The Flagellar Regulator fliT Represses Salmonella Pathogenicity Island 1 through flhDC and fliZ

Salmonella pathogenicity island 1 (SPI1), comprising a type III section system that translocates effector proteins into host cells, is essential for the enteric pathogen Salmonella to penetrate the intestinal epithelium and subsequently to cause disease. Using random transposon mutagenesis, we found that a Tn10 disruption in the flagellar fliDST operon induced SPI1 expression when the strain was grown under conditions designed to repress SPI1, by mimicking the environment of the large intestine through the use of the intestinal fatty acid butyrate. Our genetic studies showed that only fliT within this operon was required for this effect, and that exogenous over-expression of fliT alone significantly reduced the expression of SPI1 genes, including the invasion regulator hilA and the sipBCDA operon, encoding type III section system effector proteins, and Salmonella invasion of cultured epithelial cells. fliT has been known to inhibit the flagellar machinery through repression of the flagellar master regulator flhDC. We found that the repressive effect of fliT on invasion genes was completely abolished in the absence of flhDC or fliZ, the latter previously shown to induce SPI1, indicating that this regulatory pathway is required for invasion control by fliT. Although this flhDC-fliZ pathway was necessary for fliT to negatively control invasion genes, fliZ was not essential for the repressive effect of fliT on motility, placing fliT high in the regulatory cascade for both invasion and motility.


Introduction
Salmonella is an important bacterial pathogen that is a leading source of food-borne illness, causing diseases ranging from transient enteritis to life-threatening septicemia. To infect its animal hosts, Salmonella first must penetrate the intestinal epithelium, a process termed invasion. Most of the genes required for invasion lie within a 40 kb gene cluster at centisome 63 termed Salmonella Pathogenicity Island 1 (SPI1), which is used by Salmonella to construct a type III secretion apparatus, the needle complex, to deliver secreted effector proteins into the host cell cytoplasm [1,2,3,4,5,6]. Once these proteins are translocated into a targeted epithelial cell, they induce cytoskeleton rearrangement and membrane ruffling, resulting in internalization of Salmonella by the host cell [7,8,9,10]. SPI1 genes are known to be controlled by several transcriptional regulators encoded within and outside SPI1 through a complex network. Four transcriptional regulators, hilD, hilC, hilA and invF are present within SPI1 [4,11,12,13,14]. Among these, hilD is at the top of the regulatory cascade and controls hilC as well as a regulator located outside SPI1, rtsA [15,16]. HilD, HilC, and RtsA are able to regulate their own gene expression and can activate expression of hilD, hilC and rtsA independent of each other to constitute a regulatory circuit for the control of the SPI1 central regulator hilA [17]. HilA controls the sic/sip operon, encoding effector proteins, and the prg/org and inv/spa operons that encode proteins composing the type III secretion apparatus [13,14]. HilA also induces the expression of the transcriptional regulator invF, encoding a member of the AraC family that activates the expression of genes encoding effector proteins within and outside SPI1 [13,14]. In addition, invF has been shown to be directly regulated by HilD and HilC through a HilA-independent pathway [18]. Several genetic regulators outside SPI1 have also been shown to transcriptionally or post-transcriptionally control invasion gene expression. Regulators affecting SPI1 at the level of transcription include the two-component regulators PhoP/PhoQ, EnvZ/ OmpR, PhoB/PhoR and BarA/SirA [1,19,20]. In addition, the DNA binding proteins H-NS and Hha have been demonstrated to bind to multiple A-T rich sequences in SPI1, occupying the binding sites of positive regulators, and consequently preventing transcription [21,22,23]. Among the post-transcriptional regulators of SPI1, the Csr system, PNPase, Lon protease and HilE have been shown to control invasion genes by affecting protein production or by manipulating the level or activity of HilD [24,25,26,27,28].
In addition to the mechanisms of control described above, two regulators of the flagellar regulon, flhDC and fliZ, have been described as inducers of SPI1 [19,29]. In the Salmonella flagellar regulatory cascade, the FlhD 4 C 2 complex, encoded by flhDC, functions as a master regulator that binds to the class 2 flagellar promoters and to its own promoter to induce downstream flagellar gene expression [30,31]. However, the function of FlhD 4 C 2 is antagonized by another flagellar protein, FliT, which associates with FlhD 4 C 2 and neutralizes its activity [32,33]. fliZ has been characterized as a class 2 flagellar gene [34]. Previous studies have shown that mutation of fliZ significantly reduces hilA expression and Salmonella intestinal colonization in mice. In addition, overexpression of fliZ increases hilA expression only when hilD is present, indicating that fliZ controls invasion gene expression through hilD [29]. Although fliZ has been demonstrated to negatively control hilD, the mechanism by which this is accomplished remains uncertain [29,35].
