|Home | About | Journals | Submit | Contact Us | Français|
The facultative intracellular pathogen Salmonella enterica serovar Typhimurium relies on its Salmonella pathogenicity island 2 (SPI2) type III secretion system (T3SS) for intracellular replication and virulence. We report that the oxidoreductase thioredoxin 1 (TrxA) and SPI2 are coinduced for expression under in vitro conditions that mimic an intravacuolar environment, that TrxA is needed for proper SPI2 activity under these conditions, and that TrxA is indispensable for SPI2 activity in both phagocytic and epithelial cells. Infection experiments in mice demonstrated that SPI2 strongly contributed to virulence in a TrxA-proficient background whereas SPI2 did not affect virulence in a trxA mutant. Complementation analyses using wild-type trxA or a genetically engineered trxA coding for noncatalytic TrxA showed that the catalytic activity of TrxA is essential for SPI2 activity in phagocytic cells whereas a noncatalytic variant of TrxA partially sustained SPI2 activity in epithelial cells and virulence in mice. These results show that TrxA is needed for the intracellular induction of SPI2 and provide new insights into the functional integration between catalytic and noncatalytic activities of TrxA and a bacterial T3SS in different settings of intracellular infections.
In Escherichia coli, thioredoxin 1 (TrxA, encoded by trxA) is an evolutionary conserved 11-kDa cytosolic highly potent reductase that supports the activities of various oxidoreductases and ribonucleotide reductases (1, 29) and interacts with a number of additional cytoplasmic proteins through the formation of temporary covalent intermolecular disulphide bonds (32). Consequently, as trxA mutants of E. coli (51), Helicobacter pylori (13), and Rhodobacter sphaeroides (34) show increased sensitivity to hydrogen peroxide, TrxA has been defined as a significant oxidoprotectant. In addition, TrxA possess a protein chaperone function that is disconnected from cysteine interactions (30, 32).
Salmonella enterica serovar Typhimurium is closely related to E. coli. During divergent evolution, the Salmonella genome acquired a number of virulence-associated genes (20). Many of these genes are clustered on genetic regions termed Salmonella pathogenicity islands (or SPIs). Of these, SPI1 and SPI2 code for separate type III secretion systems (T3SSs). T3SSs are supramolecular virulence-associated machineries that, in several pathogenic gram-negative bacterial species, enable injection of effector proteins from the bacteria into host cells (22, 57). The effector proteins, in turn, manipulate intrinsic host cell functions to facilitate the infection.
The SPI1 T3SS of S. serovar Typhimurium is activated for expression in the intestine in response to increased osmolarity and decreased oxygen tension (22, 57). SPI1 effector proteins are primarily secreted into cells that constitute the epithelial layer and interfere with host cell Cdc42 and Rac-1 signaling and actin polymerization. This enables the bacteria to orchestrate their own actin-dependent uptake into nonphagocytic cells (57). SPI1 effector proteins also induce inflammatory signaling and release of interleukin-1β from infected cells (25, 26).
Subsequent systemic progression of S. serovar Typhimurium from the intestinal tissue relies heavily on an ability to survive and replicate in phagocytic cells (18, 46, 53, 54). S. serovar Typhimurium uses an additional set of effector proteins secreted by the SPI2 T3SS for replication inside host cells and for coping with phagocyte innate responses to the infection (10, 11, 54). The functions of SPI2 effectors include diversion of vesicular trafficking, induction of apoptotic responses, and manipulation of ubiquitination of host proteins (28, 40, 45, 53). Hence, SPI2 effector proteins create a vacuolar environment that sustains intracellular replication of S. serovar Typhimurium (28).
In addition to pathogenicity islands, the in vivo fitness of Salmonella spp. relies on selected functions shared with other enterobacteria. Thus, many virulence genes are integrated into “housekeeping” gene regulatory networks, coded for by a core genome, which steer bacterial stress responses (12, 17, 27, 55). Selected anabolic pathways also contribute to virulence of S. serovar Typhimurium (18, 27), evidently by providing biochemical building blocks for bacterial replication (36).
