Contribution of the Cpx envelope stress system to metabolism and virulence regulation in Salmonella enterica serovar Typhimurium

The Cpx-envelope stress system regulates the expression of virulence factors in many Gram-negative pathogens. In Salmonella enterica serovar Typhimurium deletion of the sensor kinase CpxA but not of the response regulator CpxR results in the down regulation of the key regulator for invasion, HilA encoded by the Salmonella pathogenicity island 1 (SPI-1). Here, we provide evidence that cpxA deletion interferes with dephosphorylation of CpxR resulting in increased levels of active CpxR and consequently in misregulation of target genes. 14 potential operons were identified to be under direct control of CpxR. These include the virulence determinants ecotin, the omptin PgtE, and the SPI-2 regulator SsrB. The Tat-system and the PocR regulator that together promote anaerobic respiration of tetrathionate on 1,2-propanediol are also under direct CpxR control. Notably, 1,2-propanediol represses hilA expression. Thus, our work demonstrates for the first time the involvement of the Cpx system in a complex network mediating metabolism and virulence function.


Introduction
An important group of bacterial regulatory sensing systems are the two-component systems, each of which enable bacteria to sense and respond to a specific subset of environmental changes and stress factors [1][2][3]. Two-component systems recognize environmental changes a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 provide evidence that cpxA deletion interferes with dephosphorylation of CpxR under invasion inducing conditions. We propose that in the absence of CpxA under invasion inducing conditions CpxR may be constitutively phosphorylated by another kinase or phosphodonor leading to repression of the SPI-1 regulator HilA. Accordingly, dephosporylation of CpxR by CpxA restores hilA expression by preventing the negative effects of active CpxR on this gene. Thus, this study suggests that both activities of CpxA-phosphorylation and dephosphorylation of CpxR are critical to control CpxR-mediated virulence gene regulation.

Strain and plasmid construction
Bacterial strains and plasmids used in this study are listed in Table 1. E. coli strain JM109 was used for cloning and E. coli strain BL21DE3 for protein expression. SL1344 is a standard virulent strain of S. Typhimurium. The cpxA, cpxR and cpxRA deletion mutants were constructed in S. Typhimurium LT2 as described [25][26][27], confirmed by PCR and P22 transduced into the parent SL1344.
The coding region of cpxR was cloned into the NcoI and BamHI sites of pIVEX2.4, resulting in pIStmCpxR. pSSS11 was achieved by cloning the cpxRA coding region into the BamHI and SalI sites of pACYC184. All constructed plasmids were confirmed by sequencing. DNA manipulation, restriction digestion, ligations and transformations were performed using standard genetic and molecular techniques [35].

Measurement of gene expression
The activity of the hilA-GFP fusion encoded on the plasmids pD2E, was measured under SPI-1 inducing growth conditions (LB-pH7.0 with 0.3M NaCl under oxygen limiting static conditions). A fresh colony was inoculated into 5 ml SPI-1 media and grown for 2 hours (OD600 of 0.5). For fluorimeter measurements cultures were diluted 1:50 in 150 μl SPI-1 Media and transferred into Costar 96 black clear bottom plates (Corning Life Sciences; The Netherlands). Plates were sealed with adhesive sealing films (Roth, Germany) and cell growth (OD620) and GFP-production were monitored every 17 minutes over a timeline of 6 h in a Fluorimeter (BD Biosciences). The fluorescent values were measured at excitation values set at 485 nm and emission values set at 540 nm. Assays were performed with an n = 6 and normalized to the wildtype control.

RNA isolation and labelling for DNA microarray
Total bacterial RNA was isolated from bacteria grown under different growth conditions after killing by the addition of 0.2 volumes of 95% ethanol, 5% phenol, pH 4.3. Pellets were resuspended in 10 mM Tris, 1 mM EDTA containing 2 mg ml-1 lysozyme and incubated at 37˚C for 30 min. Cell lysis solution (Qiagen, Hilden, Germany) was added and the mixture was incubated at 65˚C for 5 min and at room temperature for 10 min. After the addition of precipitation solution (Qiagen, Hilden, Germany) and incubation on ice for 5 min cell debris, proteins and DNA were pelleted. The RNA containing supernatant was mixed with ethanol and loaded on a spin column (Promega). Further RNA purification and DNase digestion was done as described by the manufacturer.
A total of 50 μg of RNA of six separate experiments was reverse transcribed to cDNA and labelled with Cy3-or Cy5-conjugated dCTP (GE Healthcare) using reverse transcriptase (SupersciptII, Invitrogen) and random hexamers as primers. RNA was removed by hot-alkali treatment. Labelled cDNA was purified using a Qiaquick PCR purification kit and quantified by Nano-Drop analysis (ND-1000 Spectrophotometer, Peqlab).