Expression of invasion genes can also be induced using various laboratory conditions that mimic the host intestinal environment, such as low oxygen, high osmolarity, and a near neutral pH [36,37,38]. In addition, short-chain fatty acids, produced by the intestinal microbiota through fermentative metabolic pathways, have been shown to play important roles in controlling Salmonella invasion [39,40]. Among these, butyrate, which exists in high concentration in the large intestine where salmonellosis rarely occurs, represses SPI1 gene expression [40,41].
To identify additional genetic elements involved in Salmonella invasion control, here we applied a transposon mutagenesis approach and identified a mutation in the flagellar gene fliT that affects the expression of SPI1 genes. As fliT was known to be a negative regulator of the flagellar regulon, we used genetic approaches to study the role of fliT and associated regulatory elements in the repression of invasion. Here, we demonstrate that fliT controls Salmonella invasion genes through flhDC and the flhDCregulated gene, fliZ.

Results
Identifying fliT, a novel negative regulator of Salmonella invasion, using random transposon mutagenesis screening Salmonella Pathogenicity Island 1 (SPI1) gene expression is controlled by various regulatory elements inside and outside the island, and is also affected by environmental cues [42]. To identify novel regulators that negatively control Salmonella invasion, we used random Tn10 transposon mutagenesis in a strain carrying a gfp reporter fusion to the SPI1 gene sipC, with the strain grown in the presence of butyrate, a short-chain fatty acid found in abundance within the mammalian intestine. As butyrate has been shown to repress SPI1 genes [40,41], the bacterial colonies carrying the sipC::gfp reporter showed little fluorescence on LB agar containing 10 mM butyric acid. We surmised that transposon insertions in negative regulators of invasion would increase sipC::gfp expression, producing fluorescent colonies. The strain used for this screen also carried a deletion of ackA, encoding acetyl kinase, as our studies showed that the ackA mutation partially restored sipC expression in media containing butyric acid ( Fig. 1 and data not shown). This strain allowed the screen to be performed without the repeated isolation of ackA mutants, and thus provided the possibility of identifying novel regulators of invasion. In total, we screened approximately 40,000 colonies, representing an 8-fold screening of the genome, with 31 fluorescent colonies being found. We next sought to determine the transposon insertion sites in candidate mutants. Previously, it had been reported that Tn10 insertions near the promoter region of the SPI1 regulator hilD could cause increased expression of the downstream regulator hilA, which is essential to induce sipC [43]. To rule out these and other potential mutations within SPI1, we examined the linkage of Tn10 insertions to sipC using P22 bacteriophage-mediated transductional mapping. The results showed that 22 candidates possessed a Tn10 insertion linked to sipC; these mutants were not further characterized.
For the remaining nine candidate colonies, the Tn10 insertions were moved by transduction into an ackA mutant carrying a MudJ insertion encoding a lacZY fusion to the sipBCDA operon to quantify the increase in invasion gene expression using bgalactosidase assays. Based upon the increased level of sipBCDA:: Figure 1. Mutation of fliT restores sipC invasion gene expression in the ackA mutant under SPI1-repressing conditions. Wild type and mutants strains carrying the sipC::lacZY fusion were grown in LB broth buffered to pH 6.7 with 100 mM MOPS overnight without aeration with no additive (black bars) or with 10 mM butyric acid (white bars). sipC::lacZY expression was measured using b-galactosidase assays. The value of individual bars represents means for samples tested in triplicate, and the error bars represent standard deviations. An asterisk (*) indicates a statistically significant difference due to butyrate as compared to the same strain with no additive at p,0.05. A plus (+) indicates a statistically significant difference due to deletion of a gene as compared to the wild type when grown under the same conditions at p,0.05. doi:10.1371/journal.pone.0034220.g001 lacZY expression, candidates were categorized into two classes; those with increased expression only when butyrate was present (six mutants), and those with increased expression under both repressing and inducing conditions (three mutants). As individual mutants in each group possessed a similar effect on sipBCDA expression, their phenotypes suggested that they might carry Tn10 disruptions in the same gene or operon. To identify the sites of transposon insertion, we amplified the region flanking the Tn10 for one candidate from each of the two groups by arbitrary PCR [44]. We found that the mutant affected only under repressing conditions, in the presence of butyrate, carried a Tn10 insertion in fliD, the first gene of the fliDST operon. A representative of the second class, showing increased sipC expression under both repressing and inducing conditions, carried a Tn10 insertion in pnp. Further, we determined the genetic linkage of Tn10 in all of the remaining candidates of both groups to fliD and pnp by transductional mapping, finding that all insertions within a group were 100% linked to these respective genes. These results, taken together, demonstrate that all of the mutations residing outside SPI1 that induced the expression of sipC under our tested conditions resulted from disruptions in or near either fliD or pnp.