In S. serovar Typhimurium, TrxA is a housekeeping protein that strongly contributes to virulence in cell culture and mouse infection models (8). However, the mechanism by which TrxA activity adds to virulence has not been defined. Here we show that the contribution of TrxA to virulence of S. serovar Typhimurium associates with its functional integration with the SPI2 T3SS under conditions that prevail in the intracellular vacuolar compartment of the host cell. These findings ascribe a novel role to TrxA in bridging environmental adaptations with virulence gene expression and illuminate a new aspect of the interaction between evolutionary conserved and horizontally acquired gene functions in bacteria.
S. serovar Typhimurium strain 14028 (ATCC, Manassas, VA) was used throughout the study. ssaV and two trxA mutants, trxA::Km and trxA, have been described previously (6, 8). The trxA mutant strain was derived from the trxA::Km mutant by removing the antibiotic resistance cassette as described by Datsenko and Wanner (15). Both trxA mutants display the same phenotype both in vitro and in vivo (8). The hilA mutant strain was obtained from C. A. Lee (2), whereas S. serovar Typhimurium SL1344 mutant strains showing constitutive expression (one strain), promoterless expression (one strain), and ssaG-directed expression (one strain) of green fluorescent protein (GFP) were constructed by I. Hautefort et al. (23). Transfer of GFP expressing genetic entities and construction of hilA and trxA/ssaV double mutants of S. serovar Typhimurium 14028 were conducted by P22 int transduction (48). For routine purposes, bacteria were grown in complex Luria Bertani (LB) medium and all viable counting experiments, selection of mutants, and propagation of strains were conducted using Luria agar plates.
To induce SPI1-dependent invasiveness, bacteria were grown in LB medium to the late logarithmic phase of growth as described previously (8). To induce expression of SPI2, bacteria were taken from an overnight LB agar plate and grown in either of two low pH media, namely, MM5.8 minimal medium (31, 56) or 2-(N-morpholine)-ethane sulfonic acid-based MES5.0 medium (4).
When necessary, the growth media were supplemented with antibiotics or inducers (Sigma Aldrich) at the following final concentrations: kanamycin at 50 mg/ml, ampicillin at 100 mg/ml, chloramphenicol at 10 mg/ml, and l-arabinose at 20 mM.
A pBAD33 plasmid-based construct containing either wild-type trxA or a genetically engineered variant of trxA positioned under the control of the E. coli arabinose promoter (3, 41, 42) was used to complement trxA mutants. As these trxA sequences were derived from E. coli, it should be noted that the primary structures of S. serovar Typhimurium LT2 and E. coli TrxA are identical (9, 38). The plasmids coding for the hemagglutinin-tagged SPI2 effector protein SseJ and GFP were kindly provided by Michael Hensel (21) and David Holden (35).
Isolation of SPI1-secreted effector proteins was conducted using a two-step induction procedure as defined by Clements et al. (12). Briefly, bacteria were propagated in LB medium to induce invasion competence; after this procedure, bacteria were grown for 1 h in HEPES-buffered Dulbecco's modified Eagle medium. Bacterial cultures were adjusted to the appropriate optical density at 600 nm. Bacteria and bacterial debris were removed by repeated centrifugation (18,620 × g for 10 min). Proteins from the resulting supernatants were precipitated with trichloroacetic acid (Sigma Aldrich) at a final concentration of 10%. Pellets were washed in ice-cold acetone, dissolved in reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer, and separated on 12% SDS-PAGE gels. Proteins were detected by staining the gel with Coomassie blue.
J774-A.1 cells and RAW264.7 murine macrophage-like cells (catalog no. TIB 67; American Type Culture Collection, Rockville, MD) were cultivated in RPMI medium (Gibco) supplemented with 10% fetal bovine serum (Gibco), l-glutamine (10 mM final concentration), HEPES (10 mM final concentration), and gentamicin (10 μg/ml final concentration). The l-glutamine, HEPES, and gentamicin were all from Sigma Aldrich. For propagation of epitheloid MDCK cells, RPMI medium was replaced by Dulbecco's modified Eagle medium (Gibco).
To probe for functionality of SPI1, dilutions of overnight cultures were grown for 2 h in LB medium and then applied on MDCK cells as defined by Bjur et al. (8). To analyze the dependency of invasiveness on oxidative stress, hydrogen peroxide (Sigma Aldrich) was added throughout the 2-h incubation.
