Abstract
The human-pathogenic yersiniae represent an ideal species group to study the evolution of highly virulent bacteria, with Yersinia pestis having emerged from the enteropathogen Y. pseudotuberculosis an estimated 20 000 years ago. Sequence data reveal that the Y. pestis genome is in the early stages of decay and contains hundreds of non-functioning pseudogenes, some of which may be important in the enteric lifestyle of Y. pseudotuberculosis. Bioinformatic analysis of pseudogenes from seven Y. pestis genome sequences identified rcsD as a gene disrupted early in the evolution of this organism. This phosphotransfer protein is part the of the Rcs phosphorelay, a two-component system present in the Enterobacteriaceae which has been shown to regulate the expression of capsular polysaccharide and other virulence determinants in several species including Escherichia coli and Salmonella. Using microarray analysis, we determined that the Y. pseudotuberculosis Rcs phosphorelay regulates the expression of 136 coding sequences, of which 60 % are predicted to affect the cell envelope. Several putative virulence determinants were identified as being regulated by this phosphorelay, along with proteins involved in biofilm formation, motility, mammalian cell adhesion and stress survival. Phenotypic assays on defined mutants confirmed a role for the phosphorelay in these processes in both Y. pseudotuberculosis and Y. enterocolitica.
-
↵†Present address: Department of Microbiology, University College Cork, Cork, Ireland.
-
The microarray data for this study have been submitted to BUGSBase and ArrayExpress with accession number E-BUGS-57.
-
The online version of this paper includes supplementary material on the verification of the microarray data by the use of Lux constructs of two members of the proposed Y. pseudotuberculosis Rcs regulon.
Edited by: P. van der Ley
INTRODUCTION
The human-pathogenic yersiniae, consisting of the enteropathogenic species Yersinia enterocolitica and Yersinia pseudotuberculosis and the highly virulent plague bacillus Yersinia pestis, represent ideal species to study how bacteria adapt from different environments and evolve to be highly virulent. Population-genetics studies have suggested that Y. pestis recently emerged from Y. pseudotuberculosis (1500 to 20 000 years ago) and thus these bacteria are highly similar at the genomic level (Achtman et al., 1999). However, genomic sequencing of these species has revealed that Y. pestis, despite being highly virulent, is in the early stages of genome decay, with up to 10 % of its genes represented as pseudogenes and presumably non-functional (Parkhill et al., 2001; Chain et al. 2004; Lerat & Ochman, 2005). Amongst these genes are several known to be important in the enteric life style of Y. pseudotuberculosis, including genes for flagella, invasin, urease and cytotoxic necrotizing factor.
All seven Y. pestis genome sequences available on the NCBI blast microbial genomes database (CO92, KIM, 91001, Nepal516, Antiqua, Angola and IP275; ) were therefore analysed for common pseudogenes. This analysis identified rcsD, which contains an identical frameshift mutation in all seven Y. pestis strains but is not frameshifted in either of the two Y. pseudotuberculosis genomes available on the same database (IP32953, IP31758). This indicates that this mutation occurred very early in the evolution of Y. pestis.
RcsD is an intermediate phosphotransfer protein in the Rcs phosphorelay system which was originally identified in Escherichia coli as a regulator of the capsular polysaccharide synthesis (cps/wca) operon responsible for the production of colanic acid (Stout & Gottesman, 1990). Subsequent orthologous systems have been identified in a range of Gram-negative bacteria, mostly in members of the Enterobacteriaceae, including the yersiniae (Huang et al., 2006; Erickson & Detweiller, 2006).
The sensor kinase of this phosphorelay is RcsC, an inner-membrane-associated protein with a single periplasmic domain and several cytoplasmic domains including a histidine kinase domain and a phosphoreceiver domain (Rogov et al., 2006; Majdalani & Gottesman, 2005). Activation of RcsC results in autophosphorylation and the subsequent transfer of the phosphate group to RcsD and finally onto a receiver domain in the RcsB response regulator. Phosphorylated RcsB functions as a transcriptional regulator independently or associated with the unstable auxiliary regulator RcsA, with two different promoter recognition sequences (Wehland & Bernhard, 2000; Majdalani et al., 2002; Sturny et al., 2003). There is some evidence that RcsB is also dephosphorylated by RcsC and possibly also by RcsD as a negative-feedback mechanism (Majdalani et al., 2002).
Several other proteins have been demonstrated to have direct effects on the Rcs phosphorelay. A negative regulator of the Rcs phosphorelay, IgaA, was originally discovered in a screen for Salmonella mutants, which were unable to support intracellular growth in macrophages (Cano et al., 2001). Null mutations of the igaA orthologues in E. coli and Proteus mirabilis (yrfF and umoB, respectively) have been shown to have a high-level constitutive induction of the pathway (Majdalani & Gottesman, 2005).
Under certain circumstances an outer-membrane lipoprotein, RcsF, appears to be important for regulation of the Rcs phosphorelay and acts upstream of RcsC (Majdalani et al., 2005). Also the two-component system PhoP/Q has been shown to modify the Rcs regulon in the presence of high concentrations of exogenous Zn2+ ions (Hagiwara et al., 2003). The mechanisms by which RcsF and PhoP/Q affect the Rcs pathway are unknown and require further study.
