Influence of PhoP and Intra-Species Variations on Virulence of Yersinia pseudotuberculosis during the Natural Oral Infection Route

The two-component regulatory system PhoP/PhoQ has been shown to (i) control expression of virulence-associated traits, (ii) confer survival and growth within macrophages and (iii) play a role in Yersinia infections. However, the influence of PhoP on virulence varied greatly between different murine models of infection and its role in natural oral infections with frequently used representative isolates of Y. pseudotuberculosis was unknown. To address this issue, we constructed an isogenic set of phoP + and phoP − variants of strain IP32953 and YPIII and analyzed the impact of PhoP using in vitro functionality experiments and a murine oral infection model, whereby we tested for bacterial dissemination and influence on the host immune response. Our results revealed that PhoP has a low impact on virulence, lymphatic and systemic organ colonization, and on immune response modulation by IP32953 and YPIII, indicating that PhoP is not absolutely essential for oral infections but may be involved in fine-tuning the outcome. Our work further revealed certain strain-specific differences in virulence properties, which do not strongly rely on the function of PhoP, but affect tissue colonization, dissemination and/or persistence of the bacteria. Highlighted intra-species variations may provide a potential means to rapidly adjust to environmental changes inside and outside of the host.


Introduction
The lifestyle of most enteric bacterial pathogens, particularly those frequently circulating between external reservoirs and warmblooded hosts, demand efficient strategies to survey and respond to their permanently varying environments.Enteropathogenic bacteria are continuously confronted with rapid changes in nutrient and ion availability during their existence outside and within the host.In order to cope with this situation they evolved sophisticated sensory and signal transduction systems to monitor a large variety of chemical parameters and to convert them into intracellular signalling cascades to adjust their gene expression profile.The most rapid and efficient strategy of adjusting gene transcription involves the highly conserved two-component regulatory systems (TCSs).These TCSs typically comprise a membrane-spanning sensor kinase, which monitors extracellular components (e.g.certain nutrients or ions) and transduces the signal to a DNAbinding cognate response regulator via a histidine to aspartate phosphorelay [1][2][3].
Multiple TCSs have been identified in different enteric pathogens and many of them play an important role in the regulation of bacterial pathogenesis.Based on genome-wide in silico analysis it was predicted that the entero-pathogenic bacterium Yersinia pseudotuberculosis encodes 24 different TCSs [4,5].Several of these systems were found to promote tolerance towards certain environmental stresses (e.g.inorganic/organic acids/low pH, high salinity, low iron/magnesium, antimicrobial peptides such as polymyxin B, and hydrogen peroxide) and were shown to be important for virulence [4,6].
One of the most important TCSs implicated in Yersinia virulence is the PhoP/PhoQ system.This system has been extensively characterized in Yersinia and related pathogens for its ability to sense low Mg 2+ concentrations, low pH and the presence of cationic antimicrobial peptides, i.e. typical attributes of professional phagocytes [6][7][8][9].These conditions are sensed by the sensor kinase PhoQ, which promotes phosphorylation and activation of the response regulator PhoP.The regulatory cascade elicited by PhoP confers the ability to survive and replicate within macrophages [6,[10][11][12].Furthermore, phoP-deficient Y. pseudotuberculosis mutants were found to be attenuated in different murine infection models, suggesting that growth within macrophages is an important feature of pathogenesis [6,[12][13][14].
Y. pseudotuberculosis is a zoonotic food-borne pathogen responsible for the onset of several gut-associated diseases (yersiniosis) ranging from self-limiting enteritis, enterocolitis, watery diarrhea to mesenterial lymphadenitis.In rare cases, the bacteria also induce autoimmune disorders such as reactive arthritis and other systemic infections [15][16][17].Y. pseudotubercu-losis is usually transmitted via the fecal-oral route.After ingestion, the bacteria travel to the ileum and cross the intestinal epithelial layer via M-cells overlaying the lymphoid follicles, known as Peyer's patches.Within the lymphatic tissue, the bacteria are immediately confronted by resident and recruited phagocytes, mainly neutrophils, macrophages and dendritic cells (DCs) [18].At this stage of the infection the bacteria prevent phagocytosis by these innate immune cells, which would otherwise destroy the bacteria once they are internalized.For this purpose, they use a type III secretion system (T3SS) to inject multiple effector proteins (the Yops and LcrV) into the phagocytes in order to disrupt the dynamics of the cell cytoskeleton, trigger apoptosis of macrophages and suppress the production of proinflammatory cytokines [19].As a result, the pathogenic yersiniae are mainly found as aggregates of extracellular bacteria within abscesses or necrotic lesions where they effectively resist phagocytosis by neutrophils [20,21].Although the majority of bacteria appear to multiply in extracellular locations, there is also evidence that they can survive and replicate within macrophages.This process is likely to be important during the very early stages of the infection (,12 h).In a study by Fujimura et al. [22] Y. pseudotuberculosis was detected in rabbits within follicle-associated macrophages underneath M cells 4 h postinfection, and also the closely related Yersinia species Y. pestis and Y. enterocolitica have been identified within macrophages during the very early phases of an infection [23,24].Moreover, it was shown that all three pathogenic yersiniae are able to replicate in naı ¨ve and activated macrophages in vitro [25][26][27][28].Based on this observation it was speculated that macrophages could provide a replicative niche in which the bacteria can (i) proliferate while protected from neutrophils that are recruited to the site of infection, (ii) avoid antigen presentation and therefore delay the development of a specific immune response, and (iii) be transported from the initial site of infection to deeper lymph tissues (Trojan horse model) [29].Consequently, the ability to survive and replicate in macrophages would provide a significant selective advantage during pathogenesis.
Several studies have clearly demonstrated that the PhoP/PhoQ TCS is required for survival and replication of pathogenic yersiniae in macrophages and/or neutrophils [6,12,25,[29][30][31].However, the role of the PhoP/PhoQ system and how the ability to replicate within macrophages affects the infection process in different Yersinia strains and species is less clear.Although, the PhoP-dependent ability to proliferate in phagocytes is conserved among different Yersinia species and strains, the defect in virulence observed with corresponding ortholog phoP mutants was very different.For instance, a strong attenuation (75-fold higher LD 50 ) was reported for a phoP mutant of the Y. pestis strain GB, but only a modest defect in virulence was seen with a phoP 2 mutant of Y. pestis CO92 upon subcutaneous injections in mice (bubonic model) [12,14].In addition, PhoP does not seem to play a role for virulence of Y. pestis CO92 during aerosol infections of mice (pneumonic model) [14], whereas a much greater influence was observed after an aerosol challenge with phoP-deficient mutants of the Y. pseudotuberculosis strains IP32953 and IP2666 [13,14].Hence, the impact of a phoP mutation appears to depend on species and strain-specific differences that remodel regulatory networks and/or virulence-associated traits with different effects on their virulence phenotype.
One of the best characterized isolates of Y. pseudotuberculosis, the serotype O:3 strain YPIII, encodes a mutated and nonfunctional form of the phoP gene associated with an intracellular survival defect in macrophages and yet retains its ability to cause a lethal disease in the murine infection model [6,32].Due to the inactivating mutation of the phoP gene, the use of YPIII as a model pathogen for murine infections has often been criticized.However, an extensive analysis of the influence of phoP on the virulence properties of the YPIII strain is so far missing.In the present study, we took YPIII (phoP 2 ) and the more virulent human isolate IP32953 (phoP + ) as models of Y. pseudotuberculosis infections, constructed a YPIII phoP + and a IP32953 phoP 2 derivative, and determined the relevance of PhoP functionality both in vitro and in vivo during the natural oral infection route with particular interest on bacterial dissemination and modulation of the host immune response.Our results demonstrated that the impact of PhoP on the virulence of IP32953 and YPIII during oral infections is rather low.However, major strain-specific differences in tissue colonization and the induction of the innate and adaptive immune response were observed, indicating that different virulence properties do not mainly rely on the functionality of PhoP.