fliT is a negative regulator of Salmonella invasion pnp, encoding a 3-to 5-phosphorolytic exonuclease, a subunit of RNA degradosome, has been shown to affect SPI1 genes expression by interfering with RNA half-life [28]. However, genes in the fliDST operon have not been reported to control invasion by Salmonella. For this reason, we focused our study on the role of the fliDST operon in the control of SPI1. To quantify the effects of the Tn10 disruption of fliD on invasion, we compared sipC::lacZY expression in various mutants grown with or without butyric acid by b-galactosidase assays. In the wild type, sipC expression decreased 3.5-fold when the strain was grown in media containing 10 mM butyric acid compared to media with no additive (with all media stably buffered to pH 6.7), and an ackA mutant, as expected, demonstrated a lesser, 1.5-fold repression due to the presence of butyric acid (Fig. 1). Importantly, sipC expression was unaffected by butyric acid in the ackA, fliD::Tn10 double mutant (Fig. 1). As fliD is the first gene in the fliDST operon, the increase of sipC expression caused by the disruption of fliD compared to the wild type grown under the same repressive conditions may have resulted from polar effects on any of the downstream genes in the operon. Thus, we next determined which genes played important roles in control of sipC expression by testing the effects of mutations of operon genes, singly and in combination. The results showed that ackA strains with an additional deletion of fliDST, fliST, or fliT restored sipC expression in the presence of butyric acid compared to the same strains without additive (Fig. 1). There remained, however, a significant decrease in sipC expression by butyrate in strains with disruptions of fliD (data not shown) or fliS (Fig. 1), the first two genes of the fliDST operon. From these results, we concluded that the last gene of the fliDST operon, fliT, is required for the negative control of SPI1 gene expression. The increased sipC expression caused by deletion of fliT was, however, seen only in the ackA mutant and with the repression of SPI1 genes provided by butyric acid (Fig. 1). In addition to the repressive effects that required fliT, we also found that the loss of some members of the operon in the ackA null strain reduced sipC expression irrespective of the media conditions employed ( Fig. 1; compare white bars to each other). This suggests that components of this operon under some conditions may exhibit a positive effect on invasion gene expression, but in this work we further examined only the genesis of invasion gene repression caused by these genes.
To confirm the negative effect of fliT on invasion, we cloned the fliT ORF onto a low-copy number plasmid, on which fliT was constitutively expressed under the control of an exogenous promoter. Again using a sipC::lacZY fusion, we found a significant 3.1-fold decrease in sipC expression in the wild type strain with the fliT plasmid compared to the isogenic strain carrying the control plasmid, pACYC177 (Fig. 2). This repressive effect of fliT on invasion gene expression, however, was limited to conditions of over-expression as a fliT mutant carrying a sipC::lacZY fusion in an otherwise wild type background and grown under SPI1-inducing conditions demonstrated no significant change in sipC expression (data not shown), identical to the phenotype of the fliD::Tn10 insertion shown in Figure 1. To verify that the repressive effect of fliT on gene expression manifested itself as a significant virulence phenotype, we next characterized changes in the levels of effector proteins of SPI1 produced and secreted by this strain. SPI1 invasion proteins encoded by the sipBCDA operon have been shown to be secreted into the culture medium when Salmonella is grown in laboratory media [45]. We extracted the secreted proteins from overnight bacterial cultures and examined the SPI1 effector protein profile using SDS-PAGE with Coomassie blue staining (Fig. 3). Four bands had molecular weights equivalent to the invasion proteins SipA (89 kDa), SipB (67 kDa), SipC (43 kDa) and SipD/InvJ (38 kDa) (protein sequences of these bands were determined by mass spectrometry, with the band for SipD overlapping that of another invasion protein, InvJ, due to their similar molecular weights). These bands were significantly diminished in the wild type strain carrying the fliT plasmid compared to the strain with the control plasmid ( Fig. 3, lanes 1 and 2). In addition, we examined the invasion of the wild type strain carrying the fliT plasmid using a gentamicin protection assay Figure 2. Over-expression of fliT negatively controls sipC invasion gene expression. The expression of sipC::lacZY was measured in strains carrying a low copy-number plasmid expressing fliT (pFliT) or its vector control (pACYC177). Bacterial strains were cultured in LB broth buffered to pH 6.7 with 100 mM MOPS overnight without aeration with no additive (black bars) or with 10 mM butyric acid (white bars). sipC::lacZY expression was measured using bgalactosidase assays. The value of individual bars represents means for samples tested in triplicate, and the error bars represent standard deviations. An asterisk (*) indicates a statistically significant difference due to butyrate as compared to the same strain with no additive at p,0.05. A plus (+) indicates a statistically significant difference due to pFliT as compared to the isogenic strain carrying the control plasmid pACYC177 at p,0.05. doi:10.1371/journal.pone.0034220.g002 with the HEp-2 epithelial cell line. We found that over-expression of fliT significantly reduced the ability of Salmonella to penetrate these cells, reducing invasion by 23-fold (Fig. 4). Based upon the results of b-galactosidase assay, the secreted protein profile assays, and this invasion assay, we thus demonstrated that fliT acts as a repressor of SPI1 gene expression and Salmonella invasion when it is over-expressed.
