aerobic conditions; (B) LB medium under anaerobic conditions.

DISCUSSION

A previous transcriptome analysis of PT149, a strain overexpressing MexEF-OprN, revealed increased expression levels of some genes involved in nitrate respiration in spite of the presence of oxygen in the medium (12). In this study, we uncovered the correlation between this transcriptional change and the overexpression of MexEF-OprN as well as the physiological consequences associated with the overexpression of this system under various conditions. The family of RND efflux pumps to which MexE belongs are proton antiporters (56). Overexpression of MexEF-OprN does not necessarily mean an increase in its antiporter activity, as this is achieved only in the presence of a proper substrate. In a previous study, we demonstrated that kynurenine, an intermediate of the PQS (Pseudomonas quinolone signal) biosynthetic pathway, is a natural substrate for MexEF-OprN and that a strain overexpressing this system extrudes l-kynurenine at higher levels than the wild type (12). These results indicate that under our experimental conditions, the strain that overexpresses this system also presents increased antiporter activity.

Transcriptional regulation of the nitrate respiratory chain is carefully controlled to allow expression when the pathway is needed the most, that is, under anaerobic or microaerobic conditions in the presence of nitrate or nitrite (35). In addition, the presence of nitrate in millimolar concentrations under aerobic conditions or microaerobic conditions alone is sufficient to induce the expression of the denitrification genes; the latter excludes the nar operon, encoding the Nar nitrate reductase (35, 49, 57).

We believe that the increased oxygen consumption rate observed for PT149 grown under aerobic conditions may lead to a decrease in environmental oxygen in the culture. PT149 cells sense this drop in oxygen as a cue to start expressing genes required for nitrate respiration. The fact that we observed increases in the expression levels of only the genes in the nir, nor, and nos operons, but not those in the nar operon, and increases in the consumption of nitrate and the production of nitric oxide, supports this view. Furthermore, both phenotypes are related directly to the overexpression of the MexEF-OprN efflux pump and are independent of the MexT transcriptional regulator, as no changes were observed in oxygen consumption, expression of the nitrate respiratory chain genes, or nitrate consumption and nitric oxide production in PT637, a PT149 derivative carrying active MexT and an inactive MexEF-OprN system (17).

Quorum-sensing molecules have also been implicated in regulating the expression of the nitrate respiratory chain genes (58–60). We have previously demonstrated that the overexpression of MexEF-OprN in the PT149 mutant results in the extrusion of l-kynurenine, a precursor of the PQS quorum-sensing signal, with a concomitant effect on the expression of a subset of the quorum-sensing regulon (12). To verify that the effect observed in this study was independent of the quorum-sensing signals, we restored PQS production in PT149 by adding anthranilate to LB cultures (12) and confirmed that overexpression of the nitrate respiratory genes prevailed under these conditions.

ANR is a global transcriptional regulator responsible for the adaptation of P. aeruginosa to oxygen-limited conditions (50, 61, 62). Interestingly, expression levels of anr and the cbb₃-2 high-affinity terminal oxidase were increased in PT149 exclusively when the cells were grown under low-oxygen conditions. P. aeruginosa is known to adapt to decreasing oxygen concentrations in a continuum by increasing the expressions of different sets of genes (49, 62). Our results indicate that although physiologically relevant, the impact of MexEF-OprN overexpression on oxygen consumption is restricted to a specific limit. These findings along with the capacity of P. aeruginosa to grow in low oxygen concentrations conferred by a branched respiratory chain that includes two oxidases with a high affinity for oxygen (35, 51, 63) would explain why PT149 did not show a fitness decrease when grown under microaerobic conditions (1% oxygen) (data not shown).

The MexEF-OprN system belongs to the RND family of efflux pumps. These pumps function as proton/drug antiporters, and the proton entrance serves to energize substrate extrusion (56). Overexpression of these systems would result in a constant proton influx into the cytoplasm in the presence of a proper substrate, such as the metabolic intermediate l-kynurenine (12) or 4-hydroxy-2-heptylquinoline, the precursor of the Pseudomonas quinolone signal (64), in the case of MexEF-OprN. Two possible scenarios could link this situation to the observed increase in oxygen consumption. The first one would encompass an increase in the respiration rate as a strategy to eliminate the proton accumulation that could lead to acidification of the cytoplasmic environment (65). RND-type efflux systems share certain features with the ATP synthase complex responsible for ATP synthesis, including the translocation of protons down a proton potential gradient (39, 66). In a second scenario, MexEF-OprN-mediated proton influx would mildly disturb the synthesis of ATP by perturbing the proton potential gradient, and an increase in oxygen respiration could ensue to compensate for this effect. These two scenarios are not mutually exclusive.

