|Home | About | Journals | Submit | Contact Us | Français|
Serratia marcescens is an important opportunistic pathogen in hospitals, where quaternary ammonium compounds are often used for disinfection. The aim of this study is to elucidate the effect of a biocide on the emergence of biocide- and antibiotic-resistant mutants and to characterize the molecular mechanism of biocide resistance in Serratia marcescens. A quaternary ammonium compound-resistant strain, CRes01, was selected by exposing a wild-type strain of S. marcescens to cetylpyridinium chloride. The CRes01 cells exhibited 2- to 16-fold more resistance than the wild-type cells to biocides and antibiotics, including cetylpyridinium chloride, benzalkonium chloride, chlorhexidine gluconate, fluoroquinolones, tetracycline, and chloramphenicol, and showed increased susceptibilities to β-lactam antibiotics and N-dodecylpyridinium iodide. Mutant cells accumulated lower levels of norfloxacin than the parent cells in an energized state but not in a de-energized state, suggesting that the strain produced a multidrug efflux pump(s). To verify this assumption, we knocked out a putative efflux pump gene, sdeAB, in CRes01 and found that the knockout restored susceptibility to most quaternary ammonium compounds and antibiotics, to which the CRes01 strain showed resistance. On the basis of these and other results, we concluded that S. marcescens gains resistance to both biocides and antibiotics by expressing the SdeAB efflux pump upon exposure to cetylpyridinium chloride.
Serratia marcescens is a gram-negative bacterium ubiquitous in nature that occurs in hospital environments and is a nosocomial and opportunistic pathogen (10). Infections with this bacterium often cause septicemia, meningitis, endocarditis, and wound infections (2, 10). The high and broad intrinsic resistance of this organism to various antibiotics makes S. marcescens infections difficult to treat (9, 11). Resistance of bacteria to antibiotics is generally attributable to alterations of the drug target, enzymatic modifications of antibiotics, or reduced drug accumulation as a result of efflux pump-mediated drug exclusion or a membrane barrier(s) (17, 19, 20). Expression of the resistance-nodulation-cell division (RND)-type efflux pump is a major mechanism for multidrug resistance in gram-negative bacteria (17, 18).
S. marcescens encodes at least three RND-type efflux pumps, SdeAB, SdeCDE, and SdeXY, which play an important role in the resistance of this organism to antibiotics. The SdeAB pump transports ciprofloxacin, norfloxacin, ofloxacin, chloramphenicol, and surfactants; SdeCDE transports novobiocin; and SdeXY transports erythromycin, tetracycline, norfloxacin, ampicillin, and biocides (5). The wild-type S. marcescens UOC-67 expresses only SdeAB and SdeCDE (2, 3).
Biocides have been widely used in hospitals for disinfection and in the food industry to kill bacteria that cause food poisoning. Despite extensive implementation of new strategies to control multidrug-resistant bacteria, current practices may not be satisfactory (4, 8, 23). Although the mechanisms of antibiotic resistance have been studied extensively, the molecular mechanism of resistance to disinfectants is poorly understood. It is likely that extensive exposure of hospital pathogens to biocides may cause the emergence of bacteria resistant to antibiotics (19, 20, 25), or vice versa.
In this study, we isolated an S. marcescens mutant resistant to cetylpyridinium chloride, a quaternary ammonium compound that is widely used for disinfection in hospitals all over the world. The genes and their products responsible for the cetylpyridinium chloride resistance were identified.
Bacterial strains and plasmids used are listed in Table Table1.1. Escherichia coli strains used were DH5α and S17-1. LB broth contains 1% tryptone, 0.5% yeast extract, and 0.5% NaCl per liter, adjusted to pH 7.0. NB broth contains 0.8% strength nutrient broth (Difco).
We manipulated cloned DNA according to standard methods (21). Chromosomal DNA from S. marcescens cells was isolated by the procedure described by Ausubel et al. (1). SOC medium contains 0.5% yeast extract, 2.0% tryptone, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl2, 20 mM MgSO4, and 20 mM glucose.