The fact that the loss of fliT in the ackA deletion mutant could relieve the butyrate-induced repressive effect on the SPI1 gene sipC (Fig. 1) led us to speculate that butyrate might function through the induction of fliT itself. To test this hypothesis, we used a fliT-lacZ transcriptional fusion in the wild type and the ackA mutant, and examined whether fliT expression was increased by butyrate. We found that there was no significant difference in fliT expression in either strain background with the addition of butyric acid (data not shown), indicating that fliT expression is not affected by butyrate at the transcriptional level. To further investigate whether the negative effects of fliT and butyrate on invasion genes were independent, we determined whether the addition of butyrate promoted the repressive effect on SPI1 when fliT was over-produced. We over-expressed fliT in the wild type and the ackA mutant strains carrying the fliT plasmid and compared sipC expression with or without the addition of butyric acid. As expected, we found that there was a significant further reduction of sipC expression by butyrate in these strains, 3.6-fold for the wild type and 3.3-fold for the ackA mutant (Fig. 2). These results therefore demonstrate that fliT is not involved in the negative control of butyrate on SPI1 gene expression.
fliT negatively controls invasion genes through the flagellar regulators flhDC and fliZ Having shown that over-expression of fliT from an exogenous promoter repressed invasion, we further asked how this member of the flagellar regulon exhibited this control. FliT has been shown to function as a chaperone to facilitate export of the flagella capping protein, FliD, in the assembly of flagella [46,47]. More importantly, FliT has also been demonstrated to negatively control flagellar gene expression by binding to the class 1 flagellar regulator, the FlhD 4 C 2 complex, and preventing this transcriptional activator from binding to class 2 flagellar promoters, consequently reducing downstream flagellar gene expression [32,33]. Since FliT can function as a negative regulator of the flagellar regulon, it is possible that the repressive effect of FliT on invasion may result from its negative effects on other flagellar genes that can positively control invasion gene expression. In Salmonella, two flagellar genes, flhDC and the flhDC-controlled downstream regulator fliZ, have been shown to positively regulate  carrying the fliT expression plasmid pFliT (white bars) or control plasmid pACYC177 (black bars) were grown overnight without aeration in LB with 100 mM MOPS, pH 6.7, and 100 mg/ml ampicillin. Invasion of HEp-2 cells by each strain (MOI<10) was assessed using a gentamicin protection assay. Invasion of all strains is shown relative to the wild type carrying the control plasmid, which was set to 100%. The value of individual bars represents means for samples tested in quadruplicate, and the error bars represent standard deviations. An asterisk (*) indicates a statistically significant difference due to the pFliT plasmid compared to the same strain with the control plasmid at p,0.05. A plus (+) indicates a statistically significant difference due to deletion of a gene as compared to the wild type at p,0.05. doi:10.1371/journal.pone.0034220.g004 SPI1 [19,29]. In addition, fliZ has been shown to regulate invasion genes through the control of the SPI1 regulator, HilD [29,35]. To test whether fliT affected Salmonella invasion through the negative control of this flhDC-fliZ pathway, we first examined the abilities of the flhDC and fliZ mutants, each carrying the fliT plasmid or the control plasmid pACYC177, to invade cultured HEp-2 epithelial cells using a gentamicin protection assay (Fig. 4). We found that there was a significant reduction in Salmonella invasion, 26-fold for the flhDC mutant and 22-fold for the fliZ mutant (Fig. 4). These results demonstrate that these two flagellar genes are positive regulators of invasion, and are consistent with results published by other groups [19,29]. We also found that there was no significant difference in invasion between the flhDC mutant carrying the fliT plasmid or the control plasmid. However, a 1.4-fold decrease in invasion was observed in the fliZ mutant carrying the fliT plasmid compared to the same strain carrying the control plasmid (Fig. 4). These results suggested that flhDC is required for fliT to control invasion, but fliZ may be dispensable. As flagella have been shown to affect Salmonella invasion [48], mutation of the master flagellar regulator flhDC might cause greater effects on flagella production than mutation of fliZ, and consequently affect the results of invasion assays. Therefore, to more precisely test whether fliT affected invasion genes through the negative control of this flhDC-fliZ pathway, we next examined sipC expression in the wild type, the flhDC mutant, and the fliZ mutant, each carrying the fliT plasmid or the control plasmid. Using the sipC::lacZY fusion, there was a significant reduction of sipC expression, in the flhDC mutant (4.6-fold) and the fliZ mutant (6.8-fold) (Fig. 5A). Additionally, over-expression of fliT did not further reduce sipC expression in the flhDC or the fliZ mutant (Fig. 5A). These results suggest that fliT negatively controls sipC through this recognized pathway of regulation. To confirm the negative effect of fliT on SPI1 genes through flhDC and fliZ, we further examined the secreted invasion protein profiles using culture conditions identical to those employed for the b-galactosidase assays. The result showed that the secreted invasion effector proteins SipA, SipB, SipC and SipD were significantly diminished in the flhDC and the fliZ mutants compared to the wild type (Fig. 3, lanes 1, 3 and 5). Additionally, there was no further reduction in these proteins in the flhDC or fliZ mutant carrying the fliT plasmid (Fig. 3, lanes 4 and 6). As we had shown that downstream SPI1 effector proteins were affected by over-expression of fliT, in parallel we also determined whether their upstream regulator, hilA, was affected. As for the previous bgalactosidase results using sipC, hilA expression was significantly reduced in the flhDC and fliZ mutants carrying the control plasmid compared to the wild type with the same plasmid. A 2.8-fold decreased in hilA expression was also observed due to the expression of fliT, and there was no additional decrease in hilA expression in the flhDC and fliZ mutants carrying the fliT plasmid compared to the same strains with the control plasmid (Fig. 5B). Based upon these results, we conclude that fliT negatively affects Salmonella invasion gene expression through flhDC and fliZ.