Capsular polysaccharide expression and major cell envelope modifications appear to be a common theme of Rcs phosphorelay activation and it has been also proposed to play a role in regulating bacterial virulence and biofilm formation (Arricau et al., 1998; Ferrières & Clarke, 2003; Francez-Charlot et al., 2005; Tobe et al., 2005; Vianney et al., 2005). As yet the precise signal for pathway activation has not been determined, but it appears that the RcsC sensor responds to membrane perturbations such as osmotic stress, desiccation, bile salts and the aberrant expression of certain membrane-associated proteins (Ophir & Gutnick, 1994; Sledjeski & Gottesman, 1996; Clarke et al., 1997; Bernstein et al., 1999; Kelley & Georgopoulos, 1997; Potrykus & Wegrzyn, 2004; Zuber et al., 1995). Activation of this pathway has also been shown to occur during growth on a solid medium (Ferrières & Clarke, 2003).
Given that RcsD appears to have been naturally deselected in Y. pestis as it moved away from its enteric lifestyle dependency, it is reasonable to speculate that RcsD may be important in the enteric survival of Y. pseudotuberculosis and Y. enterocolitica. With this in mind, the aim of this study was to use mutagenesis, microarray analysis and phenotypic assays to determine the role of the Rcs phosphorelay in the enteropathogenic yersiniae.
METHODS
Bacterial strains and growth media.
All bacterial strains are shown in Table 1⇓. Y. pseudotuberculosis strains IP32953 and YPIII pIB1 were obtained from E. Carniel (Institut Pasteur, Paris, France) and H. Wolf-Watz (Umeå University, Umeå, Sweden) respectively. Y. enterocolitica strains WA-C and JB580v were obtained from Alexander Rakin (Max von Pettenkofer-Institut für Hygiene und Medizinische Mikrobiologie, Munich, Germany) and Virginia Miller (Washington University, St Louis, MO, USA) respectively. All strains were maintained on CIN agar supplemented with Yersinia selective supplement (Oxoid) and cultures were grown in LB broth with appropriate antibiotics unless otherwise stated.
Strains, plasmids and mutagenesis primers
Generation of rcsC and rcsD insertion mutants.
The mutation method was a variation on the λ-Red recombinase method (Datsenko & Wanner, 2000) and utilized plasmid pAJD434, which has been optimized for efficiency in Yersinia spp. (Maxson & Darwin, 2004). Plasmids and primers used for the mutant generation are listed in Table 1⇑. Gene-specific primers were designed to introduce a kanamycin-resistance cassette into the intended coding sequence (CDS). These 70-mer oligonucleotides consisted of 50 bases of gene-specific sequence at the 5′ end of the primer whilst the 3′ end consisted of 20 bases designed to amplify the kanamycin cassette from either the pUC4K or the pKD4 plasmid. After optimization, the single PCR product was purified using a QIAgen PCR Purification kit and digested with DpnI to remove the plasmid template.
Wild-type strains were electroporated with the λ-Red plasmid pAJD434 and grown at 28 °C in LB broth containing 0.8 % (w/v) l-arabinose to maintain a high level of expression of the bacteriophage red-gam genes. Once cells had reached OD600 ∼0.7, they were incubated on ice for 20 min before being washed twice in ice-cold H2O and resuspended with the 50 μl purified PCR product. Cells were then electroporated, transferred to a bijou bottle containing 500 μl LB+0.8 % w/v l-arabinose and incubated overnight at 28 °C before being plated out onto selective plates. Mutations were verified by PCR and Southern blotting. Retention of the virulence plasmid pYV was confirmed by PCR and by colony appearance on CRMOX plates (Riley & Toma, 1989). The pAJD434 plasmid is thermosensitive and thus retention of the plasmid was selected against by growth of mutants at 37 °C in LB broth containing 2.5 mM CaCl2 to prevent the loss of the virulence plasmid pYV. For complementation studies the Y. pseudotuberculosis YPIII pIB1 rcsD CDS was cloned into the low-level constitutive expression plasmid pTRC99a whilst a 4.4 kb region of the Y. enterocolitica 8081 genome containing the rcsD CDS along with ORF YE1399 and 421 bp of upstream sequence was cloned into pBAD33.
RNA isolation.
Wild-type YPIII pIB1 and rcs mutants were all transformed with pPSG961-31, a pBAD33-derived plasmid containing the E. coli djlA gene under the control of the ParaBAD promoter (Guzman et al., 1995; Clarke et al., 1996). Wild-type YPIII pIB1 was also transformed with pBAD24 or pBMM608, a pBAD24-derived plasmid containing the Y. pseudotuberculosis rcsB CDS. Cells were grown and maintained in LB broth at 28 °C in the presence of 0.1 % (w/v) glucose to inhibit expression.