Ethics Statement
All animal experiments were performed in strict accordance with the German Recommendations of the Society of Laboratory Animal Science (GV-SOLAS) and the European Health Recommendations of the Federation of Laboratory Animal Science Associations (FELASA).The animal protocols were approved by the Niedersa ¨chsisches Landesamt fu ¨r Verbraucherschutz und Lebensmit-telsicherheit: animal licensing committee permission no.33.9.42502-04-12/1010.Animals were handled with appropriate care and all efforts were made to minimize suffering.

Bacterial strains, cell culture, culture media and growth conditions
All bacterial strains used in this study are listed in Table 1.If not stated differently, cultures of E. coli and Y. pseudotuberculosis were grown at 37uC and 25uC in LB (Luria-Bertani) broth, respectively.For bacterial selection, antibiotics were used in the following concentrations: chloramphenicol 30 mg/ml, kanamycin 50 mg/ml, carbenicilin 100 mg/ml.For Mg 2+ -dependent growth analysis of the Y. pseudotuberculosis strains, bacteria were grown in TMH medium at 25uC and 37uC [33,34] without MgCl 2 (referred to as low magnesium medium) or supplemented with 5 mM MgCl 2 .

DNA manipulations and plasmid construction
All DNA manipulations, PCR, restriction digestions, ligations and transformations were performed using standard techniques as described previously [35,36].Plasmids used in the assays are shown in Table 1.For the construction of a kanamycin resistance cassette with 500 bp flanking regions of the phoPQ locus on plasmid pVP3, the kanamycin resistance cassette from chromosomal DNA of YP56 was amplified (primer pair III925/III926) and ligated into the SpeI/SacI sites of pDM4.The plasmid pWH1 was constructed by amplification of phoPQ of Y. pseudotuberculosis strain IP32953 with its 500 bp flanking regions using the primers III926 and III927.The fragment was ligated into the XhoI/SacI sites of pDM4.

