A phosphotransferase system permease is a novel component of CadC signaling in Salmonella enterica

Authors


Correspondence: Yong Keun Park, School of Life Sciences and Biotechnology, Korea University, Seoul 136-701, South Korea. Tel.: +82 2 3290 3422; fax: +82 2 927 9028; e-mail: ykpark@korea.ac.kr

Abstract

In Salmonella enterica serovar Typhimurium, proteolytic cleavage of the membrane-bound transcriptional regulator CadC acts as a switch to activate genes of the lysine decarboxylase system in response to low pH and lysine signals. To identify the genetic factors required for the proteolytic activation of CadC, we performed genome-wide random mutagenesis. We show that a phosphotransferase system (PTS) permease STM4538 acts as a positive modulator of CadC function. The transposon insertion in STM4538 reduces the expression of the CadC target operon cadBA under permissive conditions. In addition, deletional inactivation of STM4538 in the wild-type background leads to the impaired proteolytic cleavage of CadC. We also show that only the low pH signal is involved in the proteolytic processing of CadC, but the lysine signal plays a role in the repression of the lysP gene encoding a lysine-specific permease, which negatively controls expression of the cadBA operon. Our data suggest that the PTS permease STM4538 affects proteolytic processing, which is a necessary but not sufficient step for CadC activation, rendering CadC able to activate target genes.

Introduction

Transmembrane signaling is an essential feature that is common to all living cells and has become an increasingly attractive target for the development of new antimicrobial drugs (Rasko et al., 2008; Dougan, 2009). During the last decade, the regulated proteolysis of membrane-associated transcription factors has emerged as an important signaling mechanism conserved from bacteria to humans (Brown et al., 2000). This proteolytic switch produces a rapid cellular response by activating pre-existing pools of dormant transcription factors. In bacteria, one of the best studied examples is the activation of the alternative sigma factor σE, which is involved in the envelope stress response in Escherichia coli. The membrane-spanning anti-σ factor RseA is first cleaved by DegS and then by YaeL, thereby releasing σE from anti-σ factor sequestration (Alba et al., 2002; Chaba et al., 2007). Another example is the activation of the Bacillus subtilis σW, in which case the transmembrane anti-σ RsiW is sequentially cleaved by PrsW and RasP in the same manner as the E. coli RseA (Schobel et al., 2004; Heinrich & Wiegert, 2006).

The bacterial phosphotransferase system (PTS) catalyses the transport and phosphorylation of a number of sugar substrates. It consists of two general cytoplasmic proteins (Enzyme I and phosphocarrier protein HPr) and membrane-bound sugar-specific multiprotein permeases (Enzymes II), which are composed of three or four domains (EIIA, EIIB, EIIC and sometimes EIID). EIIA and EIIB are part of a phosphotransfer cascade, whereas EIIC (and sometimes EIID) is involved in sugar transport. The PTS uses phosphoenolpyruvate as an energy source and phosphoryl donor and transfers the phosphoryl group sequentially via Enzyme I, HPr, EIIA and EIIB to the transported sugar (Barabote & Saier, 2005; Deutscher et al., 2006). The PTS is also known to play a direct role in transcriptional control through modulation of the activities of specific multidomain transcriptional activators and antiterminators, DNA- and RNA-binding proteins, that contain homologous phosphorylation domains (Tortosa et al., 1997; Martin-Verstraete et al., 1998; Stulke et al., 1998).

Salmonella enterica serovar Typhimurium (S. Typhimuium) CadC is a membrane-spanning transcriptional activator with a cytoplasmic DNA-binding domain and a periplasmic signal-sensing domain. CadC activates the cadBA operon in response to external acidification and exogenous lysine (Watson et al., 1992; Lee et al., 2007). Following induction, CadA-mediated lysine decarboxylation produces cadaverine, which is excreted through the lysine-cadaverine antiporter CadB, contributing to the acid tolerance response (Park et al., 1996; Foster, 1999). In E. coli, the nucleoid-associated DNA-binding protein H-NS negatively regulates expression of the cadBA operon through the formation of a repression complex at the cadBA promoter region under noninducing conditions (Shi et al., 1993; Kuper & Jung, 2005). Our previous study clearly demonstrated that in S. Typhimurium CadC is produced as a dormant membrane-localized precursor that is rapidly cleaved in response to low pH and lysine signals. Site-specific proteolysis at the periplasmic domain of CadC generates a biologically active form of the N-terminal DNA-binding domain, which binds to the target gene promoter (Lee et al., 2008). However, the identity of the proteases involved and the precise role of each individual signal remain unknown.

