|Home | About | Journals | Submit | Contact Us | Français|
The twin arginine translocation (Tat) system targets folded proteins across the inner membrane and is crucial for virulence in many important human-pathogenic bacteria. Tat has been shown to be required for the virulence of Yersinia pseudotuberculosis, and we recently showed that the system is critical for different virulence-related stress responses as well as for iron uptake. In this study, we wanted to address the role of the Tat substrates in in vivo virulence. Therefore, 22 genes encoding potential Tat substrates were mutated, and each mutant was evaluated in a competitive oral infection of mice. Interestingly, a ΔsufI mutant was essentially as attenuated for virulence as the Tat-deficient strain. We also verified that SufI was Tat dependent for membrane/periplasmic localization in Y. pseudotuberculosis. In vivo bioluminescent imaging of orally infected mice revealed that both the ΔsufI and ΔtatC mutants were able to colonize the cecum and Peyer's patches (PPs) and could spread to the mesenteric lymph nodes (MLNs). Importantly, at this point, neither the ΔtatC mutant nor the ΔsufI mutant was able to spread systemically, and they were gradually cleared. Immunostaining of MLNs revealed that both the ΔtatC and ΔsufI mutants were unable to spread from the initial infection foci and appeared to be contained by neutrophils, while wild-type bacteria readily spread to establish multiple foci from day 3 postinfection. Our results show that SufI alone is required for the establishment of systemic infection and is the major cause of the attenuation of the ΔtatC mutant.
The genus Yersinia includes three species that are pathogenic to humans: Yersinia pestis, the causative agent of plague, and the two enteropathogens Y. enterocolitica and Y. pseudotuberculosis, which normally cause a self-limiting disease with symptoms ranging from mild diarrhea, enterocolitis, septicemia, and mesenteric lymphadenitis to reactive arthritis in humans after ingestion of contaminated food or water (1). Once it is ingested, Y. pseudotuberculosis traverses the epithelial barrier through M cells and infects the associated lymphoid tissues, such as Peyer's patches (PPs) and cecal patches, and later spreads to the mesenteric lymph nodes (MLNs). Although the infection is usually self-limited in humans, the infection caused by the two enteropathogenic Yersinia species in mice readily progresses to become systemic and disseminates to the spleen and liver (2). The main virulence arsenal of Yersinia is the type III secretion system (T3SS), which is encoded by an ~70-kb virulence plasmid. The T3SS enables the translocation of virulence effector proteins directly into the cytosol of the target host cell, which results in the disruption of host signaling and early immune responses, such as inhibition of phagocytosis and the downregulation of proinflammatory responses (3).
Designated protein secretion systems have evolved to be important mechanisms for different processes ranging from physiological adaptation for survival in harsh environments to pathogenicity. The twin arginine translocation (Tat) pathway is found in bacteria, archaea, and plants, where it enables the transportation of fully folded proteins across the cytoplasmic membranes (4). Tat substrates contain in their N-terminal region a conserved, distinctive signal peptide [(S/T)-R-R-X-F-L-K] with a nearly invariant twin arginine motif (5). The minimal translocation complex found in Gram-positive bacteria and archaea consists of two proteins of the TatA and TatC family of integral membrane proteins. For some Gram-negative bacteria, including Escherichia coli, another TatA family protein denoted TatB is also required (6, 7). The translocation process of the Tat pathway is energized by the electrochemical potential (Δp) of the membrane (8).
Some of the Tat substrates are cofactor-containing proteins that function in redox reactions; others are involved in metal ion acquisition and/or resistance, cell envelope maintenance, and metabolism (9). Furthermore, the Tat pathway has also been shown to be essential for the virulence of many different human-, animal-, and plant-pathogenic bacteria, such as Pseudomonas aeruginosa (10), P. syringae pv. tomato DC3000 (11), Burkholderia thailandensis (12), Salmonella enterica serovar Enteritidis (13), Campylobacter jejuni (14), and Vibrio cholerae (15). In these pathogens, mutants lacking a functional Tat exhibit different virulence-related phenotypes, such as decreased motility, toxin production, and biofilm formation; sensitivity to detergents and bile; and decreased T3SS secretion, cell growth, and division (16, 17). In many cases, these Tat-related phenotypes are the result of indirect effects on regulation/expression. Only a subset of the pathogens encodes Tat-dependent virulence factors, such as phospholipase toxins and PvdN in P. aeruginosa (18, 19) and phospholipase C in Legionella pneumophila, that directly contribute to the virulence attenuation of Tat mutants (20). A recent study of S. enterica serovar Typhimurium disclosed that Tat mutants are attenuated due to envelope defects caused by the absence of three periplasmic proteins, AmiA, AmiC, and SufI. Still, additional Tat substrates are likely to be involved in virulence, as even the triple mutant was not as attenuated as the tat mutant (21).