For microarray analysis of global RNA expression levels the cells were grown at the appropriate temperature to OD600 ∼0.5 before addition of l-arabinose to a final concentration of 0.2 % (w/v) in order to induce overproduction of either DjlA or RcsB. Cells were then grown for a further 2 h before 2 ml aliquots were removed and added to 4 ml RNAprotect reagent (Qiagen) for RNA stabilization. Total bacterial RNA was then isolated using the RNeasy mini-kit (Qiagen) and all RNA concentrations were determined using a spectrophotometer.
Microarray analysis.
The original Y. pestis CO92 microarray was constructed from spotted PCR products designed and printed at the Bacterial Microarray facility at St George's Hospital Medical School as previously described (Hinchliffe et al., 2003; Stabler et al., 2003). For this set of experiments, DNA sequences representing all 4221 predicted CDSs (4012 chromosomal and 209 plasmid-encoded) from Y. pestis CO92 (biovar Orientalis) were amplified and spotted in duplicate onto glass microscope slides to produce the CO92 gene-specific microarray version YPv01.
RNA (10 μg) was denatured at 95 °C in the presence of 3 μg Random Primers (Invitrogen) and snap cooled on ice. The RNA was then reverse transcribed to generate Cy-dye-labelled cDNA using 500 U Superscript II reverse transcriptase (Invitrogen) in the presence of 1× First Strand Buffer, 10 mM DTT, 460 μM dATP, 460 μM dGTP, 460 μM dTTP, 184 μM dCTP and 850 pM of either Cy5-dCTP or Cy3-dCTP (Amersham Pharmacia). After incubation at 42 °C in the dark for 90 min, labelled cDNAs from comparative samples were mixed and purified together using a single Qiagen mini-elute column. Microarray slides were pre-hybridized in 3.5× SSC, 0.1 % SDS, 10 mg BSA ml−1 at 65 °C for 20 min before being washed for 1 min in H2O followed by 1 min in 100 % 2-propanol. The purified cDNA mixture was denatured at 95 °C before being applied to the microarray slide in hybridization solution (4× SSC, 0.3 % SDS). Hybridization was for 18 h at 65 °C, prior to washing the slides once in 1× SSC, 0.05 % SDS at 65 °C, and twice in 0.06× SSC for 2 min. Dye-swaps were performed with the three independent RNA isolations being hybridized to the arrays on two separate occasions. With duplicate spots on the arrays this resulted in a total of 12 replicates for each of the represented reporter elements.
Microarray slides were scanned using a Genetic Microsystems GMS 418 scanner (MWG Biotech), and images analysed using Imagene (BioDiscovery) and Genespring (Silicon Genetics) software. Forty per cent of the data was used to calculate the Lowess curve which was fitted to the log ratio plot. This curve was used to adjust the control value for each measurement. Only data points determined to be ‘Present or Marginal’ by the Imagene software were used in the final analysis. Statistically significant CDSs were determined using a P-value of <0.05 when analysed by t-test with Benjamini and Hochberg false discovery rate as a cut-off.
Adhesion and invasion.
Human HEp-2 cells were grown and maintained in MEM medium supplemented with 10 % (v/v) fetal bovine serum (FBS), 2 mM l-glutamine, 1× MEM non-essential amino acids and 100 U penicillin/streptomycin ml−1. Cells were seeded in 6-well plates at 5×105 cells per well and allowed to grow for 40 h at 37 °C in a humidified CO2 incubator. The tissue culture supernatant was removed and replaced by 2 ml MEM medium, supplemented with 10 % (v/v) FBS only, containing a 10−6 dilution of an overnight bacterial culture (YPIII pIB1 was diluted 10−4 due to its low invasiveness) and serial dilutions were plated out to determine exact bacterial numbers in the inocula. Cells were then incubated for 3 h at 37 °C in a humidified CO2 incubator, after which time the supernatant was removed and cells were washed twice with 3 ml sterile PBS. For the invasion assay, cells were then incubated in 2 ml gentamicin solution (100 μg ml−1 in MEM medium supplemented with 10 % FBS) for a further 2 h at 37 °C in a humidified CO2 incubator before being washed twice in sterile PBS. For the adhesion assays this gentamicin step was omitted. Finally, the HEp-2 cells were lysed by the addition of 1 ml 1 % Triton X-100 in PBS. Cell scrapers were used to ensure all cell debris was removed from the 6-well plate, and 100 μl of the lysed cell solution was plated out. Colonies were counted after 48 h growth at 28 °C. Data were analysed from three separate experiments with triplicate samples for each strain. Statistical analysis was performed using a Student's two-sided t-test assuming equal variances.
Stress survival assays.
Overnight cultures were diluted 10−3 in 100 ml LB broth containing either 1 % Bile Salts no. 3 (Oxoid) or various concentrations of NaCl or sucrose. Cells were then grown with shaking at 37 °C for the bile salt survival assay, or at 20 °C for the NaCl and sucrose survival assays. At various time points aliquots were taken, serially diluted and plated out on Yersinia selective agar in order to count viable bacteria. Data were analysed from a single experiment with triplicate samples for each strain. Statistical analysis was performed using a Student's two-sided t-test assuming equal variances.