The flhDC-fliZ pathway is required for the repressive effects of fliT on invasion gene expression, but not for its effects on flagellar regulation Our results demonstrate that fliT acts as a negative regulator of invasion, and previous studies have shown that fliT affects flagellar control in Salmonella [32,33,49]. Additionally, our data suggest that the repressive effect of fliT on invasion genes is accomplished through the flhDC-fliZ pathway. Since flhDC and fliZ have been implicated as regulators in the flagellar regulon [50], we further asked whether this flhDC-fliZ route is used by fliT in its control of flagella. To test this, we used the wild type, the flhDC mutant, and the fliZ mutant carrying either the control plasmid or the fliT plasmid, and examined their swimming ability on 0.35% LB agar plates (Fig. 6). We found that in the wild type strain overexpression of fliT completely eliminated Salmonella motility. The same phenotype was also observed in the flhDC mutant whether it carried the control plasmid, pACYC177, or the fliT plasmid. As previously described, fliT is able to negatively control flagellar gene expression by post-translational regulation of FlhD 4 C 2 activity. Our results thus suggest that fliT controls Salmonella motility The fliT expression plasmid pFliT (white bars) and the control plasmid pACYC177 (black bars) were tested in the wild type, the flhDC mutant and the fliZ mutant carrying A) the sipC::lacZY fusion, and B) the hilA::lacZY fusion. Strains were cultured in LB broth with 100 mM MOPS, pH6.7, and 100 mg/ml ampicillin overnight without aeration and lacZY expression was measured using b-galactosidase assays. The value of individual bars represents means for samples tested in triplicate, and the error bars represent standard deviations. An asterisk (*) indicates a statistically significant difference due to the fliT expression plasmid pFliT as compared to the same strain with the control plasmid at p,0.05. A plus (+) indicates a statistically significant difference due to deletion of a gene as compared to the wild type at p,0.05. doi:10.1371/journal.pone.0034220.g005 through flhDC, and are consistent with other studies. However, unlike flhDC, the fliZ mutant showed only a slight reduction in swimming ability, and over-expression of fliT in this strain fully inhibited its motility, suggesting that the repression of motility by fliT was not mediated through fliZ. Therefore, our results, taken together, suggested that the flhDC-fliZ pathway is specific for repression of Salmonella invasion gene expression by fliT, but this pathway is not required for fliT repression of the flagellar regulon.

Discussion
For serovars of Salmonella, the genes of SPI1 are key elements that dictate the ability of the pathogen to penetrate the intestinal epithelium and cause further systemic infection. The control of SPI1 gene expression has been shown to be evoked by complex interrelated regulatory networks. In this work, using a random transposon mutagenesis strategy, we discovered that the flagellar regulator fliT, encoded within the fliDST operon, can negatively control SPI1 gene expression (Fig. 1). In addition, we showed that fliT over-expression reduced invasion gene expression (Figs. 2, 3, and 5) and Salmonella invasion (Fig. 4), and that this repressive effect on SPI1 functioned through the negative control of the flhDC-fliZ pathway (Figs. 3 and 5). fliZ has been shown to positively control invasion genes by regulating the SPI1 regulator HilD, which exists high in the regulatory hierarchy of this pathogenicity island [16,29]. It has been suggested that FliZ post-transcriptionally controls HilD through an unidentified mechanism, rather than affecting hilD expression at the level of transcription [29]. Kage and coworkers showed that HilD protein level, when expressed from a constitutive promoter, was significantly decreased by the deletion of fliZ. However, the half-life of HilD was not changed when fliZ was missing. Their studies thus suggest that fliZ controls HilD at the translational level [35]. In contrast, Chubiz and colleagues showed that HilD, when constitutively expressed from a single-copy chromosomal tetracycline-inducible promoter, was only slightly reduced in the fliZ null strain compared to that in the wild type. Additionally, they measured the stability of HilD in the fliZ mutant and in the wild type and showed that the stability of HilD was not significantly altered in this mutant. Therefore, they suggested that the mechanism by which fliZ regulates HilD is by post-translationally affecting HilD activity [29]. As we have shown that fliT negatively controls invasion through fliZ, we suggest that fliT negatively controls Salmonella invasion by changing the amount or activity of HilD, and subsequently affects expression of downstream invasion genes.
