Oxidative Stress Enhances Cephalosporin Resistance of Enterococcus faecalis through Activation of a Two-Component Signaling System

Abstract

Enterococcus faecalis is a low-GC Gram-positive bacterium, a normal resident of the gastrointestinal (GI) tract, and an important hospital-acquired pathogen. An important risk factor for hospital-acquired enterococcal infections is prior therapy with broad-spectrum cephalosporins, antibiotics that impair cell wall biosynthesis by inhibiting peptidoglycan cross-linking. Enterococci are intrinsically resistant to cephalosporins; however, environmental factors that modulate cephalosporin resistance have not been described. While searching for the genetic determinants of cephalosporin resistance in E. faecalis, we unexpectedly discovered that oxidative stress, whether from external sources or derived from endogenous metabolism, drives enhanced intrinsic resistance to cephalosporins. A particular source of oxidative stress, H₂O₂, activates signaling through the CroR-CroS two-component signaling system, a known determinant of cephalosporin resistance in E. faecalis. We find that CroR-CroS is required for adaptation to H₂O₂ stress and that H₂O₂ potentiates the activities of cephalosporins against E. faecalis when the CroR-CroS signaling system is nonfunctional. Rather than directly detecting H₂O₂, our data suggest that the CroR-CroS system responds to cell envelope damage caused by H₂O₂ exposure in order to promote cell envelope repair and enhanced cephalosporin resistance.

INTRODUCTION

Enterococci are ubiquitous inhabitants of the gastrointestinal tract in healthy animals, including humans. However, antibiotic-resistant enterococci are also major causes of hospital-acquired infections (1, 2) and therefore represent a serious public health problem. One well-known risk factor for the acquisition of enterococcal hospital-acquired infections is prior therapy with broad-spectrum cephalosporins (3), antibiotics that belong to the β-lactam family and interfere with cell wall biosynthesis by inhibiting the penicillin-binding proteins (PBPs) that cross-link peptidoglycan. Enterococci exhibit intrinsic resistance to cephalosporins, enabling them to proliferate and achieve abnormally high densities in the gastrointestinal (GI) tract of patients during cephalosporin therapy (4), thereby promoting dissemination to other sites, where they cause infection (5). This intrinsic cephalosporin resistance is a trait shared by essentially all isolates of Enterococcus faecalis; however, our understanding of the genetic and biochemical basis underlying cephalosporin resistance in enterococci remains incomplete. One important well-characterized factor that is required for cephalosporin resistance is a specialized low-affinity PBP (Pbp5) that does not get inactivated by cephalosporins and can therefore perform cross-linking of peptidoglycan to permit growth in the presence of the antibiotic (6, 7). Although required, Pbp5 is not sufficient for resistance, as mutations in other loci render E. faecalis susceptible to cephalosporin by mechanisms that have not been fully worked out (8–12).

Among the additional determinants required for intrinsic cephalosporin resistance in E. faecalis is the CroR-CroS two-component signal transduction system (TCS) (8). This TCS is composed of a histidine kinase, CroS, and its cognate DNA-binding response regulator, CroR. CroS autophosphorylates and subsequently transfers its phosphoryl group to CroR (8). CroR can bind promoters of genes in order to influence gene expression. Only 3 such promoters have been identified to date: salB, encoding a secreted protein of poorly understood function that does not appear to play a role in cephalosporin resistance (13), glnQ, the first gene of an operon predicted to encode a glutamine uptake apparatus (14), and croR itself (13). Although these known targets for CroR regulation do not offer any insight into the mechanism by which CroR-CroS influences cephalosporin resistance, mutants of E. faecalis lacking CroR exhibit a loss of intrinsic resistance to cephalosporins (8, 9), suggesting that as-yet-unknown members of the CroR regulon are important determinants for cephalosporin resistance. Moreover, treatment with antibiotics that impair peptidoglycan biosynthesis, but not other cellular processes, leads to the induction of the croR promoter (8), suggesting that the CroR-CroS system responds to the inhibition of peptidoglycan biosynthesis. Given the involvement of the CroR-CroS TCS in the control of cephalosporin resistance, it might be expected that environmental stimuli that activate CroR-CroS enhance enterococcal cephalosporin resistance. However, such an effect has not yet been described.

In recent years, the ongoing emergence of antibiotic-resistant bacterial pathogens has prompted efforts to more fully understand the mechanisms of action for basic antibiotics, with the goal of exploiting that knowledge to enhance the efficacies of existing drugs. Most classical clinically used antibiotics target cell wall biosynthesis, protein synthesis, or DNA metabolism to kill bacteria or inhibit their proliferation. Recent studies have reported that many antibiotics owe their lethal consequences, at least in part, to oxidative stress and damage incurred as a result of the antibiotic-induced generation of reactive oxygen species (15–20). Although this model remains controversial (21–23), some workers have proposed that enhancing the production of endogenous oxidants in order to potentiate the activities of existing antibiotics may be a viable therapeutic strategy to enhance our antibiotic arsenal (24). To be broadly applicable, such a strategy inherently relies on the notion that oxidants potentiate antimicrobial activities for all pathogens, although it remains unclear if this is the case. In E. faecalis, the reactive oxygen species superoxide has been proposed to mediate the lethal effects of several classes of antibiotics, based on the observation that an E. faecalis mutant lacking the ability to detoxify superoxide (i.e., ΔsodA mutant) is killed by antibiotics to which wild-type E. faecalis is normally tolerant, such as vancomycin and penicillin (25, 26). However, E. faecalis appears to deviate somewhat from the emerging oxidant potentiation model, as H₂O₂ accumulation, in contrast to that of superoxide, did not enhance killing by vancomycin (25). Whether superoxide or other types of reactive oxygen species play a role in the intrinsic cephalosporin resistance of enterococci is not known.