Infection of nonactivated J774-A.1 or RAW264.7 cells was carried out as described previously (8). Briefly, bacteria were cultured to obtain a noninvasive phenotype and opsonized with 10% preimmune mouse serum and then used at a multiplicity of infection (MOI) of 10:1. At 1 h postinfection, medium containing gentamicin at a concentration of 50 μg/ml was applied and cells were incubated for 45 min to kill extracellular bacteria. For continued incubations, the killing medium was replaced by maintenance medium containing 10 μg of gentamicin/ml. Cells were lysed by hypotonic lysis at indicated time points, and the amounts of live intracellular bacteria were determined by CFU counts.
For analysis of SseJ-2HA secretion, proteins were isolated from bacteria grown in MM5.8 minimal medium with high-level aeration. The samples were adjusted to the same optical density at 600 nm, and the bacteria were pelleted by repeated centrifugation (twice for 10 min each time at 18,620 × g). Thereafter, proteins in the supernatants were precipitated with trichloroacetic acid at a final concentration of 10% (wt/vol). Pellets were washed in ice-cold acetone and dissolved in reducing SDS-PAGE sample buffer. Proteins were separated on either 12% or 15% SDS-PAGE gels (33).
For analysis of SseJ-2HA, proteins were transferred onto a Hybond-P polyvinylidene difluoride membrane (Amersham Pharmacia Biotech) as described previously (52). Hemagglutinin-tagged (HA-tagged) SseJ was detected by immunoblotting with a mouse monoclonal anti-HA-tag antibody (HA-probe F-7; Santa Cruz Biotechnology and Covance) (1:1,000). Bound antibodies were detected using the SuperSignal kit (Pierce). All procedures following the transfer to the Hybond-P polyvinylidene difluoride membrane were carried out according to the descriptions provided by the manufacturer of the chemoluminescence detection kit (Pierce). A Pageruler Plus prestained protein ladder (Fermentas) was used as a molecular mass marker in the immunoblot experiments.
RAW264.7 and MDCK cells were grown as described above and seeded on glass coverslips. Infection was carried out with wild-type S. serovar Typhimurium, a trxA mutant, and a trxA mutant complemented with pFA3 (or pFA8 in the case of MDCK cells), all transformed with the plasmid p2777 (21) and applied at an MOI of 10 or 100. To induce TrxA expression, cell culture medium was supplemented with l-arabinose (20 mM). At 8 h postinfection, the cells were fixed in phosphate-buffered 4% formaldehyde (pH 7.2), permeabilized with 0.2% saponin, and incubated with blocking solution (10 mg/ml bovine serum albumin, 10% fetal bovine serum, and 0.2% saponin). Detection of SseJ-2HA was performed using a mouse anti-HA primary antibody (Covance) (1:50) and an anti-mouse immunoglobulin G Cy3-conjugated secondary antibody (Jackson ImmunoResearch) (1:50). Coverslips were mounted on microscopy slides and observed using a fluorescence microscope.
Expression of the SPI2 ssaG gene was followed by measurement of the increase in fluorescence of an S. serovar Typhimurium strain carrying a single-copy ssaG::gfp+ fusion in the putA locus (23). As controls, we applied isogenic strains with the rpsM directing constant gfp+ expression or a strain with a promoterless gfp+ inserted into proU (23). For measurement of GFP expression, bacteria were grown overnight in LB medium and subsequently washed and diluted 50-fold in MM5.8 minimal medium. Time zero corresponded to the beginning of incubation in MM5.8 minimal medium, and samples were collected and fixed for 5 min at room temperature in 4% (wt/vol) phosphate-buffered formaldehyde (pH 7.2). Fixed bacteria were washed in phosphate-buffered saline, and fluorescence levels were measured using a BD Biosciences FACScan system and analyzed using CellQuest software.