Biofilm formation.
Bacterial cultures were grown overnight at 28 °C in LB broth containing appropriate antibiotics. Cultures were then diluted 10−1 in Terrific broth and 100 μl aliquots were transferred to a Nunclon Delta 96-well round-bottom microtitre polystyrene plate and grown overnight at 28 °C with shaking (180 r.p.m.). All strains were grown in triplicate on the same plate. Bacterial cultures were discarded and plates were washed three times with water, fixed with 2.5 % glutaraldehyde for 15 min, washed twice with water and stained with 0.4 % crystal violet for 15 min. After three further washes with water, the bacterial biofilm was destained with 200 μl ethanol/acetone (80 : 20, v/v) for 2 min and the eluted stain was added to 800 μl 70 % ethanol. Biofilm formation was quantified by measuring A570 using a S2000 Lightwave UV-Vis Diode Array spectrophotometer (WPA).
The analysis of biofilms on a biotic surface was performed using the Caenorhabditis elegans model as previously described (Joshua et al., 2003). C. elegans were maintained on NGM agar with E. coli strain OP50 as a food source. For each Yersinia mutant a 1 ml aliquot of an overnight LB broth culture was transferred onto a fresh NGM agar plate and grown overnight at 28 °C. Young adult C. elegans were harvested and washed in M9 medium, and approximately 20 nematodes were transferred to each plate. Plates were then incubated overnight at 20 °C and the presence of biofilms was determined using a light microscope.
Swimming and swarming.
Swimming and swarming phenotypic assays were performed using motility agar consisting of 10 g tryptone l−1 with the addition of 0.3 % Bactoagar (swim) or 0.6 % Bactoagar (swarm). Media for the Y. pseudotuberculosis assays were also supplemented with 5 g NaCl per litre. Filter-sterilized glucose (10 mM final concentration) was added to swarm motility agar immediately before pouring plates. Swim agar plates were inoculated by injecting a 1 μl aliquot of an overnight culture halfway into the agar. Swarm agar plates were inoculated by spotting a 1 μl aliquot of culture onto the agar surface. Plates were then incubated upright at 28 °C.
RESULTS
In order to determine the role of the Rcs phosphorelay in the enteropathogenic yersiniae rcsC : : Km and rcsD : : Km insertion mutants were generated in Y. pseudotuberculosis strains IP32953 and YPIII pIB1 whilst rcsD : : Km insertion mutants were made in Y. enterocolitica strains WA-C and JB580v.
Microarray analysis of the Rcs regulon in Y. pseudotuberculosis strain YPIII pIB1
The Rcs phosphorelay of E. coli can be induced either by overproduction of several membrane-associated proteins such as DjlA, YpdI, LolA and OmpG or by overproduction of the response regulator RcsB (reviewed by Majdalani & Gottesman, 2005). It was therefore assumed that these stimuli would also activate the phosphorelay in Y. pseudotuberculosis. However, to ensure thorough analysis of the Y. pseudotuberculosis Rcs regulon two separate sets of microarray experiments were performed. In the first set of experiments the E. coli DjlA protein was overproduced in wild-type YPIII pIB1, a rcsC : : Km and a rcsD : : Km insertion mutant and their transcriptional responses were compared. In a further set of microarray experiments the RcsB protein was overproduced in the wild-type YPIII pIB1 strain and RNA expression levels were compared to cells containing vector alone. Both the DjlA and RcsB overproduction studies were performed at 37 °C and 28 °C to determine any effects of temperature on the Rcs regulon. Experiments with Lux constructs to verify the microarray data for the putative metalloprotease YPO3973 (YPTB3814) and ShET2-like enterotoxin YPO1002 (YPTB3305) genes are described in the supplementary data available with the online version of this paper.
Comparison of overproduction of DjlA in the wild-type YPIII pIB1 and rcsC : : Km strain identified 40 CDSs as being statistically different at both 28 °C and 37 °C, all of which were more than twofold upregulated in the wild-type compared to the mutant. A further 126 were statistically different at 37 °C only, whilst only 10 were statistically different at 28 °C only. Comparison of overproduction of DjlA in the rcsC : : Km and rcsD : : Km strains resulted in no statistically significant differences, indicating that gene regulation following the activation of Rcs phosphorelay in Y. pseudotuberculosis has a definite requirement for the RcsD phosphotransfer protein.
Overproduction of RcsB in the wild-type YPIII pIB1 resulted in a much larger number of CDSs being regulated, with the expression of 406 CDSs being statistically different at both 28 °C and 37 °C. A further 978 CDSs were statistically different at 37 °C only, whilst 268 were statistically different at 28 °C only. The large number of CDSs affected by RcsB overproduction is probably due to its DNA-binding ability: when expressed in excess it may bind to partial binding motifs in promoter sequences, albeit with a weaker affinity than to genuine RcsB binding motifs.
A comparison of the statistically significant CDSs from both sets of microarray experiments revealed that 136 CDSs were regulated by both RcsB and DjlA overproduction (Table 2⇓). Greater levels of change were noticeable in the RcsB overproduction datasets compared to the DjlA overproduction datasets, presumably due to the fact that these promoters were highly active in the presence of excess RcsB protein.
