Serratia marcescens RamA Expression Is under PhoP-Dependent Control and Modulates Lipid A-Related Gene Transcription and Antibiotic Resistance Phenotypes

ABSTRACT

Serratia marcescens is an enteric bacterium that can function as an opportunistic pathogen with increasing incidence in clinical settings. This is mainly due to the ability to express a wide range of virulence factors and the acquisition of antibiotic resistance mechanisms. For these reasons, S. marcescens has been declared by the World Health Organization (WHO) as a research priority to develop alternative antimicrobial strategies. In this study, we found a PhoP-binding motif in the promoter region of transcriptional regulator RamA of S. marcescens RM66262. We demonstrated that the expression of ramA is autoregulated and that ramA is also part of the PhoP/PhoQ regulon. We have also shown that PhoP binds directly and specifically to ramA, mgtE1, mgtE2, lpxO1, and lpxO2 promoter regions and that RamA binds to ramA and lpxO1 but not to mgtE1 and lpxO2, suggesting an indirect control for the latter genes. Finally, we have demonstrated that in S. marcescens, RamA overexpression induces the AcrAB-TolC efflux pump, required to reduce the susceptibility of the bacteria to tetracycline and nalidixic acid. In sum, we here provide the first report describing the regulation of ramA under the control of the PhoP/PhoQ regulon and the regulatory role of RamA in S. marcescens.

IMPORTANCE We demonstrate that in S. marcescens, the transcriptional regulator RamA is autoregulated and also controlled by the PhoP/PhoQ signal transduction system. We show that PhoP is able to directly and specifically bind to ramA, mgtE1, mgtE2, lpxO1, and lpxO2 promoter regions. In addition, RamA is able to directly interact with the promoter regions of ramA and lpxO1 but indirectly regulates mgtE1 and lpxO2. Finally, we found that in S. marcescens, RamA overexpression induces the AcrAB-TolC efflux pump, required to reduce susceptibility to tetracycline and nalidixic acid. Collectively, these results further our understanding of the PhoP/PhoQ regulon in S. marcescens and demonstrate the involvement of RamA in the protection against antibiotic challenges.

INTRODUCTION

Serratia marcescens is a Gram-negative bacterium that belongs to the Enterobacteriaceae family. S. marcescens is widely distributed in the environment and in a wide range of host organisms. In addition to its ubiquity, S. marcescens is an emergent health-threatening nosocomial pathogen, due to the acquisition of antibiotic resistance mechanisms, the ability to survive for months on inanimate surfaces, and its resistance to conventional disinfection procedures (1–3). This bacterium has been declared by the WHO as a research priority to develop alternative antimicrobial strategies (4). In the last years, multidrug resistance strains outbreaks and high incidence in intensive and neonatal care units have increasingly been reported (5–7). Moreover, recent work identified S. marcescens as one of the three most abundant microbial species that colonize dysbiotic gut in patients with Crohn’s disease (8). S. marcescens can also develop either symbiotic or pathogenic interactions with plants and insects (9). The ability of S. marcescens to produce myriad extracellular enzymes and various secondary metabolites allows it to adapt to and survive in both hostile and changing environments (9). Two-component systems (TCS) are one of the most ubiquitous mechanisms by which bacteria generate adaptive responses to environmental or intrahost challenges. In these signal transduction systems, the activation of a sensor histidine kinase leads to autophosphorylation followed by transfer of the phosphoryl group to a cognate response regulator in an aspartate residue (10).

PhoP/PhoQ is a broadly conserved TCS among many pathogenic and nonpathogenic bacteria. In vitro, this system can be activated by acidic pH and antimicrobial peptides (APs) and is repressed by millimolar magnesium or calcium concentrations and by long-chain unsaturated free fatty acids (11–14). In various pathogenic bacteria it was demonstrated that PhoP/PhoQ has the ability to sense host intracellular signals and regulate bacterial lifestyle adaptation during infection (15). Although PhoP/PhoQ displays similar functions in regulating the virulence capacity of pathogenic bacteria, such as Salmonella enterica (16, 17), Mycobacterium tuberculosis (18), Yersinia pestis (19), Shigella flexneri (20), and extraintestinal pathogenic Escherichia coli (21), the set of regulated genes recruited under its control vary in a species-specific manner.

In our previous work, we have shown that in S. marcescens clinical isolate RM66262 (22), the PhoP/PhoQ system is involved in the adaptation of this bacterium to grow in scarce environmental Mg²⁺, at acidic pH, and in the presence of polymyxin B. Furthermore, we have shown that the PhoP/PhoQ system is implicated in the avoidance strategy that allows Serratia to survive and multiply inside epithelial cells (23).

The lipopolysaccharide (LPS), composed of lipid A, core, and O antigen, contains a molecular pattern recognized by the innate immune system, thereby promoting host defense responses (24). The ability of Gram-negative bacteria to modify the LPS is implicated in the avoidance of the host immune system and the resistance to killing by APs. In many bacterial species, modifications in the LPS that confer resistance to antimicrobial peptides are regulated by the PhoP/PhoQ system (25–27). In S. marcescens, it was demonstrated that the expression of the arn operon, which is involved in LPS modification, is under PhoP control (28). On the other hand, in Klebsiella pneumoniae, the transcriptional regulator RamA (resistance antibiotic multiple A) functions as an alternative regulator to PhoP to modulate the expression of the lpxC, lpxL-2, and lpxO genes, which are associated with lipid A biosynthesis (29). It has been reported that Klebsiella remodels its lipid A in vivo, in the infected lung tissues of mice. The lipid A species found in the lungs have modifications dependent on the PhoP/PhoQ-regulated oxygenase LpxO. In addition, an lpxO mutant is attenuated in vivo, highlighting the importance of this lipid A modification for the Klebsiella infection process (30).