In our previous study, we showed that a ΔtatC mutant of Y. pseudotuberculosis is highly impaired in colonization of the spleen (22). We recently reported that the loss of Tat function leads to drastic changes in the transcriptome, including genes involved in metabolism, stress responses, and virulence, especially at 26°C during stationary-phase growth. We could also correlate the changes in transcription to phenotypic changes, including defects in iron acquisition, downregulation of the expression of the adhesin YadA, sensitivity to copper, and sensitivity to the detergent sodium dodecyl sulfate (SDS) (23). However, none of the strains with mutations in Tat substrates predicted to be functionally related to these phenotypes showed the same level of phenotypic defect as the Tat-deficient strain (23). Therefore, in this study, we sought to investigate the specific role of the putative Tat substrates encoded by the Y. pseudotuberculosis genome in in vivo virulence. Among the 22 potential Tat substrates that were mutated, only the ΔsufI mutant was significantly attenuated and exhibited essentially the same level of attenuation as the ΔtatC mutant. By using bioluminescent in vivo imaging (BLI), we could show that both mutants replicated in vivo and were able to infect MLNs, but unlike the wild-type (wt) strain, both the ΔtatC and ΔsufI mutants appeared to be contained by neutrophils in these lymphoid organs and were eventually cleared. These results show that the Tat substrate SufI is critical for the in vivo virulence of Y. pseudotuberculosis and is mostly responsible for the attenuation of the ΔtatC mutant.
In our previous studies, we showed that a functional Tat system is required for establishment of systemic infection and colonization of the spleen (22) in mice and that the loss of Tat function leads to a growth defect in the presence of low-iron medium, increasing copper concentrations, acidic medium, and SDS (23). However, the Tat substrates functionally related to these phenotypes only partly contributed to the respective phenotypes. Therefore, in this study, we wanted to expand the previous study to investigate the role of additional Tat substrates in the in vivo virulence of Y. pseudotuberculosis. We initially performed a new in silico analysis using the Tat substrate prediction tool PRED-TAT, which has been suggested to be more sensitive and specific than the TatP and TATFIND programs that we used in our previous study (24). The PRED-TAT analysis resulted in a total of 27 potential Tat substrates, with an additional 10 new potential substrates being found and with 6 substrates from our previous analysis with the TatP program being eliminated (22) (Table 1). All of the six eliminated substrates appeared to have a transmembrane segment instead of a Tat signal peptide, so we decided to base this study on the results from the PRED-TAT analysis. The genes encoding each putative Tat substrate were mutated to facilitate studies on the role of each substrate in in vivo virulence. Among those 27 potential substrates on our list, we decided to remove 2 proteins, which were YPTB0195 and NapF. The YPTB0195 gene encodes a TetR family regulatory family protein and has a helix-turn-helix TetR-type DNA-binding domain, indicating a probable cytoplasmic location. NapF is an iron-sulfur protein that is a subunit of a nitrate reduction system and was found to localize and function in the cytoplasm and confirmed not to be Tat dependent (25). Among those 25 potential Tat substrates, we were able to mutate 22. In spite of many efforts using different methods, we were unable to generate insertion mutants in the genes mdfA, fliY, and YPTB3261 (Table 1). It is possible that these genes or downstream genes are somehow required for viability or that for some reason recombination frequencies were very low for these genes.
In order to evaluate the role of each of the Tat substrates in in vivo virulence, a competitive oral infection between the wild-type and the 22 Tat substrate mutants was performed. For this, equal numbers of CFU of wild-type bacteria and each of the 22 strains were mixed and groups of four BALB/c mice were infected orally with each mutant strain–wild-type strain mixture. Spleens were collected at day 4 or 5 postinfection (p.i.), and the numbers of wild-type and mutant bacteria in the input and output mixtures in homogenized spleen samples were compared. Similar to the findings of our previous studies, the ΔtatC mutant was found to be highly attenuated and unable to colonize the spleen and cause systemic infection (22) (Table 2). Seventeen of the mutants tested were found to be as virulent as the parental strain. For some mutants, the competitive index (CI) was greater than 1, and these mutants appeared to be somewhat more virulent than the wild type (Table 2). We also tested these mutants in single infections but did not observe any symptoms in the mice infected with these mutants, suggesting that they are not significantly more virulent in vivo (data not shown). Five mutants with mutations in genes that encode the cell division protein SufI, the iron transporter YbtP, the oxidoreductase YedY, a putative sulfatase with GenBank accession no. YP_071579 (YPTB3074), and the putative urea transporter UrtA showed different levels of attenuation (Table 2). The ΔsufI mutant was essentially as attenuated as the ΔtatC mutant and unable to colonize the spleen. The ΔyptP mutant was the second most attenuated mutant (62-fold) and showed a decreased level of spleen colonization (Table 2). As the level of attenuation for the ΔyedY, ΔYPTB3074, and ΔutrA mutants was quite low (1.4-fold, 1.8-fold, and 1.5-fold, respectively) (Table 2), we decided to focus on the ΔsufI and ΔybtP mutants to further analyze their contributions to the attenuation of the Tat-deficient strain. For this, it was first important to know whether these two proteins are true Tat substrates in Y. pseudotuberculosis.