The Y. pseudotuberculosis Rcs regulon
Normalized data and P-values from all microarray experiments are given for the 136 CDSs identified as being significantly regulated by Rcs pathway activation. ns, Not significant.
Of these 136 CDSs, just over 60 % either localized to the cell envelope as predicted by Protein Subcellular Localization Prediction for Gram Negative Bacteria @BII (), or were involved in modifying cell envelope structures. Many of the regulated genes were putative lipoproteins and membrane-bound proteins, whilst others were involved in peptidoglycan modification (YPO1615, YPO3570, murA, mltB/C and mrcB). Approximately 10 % of the Rcs-regulated genes were involved in biosynthesis and modification of the LPS core (YPO0187, rffC/H/G, wzzE, rfe, rfaD/F, YPO3578) and the O-antigen (YPO3096-3102 and YPO3112-3118).
Most bacterial systems contain a feedback mechanism; however, the expression of rcsB, rcsC and rcsD was unaffected by activation of the phosphorelay. The expression of rcsF was slightly increased (∼1.5-fold) at both 28 °C and 37 °C, but this was not statistically significant. Of the two potential RcsA paralogues, only expression of YPO2955 was significantly upregulated by phosphorelay activation, suggesting that this is the true RcsA. However, expression of YPO0142, the paralogue of IgaA, was also significantly upregulated by phosphorelay activation at 37 °C, indicating that there is a feedback mechanism to inhibit the phosphorelay after activation.
In this study many of the most highly upregulated CDSs are hypothetical proteins but several CDSs could be identified which may be virulence determinants in Y. pseudotuberculosis. The most significantly upregulated CDS was YPO3973, encoding a putative metalloprotease, which was upregulated over 300-fold by RcsB overproduction and 15–20-fold by DjlA overproduction.
Adherence and invasion
The microarray data also identified a range of putative virulence determinants which could be grouped according to putative function. One group comprised CDSs which are involved in binding to mammalian cells: a putative intimin (YPO1562) (Strong et al., 2006), a fimbrial protein (YPO3877), its adjacent chaperone (YPO3878) and a putative fimbriae regulatory protein (YPO2593). Therefore rcs mutants were assayed for adhesion and invasion of the human epithelial cell line HEp-2. The ability of Y. pseudotuberculosis IP32953 to adhere and invade was approximately 100-fold higher than that of the other Y. pseudotuberculosis strain, YPIII pIB1 (Fig. 1a, b⇓). Surprisingly, for both Y. pseudotuberculosis strains, over 90 % of adherent bacterial cells managed to invade the HEp-2 cells, indicating that adhesion is the rate-limiting step. In both Y. pseudotuberculosis strains the rcsC : : Km or rcsD : : Km mutation significantly decreased adhesion, and therefore invasion (two-sided t-test; P<0.01). This was fully restored by complementation.
Adhesion to and invasion of human HEp-2 cells by (a) Y. pseudotuberculosis strain IP32953, (b) Y. pseudotuberculosis strain YPIII pIB1 and (c) Y. enterocolitica strains WA-C and JB580v and their respective Rcs mutants. Complemented strains are designated by pTRCrcsD or pBADrcsD. Results (white bars, adhesion; black bars, invasion) are presented as a percentage of the total initial inoculum and error bars are the standard deviation of triplicate samples from three separate experiments. Statistical analysis was performed using a Student's two-sided t-test assuming equal variances, and for all mutants compared to wild-type P<0.01.
Similarly there was a 10-fold difference between the two Y. enterocolitica strains in their ability to invade HEp-2 cells, and mutation of rcsD significantly decreased invasion of the WA-C strain (Fig. 1c⇑). Complementation of the JB580v rcsD : : Km mutant with the rcsD CDS cloned into a pBAD vector actually increased invasion 10-fold compared to the wild-type. This is probably due to increased expression of rcsD from this vector, indicating that overproduction of RcsD increases the ability of Y. enterocolitica to invade mammalian cells.
Biofilm formation
As previously reported for other species, several of the regulated genes are potentially involved in biofilm formation (uvrY, YPO3570, tolB/Q). Using plastic 96-well plate assays, no biofilms were evident with either of the two Y. enterocolitica strains or the YPIII pIB1 strain of Y. pseudotuberculosis. However, the IP32953 strain of Y. pseudotuberculosis readily formed a biofilm on coated tissue-culture plates. Fig. 2⇓ shows relative biofilm formation of the wild-type, mutant and complemented IP32953 strains grown at 37 °C and 28 °C as measured by crystal violet staining. Biofilm formation at 28 °C proceeded much more slowly than at 37 °C and reached a peak at approximately 24 h, after which time the biofilm did not increase in mass. At both temperatures, and at all time points, the rcsC : : Km and rcsD : : Km mutants produced significantly less biofilm than the wild-type and complemented strain. However, biofilm formation was not completely abolished in the mutant strains. In contrast, using our C. elegans model of biofilm formation on a biotic surface (Joshua et al., 2003) only the YPIII pIB1 strain of Y. pseudotuberculosis was able to form a biofilm and no difference was observed between wild-type and the rcs mutants (data not shown).