The aim of the current study was to identify candidate genes associated with the proteolytic activation of CadC. We employed a genetic screen and identified the PTS permease STM4538 as a novel modulator of CadC function. We further addressed the individual roles of low pH and lysine signals in the proteolytic activation of CadC. These findings reveal previously unrecognized regulatory aspects of CadC signaling in S. Typhimurium.

Materials and methods

Bacterial strains and growth conditions

The S. Typhimurium strains used in this study are listed in Table 1. The cells were routinely cultured at 37 °C in Luria–Bertani (LB) complex medium or Vogel and Bonner E minimal medium supplemented with 0.4% glucose (Vogel & Bonner, 1956; Maloy & Roth, 1983). Lysine decarboxylase (LDC) broth (0.5% peptone, 0.3% yeast extract, 0.1% dextrose, 0.5% l-lysine and 0.002% bromcresol purple) was used for the LDC assay. The following antibiotics were used when appropriate: ampicillin (Ap; 60 μg mL−1), kanamycin (Km; 50 μg mL−1) and chloramphenicol (Cm; 30 μg mL−1). Acid stress (pH 5.8, 10 mM lysine) was applied to cells grown in E glucose medium to an OD600 nm of 0.6.

Table 1. Bacterial strains and plasmids used in this study
Strain or plasmidRelevant characteristicsSource or reference
S. enterica serovar Typhimurium
SF530 (χ3761)Wild-type UK1Curtiss & Hassan (1996)
JF3068UK1 cadA::lacZPark et al. (1996)
YK5002UK1 ΔcadCLee et al. (2007)
YK5004UK1 ΔlysPThis study
YK5005UK1 cadA::lacZ ΔcadCThis study
YK5006UK1 cadA::lacZ ΔlysPThis study
YK5007UK1 cadA::lacZ STM4538::Tn10dCmThis study
YK5009UK1 STM4538::Tn10dCmThis study
YK5010UK1 ΔSTM4538This study
YK5011UK1 cadA::lacZ ΔSTM4538This study
Plasmid
pKD4Source of resistance cassette, Ampr KmrDatsenko & Wanner (2000)
pKD46oriR101 repA101(ts) araBp-gam-bet-exo ap, AmprDatsenko & Wanner (2000)
pCP20λcI857(ts) ts-rep, AmprDatsenko & Wanner (2000)
pACYC184Low-copy-number cloning vector, Tetr CmrNew England Biolabs
pACYC184-HA-CadCpACYC184 with native promoter and HA-cadC gene, CmrLee et al. (2008)
pACYC184-LysP-HApACYC184 with native promoter and lysP-HA gene, CmrThis study
pMW118Very-low-copy-number cloning vector, AmprNippon Gene
pMW118-STM4538pMW118 with native promoter and STM4538 gene, AmprThis study

Construction of S. Typhimurium strains

Knockout mutants were constructed using the lambda red recombinase system (Datsenko & Wanner, 2000). For construction of the STM4538 mutant, the KmR cassette was amplified from pKD4 using primers STM4538-Mu-F (5′-GATTTACGCCGCGTCTTCTGGCGGTCATTCCAGATGGAGTGTGTAGGCTGGAGCTGCTTC-3′) and STM4538-Mu-R (5′-CAGACAAGGCATGATGTCGTTAATAATGTCCTGAACATGGCATATGAATATCCTCCTTAG-3′), and the resulting PCR product was electroporated into the UK1 wild-type strain carrying plasmid pKD46. The genotype of the generated mutant was verified using PCR and DNA sequencing, and then the KmR cassette was removed using plasmid pCP20. The lysP gene was disrupted in the same way using primers lysP-Mu-F (5′-TTATAACCGCGCATTTGTGTCGGAAGGATAGTATTTCGTCGTGTAGGCTGGAGCTGCTTC-3′) and lysP-Mu-R (5′-ACCGGAGGTGTTTAACAGCCACAGATAGACCGTCTGGTTGCATATGAATATCCTCCTTAG-3′). Salmonella strains YK5005 (cadA::lacZ ΔcadC), YK5006 (cadA::lacZ ΔlysP) and YK5009 (STM4538::Tn10dCm) were constructed using phage P22-mediated generalized transduction as previously described (Davis et al., 1980). Bacteriophage P22HT int 105 was propagated in a donor strain (JF3068 or YK5007) and used to infect the recipient strain (YK5002, YK5004 or UK1 wild-type). The transductants were selected on LB agar containing Km (50 μg mL−1) or Cm (30 μg mL−1). P22 H5 was used to confirm that transductants were phage-free and not P22 lysogens (Maloy et al., 1996).