Here, we report the unexpected finding that oxidative stress enhances intrinsic cephalosporin resistance in the hospital-acquired pathogen E. faecalis. In particular, oxidative stress imposed by H₂O₂, whether from external sources or derived from endogenous metabolism, drives enhanced cephalosporin resistance via a pathway that requires signaling through the CroR-CroS two-component signal transduction system. We propose a model in which H₂O₂ treatment leads to cell envelope damage that is sensed by the CroS sensor kinase to trigger an adaptive response that promotes envelope repair and elevated cephalosporin resistance.

MATERIALS AND METHODS

Bacterial strains, growth media, oligonucleotides, and chemicals.

The strains and plasmids used in this study are listed in Table 1. The oligonucleotides used for plasmid construction and reverse transcription-quantitative PCR (qRT-PCR) were synthesized by Integrated DNA Technologies, Inc. The Escherichia coli strains were cultured in LB or half-strength brain heart infusion (hBHI) medium (Difco) at 30°C with shaking at 225 rpm. The E. faecalis strains were cultured in MM9YE medium (27) or Mueller-Hinton (MH) broth prepared according to the manufacturer's instructions (Difco). When required, antibiotics were added at the following concentrations: 100 μg/ml (for E. coli, in hBHI) or 10 μg/ml (for E. faecalis) erythromycin (Em) and 10 μg/ml chloramphenicol (Cm).

TABLE 1.

Strains and plasmids used in this work

Strain or plasmid	Relevant genotype or descriptiona	Reference or source
Strains
E. coli
TOP10	E. coli host for routine cloning	Invitrogen
LE392	E. coli cloning host for pCJK218-based plasmids	Promega
E. faecalis
OG1	Wild-type reference strain (MLST 1)	49
OG1RF	Rifampin- and fusidic acid-resistant OG1 derivative	50
T1 (SS498)	Wild-type (MLST 21), CDC reference strain	51
V583	Clinical isolate (MLST 6)	52
CK221	V583 ΔermB (Ems)	C. Kristich, unpublished data
DV75	OG1RF ΔmurAA2	10
DV87-4	OG1 Δrex	28
DV101-A2	OG1 Δnpr2	This work
DV106	DV87-4 (OG1RF_10894-5)2:rex	28
DV116	OG1 ΔperR2	This work
DV118	OG1 ΔhypR2	This work
JL339	OG1 Δpbp5	53
SB23	OG1 ΔcroR2 (in-frame deletion)	54
SB43	CK221 ΔcroS2 (in-frame deletion)	S. Kellogg and C. Kristich, unpublished data
SB45	CK221 ΔcroR2 (in-frame deletion)	S. Kellogg and C. Kristich, unpublished data
HH29	OG1RF ΔsodA::ermC (deletion/insertion of Emr)	H. He and C. Kristich, unpublished data
HH45	OG1RF ΔcroR ΔsodA:ermC (in-frame deletion of croR and deletion/insertion of Emr at the sodA locus)	H. He and C. Kristich, unpublished data
Plasmids
pCJK218	E. faecalis allelic exchange vector (Cmr)	28
pDV66-2	Δnpr2 (ΔG7-E441) in pCJK218	This work
pDV90	ΔperR2 (ΔK7-Q141) in pCJK218	This work
pDV91	ΔhypR2 (ΔH7-K288) in pCJK218	This work
pSLB31	E. faecalis expression plasmid carrying gene encoding CroR (Emr)	S. Kellogg and C. Kristich, unpublished data
pCJK4	Plasmid carrying promoterless lacZ (Emr)	28
pCJK106	Derivative of pCJK4 carrying croR′-lacZ	This work
pCJK205	Constitutive expression of lacZ (Emr)	C. Kristich, unpublished data

^aMLST, multilocus sequence type; Em^s, erythromycin susceptible; Em^r, erythromycin resistant; Cm^r, chloramphenicol resistant.

Construction of deletion mutants.

The deletion of OG1RF_10983 (npr), OG1RF_11304 (perR), or OG1RF_12241 (hypR) was performed using markerless exchange, as described previously (28). Briefly, derivatives of pCJK218 carrying in-frame deletion alleles of either OG1RF_10983 (pDV66-2), OG1RF_11304 (pDV90), or OG1RF_12241 (pDV91) were constructed using isothermal assembly of Gibson et al. (29). The deletion alleles were designed such that the first 6 (first 7 in the case of OG1RF_10983) and last 6 codons of each gene remained (97% of OG1RF_10983, 92% of OG1RF_11304, and 96% of OG1RF_12241 were deleted in the corresponding mutants). The deletion alleles were transferred to their native loci in the E. faecalis OG1 genome by allelic exchange with counterselection on MM9YE medium containing p-chlorophenylalanine (27) at 30°C for 2 days. Two independent mutants for each of the deletion strains were examined and found to exhibit identical phenotypes.

Antibiotic susceptibility assays.

MICs were determined after 24 h at 37°C with 2-fold serial dilutions of antibiotic in MH broth, supplemented as indicated. Microtiter plates were inoculated from stationary-phase cultures diluted to ∼5 × 10⁴ CFU/ml, and growth was monitored using a Bioscreen C plate reader. The anaerobic experiments were performed in a Coy anaerobic chamber with an atmosphere of 90% nitrogen, 5% hydrogen, and 5% carbon dioxide. Following sterilization, the growth medium used in the anaerobic experiments was transferred to the anaerobic chamber while still warm and allowed to equilibrate for a minimum of 24 h before use. The lowest antibiotic concentration that prevented growth was recorded as the MIC. Plasmid-bearing strains were cultured in the presence of 10 μg/ml Em for plasmid maintenance.

Quantitation of hydrogen peroxide.

H₂O₂ present in the supernatant of E. faecalis cultures was quantified using Amplex Red (Invitrogen), as described previously (28). Briefly, stationary-phase cultures were grown under static conditions in MM9YE supplemented with 0.3% glucose, diluted to an optical density at 600 nm (OD₆₀₀) of 0.01, and cultured with aeration at 37°C until mid-exponential phase. The sterile-filtered culture supernatants were analyzed using Amplex Red. A standard curve for the Amplex Red measurements was generated by diluting the standard in equivalent culture medium. Control experiments in which aliquots of the culture supernatants were treated with bovine catalase before Amplex Red analysis established that the Amplex Red signal was indeed due to H₂O₂. The experiments were performed a minimum of two times. The data are presented as the mean ± standard error.