In this study, we identified a recognition site for the PhoP regulator in the putative promoter region of the S. marcescens RM66262 ramA homolog. Furthermore, we also found two lpxO orthologous genes (which we have named lpxO1 and lpxO2). We characterized the regulation cascade that involves S. marcescens ramA and defined its participation in the regulation of the lpxO and mgtE genes and in the control of bacterial resistance to antibiotics. To our knowledge, this is the first report describing the recruitment of ramA under the PhoP/PhoQ regulon and the regulatory role of RamA in S. marcescens.

RESULTS AND DISCUSSION

S. marcescens ramA is a PhoP-regulated gene.

Previous reports demonstrated that the PhoP/PhoQ system is required for S. marcescens to tolerate the challenge of an environment deficient in Mg²⁺, of acidic pH, or of the presence of polymyxin B (23). We have also shown that within the host, this TCS is involved in preventing the intracellular delivery of the bacteria to degradative/acidic compartments (23). In search of S. marcescens RM66262 PhoP-regulated genes, we previously performed a bioinformatic search by using the MEME/MAST motif detection program (31, 32) and identified a set of genes that display putative PhoP-binding sites in their promoter regions. Among these genes, we found that the two S. marcescens mgtE orthologs displayed a conserved PhoP-binding motif and demonstrated that mgtE1 expression is PhoP dependent (23). Following this strategy, we were able to additionally detect a conserved motif for PhoP recognition in the promoter regions of genes encoding an AraC-type transcriptional regulator (which we have named ramA) and two lpxO orthologs (which we have named lpxO1 and lpxO2) (Fig. 1A). By sequence homology analysis, lpxO genes are predicted to encode proteins with oxygenase activity that would 2-hydroxylate specific lipid A acyl chains (33). In order to examine the PhoP-dependent expression of ramA, mgtE1, mgtE2, lpxO1, and lpxO2, we constructed reporter plasmids that harbor the gfp gene, which encodes the green fluorescent protein (GFP), under transcriptional control of the putative promoter regions (500 bp upstream of the translational ATG start) of each gene. Because ramA encodes a response regulator that belongs to the PhoP/PhoQ regulon, we carried out assays not only in the phoP background but also in the ramA and the double phoP ramA mutant background strains.

FIG 1.

(A) PhoP-binding sites in Serratia marcescens RM66262 genome. A consensus motif for the PhoP-binding site was generated using a collection of previously defined promoter regions of PhoP-regulated genes in S. Typhimurium and E. coli as a database and the MEME software tool. The logo (obtained by using WebLogo [weblogo.berkeley.edu]) shows the consensus motif for the PhoP-binding site, and the putative PhoP-binding site sequences with high scores identified in the S. marcescens RM66262 genome by MAST are listed at the bottom. The PhoP boxes are shown in boldface; arrows indicate orientation relative to each translational start site. The name of the gene and the distance (in base pairs) from the translational start site are also indicated. (B) Transcriptional expression of phoP, ramA, mgtE1, and mgtE2. Bacteria were grown for 16 h in N medium with 10 μM MgCl₂ for low-Mg²⁺ conditions or 10 mM MgCl₂ for high-Mg²⁺ conditions, in 96-well microplates, at 37°C with agitation. Transcriptional activity was calculated as the ratio of GFP fluorescence values and OD₆₀₀ (FU/OD₆₀₀) measured from the S. marcescens wild-type (WT), phoP, ramA, and ramA phoP strains carrying the PphoP-gfp, PramA-gfp, PmgtE1-gfp, and PmgtE2-gfp reporter plasmids. Means ± SDs from three independent experiments performed in duplicate in each case are shown. Statistical analysis was performed using two-way analysis of variance with Bonferroni’s posttest. **, P < 0.01; ****, P < 0.0001 (statistically significantly different from WT S. marcescens). ns, not significant.

As shown in Fig. 1B, in the wild-type (WT) background, ramA transcriptional levels are not affected by the Mg²⁺ concentration of the bacterial growth medium. However, under low-Mg²⁺ conditions, ramA expression is circa 50% downregulated in the phoP, ramA, and phoP ramA strains, while under high-Mg²⁺ conditions, ramA expression is PhoP independent but dependent on ramA integrity. These results suggest that (i) ramA expression is autoregulated, (ii) under low-Mg²⁺ conditions, ramA transcriptional activity could be repressed by an unknown PhoP-independent mechanism, or (iii) under low-Mg²⁺ conditions, integrity of both ramA and phoP is required for full ramA transcriptional expression.

In enterobacteria, PhoP drives expression from an array of promoters with sequence diversity, and in doing so, it coordinates the expression of a variety of gene products that are required in different amounts and/or for different extents of time, according to environmental characteristics (34). In S. marcescens the phoPQ operon is autoregulated, as was previously reported (23), and phoP transcription is repressed with increasing concentrations of Mg²⁺, and our results show that RamA is dispensable for phoP expression (Fig. 1B). As previously mentioned, we have shown that in S. marcescens, the transcription of mgtE1 is PhoP regulated in a Mg²⁺-dependent fashion (23) (Fig. 1B). In addition, our results show that mgtE1 expression depends on RamA as a positive regulator, independent of the Mg²⁺ concentration of the growth medium (Fig. 1B), indicating that mgtE1 transcriptional levels are dependent on the simultaneous presence of both PhoP and RamA.

Consistent with our in silico screening of putative Serratia PhoP-regulated genes, we also found that transcription of mgtE2 is differentially regulated at low/high-Mg²⁺ concentrations. As shown in Fig. 1B, the transcriptional level of mgtE2 was 2-fold higher at low Mg²⁺, either in the WT or in the ramA strain, than the levels detected for either phoP or ramA phoP mutant strains. At high concentrations of Mg²⁺, the transcription level of mgtE2 was not altered in the ramA phoP or ramA phoP strains compared to the expression levels obtained for the WT strain. In sum, the transcriptional levels of ramA were downregulated in the phoP mutant, under either high- or low-Mg²⁺ conditions, while the expression of ramA did not affect mgtE2 transcriptional levels. These results demonstrate that in contrast to mgtE1, mgtE2 is a PhoP-dependent, RamA-independent gene.