DNA microarrays and data analysis
Slides containing three replicate arrays spotted onto CMT-UltraGAPS (Corning) slides were prehybridized in 25% formamide, 5 x SSC and 0.1% SDS at 42˚C. Probes were prepared by mixing equal amounts of Cy3-or Cy5-labelled cDNA from wild type and the cpxA mutant strain with an equal volume of hybridization solution (50% formamide, 10 x SSC and 0.2% SDS). After hybridization and washing, arrays were scanned with a Microarray Scanner BA (Agilent Technologies) at 5 μm resolution. Raw microarray image data were processed with the Image Analysis / Feature Extraction software G2567AA (Version A.7.5, Agilent technologies). Data analysis was carried out on the Rosetta Inpharmatics platform Resolver (Rosetta Biosoftware, Built 5.1). A color-swap dye reversal experimental setting was applied [36]. Ratio profiles comprising single hybridizations were combined in an error-weighted fashion to create ratio experiments. A two fold change expression cut-off ratio was applied together with anti-correlation of ratio profiles rendering the microarray analysis set highly significant (P-value > 0.01), robust and reproducible. Bioinformatics A S. Typhimurium specific weight matrix which predicts the affinity of phosphorylated CpxR for a given DNA sequence was developed using as input files the promoter regions of operons cpxRA, cpxP, motABcheAW, tsr, spy, yihE-dsbA, degP (htrA), ppiA, yccA and rpoE-rseABC. These promoters were chosen because among the best defined CpxR targets [37] these were identified to be strongest affected in SHS01 under invasion-inducing conditions by our microarray analysis (S1 Table). We aligned 500-bp of the promoter regions upstream of the start codons of these operons with the motif-finding program Target Explorer which is based on the Gibbs sampling algorithm [38]. A conserved 15-bp motif was identified for each promoter (see S2 Table). These 10 motifs were used to calculate a S. Typhimurium specific weight matrix which was then used to search both strands of the genome (GenBank accession number AE006468) and the pSLT plasmid (GenBank accession number AE006471) of S.

RNA extraction and RT-qPCR
Total RNA from cells grown under invasion inducing conditions was isolated after stabilizing the RNA with RNAprotect (Qiagen, Hilden, Germany). Cell pellets were resuspended in 10 mM Tris, 1 mM EDTA containing 2 mg ml-1 lysozyme and incubated at 37˚C for 30 min. Cell lysis solution (Qiagen, Hilden, Germany) was added and the mixture was incubated at 65˚C for 5 min and at room temperature for 10 min. After the addition of precipitation solution (Qiagen, Hilden, Germany) and incubation on ice for 5 min cell debris, proteins and DNA were pelleted. The RNA containing supernatant was mixed with ethanol and loaded on a spin column (Promega, WI, USA). Further RNA purification and DNase digestion was done as described by the manufacturer (SV total RNA isolation system, Promega, WI, USA). In total, 1 mg of total RNA was reverse transcribed with the reverse transcription kit (Qiagen, Hilden, Germany). Diluted cDNA samples were used as templates in Real-time qPCR analysis using specific primer pairs and SYBR Green fluorescent dye. Real-time PCR was performed using PowerSYBR Green PCR Mastermix on a 7500 Fast PCR Cycler (AppliedBiosystems, Carlsbad, CA, USA). Uniformity of the product was checked for every PCR by the determination of a dissociation curve. Pairs of primers with lengths of 19-21 nucleotides were optimized for use at an annealing temperature of 51˚C. Each primer pair amplified a fragment of 150-250 bp. Relative expression ratios were determined by the DDCt method as described [43] and normalized to the level of 16S-RNA as a constitutive control. Experiments were repeated nine times with biological triplicate performed in technical triplicate reactions/cDNA dilution.

Gel shift assay
The promoter regions of the genes eco (301 bp), pgtE (251 bp), pocR (301 bp), ssrB (301 bp) and tatA (251 bp) were generated by PCR using the Salmonella SL1344 genomic DNA as template. As control a 156 bp fragment of the Salmonella cpxP promoter region without the CpxR~P recognition motif was used. Binding reactions were performed as previously described for RovM using the purified His6-CpxR [44].

Polymyxin B survival assay
The sensitivity to polymyxin B was performed as described previously [45]. The overnight cultures of SL1344, NOS01 and NOS01 <pIStmCpxR> were diluted into fresh LB pH 8.0 medium with the colony-forming units (CFU) being kept constant at 4 x 105 CFU ml-1. Cells were allowed to grow to OD600 = 0.14 at 37oC. 1 ml of the cultures were incubated with six different concentrations (ng ml -1 ) of polymyxin B: 0, 100, 101, 102, 103, 104, 105, 106 in polypropylene tubes and allowed to stand at 37˚C for 1 h. After incubation cultures were washed twice with 1 x PBS to remove the remnant polymyxin B hindering the growth of bacteria when plated. During plating, 20 μl of the bacterial cells treated with the appropriate concentrations of polymyxin B was mixed with 20 ml of molten Tryptic Soya Agar (55˚C) and the plates were incubated overnight at 37˚C. All experiments were done in triplicates. For control purpose the experiments were performed for a phoP::kan strain. The number of CFU from each plate was counted and the % survival was determined taking the CFU counted from the plates with nonpolymyxin B treated bacteria.