SufI is a well-characterized Tat substrate in E. coli, and we wanted to establish if SufI also showed a Tat-dependent periplasmic localization in Yersinia (26, 27). Therefore, we constructed a gene fusion between the coding region of the Tat signal peptide of the Y. pseudotuberculosis sufI gene and the mCherry gene under the control of the araB promoter. The plasmid was introduced into both the parental wild-type strain and the ΔtatC mutant strain. Following arabinose induction, the two strains were examined by fluorescence microscopy to monitor mCherry localization. In wild-type bacteria, fluorescence was seen in the periplasm/membrane region, while in the ΔtatC mutant, the protein was evenly localized in the cytoplasm (Fig. 1A). Our interpretation of these results is similar to our interpretation of the results for E. coli: SufI is also translocated to the periplasm by the Tat pathway in Y. pseudotuberculosis (26). To analyze the Tat dependency of YbtP, we took a different approach. YbtP is predicted to be an inner membrane permease protein (28), and therefore, the use of mCherry fusions and fluorescence microscopy would not be an optimal strategy. Instead, we cloned full-length ybtP in frame with the FLAG tag sequence into an IPTG (isopropyl-β-d-1-thiogalactopyranoside)-inducible plasmid. The resulting plasmid was introduced into the wild-type and ΔtatC mutant strains. Bacteria expressing YbtP-FLAG were fractionated into cytoplasm, periplasm, and membrane fractions, and YptP-FLAG expression was monitored by Western blotting using a FLAG tag antibody. YptP was found to localize in the membrane fraction both in the parental strain and in the ΔtatC mutant strain (Fig. 1B). As a control, the localization of the membrane protein FtsH was used to verify the accuracy of the fractionation method (Fig. 1B). These results suggest that YbtP is not a Tat-dependent protein and, therefore, that YbtP does not contribute to the attenuation of the ΔtatC mutant.
As YbtP was found not to be a Tat substrate, we decided to focus on the ΔsufI mutant and to investigate the contribution of SufI to the in vivo attenuation of the ΔtatC mutant. Bioluminescent in vivo imaging (BLI) allows the detection, visual characterization, and semiquantification of bacteria during infection of mice in real time (29). This method has widely been used to monitor Y. pseudotuberculosis YPIII infections using an in vivo imaging spectrum (IVIS) for BLI analysis (30,–33). To allow similar studies in strain IP32953, we constructed bioluminescent variants of the wild-type strain and the ΔtatC and ΔsufI mutant strains by integration of the complete lux operon from Photorhabdus luminescens into the 16S rRNA gene using a thermosensitive plasmid, p16Slux. The correlation between luminescence, growth, and numbers of CFU in all bioluminescent variants of the parental wild-type strain (strain wtL) and the ΔtatC and ΔsufI mutant strains (the ΔtatCL and ΔsufIL mutant strains, respectively) was similar over time (Fig. 2A and andB).B). There was no significant difference in the luminescence levels between the strains, suggesting that a single copy of the lux operon was inserted into the chromosomes of all strains. Furthermore, bioluminescent strains expressed similar levels of bioluminescence for more than 100 generations (data not shown), indicating that the integrated genes were not readily lost during prolonged growth without antibiotic selection.
Next, the bioluminescent variant of the parental strain (IP32953 wtL) was compared with the original wild-type parental strain (IP32953) during in vivo infection in mice to determine if virulence was affected by the lux operon insertion. Accordingly, we infected BALB/c mice orally with the wt and wtL strains and monitored their survival. Neither the establishment nor the progression of the infection was significantly affected in the wtL strain compared to that in the wt strain, and mice infected with both strains showed escalating symptoms of disease starting from day 5 p.i. All mice exhibited the characteristic symptoms of systemic infection (diarrhea, weight loss, fuzzy fur, and hunched postures) between days 5 and 9 p.i. and were euthanized when they showed severe symptoms of disease (Fig. 2C). The progression of infection in mice that were infected with the wtL strain was also monitored by BLI (Fig. 2D). BLI analyses of the mice that were euthanized due to signs of systemic infection revealed that the bacteria were indeed present in the Peyer's patches (PPs), mesenteric lymph nodes (MLNs), spleen, and liver (Fig. 2D). The organs from infected mice were severely affected; e.g., they had a small empty cecum, blood-filled mesenterium, enlarged PPs, empty stomach, enlarged spleen, and necrotic liver. Overall, these results indicate that the bioluminescent parental strain is fully virulent, able to cause a systemic infection, and well suited to be used for BLI analyses.