Biofilm formation on an abiotic surface by Y. pseudotuberculosis strain IP32953, the Rcs phosphorelay mutants (rcsC : : Km and rcsD : : Km) and the complemented ΔRcsD mutant (pTRCrcsD). Wild-type and mutant strains were grown in Terrific broth in plastic 96-well plates over a time-course and resultant biofilms were stained with crystal violet. Results and error bars are the mean and standard deviation of triplicate samples from a single experiment.
Resistance to bile
Resistance to stress such as bile salts has been implicated as a potential role of the Rcs phosphorelay system in enterobacterial pathogenicity. This is further backed up by the regulation of tolB/Q by the Rcs phosphorelay in Y. pseudotuberculosis. Mutations of tolB/Q in E. coli resulted in increased sensitivity to bile salts and leakage of periplasmic proteins (Ray et al., 2000). In order to determine if the Rcs phosphorelay plays a role in bile salt resistance in the enteric yersiniae, growth kinetics of the rcsD : : Km mutants of the Y. enterocolitica and Y. pseudotuberculosis strains were compared to wild-type when grown in LB broth containing 1 % Bile Salts no. 3 (Oxoid). The two Y. enterocolitica rcsD : : Km mutants and the Y. pseudotuberculosis IP32953 rcsD : : Km mutant showed significantly increased sensitivity to bile salts compared to wild-type (Fig. 3a, b⇓). However, the Y. pseudotuberculosis YPIII pIB1 rcsD : : Km mutant showed a significant increase in resistance to bile (Fig. 3a⇓).
Growth of Y. pseudotuberculosis (a) and Y. enterocolitica (b) wild-type and rcsD : : Km mutants in LB broth supplemented with 1 % bile salts. Complemented strains are designated by pTRCrcsD or pBADrcsD. Results and error bars are the mean and standard deviation of triplicate samples from a single experiment.
Osmotic stress
In order to determine whether the Rcs phosphorelay also plays a role in survival during osmotic stress, growth kinetics of the rcsD : : Km mutants of the Y. pseudotuberculosis and Y. enterocolitica strains were compared to wild-type when grown in LB broth containing various concentrations of NaCl or sucrose. In high-sucrose media, no difference between mutant and wild-type was seen for either strain (data not shown). Similarly, no difference was observed between mutant and wild-type Y. enterocolitica strains in high-NaCl broth. However, a significant difference was seen between mutant and wild-type for both Y. pseudotuberculosis strains at high NaCl concentrations (Fig. 4⇓). Surprisingly, the two Y. pseudotuberculosis strains showed different growth patterns, with the IP32953 rcsD : : Km mutant showing significantly reduced growth in high-NaCl media compared to the wild-type, whereas the YPIII pIB1 rcsD : : Km mutant showed a significantly increased growth rate in high-NaCl media compared to the wild-type (Fig. 4⇓).
Growth of Y. pseudotuberculosis wild-type and rcsD : : Km mutants in LB broth supplemented with NaCl. (a) IP32953 with 1 M NaCl; (b) YPIII pIB1 with 0.6 M NaCl. Complemented rcsD : : Km mutants are designated by pTRCrcsD. Results and error bars are the mean and standard deviation of triplicate samples from a single experiment.
One of the most highly upregulated CDSs was YPO1624, a gene that is similar to umoD in Proteus mirabilis. Disruption of any of the four chromosomal umo genes in P. mirabilis resulted in reductions in swarming motility and cell elongation (Dufour et al., 1998). The expression profile of umoD was noted to be similar to that of flhDC and increased in parallel with flhDC expression during differentiation into elongated hyperflagellated swarm cells. Indeed, further studies revealed that umoD in P. mirabilis is upregulated by FlhDC (Dufour et al., 1998). In this study, the two flagella operons (including flhDC) were unaffected by Rcs phosphorelay activation, indicating that this high-level expression of the UmoD orthologue in Y. pseudotuberculosis is not due to the FlhDC master regulator. Although there was no evidence from the microarrays that flagella genes were regulated by the Rcs phosphorelay, altered motilty is a common feature of rcs mutants in other species. Therefore the Y. pseudotuberculosis and Y. enterocolitica wild-type and mutant strains were tested for both swimming and swarming motility. Both Y. pseudotuberculosis wild-type strains demonstrated swimming motility at 28 °C, although the IP32953 strain was considerably less motile than the YPIII pIB1 strain. The rcsD : : Km mutation had no effect on swimming in the YPIII pIB1 strain, but in the IP32953 strain the rcsD : : Km mutant was hypermotile compared to the wild-type (Fig. 5a⇓). Neither of the Y. pseudotuberculosis strains was able to swarm at any temperature; however, both of the Y. enterocolitica strains were capable of swarming at 28 °C. Mutation of rcsD in both Y. enterocolitica strains completely abrogated swarming motility, a phenotype which was fully restored by complementation (Fig. 5b, c⇓).