We amplified a 3,397-bp fragment containing the SdeB coding region by PCR using the following pair of primers: SdeB1Eco and SdeB4Hin (Table (Table2).2). The PCR mixture containing PrimeStar polymerase was subjected to the following reactions: 94°C for 2 min once; 30 cycles of 98°C for 10 s, 62°C for 5 s, and 72°C for 3.5 min; and finally 70°C for 10 min. The PCR product was treated with EcoRI and HindIII and was then inserted into pHSG398. The resulting plasmid was designated pHSG398-sdeB. To disrupt the chromosomal sdeB gene, the DNA was treated with PstI and BamHI. The DNA fragment containing the xylE gene was amplified using X1918 as a template with XylE1Pst and XylE4Bam (Table (Table2),2), and the product was treated with PstI and BamHI. The xylE DNA fragment was ligated to pHSG398-sdeB treated with PstI and BamHI. The resulting plasmid, pHSG398-sdeB::xylE, was treated with EcoRI and HindIII, and the DNA fragment containing the xylE gene was ligated with pGMS1 treated with EcoRI and HindIII, yielding pGMS1-sdeB. This plasmid was transferred to S. marcescens ATTC 13880 by conjugation via a mobilizer strain, E. coli S17-1, as reported earlier (16). The pGMS1-sdeB::xylE plasmid was inserted into the chromosomal sdeB gene by homologous recombination. Transconjugant cells showing both gentamicin resistance and sucrose sensitivity were selected on an LB plate containing 12.5 μg/ml of gentamicin or 10% sucrose. However, the transconjugant cells contained an extra DNA fragment coding for gentamicin-resistance. Therefore, cells without this unwanted fragment were obtained by selecting for sucrose resistance and gentamicin susceptibility, yielding strains HMS011 and CRes011 (16).
The MICs of biocides were determined by the standard broth dilution method (14). The biocide solution was diluted in a stepwise manner with fresh nutrient broth to the desired concentrations. A twofold-diluted disinfectant solution was prepared by mixing 0.5 ml of the biocide with an equal amount of nutrient broth containing 1.0 × 106 cells. The MIC of the biocide was determined visually after incubation at 37°C for 20 h.
The MICs of antibiotics were determined by the agar dilution method using Mueller-Hinton agar II (Difco) according to the method recommended by the Japanese Society of Chemotherapy (16).
S. marcescens ATCC 13880 cells were incubated in nutrient broth at 37°C for 18 h without shaking, and a 50-μl aliquot was transferred to tubes containing 2 ml of serially diluted cetylpyridinium chloride (1 μg/ml to 9.31 μg/ml). Tubes were incubated at 37°C for 18 h. A 50-μl aliquot of the cell suspension grown at the highest concentration of cetylpyridinium chloride was subcultured in broth containing 1-, 1.25-, 1.56-, 1.95-, 2.44-, and 3.05-fold higher concentrations of cetylpyridinium chloride than the tubes from which the cells were taken. Cells grown in the 13th cycle of the culture were streaked onto nutrient agar plates (24). Cells were incubated in the cetylpyridinium chloride-free broth for 60 generations of culture and were streaked onto cetylpyridinium chloride-free agar plates. One thousand colonies on the agar plate were tested for cetylpyridinium chloride susceptibility.
The method reported by Yoneyama et al. was adopted (27). Mid-log-phase cells grown in LB broth were harvested, washed once with a solution of 100 mM NaCl-50 mM sodium phosphate (pH 7.0), and resuspended in the same solution adjusted to 50 mg (wet weight) of cells per ml. After incubation in a shaking water bath at 37°C for 10 min, 50 μg/ml of norfloxacin was added. At the desired time intervals, a 0.6-ml sample was withdrawn and centrifuged at 4°C for 1 min, and the cells were washed twice with the uptake buffer kept at 4°C. The cell pellet was suspended in 1.5 ml of 100 mM glycine-HCl buffer (pH 3.0) and was incubated at 23°C for 1 h. Samples were centrifuged, and the supernatant was measured by a spectrofluorometer at excitation and emission wavelengths of 281 and 440 nm, respectively. The value obtained from the cells that were not exposed to norfloxacin was subtracted as background.
EZ::TN KAN-2 Transposomes (Epicenter, Madison, WI) were electroporated into electrocompetent S. marcescens cells, which had been prepared by washing log-phase cells (optical density at 600 nm, ~0.7) three times with ice-cold sterile 1 mM morpholinepropanesulfonic acid (MOPS) buffer containing 15% glycerol, and the cells were suspended in ice-cold sterile 1 mM MOPS buffer containing 15% glycerol. Electrocompetent cells were stored in a 43-μl single-use aliquot at −80°C until use. These cells were placed in a 2-mm-wide electroporation cuvette and exposed to microwaves of 25 μF, 100 Ω, and 12.5 kV cm−1 by a Gene Pulser 2 (Bio-Rad); then they were suspended in 1.0 ml of SOC medium and allowed to recover at 37°C for 1 h with shaking. Cells subjected to electroporation were plated onto LB agar containing 15 μg per ml of kanamycin. Transformants resistant to kanamycin were selected and patched on an LB master plate and an LB plate containing 0.5 μg/ml of norfloxacin as an indicator of increased norfloxacin susceptibility. Slow-growing colonies on the antibiotic-impregnated plates were selected from the master plate. The cells were tested for susceptibility to cetylpyridinium chloride.