In our first analysis with the bioluminescent variants of the wt, ΔsufI, and ΔtatC strains, we compared the initial colonization abilities of the two mutants at early time points of infection. BALB/c mice were infected with the wtL strain and the ΔtatCL and ΔsufIL mutant strains, and the infection was followed for 24 h postinfection (hpi) by BLI analysis. BLI analyses using specified regions of interest (ROIs) to determine the total flux emitted from the abdomen of each mouse showed that the total flux emitted at 6 hpi in mice infected with the ΔtatCL and ΔsufIL mutants was lower than that in mice infected with the wtL strain (Fig. 3A and andB).B). Interestingly, at 24 hpi, mice infected with the wtL strain emitted a total level of flux (2.4 × 107 photons/second) 57 times higher than mice infected with the ΔtatCL mutant (4.2 × 105 photons/second) and 16 times higher than mice infected with the ΔsufIL mutant (1.5 × 106 photons/second) (Fig. 3B). BLI analysis of dissected organs demonstrated that all strains were able to colonize the cecum as early as 6 hpi (Fig. 3C). For a more detailed investigation of the early colonization of the different strains, the bacterial loads in the stomach, cecum, small intestine, cecal contents, and feces were determined at 6, 12, and 24 hpi. We included the stomach in the samples, as we have previously shown that the ΔtatC mutant has a lower rate of survival under in vitro acidic pH conditions (23). Indeed, the ΔtatCL mutant showed a bacterial load in the stomach 200 times lower than that of the wtL strain at 6 hpi (3.5 × 103 versus 8.5 × 105 CFU/g) (Fig. 3D). At 6 hpi, the bacterial loads in the cecum were quite similar for all strains (~105 CFU/g). At 12 hpi, the numbers of wtL strain bacteria showed a rapid increase (~106 CFU/g), while the levels for the ΔtatCL and ΔsufIL mutants remained more or less similar to those at 6 hpi. Intriguingly, at 24 hpi, the bacterial load for the wtL strain continued to increase (~108 CFU/g), while the bacterial load for the ΔtatCL mutant remained similar to the level at 12 hpi. However, the load for the ΔsufIL mutant increased to ~106 CFU/g at 24 hpi, but it was still about 100-fold lower than that for the wild type, which indicates that the ΔsufIL mutant more efficiently colonizes the cecum than the ΔtatCL mutant (Fig. 3E). The same pattern was also seen for the bacterial load in the small intestine, cecal contents, and feces (Fig. 3F and andG;G; data not shown). Overall, these results were quite consistent with the ROI measurements shown in Fig. 3B. SufI has been suggested to be involved in cell division (34), and the loss of Tat also confers impaired septation in E. coli and S. Typhimurium (21, 35). In order to establish if the impaired colonization ability of the mutants is due to a replication and/or cell division defect, we decided to monitor in vitro growth in more detail, where we observed the morphology of cells along with growth at different time points (Fig. 4). Both the ΔtatC and ΔsufI mutants were slightly elongated compared to the morphology of the wild-type strain during the first 6 h of growth, when the bacteria replicated logarithmically. However, after 8 h, when the bacteria entered into stationary phase, both mutants gradually showed a morphology similar to that of the wild type (Fig. 4). The growth properties and morphology of the bioluminescent strains were similar to those of the nonbioluminescent strains, and also here, there was no significant difference in growth rates and the numbers of CFU between the wild-type and the mutant strains (data not shown and Fig. 2).
Together, these data demonstrate that the bacterial numbers of both the ΔsufIL and ΔtatCL mutant strains were lower than those of the wild-type strain during early time points of infection and the ΔsufIL mutant showed an intermediate phenotype. Importantly, both mutant strains were able to colonize the cecum, although they did so less efficiently than the parental strain.
Next, we wanted to compare the ability of the two mutants to disseminate at the later time points of infection. Mice (4 in each group) were infected with the wtL strain and the ΔtatCL and ΔsufIL mutant strains, and the infection was monitored using BLI. The parental strain disseminated rapidly and caused systemic infection on days 5 to 7 p.i., while the two mutant strains were unable to spread and were eventually cleared at about 18 days postinfection (dpi) (Fig. 5A). Mice infected with the ΔtatCL and ΔsufIL strains showed only a small degree of weight loss and diarrhea at about day 2 to 3 p.i. Thereafter, they gained weight and appeared healthy after 5 dpi and onwards. ROI analysis from BLI showed a higher total flux emitted from mice infected with the wtL strain than from those infected with the two mutant strains throughout the experiment. The signals for the ΔtatCL mutant (1.3 × 106 photons/second) and the ΔsufIL mutant (1.7 × 106 photons/second) decreased significantly compared to the signal for the wtL strain (9.4 × 107 photons/second) and continued to decrease until 18 dpi (Fig. 5A and andB).B). Organ analysis by BLI on days 3, 5, and 7 p.i. showed that all three strains could be found in PPs and MLNs (Fig. 5C). On day 5 p.i., the wtL strain further disseminated to the spleen and liver to cause systemic infection, whereas the ΔtatCL and ΔsufIL mutants remained within MLNs and did not spread systemically (Fig. 5C). Accordingly, on days 5 and 7 p.i., ROI measurements from MLNs that were infected with the mutants showed the same trend with a decrease in the total flux emitted (Fig. 5D). On day 18 p.i., all mice except for one mouse which was infected with the ΔtatCL mutant had cleared the infection (Fig. 5A).
Taken together, these data show that both the ΔtatCL and ΔsufIL mutants have the ability to colonize the cecum and PPs and disseminate to the MLNs but that they cannot spread to the spleen and liver.
