Role of Cross Talk in Regulating the Dynamic Expression of the Flagellar Salmonella Pathogenicity Island 1 and Type 1 Fimbrial Genes

Abstract

Salmonella enterica, a common food-borne pathogen, differentially regulates the expression of multiple genes during the infection cycle. These genes encode systems related to motility, adhesion, invasion, and intestinal persistence. Key among them is a type three secretion system (T3SS) encoded within Salmonella pathogenicity island 1 (SPI1). In addition to the SPI1 T3SS, other systems, including flagella and type 1 fimbriae, have been implicated in Salmonella pathogenesis. In this study, we investigated the dynamic expression of the flagellar, SPI1, and type 1 fimbrial genes. We demonstrate that these genes are expressed in a temporal hierarchy, beginning with the flagellar genes, followed by the SPI1 genes, and ending with the type 1 fimbrial genes. This hierarchy could mirror the roles of these three systems during the infection cycle. As multiple studies have shown that extensive regulatory cross talk exists between these three systems, we also tested how removing different regulatory links between them affects gene expression dynamics. These results indicate that cross talk is critical for regulating gene expression during transitional phases in the gene expression hierarchy. In addition, we identified a novel regulatory link between flagellar and type 1 fimbrial gene expression dynamics, where we found that the flagellar regulator, FliZ, represses type 1 fimbrial gene expression through the posttranscriptional regulation of FimZ. The significance of these results is that they provide the first systematic study of the effect of regulatory cross talk on the expression dynamics of flagellar, SPI1, and type 1 fimbrial genes.

Salmonella enterica causes a large number of diseases ranging from self-limiting gastroenteritis to life-threatening systemic infection (21, 62). Previous studies have identified multiple factors involved in Salmonella pathogenesis, including those related to motility, adhesion, invasion, and intestinal persistence (12, 18, 39, 43, 45, 55, 66, 83, 86-88). Key among them is a type 3 secretion system (T3SS) encoded within a 40-kb region of the chromosome called Salmonella pathogenicity island 1 (SPI1) (46-48, 51, 63, 76). The SPI1 T3SS functions as a molecular hypodermic needle, enabling Salmonella to inject proteins into host cells (11-13). These injected proteins both commandeer the actin cytoskeleton to facilitate the invasion of host cells and induce inflammatory diarrhea (27, 29, 33, 52, 59, 92).

In addition to the SPI1 T3SS, other systems, including flagella and type 1 fimbriae, have been implicated in Salmonella pathogenesis (31, 36, 75). Briefly, flagella are long helical filaments attached to rotary motors embedded within the membrane that enable the bacterium to swim in liquids and swarm over surfaces (9). Flagella are thought to facilitate invasion by enabling Salmonella to swim to sites of invasion (39, 75). In addition to motility, flagellin activates the expression of proinflammatory cytokines (26, 60, 61, 77, 84). Type 1 fimbriae, on the other hand, are hairlike appendages that carry adhesins specific for mannosylated glycoproteins on eukaryotic cell surfaces (2, 25, 54, 78). They are thought to be involved in pathogenesis by facilitating binding to intestinal epithelial cells (4, 24, 35, 40, 78). As with flagella, type 1 fimbriae do not appear to play a direct role in intestinal invasion but rather are thought to contribute to intestinal colonization and persistent infections (1, 19).

Multiple studies have shown that extensive regulatory cross talk exists between these three systems (5, 10, 22, 38, 53, 58, 72, 79). While the molecular details have been studied extensively (Fig. 1), the role and significance of these cross talk interactions are still relatively unknown. As all three systems play unique and potentially mutually exclusive roles during the infection cycle, we hypothesized that regulatory cross talk controls their dynamic expression. In particular, regulatory cross talk controls the timing of induction and the duration of gene expression for these three systems. To test the hypothesis, we monitored SPI1, flagellar, and type 1 fimbrial gene expression dynamics in a number of mutants where different regulatory links between these three systems had been selectively removed. Based on these results, we demonstrate that there is a natural hierarchy in the expression dynamics of the three, beginning with flagella, followed by the SPI1 T3SS, and ending with type 1 fimbriae. Our results also indicate that the regulatory cross talk between the three systems serves to tune the timing of gene expression with regard to their temporal activation and deactivation.

FIG. 1.

Coordinate regulation of the flagellar, SPI1, and type 1 fimbrial genes. The master regulator for flagellar gene expression is FlhD₄C₂ (9). The FlhD₄C₂ complex, in turn, activates two additional regulators, FliA and FliZ, encoded within the fliAZ operon. FliA is a flagellum-specific alternate sigma factor essential for the expression of the motor, filament, and chemotaxis genes. FliZ is a posttranslational activator of FlhD₄C₂ (71). FliZ also activates HilD (38, 41, 53, 57, 82) and represses FimZ posttranscriptionally (this study). In the SPI1 T3SS, HilD, HilC, and RtsA form three interlocking positive-feedback loops where all three activate each other's and their own expression (23). In addition, they can independently activate HilA expression. HilA is required for the expression of the genes encoding the SPI1 T3SS. SPI1 gene expression is negatively regulated by HilE, which binds to HilD and prevents it from activating the SPI1 promoters. RtsB, encoded within the same operon as RtsA, binds to the P_flhDC promoter and represses motility (22). In type 1 fimbriae, FimW and FimZ form a coupled feedback loop where they can activate their own and each other's expression (72). They can also independently activate the expression of the P_fimA promoter, which controls the expression of genes encoding type 1 fimbriae. FimY and FimW also participate in a negative-feedback loop, where FimY activates FimW expression and FimW represses FimY expression. FimZ also binds to the P_flhDC promoter and represses the expression of the flagellar genes (10) and induces the expression of HilE to repress SPI1 gene expression (5, 72).