In order to examine if these differentially expressed genes were under direct control of PhoP, we performed electrophoretic mobility shift assay (EMSA) using purified recombinant PhoP protein. When DNA fragments containing the promoter sequences of ramA, mgtE1, and mgtE2 were incubated in the presence of PhoP, retarded bands were detected and their intensities were enhanced when the amount of PhoP was increased, with a concomitant intensity reduction in the band that corresponds to the unbound probe (Fig. 2A, left, and B). These results demonstrate that, as predicted, PhoP was able to directly interact with the promoter sequences of ramA, mgtE1, and mgtE2 that contain a PhoP-binding motif. A 25- to 100-fold excess of nonspecific nucA DNA fragment (a 441-bp DNA region that codes for the S. marcescens NucA nuclease) did not affect the interaction with the labeled probe (Fig. 2A, middle), while the shifted band was progressively lost when increasing amounts of unlabeled ramA promoter fragment were included in the mixture to compete with the labeled probe (Fig. 2A, right). This result indicates that the interaction of PhoP with the ramA promoter region was specific. In addition, the presence of competing nonlabeled nucA DNA fragment did not affect the interactions of PhoP with the labeled probes containing the promoter region of phoP, mgtE1, or mgtE2 (Fig. 2B). This result indicates that the interaction of PhoP with these promoter regions was specific. Because phoP transcription is autoregulated, the interaction of PhoP with the promoter region of phoP was used as a positive control for EMSA.

FIG 2.

PhoP and RamA interactions with ramA, mgtE1, and mgtE2 gene promoter regions of S. marcescens. (A and C) Electrophoretic mobility shift assays (EMSAs) were performed using different amounts of purified PhoP-6×His (A) or RamA-6×His (C). Target DNA was a ³²P-labeled PCR fragment that included the ramA promoter region (PramA). Binding specificity was assessed by competition reactions using 20 pmol of purified PhoP-6×His (A) or RamA-6×His (C) in which increasing amounts (50, 100, and 200 ng) of nonspecific (nucA; middle) or specific (PramA; right) unlabeled DNA template competed with labeled DNA for binding to PhoP-6×His (A) or RamA-6×His (C). (B and D) EMSAs were performed using nonlabeled PCR fragments carrying the complete phoP, mgtE1, and mgtE2 promoter DNA sequences and purified protein PhoP-6×His (B) or RamA-6×His (D) (10 or 20 pmol, as indicated). When indicated, a 441-bp DNA fragment was used as a nonspecific competitor.

To assess whether RamA is able to directly interact with the promoter regions upstream of ramA and mgtE1, EMSA was performed using purified recombinant RamA protein. When the probe containing the promoter region of ramA was incubated in the presence of RamA, a retarded band was detected and its intensity was enhanced with increasing amounts of RamA, with the concomitant intensity reduction in the lower band that corresponds to the unbound probe (Fig. 2C, left). A 25- to 100-fold excess of nonspecific nucA DNA fragment did not affect the interaction (Fig. 2C, middle), while the shifted band was progressively lost when increasing amounts of unlabeled ramA promoter fragment were included in the mixture (Fig. 2C, right), indicating that the interaction of RamA with the ramA promoter region was specific. These results show that RamA is able to directly bind to the promoter region of its own gene, ramA. When the promoter regions of mgtE1 and mgtE2 were incubated in the presence of RamA, no shift in the DNA probes was detected (Fig. 2D). This result suggests that mgtE1 regulation is under indirect control of RamA.

Overall, these results show that mgtE1, mgtE2, and ramA are members of the S. marcescens PhoP regulon. They also indicate that RamA is able to control its own expression in an autoregulatory positive loop and that PhoP and RamA simultaneously control mgtE1 transcription, while mgtE2 expression depends only on PhoP.

The PhoP/PhoQ system and RamA regulate genes involved in lipid A modifications.

In order to analyze the influence of PhoP and RamA on the expression of lpxO1 and lpxO2, we determined their transcriptional expression levels detecting fluorescence over time from the PlpxO1-gfp or PlpxO2-gfp transcriptional reporter. As shown in Fig. 3, lpxO1 transcript levels were not affected by the Mg²⁺ concentration of the bacterial growth medium. However, under low-Mg²⁺ conditions, lpxO1 expression was circa 50% downregulated in the phoP, ramA, and phoP ramA strains, indicating that both PhoP and RamA are required for their full expression. In contrast, under high-Mg²⁺ conditions, lpxO1 expression was independent of either PhoP or RamA. Under low-Mg²⁺ conditions, the lpxO2 transcript level was at least 4-fold decreased in the phoP, ramA, and ramA phoP background strains compared to wild-type levels, indicating that both PhoP and RamA are required for their full expression. In contrast, at high Mg²⁺ concentrations, lpxO2 transcription levels were independent of the integrity of either phoP or ramA.

FIG 3.

Transcriptional expression of lpxO1 and lpxO2. Bacteria were grown for 16 h in N medium with 10 μM MgCl₂ for low-Mg²⁺ conditions or 10 mM MgCl₂ for high-Mg²⁺ conditions in 96-well microplates, at 37°C with agitation. Transcriptional activity was calculated as the ratio of GFP fluorescence values and OD₆₀₀ (FU/OD₆₀₀) measured from the S. marcescens WT, phoP, ramA, and ramA phoP strains carrying the PlpxO1-gfp and PlpxO2-gfp reporter plasmids. Means ± SDs from four independent experiments performed in duplicate in each case are shown. Statistical analysis was performed using two-way analysis of variance with Bonferroni’s posttest. ****, P < 0.0001 (statistically significantly different from WT S. marcescens).