Mutant strain constructions
In order to generate the Y. pseudotuberculosis IP32953 phoPQ knock-out mutant YPIP04, pVP3 was integrated into the phoPQ locus of IP32953 via conjugation as previously described [37].Chromosomal integration of the fragment was selected by plating on Yersinia selective agar supplemented with kanamycin.Conjugants with an excision of the plasmid including the functional phoPQ operon of IP32953 were identified after plating on 10% sucrose and selection of fast-growing, carbenicillin sensitive strains.Absence of the phoPQ operon was tested by PCR.To construct Y. pseudotuberculosis IP32953 phoP 2 strain YPIP06 harbouring the non-functional phoP gene of YPIII, the plasmid pWH1 was conjugated into YPIP04.Chromosomal integration of the plasmid was selected by plating on Yersinia selective agar supplemented with chloramphenicol.The correct mutant was identified after plating on 10% sucrose as following: (i) fast-growing bacteria were isolated and tested for loss of the kanamycin cassette; (ii) presence of the phoPQ operon of YPIII was analyzed by PCR with primers III964, III965, III966, and III967; and (iii) presence of the phoPQ locus including the mutation of YPIII in phoP was confirmed by PCR and sequencing.
Intracellular survival of Y. pseudotuberculosis in J774A.1 macrophages Aliquots of 2610 5 J774A.1 macrophages were seeded into 24well plates and activated with Phorbol-12-myristate-13-acetate (PMA) 48 h prior to infection.One hour before infection, the culture medium was replaced as previously described [38].Overnight cultures of bacteria grown in LB medium at 25uC were washed with PBS and added to the J774A.1 macrophages in a multiplicity of infection (MOI) of 1 or 10.The number of living bacteria for infection was determined by plating on LB medium and assessment of colony forming units (CFU).The infected cells were incubated for 1 h at 37uC in an atmosphere consisting of 5% CO 2 followed by a further 1 h incubation in medium supplemented with 8 mg/ml gentamicin.Bacterial uptake was calculated as the percentage of intracellular bacteria at t 0 relative to the number of infecting bacteria.To assess bacterial uptake 2 h postinfection, cells were extensively washed with PBS, lysed with 0.1% Triton X-100 in PBS and lysates were plated on LB medium to determine CFU counts.At 8 h and 24 h postinfection, survival of bacteria was assessed as described previously [38].For each strain, the proportion of bacterial uptake was determined by calculating the number of CFUs at uptake relative to total number of CFUs used for infection.Survival rates were calculated by the number of CFUs at 8 and 24 h relative to the number of CFUs at uptake, respectively.Statistical analysis was performed using the student's t-test.Infection of mice BALB/c female mice aged between 10-12 weeks were purchased from Janvier (Le Genest Saint Isle, St Berthevin Cedex, France) and housed in individually ventilated cages on a 12 h light: dark cycle under specific pathogen-free conditions according to the FELASA recommendations in the animal facility of the Helmholtz Centre for Infection Research, Braunschweig, Germany.Animals were given food and water ad libitum; 14 hours prior to infection food was removed and made available again after the infection.Y. pseudotuberculosis strains used for infections were grown over night in LB medium at 25uC, washed and resuspended in PBS.For survival assays, groups of five weightmatched mice were infected intragastrically using a gavage needle with approximately 5610 8 CFUs of Y. pseudotuberculosis YPIII, IP32953, YPIII phoP + and IP32953 phoP 2 .The infected mice were monitored every day for 14 days to determine the survival rate, body weight and health status.Survival experiments were performed in triplicates.As mice lost more than 20% of their original body weight they were sacrificed by CO 2 asphyxiation.Statistical analysis was performed using the Log-rank (Mantel-Cox) test.
For CFU assessment in infected organs, groups of 5 weightmatched mice were orally infected with approximately 2610 8 CFUs of Y. pseudotuberculosis phoP + or phoP 2 variants of strains YPIII and IP32953 in three independent experiments.At day three postinfection, mice were sacrificed by CO 2 asphyxiation.Peyer's patches (PPs), mesenteric lymph nodes (MLNs), spleen and liver were excised aseptically.Subsequently, all organs were weighed and homogenized in sterile PBS at 15,000 rpm for 30 seconds using a Polytron PT 2100 homogenizer (Kinematica, Switzerland).Bacterial organ burden was determined by plating serial dilutions of the homogenates on Yersinia selective agar (Oxoid).The CFUs were counted 48 h after plating and are given as the number of CFUs per gram of organ tissue.The data was statistically analyzed using One-way ANOVA followed by Tukey's multiple comparisons test.

Immune response analysis
Groups of 5 weight-matched mice were challenged intragastrically with 2610 8 bacteria of Y. pseudotuberculosis YPIII, IP32953, YPIII phoP + or IP32953 phoP 2 , in four independent experiments.Groups of 5 uninfected mice in three independent experiments were included as controls.At day 3 postinfection, the infected and the uninfected mice were euthanized via CO 2 asphyxiation.Subsequently, PPs, MLNs and spleen were isolated and single cell suspensions were generated in PBS supplemented with 0.2% BSA by meshing the organs through a cell strainer (100 mm, BD).To eliminate the erythrocytes, splenocytes were incubated in ACKbuffer (150 mM NH 4 Cl, 10 mM KHCO 3 , 1 mM EDTA, pH 7.4) for 3 min.Cell suspensions of all analyzed tissues were counted using a Beckman Coulter Counter.Subsequently, 1610 6  were inoculated into TMH medium with either low or high magnesium concentrations (5 mM) and grown at either 25uC or 37uC.The graph represents one independent experiment out of three replicates.Statistical analysis was performed using Two-way ANOVA with Tukey's multiple comparisons test at 24 h of growth (*, p,0.05; ***p,0.001).(B) Activated J774A.1 macrophages were infected with Y. pseudotuberculosis strains YPIII (phoP 2 ), YP149 (YPIII phoP + ), IP32953 (phoP + ) and YPIP06 (IP32953 phoP 2 ) at an MOI 1 for 1 h.The infection was stopped with gentamicin and bacterial uptake was determined 2 h postinfection.Bacterial survival inside the macrophages was determined over time.The graph represents a summary of three independent experiments with four replicates respectively.Statistical analysis was performed using Two-way ANOVA with Tukey's multiple comparisons test at 24 h post assessment of bacterial uptake (***, p,0.001).Error bars represent SEM.doi:10.1371/journal.pone.0103541.g001