MATERIALS AND METHODS

General techniques and growth conditions.

All experiments were performed in Luria-Bertani (LB) broth at 37°C unless noted otherwise. Antibiotics were added at the following concentrations: ampicillin at 100 μg/ml, chloramphenicol at 20 μg/ml, kanamycin at 40 μg/ml, and tetracycline at 15 μg/ml. All experiments involving growth of strains carrying the helper plasmid pKD46 or pCP20 were performed at 30°C as previously described (14). Loss of these temperature-sensitive plasmids was achieved by growth at 42°C. Removal of the antibiotic from the FRT-Cm/Kan-FRT insert was achieved by passing pCP20 through the isolated mutants (8). Enzymes were purchased from Fermentas or New England Biolabs and used according to the manufacturer's recommendations. Primers were purchased from IDT Inc.

Strain and plasmid construction.

Bacterial strains and plasmids are described in Tables 1 and 2, respectively. All S. enterica serovar Typhimurium strains are isogenic derivatives of strain 14028 (American Type Culture Collection). The S. enterica serovar Typhimurium generalized transducing phage P22 HT105/1 int-201 was used in all transductional crosses (15). Chromosomal mutations were introduced using standard λ Red recombination in cells carrying the helper plasmid pKD46 as described by Datsenko and Wanner (14). All mutations were checked using primers that target sequences outside the deleted region. Prior to removal of the antibiotic resistance marker, the constructs resulting from this procedure were moved into a clean wild-type background (14028) by P22 transduction.

TABLE 1.

List of strains used in this study

Strain	Genotype or characteristica	Source or referenceb
14028	Wild-type S. enterica serovar Typhimurium	ATCCc
JS481	Δ(invH-avrA)2916::Cm (called ΔSPI1)	20
CR201	ΔfliZ::FRT	71
CR222	ΔflhDC::FRT	71
CR312	ΔfimZ::Cm	72
CR322	ΔfimY::FRT	72
CR334	ΔfimYZ::FRT	72
CR314	ΔfimW::Cm	72
CR800	ΔrtsB::Cm
CR801	ΔrtsB::FRT
CR802	ΔfimZ::Cm ΔfliZ::FRT
CR803	ΔfimZ::FRT ΔfliZ::FRT
CR804	ΔPfimZ::tetRA
CR805	ΔPfimZ::tetRA ΔfimY::FRT
CR806	ΔPfimZ::tetRA ΔfimYZ::FRT
CR807	ΔrtsB::Cm ΔfimZ::FRT
CR808	ΔrtsB::FRT ΔfimZ::FRT
CR809	ΔfimYZ::FRT ΔfimW::FRT

^aAll Salmonella strains are isogenic derivatives of S. enterica serovar Typhimurium strain 14028.

^bStrains are from this study unless specified otherwise.

^cATCC, American Type Culture Collection.

TABLE 2.

Plasmids used in this study

Plasmid	Characteristic	Reference or sourcea
pKD46	bla PBADgam beto exo pSC101 oriTS	14
pKD3	bla FRT cm FRT oriR6K	14
pCP20	bla cat cI857 λPRflp pSC101 oriTS	8
pSS009	kan luxCDABE ori p15a	71
pSS010	kan PflgA-luxCDABE ori p15a	71
pSS077	kan PhilA-luxCDABE ori p15a
pSS222	kan PfimA-luxCDABE ori p15a
pPROTet.E	cm PLtetO-1ori ColE1	Stratagene
pSS013 (pFliZ)	cm PLtetO-1fliZ ori ColE1	71
pRtsB	cm PLtetO-1rtsB ori ColE1
pFimZ	cm PLtetO-1fimZ ori ColE1
pPROBE-Venus	kan venus ori p15a	73
PfimA-Venus	kan PfimA-venus ori p15a	73

^aPlasmids were made in this study unless specified otherwise.

Standard “scarred” FLP recombination target (FRT) mutants were produced as previously described, using pKD3 as the PCR template (14). Strain CR800 (ΔrtsB::Cm) was made using primers SS217F (TTTT AGC GTT TTT ATC TTC CTC TCG TCA TCA ATA TGT TAA GTG TAG GCT GGA GCT GCT TC) and SS217R (AGT TGC CTT GCC TAC CAC TCT ACC AAC ATT TTA GGA AAA ACA TAT GAA TAT CCT CCT TAG). The antibiotic marker was removed by passing pCP20 through strain CR800, resulting in strain CR801 (ΔrtsB::FRT).