Gene expression measurements.

Quantitative reverse transcription-PCR (RT-PCR) was performed on RNA extracted from exponentially growing cultures. Calculations of the fold change in gene expression used the Pfaffl method and gyrB as a reference gene. β-Galactosidase activity from a croR′-lacZ fusion was measured as previously described (28, 30). Briefly, stationary-phase cultures of plasmid-bearing strains were diluted to an OD₆₀₀ of 0.01 in MH broth supplemented with 10 μg/ml Em and cultured to exponential phase. The cells were permeabilized with SDS and chloroform, and β-galactosidase activity was measured with ortho-nitrophenyl-β-galactoside. The samples were analyzed in triplicate and the experiments performed a minimum of two times. The plasmid carrying the croR′-lacZ fusion (pCJK106) carries a 298-bp fragment encoding 250 bp upstream of croR and the first 16 codons of the croR open reading frame (ORF) upstream of the promoterless lacZ in pCJK4.

Time-kill analysis and growth curves.

Stationary-phase cultures were diluted to an OD₆₀₀ of 0.004 and grown in MH broth at 37°C with shaking until exponential phase (OD₆₀₀, 0.13 to 0.14). Aliquots were treated with 5 mM H₂O₂ or left untreated. Samples were removed at intervals and CFU determined by plating serial dilutions on MH agar. The data represent the geometric mean ± standard error from independent experiments. Growth curves were obtained in Erlenmeyer flasks at 37°C and 225 rpm. The stationary-phase cultures were diluted to an OD₆₀₀ of 0.01 into MH broth and the OD₆₀₀ measured every 30 min.

Cell envelope permeability by CPRG.

Stationary-phase cultures grown in MH broth supplemented with 10 μg/ml Em (for the maintenance of pCJK205) were patched onto MH agar plates supplemented with Em and 40 μg/ml chlorophenol red-β-d-galactopyranoside (CPRG) in the presence or absence of 1 mM H₂O₂ or antibiotics, as indicated, and incubated overnight at 37°C. To analyze CPRG hydrolysis in the liquid cultures, the cultures were grown at 37°C to stationary phase in MH broth supplemented with Em and 40 μg/ml CPRG in the presence or absence of 1 mM H₂O₂. Viable bacteria were enumerated on MH agar. CPRG hydrolysis was quantified by measuring the absorbance at 570 nm after removing the bacteria by centrifugation.

RESULTS

To identify new factors that influence intrinsic cephalosporin resistance in E. faecalis, we screened a library of transposon mutants (31) for isolates exhibiting altered resistance toward ceftriaxone, a broad-spectrum cephalosporin. One mutant exhibiting elevated resistance was found to carry a transposon insertion in OG1RF_12010, a gene that we previously characterized as encoding a redox-sensing transcriptional repressor known as Rex (28). Similar to the transposon mutant, an in-frame deletion mutant lacking OG1RF_12010 (Δrex) also exhibited substantially elevated resistance to ceftriaxone (Table 2) compared to that of the otherwise isogenic wild-type strain (E. faecalis OG1). Furthermore, the hyperresistant phenotype of the deletion mutant was complemented by providing Rex from an ectopic chromosomal locus (Table 2), indicating that loss of the Rex repressor results in elevated ceftriaxone resistance in E. faecalis.

TABLE 2.

Susceptibility analyses: MICs for ceftriaxone^a

Strainb	MIC (μg/ml) under the indicated condition
	Aerobic				Anaerobic
	Without H2O2	With H2O2c	500 U/ml catalase	With 35 mM paraquat	Without paraquat	With 35 mM paraquat
E. faecalis OG1
Wild-type	32	512	16	1,024	16	512
OG1 Δrex	512	512	128	512	32	NDd
OG1 Δrexcomp	32	ND	ND	ND	ND	ND
OG1 Δnpr	64	512	16	512	16	ND
OG1 Δpbp5	1	1	ND	1	1	1
OG1 ΔcroR	4	0.5	ND	1	4	4
OG1 ΔcroRcomp	32	512	ND	ND	ND	ND
E. faecalis V583
Wild-type	256	512	256	1,024	32	1,024
CK221 ΔcroR	2	0.5	ND	0.5	ND	ND
CK221 ΔcroS	2	0.5	ND	0.5	ND	ND
E. faecalis T1	16	512	8	512	16	512

^aMICs for ceftriaxone determined in MH broth, supplemented as indicated, after 24 h at 37°C for three distinct genetic lineages of E. faecalis. All values are the median MICs determined from a minimum of three independent experiments.

^bMutant strain designations are as follows: OG1 Δrex, DV87-4; OG1 Δrex_comp, DV106; OG1 Δnpr, DV101-A2; OG1 Δpbp5, JL339; OG1 ΔcroR, SB23; OG1 ΔcroR_comp, SB23(pSLB31); CK221 ΔcroR, SB45; CK221 ΔcroS, SB43.

^cH₂O₂ was included at a concentration 0.5 mM for the CK221 derivatives and 1 mM for all others.

^dND, not determined.

Accumulation of hydrogen peroxide drives cephalosporin resistance.

Previously, we showed that the Δrex mutant accumulates substantially more endogenously produced H₂O₂ in the culture medium than an otherwise wild-type strain (28). Because H₂O₂ diffuses across bacterial membranes (32), accumulation in the culture supernatants likely reflects, in part, an impaired ability of the mutant to detoxify the H₂O₂ generated during the course of normal metabolism. To determine if this H₂O₂ contributed to the elevated ceftriaxone resistance phenotype, we performed ceftriaxone susceptibility measurements on the Δrex mutant in the presence of exogenously provided catalase (to detoxify H₂O₂) and found that resistance was considerably reduced (Table 2). Additionally, susceptibility analysis performed under anaerobic conditions (where H₂O₂ cannot be generated) essentially eliminated the hyperresistant phenotype of the Δrex mutant. Together, these results suggest that H₂O₂ promotes enhanced resistance to cephalosporins in E. faecalis.