The gene whose transposon (Tn) disruption compromised cetylpyridinium chloride resistance was identified by sequencing DNA flanking the transposon insertion site. Briefly, the chromosomal DNA was isolated by the procedure described by Ausubel et al. (1). The isolated chromosomal DNA was digested by PstI, self-ligated, and then used as a template for inverse PCR with the transposon-specific primers KAN-2GWF1 and KAN-2GWR1 (see Table Table2).2). Samples were subjected to the following PCR conditions with Ex Taq polymerase: 94°C for 3 min; 30 cycles of 94°C for 30 s and 68°C for 3.5 min; and finally 70°C for 10 min. The nucleotide sequence of the genomic DNA flanking the transposon element was determined by an automated DNA-sequencing system with primers KAN-2GWF1 and KAN-2GWR1, complementary to this insertion element, which were supplied with the mutagenesis kit. The nucleotide sequence flanking each transposon insertion site was referred to the S. marcescens strain Db11 genome library. The genomic sequence of S. marcescens was implemented by the Pathogen Sequencing Group at the Sanger Institute, Hinxton, Cambridge, United Kingdom (http://www.sanger.ac.uk/Projects/S_marcescens/). To determine the full-length coding sequence, oligonucleotide primers HASF1 and HASF2 (Table (Table2)2) were designed to amplify the hasF gene. Samples were subjected to the following PCR conditions with LA Taq polymerase: 94°C for 2 min; 30 cycles of 98°C for 10 s, 58°C for 30 s, and 72°C for 1 min; and finally 70°C for 10 min.
The cetylpyridinium chloride-resistant S. marcescens strain became increasingly resistant to cetylpyridinium chloride with the stepwise increase in its concentration. The cetylpyridinium chloride MIC for the resistant strain eventually obtained, CRes01, was 50 μg/ml (Table (Table3).3). This level of cetylpyridinium chloride resistance remained unchanged as the cells were incubated in the cetylpyridinium chloride-free medium for about 60 generations and 1,000 colonies were examined, suggesting that the mutation is stable. A single-step selection yielded mutants only eightfold resistant to cetylpyridinium chloride. The CRes01 cells showed resistance not only to cetylpyridinium chloride but also to chlorhexidine gluconate and benzalkonium chloride, yet their susceptibility to N-dodecylpyridinium iodide remained unchanged. Interestingly, this strain showed twofold higher susceptibility to dodecylbis(aminoethyl)glycine hydrochloride. Since the cetylpyridinium chloride-resistant CRes01 cells showed increased resistance to structurally unrelated compounds, such as chlorhexidine gluconate and dodecylbis(aminoethyl)glycine hydrochloride, but unchanged susceptibility to a structurally analogous compound, N-dodecylpyridinium iodide, it is unlikely that the resistance is due to the expression of a drug-modifying enzyme.
The antibiotic susceptibility of the CRes01 cells was tested, and the results could be grouped into three categories. The mutant showed (i) 2- to 32-fold-higher resistance to antibiotics including fluoroquinolones, tetracycline, chloramphenicol, pipemedic acid, and enoxacin than the parent strain; (ii) unchanged susceptibility to aminoglycosides and nalidixic acid; and (iii) 2 to 16 times higher susceptibility to β-lactam antibiotics than the parent strain (for details, see Table Table3).3). The multidrug-resistant nature of the mutant suggests that the resistance is attributable to elevated expression levels of multidrug efflux pumps. It is also conceivable, however, that the mutant gained multiple mutations, given its hypersusceptibility to β-lactams and dodecylbis(aminoethyl)glycine hydrochloride.