DNaseI footprinting
For DNase I footprinting, different segments of the tatA and ssrB promoter regions were amplified by PCR using a digoxigenin (DIG)-labelled primer and a non-labelled primer. Different primer combinations were chosen for the footprinting analysis of coding and non-coding strands ( Table 2). The amplified promoter regions were 529 and 530 bp for ssrB-A (DIGcoding strand) and ssrB-B (DIG-Non-coding strand) respectively. The amplified promoter region was 524 bp, for tatA-A (coding strand) and tatA-B (non-coding strand). PCR fragments were purified by Nucleospin plasmid kit (Macherey-Nagel) and incubated with the purified phosphorylated His6-CpxR protein in 20 μl of DNA-binding buffer as described for the gel shift assays. The PCR products were digested with DNaseI of an appropriate dilution and the resulting products were separated and visualized as described [44]. The protected bands were identified by comparison with a sequence ladder generated with the same DIG-labelled primer used for PCR amplification of the fragment by using the Thermo Sequenase cycle sequencing kit (USB).

Cell growth on 1,2-propandiol
Anaerobic growth on 1,2-propaendiol was determined according to an established protocol [24]. In brief, tubes were filled with NCE glycerol with or without 80 mM propanediol and preincubated in an anaerobic chamber (Oxoid) with N2 gas for 24 h. Cells were grown aerobically in NCE glycerol to stationary phase, washed in NCE and diluted to an turbidity at 650 nm of 0.1. The tubes were crimp capped and flushed with N2 gas. The cultures were incubated at 37˚C with shaking and turbidity was monitored with a tube spectrophotometer (Riele PM310) at 650 nm.

Statistical analyses
Statistical analyses were performed using the Student's t-test (two-tailed). A P-value < 0.05 was considered significant.

CpxA-mediated dephosphorylation of CpxR is required for hilA expression
The Cpx envelope stress system is implicated in the invasion process of Salmonella into nonphagocytic cells which depends on the SPI-1 T3SS with HilA as the key transcriptional regulator [19,20,46]. The expression of hilA at low pH requires the sensor kinase CpxA but not the response regulator CpxR [47]. We asked whether hilA expression might also be dependent on the Cpx envelope stress system under in vitro conditions that mimic invasion. Therefore, we investigated the expression of a plasmid-coded hilA-GFP fusion under invasion inducing conditions in a high-salt LB medium under low oxygen tension for S. Typhimurium knock out strains SHS01 (cpxA::kan), NOS01 (cpxR::kan) and VSM01 (cpxRA::kan) in comparison to the wild-type strain SL1344. We found under these invasion inducing conditions the expression of hilA to be dependent on CpxA but not on CpxR (Fig 1) and confirmed that this effect is not traceable in standard LB medium (S1 Fig) [20]. Interestingly, we observed no effect on hilA expression for the S. Typhimurium cpxAR double deletion mutant VSM01 (Fig 1, dark grey bars). It is well established that CpxR can be phosphorylated independent on CpxA by the small phosphodonor acetyl-phosphate in vivo [48-50] and that CpxA acts under non-inducing conditions as a phosphatase of phosphorylated CpxR [6,51]. We would like to propose that CpxR is constitutively phosphorylated in a S. Typhimurium cpxA knock out grown under invasion inducing conditions leading to repression of hilA transcription. Therefore, our finding suggests that dephosphorylation of CpxR by CpxA might be critical for hilA expression.