Rex is a transcription factor known to directly or indirectly regulate >20 genes involved in various aspects of metabolism (28). To exclude the possibility that the altered expression of these genes influenced ceftriaxone resistance in unknown ways, we sought to examine the role of endogenously produced H₂O₂ independent of the effect of derepression of the Rex regulon. To do so, we constructed an E. faecalis mutant (Δnpr) lacking the NADH-dependent peroxidase that catalyzes the reduction of H₂O₂ to H₂O, thereby contributing to H₂O₂ resistance in E. faecalis (33, 34). A mutant of E. faecalis strain JH2-2 lacking Npr was previously shown to accumulate H₂O₂ due to its inability to use Npr to efficiently detoxify H₂O₂ (34). Our Δnpr mutant also accumulated elevated levels of H₂O₂ under aerobic conditions to levels of about half of that observed in the supernatants of the Δrex mutant (Fig. 1A). The Δnpr mutation also resulted in enhanced ceftriaxone resistance compared to that of the otherwise isogenic wild-type strain (Table 2), although the effect was modest. This enhancement was due to H₂O₂, because the elevated ceftriaxone resistance of the Δnpr mutant was eliminated by catalase treatment or anaerobic growth (Table 2). The observation that the ceftriaxone resistance of the Δrex mutant was significantly higher than that of the Δnpr mutant is likely a result of the elevated level of H₂O₂ accumulation by the Δrex mutant compared to that of the Δnpr mutant. Additionally, H₂O₂ appears to impact the cephalosporin resistance of wild-type E. faecalis, because anaerobic growth of otherwise wild-type E. faecalis diminished ceftriaxone resistance (Table 2), for both the wild-type lab reference strain (OG1) and a clinical isolate (E. faecalis V583). Collectively, these results are consistent with the hypothesis that endogenous H₂O₂ stress promotes enhanced cephalosporin resistance.

FIG 1.

Hydrogen peroxide enhances cephalosporin resistance. (A) E. faecalis wild-type (WT) strain (OG1), Δnpr mutant (DV101-A2), and Δrex mutant (DV87-4) were grown to exponential phase at 37°C at 225 rpm in MM9YE medium supplemented with 0.3% glucose. Cell-free culture supernatants were collected and H₂O₂ measured using Amplex Red. The data represent the mean ± standard error from at least two independent experiments. H₂O₂ does not accumulate in the wild-type supernatant above background, presumably due to efficient detoxification by NADH peroxidase. nd, not detected. (B) Susceptibility assays to determine MIC for ceftriaxone after 24 h at 37°C were performed in the presence of various concentrations of exogenously added hydrogen peroxide in MH broth. Black bars, wild type; white bars, Δnpr mutant. The median MICs were determined from a minimum of three independent experiments; the y axis is plotted on a log₂ scale.

Exogenous hydrogen peroxide induces an increase in cephalosporin resistance.

Enterococci are likely to encounter H₂O₂ in a variety of natural environments, whether produced by competing microbes, environmental chemistry, or the innate immune system of eukaryotic hosts. To examine if exogenous H₂O₂ stress can promote ceftriaxone resistance, we performed susceptibility analyses in the presence of exogenously added H₂O₂, which revealed a dose-dependent increase in ceftriaxone resistance upon H₂O₂ exposure for wild-type E. faecalis (Fig. 1B). A similar effect was apparent at lower doses of H₂O₂ for the Δnpr mutant, consistent with the fact that the mutant is impaired at H₂O₂ detoxification and therefore will experience a higher effective dose of H₂O₂ than the wild-type strain at a given concentration of H₂O₂. Using a series of control experiments in which H₂O₂ was preincubated with ceftriaxone, we confirmed that the effect of H₂O₂ on cephalosporin resistance was not due to the inactivation of the antibiotic by H₂O₂ (see Fig. S1 in the supplemental material). In addition, we showed that the H₂O₂-dependent enhancement of ceftriaxone resistance is a trait that is common among diverse E. faecalis strains, including clinical isolates, as 2 distantly related clinical E. faecalis isolates (V583 and T1) both exhibited increased cephalosporin resistance in the presence of H₂O₂ (Table 2). H₂O₂-dependent enhancement of ceftriaxone resistance required continuous exposure to H₂O₂, because wild-type E. faecalis OG1 cells that had been precultured in the presence of 1 mM H₂O₂, washed, and subjected to MIC determination for ceftriaxone (in the absence of H₂O₂) did not exhibit the enhanced resistance phenotype (not shown).

H₂O₂-induced increase in resistance is cephalosporin specific.

If H₂O₂ triggered a global stress response, treatment with H₂O₂ might influence resistance to many classes of antibiotics with diverse cellular targets. To test this, we performed susceptibility measurements for a panel of antibiotics in the presence and absence of H₂O₂. Although H₂O₂ treatment elicited enhanced resistance to all expanded- and broad-spectrum cephalosporins tested (Table 3), no such effect was observed for noncephalosporin antibiotics, including others that target cell wall biosynthesis. Thus, H₂O₂ exposure does not appear to globally activate antibiotic resistance mechanisms via a nonspecific general stress response; rather, it appears to enhance cephalosporin resistance by triggering a specific signaling pathway to promote cephalosporin resistance.

TABLE 3.

H₂O₂- and paraquat-induced enhancement of resistance is specific for cephalosporins

Antibiotic by target	MIC (μg/ml) fora:
	Without H2O2	With 1 mM H2O2	With 35 mM PQb
Cell wall targets
Expanded- and broad-spectrum cephalosporins
Ceftriaxone	32	512	1,024
Ceftazidime	256	1,024	1,024
Cefuroxime	32	256	512
Other cell wall targets
Cefadroxil	64	64	128
d-Cycloserine	128	128	128
Bacitracin	64	64	64
Ampicillin	1	1	1
Vancomycin	2	2	2
Non-cell wall targets
Norfloxacin	4	4	4
Tetracycline	0.25	0.25	0.25
Erythromycin	1	1	1
Chloramphenicol	4	4	4
Spectinomycin	64	64	64
Kanamycin	64	64	64

^aMICs for different antibiotics for wild-type E. faecalis OG1 were determined in MH broth after 24 h at 37°C from a minimum of two independent experiments.