Strain CR803 (ΔfimZ::FRT ΔfliZ::FRT) was constructed by transducing the antibiotic resistance marker from CR312 to CR201 using P22 transduction (resulting in strain CR802) and then removing the resistance marker using pCP20. Strain CR804 (P_fimZ::tetRA) was constructed by replacing the native P_fimZ promoter with a tetRA element from transposon Tn10. The tetRA element was amplified using primers SS178F (CCA TTA AAT GTA AAT ATT TCA CAT AAA ATT AAT ATT TAC AAG AGT AGG GAA CTG CCA) and SS178R (CTG TTA TGC GTC CTT CGT TTT ATA ATA AGC GTC AGA CAC CCT AAG CAC TTG TCT CCT), with TH8094 as the template (44). The resulting strain, CR804 (P_fimZ::tetRA), places FimZ under the control of a tetracycline-inducible promoter. Strains CR805 (ΔP_fimZ::tetRA ΔfimY::FRT) and CR806 (ΔP_fimZ::tetRA ΔfimYZ::FRT) were constructed by transducing the tetRA element from CR804 into strains CR322 and CR334, respectively.

Strain CR808 (ΔrtsB::FRT ΔfliZ::FRT) was made by first transducing the chloramphenicol resistance marker from strain CR800 (ΔrtsB::Cm) to CR201 (ΔfliZ::FRT) and then removing the antibiotic resistance marker from the resulting strain, CR807 (ΔrtsB::Cm ΔfliZ::FRT). Strain CR809 (ΔfimYZ::FRT ΔfimW::FRT) was made by first transducing the chloramphenicol resistance marker from strain CR314 (ΔfimW::Cm) into strain CR334 (ΔfimYZ::FRT), resulting in the ΔfimYZ::FRT ΔfimW::Cm strain, and then removing the resistance marker from the ΔfimYZ::FRT ΔfimW::Cm strain by using pCP20.

Transcriptional fusions of the promoters of interest were made in the following manner. The P_hilA promoter was amplified using primers PhilAF (GGG GGA TCC ACT TGT CAT CGC TAT GAT GA) and PhilAR (GGG GAA TTC ACA GGA TTA AAA TGT GGC AT). The P_fimA promoter was amplified using primers SS104F (TTT GGT ACC AAA TCT GTG AGG CCG GAT TG) and SS104R (GGG GAA TTC GTA GAG GTC ATT AAT TTA TG). The PCR fragments from both were digested with KpnI and EcoRI (underlined sequences) and cloned into the multiple-cloning site of plasmid pSS009. The resulting plasmids were called pSS077 and pSS222, respectively.

Expression vectors for RtsB and FimZ were constructed by PCR amplifying the gene of interest and cloning it in the multiple-cloning site of the vector, pPROTet.E. The resulting arrangement put the gene under a constitutively active P_LtetO-1 promoter. The rtsB gene was amplified using primers SS198F (GGC GAA TTC TTA TAA GGA GGA AAA ACA TTT GAG ATA TCT GAC AAT GCA) and SS198R (TTT GGT ACC TTA CGT AAT ATC GAC TGA TA). The fimZ gene was amplified using primers SS106F (GGG GAA TTC TAA CAG TCT GAG GCA TAC AA) and SS106R (TTT GGT ACC TTA CAA TAA TTC GTG TGA TT). The resulting plasmids were called pRtsB and pFimZ, respectively. Prior to transformations in wild-type and mutant strains, all constructs were verified by sequencing.

Fluorescence and luminescence assays.

Endpoint fluorescent measurements and dynamic luminescence measurements were made using a Tecan Safire2 microplate reader. For fluorescence endpoint measurements for type 1 fimbrial gene expression (P_fimA promoter activity), a 1-ml culture was grown at 37°C overnight and then subcultured 1:1,000 in fresh medium and grown under static conditions for 24 h at 37°C. One hundred microliters was then transferred to a 96-well microplate, and the relative fluorescence and optical density at 600 nm (OD₆₀₀) were measured. The fluorescence readings, given as relative fluorescence units (RFU), were normalized with the OD₆₀₀ to account for cell density. All endpoint experiments were done on three different days, with six repeats on each day. The average values and the standard deviations of the data are reported. Statistical analysis was performed using Student's t test, where the reported P values are based on the average fluorescence value for each separate day.

Dynamic luminescence experiments were performed as follows. Cells were grown overnight at 37°C in LB without salt and with vigorous shaking. The overnight culture was then subcultured 1:500 in fresh LB medium (with 1% salt). A 100-μl aliquot of the subcultured cells was then transferred to a 96-well plate and covered with a Breathe-Easy membrane to prevent evaporation. Luminescence and optical density readings were then taken every 20 min. All experiments were done in three independent repeats, with six samples in each experiment. The average values and the standard deviations of the data at each time point are reported.

RESULTS

Flagellar, SPI1, and type 1 fimbrial genes are expressed in a temporal hierarchy.