In order to examine if these genes are under direct control of PhoP and RamA, we performed EMSA using purified recombinant PhoP and RamA proteins. When the promoter regions of lpxO1 and lpxO2 were incubated in the presence of PhoP, retarded bands were detected and their intensities were enhanced when the amount of PhoP was increased, with the concomitant intensity reduction in the band that corresponds to the unbound probe (Fig. 4A, left, and B). These results demonstrate that, as predicted, PhoP is able to directly interact with the promoter sequences that contain PhoP-binding motifs of lpxO1 and lpxO2. A 25- to 100-fold excess of unlabeled nonspecific nucA DNA fragment did not affect the interaction (Fig. 4A, middle), while the shifted band was progressively lost when increasing amounts of the unlabeled lpxO1 promoter fragment were included in the mixture (Fig. 4A, right), showing the specificity of the interaction. The presence of unlabeled nonspecific nucA DNA fragment did not affect the interactions (Fig. 4B), indicating that the interaction of PhoP with the lpxO2 promoter region is specific.

FIG 4.

PhoP and RamA interactions with lpxO1 and lpxO2 promoter regions in S. marcescens. (A and C) EMSAs were performed using different amounts of purified PhoP-6×His (A) or RamA-6×His (C). Target DNA was a ³²P-labeled PCR fragment that included the lpxO1 promoter region (PlpxO1). Binding specificity was assessed by competition reactions using 20 pmol of purified PhoP-6×His or RamA-6×His in which increasing amounts (50, 100, and 200 ng) of nonspecific (nucA; middle) or specific (PlpxO1; right) unlabeled DNA template competed with labeled DNA for binding to PhoP-6×His or RamA-6×His. (B and D) EMSAs were performed using nonlabeled PCR fragments carrying the complete lpxO2 promoter and purified PhoP-6×His (B) or RamA-6×His (D) (10 or 20 pmol, as indicated). When indicated, a 441-bp DNA fragment was used as a nonspecific competitor.

On the other hand, when the promoter region of lpxO1 was incubated in the presence of RamA, retarded bands were detected and their intensities were enhanced when the amount of RamA was increased, with concomitant intensity reduction in the band that corresponds to the unbound probe (Fig. 4C, left). The results show that RamA directly binds to the promoter region of the lpxO1 gene. A 25- to 100-fold excess of competing nonspecific nucA DNA fragment did not affect the interaction (Fig. 4C, middle), while the shifted band was progressively lost when increasing amounts of unlabeled lpxO1 promoter fragment were included in the mixture (Fig. 4C, right), indicating that the interaction of RamA with the lpxO1 promoter region is specific. In contrast, when the promoter region of lpxO2 was incubated in the presence of RamA, no shift in the mobility of the DNA probe was detected (Fig. 4D). This result reinforces the notion that that lpxO2 regulation would be under indirect control of RamA. Together, these results demonstrate that in S. marcescens, RamA functions together with PhoP as a regulator of genes that are predicted to be involved in lipid A modification.

RamA overexpression alters S. marcescens susceptibility to antibiotic compounds.

In Salmonella enterica serovar Typhimurium (35, 36) as well as in other Enterobacteriaceae, including Klebsiella (37) and Enterobacter (38) spp., RamA was described to regulate the expression of the genes encoding the AcrAB-TolC resistance-nodulation-division multidrug efflux system. AcrAB-TolC multidrug efflux pumps restrict the intracellular concentrations of various antibiotics, including β-lactams, tetracyclines, chloramphenicol, and quinolones (39). Therefore, we sought to analyze whether RamA is involved in conferring tetracycline resistance to S. marcescens. To this end, we compared the growth capacities of WT, phoP, ramA, and ramA phoP strains in Luria-Bertani (LB) medium using concentrations of tetracycline between 0 and 8 μg/ml. No significant differences were obtained in susceptibility to tetracycline in these strains (Fig. 5, strains carrying an empty plasmid pSU36). Because for K. pneumoniae, Enterobacter aerogenes, and S. enterica serovar Typhimurium it has been previously reported that overexpression of RamA increases resistance to antibiotics (38, 40, 41) and we do not know the conditions that induce RamA expression or activation in Serratia, we conjectured that overexpression of the regulator could bypass or mimic inducing conditions. To assess whether RamA overexpression was able to increase the resistance of S. marcescens to tetracycline, nalidixic acid (a quinolone), or chloramphenicol, we compared the growth capacities of strains when we expressed RamA from pSU36::ramA, in the presence of increasing concentrations of these drugs. The half-inhibitory concentrations (IC₅₀) were estimated at 1.0 μg/ml for tetracycline, 1.7 μg/ml for nalidixic acid, and 0.7 μg/ml for chloramphenicol in the strains harboring the empty pSU36 vector. The IC₅₀ values increased to 2.0 μg/ml for tetracycline, 3.3 μg/ml for nalidixic acid, and 4.4 μg/ml for chloramphenicol in the strains in which RamA was overexpressed (Fig. 5). The results showed that RamA overexpression reduced the susceptibility to tetracycline, nalidixic acid, and chloramphenicol.

FIG 5.

Susceptibility of S. marcescens to antibiotics. The OD₆₀₀ was determined for overnight cultures of the WT, phoP, ramA, and ramA phoP Serratia strains carrying the pSU36 or pSU36::ramA plasmid grown in LB medium in the presence of different concentrations of tetracycline (A), nalidixic acid (B), or chloramphenicol (C). Results are averages from three independent assays performed in duplicate.