Female BALB/c J mice at the age of 6 to 8 weeks were purchased from Taconic Europe (Denmark) and subjected to either of two infection protocols. To analyze visceral dissemination, mice were challenged orally with 108 bacteria per mouse as described previously by Sukupolvi et al. (50). For in vivo competition experiments, mice were infected intraperitoneally with a mixture that consisted of either (i) S. serovar Typhimurium trxA complemented with pFA3 (coding for wild-type trxA) and trxA::Km complemented with the empty pBAD33 vector or (ii) trxA complemented with pFA8 (coding for catalytically inactive TrxA) and trxA::Km complemented with the empty pBAD33 vector. Bacteria were given in a 1:1 ratio, with a total dose of 104 bacteria per mouse. Five mice from each group were sacrificed at 24 h and 72 h postinfection. The pBAD33/pFA3 or pBAD33/pFA8 ratios in liver and spleen samples were determined by CFU counts of viable bacteria. To distinguish between the complemented trxA mutants, and to assure that the complementing plasmids were retained at high efficiency, samples were plated on ordinary LB agar plates and on plates supplemented with kanamycin or chloramphenicol or kanamycin plus chloramphenicol.
Mice were housed at Karolinska Institutet, Stockholm, Sweden, under normal conditions in accordance with both institutional and national guidelines. Animal experiments were conducted under ethical permits N 284/04, N 107/06, and N 345/08 (provided by Stockholms Norra Djurförsöksetiska nämd).
Data from cell culture experiments and in vivo experiments were analyzed by a two-sided t test. All P values from the t test or Mann-Whitney test analyses are specified in the figure legends.
Expression of SPI2 is repressed during growth in rich laboratory medium but induced through growth in medium characterized by low pH and low concentrations of Mg++ and phosphate (4, 31, 56). These parameters mirror the SPI2-inducing environmental cues prevailing in the intracellular vacuolar compartment where Salmonella bacteria reside and replicate (56).
The gene ssaG codes for an essential component of the SPI2 T3SS (22). To monitor SPI2 induction, we used a chromosomally integrated ssaG::gfp+ transcriptional gene fusion in wild-type S. serovar Typhimurium. The bacteria were shifted from growth in LB medium to SPI2-inducing MM5.8 growth medium, which is characterized by low pH and a low concentration of Mg++ and phosphate (31). At 5 h after the shift, the ssaG::gfp+-tagged S. serovar Typhimurium bacteria showed an apparent increase in fluorescence compared to the fluorescence observed at 2 h after the shift (Fig. (Fig.1A),1A), which is consistent with an induction of ssaG expression in MM5.8 medium.
To analyze whether expression of TrxA was affected under SPI2-inducing growth conditions, TrxA levels were quantified by immunoblotting using wild-type cultures grown in LB, MM5.8, or MES5.0 medium (MES5.0 medium represents another low-pH growth medium that activates expression of SPI2) (4). This analysis showed that TrxA was expressed in LB medium and yet that TrxA expression was substantially increased upon growth in MM5.8 or MES5.0 medium (Fig. (Fig.1B).1B). Thus, TrxA and SPI2 are coinduced under growth conditions that promote SPI2 expression.
We proceeded to probe for functional interconnections between TrxA and SPI2 under SPI2-inducing growth conditions. To address whether TrxA affected SPI2 induction we compared the fluorescence induction of the ssaG::gfp+-tagged wild-type and trxA mutant strains upon a shift of growth from LB medium to MM5.8 medium. At 2 and 5 h postshift, the two strains expressed equally low and high levels of fluorescence intensity (Fig. (Fig.1A).1A). Interestingly, although the growth of the trxA mutant was not affected in MM5.8 medium, a kinetic follow-up of the induction of ssaG::gfp+ between the 2- and 5-h time points revealed a marked delay in ssaG::gfp+ expression in the trxA mutant background (Fig. (Fig.1C).1C). The fluorescence emitted from a constitutively expressed chromosomal control gfp+ gene was not affected in the trxA mutant (Fig. (Fig.1A).1A). These results showed that TrxA is needed for proper induction of the SPI2 T3SS ssaG gene in MM5.8 medium.
To define whether the dysregulation of ssaG in the trxA mutant is associated with any functional consequences regarding SPI2 activity, we set out to analyze secretion of SPI2 effector proteins in the trxA mutant. This was achieved using an S. serovar Typhimurium 14028 strain harboring plasmid p2777. Plasmid p2777 codes for a constitutively expressed GFP and a 2HA-tagged SPI2 SseJ effector protein expressed by the native sseJ promoter. When we probed for the HA-tagged SseJ SPI2 effector protein, we recorded significant expression of the fusion protein in both the wild type and the trxA mutant of S. serovar Typhimurium upon growth in MM5.8 medium (Fig. (Fig.2).2). Yet, while we were able to detect the SseJ-2HA fusion protein in both the cellular fraction and the extracellular fraction in the wild-type strain, the fusion protein was detected only in the cellular fraction in the trxA mutant. Introduction of the recombinant pFA3 plasmid that codes for TrxA into the trxA mutant restored SseJ-2HA secretion (Fig. (Fig.22).