Our governing hypothesis is that regulatory cross talk controls the dynamic expression of flagellar, SPI1, and type 1 fimbrial genes. As a first step toward testing this hypothesis, we measured gene expression dynamics for the three systems in wild-type cells. In these experiments, we first grew the cells overnight in Luria-Bertani (LB) medium without salt and then subcultured them into fresh LB-1% NaCl medium in the absence of shaking, thus inducing a transition from SPI1-repressing to SPI1-inducing conditions. Growth under low-oxygen and high-salt conditions has previously been shown to induce SPI1 gene expression in vitro (3, 51). To measure gene expression dynamics, we employed plasmid-based transcriptional fusions of the P_flgA, P_hilA, and P_fimA promoters to the luciferase operon luxCDABE from Photorhabdus luminescens (71, 89). The P_flgA promoter controls the expression of the genes encoding the structural components of the flagellar P-ring protein (65) and thus provides a proxy measure for flagellar gene expression. The P_hilA promoter controls the expression of HilA, the master regulator for the genes encoding the SPI1 T3SS (56), and similarly provides a proxy for the expression of SPI1 genes (20). The P_fimA promoter controls the expression of the six genes encoding type 1 fimbriae (69, 70). The reason that we chose bacterial luciferase as opposed to other reporters for these experiments is that it is sensitive to dynamic changes in promoter activity (30, 71).

As shown in Fig. 2, we found that flagellar, SPI1, and type 1 fimbrial genes are expressed in a temporal hierarchy. Specifically, the cells first express flagellar genes, followed by SPI1 genes and, lastly, type 1 fimbrial genes. Hierarchical expression, however, is not entirely surprising as previous studies have shown that the growth phase plays an important role in the timing of activation of these three systems (7, 42, 67). In particular, the flagellar genes are maximally expressed during the early log phase, the SPI1 genes during the late exponential phase, and the type 1 fimbrial genes upon entry into stationary phase. The immediate question then is whether regulatory cross talk plays any role in dictating this transcriptional hierarchy.

FIG. 2.

Dynamic expression of the flagellar, SPI1, and type 1 fimbrial genes. Time course dynamics of the P_flgA (flagellar), P_hilA (SPI1), and P_fimA (type 1 fimbrial) promoter activities in wild-type cells as determined using luciferase transcriptional reporters. For reference purposes, the optical density (OD600) was plotted to illustrate how each system is activated during different phases of growth. In these experiments, cells were first grown overnight at 37°C in LB without salt and then subcultured 1:500 in fresh LB-1% salt. Cells were then grown statically with luminescence, and optical density readings were taken every 20 min. Average promoter activities from three independent experiments on separate days are reported. For each experiment, six samples were tested. Error bars denote standard deviations. A.U., arbitrary units.

Regulatory cross talk tunes gene expression dynamics.

To determine whether the transcriptional hierarchy is due to cross talk, we measured gene expression dynamics in mutants where each system was selectively inactivated. First, we measured how inactivating flagellar gene expression using the ΔflhDC mutant affected SPI1 and type 1 fimbrial gene expression. FlhD and FlhC form the FlhD₄C₂ complex, the master regulator of flagellar gene expression in enteric bacteria (37, 85). Deleting flhD and flhC shuts off flagellar gene expression (Fig. 3 A). As shown in Fig. 3B, SPI1 gene expression is significantly attenuated in the ΔflhDC mutant. However, the dynamics of SPI1 gene expression are unchanged; the times when SPI1 gene expression is first turned on and then off are the same in the ΔflhDC mutant as in the wild type. The only change in gene expression is that the magnitude is reduced. With regard to type 1 fimbriae, we found that gene expression was induced prematurely by roughly an hour in the ΔflhDC mutant compared to that in the wild type (Fig. 3C).

FIG. 3.

Effect of transcriptional cross talk on flagellar, SPI1, and type 1 fimbrial gene expression dynamics. (A to C) The flagellar genes amplify SPI1 gene expression and delay type 1 fimbrial gene expression. P_flgA (A), P_hilA (B), and P_fimA (C) promoter activities in the wild type and the ΔflhDC mutant are shown. (D to F) The SPI1 genes reduce the duration of flagellar gene expression and accelerate the induction of type 1 fimbrial gene expression. P_flgA (D), P_hilA (E), and P_fimA (F) promoter activities in the wild type and the ΔSPI1 mutant are shown. (G to I) Type 1 fimbrial genes do not affect flagellar gene expression and reduce the duration of SPI1 gene expression. P_flgA (G), P_hilA (H), and P_fimA (I) promoter activities in the wild type and the ΔfimYZW mutant are shown. Experiments were performed as described for Fig. 2. The mutants were also tested to see whether they affected growth. However, no change in optical density as a function of time was observed (data not shown).

Next, we tested the effect of inactivating SPI1 gene expression. To do this, we employed a mutant where the entire pathogenicity island was deleted (ΔSPI1) (20). Deleting SPI1 inactivates P_hilA promoter activity (Fig. 3E). The reason is that hilD is required for the transcription of hilA (20, 74). When we measured gene expression in the ΔSPI1 mutant, we found that the flagellar genes were expressed for an hour longer than they were in the wild type (Fig. 3D). In the case of the type 1 fimbrial genes (Fig. 3F), we found that their induction was delayed by 30 min in the ΔSPI1 mutant compared to that in the wild type.