It is known that neutrophils are major targets for the T3SS effectors and Yersinia can be found inside neutrophils in the PPs, MLNs, and spleen (36). As our results showed that MLNs could serve as a barrier to inhibit further dissemination and systemic infection by the ΔtatCL and ΔsufIL mutants, the interaction of the mutants and wild-type bacteria with neutrophils in this tissue was examined. MLNs isolated on days 3, 5, and 7 p.i. from the mice infected with the wtL, ΔtatCL, and ΔsufIL strains (positive for bioluminescence) were stained using anti-Yersinia and anti-Ly6G antibodies. On day 3 p.i., all three strains formed distinct foci in MLNs that were surrounded by neutrophils (Fig. 6). At this time point, there were no significant differences between the strains in terms of the colonization of MLNs and the recruitment of neutrophils. On days 5 and 7 p.i., several foci and also single bacteria were detected in MLNs from wild-type-infected mice. However, not all wild-type foci were surrounded by neutrophils, suggesting an ability of the wild-type bacteria to escape from neutrophil attack. In distinct contrast, fewer foci were detected in MLNs from mice infected with the ΔtatCL and ΔsufIL strains on days 5 and 7 p.i., and at all times, these foci were surrounded by large numbers of neutrophils that appeared to completely contain the bacteria.
Together, these data demonstrate that although the ΔsufIL and ΔtatCL strains are initially able to colonize MLNs, a functional Tat system and its substrate, SufI, are required for Y. pseudotuberculosis to disarm the early immune response and survive to spread systemically.
Twin arginine translocation has been implicated in the virulence of many different pathogenic bacteria, but so far only a few Tat substrates have been directly linked to virulence. Our previous studies suggested that the attenuation was the result of several contributing virulence-related phenotypes linked to Tat deficiency. Here we show that the virulence attenuation essentially can be attributed to a single Tat substrate: the periplasmic cell division protein SufI. The ΔtatC mutant as well as the ΔsufI mutant was highly attenuated and unable to spread from the MLNs to cause systemic infection. Our finding is in line with the findings of previous mutant screening studies where sufI mutants of Y. pseudotuberculosis YPIII were found to be avirulent (37, 38). SufI has been used in mechanistic studies of Tat structure and function in E. coli (27, 39, 40), and we confirmed that Tat also promotes SufI translocation to the periplasm in Y. pseudotuberculosis. This highlights that the periplasmic localization of SufI is essential for its function. The definite function of SufI is not clear, but overexpression of SufI can suppress the thermosensitivity of an ftsI123 mutant in E. coli, which suggests a function in cell division (41). Similarly, overexpression of SufI can also suppress the growth defect in ftsEX mutants with low-osmolarity and filamentation phenotypes, suggesting a role for stabilizing the divisome (34, 42). SufI localizes to the septal ring, and it is recruited to the septal ring in the late phases of constriction. However, SufI does not appear to be required for functional cell division and growth (26). Interestingly, SufI is found only in organisms in the Enterobacteriales and Pasteurellales of the Gammaproteobacteria (26). In S. Typhimurium, a triple sufI, amiA, and amiC mutant of Tat substrates conferred attenuation similar to that conferred by a single sufI mutant in our study, and this was attributed to a defect in envelope structure (21). AmiA and AmiC are two amidases that cleave peptidoglycan chains in constricting cells, and they are both Tat substrates in E. coli and in Salmonella spp. In these bacteria, ΔtatC or ΔamiA ΔamiC mutants have been shown to be sensitive to SDS, bile, and antimicrobial peptides (35, 43). However, according to our findings in Y. pseudotuberculosis, only AmiC and SufI are potential Tat substrates, but neither of them contributes to envelope stress and only the ΔtatC mutant is important for cell envelope maintenance (23). Even though the specific role of these cell division proteins and their contribution to virulence differ, these findings indicate a central role for cell division proteins in the virulence of enteric pathogens.
In our competitive mouse infection studies, we could also identify four additional mutants, the ΔyedY, ΔYPTB3074, ΔurtA, and ΔybtP mutants, which were attenuated at different levels in vivo. The low levels of attenuation seen for the ΔyedY, ΔYPTB3074, and ΔurtA mutants make it unlikely that they would each significantly contribute to the attenuation of the Tat-deficient strain. ybtP encodes an inner membrane ABC transporter involved in yersiniabactin transport (28), and a Y. pestis KIM6 ΔybtP mutant strain has also been shown to be avirulent (28). Our analysis on the subcellular localization of YptP in wild-type and ΔtatC mutant strains revealed that YbtP is not a true Tat substrate since YbtP localized to the membrane both in the wild-type strain and in the ΔtatC mutant strain. There is evidence that a subset of inner membrane proteins is Tat dependent. This has been shown, for instance, for [NiFe] hydrogenases that contain hydrophobic C-terminal helices with topologies in which the N terminus localizes to the periplasm and the C terminus localizes to the cytoplasm (44). YbtP, on the other hand, has transmembrane domains in the N terminus and has no periplasmic domain. The C terminus encodes an ATPase domain suggested to localize to the cytoplasm, and altogether, these findings indicate that YptP is not a Tat substrate (45). Thus, similar to what was shown in Y. pestis, this protein is important for virulence, but as it is not a Tat substrate, it does not contribute to the attenuation of the tat mutant.