As TrxA was needed for SPI2 activity in vitro, we set out to analyze to what extent the functionality of SPI2 was affected by TrxA during infection of macrophage-like RAW264.7 or epithelial MDCK cells. We have previously shown that SPI2 genes are already induced in macrophage-like and epithelial cells at 4 h postinfection (16, 24) and that SPI2-mediated intracellular replication is not manifested before 8 h postinfection in our settings of cell culture infections (44). Thus, in order to avoid any effects related to bacterial intracellular replication while enabling the recording of significant numbers of intracellular bacteria, we infected RAW264.7 or MDCK cells with wild-type or TrxA-deficient S. serovar Typhimurium carrying plasmid p2777 at an MOI of 100:1 and limited the infection to an 8 h period of time.
Fluorescence microscopy of wild-type bacteria carrying p2777 revealed that infected RAW264.7 cells contained perinuclear ensembles of S. serovar Typhimurium at 8 h postinfection (Fig. (Fig.3A).3A). These bacterial aggregates stained for secreted SseJ-2HA (Fig. (Fig.3A),3A), reflecting the functionality of the SPI2 T3SS. The TrxA-deficient S. serovar Typhimurium/p2777 strain revealed more scattered ensembles of bacteria, with none of them staining for the HA tag (Fig. (Fig.3A).3A). Complementation with pFA3 (coding for wild-type TrxA) restored a wild-type perinuclear appearance and staining for secreted SseJ-2HA of the trxA mutant in infected RAW264.7 cells (Fig. (Fig.3A).3A). In contrast, complementation of the trxA mutant with pFA8 (which codes for a catalytically inactive TrxA) did not restore SseJ-HA secretion in RAW264.7 cells, even at 16 h postinfection, when using an MOI of 1:10 (data not shown).
Subsequent immunoblotting for SseJ-2HA from wild-type S. serovar Typhimurium/p2777 isolated from RAW264.7 cells 8 h postinfection demonstrated a clear signal for SseJ-2HA (Fig. (Fig.3B).3B). Significantly, no SseJ-2HA signal could be retrieved from the same amount of trxA mutant/p2777 bacteria isolated from RAW264.7 cells (Fig. (Fig.3B).3B). This strongly implied that TrxA was needed for induction of SseJ in RAW264.7 cells. Complementation with pFA3 restored the ability to express the SseJ-2HA fusion protein in the trxA mutant (Fig. (Fig.3B).3B). We also noted that pFA8 did mediate a small restoration of SseJ-2HA expression in RAW264.7 cells infected with a trxA mutant of S. serovar Typhimurium/p2777 (Fig. (Fig.3B3B).
The same set of strains showed similar SseJ-2HA staining results in infected MDCK cells 8 h postinfection, but with one significant deviation: S. serovar Typhimurium trxA mutant strain complementation with pFA8 resulted both in a wild-type intracellular growth pattern and in prominent staining for secreted SseJ-2HA (Fig. (Fig.3C).3C). These data proved that secretion of SseJ in the macrophage-like RAW264.7 cell line is strictly dependent on redox-active TrxA, whereas the noncatalytic variant of TrxA significantly contributes to SseJ secretion and intracellular replication in MDCK cells.
We proceeded by analyzing the abilities of the complemented trxA mutant bacteria to replicate in RAW264.7 and MDCK cells. For this, we used an MOI of 10:1 and allowed the infections to proceed for 16 h. As expected, the trxA mutant showed a severely reduced ability to replicate in RAW264.7 cells (Fig. (Fig.4A).4A). When RAW264.7 cells were infected with a trxA mutant complemented with pFA3 (whichcodes for wild-type TrxA), we recorded an almost complete restoration of intracellular replication (Fig. (Fig.4A).4A). In contrast, trxA mutant complementation with pFA8 (which codes for a catalytically inactive TrxA) or with the vector control pBAD33 gave results similar to those seen with the noncomplemented mutant in RAW264.7 cells (Fig. (Fig.4A4A).