Last, we tested the effect of repressing type 1 fimbrial gene expression. Here, we employed the ΔfimWYZ mutant. This mutant lacks the three regulatory proteins known to directly regulate type 1 fimbrial gene expression (72, 80, 81, 90). While P_fimA promoter activity is not completely repressed in this mutant (Fig. 3I), these three proteins are the ones that would most likely participate in any transcriptional cross talk. The other known genes in the type 1 fimbrial cluster encode the structural elements of type 1 fimbriae. When we measured gene expression dynamics in the ΔfimWYZ mutant, we observed no change in flagellar gene expression (Fig. 3G). However, in the case of the SPI1 genes (Fig. 3H), we found that they are expressed for an hour longer in the ΔfimWYZ mutant than in the wild type.

Collectively, these results demonstrate that the transcriptional hierarchy is not due to regulatory cross talk alone, as initially hypothesized, but rather is controlled predominantly by the growth phase and external regulatory factors. Regulatory cross talk, on the other hand, appears to tune gene expression dynamics. In particular, cross talk primarily regulates the transitions between the different phases of gene expression, with the notable exception being the requirement of flagellar gene expression for maximal SPI1 gene expression. Next, we tested the role of specific regulators in establishing this cross talk.

FliZ regulates the magnitude of SPI1 gene expression and the timing of type 1 fimbrial gene expression.

The flagellar regulator FliZ is a posttranslational activator of FlhD₄C₂ (71). In addition, FliZ is a positive regulator of SPI1 gene expression (38, 41, 53, 57, 82). While the specific mechanism is unknown, FliZ-dependent activation of SPI1 gene expression is known to occur through HilD (41, 53). To test the specific role of FliZ in regulatory cross talk, we measured flagellar, SPI1, and type 1 fimbrial gene expression dynamics in the wild type, a ΔfliZ mutant, and a ΔfliZ mutant constitutively expressing FliZ from a plasmid.

Comparing gene expression dynamics in these different strains, we found that FliZ is a positive regulator of flagellar and SPI1 gene expression and a negative regulator of type 1 fimbrial gene expression. In particular, we found that deleting FliZ decreases the relative magnitudes of flagellar and SPI1 gene expression but does not affect their dynamics (Fig. 4 A and B). Moreover, deleting FliZ accelerates the induction of type 1 fimbrial gene expression (Fig. 4C). We also found that overexpressing FliZ increases both the magnitudes and durations of flagellar and SPI1 gene expression (Fig. 4D and E), whereas it represses and delays type 1 fimbrial gene expression (Fig. 4F). Comparison of these results with those obtained using the ΔflhDC mutant suggests that the effect of flagellar gene expression on SPI1 and type 1 fimbrial gene expression is due to FliZ.

FIG. 4.

FliZ controls the magnitude of SPI1 gene expression and the dynamics of type 1 fimbrial gene expression. (A to C) Deleting FliZ represses flagellar and SPI1 gene expression and accelerates the induction of type 1 fimbrial genes. P_flgA (A), P_hilA (B), and P_fimA (C) promoter activities in the wild type and the ΔfliZ mutant are shown. (D to F) Overexpressing FliZ increases the magnitude of flagellar and SPI1 gene expression and delays the induction of type 1 fimbrial genes. P_flgA (D), P_hilA (E), and P_fimA (F) promoter activities in the wild type and the ΔfliZ mutant constitutively expressing FliZ from a P_LtetO-1 promoter on a plasmid (pFliZ) are shown. Experiments were performed as described for Fig. 2.

Repression of type 1 fimbrial gene expression by FliZ is through FimZ.

While FliZ has previously been shown to regulate SPI1 gene expression in a number of studies (38, 41, 53, 57, 82), its effect on type 1 fimbrial gene expression has not previously been reported to the best of our knowledge. Therefore, to further characterize this FliZ-dependent repression of type 1 fimbrial gene expression, we sought to determine the genetic target. To do this, we first compared P_fimA promoter activities in the wild type, a ΔfimZ mutant, a ΔfliZ mutant, a ΔfimZ ΔfliZ mutant, a ΔfliZ mutant constitutively expressing FliZ from a plasmid, and a ΔfimZ ΔfliZ mutant also containing the FliZ plasmid. We performed endpoint measurements after 24 h of growth. In these experiments, we employed promoter fusions to the fluorescent protein Venus on a plasmid that were identical to the fusions used for the luciferase reporters (32). The reason for using a fluorescent reporter rather than bacterial luciferase is that it is much more stable and thus provides a better measure of integrated promoter activity, as is desired in an endpoint measurement (32).