Among the 25 potential Tat substrates identified, we were unable to mutate three genes, mdfA, fliY, and YPTB3261, in spite of several attempts. MdfA is an inner membrane protein associated with drug efflux with 12 transmembrane domains. MdfA is well characterized in E. coli, and structural studies have shown that the very N terminus of the protein faces toward the cytoplasm (46, 47). The alignments of MdfA from Y. pseudotuberculosis and E. coli revealed that they had a very high level of similarity (73.3% identity), but the E. coli MdfA does not have an RR motif in the N terminus and the Yersinia MdfA has 6 additional amino acid residues in the N-terminal sequence with an RR motif. Since the protein is not predicted to have any periplasmic domains, we find it unlikely that MdfA is a Tat substrate (44, 48). fliY is the first gene in a putative operon (fliY yecS yecC) encoding proteins with a suggested function in cysteine transport that were previously identified in E. coli (49). These genes do not appear to be essential in E. coli, but this could be different in Yersinia. YPTB3261 encodes a putative exported protein with one domain similar to dioxygenases. The major activity of this domain is to cleave aromatic rings for degradation of aromatic compounds (50). Interestingly, the upstream gene YPTB3260 also encodes a putative Tat substrate, but in this case, gene deletion was successful. YPTB3260 encodes a poly-3-hydroxybutyrate depolymerase which could have a function similar to or overlapping that of YPTB3261. While we cannot rule out the possibility that these three genes could be Tat substrates and contribute to the attenuation of the Tat mutant, we still find it unlikely that they would have a major impact on virulence, given the unique phenotype of the ΔsufI mutant among the mutants with mutations in the 22 genes analyzed in the study.
For a more comprehensive analysis and in vivo characterization of the ΔtatC and ΔsufI mutants, we generated bioluminescent variants of these mutants to enable the monitoring of infection in real time. During the first 24 hpi, the intestine was rapidly colonized, after 6 hpi the bioluminescent wild-type strain as well as the two mutants had already colonized the cecum, and the numbers of CFU of the wild-type strain increased about 100-fold from 6 to 24 hpi. Even though the two mutants were able to colonize the cecum, the number of bacteria did not increase but remained more or less constant during the first 24 h, indicating a struggle in the initial adaptation of those mutants in the cecum. Moreover, we also showed that this struggle was independent of any cell division and/or replication defect of the mutant strains since in both mutants the in vitro growth and number of CFU correlated and did not show any difference from the in vitro growth and number of CFU of the wt strain. Accordingly, there was an initial drop in numbers of ΔtatCL mutant bacteria in the stomach, a finding which is in line with our previous finding on the growth defect of the ΔtatC mutant at acidic pH (23), suggesting that some population of this strain could be lost during the intestinal passage. However, it is very likely that the ΔtatCL mutants that survived in the acidic environment in the stomach were still able to colonize the cecum and PPs at the later time points of infection. Our results also show that, in contrast to the two mutants, wild-type Y. pseudotuberculosis is able to replicate in the host intestine and also reach to MLNs already at 24 hpi, which is in line with the findings of previous studies (31).
At later time points of infection, both the ΔsufIL and ΔtatCL mutants disseminated by 3 dpi and colonized PPs and MLNs. However, neither mutant was able to cause systemic infection, and the mice infected with the mutants showed limited or no symptoms of infection. Thus, the in vivo characteristics of both mutants at later time points of infection were very similar. The small difference seen between the ΔsufIL and ΔtatCL mutants early in infection suggests a role for other Tat substrates possibly linked to the defects in iron uptake, copper resistance, and envelope structure (23). These differences, however, seem to be less significant at later time points, when the ability to survive and spread from MLNs to cause systemic infection strictly requires the SufI function in the periplasm. In contrast, a previous study by Mecsas et al. found that a sufI mutant of the Y. pseudotuberculosis YPIII strain was unable to colonize the cecum, PPs, and MLNs, which indicates that there is a difference between the IP32953 and YPIII strains in terms of in vivo colonization by their sufI mutants (37). Moreover, SufI was also identified in a genetic screen to be important for colonization of the liver by virulence plasmid-deficient Y. pseudotuberculosis following intravenous infection (38). One significant difference between these strains is that YPIII is known to carry a mutation that renders PhoP nonfunctional. PhoP is important for intercellular survival, and it is likely that a lack of a functional PhoP could at least in part affect the ability of YPIII to efficiently colonize PPs and MLNs (51, 52). We have also observed a similar colonization defect in PPs and MLNs in mice infected with a ΔtatC mutant of the YPIII strain (U. Avican, unpublished results).
Neutrophils are key players in innate immunity, and they are rapidly recruited to the infection site to eliminate the invading pathogenic bacteria (53). Neutrophils have also been shown to control Yersinia infections (54), and Yersinia species primarily target neutrophils (55). Our analysis showed that on day 5 and day 7 p.i., both the ΔtatCL and the ΔsufIL mutants had formed fewer infection foci that were surrounded by large numbers of neutrophils. In contrast, we found that the wtL strain formed large numbers of foci and that some of them were not surrounded by neutrophils, which could make it possible for these bacteria to spread and cause systemic infection. The T3SS effectors YopH, YopE, and YopK are essential for the targeting of and escape from neutrophils during early infection (30). Our previous studies showed that the T3SS function was not affected in the ΔtatC mutant (22, 23). T3SS expression has been shown to be downregulated in S. Typhimurium and in P. syringae Tat-deficient strains, a finding which is similar to our findings. However, in our case attenuation could not be attributed to the T3SS function (11, 21). Additionally, we did not see any effects of the ΔsufI mutation on T3SS (data not shown), verifying that the decreased survival in MLNs seen for both mutants is T3SS independent. It was recently found that during persistent Y. pseudotuberculosis infection, gene expression is reprogrammed toward more stress- and metabolism-related gene functions and that T3SS is indeed downregulated (56). Therefore, predicting the outcome of in vivo infections on the basis of different in vitro studies of virulence-related genes is difficult, and more in vivo infection studies at critical sites like MLNs are therefore needed.