The trxA mutant was also defective for replication in MDCK cells (Fig. (Fig.4B).4B). Both pFA3 and pFA8, but not pBAD33, increased the replication of the trxA mutant (Fig. (Fig.4B).4B). Thus, restoration of intracellular replication by the trxA plasmid constructs paralleled the complementation of SseJ-2HA secretion.
The series of experiments described above strongly suggested that the intracellular growth defect of the trxA mutant was related to a lack of SPI2 activity. Hence, if TrxA and SPI2 contribute to intracellular replication through a shared pathway, then a trxA mutant, a SPI2 T3SS mutant, and a trxA/SPI2 T3SS double mutant should show equal levels of attenuation in growth, whereas if trxA and SPI2 act through divergent pathways, one might expect an additive reduction in intracellular fitness for a trxA/SPI2 T3SS double mutant. This led us to compare the decreases in intracellular growth shown by the trxA mutant, an ssaV mutant defective in the SPI2 T3SS apparatus, and a trxA/ssaV double mutant of S. serovar Typhimurium. Growth measurements were conducted using epitheloid MDCK cells (Fig. (Fig.5A)5A) and the macrophage-like J774-A.1 (Fig. (Fig.5B)5B) and RAW264.7 (Fig. (Fig.5C)5C) cell lines commonly used for assaying Salmonella intracellular fitness and replication.
In this set of experiments, all three mutant strains revealed decreased intracellular growth yields in comparison to the wild-type S. serovar Typhimurium strain (Fig. (Fig.5).5). Significantly, the trxA and ssaV mutants showed equal decreases in intracellular replication, without any apparent additive attenuation in growth or fitness revealed by the trxA/ssaV double mutant.
The results presented in Fig. Fig.55 are consistent with an assumption that TrxA and SsaV contribute to intracellular fitness of S. serovar Typhimurium through a convergent pathway. We therefore considered the possibility that TrxA itself could act as an SPI2-secreted effector protein. However, we could not detect TrxA by immunoblotting in experiments performed with the SPI2-secreted protein fraction (Fig. (Fig.6).6). This in turn suggested that TrxA contributes to the functionality of the SPI2 T3SS, rather than acting as a secreted SPI2 effector.
To compare the contributions of TrxA and SPI2 with respect to virulence, BALB/c mice were challenged orally with wild-type, trxA, or ssaV mutant strains of S. serovar Typhimurium. Cohorts of mice were sacrificed at 1, 2, and 3 days postinfection with wild-type bacteria as well as with trxA or ssaV mutants. Bacteria from mesenteric lymph nodes (mLNs), Peyer's patches (PPs), livers, and spleens were enumerated. In this analysis, the trxA and ssaV mutants showed comparable strong levels of attenuation with respect to colonization of the mLNs, livers, and spleens (Fig. (Fig.7).7). While both mutants colonized PPs to a significant level, neither mutant underwent replication in PPs (Fig. (Fig.77).
As the levels of trxA and ssaV mutants detected in the mLNs, PPs, liver, and spleen remained very low, we compared the contributions to virulence of trxA and ssaV in competition experiments. Such competition experiments provide a more sensitive assay for measuring differences in net in vivo growth (5). For this we carried out infections using BALB/c mice and mixtures of mutant strains in a 1:1 ratio. The infection was given intraperitoneally to reduce variation and to increase the bacterial load in the liver and spleen. At 1 and 3 days postinfection, cohorts of mice were sacrificed and the proportions of the mutants from the liver and spleen were enumerated.
In these experiments, the competitive indices for the trxA/ssaV double mutant and the trxA mutant were slightly but repeatedly decreased when measured in mixtures with the ssaV mutant (Fig. (Fig.8).8). In contrast, the trxA/ssaV and trxA mutants showed almost equal competitive indices (Fig. (Fig.8).8). These results showed that TrxA provided additional fitness in a SPI2-deficient background. Yet the level of fitness added by TrxA in the SPI2-deficient background remained small in relation to the drastic decrease in virulence of these mutants (Fig. (Fig.7)7) and cannot alone account for the strong attenuation of the trxA mutant (Fig. (Fig.77).