Cytokine/chemokine multiplex assay
Heart blood was taken from 8 weight-matched mice either infected with Y. pseudotuberculosis YPIII, IP32953, YPIII phoP + or IP32953 phoP 2 at 3 days post infection and from 5 uninfected mice.Samples were obtained from three independent experiments and the serum was separated from the cellular part by two-step centrifugation at 5,000 rpm for 8 min after clotting of the blood.The serum was analyzed using the MILLIPLEX Mouse cytokine/ chemokine panel (Merck Millipore).Cytokine levels were statistically analyzed by using One-way ANOVA with Tukey's post hoc test.

Luciferase assay
Y. pseudotuberculosis YPIII and YPIII phoP + (YP149) carrying either pJNS02 [32] or the empty vector pFU54 [39] were inoculated 1:50 from overnight cultures and grown in LB medium with antibiotics at 25uC.Every 2 h, samples were taken and analyzed for luciferase activity and optical density (OD) at 600 nm in technical duplicates with Varioskan flash (Thermo scientific).Relative luciferase units were normalized to the corresponding OD 600 .

Results
Comparative analysis of Y. pseudotuberculosis YPIII and IP32953 phoP + and phoP 2 variants Previous work demonstrated that the PhoP/PhoQ TCS affects replication of Y. pseudotuberculosis in macrophages, but its role in virulence, in particular upon the natural oral infection route is not clear.Over the last decades, mainly Y. pseudotuberculosis isolates YPIII (serotype O:3) and IP32953 (serotype O:1b) were used to elucidate Y. pseudotuberculosis pathogenicity.However, it was recently found that YPIII carries a mutation in the PhoP-encoding gene YPK_1715, leading to a non-functional version of PhoP [6].
Although it has been reported that PhoP/PhoQ is required for virulence in Y. pseudotuberculosis 32777 by the intragastric route [6], strain YPIII is still able to cause lethal disease in the oral murine mouse infection model [40].
In order to test the role of PhoP in Y. pseudotuberculosis and compare its impact between the representative isolates YPIII and IP32953, we first constructed an isogenic phoP + and phoP 2 derivative of both strains: YP149 (YPIII phoP + IP32953 ) and YPIP06 (IP32953 phoP 2 YPIII ) (for further details see Material and Methods), tested their growth in low and high Mg 2+ environments, and determined their ability to replicate within macrophages (Fig. 1).In accordance with previous studies [6], only YPIII phoP + and IP32953 were able to grow under low Mg 2+ concentrations, while YPIII and IP32953 phoP 2 were both growth defective at 25uC and 37uC, and supplementation with 5 mM MgCl 2 enabled growth of all tested strains (Fig. 1A).Bacterial uptake inside macrophages was similar among the four tested strains, and corresponded to approximately 40% at MOI 1 and 20% at MOI 10 of the inoculated bacteria, respectively (data not shown).As expected, YPIII was clearly defective for replication in macrophages (Fig. 1B), whereas its phoP + derivative displayed significantly increased survival, replicating during the first day of infection despite the presence of the virulence plasmid.A similar reduction in survival and replication in macrophages due to the loss of phoP was observed for the Y. pseudotuberculosis strain IP32953 (Fig. 1B).However, the bacterial counts of IP32953 were considerably higher compared to YPIII phoP + , indicating that IP32953 is much less sensitive to killing by the macrophage phagosome.Taken together, these results are in general agreement with previously published studies [6], and confirm the correct replacement of the phoP alleles.