Swimming and swarming motility assays. (a) Swimming motility of Y. pseudotuberculosis strain IP32953: A, wild-type; B, rcsD : : Km mutant; C, complemented rcsD : : Km mutant. (b) Swarming motility of Y. enterocolitica strain WA-C: A, wild-type; B, rcsD : : Km mutant; C, complemented rcsD : : Km mutant. (c) Swarming motility of Y. enterocolitica strain JB580v: A, wild-type; B, rcsD : : Km mutant; C, complemented rcsD : : Km mutant.
DISCUSSION
The rapid increase of bacterial genome sequencing projects has led to the identification of the Rcs phosphorelay in multiple Gram-negative bacterial species, including the three human-pathogenic Yersinia species. Intriguingly, the CDS of the intermediate phosphotransfer protein RcsD is predicted to be a pseudogene and therefore non-functional in all seven sequenced strains of Y. pestis, including the human-avirulent strain 91001 (Song et al., 2004). In all Y. pestis strains rcsD gene function has been abrogated due to an identical frameshift mutation. This indicates that this particular mutation occurred early in the evolution of Y. pestis from its Y. pseudotuberculosis progenitor and may be involved in the initial adaptation of the ancestral Y. pestis to a new lifestyle as a highly virulent blood-borne pathogen transmitted by a flea vector. Furthermore, given that Y. pestis has essentially lost its intestinal pathogenic potential, as evidenced by numerous examples of pseudogenes in enteric pathogenic factors such as ail and ureA, it can be speculated that rcsD may be important for the enteric survival of Y. pseudotuberculosis and Y. enterocolitica. Additionally, the fact that neither rcsC nor rcsB has accumulated mutations resulting in them becoming pseudogenes in any of the sequenced Y. pestis strains raises the possibility that RcsC and RcsB are still required by Y. pestis and may still signal via an unknown phosphotransfer protein or by cross-talk with other sensory systems (Zhou et al., 2003).
To determine the role of the Rcs phosphorelay in the enteric yersiniae, two strains of Y. pseudotuberculosis and two strains of Y. enterocolitica were chosen. For all strains, mutations were generated in either rcsD or rcsC by the partial deletion of the CDS and insertion of a kanamycin-resistance cassette. Microarray analysis was used to determine the effect of these mutations on gene regulation in the YPIII pIB1 strain of Y. pseudotuberculosis. This strain was chosen for all microarray work as its natural phoP mutation allows for the identification of the core regulon without any potential phoP-mediated effects. To ensure correct identification of the Rcs regulon in Y. pseudotuberculosis the effects of overproduction of DjlA and RcsB were compared during the microarray analysis. DjlA was chosen because overproduction has previously been shown to activate the phosphorelay in an rcsC-deficient strain of E. coli when complemented by rcsC from Y. pestis (Y. Huang & D. J. Clarke, unpublished data). It was therefore assumed that DjlA overproduction would activate the phosphorelay in Y. pseudotuberculosis.
The microarray comparison of gene regulation after activation of the Rcs phosphorelay by overproduction of DjlA (mutant vs wild-type) or RcsB (wild-type vs control) resulted in the identification of 136 CDSs that were significantly regulated by both conditions. As expected the majority (∼60 %) of the gene products were predicted to be either located in the cell envelope or involved in cell envelope modifications. Thus activation of the Rcs phosphorelay in Yersinia is predicted to result in gross modifications of the cell envelope. This is similar to the E. coli Rcs regulon, where ∼50 % of CDSs were deemed to affect the cell envelope (Ferrières & Clarke, 2003). Almost 20 % of the Y. pseudotuberculosis Rcs regulon has no predicted function, including many of the most highly regulated CDSs. However, several putative enteropathogenic virulence factors can be identified from the Rcs regulon, including a metalloprotease (YPO3973), a putative intimin (YPO1562) and a putative ShET2-like enterotoxin (YPO1002). Metalloproteases have a variety of putative roles, and thus the significance of this high-level expression can only be hypothesized. Metalloprotease expression has previously been shown to affect surface adherence and colonization by Vibrio vulnificus due to an effect on swarming behaviour (Kim et al., 2007); however, metalloproteases may also be involved in mammalian tissue toxicity (Bowen et al., 2003) and may be important in disease pathology. The putative intimin is likely to act as an adhesin in the gut whilst the putative ShET2-like enterotoxin (YPO1002) is likely to play a role in the pathology of Y. pseudotuberculosis infection by altering the permeability of the epithelial cells of the intestinal wall and causing diarrhoea.
Y. pestis appears to have lost rcsD early in its evolution and thus microarrays were performed to determine if the Rcs phosphorelay was able to signal in the absence of RcsD. There were no significant differences in expression between the rcsC : : Km mutant and the rcsD : : Km mutant, indicating that, in Y. pseudotuberculosis at least, signalling by the Rcs phosphorelay cannot occur without RcsD. It is possible that Y. pestis has acquired an alternative phosphotransfer protein, but no obvious candidates can be determined from genome sequence data. The question remains as to why the other components of the Rcs phosphorelay have not acquired any mutations. Indeed the rcsC from Y. pestis has been shown to complement an rcsC deletion in E. coli and actually appears to be more active than its E. coli orthologue (Y. Huang & D. J. Clarke, unpublished data).