In the next set of experiments, the trxA mutant carrying plasmid pFA3 coding for wild-type TrxA or pFA8 coding for the engineered catalytically inactive form of TrxA was used in competition experiments with BALB/c mice with a trxA::Km mutant carrying pBAD33. Cohorts of mice were infected intraperitoneally and sacrificed 1 and 3 days postinfection, whereupon the proportions of the two competing strains collected from the liver and spleen were measured. If both catalytic and noncatalytic TrxA contribute to visceral replication, then the level of the pBAD33-complemented strain should decrease in proportion. If the noncatalytic TrxA does not contribute to visceral replication, the ratio of the two competing strains should remain the same.
The trxA mutant complemented with wild-type TrxA strongly outcompeted the vector-complemented control strain in both the liver and the spleen (Fig. (Fig.9).9). Interestingly, a trxA mutant complemented with a noncatalytic TrxA also outcompeted the vector-complemented strain, albeit to a lesser degree than the mutant complemented with pFA3 (Fig. (Fig.99).
As TrxA is connected to SPI2 activity and virulence, we continued by examining whether these activities of TrxA were restricted to the SPI2 T3SS or whether the SPI1 or T3SS-related flagellar systems also respond to TrxA activity.
The functionality of the SPI1 T3SS can be determined through monitoring SPI1 effector protein secretion in vitro and invasiveness of S. serovar Typhimurium bacteria in cell culture-based infection assays (14, 39, 43, 49). To probe for a possible role of TrxA in SPI1 activity, we set out to isolate and analyze SPI1 effector proteins secreted by wild-type S. serovar Typhimurium and its isogenic trxA mutant under SPI1-inducing growth conditions. When isolated effector proteins were analyzed on SDS-PAGE gels after protein staining, we did not note any apparent differences in the SPI1 effector protein profiles produced by wild-type and trxA strains (Fig. 10A). This was in contrast to the hilA mutant, which lacks expression of a key SPI1 gene activator and whose results revealed flagellin alone in the secreted protein fraction (Fig. 10A). The trxA mutant strain also retained the ability to produce flagellin (Fig. 10A).
To reveal any functional defect in the SP1 T3SS, we conducted invasion assays by infecting MDCK cells with wild-type and trxA mutant strains in a gentamicin protection assay (49). Whereas in this assay the trxA mutant showed a minor reduction in invasion capability in comparison to the wild-type strain, this difference was not statistically significant (Fig. 10B). To probe whether the noted difference became amplified by oxidative stress, invasive cultures of the wild type and the trxA mutant were exposed to sublethal concentrations of hydrogen peroxide prior the invasion experiment. As shown in Fig. 10B, sublethal hydrogen peroxide concentrations had a minimal effect on the invasiveness of the wild-type or trxA strains. However, these alterations were not statistically significant.
This series of observations indicated that the effect of TrxA on the S. serovar Typhimurium SPI1 activity is minor and that TrxA does not contribute to expression of the flagella (Fig. 10A) that are assembled through a T3SS-related protein transport system (37).
As the innate defense responses of higher vertebrates include the production of reactive oxygen radicals, TrxA has been suggested to act as a bacterial oxidoprotectant and virulence factor (13, 51). We recently tested this idea by constructing mutants defective in the various parts of the thioredoxin pathway in the facultative intracellular pathogen S. serovar Typhimurium and concluded that trxA mutants are attenuated in both cell culture and murine infection models (8). Here we show that TrxA contributes to S. serovar Typhimurium virulence by determining the functionality and expression of the S. serovar Typhimurium virulence-associated SPI2 T3SS.
Under in vitro SPI2-inducing conditions, the trxA mutant revealed delayed induction of the ssaG SPI2 T3SS apparatus gene and a lack of secretion of the SseJ SPI2 effector protein. Upon infection of cultured cells, the trxA mutant failed to induce expression of the SseJ protein. As we could not detect TrxA among the secreted SPI2 effector proteins, we assumed that under in vitro conditions in cell culture infections, TrxA is required for the functionality of SPI2 rather than acting as a SPI2-secreted effector protein. By using either a trxA mutant or an SPI2 mutant, we demonstrated that TrxA and SPI2 contributed equally to bacterial intracellular replication and that a trxA SPI2 double mutant did not show further attenuation with respect to intracellular replication. While these observations are consistent with the idea of a direct convergent contribution of TrxA and SPI2 to intracellular replication of S. serovar Typhimurium, we cannot at this stage exclude the possibility of alternative more complex interactions. The visceral spread of the ssaV mutant (defective in SPI2 T3SS activity) and that of the trxA mutant were severely and similarly reduced in the salmonella-susceptible BALB/c mouse line. Combined with the results obtained with the investigations of cell cultures, these observations suggest that the attenuation caused by the trxA mutation mechanistically connects to a defect in SPI2 activity.