As shown in Fig. 5 A, deleting FliZ increases the activity of the P_fimA promoter (P < 0.01). Similarly when FliZ is expressed from a constitutive promoter on a plasmid in an otherwise ΔfliZ background, P_fimA promoter activity is repressed (P < 0.01). However, in the ΔfimZ mutant, we did not observe any additional change in promoter activity when FliZ was also removed (P > 0.20). When FliZ was expressed from a plasmid in an otherwise ΔfimZ ΔfliZ background, we observed a minor decrease in P_fimA promoter activity, though the effect was only marginally significant (P = 0.04). We also performed similar experiments with FimY and found that FliZ is still able to regulate the P_fimA promoter in the absence of FimY (data not shown). These results suggest that FliZ does not directly repress P_fimA promoter activity. We next tested whether FliZ represses type 1 fimbrial gene expression through FimZ. To test this hypothesis, we employed a strain where FimZ was constitutively expressed. Specifically, we replaced the P_fimZ promoter with the tetracycline-inducible tetRA element from transposon Tn10 (P_fimZ::tetRA). This arrangement decouples FimZ expression from its native regulation and causes it to be constitutively expressed from its native chromosomal locus in the presence of tetracycline. In addition, we also deleted FimY so that regulation was dependent entirely on FimZ (FimW is a negative regulator of type 1 fimbrial gene expression and operates through FimY [72]). When we tested the effect of FliZ in a P_fimZ::tetRA ΔfimY genetic background, we found that deleting FliZ increases P_fimA promoter activity (P < 0.01) whereas constitutively expressing FliZ represses it (P < 0.01) (Fig. 5B). We also performed similar experiments with FimY and found that FliZ has no effect (data not shown). These results indicate that FliZ regulates type 1 fimbrial gene expression through FimZ. One possibility is that FliZ posttranscriptionally regulates FimZ in a manner similar to FlhD₄C₂ and HilD. Alternatively, FliZ may prevent FimZ from activating the P_fimA promoter.

FIG. 5.

FliZ regulates type 1 fimbrial gene expression though FimZ. (A) FliZ is unable to regulate P_fimA promoter activity in the absence of FimZ. P_fimA promoter activities in the wild type, the ΔfimZ mutant, the ΔfliZ mutant, the ΔfimZ ΔfliZ mutant, the ΔfliZ mutant expressing FliZ from the constitutive P_LtetO-1 promoter on a plasmid (pFliZ), and the ΔfimZ ΔfliZ mutant harboring pFliZ are shown. (B) FliZ regulates FimZ posttranscriptionally. P_fimA promoter activities in the P_fimZ::tetRA ΔfimY mutant, the P_fimZ::tetRA ΔfimY ΔfliZ mutant, and the P_fimZ::tetRA ΔfimY ΔfliZ mutant expressing FliZ from the constitutive P_LtetO-1 promoter on a plasmid are shown. In the genetic background P_fimZ::tetRA, FimZ is under the control of a tetracycline-inducible promoter. Overnight cultures were subcultured 1:1,000 in fresh LB and then grown statically at 37°C for 24 h. FimZ expression was induced by addition of 15 μg/ml tetracycline. Fluorescence and optical density (OD600) values were then measured for each sample. Average promoter activities from three independent experiments on separate days are reported. For each experiment, six samples were tested. Error bars denote standard deviations.

RtsB directly regulates the dynamics of flagellar gene expression and indirectly regulates the dynamics of SPI1 and type 1 fimbrial gene expression through FliZ.

RtsB has previously been shown to bind to the P_flhDC promoter and repress flagellar gene expression (22). While RtsB does not reside within SPI1, it is located in the same operon as RtsA, an AraC-like regulator of SPI1 gene expression. This two-gene operon requires HilD for expression, and its expression is correlated with the other SPI1 genes (20, 22). To understand how RtsB contributes to transcriptional cross talk, we measured the P_flgA, P_hilA, and P_fimA promoter activities in the wild type, an ΔrtsB mutant, and an ΔrtsB mutant where RtsB was constitutively expressed from a plasmid. Comparing gene expression dynamics in these different mutants, we found that RtsB represses the dynamics of both flagellar and SPI1 gene expression and accelerates type 1 fimbrial gene expression. In the ΔrtsB mutant, we found that flagellar genes are expressed for roughly an hour longer (Fig. 6 A) and that the induction of type 1 fimbrial gene expression is weakly delayed (Fig. 6C). No change in SPI1 gene expression, however, was observed (Fig. 6B). When RtsB is constitutively expressed, we found that it completely inhibits flagellar gene expression (Fig. 6D). It also represses SPI1 gene expression (Fig. 6E) and accelerates the induction of type 1 fimbrial genes by roughly 1 h (Fig. 6F).

FIG. 6.

RtsB controls the dynamics of flagellar and type 1 fimbrial gene expression. (A to C) Deleting RtsB increases the duration of flagellar gene expression and slows the induction of type 1 fimbrial genes. P_flgA (A), P_hilA (B), and P_fimA (C) promoter activities in the wild type and the ΔrtsB mutant are shown. (D to F) Overexpressing RtsB inactivates flagellar gene expression and accelerates the induction of type 1 fimbrial genes. P_flgA (D), P_hilA (E), and P_fimA (F) promoter activities in the wild type and the ΔrtsB mutant constitutively expressing RtsB from the constitutive P_LtetO-1 promoter on a plasmid (pRtsB) are shown. Dynamic luminescence experiments were performed as described for Fig. 2. (G) RtsB regulates type 1 fimbrial gene expression through FliZ. P_fimA promoter activities in the wild type, the ΔrtsB mutant, the ΔfliZ mutant, the ΔfliZ ΔrtsB mutant, and the ΔrtsB and ΔfliZ ΔrtsB mutants constitutively expressing RtsB are shown. Endpoint fluorescence experiments were performed as described for Fig. 5.