Phenotypic analysis of Rcs mutants of both Y. pseudotuberculosis and Y. enterocolitica was performed based upon the microarray data, which had revealed possible roles of the Rcs phosphorelay in adhesion/invasion, biofilm formation, motility and stress survival. Overall trends between the two species could be determined, but in some instances there was variation, even between strains of the same species. This highlights the difficulties of drawing conclusions from data produced from a single strain as being representative of a given species. For example the two Y. enterocolitica rcsD : : Km mutants and the Y. pseudotuberculosis IP32953 rcsD : : Km mutant all showed greater sensitivity to bile salts than their respective wild-type strains, which was not observed with the Y. pseudotuberculosis YPIII pIB1 rcsD : : Km mutant. These results are in contrast to Rcs mutants in Y. pseudotuberculosis strain 32777, which have a reported increase in tolerance to bile salts (Flamez et al., 2007). Similarly the rcsD : : Km mutants of the two Y. pseudotuberculosis strains showed differences in tolerance to high NaCl concentrations, the IP32953 rcsD : : Km mutant being less tolerant and the YPIII pIB1 rcsD : : Km mutant more tolerant than the wild-type, whilst the Rcs mutants in strain 32777 showed no difference. This indicates that there are other factors involved in the stress responses that differ between the strains. The microarray data indicate that the YPIII pIB1 Rcs phosphorelay is a major regulator of stress-response genes, presumably due to its ability to respond to perturbations of the membrane as an early indication of stress. The Rcs regulon includes genes encoding DNA repair enzymes, mutY, mutH, mutT and deoB, along with YPO0917 and YPO0953, which are paralogues of YggE and YggX that have a function in restoring physiological defects of E. coli after oxidative stress (Pomposiello et al., 2003; Skovran et al., 2004; Kim et al., 2005).
The microarray data also revealed a potential role for the Y. pseudotuberculosis Rcs regulon in motility and biofilm formation. In our study the rcsD : : Km mutants of both Y. enterocolitica strains showed reduced swarming motility, similar to Rcs mutants in Salmonella enterica serovar Typhimurium, but in contrast to mutants in Proteus and E. coli (Belas et al., 1998; Toguchi et al., 2000; Takeda et al., 2001). Neither of the Y. pseudotuberculosis strains was able to swarm, but disruption of the Rcs phosphorelay did affect the swimming motility of Y. pseudotuberculosis IP32953 by making it hypermotile. Strain IP32953 is poorly motile and even this hypermotile mutant took 96 h to reach the edge of the Petri dish. This is in contrast to the YPIII pIB1 wild-type and mutant strains, which only took 48 h to reach the edge of the dish.
The rcsD : : Km mutant of Y. pseudotuberculosis IP32953 showed a reduced ability to form biofilms in plastic plates; however, loss of the Rcs phosphorelay in Y. pseudotuberculosis strain YPIII pIB1 did not affect biofilm formation on a biotic surface in the C. elegans model. Interestingly, the Rcs regulon includes paralogues of UvrY and BolA. In E. coli, the BarA/UvrY two-component regulatory system can activate biofilm formation, and ectopic expression of UvrY stimulated biofilm formation (Suzuki et al., 2002). Overproduction of BolA in E. coli induces biofilm development, while BolA deletion decreases biofilms (Vieira et al., 2004). BolA expression in E. coli has also been shown to trigger the formation of osmotically stable round cells and acts as a regulator of cell wall biosynthetic enzymes with different roles in cell morphology and cell division (Santos et al., 2002). Several other CDSs in the Y. pseudotuberculosis Rcs regulon also encode proteins that appear be linked to cell division and morphology, including the actin-like ATPase YPO0760, MreB/C and MukE. Thus the Rcs phosphorelay may play an important role in inducing changes in cellular morphology as well as the structure of the cell envelope which may also be important in stress survival.
Thus we have shown that the Rcs phosphorelay of the enteropathogenic yersiniae may play an important role in stress survival within the host gut and the external environment and may also be associated with the regulation of certain virulence determinants. Strikingly, the effects of the loss of the Rcs phosphorelay differ between different strains. This indicates that the Rcs phosphorelay is part of a complex network of signalling systems which control the overall phenotype of a particular strain. The role, if any, of the Rcs phosphorelay in Y. pestis remains to be determined.
Acknowledgments
Plasmid pAJD434 was a gift from Andrew Darwin (New York University School of Medicine). The work in D. J. C.'s lab was supported by the BBSRC (reference P16371) and a studentship award (to Y. H. H.) from the Government of Taiwan. The work in B. W. W.'s lab was funded by DSTL, UK. We acknowledge the Bacterial Microarray Group at St George's Medical School (BUGS) for the design and production of the microarrays, especially Lucy Brooks and Adam Whitney for their help with the BUGSBase microarray database.