Our observations therefore gave rise to three mechanistic hypotheses connecting TrxA with SPI2 activity. The first hypothesis, implicating a TrxA-supported disulphide-bond isomerization pathway, could be discarded based on genetic evidence showing that dsbC mutants did not express a trxA mutant phenotype (8). The second hypothesis assumed that the trxA mutant phenotype was linked to abrogation of the electron transport to methionine sulfoxide reductase (MsrA), a well-known virulence factor in bacteria (47). Yet this hypothesis was also discarded, as an msrA mutant lacked a trxA mutant phenotype in S. serovar Typhimurium (Horst et al., unpublished observations). Furthermore, contrary to expectations, S. serovar Typhimurium trxA mutants were not sensitized to in vitro nitrosative or oxidative stress (8).
As the periplasmic thioredoxin-related DsbA oxidoreductase has been implicated in T3SS activity and virulence of S. serovar Typhimurium (39), an alternative possibility would be that TrxA in S. serovar Typhimurium more directly lodges on DsbA to ensure SPI2 activity. Yet DsbA and TrxA seem to diverge with respect to their contributions to virulence. First, DsbA strongly affects the in vitro expression of invasion-associated SPI1, SPI2, and the T3SS-related flagella (39), whereas TrxA selectively affected the SPI2 T3SS. Second, a dsbA mutant showed a 10-fold reduction in virulence, as assayed by growth competition in BALB/c mouse studies (39) whereas our trxA mutant revealed 1,000-fold attenuation in an identical virulence assay. Third, DsbA acts as a periplasmic oxidase, being reoxidized by DsbB rather than reduced by TrxA (29). Therefore, it is unlikely that the contribution of TrxA to virulence would directly relate to the presence of DsbA and thus to a periplasmic assembly component of the SPI2 T3SS. Rather, our results are consistent with TrxA assisting the very activity or induction of SPI2 under conditions defined by low pH and low concentration of Mg++ and phosphate or when the bacteria reside inside host cells.
We demonstrated that the in vivo replication defect of the trxA mutant could be complemented with wild-type TrxA and, intriguingly, to a measurable extent with a trxA allele coding for a catalytically inactive TrxA. Catalytically inactive TrxA also partially improved intracellular SPI2 activity and replication of the trxA mutant in an epitheloid cell line but not in macrophage-like cells.
The explanation for the differences in TrxA dependency of different cell types may come from distinct bacterial growth demands (7, 19) and gene expression profiles of S. serovar Typhimurium in phagocytic versus nonphagocytic cells (24). Yet even in epitheloid cells, TrxA and SPI2 showed similar and apparent convergent contributions to intracellular replication. As TrxA associates with cytoplasmic proteins, both through catalytic intermolecular disulfide formations and noncatalytic chaperone interactions, it remains possible that, in fact, both activities of TrxA support SPI2 and that the contribution of the redox activity is less prominent in epithelial cells. Such a dual contribution of TrxA to SPI2 activity could also explain why the partial SseJ SPI2 effector protein expression evoked by a catalytically inactive TrxA did not translate to bacterial replication or effector protein secretion in phagocytic cells despite the fact that the same TrxA variant did increase SPI2 activity and bacterial intracellular replication in epithelial cells. Finally, our findings do not exclude functional interactions between TrxA and additional virulence factors.
We are grateful to David W. Holden, Michael Hensel, Catherine A. Lee, and Isabelle Hautefort for providing genetic constructs. We thank Laura Plant for critically reading the manuscript.
This study was supported by a Vetenskapsrådet (Swedish Research Council) grant and a Lars Hiertas foundation grant.
Published ahead of print on 18 September 2009.