To analyze whether the reduction in tetracycline, nalidixic acid, and chloramphenicol susceptibility observed by overexpressing RamA could be associated with an induced expression of the AcrAB-TolC efflux pump, we determined transcriptional levels of acrA and tolC by reverse transcription-quantitative PCR (qRT-PCR). As shown in Fig. 6, either acrA or tolC transcript levels were significantly lower in the ramA mutant than in the wild-type strain. Moreover, in the otherwise isogenic strains that overproduce RamA, we observed that the transcript levels of acrA or tolC increased more than 2-fold compared to those in strains carrying the empty plasmid. These results indicate that RamA is able to activate AcrAB-TolC efflux pump expression and suggest that in S. marcescens, the resistance to diverse antibiotics could be increased by RamA-mediated enhanced expression of AcrA/TolC. In order to confirm the role of the AcrAB-TolC efflux pump in resistance to these antibiotics, we constructed tolC and acrA mutant strains and compared the growth capacities of WT, tolC, and acrA strains in LB medium using concentrations of either tetracycline or nalidixic acid between 0 and 8 μg/ml. As expected, tolC and acrA mutants were less resistant to either tetracycline or nalidixic acid than the wild-type strain (Fig. 7A). The overexpression of RamA in tolC and acrA mutants was not able to restore the levels of wild-type resistance to either antibiotic (Fig. 7B), demonstrating that RamA-dependent upregulation of AcrAB-TolC expression is responsible for enhanced levels of antibiotic resistance in S. marcescens.

FIG 6.

qRT-PCR assays showing relative expression data for the tolC and acrA genes analyzed. The strains carrying the pSU36 or pSU36::ramA plasmid were grown in LB medium for 4 h at 37°C. mRNA levels were normalized to the 16S rRNA gene, and relative expression was calculated using the 2^−ΔΔCT method. Means and standard errors of four independent experiments are shown. Significant differences versus reference condition calculated by paired t test are indicated as follows: *, P < 0.05, and **, P < 0.01.

FIG 7.

Susceptibility of S. marcescens to antibiotics. (A) The OD₆₀₀ was determined for overnight-grown cultures in LB medium in the presence of different concentrations of tetracycline or nalidixic acid. (A) WT, tolC, and acrA Serratia strains. (B) WT, tolC, and acrA Serratia strains carrying the pSU36 or pSU36::ramA plasmid. Results are averages from four (A) or three (B) independent assays performed in duplicate.

Overall, our results show that in S. marcescens, RamA expression has been recruited under PhoP-dependent regulation. In addition, although the identity of the inducing signal is unknown, increased RamA expression is able to induce the expression of the AcrA/TolC efflux pump, which, in turn, enhances antibiotic resistance levels by restricting the intracellular concentration of the assayed antibiotics.

Concluding remarks.

S. marcescens can be isolated in host and nonhost environments. We have previously demonstrated that S. marcescens is able to invade, survive inside, and proliferate inside nonphagocytic cells (42). We have also shown that the PhoP/PhoQ system is implicated in the avoidance strategy that allows Serratia to survive and proliferate inside host cells. Furthermore, we have shown that the PhoP/PhoQ system is involved in the adaptation of this bacterium to growth in the context of scarce environmental Mg²⁺, at acidic pH, and in the presence of polymyxin B (23). Therefore, the PhoP/PhoQ system allows S. marcescens to detect and respond to both ambient and host-associated signals.

In this report, we show that the S. marcescens ramA gene, which codes for the AraC-type transcriptional regulator RamA, is autoregulated and that it is also part of the S. marcescens PhoP/PhoQ regulon. However, the ramA gene is atypically regulated by PhoP, because our results indicate that even high-Mg²⁺-concentration conditions, which would imply low concentrations or inactive PhoP, would be sufficient to promote ramA transcription. Curiously, this is also the case for lpxO1. The facts that in the wild-type strain, ramA and lpxO1 expression levels are not affected by the Mg²⁺ concentration of the growth medium and that PhoP integrity alters ramA and lpxO1 transcriptional levels only under low-Mg²⁺ conditions indicate that the phosphorylation status of PhoP does not influence this regulation and suggest the involvement of an additional unknown Mg²⁺-modulated factor. This unknown factor might contribute to activate ramA and lpxO1 transcription under high-Mg²⁺ conditions and/or repress their expression under low-Mg²⁺ conditions (Fig. 8).

FIG 8.

Proposed model for the S. marcescens PhoP/PhoQ-RamA regulatory cascade. Under high-Mg²⁺ conditions (left), ramA expression is PhoP independent but is dependent on ramA integrity. Under low-Mg²⁺ conditions (right), full ramA transcriptional expression is both RamA and PhoP dependent. RamA positively regulates the expression of the AcrAB-TolC efflux pump, responsible for detoxification of a wide range of substrates, including antimicrobials. ramA, mgtE1, mgtE2, lpxO1, and lpxO2 are members of the S. marcescens PhoP regulon. mgtE2 is only PhoP dependent, while ramA, mgtE1, lpxO1, and lpxO2 are PhoP and RamA dependent. PhoP is able to directly and specifically bind to ramA, mgtE1, mgtE2, lpxO1, and lpxO2 promoter regions, RamA is able to recognize ramA and lpxO1 but not mgtE1 and lpxO2 promoter regions, suggesting an indirect regulatory mechanism for the expression of the latter genes. Still-uncovered regulatory mechanisms (represented by “X”) might contribute to activate ramA and lpxO1 genes transcription under high-Mg²⁺ conditions and/or repress its expression under low-Mg²⁺ conditions.

In addition, our results demonstrate that ramA, mgtE1, mgtE2, lpxO1, and lpxO2 are members of the S. marcescens PhoP regulon. While mgtE2 is only PhoP dependent, ramA, mgtE1, lpxO1, and lpxO2 are PhoP and RamA dependent. We also show that while PhoP is able to directly and specifically bind to ramA, mgtE1, mgtE2, lpxO1, and lpxO2 promoter regions, RamA is able to recognize ramA and lpxO1 but not mgtE1 and lpxO2, suggesting an indirect regulatory mechanism for the expression of the latter genes. LpxO has been identified as the oxygenase that 2-hydroxylates the acyl chains of lipid A (25). The addition of a 2-hydroxyl group in acyl chain would stabilize the outer membrane, contributing to resistance to antimicrobial peptides (30). According to the sequence homology of the S. marcescens lpxO genes with genes involved in lipid A remodeling in other enterobacteria, we can conjecture that PhoP and RamA would be relevant for the regulation of S. marcescens envelope properties.