As we observed similar dynamic responses in the ΔrtsB and ΔSPI1 mutants, we conclude that the effect of SPI1 genes on flagellar and type 1 fimbrial gene expression is most likely due to RtsB. To explain our overexpression experiments, we note the following. While RtsB has been shown to directly repress flagellar gene expression, it is not believed to directly regulate SPI1 gene expression (22). Most likely, its effect on SPI1 gene expression is through FliZ. In particular, when RtsB is constitutively expressed, SPI1 gene expression dynamics mirror those in the ΔflhDC mutant. Similarly, RtsB is likely also affecting type 1 fimbrial gene expression through FliZ. To prove this, we compared P_fimA promoter activities in the wild type, the ΔrtsB mutant, the ΔfliZ mutant, the ΔfliZ ΔrtsB mutant, and the ΔrtsB and ΔfliZ ΔrtsB mutants constitutively expressing RtsB, using endpoint measurements after 24 h of growth. Consistent with our hypothesis, we found that RtsB has no effect on type 1 fimbrial gene expression in the absence of FliZ (Fig. 6G). Specifically, we found no change in P_fimA promoter activity when RtsB was deleted (P > 0.8) or expressed on a plasmid (P > 0.95). However, we qualify this analysis by noting that RtsB itself has only a marginal statistical effect on P_fimA promoter activity as determined using endpoint measurements (P = 0.03).

FimZ regulates the dynamics of SPI1 gene expression and is the link between the type 1 fimbrial genes and the flagellar and SPI1 genes.

FimZ is a positive regulator of type 1 fimbrial gene expression (90, 91). FimZ has also been shown to bind the P_flhDC promoter and repress flagellar gene expression (10). In addition, it represses SPI1 gene expression by increasing the expression of HilE, a negative regulator of SPI1 gene expression (72). To determine the contribution of FimZ to regulatory cross talk, we compared gene expression dynamics in the wild type, the ΔfimZ mutant, and a strain where FimZ is constitutively expressed (ΔP_fimZ::tetRA).

In the ΔfimZ mutant, we found that flagellar gene expression is unchanged (Fig. 7 A), whereas the SPI1 genes were expressed for an additional 30 min relative to their expression in the wild type (Fig. 7B). We also found that type 1 fimbrial gene expression was reduced roughly 2-fold (Fig. 7C). When FimZ was constitutively expressed, however, we found that both the flagellar and SPI1 genes were repressed (Fig. 7D and E). Also, as expected, type 1 fimbrial gene expression was significantly accelerated when FimZ was constitutively expressed.

FIG. 7.

FimZ controls the dynamics of SPI1 gene expression. (A to C) Deleting FimZ increases the duration of SPI1 gene expression. P_flgA (A), P_hilA (B), and P_fimA (C) promoter activities in the wild type and the ΔfimZ mutant are shown. (D to F) Overexpressing FimZ represses both flagellar and SPI1 gene expression. P_flgA (D), P_hilA (E), and P_fimA (F) promoter activities in the wild type and the P_fimZ::tetRA mutant, where FimZ is under the control of a tetracycline-inducible promoter, are shown. FimZ expression was induced by addition of 15 μg/ml tetracycline. Experiments were performed as described for Fig. 2.

Comparison of these results with those obtained using the ΔfimWYZ mutant suggests that FimZ is the link between type 1 fimbrial genes and flagellar and SPI1 gene expression as similar dynamic behaviors were observed in the two mutants. Interestingly, we found that deleting FimZ had no significant effect on flagellar gene expression. This is probably due to the fact that, under physiological conditions, the type 1 fimbrial genes are expressed long after the flagellar genes no longer are. Only when FimZ is constitutively expressed is repression seen. Similarly, the repression of the SPI1 gene is likely due to a combination of increased HilE expression and reduced flagellar gene expression.

DISCUSSION

Salmonella enterica needs to coordinate the expression of a diverse number of cellular systems during the infection cycle (28, 31, 34, 36, 50, 63, 75). In this study, we investigated the dynamic regulation of three of these systems, namely, flagella, the SPI1 T3SS, and type 1 fimbriae. We were able to demonstrate that these three systems are expressed sequentially during in vitro growth. This hierarchy in gene expression could mirror the roles of these three systems during the infection cycle. According to this simplified model, Salmonella first needs to swim to the sites of invasion in the distal small intestine. Logically then, the flagellar genes are expressed first. Upon reaching its target sites for invasion, Salmonella stops synthesizing flagella as movement is no longer required and starts to synthesize the SPI1 T3SSs necessary for invasion. Once the bacterium enters the stationary phase, it stops synthesizing the SPI1 T3SSs and begins to express the type 1 fimbrial genes involved in intestinal colonization and persistence (1, 19, 49). In vivo, we imagine that this last step is necessary only in those bacteria that were unable to breach the intestinal epithelium.