Earlier papers reported that the S. marcescens strain gained multiantibiotic resistance as it expressed RND-type efflux pumps (13, 19). To ascertain whether the multi-antibiotic-biocide resistance of CRes01 is due to the expression of efflux pumps, the intracellular accumulation of norfloxacin was determined in the presence and absence of an energy uncoupler, CCCP (carbonyl cyanide m-chlorophenyl hydrazone). The accumulation of norfloxacin in wild-type cells reached about 60 arbitrary units in approximately 30 min of incubation, and that in CRes01 cells reached about 10 arbitrary units, indicating that the mutant accumulated about one-sixth of the amount of norfloxacin in the parent cells (Fig. (Fig.1).1). However, these results do not necessarily mean that the low-level accumulation of norfloxacin in the mutant cells is attributable solely to the efflux pump. To see if this difference is dependent on cellular energy, 10 mM CCCP, a well-characterized proton conductor, was added at 10 min of incubation. The results shown in Fig. Fig.11 clearly demonstrate that both the parent and the mutant cells accumulated about 250 arbitrary units of norfloxacin in 20 min after the addition of CCCP. The values were about 5 and 25 times higher than those in the absence of CCCP in wild-type and mutant cells, respectively. These results also suggested that the influx of norfloxacin into the mutant in the presence of CCCP was comparable to that in the parent strain. These results therefore suggest that the mutant likely expressed the efflux pump responsible for multidrug-biocide resistance.
In order to identify the gene(s) responsible for cetylpyridinium chloride resistance in S. marcescens, cetylpyridinium chloride-susceptible mutants were isolated from CRes01 cells by the random mutagenesis method using EZ::TN KAN-2 transposomes (Epicenter, Madison, WI). About 10,000 kanamycin-resistant EZ-Tn5 insertional mutants were streaked onto LB agar plates containing 0.5 μg/ml of norfloxacin (1/64 of the MIC of norfloxacin for CRes01) as an indicator of increased norfloxacin susceptibility. Two mutants with increased norfloxacin susceptibility, designated CRNS71 and CRNS79, were selected.
To be sure that the mutants were susceptible to antibiotics and biocides to which CRes01 showed resistance, we determined the MICs of various antibiotics and biocides. The susceptibility profiles of CRNS71 and CRNS79 showed that the mutants were 8- to 64-fold more susceptible to antibiotics and biocides than CRes01 cells. Unexpectedly, CRNS71 and CRNS79 cells were 2- to 16-fold more susceptible to antibiotics and biocides than wild-type cells (Table (Table3).3). These results suggested that the wild-type strain might express the efflux pump weakly.
To identify the gene responsible for the biocide-antibiotic resistance, we determined the DNA sequence flanking the transposon insertion site in both CRNS71 and CRNS79 cells and found that the transposon was inserted into a 1,524-bp open reading frame, which encodes a polypeptide with 507 amino acid residues. Mutations in CRNS79 and CRNS71 were found at the 449th and 939th bp, respectively. The deduced amino acid sequence of the polypeptide showed 98% identity with the outer membrane protein HasF in S. marcescens SM365, an E. coli TolC homologue.
To confirm that hasF is responsible for the biocide and antibiotic resistance, a complementation experiment was carried out. pSTV29 carrying hasF was introduced into CRNS71 and CRNS79 cells, and antibiotic-biocide susceptibility was determined. The transformants, CRNS71(pSTV29-hasF) and CRNS79(pSTV29-hasF), showed fully restored resistance to biocides and antibiotics (Table (Table3).3). The result confirmed that the gene impaired in CRNS71 and CRNS79 is identical to hasF.
The TolC family outer membrane proteins are multifunctional transporters interacting with several different types of inner membrane (and periplasmic) transporters. Although HasF has been shown to interact with SdeAB (3), there is no guarantee that SdeAB is indeed involved in the biocide-antibiotic resistance reported in this paper and that other transport proteins are not involved.
To identify the transporter protein(s), which forms a functional unit with HasF, we constructed sdeB::xylE fusion strains in ATCC 13880 and CRes01, designated HMS011 and CRes011, respectively. These strains express catechol 2,3-dioxygenase in place of the sdeB gene. CRes011 cells carrying an sdeB::xylE fusion in CRes01 showed two- to fourfold hypersusceptibility to antibiotics and cetylpyridinium chloride compared with CRes01 cells (Table (Table4).4). This result indicated that CRes01 cells expressed the SdeAB efflux pump. On the other hand, the MICs of biocides and antibiotics for HMS011 appeared identical to those for ATCC 13880. These results imply that sdeB is responsible for the biocide-antibiotic resistance in CRes01 and that the SdeAB efflux pump is quiescent in the wild-type strain ATCC 13880. The latter is in striking contrast with the finding that SdeAB was expressed in another wild-type strain of S. marcescens, UOC-67. Yet CRes011 cells were still more resistant to norfloxacin, lomefloxacin, chloramphenicol, pipemidic acid, and cetylpyridinium chloride than ATCC 13880 cells. The results suggested that the CRes01 strain expressed SdeAB plus at least one additional efflux pump connected to HasF.