It has been previously shown that in pathogenic bacteria, RamA regulates the expression of efflux pumps like AcrAB-TolC, responsible for the detoxification of a wide range of substrates, including antimicrobials, heavy metals, and detergents outside the cell (43). Overexpression of RamA from a plasmid was found to reduce the susceptibility to antibiotics by increasing the levels of AcrAB-TolC efflux pump in K. pneumoniae, E. aerogenes, and S. enterica serovar Typhimurium (38, 40, 41). Furthermore, it was found that clinical isolates of multidrug-resistant (MDR) bacterial pathogens overexpress regulators such as RamA with subsequent overproduction of AcrAB-TolC, which confers higher resistance to antibiotics (44–48). It was recently shown that in S. marcescens, the MacAB efflux pump is essential for survival during oxidative stress and confers resistance to polymyxins and aminoglycoside antibiotics (49). These results show the crucial role played by efflux pumps in the intrinsic resistance of bacteria to antimicrobials.

Our results describe for the first time a regulatory cascade between the PhoP/PhoQ system and the RamA regulator in S. marcescens. We here show that RamA is PhoP regulated and that RamA overexpression induces the AcrAB-TolC efflux pump, required to reduce the susceptibility of the bacteria to tetracycline, nalidixic acid, and chloramphenicol. The PhoP/PhoQ system in many bacteria regulates virulence genes and LPS modifications that give them resistance to antimicrobial cationic peptides. The fact that PhoP regulates RamA expression in Serratia suggests that the PhoP regulon could have adopted RamA regulation to simultaneously acquire the ability to resist the action of antibiotics used in the clinic for the treatment of bacterial infections and fight the effect of antimicrobial cationic peptides produced as defense by the host.

Overall, our work provides new insights into the PhoP/PhoQ signal transduction system regulon in S. marcescens and demonstrates the involvement of RamA in the protection against antibiotic challenges.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Serratia marcescens RM66262 is a nonpigmented clinical isolate from a patient with a urinary tract infection (GenBank accession no. NZ_JWLO00000000.1) (22). The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were routinely grown in Miller’s Luria-Bertani (LB) medium or on LB agar plates overnight at 30°C or 37°C. To assay Mg²⁺ regulation, strains were grown overnight under low-Mg²⁺ conditions {N medium [0.1 M Tris-HCl (pH 7.4), 0.1% (wt/vol) Casamino Acids, 38 mM glycerol, 0.37 g/liter of KCl, 0.99 g/liter of (NH₄)₂SO₄, 0.087 g/liter of K₂SO₄, 0.14 g/liter of KH₂PO₄] [50] plus 10 μM MgCl₂} or high-Mg²⁺ conditions (N medium plus 10 mM MgCl₂). The antibiotics used were ampicillin (100 μg/ml), kanamycin (50 μg/ml), chloramphenicol (20 μg/ml), streptomycin (100 μg/ml), and gentamicin (15 μg/ml).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype and/or comments	Reference or source
S. marcescens strains
RM66262	WT; clinical isolate	22
phoP strain	phoP::pKNOCK-Gm	23
lpxO1 strain	lpxO1::pKNOCK-Gm	This work
lpxO2 strain	lpxO2::pKNOCK-Gm	This work
ΔramA strain	ΔramA	This work
ΔramA phoP strain	ΔramA phoP::pKNOCK-Gm	This work
WT/pSU36	WT/pSU36	This work
WT/pSU36::ramA	WT/pSU36::ramA	This work
phoP/pSU36 strain	phoP::pKNOCK-Gm/pSU36	This work
phoP/pSU36::ramA strain	phoP::pKNOCK-Gm/pSU36::ramA	This work
ΔramA/pSU36 strain	ΔramA/pSU36	This work
ΔramA/pSU36::ramA strain	ΔramA/pSU36::ramA	This work
ΔramA phoP/pSU36 strain	ΔramA phoP::pKNOCK-Gm/pSU36	This work
ΔramA phoP/pSU36::ramA strain	ΔramA phoP::pKNOCK-Gm/pSU36::ramA	This work
tolC strain	tolC::pKNOCK-Gm	This work
acrA strain	acrA::pKNOCK-Gm	This work
tolC/pSU36::ramA strain	tolC::pKNOCK-Gm/pSU36::ramA	This work
acrA/pSU36::ramA strain	acrA::pKNOCK-Gm/pSU36::ramA	This work
E. coli strains
One Shot Top10	F− mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 nupG recA1 araD139 Δ(ara-leu)7697 galE15 galK16 rpsL endA1 Smr	Invitrogen
SM10 λpir	thi J thr leu tonA lacY61 lic recA::RP4‐2‐Tc::Mu λpir Kmr	52
Plasmids
pSU36	Kmr	55
pSU36::ramA	pSU36::ramA; Kmr	This work
pPphoP-gfp	pPROBE(NT)::promphoP Kmr	Laboratory stock
pPramA-gfp	pPROBE(NT)::promramA Kmr	This work
pPlpxO1-gfp	pPROBE(NT)::promlpxO1 Kmr	This work
pPlpxO2-gfp	pPROBE(NT)::promlpxO2 Kmr	This work
pPmgtE1-gfp	pPROBE(NT)::prommgtE1 Kmr	Laboratory stock
pPmgtE2-gfp	pPROBE(NT)::prommgtE2 Kmr	Laboratory stock
pET22b::phoP	Expression vector for PhoP-6×His	This work
pET22b::ramA	Expression vector for RamA-6×His	This work

Insertion mutations in lpxO1 (RT90_RS21050), lpxO2 (RT90_RS03150), tolC (RT90_RS20310), and acrA (RT90_RS11255) were constructed with the pKNOCK-Gm suicide plasmid (51). For each gene, an internal 500-bp region was amplified using primers lpxO1-fw.BamHI and lpxO1-rv.XhoI, lpxO2-fw.BamHI and lpxO2-rv.XhoI, tolC-fw.XbaI and tolC-rv.XhoI, and acrA-fw.XbaI and acrA-rv.XhoI (Table 2). The purified PCR products were digested with the restriction enzymes indicated in the primer names and cloned into the pKNOCK-Gm plasmid. The resulting plasmids were introduced into competent E. coli SM10 λ_pir (52) cells by electroporation and then mobilized into S. marcescens RM66262 by conjugation. Insertional mutants were confirmed by PCR analysis using primers.