FliT has been shown to possess two functions in Salmonella, acting both as a regulator and a chaperone [46,49,51]. In its chaperone function, FliT directly interacts with several flagellar proteins, including FliD, FliI and FliJ, preventing their prematuration and aggregation within the bacterial cytoplasm and thus facilitating flagellar assembly [46,47]. As a regulator of flagellar expression, FliT binds to FlhC as part of the FlhD 4 C 2 complex and inhibits FlhD 4 C 2 from binding to its target promoters, consequently repressing downstream flagellar gene expression [32]. In our work, we showed that fliT, when deleted, was the only gene of the fliDST operon able to restore invasion gene expression under our test conditions. We would expect that if FliT had affected invasion genes through its role as a chaperone, by interacting with FliD, the deletion of fliD would similarly restore sipC expression, as did the fliT mutant (Fig. 1). However, restoration of sipC expression was not observed in the fliD mutant under the conditions used (data not shown), suggesting that FliT does not affect SPI1 gene expression through its function as a chaperone. Instead, our results indicate that fliT acts on invasion genes in its role as a regulator, as we have demonstrated that the flhDC-fliZ pathway with which FliT is known to interact is required for its negative control of invasion genes. In addition to the repressive effects that we identified for fliT, we also found that the loss of some members of the fliDST operon caused a reduction in sipC expression when tested in the ackA null mutant, a phenotype that was independent of the media conditions used (Fig. 1). This may suggest that different components of this operon can have opposing actions. Specifically, it is possible that fliT functions as a repressor, but that other products of the operon, alone or in combination with fliT, might act as inducers. Work in this area will require further efforts.
In agreement with previous studies [32,33], our work suggests that flhDC is also required for fliT to control Salmonella motility (Fig. 6). However, we found that fliZ was required for the effects of fliT only on SPI1 gene expression, and not on motility, as only a slight reduction in swimming was observed in the fliZ mutant (Fig. 6). These results, taken together, thus demonstrate that fliT is able to coordinately regulate invasion and flagellar gene expression through the single flagellar master regulator flhDC, but that the control of these two regulons diverges at subsequent steps in their regulatory cascades.
In Salmonella, flagella and invasion have been shown to be coordinately regulated by regulators in addition to fliT through the flhDC-fliZ pathway. ClpXP is an ATP-dependent protease and has been demonstrated to repress both flagellar and SPI1 gene expression [35]. ClpXP negatively controls the flagellar regulon by facilitating the degradation of the master flagellar regulators FlhD and FlhC and subsequently repressing downstream flagellar genes [52]. Kage and coworkers have demonstrated that the repressive effect of ClpXP on the flhDC-fliZ cascade is required for this protease to negatively control invasion genes [35]. TviA is another regulator that negatively co-regulates invasion and flagellar genes through this pathway. TviA is a regulator within Salmonella Pathogenicity Island 7(SPI7) unique to Salmonella serovar Typhi that does not exist in S. Typhimurium. This regulator has been shown to respond to stimulation by low osmolarity and also negatively controls both flagellar and invasion gene expression. TviA affects flagellar genes by repressing the transcription of flhDC. This inhibitory effect on flhDC has been suggested to consequently cause the reduction of invasion gene expression through the flhDC-fliZ pathway [53]. The above two regulators and FliT have thus been demonstrated to either transcriptionally or post-translationally affect the flagellar master regulator flhDC, while previous studies and the results we present here demonstrate that the flhDC-fliZ pathway is essential for these regulators to control invasion genes. Based on this evidence, we suggest that the flhDC-fliZ pathway is an important common route used by Salmonella to allow the flagellar regulon to coordinately control invasion gene expression.