Genome wide screen for Cpx interactions
From our above results we conclude that CpxR is constitutively active in a cpxA knock out strain (SHS01) under invasion inducing conditions. Consequently, we determined the extent of the Cpx envelope stress system in S. Typhimurium by global transcriptional analysis of a cpxA knock out strain (SHS01) grown under invasion inducing conditions with high osmolarity (0.3 M NaCl) and found 392 genes to be differently regulated (S1 Table). The CpxR target gene cpxP was the most strongly affected gene supporting our hypothesis that deletion of CpxA results in high level of phosphorylated CpxR. This supports the widely accepted hypothesis that CpxP can be only switched on by activation via the stress pathway [53]. To predict among these genes those that are under direct control of CpxR, we compared the data of the global transcriptional analysis with a bioinformatic approach using a S. Typhimurium specific CpxR recognition motif ( Fig 2B; S2 Table). This S. Typhimurium specific CpxR recognition motif was generated by using among the best defined CpxR targets in E. coli [37] the promoter regions of those 10 CpxR targets that were strongest affected in the S. Typhimurium cpxA strain (SHS01) under invasion inducing conditions as determined by the transcriptional analysis (S1 Table) (cpxRA, cpxP, motABcheAW, tsr, spy, yihE-dsbA, degP, ppiA, yccA and rpoE-rseABC). For each promoter a conserved 15-bp motif was identified (S2 Table) and these 10 motifs were used to calculate a S. Typhimurium specific weight matrix ( Fig 2B). conditions. Genes for which the ratio of the cpxA-null strain to the wild-type strain is e 4 fold with a P value of <0.005 were clustered using Tibco Spotfire DecisionSite default. B. Bioinformatics-based screening for direct CpxR targets. The sequence logo for the CpxR~P recognition weight matrix in S. Typhimurium is given with the left (I) and right pentamers [28] and the 5-bp spacer marked by roman numerals. The base conservation measured in bits is shown as the relative height of each base [54]. C. Potential novel Cpx regulon members of S. Typhimurium identified by global transcriptional analysis and bioinformatic screening. The left (I) and right [28] pentamers and the 5-bp spacer [28] are displayed as a heat-map to show bases of high conservation (orange) from degenerate regions (light grey/white). The genes controlled by these promoters are indicated to the left of the sequences. https://doi.org/10.1371/journal.pone.0211584.g002 We employed this weight matrix to screen the whole S. Typhimurium LT2 genomic and pSLT sequences for possible CpxR recognition motifs by the use of three different motif-finding programs applying different algorithms each: ScanACE, Prodoric Virtual Footprint and GeneSoap [39, 40, 42]. 106 potential CpxR recognition motifs were commonly identified by all three programs in an appropriate distance of less than 500 bp to an annotated open reading frame (S3 Table). The comparison of bioinformatics and global transcriptional analysis data resulted in 25 possible CpxR target genes (Fig 2C). Along with 11 known genes (chaA, cpxP, cpxR, dsbA, motA, psd, tsr, ppiA, rpoE, spy, yccA) [37, 55-57] 14 potential novel CpxR target genes could be identified (aldB, bioA, eco, glpT, pfkB, pgtE, ssrB, STM1269, STM2613, STM 3681,STM4562, tatA, ybiJ, yciG) (Fig 2C). These potential novel CpxR target genes were clustered into genes with uncharacterized products (STM1269, STM2613, STM3681, STM4562, ybiJ, yciG), products implicated in metabolism (aldB, bioA, glpT, pfkB) or Salmonella virulence (eco, pgtE, ssrB, tatA). For our further studies we focused on four genes that have been demonstrated to be important for Salmonella virulence: eco, pgtE, ssrB and tatA.
The eco gene encodes the serine protease inhibitor ecotin. Ecotin orthologues are present in many Gram-negative bacteria and have been shown to be important for protecting the bacteria against eukarytotic proteases that have translocated across the damaged outer membrane [58]. PgtE is a member of the omptin family of outer membrane aspartic proteases identified in Gram-negative bacteria [23]. PgtE and its closest homologue of Pla of Yersinia pestis attack on the innate immunity for instance by means of inactivating antimicrobial peptides or by affecting the plasminogen/plasmin system by cleavage of the plasminogen activator inhibitor 1 [59][60][61]. SsrB is the response regulator of the SsrAB two-component system which is essential for the coordinated expression of a second T3SS encoded on the SPI-2 and almost all of its accessory effector proteins [62,63]. Salmonella requires the SPI-2 T3SS for intracellular survival and persistence in macrophages [62,64]. The tatA gene belongs to the operon for the twinarginine translocation [23] complex TatABC which promotes the secretion of folded proteins across the cytoplasmic membrane and has recently been demonstrated to be implicated into Salmonella invasion and resistance to antimicrobial peptides [22,65].
To validate the identification of eco, pgtE, ssrB and tatA as direct CpxR target genes in S. Typhimurium, real-time qPCR and electrophoretic mobility shift assays (EMSA) were performed (Fig 3). Cells were grown under invasion inducing conditions and harvested at the end of mid log phase for RNA preparation. The genes cpxP and motA served as controls. The expression of all tested genes was dependent on CpxR ( Fig 3A). We found eco (1.9 fold), ssrB (8.2 fold) and tatA (2.6 fold) to be under positive and pgtE (-2.1-fold) to be under negative CpxR control. In order to demonstrate direct binding of CpxR to the promoter regions of eco, pgtE, ssrB and tatA we performed electrophoretic mobility shift assays (EMSA) with purified, phosphorylated His6-CpxR (CpxR~P) tagged protein (Fig 3B). The mobility of all DNA fragments covering the single promoter regions were retarded in the presence of phosphorylated His6-CpxR (Fig 3B). To confirm the species specificity displayed by CpxR~P in binding to tatA promoter of S. Typhimurium, we performed the EMSA for E. coli tatA promoter using CpxR~P and found no significant shift (S2 Fig). Together, these results demonstrate that the eco, pgtE, ssrB and tatABC operons are direct CpxR targets in S. Typhimurium.