PhoP influence on virulence of Y. pseudotuberculosis YPIII and IP32953
It was previously shown that the loss of a functional phoPQ locus results in the increased survival of mice orally infected with Y. pseudotuberculosis serogroup O:1 strain 32777 (100-fold less virulent estimated by LD 50 ) [6].To better understand the role of PhoP and to determine the impact of the non-functional phoP 2  YPIII allele on virulence of Y. pseudotuberculosis, mice were orally infected with approximately 5610 8 CFU of phoP + and phoP 2 variants of YPIII and IP32953.Presence of a functional PhoP resulted only in a mild increase in lethality for the YPIII phoP + strain in comparison to YPIII, whereas statistical comparison of the survival curves did not highlight any PhoP-dependent difference between the IP32953 phoP + and phoP 2 variants (Fig. 2).
Upon oral infection, Y. pseudotuberculosis disseminates from the PPs and/or small intestinal epithelium to the MLNs and systemic organs like the liver and spleen.To explain the observed virulence properties in the oral infection model, we investigated the influence of PhoP on colonization and dissemination of YPIII and IP32953 in more detail.For each strain, mice were orally challenged with 2610 8 CFUs.At day 3 postinfection, mice were sacrificed and the bacterial burden in the PPs, MLNs, spleen and liver was assessed (Fig. 3).Analysis of the organ burden revealed significant strain specific differences in the colonization and dissemination ability.IP32953 and IP32953 phoP 2 populated all assessed organs with higher efficiency than YPIII and YPIII phoP + (Fig. 3).Furthermore, the YPIII and YPIII phoP + strains disseminated less efficiently to the MLNs, liver and spleen (Fig. 3).Despite the small differences observed for the survival of the mice (Fig. 2), organ colonization by YPIII was not affected by the presence of a functional copy of phoP.Indeed, YPIII phoP + and YPIII phoP 2 bacteria were found at similar levels in all analyzed organs.In contrast, the lymphatic tissues (PPs and MLNs) were populated slightly more efficiently by IP32953 in the absence of a functional PhoP (Fig. 3A, B), while less bacteria were isolated from the liver and spleen (Fig. 3C, D).In summary, this data indicates that PhoP appears to be dispensable for host tissue colonization by YPIII, and only minor changes were observed for IP32953.PhoP is not important for induction of the immune response Many studies have shown that cells of the innate immunity like macrophages and neutrophils, and adaptive immunity like CD4 + T h 1 and CD8 + T cells are crucial in the control of Yersinia infections [41][42][43][44].In 2009, Kumar et al., showed that a virulenceplasmid cured Y. pseudotuberculosis DphoPQ mutant is diminished in the T h 1-type immune response in a murine systemic infection model following intravenous inoculation [45].However, a deeper dissection of the PhoP-driven immune response in mice infected by the natural intragastric route is so far missing.To this purpose, the influence of the response regulator PhoP and strain-driven differences on innate and adaptive immune responses to Y. pseudotuberculosis infection were assessed.Three days after oral infection, mice were sacrificed to analyze the cell type composition of the PPs, MLNs and spleen.To this purpose, living cell levels of natural killer cells (NK cells), neutrophils, monocytes, dendritic cells (DCs), macrophages, B cells and T cells were evaluated (Fig. 4-6; Fig. S3-S5).
In the PPs and MLNs of Y. pseudotuberculosis infected mice, a significant infiltration/increase of innate and adaptive immune cells was observed (Fig. 4, 5).In particular, neutrophils, monocytes and NK cells increased substantially and to a similar extent within the lymphoid tissues in YPIII and IP32953 infected mice (Fig. 4C, D, E, Fig. 5C, D, E).Only pDC levels did not raise in the PPs of IP32953 infected mice (Fig. 4B), irrespective of the functionality of PhoP.Analysis of the mesenteric lymph node composition highlighted a stronger increase in the relative abundance of macrophages and neutrophils in mice infected with the strains IP32953 and IP32953 phoP 2 in comparison to YPIII and YPIII phoP + infected ones (Fig. S4).Overall, the presence of PhoP had no, or only a very small effect on the infiltration of immune cells into the gut-associated lymphatic tissues.
Similarly in the spleen, no significant difference in the levels of cells of the adaptive and innate immune systems were observed during infection with Y. pseudotuberculosis strains IP32953 and IP32953 phoP 2 (Fig. 6).However, YPIII and YPIII phoP + induced a significant decrease (likely due to a cytotoxic effect) in most of the immune cells, including NK cells, pDCs, macrophages and monocytes, splenic B cells and T cells.Thereby, YPIII seems to influence exclusively the number of pDCs and not cDCs (Fig. 6B).Furthermore, reduction/killing of immune cells was enhanced in the phoP + derivative of YPIII (Fig. 6).Notably, spleens of mice infected with the IP32953 derived strains displayed a higher percentage of neutrophils and monocytes in comparison to the YPIII infected mice (Fig. S5).Taken together, these results show profound differences in the modulation of the immune response elicited by YPIII and IP32953, whereby only YPIII leads to a PhoP-dependent reduction of immune cells in the spleen during intragastric infection.