As previously described, FliT has been shown to negatively control FlhD 4 C 2 activity by its interaction with FlhC and subsequently to inhibit the binding of FlhD 4 C 2 to target DNA [32]. Interestingly, Aldridge and coworkers showed that FliT was able to interact with FlhD 4 C 2 that has not bound to its target DNA leading to the dissociation of the FlhD 4 C 2 complex in vitro [33]. However, when FlhD 4 C 2 was pre-associated with its target DNA, this protein-DNA complex was insensitive to FliT [33]. Their studies suggest a means by which Salmonella can efficiently control the flagellar regulon in response to rapidly changing environments. When FliT is produced, it binds existing FlhD 4 C 2 , dissociating the FlhD 4 C 2 complex and resulting in a quick down-regulation of flagellar gene expression. When the level of FliT is low, however, the FlhD 4 C 2 complex associates with its target DNA and thus efficiently activates the flagellar regulon. A recent study has also shown that flagella and invasion are coordinately regulated in response to growth phase. Both are highly expressed in the early stages of growth in laboratory medium [54]. However, these two regulons were both repressed in late stationary phase, and alternatively fimbrial genes were highly expressed [54]. This phenomenon may be relevant to the control of Salmonella gene expression within the intestine of an animal host. Infecting bacteria first utilize flagella and express invasion genes to reach and invade the intestinal epithelium. For those unable to penetrate the epithelium, expression of fimbrial genes would allow bacteria to better colonize within the intestine [54]. Thus, although the environmental and genetic cues of the intestinal tract that elicit control of flagella and invasion, including that mediated by fliT, are not well known, the coordinated regulation of these two important functions is clearly essential to productive infection and disease.

Construction of mutant strains
Salmonella enterica serovar Typhimurium strain ATCC 14028S and isogenic mutants were used throughout this study, and are shown in Table 1. Gene deletions were made using the previously reported one-step inactivation method [55]. In brief, PCR reactions were performed to amplify the fragments containing the FRT sequences flanking the antibiotic resistance markers from plasmids pKD3 or pKD4 using primers carrying 40 bases of homologous sequence flanking the coding region of the target gene. The resulting PCR product was purified and transformed into a Salmonella strain carrying the plasmid pKD46, which expresses the Red l recombinase, allowing allelic exchange. The resulting deletion mutants were cultured at 42uC to remove the temperature-sensitive pKD46 plasmid, and the loss of the target gene was determined by PCR. The chromosomal sipC::gfp translational fusion was created using the above one-step gene exchange method. A promoterless gfp linked to a chloramphenicol resistance marker was amplified from the plasmid pZEP07 [56] with primers TGAGACGTTGATCGGCACGTAAGAGGTTC-CAACTTTCACCTGTAGGCTGGAGCTGCTTCG and TT-AAATCACACCCATGATGGCGTATAGATGACCTTTCAG-ACATATGAATATCCTCCTTAG, which encode DNA homologous to the regions immediately adjacent to the sipC open reading frame. The resulting PCR product was purified and treated with DpnI to remove the pZEP07 template and transformed into the Salmonella strain carrying pKD46 with selection on 25 mg/ml chloramphenicol to allow recombination of gfp, creating a translational fusion to sipC with an adjacent chloramphenicol cassette. To create the fliT expression plasmid (pFliT), a PCR product was produced that included a synthetic ribosome binding site, based upon that of lacZ, and the fliT ORF with an additional 175 bp 39 of the fliT sequence to include the predicted transcriptional termination site. This product was amplified using primers CCCATCGATCAATTTCACACAGGAAACAGCTA-TGACCTCAACCGTGGAGTTTATCAAC and TCCCCCGG-GGATATCATTCAGCCCATCAGCACG. The PCR product was then cloned into the unique ClaI and SmaI sites of pACYC177 to place fliT under the control of the kanamycin resistance gene (npt) promoter on this vector.

Tn10 random transposon mutagenesis screening
To create a random transposon Tn10 library, a wild type strain carrying the IPTG-inducible Tn10 plasmid pNK2883 [57] and an additional plasmid pMS421 [58] that expresses lacI q was used. The strain was grown overnight in LB broth with 100 mg/ml of ampicillin, 100 mg/ml of spectinomycin, and 20 mg/ml streptomycin at 37uC with shaking, and sub-cultured in the same medium to midlog phase. To induce transposon insertion, IPTG was added at a final concentration of 0.1 mM to the mid-log culture, which was grown for another 16 hours. The resulting random Tn10 insertion library was moved into the ackA mutant strain carrying the chromosomal sipC::gfp translational fusion by P22 phage transduction [59]. Transductants were plated on LB agar with 25 mg/ml tetracycline, 100 mM 3-(N-morpholino)propanesulfonic acid (MOPS) pH 6.7, 10 mM ethylene glycol tetraacetic acid (EGTA), and 10 mM butyric acid and incubated at 37uC overnight. The green florescence of individual colonies was determined using an OV100 Observation Intravital System (Olympus Corp., Tokyo, Japan).