The Cpx envelope stress system activates Salmonella motility
To our surprise, global transcriptional analysis and qRT-PCR revealed motA to be under positive control (50 fold) of the Cpx envelope stress system. In contrast, the genes for the flagellar motor (motABcheAW) are under negative CpxR control in E. coli [37,66]. We verified phenotypic difference between both closely related organisms by motility assay (Fig 4). Motility of S. Typhimurium was drastically repressed in a cpxR deletion strain and could be restored by the overexpression of CpxR. This finding underscores the remarkable difference in core gene regulation between Salmonella and Escherichia as has also been described for the tolQRA cluster, SlyA, DegP, RpoS stability and copper homeostasis [67][68][69][70][71].

A cpxR mutant displays increased sensitivity to polymyxin B
We identified the omptin PgtE as a direct Cpx target. PgtE confers resistance towards the cationic antimicrobial peptides (CAMPs) protamine and polymyxin B [60]. Very recently, it has been proven that the Cpx-system contributes to resistance to the CAMPs protamine, magainin-2 and melittin through regulated expression of the two Tat-system dependent amidases AmiA and AmiC [22]. AmiA and AmiC do not contribute to bacterial resistance to the CAMPs HNP-1 and polymyxin B [22]. In order to investigate whether the Cpx pathway also contributes resistance to polymyxin B, we performed a polymyxin B resistance assay. As shown in Fig 5 resistance against   be partially restored by the overexpression of CpxR indicating that the Cpx-system modulates the resistance to CAMPs in S. Typhimurium by regulating PgtE.

Detection of the CpxR binding sites for the ssrB and tatABC promoter regions
To map the precise location of the CpxR binding sites for the ssrB and tatABC promoter regions we performed DNase I footprint assays. For the ssrB promoter, a 250 bp DIG-labelled DNA fragment including the CpxR recognition motif was incubated with increasing concentrations of purified and phosphorylated His6-CpxR (CpxR~P) prior to DNaseI on the noncoding strand from positions +19 to +51 downstream of the transcriptional start site treatment (Fig 6A). The analysis of both strands revealed one binding region for phosphorylated His6-CpxR overlapping the CpxR recognition motif (Fig 6A and 6B). We checked the coding strand of ssrB for binding of CpxR~P but did not observe significant binding (S3 Fig). This result also indicates that phosphorylated His6-CpxR binds to its consensus motif on ssrB's non-coding strand specifically and serves as a direct regulator of ssrB's expression.
To investigate the binding of CpxR to the tatABC promoter region, a 250 bp DIG-labelled DNA fragment including the CpxR recognition motif was subjected to DNase I footprint analysis (Fig 7A and 7B). In contrast to the ssrB promoter region, binding regions for phosphorylated His6-CpxR were located on both strands (Fig 7A and 7B). The binding regions on the coding strand flanked the Cpx recognition motif (Fig 7A and 7C) and overlapped the CpxR recognition motif on the non-coding strand (Fig 7B and 7C).
Our data clearly assign the tatABC operon as a direct Cpx target in S. Typhimurium. In contrast, the tatABC operon of E. coli has been described to be constitutively expressed [72]. In order to clarify the discrepancy of tatABC regulation between S. Typhimurium and E. coli we compared the structure and promoter sequences of the tatABC operons of both organisms. The tatABC promoter of E. coli is substantially shorter than that of S. Typhimurium and lacks the Cpx recognition motif (Fig 7C). We asked whether the tatABC operon might be CpxRdependent regulated in pathogenic E. coli strains or other pathogenic genera like Shigella, Yersinia or Legionella. The Cpx recognition motif could only be identified in the tatABC promoter region of different Salmonella strains (Fig 7D) indicating an adaptation of Salmonella to a specific niche that depends on a substrate of the Tat