PhoP influence on the production of proinflammatory cytokines
The inflammatory process responsible for immune cell recruitment and function is largely mediated by cytokines that are produced upon infection [46].Therefore, we further analyzed the levels of 32 different cytokines in sera of mice infected with YPIII, YPIII phoP + , IP32953 and IP32953 phoP 2 as well as uninfected mice via a cytokine multiplex assay.
Infection with Yersinia induced only an increase of selected cytokines at the systemic level with major differences between the different serotypes (Fig. 7).Notably, a significantly higher production of these cytokines was observed in IP32953 infected mice compared to mice challenged with YPIII regardless of the functionality of phoP.Among the most induced cytokines was G-CSF, which is responsible for the mobilization and activation of hematopoietic cells from the bone marrow (Fig. 7A), and the neutrophil-, monocyte-and DC-chemoattractant cytokines MCP-1, IP-10 and KC (Fig. 7C-E).Concentrations of these cytokines were 2-4 fold higher in the blood of IP32953-infected mice compared to YPIII-infected mice.Most likely, this inter-serotype difference in cytokine production is reflected in different neutrophil population levels within the infected organs (Fig. 4-6).Similarly, serum levels of the pleiotropic cytokine IFNc and the potent systemic inflammation inducer TNFa were only slightly increased in mice infected with YPIII and YPIII phoP + whereas significantly higher amounts of IFNc were produced in response to IP32953 and IP32953 phoP 2 .Notably, lack of PhoP in the IP32953 strain also resulted in a considerably lower TNFa, IFNc and IP-10 response in comparison to the wild-type strain.Finally, the serum analysis also highlighted a strain-dependent induction of IL-13.This cytokine is responsible for the establishment of a T h 2 immune response against extracellular pathogens.Infection with the YPIII-derived strains did not induce a systemic increase of IL-13 in comparison to uninfected mice, while a 2-fold rise in its levels was observed in mice infected with IP32953 and IP32953 phoP 2 .In contrast, levels of the potent macrophage activator cytokine GM-CSF were only slightly affected by both Y. pseudotuberculosis strains at the time-point after infection (Fig. 7B).Taken together, these results reveal major differences in the systemic cytokine response between mice infected with YPIII and IP32953, mostly in a PhoP-independent fashion.

PhoP induces expression of CNF Y in Y. pseudotuberculosis YPIII
Immune cell composition revealed a substantial increase in the capability of YPIII phoP + strain to influence the distribution of cells of the innate immune system in the spleen in comparison to the wild-type strain YPIII (Fig. 6).Given the similar titres of bacteria in this organ (Fig. 3), the mechanism seems to be independent of the quantity of bacterial cells.We recently identified the cytotoxic necrotizing factor of Yersinia (CNF Y ) as a major player in determining the depletion of immune cells in the spleen via the improvement of Yop delivery in the cytoplasm of host cells [32].Therefore, we investigated whether the presence of PhoP was able to affect cnfY expression in vitro.YPIII and YPIII phoP + carrying a plasmid with the cnfY promoter fused to the luciferase operon luxCDABE were grown from exponential to stationary phase in LB medium and the growth and bioluminescence measured.Expression of the reporter fusion was significantly Figure 5.Immune response analysis in the mesenteric lymph nodes of BALB/c mice infected with phoP + and phoP 2 derivatives of Y. pseudotuberculosis YPIII or IP32953.Mice were challenged with 2610 8 CFU of Y. pseudotuberculosis strains YPIII (phoP 2 ) (n = 20), YP149 (YPIII phoP + ) (n = 20), IP32953 (phoP + ) (n = 20) and YPIP06 (IP32953 phoP 2 ) (n = 20); a control group of uninfected mice (n = 15) was included.At day three postinfection living cell numbers of B cells and T cells (A), plasmacytoid and conventional DCs (B), NK cells (C), neutrophils (D), monocytes (E) and macrophages (F) in the MLNs were analysed.The data represents the total cell number per organ in a logarithmic scale.Data from four independent experiments were pooled.Living cell numbers were statistically analyzed using One-way ANOVA with Tukey's post hoc test (*, p,0.05; **, p,0.01; ***, p,0.001).doi:10.1371/journal.pone.0103541.g005higher in the YPIII phoP + strain compared to YPIII after exponential phase, indicating that PhoP activates cnfY expression upon entry into the stationary phase (Fig. 8).These results suggest a role for PhoP on CNF Y expression during the course of infection, and may account for the increased cytotoxic effect of the YPIII phoP + strain in comparison to the parent strain YPIII.We further investigated whether presence of CNF Y had an influence on the systemic cytokine response (Fig. S6).However, only small changes between YPIII and an isogenic cnfY mutant were observed, indicating that major differences in IFN-c, TNF-a, MCP-1, KC, IL-13 and IP-10 secretion between mice infected with YPIII and IP32953 are not caused by CNF Y .