Determining the DNA sequence flanking the Tn10 element
The sequences flanking the Tn10 insertions were identified using a method previously reported [44]. In brief, Tn10 insertion strains and a control strain (the isogenic strain without a Tn10 insertion) were grown overnight in LB broth with aeration. The overnight culture was diluted 100-fold with nuclease-free water, and bacteria were frozen and thawed three times to expose the genomic DNA. Five ml of the sample was used as a template to perform an initial PCR using primer AATTGCTGCTTATAA-CAGGCACTG in combination with arbitrary primers GGCCAGCGAGCTAACGAGACNNNNGTTGC, GGCCAG-CGAGCTAACGAGACNNNNGATAT, and GGCCAGCGA-GCTAACGAGACNNNNAGTAC with a cycle of 3 min at 95uC followed by 30 cycles of 30 sec denaturation at 95uC, 30 sec annealing at 38uC, 90 sec extension at 72uC and an additional cycle of 3 min final extension at 72uC. Five ml of this PCR reaction was next used as template to perform a second PCR using the primer set GGCCAGCGAGCTAACGAGAC and ACCTTTGGTCACCAAAGCTTT, beginning with a cycle of 3 min denaturation at 95uC followed by 30 cycles of 15 sec denaturation at 95uC, 30 sec annealing at 56uC, 90 sec extension at 72uC and a final cycle of 3 min extension at 72uC. PCR products were separated by electrophoresis on a 2% agarose gel. The DNA fragments produced from the Tn10 insertion mutants but not from the control strain were harvested from the gel. Purified DNA fragments were sequenced using the primer ACCTTTGGTCACCAAAGCTTT.

Bacteriophage-mediated transductional mapping
The bacteriophage P22 was used to map the location of genes by transduction [59]. The donor strain carrying the antibiotic marker inserted in the bacterial chromosome was grown in LB overnight at 37uC with aeration. Four hundred ml of culture was added to 2 ml of P22 phage broth containing 5610 6 pfu/ml of the phage. The mixture was grown overnight at 37uC with aeration, and 6 drops of chloroform was then added to make the phage lysate. The recipient strain, harboring a different antibiotic marker in the chromosome, was grown in LB to mid-log phase, and 10 ml of phage lysate prepared from the donor strain was added to 500 ml of mid-log culture and incubated for 30 minutes at 37uC. Then, 500 ml of 20 mM EGTA was added to the above mixture and incubated at 37uC for 1 hour. Transductants were selected on LB agar with 10 mM EGTA and the antibiotic to which the donor strain was resistant. Fifty resulting colonies were patched onto LB agar with or without the antibiotic to which the recipient strain was resistant to assess the genetic linkage between the two markers.

b-galactosidase assays
Triplicate cultures of tested bacterial strains were grown standing overnight at 37uC in LB broth buffered to pH 6.7 with 100 mM MOPS and with 10 mM butyric acid and 100 mg/ml ampicillin if needed. b-galactosidase activity was measured as described previously [60].

Secreted protein isolation and analysis
Strains were grown in LB with100 mM MOPS, pH 6.7, and 100 mg/ml ampicillin at 37uC with shaking at 60 rpm for 16 hours. Proteins secreted into the culture supernatant were prepared and analyzed as previously described [20].

HEp-2 cell invasion assays
HEp-2 cells were grown in 24 well plates to confluence (approximately 5610 5 cells) in RPMI-1640 with 10% fetal bovine serum. Bacteria were grown overnight as static cultures in LB with 100 mM MOPS, pH 6.7, and 100 mg/ml ampicillin at 37uC. Approximately 5610 6 bacteria were added to each well. Plates were centrifuged for 10 min at 8006g and incubated for 1 hour at 37uC. Medium was discarded, and the cells were washed three times with 0.5 ml PBS. One ml of cell culture media supplemented with 20 mg/ml gentamicin was added to each well, and the cells were incubated for 1 hour at 37uC to kill the extracellular bacteria. Medium was removed, and the cells were washed three times with 0.5 ml PBS. Then, 200 ml of 1% Triton X-100 in PBS was added to each well for 5 minutes to lyse the cells, and 800 ml of PBS was added to individual wells to produce a final volume to 1 ml. The bacterial titers of the lysate were determined by colony counts. Each bacterial culture was tested in quadruplicate wells.

Bacterial swimming activity
Strains were grown overnight in LB with 100 mg/ml ampicillin at 37uC with shaking at 200 rpm. Ten ml of overnight culture of each strain was dotted onto the LB swimming agar plates (containing 0.35% agar) with100 mg/ml ampicillin, and incubated at 37uC for 7 hours in a humidified incubator.

Statistical analysis
Results from b-galactosidase assays and invasion assays were analyzed using a one-way analysis of variance to determine if the mean of at least one strain or condition differed from any of the others. The Tukey-Kramer HSD multiple comparison test was then used to determine which means were statistically different. A p-value,0.05 was considered significant. Statistical analysis was performed using Jmp 9.0 software (SAS, Cary, NC).