Plasmid construction

PCR and cloning for plasmid construction were performed by using standard techniques (Sambrook & Russell, 2001). The recombinant plasmid pMW118-STM4538 was constructed using PCR amplification of the STM4538 gene and its promoter from S. Typhimurium chromosomal DNA with primers STM4538-F(5′-CCAAGCTTTTTAATCTCCGGCATTGGG-3′) and STM4538-R (5′-CGGGATCCTTAAAATAACCCTATCCAGGAACC-3′). The plasmid pACYC184-LysP-HA was constructed in a similar manner using primers LysP-HA-F (5′-CGGGATCCTGGAAGATGAGCTGGTGGTC-3′) and LysP-HA-R (5′-CCAAGCTTTTAAGCGTAGTCTGGGACGTCGTATGGGTACTTTTTAACGCGTTCCGGG-3′). The integrity of the constructs was verified through DNA sequencing.

Transposon mutagenesis and screening

The Tn10dCm transposon was mobilized into Salmonella strain JF3068 carrying a cadA::lacZ transcriptional fusion, and insertion mutants that inhibited the expression of cadA::lacZ under acid stress (pH 5.8, 10 mM lysine) were identified as white colonies on E glucose agar plates containing X-gal. The phenotype was confirmed by moving the mutations into the parent S. Typhimurium strain using P22-mediated transduction (Davis et al., 1980). The sites of Tn10dCm insertion in the chromosome were amplified using arbitrary primed PCR with primers Cat1/Arb1 and Cat2/Arb2 and sequenced using primer Cat2 (Welsh & McClelland, 1990).

β-Galactosidase assays

β-Galactosidase activity was determined using a modification of a previously described method (Miller, 1992). Briefly, cells (1 mL) were added to 1 mL Z buffer [60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, 2.7 μL mL−1 β-mercaptoethanol (pH 7.0)], disrupted with 0.1% (w/v) SDS and chloroform, and incubated with 0.4 mL of 4 mg mL−1 o-nitrophenyl-β-d-galactoside. The reaction mixture was incubated at room temperature until a yellow color developed, and subsequently the reaction was terminated with 1 mL of 1 M Na2CO3. β-Galactosidase activity was expressed in Miller units and calculated using the formula [1000 × (A420−1.75A550)]/[time (min) × culture volume (mL) × A600].

LDC assay

Bacterial colonies were inoculated into 3 mL of Moeller LDC broth (Difco) containing decarboxylase basal medium supplemented with 0.5% l-lysine and bromcresol purple indicator. Sterile mineral oil was layered over the medium to keep the pH above 7, and the culture was incubated for 36 h at 37 °C. If the dextrose is fermented, a yellow color initially develops, but the medium gradually turns purple as the decarboxylase reaction elevates pH.

Immunoblot analysis

Cells were grown in 20 mL of E glucose medium to an OD600 nm of 0.6 and subjected to stress. At different time points, samples (1.5 mL) were collected, and the cell pellets were resuspended in SDS sample buffer [60 mM Tris-HCl [pH 6.8], 30% glycerol, 2% SDS, 0.1% bromophenol blue, and 14.4 mM 2-mercaptoethanol) and boiled for 10 min to prepare the protein lysates. Proteins were separated by electrophoresis on a 12% SDS-polyacrylamide gel and transferred onto a nitrocellulose membrane. After blocking, the blots were probed with mouse monoclonal antibodies against the HA tag (Cell Signaling Technology, Danvers, MA) or DnaK (Stressgen, Victoria, Canada) as a loading control. Horseradish peroxidase-conjugated goat anti-mouse IgG was used as the secondary antibody. The proteins were visualized using a BM chemiluminescence blotting substrate (POD) (Roche, Mannheim, Germany).