Discussion
Since the discovery of a mutation in the phoP gene of the Y. pseudotuberculosis isolate YPIII, leading to a truncated PhoP protein, and given the central role of the PhoP/PhoQ TCS in the ability to survive and replicate in macrophages in vitro in all human pathogenic Yersinia species [6,12,25,[29][30][31], several concerns about the employment of YPIII as a representative example for Y. pseudotuberculosis in murine infection models were raised.Yet, the PhoP-deficient YPIII strain is still highly virulent and maintains its lethality during oral mouse infections [32], in contrast to a phoP mutant derivative of Y. pseudotuberculosis isolate 32777 [6].Overall, a very different impact of PhoP on Yersinia virulence was reported [12][13][14], suggesting that the importance of PhoP for the control of crucial pathogenicity factors differs significantly among strains of the three different human pathogenic Yersinia species.However, a direct comparison of the outcome of the infection is difficult as distinct experimental settings (e.g.different infection routes, doses, and mouse strains) were employed.
To overcome this problem, we studied the contribution of PhoP to virulence of the best-characterized Y. pseudotuberculosis strains YPIII and IP32953 during the natural oral infection route.Both strains were and are still frequently used to analyze Y. pseudotuberculosis pathogenicity.The serotype 1b strain IP32953, a close evolutionary relative to Y. pestis, is a highly virulent clinical isolate from a human patient, which efficiently colonizes lymphatic tissues and organs after oral uptake and causes a high lethality of mice [47].YPIII is a very well characterized mouse virulent isolate that induces severe acute inflammatory responses and the formation of necrotic areas in infected tissues [48].Different to most other strains, YPIII expresses the cytotoxic necrotizing factor CNF Y which plays a dominant role in the progression of disease and a deletion of the corresponding cnfY gene was shown to abrogate virulence [32].Interestingly, the cnfY gene is truncated and thus is non-functional in most Y. pestis and Y. pseudotuberculosis strains, including IP32953 [49].The present study was designed to dissect the relevance of PhoP upon oral infection with YPIII and IP32953 in terms of bacterial colonization/dissemination, and with respect to the elicited immune response.
In the present study, we show that loss of the response regulator has only a small effect on the lethality of YPIII, and no effect on IP32953.Dispensability of PhoP is further reflected in similar colonization levels (Fig. 3).No (YPIII) or only very small variations (IP32953) of the bacterial loads were observed in the different infected tissues.Such differences between the IP32953-derived strains may indicate that absence of PhoP results in a slight reduction of the dissemination of IP32953 from the lymphoid tissues (PPs and MLNs) to the systemic organs (liver and spleen).PhoP expression in three different Y. pseudotuberculosis isolates, including YPIII, was shown to repress biofilm formation in a Caenorhabditis elegans model [50].However, based on our results this property does not seem to be important for tissue colonization and dissemination of the bacteria in mammals.This hypothesis is supported by the fact that other Y. pseudotuberculosis mutants defective in biofilm formation (e.g.DhmsT) were found to be fully virulent in a murine oral infection model [50].
The analysis of the host immune response during a YPIII and IP32953 infection further demonstrated that PhoP has no or very little effect on the composition of immune cells in the infected tissues.A dominant influence was only found in the spleen infected with YPIII, but not with IP32953.YPIII infection alone caused a strong decrease of several immune cell populations, in particular of the professional phagocytes.In a previous study we showed that the CNF Y toxin secreted by YPIII is responsible for the reduction of the immune cells and causes an enormous and unique splenotoxic effect in comparison to IP32953 (DcnfY) [32].It is unclear whether this effect is due to a long-range delivery of the toxin to the spleen from the gut compartment or is locally induced.However, it seems that this effect is PhoP-dependent since the reduction of the immune cell populations is more pronounced in mice infected with the YPIII phoP + strain.This hypothesis was confirmed by expression analyses demonstrating a stronger induction of cnfY transcription in the presence of a functional phoP copy.
Interestingly, pDC distribution differed in a strain-dependent, but PhoP-independent fashion.In mice infected with the strains IP32953 and IP32953 phoP 2 the number of pDC remained almost unaltered in all analyzed organs.However, in mice infected with YPIII and YPIII phoP + we observed a decrease in the number of pDC in the spleen and a corresponding increase of this cell type in the Peyer's patches.This suggests that in response to YPIII infection pDC may migrate from the spleen to the Peyer's patches.We additionally observed that MLNs of mice infected with IP32953 strains were predominantly populated by neutrophils and macrophages and the spleens by neutrophils and monocytes, while a similar pattern was not observed in YPIII infected mice.However, this inter-strain difference might rather be due to the higher colonization of spleen and MLN achieved by the IP32953 strain than to a specific immune response (Fig. S4, S5).
Analysis of the systemic production of cytokines, however, highlighted PhoP-dependent differences in the IP32953-triggered immune response.Significantly lower levels of several cytokines, in particular IFNc and TNFa, were detected in mice infected with the phoP-deficient mutant derivative.One could hypothesize that this difference is due to a different PhoP-dependent behaviour of the bacteria inside phagocytes.However, considering that intracellular Y. pseudotuberculosis are rare [51], and that the numbers of phagocytes remains unaffected in the IP32953 infected tissues (Fig. 4-6), it is more likely that the observed increase may be due to the higher amounts of wild-type bacteria in the liver and spleen (Fig. 3).
A second striking finding from our studies is that YPIII colonizes the infected tissues and/or disseminates much less efficiently to the PPs, the MLNs, liver and spleen than IP32953.Similarly, IP32953 was found to disseminate more rapidly in a murine model of lung infections compared to IP2666, a Y. pseudotuberculosis isolate that also expresses the CNF Y toxin and robustly colonizes lungs, but only sporadically disseminates to the spleen and liver [52].The different dissemination capacity might also be reflected in the profound PhoP-independent differences observed in the immune response to YPIII and IP32953.Especially the population of professional phagocytes in the infected tissues and systemic production of cytokines varied significantly during oral infections with the two virulent isolates.The basis for these differences is still unclear.However, expression and function of important adhesins and adhesin regulators were found to be very different between the two Y. pseudotuberculosis strains [53,54] (A.M. Nuss, unpublished results).In particular, expression of invasin seems to be much more induced in IP32953 compared to YPIII [54].We further identified variations between the YadA molecules of both isolates.Interestingly, several amino acids are changed or deleted in YadA IP32953 compared to YadA YPIII clustered in a region of the YadA head domain previously shown to be crucial for autoagglutination, fibronectin binding and host cell invasion [55].These and/or other differences in the function and expression profile of adhesin molecules may account for the variances in dissemination.Furthermore, different adhesin properties could change: (i) resistance of the host complement; (ii) efficiency of Yop translocation into professional phagocytes and (iii) specificity of immune cell types translocated with the Yops.These assumptions are supported by a recent study demonstrating that the adhesins as well as host serum factors drive Yop translocation into specific immune cell types [54].Alternatively, other pathogenicity factors, i.e. the CNF Y of YPIII shown to enhance Yop translocation into the phagocytes and NK cells [32] may contribute to the differences in the colonization efficiency and the immune response observed between YPIII and IP32953.Considering the role of the Yop proteins in suppression of the host immune response, including inhibition of phagocytic uptake, induction of apoptosis of professional phagocytes, alterations of cytokine production and hindrance of immune cell recruitment [19,56], a contribution of CNF Y to the strain-specific differences seems very likely.Therefore, we favor the idea that the strain-dependent phenotype in the murine infection model observed in this study relies upon different pathogenic mechanisms between the strains rather than on the presence of a functional PhoP.
The overall, more virulent infection by IP32953 reflected by faster killing of the host correlates with higher bacterial loads in the infected tissues and the higher proinflammatory cytokine responses in infected animals.In contrast, the slow dissemination of strain YPIII associated with an overall low induction of the host immune response may be advantageous for long-term colonization of the intestinal tract.Due to the powerful and long-range effect of the CNF Y toxin compared to the local and equally strong effect of Yops, it could also be hypothesized that loss of PhoP function in YPIII might be even more advantageous for long-term colonization of the host, as its absence extends the lifetime of the host without a significant reduction of the bacterial load in the infected tissues.Furthermore, reduced expression of CNF Y in the absence of PhoP is still sufficient to stimulate Yop translocation into professional phagocytes to block uptake, but lower toxin concentrations seems to reduce rapid killing of the immune cells, which would result in a less severe form of tissue destruction and slower killing of the host compared the phoP + derivative of YPIII.
In summary, findings in this and other recent studies highlight important strain-specific differences in the virulence properties of Y. pseudotuberculosis despite the fact that the genetic variability within this species is generally limited [52,54,57,58].The three mouse-virulent strains YPIII, IP32953 and IP2666, which are frequently used for the analysis of Y. pseudotuberculosis pathogenicity vary in their individual virulence factors, YPIII (e.g.phoP 2 ), IP2666 (e.g.psaE 2 ), IP32953 (e.g.cnfY 2 ), and in the expression level of important virulence regulators and virulence-associated traits.This may explain why controversial virulence phenotypes were often reported.Combined, these observations raise questions regarding the validity and significance of infection studies using a single strain as a species representative.Nonetheless, the analysis of the pathogenicity of a particular representative virulent Y. pseudotuberculosis isolate has and will continue to yield further insights into the structure, function and regulation of virulence factors and the induced immune responses.However, we pose that a more comparative analysis of multiple representatives of a particular Yersinia species will be a prerequisite to separate general/core virulence strategies from individual or speciesspecific adaptation strategies to certain mammalian hosts or host niches, an approach which is particularly important to develop successful diagnostic tools and anti-infectives.