The Cpx envelope stress system inhibits anaerobic growth on 1,2-propanediol
From the above results we hypothesized, that the efficient transport of one or several specific substrates of the Tat-system might be the reason that the tatABC operon is adirect CpxR target in Salmonella but not in other Gram-negative pathogens. The α-subunit of the tetrathionate reductase TtrA was described as a Salmonella specific Tat-dependent substrate [74,75]. The intriguing ttrBCA operon also includes TtrB subunit with a Tat motif where the invariant arginines are swapped with lysines [76]. The tetrathionate reductase is a membrane-bound enzyme that contains the guanine dinucleotide cofactor as a prosthetic group and confers anaerobic respiration of Salmonella on tetrathionate as terminal electron acceptor [24,74,77,78].
Tetrathionate respiration is coupled with the degradation of 1,2-propanediol and ethanoleamine, respectively, that act as electron donors and that are both abundant in the human intestine [24,76]. Interestingly, 1,2-propanediol was described to repress hilA expression [47]. Among the 28 novel CpxR target genes with well annotated functions identified by bioinformatics based screening, we found two genes implicated in 1,2-propanediol utilization: glhA and pocR (Fig 8B) not being differently regulated under invasion inducing conditions. The glhA encodes a glycerol dehydrogenase with 1,2-propanediol as substrate and is a suppressor of cpxA in S. Typhimurium SL1344 rescuing hilA expression [47]. The pocR encodes an AraC type regulator modulating the cobalamine synthesis [79] and the 1,2-propanediol utilization (pdu) gene cluster [80][81][82]. EMSA verified the regulator PocR as a direct CpxR target (Fig 8C). In order to analyze the impact of the Cpx pathway on 1,2-propanediol utilization, we analyzed the fermentative growth on 1,2-propanediol according to an established protocol [24]. Of note, fermentative growth on 1,2-propanediol depends on dilute yeast extract (0.2%) as additional carbon source that cannot support anaerobic growth alone [24]. Fermentative growth on 1,2-propanediol was enhanced in a cpxR deletion strain (Fig 8D) supporting our assumption that the Cpx-system modulates through the regulator PocR the utilization of 1,2-propanediol as the electron donor for the alternative electron acceptor tetrathionate in anaerobic respiration.