As multiple studies have shown that extensive regulatory interactions exist between these three systems (5, 10, 22, 38, 72), we hypothesized that regulatory cross talk may govern the hierarchy. However, we found that transcriptional hierarchy is controlled predominately by external factors, external in the sense that the hierarchy is not due to known interactions among these three systems. Cross talk, rather, appears to tune the hierarchy. In particular, cross talk between these three systems is critical for regulating gene expression during transition phases in the hierarchy. The one exception is the effect of FliZ on SPI1 gene expression dynamics.

Among the regulators underlying this cross talk, we found that FliZ is the most significant as it regulates both SPI1 and type 1 fimbrial gene expression, where the latter is a novel finding of this study. Moreover, FliZ's effect on SPI1 gene expression is profound, reducing the expression roughly 3-fold. While FimZ also regulates flagellar and SPI1 gene expression, the effects are minor and are really seen only when the regulator is constitutively expressed. Interestingly, unlike RtsB and FimZ, FliZ does not appear to directly regulate transcription in these three systems. Rather, FliZ appears to function through another transcription factor, be it FlhD₄C₂ (71), HilD (41), or FimZ. While the underlying mechanism of action of FliZ is still unclear, our data imply that it is posttranslational in all three systems.

One outstanding question concerns the physiological role of this regulatory cross talk between the flagellar, SPI1, and type 1 fimbrial genes. In particular, cross talk has a relatively minor role in regulating the hierarchical expression of these three systems, at least under the conditions investigated in this study. As a comparison, the most well-characterized example of regulatory cross talk involves the hierarchical expression of carbohydrate transport and metabolic genes during growth on mixed substrates. In that case, transcriptional cross talk is used to enforce a strict hierarchy in carbohydrate utilization (17, 64). While we also cannot discount that other factors associated with the flagellar, SPI1, and type 1 fimbrial systems are involved in regulating the transcriptional hierarchy, we expect that the hierarchy is not due to cross talk but rather is regulated by external factors. Specifically, based on our current understanding, we believe that these three systems are regulated in response to the growth phase of the cell (Fig. 2). Why then is cross talk employed?

If we consider the logic of this cross talk, then a simple pattern emerges (Fig. 8). Specifically, cross talk appears to make the expression of these three systems mutually exclusive, though only to the degree to which they are expressed during the infection cycle. In this regard, it helps to reinforce the transcriptional hierarchy. For example, the expression of the flagellar genes represses the expression of the type 1 fimbrial genes and vice versa. This mutual repression is perfectly logical when considering that Salmonella cannot move and adhere/persist at the same time. Similarly, the expression of the flagellar genes enhances SPI1 gene expression, whereas the expression of the SPI1 gene represses flagellar gene expression. Regulation in this case would suggest that only actively motile cells try to invade. As a corollary, persisting cells do not invade, consistent with the fact that the expression of the type 1 fimbrial genes represses the expression of the SPI1 genes. Lastly, once the cells decide to invade, logically then motility is no longer required.

FIG. 8.

Salmonella invasion program. (A) Diagram of transcriptional cross talk between the flagellar, SPI1, and type 1 fimbrial gene systems. (B) Inferred logic of transcriptional cross talk, where the decision to “move” results from flagellar gene expression, the decision to “invade” results from SPI1 gene expression, and the decision to “persist” results from type 1 fimbrial gene expression.

One limitation of this model is that it does not account for all the other systems involved in the infection cycle. For example, Salmonella has at least 13 distinct fimbrial systems (68) whose expression might also be coordinated with the expression of flagellar, SPI1, and type 1 fimbrial genes. Salmonella also possesses a nonfimbrial adhesin encoded in Salmonella pathogenicity island 4 (SPI4) (28). Previously, we demonstrated that the expression of the SPI4 adhesin is regulated by SprB, a SPI1-encoded regulator (73). Aside from fimbriae and adhesins, Salmonella also possesses a second T3SS encoded within Salmonella pathogenicity island 2 (SPI2) (34). The SPI2 T3SS is used to survive and replicate within host cells during systemic phases of infection. The SPI2 genes also play a role in inducing intestinal inflammation and are known to be regulated by HilD, a SPI1 regulator (7). However, chemical and environmental cues are required to activate SPI2 gene expression, most notably low Mg²⁺ concentrations (16) and acidic pH (6); HilD is not required for SPI2 gene expression during systemic infection (20). How the cells coordinate the expression of SPI1 and SPI2 genes is still not well understood.

Clearly, this model of the coordinate expression of Salmonella virulence genes (Fig. 8) is still incomplete, as it considers only a small subset of the systems involved in the infection cycle. In addition, it is based on just one mode of growth. Likely, cross talk is more significant when growth is irregular and the environment is variable, as opposed to our in vitro experiments where growth is uninterrupted and the environment is fixed. Further investigations are necessary to fully characterize the role of regulatory cross talk in coordinating gene expression during invasion. The significance of this study is that it is the first to systemically study the effect of regulatory cross talk on the expression dynamics of flagellar, SPI1, and type 1 fimbrial genes.