Here we have shown that a single Tat substrate, SufI, is primarily responsible for the high level of virulence attenuation of a Tat-deficient strain of Y. pseudotuberculosis, even though its predicted function does not seem to be directly related to classical virulence determinants.
Eight-week-old female BALB/c mice were purchased from Taconic, Denmark, and were given food and water ad libitum. Mice were housed at the Umeå Centre for Comparative Biology, Umeå University. The study was approved by the Animal Ethics Committee of Umeå University (Dnr A144-12).
The bacterial strains and plasmids used in this study are listed in Table 3. Bacteria were routinely cultured in Luria-Bertani (LB) broth at 26°C only for Y. pseudotuberculosis IP32953, and at 37°C for E. coli with aeration. Antibiotics were used at a final concentration of 25 μg/ml for chloramphenicol and 500 μg/ml for erythromycin. IPTG (isopropyl-β-d-1-thiogalactopyranoside) and arabinose were added to final concentrations of 0.4 mM and 0.2%, respectively, to induce the expression of genes cloned under the control of the tac and araB promoters. For the in vitro growth studies, overnight cultures of different strains were diluted to an optical density at 600 nm (OD600) of 0.1 and grown in LB medium for 10 h. Samples were taken every 2 h, and the OD600 and the luminescence were measured with a Tecan Infinite 200 plate reader (Tecan Group Ltd., Switzerland). Each sample was also serially diluted and plated onto Luria agar (LA) plates for determination of the number of CFU.
Insertion and in-frame deletion mutants with mutations of putative Tat substrates were constructed as previously described (23). The E. coli S17-1 λpir strain containing plasmid pUA066 was conjugated and integrated into the strains with the in-frame deletions ΔcynT, ΔYPTB1325, ΔycdB, and ΔYPTB3260 as previously described (57) to introduce chloramphenicol resistance for competitive infection experiments in mice.
To generate a fusion between the signal peptide of SufI (SufIsp) and mCherry, the DNA residues encoding the predicted signal peptide of SufI (amino acid residues 1 to 33) and the mCherry gene from plasmid pBAD24-TorA1–48-mCherry (58) were amplified by PCR. The resulting 112-bp and 726-bp PCR products were fused by cloning into the NcoI/HindIII sites of the pBAD24 plasmid by use of an In-Fusion HD cloning kit (Clontech) according to the manufacturer's instructions and transformed into the E. coli TOP10 strain. The DNA sequence of the gene fusion was confirmed by sequencing. The resulting construct, pBAD24-sufIsp-mCherry, was then introduced into the parental and ΔtatC strains by electroporation.
For determination of SufI localization, overnight cultures of the parental and ΔtatC mutant strains containing pBAD24-sufIsp-mCherry were diluted to an OD600 of 0.1 in 5 ml LB medium with appropriate antibiotics and grown to mid-exponential growth phase. After addition of arabinose and incubation for 3 h, 5 μl of cells was placed onto poly-l-lysine-coated microscopy slides and observed with a Nikon Eclipse 90i fluorescence microscope with a 100× oil immersion lens and phase-contrast and tetramethyl rhodamine isocyanate filters. Images were captured using a Hamamatsu Orcha C4742-95 camera and NIS-Elements AR (version 3.2) software (Nikon Instruments).
For the C-terminal FLAG tagging of the YbtP protein, the DNA sequence encoding the yptP gene was amplified by PCR, and the DNA sequence encoding the FLAG tag (DYKDDDDK) was introduced into the reverse primer before the stop codon. The resulting 1,827-bp PCR product was cloned into plasmid pCR2.1-TOPO (Invitrogen) in E. coli TOP10 to generate yptP-FLAG. After confirmation of the sequence by DNA sequencing, yptP-FLAG was digested with EcoRI and HindIII and cloned into the pMMB207 plasmid that had been linearized by the same restriction enzymes, and pMMB207 plasmid was introduced into E. coli S17-1 λpir by transformation. The resulting E. coli S17-1 λpir strains were then conjugated to parental and ΔtatC strains.
The insertion of the luxABCDE operon into the 16S rRNA gene in the chromosome of the Y. pseudotuberculosis IP32953 strain was performed as previously described with some minor modifications (59). Briefly, plasmid p16Slux was introduced into the wild-type, ΔtatC, and ΔsufI strains by electroporation. The transformants were selected by plating of the cells on LA containing 500 μg/ml erythromycin and incubation at 26°C (the temperature permissive for p16Slux replication). Light emission by the Eryr colonies was verified by using a Bio-Rad ChemiDoc MP system, and the presence of p16Slux was confirmed by DNA minipreparation and restriction analyses. The positive clones were then incubated at 26°C overnight in LB medium, diluted 1:1,000 into 100 ml of LB supplemented with 500 μg/ml erythromycin and 2.5 mM CaCl2, grown for 1 h at 26°C, and then shifted to 39°C (a temperature nonpermissive for the replication of p16Slux) for incubation overnight. The overnight cultures were then serially diluted and plated on LA plates containing erythromycin and CaCl2, followed by incubation at 39°C. The colonies were streaked onto erythromycin-containing LA plates, incubated at 39°C overnight, and checked the following day for light emissions. At the nonpermissive temperature with antibiotic stress, plasmid integration into the 16S rRNA locus of the bacterial chromosome was forced via homologous recombination. The light-emitting positive clones were streaked onto LA plates with erythromycin, and the integration of the plasmid was verified by PCR.