Conclusions
In the last decades the Cpx-envelope stress system of Gram-negative bacteria has been extensively studied and assigned to be crucially involved during the invasion of host cells [4,5,14,83]. Previous studies revealed for the food-born pathogen S. Typhimurium that deletion of the sensor kinase CpxA represses expression of the SPI-1 T3SS regulator HilA and attenuates virulence but surprisingly, deletion of the response regulator CpxR had neither effect on hilA expression nor on virulence when the bacteria were grown under mild acid conditions [20]. Here, we confirmed this effect for S. Typhimurium SL1344 grown in SPI-1 inducing medium. Interestingly, we observed in addition that a S. Typhimurium strain lacking CpxA and CpxR showed no decrease in hilA expression. Notably, CpxA represents a bifunctional sensor kinase that is able not only to phosphorylate its signaling partner CpxR but also to hold the balance between active and inactive CpxR by dephosphorylation [6,51]. In previous studies it has been reported that CpxR can be phosphorylated by additional phosphate donors as the acetyl-CoA pathway, independent of CpxA [49]. In this case, CpxR would remain in an activated phosphorylated state in the absence of CpxA and therefore CpxR would be able to inhibit hilA expression. Interestingly, similar effects were reported for the QseCB quorum sensing system involved in virulence regulation of uropathogenic E. coli (UPEC) [50]. Deletion of the sensor kinase QseC but not of the response regulator QseB significantly attenuated intracellular bacterial community formation and virulence [50]. Moreover, a qseBC deletion mutant behaved like wild-type strain contradicting the hypothesis that QseC may function through different response regulators but suggesting that dephosphorylation of QseB is required for virulence gene expression [50]. In agreement with this we suggest that the phosphatase activity of CpxA is important for hilA expression by inactivating the inhibitory effect of phosphorylated CpxR.
Here, we identified 1,2-propanediol degradation system (pocR) and the Tat-system (tatABC) which are involved in the expression of hilA to be under direct control of CpxR (Fig  8) [47, 65]. Interestingly, these two operons are functionally linked. Together they promote the anaerobic respiration of tetrathionate as electron acceptor on 1,2-propanediol as electron donor [24]. Anaerobic respiration on tetrathionate is a differential ability of Salmonella and therefore used in clinical diagnostic in a standard enrichment medium [74,84,85]. The ttrABCRS cluster responsible for tetrathionate respiration is located on SPI-2 [62,74]. It was show that reactive oxygen species generated during inflammation, caused by the activity of both S. Typhimurium T3SSs [86], react with endogenous sulphur compounds to form tetrathionate which inhibits coliforms [77]. Consequently, tetrathionate respiration has been linked to Salmonella specific host adaptation that results in a growth advantage for S. Typhimurium over the competing microbiota in the lumen of the inflamed gut [77]. 1,2-Propanediol, the electron donor for the anaerobic respiration of tetrathionate, is a fermentative product from rhamnose and fucose catabolism, two sugars commonly present in the mammalian intestinal  [28] pentamers and the 5-bp spacer [28] are displayed as a heat-map to show bases of high conservation (dark grey) from degenerate regions (light grey/white) (compare Fig 2). The genes controlled by these promoters are indicated to the left of the sequences. C. EMSA analysis verifies CpxR binding to the pocR promoter. PocR promoter fragment (p) was incubated without or with increasing amounts of the purified and phosphorylated CpxR protein. The DNA-CpxR~P complexes were separated on 4% polyacrylamide gels. The corresponding molecular weights are indicated on the left. The positions of the promoter fragments are indicated (p), arrows show the higher molecular weight DNA-CpxR~P complexes. A fragment of the cpxP promoter region without the CpxR~P binding motif (-151 to -297) was used as negative control (c). D. Stimulation of anaerobic growth by 1,2-propanediol (PDO) was determined according to an established protocol [24]. Cells of wild-type S. Typhimurium strain SL1344, the cpxR mutant NOS01, and the cpxRA overexpressing strain NOS01<pSSS11> were grown anaerobically on minimal NCE medium supplemented with 0.2% yeast extract as carbon source with or without 80 mM PDO as energy source. Shown are the means ± S.E.M. of biological quadruples (t-test).
https://doi.org/10.1371/journal.pone.0211584.g008 tract by e.g. plant cell walls [87]. The degradation of 1,2-propanediol is a complex process performed in an specific 1,2-propanediol utilization (Pdu) microcompartment in a Vit B12 dependent manner [79,88]. Salmonella upregulates proteins for the utilization of 1,2-propanediol (pdu-cob cluster) under invasion-mimicking conditions [89]. Expression of the pdu-cob cluster is inhibited by PocR, induced by 1,2-propanediol and globally controlled by the ArcA and Crp systems [90]. The combined results indicate that deletion of CpxA results in active CpxR inducing expression (Fig 8). The PocR regulator inhibits expression of the pdu-cob cluster resulting in the accumulation of 1,2,-propanediol (PDO) and finally in hilA repression. Thus, our finding that the regulator for the degradation of 1,2,-propanediol PocR is under direct CpxR control suggests that the Cpx envelope stress system links metabolism with virulence regulation.
However, the exact mechanism of how the degradation of 1,2-propanediol and the Tat-system impacts the expression of hilA is not known. Transcription of hilA is controlled by a complex feed-forward loop including RtsA, HilC and HilD [46]. Environmental signals feed into this network and mediate hilA expression through post-transcriptional or post-translational control of hilD. One of the known post-transcriptional hilD regulators is the EnvZ/OmpR two-component system [46]. Recently, it was demonstrated that the CpxR target MzrA modulates the activity of EnvZ by direct protein-protein interaction in E. coli [91][92][93]. Accordingly, MzrA can be defined as an auxiliary protein that connects two two-component systems [94]. We identified the mzrA gene by global transcriptional analysis to be under global control of the Cpx envelope stress system. Of note, we also identified CpxR recognition motif in the mzrA promoter region in S. Typhimurium as described in E. coli suggesting that MzrA might play the mediator protein role between the CpxAR and the EnvZ/OmpR systems to modulate hilA expression (Fig 8 and S2 Fig) [92].