In this study, we have demonstrated that oxygen plays an important role in the maintenance of internal pH in mutant strain PT149. Although no clear differences in the internal pH were observed under aerobic conditions, in the absence of oxygen, the cytoplasmic pH dropped to values close to 6.0. These results support our hypothesis that a higher oxygen respiration rate in this mutant contributes to maintaining pH homeostasis. Unlike oxygen respiration, nitrate respiration does not consume protons. The fitness reduction displayed by PT149 under these conditions would be in line with the cytoplasm acidification theory. Neither the expression of the nitrate respiratory genes nor nitrate consumption increased in PT149 under these conditions. Since nitrate respiration is less energetically favorable than oxygen respiration (52, 67, 68), a slight perturbation of ATP synthesis could have a greater impact under these conditions.

The environmental nutrient and electron acceptor contents largely dictate the fitness cost associated with overexpression of MexEF-OprN. As mentioned above, expression of the genes encoding the Nar nitrate reductase requires the presence of nitrate (69). The trace amounts present in LB medium are sufficient to allow the activation of the nitrate respiratory chain, as indicated by the consumption of nitrate and the concomitant detection of nitric oxide (NO) in aerobic cultures. Not only does PT149 consume more nitrate than the wild type under these conditions, but the fitness decrease observed in experiments conducted with spent medium also indicates that overexpression and activation of the nitrate respiratory chain are detrimental to the mutant when grown in rich medium lacking nitrate. This respiratory chain is truncated upon the deletion of nirS, the gene encoding the structural unit of the Nir nitrate reductase (70). In the same way, the PT149 ΔnirS mutant exhibits a fitness decrease when competing with the wild type in LB medium, which further suggests that the activation of the nitrate respiratory chain is needed to avoid the fitness cost caused by the overexpression of the MexEF-OprN system. One possible approach for reducing the spread of antibiotic-resistant microorganisms is the development of strategies or drugs that amplify the fitness costs associated with the acquisition of resistance. In this regard, it is possible that inhibitors of the P. aeruginosa nitrate respiratory chain might reduce the fitness of mutants overexpressing MexEF-OprN and, consequently, the chances for the dissemination of this type of resistant mutant.

Conclusions.

The possible metabolic burden caused by the overexpression of RND-type efflux pumps and the subsequent extrusion of bacterial metabolites and/or other natural substrates has been studied and demonstrated for some pumps and some bacterial species (14, 71, 72). However, a fitness cost is not always readily detected. In addition to extruding bacterial metabolites, the overexpression of RND efflux pumps could affect the proton potential gradient, since these transporters are proton antiporters. This hypothetical effect and the impact that it may have on bacterial fitness have been widely overlooked. The transcriptional and physiological changes displayed by PT149 mirror those of a strain preparing to face less energetically favorable conditions and thus strongly suggest that overexpression of MexEF-OprN indeed affects the proton potential gradient in P. aeruginosa. However, in rich aerated medium, this effect is metabolically compensated for by increased oxygen uptake and triggering of the expression of the nitrate respiratory chain.

Given this organism's versatile energy metabolism (35, 63), it is not surprising that it is capable of overcoming the fitness costs associated with the acquisition of resistance without the need for acquiring compensatory mutations. On the other hand, it is worth mentioning that when grown under nonfavorable conditions, such as anaerobic conditions, fitness costs cannot be overcome, and MexEF-OprN-overexpressing P. aeruginosa is less fit than its wild-type parental strain. We are currently investigating whether this is a system-specific effect or if it can also be extended to other strains overexpressing different RND-type efflux systems. If this is a common situation, the observed metabolic rewiring can be used for developing drugs with improved efficacy against antibiotic-resistant bacteria.