^bPQ, paraquat.

Other oxidants induce an increase in cephalosporin resistance.

To determine if treatment with other oxidative stress agents would also elicit an increase in cephalosporin resistance, we performed susceptibility measurements for ceftriaxone in the presence of subinhibitory concentrations of different oxidants: paraquat, phenazine ethosulfate (PES), cumene hydroperoxide, and bleach. All of the oxidants tested were capable of inducing an increase in ceftriaxone resistance of the E. faecalis wild-type strain (Table 4). A survey of antibiotic susceptibility in the presence of paraquat revealed that the pattern of paraquat-enhanced resistance paralleled that observed with H₂O₂ treatment (Table 3), suggesting that these oxidants act by perturbing the activity of a common pathway.

TABLE 4.

MICs of wild-type E. faecalis OG1 toward ceftriaxone in the presence of different oxidants

Oxidative stress	MIC (μg/ml)a
No treatment	32
Paraquat (35 mM)	512
PES (0.06 mM)b	256
0.001% cumene hydroperoxide	128
0.03% bleach	64

^aMICs for ceftriaxone determined in MH broth after 24 h of incubation at 37°C.

^bPES, phenazine ethosulfate.

Redox-cycling compounds, such as paraquat and PES, are well-known to continuously generate superoxide in aerobic environments by oxidizing redox enzymes and transferring the electrons to molecular oxygen (35). The superoxide so produced can subsequently undergo dismutation by superoxide dismutase (SOD) to produce H₂O₂. Formally, then, redox-cycling compounds might exert their effect on cephalosporin resistance by promoting the production of H₂O₂. In addition, redox-cycling agents are capable of directly oxidizing cofactors in redox-sensitive transcription factors to activate signaling pathways (36). To explore the mechanism by which redox-cycling compounds enhance cephalosporin resistance in E. faecalis, we performed susceptibility measurements for ceftriaxone in an E. faecalis mutant lacking the enterococcal superoxide dismutase (encoded by sodA), which is therefore unable to convert superoxide to H₂O₂. Paraquat treatment led to enhanced ceftriaxone resistance in the mutant (Table 5), indicating that paraquat does not promote cephalosporin resistance by driving H₂O₂ production. Furthermore, paraquat treatment led to enhanced ceftriaxone resistance for wild-type E. faecalis under anaerobic conditions (Table 2), indicating that redox-cycling compounds need not act by stimulating the production of reactive oxygen species at all. Instead, paraquat itself appears capable of directly stimulating resistance to cephalosporins. Collectively, these data provide evidence that E. faecalis possesses an oxidant-sensitive pathway, or regulator, that responds to oxidative stress by activating specific cellular pathways leading to cephalosporin resistance.

TABLE 5.

Susceptibility analysis of SOD mutants under aerobic conditions

Strain characteristic(s)	MIC (μg/ml)a
	Without paraquat	With 35 mM paraquat
Wild-type	16	1,024
ΔsodA	16	1,024
ΔcroR ΔsodA	4	8

^aMICs for ceftriaxone were determined in the absence and presence of paraquat in MH broth after 24 h at 37°C from a minimum of three independent experiments. Strains: wild-type, OG1RF; ΔsodA, HH29; ΔcroR ΔsodA, HH45.

The CroR-CroS signaling system is required for oxidant-mediated enhancement of cephalosporin resistance.

To investigate the mechanism by which oxidants lead to enhanced cephalosporin resistance in E. faecalis, we constructed E. faecalis mutants lacking one of 2 distinct transcription factors (HypR or PerR) previously implicated in the response of E. faecalis to H₂O₂ stress. HypR has been shown to directly activate the expression of genes involved in the response to oxidative stress (37, 38); the role of PerR is less clear, but a mutant lacking PerR exhibits enhanced survival upon H₂O₂ challenge (38). Neither of these regulators has a known role in enterococcal cephalosporin resistance. To probe for a role of these regulators in oxidant-enhanced cephalosporin resistance, we assessed the resistance phenotypes of the ΔperR and ΔhypR mutants in the absence and presence of H₂O₂. Ceftriaxone resistance of the two mutants was indistinguishable from that of wild-type E. faecalis under all conditions tested (see Table S1 in the supplemental material), indicating that neither of those factors is involved in mediating oxidant-enhanced cephalosporin resistance in E. faecalis.

We therefore considered that a regulatory system known to be involved in resistance to ceftriaxone might be involved in the oxidant response. The CroR-CroS two-component signal transduction system is required for the resistance of E. faecalis to cephalosporins (8, 9), but a mutation in croR does not influence resistance to many other classes of antibiotics (9). The CroR-CroS TCS has not been implicated in the oxidative stress response of E. faecalis, and furthermore, the molecular signal to which the CroS sensor kinase actually responds remains unknown. However, given the overlap between the subset of antibiotics for which CroR-CroS promotes resistance and those for which oxidants enhance resistance, we investigated if the CroR-CroS signaling system mediates the oxidant-dependent enhancement of cephalosporin resistance. We constructed mutants lacking CroR in 2 distinct lineages of E. faecalis and found, as expected, that they each exhibit impaired cephalosporin resistance (Table 2). This parallels what has been observed in derivatives of E. faecalis strains V583 (9) and JH2-2 (8) carrying mutations in croR. A susceptibility analysis of our ΔcroR mutants in the presence of either H₂O₂ or paraquat revealed that the oxidants no longer enhanced cephalosporin resistance, a defect that was complemented by the expression of croR from a plasmid (Table 2). In addition, a mutant lacking CroS, the cognate kinase for CroR, does not exhibit oxidant-enhanced cephalosporin resistance (Table 2), implying that signal transduction through the CroR-CroS TCS mediates the oxidant enhancement.