Identification of the Tat-system (tatABCD) as direct CpxR target in S. Typhimurium was unexpected since the tatABCD operon was described to be constitutively expressed in E. coli [72]. The Tat-system facilitates the export of cofactor-containing proteins across the cytoplasmic membrane [95]. In E. coli, the Tat-system consists of TatA, TatB, TatC, TatD and TatE proteins that are encoded on the tatABCD operon and tatE [96] respectively. The promoter region of the tatE gene that is thought to be a cryptic gene duplication of tatA does not consist of a CpxR recognition motif. Interestingly, a CpxR recognition motif could also not be identified upstream of the tatABCD operon in several other species including pathogenic E. coli strains, indicating that the regulated expression of tatABCD by CpxR is species-specific ( Fig  6D). In line with this is the observation that for several pathogens the effects for tat mutants vary including growth rate, motility, biofilm formation, host colonization and virulence [65,[97][98][99]. As an example, tat mutants of Escherichia, Agrobacterium and Pseudomonas become non-motile [97,100,101] whereas no effect on Vibrio motility could be observed [98]. A Salmonella specific substrate of the Tat-system is the A subunit of the anaerobic tetrathionate reductase [74]. As stated above, anaerobic respiration on tetrathionate promotes Salmonella a growth advantage over the competing microbiota [77]. Moreover, two Tat-system dependent amidase (AmiA and AmiC) that confer resistance to cationic antimicrobial peptides (CAMPs) were shown to be under direct CpxR control [22]. Taking our experimental findings into consideration we conclude that the Cpx-dependent expression of the Tat-system is important in the ecology of Salmonella.
In addition to the disparate regulation of the Tat-system, we found a phenotypic variation of Cpx-dependent motility (motABcheAW). Motility of S. Typhimurium is under positive CpxR control, whereas E. coli motility is negatively regulated by the response regulator [37]. A phenotypic difference between E. coli and S. Typhimurium was first described for the resistance to the antimicrobial peptide polymyxin B, which is governed by the PmrA/PmrB system [102]. In contrast to E. coli, S. Typhimurium is resistant to polymyxin B under low Mg2+ condition [45]. In S. Typhimurium but not in E. coli dephosphorylation of the PmrA response regulator is prevented by the PmrD connector protein expressed by the PhoP/PhoQ system and responds to low extracellular Mg2+ [102,103]. Moreover, many studies have underlined the difference in regulatory strategies between Salmonella and Escherichia coli [67][68][69][70][71]. Thus, our finding underlines the remarkable diversity in regulatory circuits between the closely related species Escherichia and Salmonella and exemplifies the assumption that disparate regulation of conserved genes has consequences for the ecological niches bacterial species can colonize [102].
Phenotypic differences among related bacteria are mainly assigned to species-specific genes. Here, we identified the genes for the SPI-2 regulator SsrB and the omptin PgtE as direct CpxR targets (Fig 8). SsrB is the response regulator of the SsrAB two-component system that activates the expression of the SPI-2 T3SS. In contrast to many other two-component systems expression of the sensor kinase SsrA and the response regulator SsrB is independently regulated from each other [73]. SsrB directly controls the expression of the SPI-2 T3SS and its effectors that provide Salmonella to survive in host cells [63,104]. It has been reported that the SPI-2 T3SS is also expressed in the intestine independently of the invasion process but without substrate protein secretion [79] supporting a previous suggestion that the acidic pH environment typical for the interior of macrophages might be an essential trigger for the secretion of SPI-2 T3SS substrate proteins [104]. Consistent with this, we found no difference in the secretion pattern for the SPI-2 T3SS substrate protein SseB (S5 Fig). These data also support the observation that deletion of the Cpx-system results in only slight attenuated growth of S. Typhimurium in macrophages [21]. Accordingly, our results indicate that the Cpx-system might be involved in preparing Salmonella for its life in host cells.
PgtE belongs to the omptin family of outer membrane β-barrel proteases that promote virulence associated functions of different pathogens [105]. Pla of Yersinia pestis is located on the virulence plasmid pPCP1, and advances the migration of the plaque bacteria through tissues [106]. SopA of Shigella flexneri is encoded on the plasmid pWR100 and is important for the intracellular mobility [105]. PgtE of Salmonella enterica promotes resistance towards CAMPs like polymyxin B and protamine [107]. CAMPs as part of the innate immune system are typically amphiphatic peptides of 12-45 residues length with wide variations in their sequences and secondary structures [108,109]. Very recently, it has been proven that the Cpx-envelope stress system confers resistance to the CAMPs protamine, magainin-2 an melittin through regulated expression of the two Tat-system dependent amidase AmiA and AmiC [22]. These Tatdependent amidases contribute to bacterial resistance to the CAMPs protamine, magainin 2 and melittin but not to the CAMPs HNP-1 and polymyxin B [22]. Recent studies in S. Typhimurium carrying pgtE deletions had 2-fold lower minimum inhibitory concentrations (MICs) to two CAMPs namely human LL-37 and its murine ortholog CRAMP [107], while pgtE overexpression increased the MIC by 8-fold (Band and Weiss 2015). Here, we demonstrate that the Cpx-system contributes also resistance to polymyxin B. Other CpxR targets that are known to confer resistance to CAMPs are the inner membrane protein of unknown function YqjA (S4 Fig) and extracytoplasmic sigma factor sE (RpoE) [110,111]. As the Cpx-targets PgtE and YqjA which contributes to resistance towards protamine, our finding suggests the regulation of these two loci as the Tat-independent mechanism of Cpx pathway promoted protamine resistance. Collectively, our data point the Cpx-envelope stress system of S. Typhimurium a critical function during the early stage of infection. Starting with the competition with the invasion process (hilA, pocR, tatABC) to the preparation for the escape from the host cell immunity (eco, ssrB). Taking all findings together we are reporting here a comprehensive model depicting network regulated by Cpx pathway (Fig 9). Moreover, it seems that S.
Typhimurium is able to tolerate a complete loss of the whole Cpx system better than an impaired incomplete system [21]. A recent quantitative proteome study identified a need for 10-fold excess of CpxP to inhibit the CpxRA two-component system [112]. The interaction of Cpx-system with other regulatory networks like small non-coding RNAs (sRNAs) has been reported [113].Thus, although the Cpx system appears not to be essential it still can be assigned to an important role in fine-tuning virulence. Further studies could shed light on if the Cpx system is involved in physical interaction with other regulatory networks like sRNAs. Together with the general ability of the Cpx system to sense a wide range of different external stimuli this work supports the notion that the Cpx system plays a central role in a complex network regulating the interaction between pathogen and host.