RNA preparation and RT-PCR assay

Total RNA was isolated using the RNeasy Mini kit (Qiagen) and treated with DNase I. The quantity and purity of RNA was determined using a NanoDrop spectrophotometer (Nanodrop Tech. Inc., Wilmington, DE). cDNA was synthesized from total RNA (1 μg) using the First Strand cDNA Synthesis Kit (Roche) at the following conditions: 25 °C for 10 min, 42 °C for 60 min, 99 °C for 5 min and cooling to 4 °C. The resulting cDNA was then amplified using gene-specific primer sets. The reaction mixture was denatured (94 °C, 4 min), followed by 20 thermal cycles (94 °C for 30 s, 54 °C for 30 s, 72 °C for 50 s) and a final extension (72 °C for 10 min). The primer pair LysP-RT-F (5′-GGAAGAAGGCTTTGGTTTCG-3′) and LysP-RT-R (5′-GAGGCATACATCCCGGAGTT-3′) was used to detect the lysP transcript. The 16S rRNA gene was used as a normalization control. The amplified products were separated on a 1.5% agarose gel, stained with ethidium bromide and visualized.

Results

Genetic screening for candidate genes required for CadC activity

To identify genes involved in the proteolytic activation of CadC, we performed a genome-wide screen to isolate mutants that prevent cadBA expression, even in the presence of the cadC gene, under acid stress conditions (pH 5.8, 10 mM lysine). The Tn10dCm transposon was used to mutagenize Salmonella strain JF3068 carrying a cadA::lacZ transcriptional fusion. Of the approximately 30 000 random transposon insertions screened, 12 mutants were identified as white colonies on E glucose agar plates containing X-gal. The precise location of the transposon insertions was determined by sequencing of genomic DNA flanking the transposons (Welsh & McClelland, 1990). Ten insertions were mapped to the cad locus, and the remaining two insertions were located in STM4538 and yfhK, which encodes a PTS permease similar to the E. coli mannose-specific PTS enzyme IID and a putative sensor kinase, respectively.

To ensure linkage of the phenotype to the transposon insertion, STM4538::Tn10dCm and yfhK::Tn10dCm were moved into the parental wild-type strain using P22-mediated transduction, and LDC assays were performed. Usually, a positive test is indicated by a purple color and a negative test by a yellow color. Indeed, the resulting strain YK5009 (STM4538::Tn10dCm) showed a weak positive reaction (very faint yellow) in the LDC assay (Fig. 1a). Moreover, when STM4538 was expressed from its own promoter in the low-copy plasmid pMW118, the YK5009 strain showed an LDC-positive phenotype (Fig. 1a). However, the phenotype of the yfhK::Tn10dCm insertion was a false negative because this transposon insertion had no influence on LDC activity. We further compared the expression of a chromosomal cadA–lacZ fusion in strains JF3068 (wild-type), YK5007 (STM4538::Tn10dCm) and YK5011 (ΔSTM4538) using β-galactosidase assays. Following 30 min of acid stress, the level of cadA expression in the STM4538 mutants was approximately twofold lower than that in the wild-type (Fig. 1b). Together, these data suggest that the PTS permease STM4538 is positively involved in the control of cadBA expression.

Figure 1.

Identification of a novel gene involved in cadBA expression. (a) LDC assay showing that the YK5009 strain (STM4538::Tn10dCm) is defective in transcriptional activation of the cadBA operon. A pale yellow color indicates a weak positive result. Yellow and purple indicate the absence and presence of LDC, respectively. (b) Inactivation of STM4538 results in decreased cadA–lacZ expression. Cells were grown in E glucose medium to an OD600 of 0.6, and were then subjected to acid stress (pH 5.8, 10 mM lysine). β-Galactosidase assays were performed following 30 min of stress. Three independent experiments were performed for each strain, and the standard deviation is indicated by error bars.