Supporting Information
Figure S1 Gating strategy for T cells, B cells and NK cells.The collected flow cytometry data was compensated and analyzed with FlowJo.At first, cells were gated for living cells by plotting Live Dead versus FSC-A followed by a double doubletexclusion (FSC-A against FSC-H; SSC-H against SSC-W).The remaining singlets were gated for leukocytes by plotting FSC-A against SSC-A.At next, the auto fluorescent cells (AF + ) were excluded (auto-fluorescence (AF) versus Live Dead).The remaining autofluorescence negative (AF 2 ) leukocytes were gated for CD19 + cells (B cells).Subsequently, the CD19 2 cells were gated for CD3 + cells (T cells) and NKp46 + (NK cells).Cell proportions of living T cells, B cells and NK cells were calculated by the frequency of living cells.(PDF) Figure S2 Gating strategy for cDCs, pDCs, neutrophils, macrophages and monocytes.The collected flow cytometry data was compensated and analyzed with FlowJo.At first, cells were gated for living cells by plotting Live Dead versus FSC-A followed by a double doublet-exclusion (FSC-A against FSC-H; SSC-H against SSC-W).The remaining singlets were gated for leukocytes by plotting FCS-A against SSC-A.At next, the leukocytes were gated for CD3/CD19/panNK lineage negative cells (Lin 2 ) by plotting AF against CD3/CD19/panNK.The Lin 2 cells were gated for F4/80 hi cells (macrophages) by plotting F4/80 against Ly6G.Subsequently, the F4/80 hi negative cells were gated against auto-fluorescence and SSC-A to exclude AF + cells.Afterwards, the AF 2 cells were gated for CD11c hi cells (cDCs) by plotting Ly6G against CD11c.The CD11c hi negative cells were gated against CD45R and Ly6G for Ly6G + cells (neutrophils).There after, the Ly6G 2 cells were plotted against CD11b and Ly6C for CD11b + cells (monocytes).Finally, The CD11b 2 cells were gated for CD45R + Ly6C + cells (pDCs).Cell proportions of living macrophages, cDCs, neutrophils, monocytes and pDCs were calculated by the frequency of living cells.(PDF)