This study aimed to elucidate the mechanism(s) by which repeated exposure of S. marcescens to biocides affects resistance to both biocides and antibiotics. It revealed that repeated exposure of S. marcescens to cetylpyridinium chloride caused the emergence of mutants resistant not only to multiple species of biocides but also to structurally and functionally diverse antibiotics.
To elucidate the molecular mechanism of the biocide-antibiotic resistance of the mutants, we carried out transposon mutagenesis and found that the hasF gene, encoding an outer membrane duct protein, was impaired. The MICs of biocides and antibiotics were lower for the HasF mutant than for wild-type cells, suggesting that the wild-type cells expressed at least one efflux pump with which HasF forms a functional unit. Since it is unlikely that the HasF protein by itself functions as the xenobiotic exporter, it is reasonable to assume that is a partner subunit protein. To find the partner transporter of HasF, we carried out site-specific insertional mutagenesis in the sdeB gene of the sdeAB operon and found that the resulting sdeB-deficient mutant showed a level of biocide-antibiotic susceptibility in between that of the biocide-resistant mutant, CRes01, and that of the HasF knockout mutant. This result was interpreted to mean that HasF forms a functional unit not only with SdeAB but also with an unidentified RND transporter. A possible candidate for the additional HasF partner expressed in CRes01 cells could be the SdeXY transporter, since an earlier study revealed that SdeXY expressed in E. coli rendered the cells resistant to norfloxacin, tetracycline, ampicillin, benzalkonium chloride, chlorhexidine gluconate, and other compounds. The profile of resistance for SdeXY-positive cells is similar to that for the sdeB deletion mutant derived from cetylpyridinium chloride-resistant CRes01 cells. Thus, it is likely that SdeXY is expressed in the wild-type strain of S. marcescens and contributes to resistance. To test this hypothesis, attempts have been made to disrupt sdeXY in CRes01 cells, without success, suggesting that such mutants are lethal.
A question remaining unanswered is whether the biocide-antibiotic resistance reported in this paper is solely attributable to SdeAB or whether an unidentified factor is working under SdeAB. We cannot rule out the possibility that an unidentified factor may be associated with SdeAB. However, the fact that the SdeB knockout mutant, CRes011, restored biocide-antibiotic resistance suggested that the SdeAB pump plays a major role.
The next question to be asked is what genetic changes caused the strong expression of the sdeAB operon in CRes01. It is possible that a positive regulator, sdeR, located upstream of sdeAB, regulates its expression (3). Thus, we analyzed the nucleotide sequences of sdeR and the promoter region of sdeAB in CRes01 cells and found that they were identical to those in the parent strains. This result raises the question of whether or not the level of transcription of sdeAB was elevated in CRes01 cells. We carried out a reporter assay using an sdeB::xylE fusion expressing catechol 2,3-dioxygenase in place of SdeB. The sdeB::xylE mutant, CRes011, grown on an agar plate, had a yellow halo upon spraying of the substrate, confirming that CRes01 expressed a high level of SdeAB (data not shown). However, no yellow zone or colony was detectable for sdeB::xylE cells derived from ATCC 13880. These results implied that the sdeAB operon is likely regulated by an unidentified regulator in addition to SdeR.
Papers solely reporting resistance to biocides or antibiotics may be abundant, yet papers reporting the linkage of biocide and antibiotic resistances are less numerous.
Three groups of investigators have studied biocide and antibiotic resistance in strains exposed to biocides (6, 7, 12, 26). Two groups reported a biocide- and antibiotic-resistant Pseudomonas aeruginosa mutant derived from a parent strain whose preexisting RND-type efflux pump genes were knocked out (6, 7). Another group studied biocide and antibiotic resistance in Salmonella enterica and found that the mutant overexpressed the AcrAB efflux pump, which is constitutively expressed in the wild-type strain of S. enterica (12, 26). Therefore, it is likely that the mutant had an impaired regulator of the pump. This study is different from those reported in the two previous papers in two points: the parent strain used in this study is a wild-type strain, and our biocide- and antibiotic-resistant S. marcescens mutant overexpressed a new efflux pump, which is totally quiescent in the wild-type strain.
It must be stressed that such a mutant is resistant not only to biocides but also to a broad spectrum of antibiotics. Caution is needed to avoid the reckless use of disinfectants in hospitals, as pointed by Russell (20).
This research was partially supported by a Grant-in-Aid for Scientific Research (B and C) and a grant from the Asahi Glass Foundation.
Published ahead of print on 14 September 2009.