Inactivation of STM4538 prevents proteolytic cleavage of CadC

To assess the potential role of STM4538 in the proteolytic activation of CadC, we performed an immunoblot analysis of total protein extracts from the S. Typhimurium wild-type and ΔSTM4538 strains harboring pACYC184-HA-CadC. N-terminally HA-tagged CadC (HA-CadC) was expressed under the control of its own promoter in the low-copy plasmid pACYC184. The cells were grown in E glucose medium to an OD600 nm of 0.6 and subjected to acid stress. As shown in Fig. 2, HA-CadC levels rapidly decreased in the wild-type background, as previously reported (Lee et al., 2008). However, despite wild-type levels of cadC transcription (data not shown), HA-CadC levels were slightly increased in the ΔSTM4538 null mutant after acid stress, indicating impaired proteolytic processing of CadC. These results suggest that the PTS permease STM4538 is required for the proteolytic activation of CadC signaling in S. Typhimurium.

Figure 2.

PTS permease STM4538 is involved in the proteolytic activation of CadC. The pACYC184-HA-CadC plasmid was expressed in the S. enterica serovar Typhimurium wild-type and ΔSTM4538 strains. Cells were grown in E glucose medium to an OD600 of 0.6, followed by acid stress. The samples were collected at the indicated times, and total proteins were immunoblotted with anti-HA and anti-DnaK antibodies. DnaK was used as a loading control.

Proteolytic cleavage of CadC occurs regardless of the lysine signal

To gain further insight into the signaling mechanism of CadC, which undergoes rapid proteolytic cleavage in response to low pH and lysine signals (Lee et al., 2008), we examined whether both signals are required for this proteolytic event. Immunoblot analysis was conducted on total protein prepared from the YK5005 (cadA::lacZ ΔcadC) strain harboring pACYC184-HA-CadC. Cells were grown in E glucose medium to an OD600 nm of 0.6 and exposed to three different types of signals. The samples were collected at the indicated times and immunoblotted with anti-HA antibodies. As shown in Fig. 3(a), proteolysis of CadC occurs strictly in response to a pH shift regardless of the lysine signal. On the other hand, the lysine signal is insufficient on its own to stimulate proteolysis. To further confirm the concomitant effects of CadC proteolysis on cadBA transcription, the β-galactosidase activity from a cadA-lacZ transcription fusion was measured 30 min after each treatment. As expected, cadA transcription was induced only when cells respond to both low pH and lysine signals (Fig. 3b). These results suggest that proteolytic processing is a necessary but not sufficient step for CadC activation.

Figure 3.

Effects of low pH and lysine on CadC proteolysis. (a) Immunoblot (IB) analysis of the CadC protein in the YK5005 (cadA::lacZ ΔcadC) strain harboring pACYC184-HA-CadC. Cells were grown in E glucose medium to an OD600 of 0.6 and subjected to pH 5.8 and 10 mM lysine individually or in combination. The samples were collected at the indicated times, and total proteins were immunoblotted with anti-HA antibodies. (b) β-Galactosidase activity of cadA–lacZ expression following 30 min of stress. The assay was performed using the same cells described in (a). A representative data set is shown.

LysP inhibits induction of cadBA in S. Typhimurium

It is known that the lysine-specific permease LysP of E. coli acts as a negative regulator of the cadBA operon in the absence of exogenous lysine (Neely et al., 1994; Neely & Olson, 1996). A recent study has also shown that E. coli CadC is inactivated through an interaction with the lysine permease LysP in the absence of exogenous lysine (Tetsch et al., 2008). However, whether LysP functions similarly in Salmonella has not been determined. Prediction of the transmembrane segments using the DAS program (Stockholm University, Sweden) suggests that S. Typhimurium LysP is a multiple membrane-spanning protein (data not shown). To determine whether LysP inhibits the induction of cadBA transcription in S. Typhimurium, we compared the expression of a chromosomal cadA–lacZ fusion in the JF3068 (wild-type) and YK5006 (ΔlysP mutant) strains using β-galactosidase assays. Figure 4(a) shows that the YK5006 strain expresses a cadA–lacZ transcriptional fusion, even in the absence of exogenous lysine, indicating that a mutation in the lysP gene confers lysine-independent cadBA transcription.

Figure 4.