Rather than stimulate enhanced cephalosporin resistance, oxidant treatment of mutants with impaired CroR-CroS signaling potentiated the activity of ceftriaxone, rendering the mutants more susceptible (Table 2) and reinforcing the importance of CroR-CroS signaling. This effect does not appear to result from inherently elevated sensitivity to oxidants, as the ΔcroR mutant and wild-type E. faecalis exhibit comparable levels of killing by lethal doses of H₂O₂ in time-kill studies (see Fig. S2 in the supplemental material). Moreover, the oxidant-mediated potentiation of ceftriaxone is not due simply to the fact that the CroR-CroS mutants are inherently susceptible to cephalosporins, because another cephalosporin-susceptible E. faecalis mutant lacking Pbp5 (the low-affinity PBP required for cross-linking of peptidoglycan in the presence of cephalosporins) did not exhibit any such potentiation by oxidants (Table 2). Collectively, these findings indicate that the CroR-CroS signaling system plays a key role in mediating oxidant-inducible cephalosporin resistance.

To determine if the ceftriaxone potentiation observed upon treatment with the redox-cycling compound paraquat is due to the compound itself or to reactive oxygen species generated as a result of its redox-cycling activity, we tested for potentiation against the ΔcroR mutant upon paraquat treatment under anaerobic conditions. No potentiation was observed (Table 2), indicating that, unlike the enhancement of cephalosporin resistance in wild-type cells by paraquat, the potentiation of paraquat against the ΔcroR mutant requires redox-cycling in order to produce reactive oxygen species. To distinguish if this effect was due to superoxide or H₂O₂, we tested for potentiation against a double mutant lacking both CroR and SOD. No potentiation was observed with the double mutant (Table 5), indicating that H₂O₂ produced via SOD activity was responsible for paraquat-mediated potentiation of ceftriaxone under aerobic conditions against E. faecalis mutants with lesions in the CroR-CroS signaling system.

CroR-CroS TCS responds to H₂O₂ treatment to promote adaptation.

Although an E. faecalis mutant lacking CroR is not inherently more susceptible to lethal doses of H₂O₂ than the wild type (see Fig. S2 in the supplemental material), growth in the presence of subinhibitory levels of H₂O₂ revealed that CroR is required for adaptation to H₂O₂ stress (Fig. 2A). Wild-type E. faecalis treated with H₂O₂ exhibited a short lag compared to untreated cells but then grew robustly (albeit at a lower growth rate than that of the untreated cells). In contrast, the ΔcroR mutant exhibited a prolonged lag period and grew poorly in the presence of H₂O₂, suggesting that signaling through the CroR-CroS TCS is required to adapt for growth upon H₂O₂ stress.

FIG 2.

The ΔcroR mutant exhibits a growth defect in the presence of H₂O₂. Bacteria were grown in MH broth in the presence or absence of 1 mM H₂O₂ at 37°C and 225 rpm. The optical density at 600 nm was measured at intervals. (A) Wild-type (OG1), circles; ΔcroR (SB23), squares (closed symbols, without H₂O₂; open symbols, with H₂O₂) (B) Bacteria were grown in MH broth in the presence or absence of 35 mM paraquat at 37°C and 225 rpm. The optical density at 600 nm was measured at intervals. Wild-type (OG1), circles; ΔcroR (SB23), squares (closed symbols, without paraquat; open symbols, with paraquat). The representative results from at least two independent experiments are shown.

To test explicitly if signaling through CroR-CroS is activated in the presence of H₂O₂, we examined CroR-dependent gene expression in E. faecalis cells that had been cultured in the presence or absence of H₂O₂. Thus far, only a few CroR-dependent promoters have been identified in E. faecalis, none of which drives the expression of genes with an obvious role in either the oxidative stress response or cephalosporin resistance. However, many two-component signaling systems positively regulate their own transcription, and previous studies indicate that this is true of the CroR-CroS system (8, 14), enabling us to use activation of the croR promoter (which drives the expression of both croR and croS) as an indicator of signaling through the CroR-CroS TCS. Quantitative RT-PCR analysis of both the croR and croS genes revealed that the growth of wild-type E. faecalis in the presence of H₂O₂ led to elevated levels of croR/croS transcripts (Fig. 3A), consistent with activation of CroR-CroS signaling. The induction of croR/croS transcription in the presence of H₂O₂ was also observed using a croR′-lacZ reporter fusion (Fig. 3B). To test if H₂O₂-dependent induction specifically requires activation of CroR-CroS-dependent signaling, we examined gene expression in the ΔcroR mutant. No H₂O₂-stimulated increase was observed in the ΔcroR mutant (Fig. 3A and B), indicating that croR/croS induction in wild-type cells upon H₂O₂ treatment is the result of signaling through the CroR-CroS TCS and not another unknown regulator.

FIG 3.

H₂O₂ induces CroR-dependent gene expression. (A) RNA was purified from exponentially growing cultures of E. faecalis wild-type strain (OG1) or ΔcroR mutant (SB23) grown in MH broth with or without 1 mM H₂O₂ at 37°C and 225 rpm. The expression of the croS (left) and croR (right) genes was measured using qRT-PCR. The transcript levels were normalized to gyrB, and fold induction (treated versus untreated for each strain) was calculated. The data represent the mean ± standard error from two independent experiments, each performed in triplicate. Black bars, wild type; white bar, ΔcroR. (B) Cultures of E. faecalis wild-type strain (OG1) and ΔcroR mutant (SB23) carrying the croR′-lacZ fusion plasmid (pCJK106) were grown in MH broth supplemented with Em with (+) or without (−) 1 mM H₂O₂ or 35 mM paraquat at 37°C and 225 rpm. The samples were collected during exponential phase, and β-galactosidase activities (Miller units [MU]) were determined. The data represent the mean ± standard error from two independent experiments, each performed in triplicate. Black bars, wild type; white bars, ΔcroR. PQ, paraquat.