TABLE 2.

Primers used in this study

Primer	Sequence (5′–3′)a
lpxO1-fw.BamHI	CGGGATCCCCTGTTCGATCACTCCACC
lpxO1-rv.XhoI	CCGCTCGAGCGTTCTGCGCCCAGTG
lpxO1-fwb	CTATGGTTGCCGCGATCATT
lpxO1-rvb	AGCTGGCTGCCGACATCA
pKNOCK (ori) fwdb	TAAGGTTTAACGGTTGTGG
lpxO2-fw.BamHI	CGGGATCCCCCGTACCTGAAGCCGGAG
lpxO2-rv.XhoI	CCGCTCGAGCCAGTGGTTCACCGCCTG
lpxO2-fwb	CGCTACAACGTGTGGCGAC
lpxO2-rvb	TAACCGGTGCGGTTCCAG
tolC-fw.XbaI	TGCTCTAGAGATCACCGACGTGCAGAACG
tolC-rv.XhoI	CCGCTCGAGGGTGTGAGCTTTCCAACTGC
tolC-fwb	ATCAGCTGCAGTCGACCCG
tolC-rvb	GCCGTCGTTAATCAAGCGC
acrA-fw.XbaI	TGCTCTAGAAAACCGCTACAAGCCATTGC
acrA-rv.XhoI	CCGCTCGAGAGCGTATCGTTCGGGTTAGG
acrA-fwb	GCATCAAGGATTTGCTGTCG
acrA-rvb	TTGGCGGAAATACTCACCGC
ramA-A	CGCGGATCCGCGGCGTGCTGCAGTTCG
ramA-B	AAAAGGCCTCATGGTTTTAGTTCTCCC
ramA-C	AAAAGGCCTGCGCTGATAAACGCGG
ramA-D	CCGGAATTCGTGATCCAGTGCAATCCG
ramA-fwb	TGAACGACGACGATCGCTTC
ramA-rvb	GCTGTTCCAGATCGATGGTG
Prom 4911 Fw EcoRI	CTGAATTCGCGGCGTGCTGCAGTTCG
Prom 4911 Rv BamHI	CGGGATCCCATGGTTTTAGTTCTCCC
4911 Fw BamHI	CGGGATCCATGGATCGCGTCAATATC
4911 Rv HindIII	TCCAAGCTTTTAACGCAGCCCCATCCAG
Prom phoP serr fwd EcoRI	ACCGAATTCGCGCTTAACCCGCTCG
Prom phoP serr rev BamHI	AGGGATCCCATGGCGAACTCCTGTG
prommgtE EcoRI fwd	GCAGGAATTCCAGCAGGC
prommgtE rev BamHI	AGAGGATCCGCAGACTGCCTA
prom mgtE2 Fw EcoRI	CGGAATTCCAAAACGAACAGCGC
prom mgtE2 Rv BamHI	CGGGATCCGTTATGGGTTGCGGC
promlpxO1-Fw	CGGAATTCGTCCTGAGACTGGTGCAGCT
promlpxO1-Rv	CGGGATCCTGCGTAAAGGAAGCTGACGAT
promlpxO2-Fw	CGGAATTCCGGTCGTTATCCTGGATGAAG
promlpxO2-Rv	CGGGATCCCGTCCGCGGTAATGCACATAG
nuclease Fw (XbaI)	GCTCTAGAGGCAAGACGCGCAACTGG
nuclease Rv (XhoI)	CCGCTCGAGGAAATCGGCGCCCTTCGG
16S-F	AAACTGGAGGAAGGTGGGGATGAC
16S-R	ATGGTGTGACGGGCGGTGTG
RT-acrA-Fw	AAGCGCAACTTCGTTGAAGG
RT-acrA-Rv	ATGGCTTGTAGCGGTTTACC
RT-tolC-Fw	TGCTGCAGGTCTACAAACAG
RT-tolC-Rv	TGCTGTTAACGCCATTGCTG

^aRestriction sites are underlined.

^bPrimer used to confirm the mutant by PCR.

S. marcescens ΔramA was constructed as follows. PCR was used to generate 501 bp of DNA upstream of ramA (RT90_RS09790) using primers ramA-A and ramA-B (Table 2), and 520 bp of DNA downstream of ramA using primers ramA-C and ramA-D (Table 2). The resulting DNA fragments were digested with the BamHI-StuI and StuI-SpeI restriction enzymes, respectively, and the fragments were ligated into the BamHI and SpeI sites of pKNG101 (53). pKNG101::ΔramA recombinant plasmid was then mobilized into S. marcescens RM66262 by conjugation. Mutant strains were selected on streptomycin, and then high sucrose allowed isolation of mutants in which the deletion allele had replaced the wild-type copy. The ramA deletion was confirmed by PCR.