Overnight cultures of the parental and ΔtatC mutant strains containing the pMMB207-yptP-FLAG plasmid were diluted 1:10 in 50 ml LB medium (with appropriate antibiotic selection), and the cells were grown to mid-exponential growth phase at 26°C following addition of IPTG and incubation for an additional 3 h. Cells were harvested by centrifugation at 4,500 × g and 4°C and fractionated into cytoplasmic, periplasmic, and membrane compartments as described previously (60). The samples were separated by 12% SDS-PAGE and immunoblotted using antibodies to the FLAG tag (anti-mouse; Sigma) and FtsH (anti-rabbit).
For oral infection, mice were deprived of food and water for 16 h prior to infection. For competitive infection, 0.25 × 108 CFU/ml from overnight cultures of the isogenic wild-type strain and each mutant strain were mixed 1:1, centrifuged, and resuspended in the same volume of sterile tap water containing 150 mM NaCl. At day 4 postinfection (p.i.), the spleens were collected, homogenized in 1 ml 1× phosphate-buffered saline (PBS), serially diluted, and plated onto LA plates with or without 25 μg/ml chloramphenicol. The competitive index was calculated by dividing the output ratio of the mutant to the wild type by the input ratio of the mutant to the wild type. For oral infection with the bioluminescent strains, 5 × 108 CFU/ml was prepared from overnight cultures and resuspended in sterile tap water containing 150 mM NaCl. The mice were allowed to drink the bacterial suspensions for 6 h, after which food and water were provided. Infection doses were determined from the CFU counts of serially diluted bacterial cultures used for infection in combination with measurement of the volumes that the mice drank. At the time points indicated above and in the figures, the cecum, stomach, small intestines, feces, and cecal contents were collected, homogenized in 1× PBS, serially diluted, and plated onto LA plates containing 500 μg/ml erythromycin.
For bioluminescent in vivo imaging (BLI) analysis, the infection was monitored and analyzed by using an in vivo imaging spectrum (IVIS; Caliper Life Sciences). Mice were anesthetized with 2.5% isoflurane (IsoFlo Vet; Orion Pharma) and thereafter placed in the IVIS imaging chamber (Caliper Life Sciences) while they were under 0.5% isoflurane anesthesia. The total photon emissions from the bioluminescent bacteria inside the mice or from dissected organs were acquired, and images were analyzed by the use of Living Image (version 4.5) software (Caliper Life Sciences). The bioluminescent signals emitted by the bacteria were determined from defined regions of interest (ROIs) on the abdominal side of each mouse and expressed as the total flux (number of photons per second).
Double immunostaining of MLN sections was performed as described previously with slight modifications (30). Briefly, MLNs were frozen with prechilled isopentane on liquid nitrogen and stored at −80°C. Cryosections (10 μm) were fixed with 4% paraformaldehyde in PBS and permeabilized with 0.5% Triton X-100 (Sigma). To block nonspecific binding, the specimens were incubated with 0.1 M glycine, avidin-biotin blocking reagents (Vector Laboratories), 2% bovine serum albumin (Sigma) in PBS, and 5% serum from the secondary antibody host (goat and donkey; Jackson ImmunoResearch). First, the sectioned tissues were stained using anti-Ly6G (clone 1A8; BD Biosciences) and a biotinylated goat anti-rat immunoglobulin as a secondary antibody (Jackson ImmunoResearch), followed by streptavidin-phycoerythrin (eBioscience). Then, the sections were incubated with rabbit anti-Yersinia serum, followed by Alexa Fluor 488-labeled donkey anti-rabbit antibody (Thermo Scientific). Finally, 4′,6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei. The specimens were mounted in Mowiol mounting medium (Sigma) and examined with a Nikon Eclipse 90i microscope. Images were captured using a Hamamatsu Orcha C4742-95 camera and NIS-Elements AR (version 3.2) software (Nikon Instruments).
Statistical analyses were performed using GraphPad Prism (version 5) software. The Mann-Whitney U test, one-way analysis of variance (ANOVA), and Student's t test were used to analyze differences, with significance being set at a P value of <0.05, <0.01, or <0.001, as indicated in the figure legends. Error bars in the graphs correspond to the standard errors of the means (SEMs), as indicated in the figure legends.
We thank Matthew S. Francis for providing the FtsH antibody and Daniel O. Daley for providing plasmid pBAD24-TorA1–48-mCherry.
This work was funded by the Swedish Research Council (grant 2011-3439 to Å.F.) and the J. C. Kempe Foundation (to U.A.).