Unexpectedly, the ΔcroR mutant grew just as well as wild-type E. faecalis upon treatment with paraquat (Fig. 2B). In addition, the ability of paraquat to activate signaling through the CroR-CroS TCS was considerably reduced compared to that of H₂O₂ (Fig. 3B), suggesting that despite the requirement for CroR-CroS in the paraquat-mediated enhancement of cephalosporin resistance, the activation of CroR-CroS signaling may not be the primary mechanism by which paraquat acts. Thus, although the oxidants H₂O₂ and paraquat have a common effect on cephalosporin resistance in E. faecalis, the mechanisms underlying that effect may be distinct.

H₂O₂ treatment results in damage to the cell envelope.

The CroR-CroS TCS is activated when E. faecalis cells are treated with antibiotics that impair peptidoglycan biosynthesis (8). Because H₂O₂ treatment also drives the activation of CroR-CroS (Fig. 3), we hypothesized that H₂O₂ exposure results in damage to the cell envelope that is detected by the CroS kinase to activate signaling. To probe for such cell envelope damage, we used a recently described small-molecule permeability assay (39) that relies on the access of intracellular β-galactosidase to an otherwise nonpermeable substrate. β-Galactosidase cleaves the substrate chlorophenol red β-d-galactopyranoside (CPRG) to release the colored chlorophenol red chromophore. In wild-type cells, CPRG is excluded from cells by the permeability barrier of the cell envelope so it cannot be cleaved efficiently by intracellular β-galactosidase. However, if circumstances arise in which the cell envelope exhibits enhanced permeability (such as a mutation that compromises the envelope barrier), CPRG can penetrate the cells and be cleaved by β-galactosidase to yield the red product (39).

To test the CPRG assay as a probe for cell envelope integrity in E. faecalis, we examined the growth of wild-type E. faecalis constitutively expressing lacZ on agar plates supplemented with CPRG. In the absence of cell wall stress, wild-type E. faecalis does not produce any red color (Fig. 4A), indicating that CPRG does not penetrate the cells and the envelope barrier is intact. Supplementation of the medium with subinhibitory levels of antibiotics that impair cell wall biosynthesis (ceftriaxone or bacitracin) enabled CPRG to permeate the cells and led to CPRG hydrolysis (Fig. 4A), consistent with impaired cell envelope integrity. In contrast, supplementation of the medium with a subinhibitory level of kanamycin, an antibiotic that does not target cell wall biosynthesis, did not lead to CPRG hydrolysis. Finally, the growth of an E. faecalis mutant known to possess compromised envelope integrity (the ΔmurAA mutant [10]) yielded CPRG hydrolysis in the absence of any antibiotic treatment (Fig. 4A). Collectively, these findings establish the effectiveness of CPRG as a probe of envelope integrity in E. faecalis.

FIG 4.

H₂O₂ treatment results in damage to the cell envelope. (A) E. faecalis strains carrying pCJK205 were grown on MH agar supplemented with Em and 40 μg/ml CPRG. Where indicated, antibiotics or H₂O₂ was also included (4 μg/ml bacitracin, 4 μg/ml ceftriaxone, 8 μg/ml kanamycin, 1 mM H₂O₂). The experiment was repeated a minimum of two times. The plates were incubated overnight at 37°C. The strains were as follows: wild-type, OG1; ΔmurAA mutant, DV75; ΔcroR mutant, SB23. (B) E. faecalis strains were grown at 37°C to stationary phase in MH broth supplemented with Em and 40 μg/ml CPRG in the presence (+) or absence (−) of 1 mM H₂O₂. CPRG hydrolysis was quantified by measuring the absorbance at 570 nm after removing bacteria by centrifugation and normalized to viable CFU. The experiment was performed at least two times, and the data are represented as the mean ± standard deviation. The strains were as follows: wild-type, OG1; ΔcroR mutant, SB23. Statistical significance was evaluated by t test. *, P < 0.05 versus untreated wild type; **, P < 0.05 versus H₂O₂-treated wild type.

To probe for cell envelope damage upon H₂O₂ treatment, we examined CPRG hydrolysis in the presence or absence of H₂O₂. The ΔcroR mutant constitutively expressing lacZ produced a faint red color on CPRG plates, indicating that the cell envelope barrier may be slightly impaired in the absence of CroR function (Fig. 4A). However, in the presence of H₂O₂, CPRG is hydrolyzed robustly by the ΔcroR mutant, consistent with the hypothesis that the cell envelope has been substantially compromised by H₂O₂ treatment. Wild-type cells grown in the presence of H₂O₂ exhibited a faint red color, but this was considerably less than that for the ΔcroR mutant. To quantify the extent of CPRG permeability, we grew liquid cultures in the presence of CPRG and subjected them to H₂O₂ treatment, assessing CPRG hydrolysis by spectrophotometry. Very little CPRG was cleaved by wild-type E. faecalis in the absence of H₂O₂. H₂O₂ treatment resulted in an approximately 2.5-fold increase in CPRG hydrolysis, suggesting that H₂O₂ treatment of wild-type cells leads to slightly impaired envelope integrity (Fig. 4B). For the ΔcroR mutant, very little CPRG was cleaved in the cultures of the untreated cells, but H₂O₂ treatment led to a dramatic increase (approximately 50-fold) in CPRG hydrolysis (Fig. 4B). We infer that H₂O₂ treatment results in damage to the cell envelope of E. faecalis, and activation of the CroR-CroS TCS upon sensing this damage drives an adaptive biological response that facilitates cell wall repair or replacement in order to restore envelope integrity.