Effect of a lysP knockout on cadA expression in the presence or absence of 10 mM lysine. (a) Expression of a chromosomal cadA–lacZ fusion in the JF3068 (wild-type) and YK5006 (ΔlysP mutant) strains was determined using β-galactosidase assays. Cells were grown in E glucose medium to an OD600 of 0.6 and subjected to pH 5.8 in the presence or absence of 10 mM lysine. The samples were collected at the indicated times. Three independent experiments were performed for each strain, and the standard deviation is indicated by error bars. (b) Repression of lysP expression by the lysine signal. RT-PCR analysis of lysP transcription in the UK1 wild-type strain, and immunoblot (IB) analysis of total protein extracts prepared from the ΔlysP strain harboring pACYC184-LysP-HA. Cells were grown in E glucose medium at 37 °C to an OD600 of 0.6 and then lysine (10 mM) was added, and samples were collected at the indicated times. 16S r>RNA and DnaK were used as loading controls.

Although the lysine signal is not directly involved in the proteolytic processing of CadC, it is essential for expression of the S. Typhimurium cadBA operon (Fig. 3). To test the effect of the lysine signal on the transcriptional activity of lysP, RT-PCR analysis was conducted on total RNA isolated from UK1 wild-type cells collected at different intervals following the addition of 10 mM lysine. As shown in Fig. 4(b), expression of lysP mRNA was significantly reduced after lysine addition. To further confirm this observation, immunoblot analysis was conducted on the total protein extracts prepared from the ΔlysP strain harboring pACYC184-LysP-HA. C-terminally HA-tagged LysP (LysP-HA) was expressed under the control of its own promoter. Figure 4(b) shows that the cellular level of LysP-HA decreases rapidly after lysine addition. These results suggest that the lysine signal represses lysP expression, thereby eliminating the negative regulation of CadC activation by LysP.

Discussion

In the present study, a genome-wide search revealed a PTS permease STM4538 as a novel component of CadC signaling in S. Typhimurium (Fig. 1). In particular, we demonstrated that inactivation of STM4538 impaired the proteolytic processing of CadC (Fig. 2). Although it is now clear that STM4538 acts as a positive modulator of CadC activity, questions still remain regarding how this PTS permease affects the proteolytic processing of CadC. One likely explanation is that the PTS permease STM4538 might exert its effects either directly or indirectly by controlling the expression of a gene that encodes a CadC-specific protease. It has been recently demonstrated that bacterial enzymes can also act as regulatory proteins. Some modulate the activity of transcription factors either by protein–protein interactions or by covalent modification, and others have a DNA-binding domain that enables them to act as transcription factors (Commichau & Stulke, 2008). Importantly, the PTS permeases, which are involved in sugar transport, were shown to control the activity of transcription regulators by phosphorylating them in the absence of the specific substrate (Stulke et al., 1998). Moreover, the oligopeptide permease Opp3 affected the expression of genes encoding three major extracellular proteases in Staphylococcus aureus (Borezee-Durant et al., 2009).

Based on all the information gathered to date, we propose the following molecular mechanism of CadC activation in S. Typhimurium. Upon acid stress (low pH and lysine), the dormant membrane-bound CadC is first proteolytically cleaved at the periplasmic domain as a result of a low pH signal. This proteolytic event generates a transmembrane signal that switches on expression of the cadBA operon. The lysine signal represses expression of the lysine permease LysP, which normally blocks transmission of the conformational signal to the cytoplasmic DNA-binding domain. In addition, the PTS permease STM4538 is positively involved in regulation of CadC proteolysis through an unknown mechanism. However, details of the functional interactions between CadC, LysP, STM4538 and unidentified proteases have not yet been elucidated.

In summary, our findings suggest a novel mode of transcriptional control by bacterial enzymes. The identification of STM4538 as a positive modulator of CadC function provides important information for uncovering the molecular basis of the proteolytic activation of CadC. It will be interesting to investigate how STM4538 affects the expression or activity of the unidentified protease.

Acknowledgements

This work was supported by a grant from the Korea Research Foundation funded by the Korean Government (MOEHRD, Basic Research Promotion Fund) (KRF-2008-314-C00328).

Authors' contribution

Y.H. Lee and S. Kim contributed equally to this work.

Ancillary