To analyze the transcriptional levels of phoP (RT90_RS07370), ramA, lpxO1, lpxO2, mgtE1 (RT90_RS05070), and mgtE2 (RT90_RS17415), the promoter regions of the genes were amplified by PCR using the primers Prom phoP serr fwd EcoRI and Prom phoP serr rev BamHI, Prom 4911 Fw EcoRI and Prom 4911 Rv BamHI, promlpxO1-Fw and promlpxO1-Rv, promlpxO2-Fw and promlpxO2-Rv, prommgtE EcoRI fwd and prommgtE rev BamHI, and prom mgtE2 Fw EcoRI and prom mgtE2 Rv BamHI (Table 2). The purified PCR products were digested with the EcoRI and BamHI restriction enzymes and were ligated into the same sites of pPROBE(NT′) (54). The resulting plasmids were introduced into competent E. coli Top10 cells by transformation. The plasmids PphoP-gfp, PramA-gfp, PlpxO1-gfp, PlpxO2-gfp, PmgtE1-gfp, and PmgtE2-gfp were mobilized by conjugation into the S. marcescens wild-type strain and the phoP, ramA, and ramA phoP mutant strains.

For complementation of the S. marcescens ramA mutant strain, the ramA gene was amplified from the S. marcescens wild-type strain chromosome by PCR using primers 4911 Fw BamHI and 4911 Rv HindIII (Table 2). The PCR product was cloned into the pSU36 plasmid (55). The construction was then introduced into the S. marcescens strains by electroporation.

phoP, ramA, mgtE1, mgtE2, lpxO1, and lpxO2 gene expression assays.

Cultures of S. marcescens wild-type (WT), phoP, ramA, and ramA phoP strains carrying the pPphoP-gfp, pPramA-gfp, pPmgtE1-gfp, pPmgtE2-gfp, pPlpxO1-gfp or pPlpxO2-gfp reporter plasmids were grown with shaking overnight at 37°C. The bacterial cultures were washed two times with N medium, 10 μl was mixed with 1 ml of N medium supplemented with 10 μM or 10 mM MgCl₂ with kanamycin, and 100-μl volumes of the mixtures were incubated in a 96-well microtiter plate at 37°C with agitation for 16 h. Optical density at 600 nm (OD₆₀₀) and GFP fluorescence (excitation wavelength [λ_exc] of 485 nm and emission wavelength [λ_em] at 528 nm) were determined using a 96-microwell plate reader (Synergy 2). Transcriptional activity was calculated as the ratio of GFP fluorescence and OD₆₀₀ (FU/OD₆₀₀) measured from strains carrying the pPphoP-gfp, pPramA-gfp, pPmgtE1-gfp, pPmgtE2-gfp, pPlpxO1-gfp, and pPlpxO2-gfp reporter plasmids. The means and standard deviations for three independent assays performed in duplicate in each case were calculated.

Protein-DNA interaction analysis.

Electrophoretic gel mobility shift assays (EMSAs) were performed using 6 fmol of ³²P-labeled DNA fragments containing the ramA promoter (PramA) and lpxO1 promoter (PlpxO1) with different amounts of purified PhoP-6×His and RamA-6×His proteins following basically previously described protocols (56). Prior to addition of the DNA probe, PhoP-6×His protein was phosphorylated by incubation with 25 mM acetyl phosphate at 30°C for 30 min. The specificity of binding was assayed using the unlabeled PramA and PlpxO1 probes or a 441-bp PCR fragment corresponding to the nucA gene from S. marcescens as a nonspecific competitor. To evaluated the interaction of PhoP or RamA with the promoter regions of the mgtE1, mgtE2, and lpxO2 genes, EMSAs were performed using 30 ng of nonlabeled DNA fragments containing the complete promoter regions of the genes and 10 or 20 pmol of purified PhoP-6×His or RamA-6×His. The nonspecific competitor DNA was assayed using a 441-bp PCR fragment corresponding to the nucA gene from S. marcescens. The primers used to amplify the PramA, PlpxO1, PmgtE1, PmgtE2, PlpxO2, and PphoP regions and nucA (RT90_RS08445) are listed in Table 2. After electrophoresis, the gels were either dried and exposed to autoradiography or stained with SYBR green (Invitrogen). DNA and protein-DNA complexes were detected and captured using a Typhoon FLA7000 laser scanner (GE Healthcare).

RT-PCR and qRT-PCR.

cDNA synthesis was performed using random hexamers, 2 μg of total RNA, and 1 U of SuperScript II reverse transcriptase (Invitrogen). Five microliters of a 1/10 dilution of each cDNA was used as the template for DNA amplification in RT-PCR or quantitative RT-PCR (qRT-PCR) (20 μl), using primers tolC and acrA (Table 2). A primer set for the 16S rRNA was used as a control to confirm that equal amounts of total RNA were used in each reaction mixture. In every case, the amplified fragment was 250 bp. For RT-PCR, the number of cycles varied according to the level of expression of each mRNA to ensure that the comparison was performed in the linear range of the amplification. For the quantitative real-time PCR, the reactions were carried out in the presence of the double-stranded DNA-specific dye SYBR green (Molecular Probes) and monitored in real time with the Mastercycler ep realplex real-time PCR system (Eppendorf). The relative expression was calculated using the threshold cycle values (C_T) obtained from each sample as follows: relative expression = 2^−ΔΔ, with Δ^CT = C_{T sample} − C_{T 16S} and ΔΔC_T = ΔC_{T sample} − ΔC_{T ref sample} (where sample is the mutant strain transcript, 16S is the 16S rRNA transcript, and ref sample is the S. marcescens RM66262 transcript). The reference sample was S. marcescens RM66262. The average values were calculated from triplicate samples.

Tetracycline, nalidixic acid, and chloramphenicol susceptibility assays.

Fifty microliters of a 1:50 dilution of overnight LB cultures was mixed with 50 μl of antibiotic solution dissolved in the same medium at final tetracycline, nalidixic acid, or chloramphenicol concentrations that ranged between 0 and 8 μg/ml. The mixtures were incubated in 96-well microtiter plates at 37°C without agitation for 16 h. OD₆₀₀ readings were determined with a BioTek ELx808 microplate reader. The means and standard deviations for three independent assays performed in triplicate in each case were calculated.

Statistical analysis.

Statistical analysis was performed using one-way or two-way analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons with significance set at a P value of <0.05. Asterisks in figures indicate the values among the treatment groups in which a statistically significant difference was determined.