DISCUSSION

This work describes the unexpected finding that oxidative stress enhances cephalosporin resistance of the major opportunistic pathogen E. faecalis. Exogenous sources of multiple diverse oxidants, when present in the environment at subinhibitory levels, promote markedly increased cephalosporin resistance. In addition, E. faecalis is known to be a prolific producer of reactive oxygen species, such as superoxide and hydrogen peroxide (40–42), and our finding that cephalosporin resistance is decreased in an anaerobic environment where such ROS cannot be produced (Table 2) suggests that these self-produced oxidants are capable of stimulating their own cephalosporin resistance in an autoinducer-like fashion. We suggest that such autoinduction is autocrine in nature (i.e., the ROS-producing cell drives the enhancement of its own resistance) as opposed to being a quorum-sensing-like process (in which the accumulation of a signal in the external environment triggers a response in the population), as we did not detect significant accumulation of H₂O₂ in cultures of wild-type E. faecalis under growth conditions in which self-produced ROS is capable of driving enhanced cephalosporin resistance (Fig. 1 and Table 2).

Although several diverse oxidants were capable of stimulating cephalosporin resistance in E. faecalis, our data suggest that different oxidants may do so by distinct, but related, mechanisms. In particular, the CroR-CroS TCS is required for H₂O₂- and paraquat-mediated enhancement of cephalosporin resistance, yet only H₂O₂ treatment leads to a marked activation of signaling through CroR-CroS (Fig. 3). Moreover, CroR-CroS appears to be required to enable adaptation and growth in the presence of H₂O₂ but not paraquat (Fig. 2). One model to explain the lack of paraquat-mediated CroR-CroS activation might be that CroR-CroS possesses an alternative function in the cell, independent of its signaling role, which is altered by paraquat. Alternatively, we favor a model in which the oxidants are active against distinct intracellular targets in E. faecalis; in other words, we propose that H₂O₂ and paraquat cause damage to different molecules (or subsets of molecules) in the cell. Precedent exists for this notion; for example, paraquat, but not H₂O₂, is capable of oxidizing SoxR in E. coli (36). The idea that paraquat has one or more targets in E. faecalis that are distinct from the targets of H₂O₂ is also supported by our observation that paraquat treatment enhances cephalosporin resistance under anaerobic conditions (Table 2). Because H₂O₂ cannot be generated by redox cycling under anaerobic conditions, paraquat must exert its effect by acting directly on one or more cellular components. The requirement for CroR-CroS in the paraquat-mediated enhancement of cephalosporin resistance might suggest that one important target of paraquat action requires CroR-CroS for its synthesis. Redox-cycling compounds, including paraquat, are capable of directly oxidizing the iron-sulfur cluster of SoxR in E. coli (36), suggesting that iron-sulfur-containing enzymes or regulators might be targets for paraquat activity in E. faecalis. Thus far, no proteins known to possess iron-sulfur cofactors have been implicated in the intrinsic cephalosporin resistance of E. faecalis, so the identification of any such factor is a subject for future studies.

The CroR-CroS TCS is required for cephalosporin resistance, as mutants lacking CroR exhibit enhanced susceptibilities specifically to cephalosporin antibiotics (8, 9) (Table 2). How then does H₂O₂ stress trigger the activation of CroR-CroS signaling? It seems unlikely that either CroS or CroR directly senses H₂O₂, as neither protein contains any cysteine residues that might participate in H₂O₂ -mediated redox chemistry similar to the OxyR transcription factor (43, 44) or the ArcB-ArcA TCS of E. coli (45). Similarly, an inspection of the CroS and CroR amino acid sequences does not indicate the presence of any domains with the potential to bind redox-sensitive cofactors, such as the PAS domain (46). These observations suggest that the CroR-CroS TCS indirectly senses H₂O₂ stress, possibly as a consequence of H₂O₂-mediated perturbation of the normal cellular process or a physiological signal to which the CroR-CroS system responds. Although the specific molecular signal that is sensed by the CroS sensor kinase to activate signal transduction and gene expression has not been defined, antibiotics that impair peptidoglycan biogenesis at any of multiple steps in the synthetic pathway activate transcription from the croR promoter in a CroR-dependent manner (8), suggesting that CroS monitors some aspect of cell wall assembly to promote enhanced cell wall integrity. In light of our finding that H₂O₂ stress impairs the cell wall integrity of the ΔcroR mutant (Fig. 4), we propose a model in which H₂O₂ treatment damages the cell wall of E. faecalis, possibly by impairing the activity of one or more enzymes required for its synthesis. This damage triggers the activation of the CroR-CroS TCS and the subsequent expression of the genes in the CroR regulon, at least some of which are key downstream effectors that promote cell wall integrity and cephalosporin resistance. The identities of the CroR-dependent genes that provide cephalosporin resistance remain a mystery, as thus far, only 2 gene clusters other than CroR itself have been confirmed to be in the CroR regulon (13, 14), and neither of those has a known or predicted role in resistance. A connection between oxidative stress and cell wall integrity has been described in other Gram-positive cocci, although many of the molecular details remain unclear. For example, H₂O₂ induces the expression of the LytF cell wall hydrolase in Streptococcus mutans (47), and mutants of Streptococcus thermophilus with lesions in Pbp2b (a transpeptidase that cross-links peptidoglycan) or RodA (a membrane protein thought to help coordinate Pbp2b-mediated cell wall assembly) exhibit enhanced susceptibility to oxidative stress (48). Further work is required to determine if these phenomena are related in a mechanistic way.

Given the potent synergy observed between H₂O₂ and ceftriaxone in the ΔcroR mutant (Table 2), combining oxidative stress with the inhibition of CroR-CroS signaling may represent a viable therapeutic strategy enabling widely used cephalosporins to be effective against multidrug-resistant enterococcal infections. Therapeutics that target TCSs directly have not yet met with clinical success, but future work on the CroR-CroS signaling system will identify gene products in the CroR regulon that may be more attractive targets for therapeutic intervention than the TCS itself, while offering the same outcome. Lastly, our finding that oxidative stress enhances cephalosporin resistance indicates that an emerging model for antibiotic activity in which reactive oxygen species potentiate the activities of common antibiotics to enhance antibiotic lethality is not universal among bacteria. Thus, therapeutic strategies designed to stimulate the production of endogenous oxidants as a means of potentiating existing antibiotics (24) should be considered carefully, as they may have unintended and potentially troublesome consequences on other members of the microbiome.