|Home | About | Journals | Submit | Contact Us | Français|
Flagellar assembly proceeds in a sequential manner, beginning at the base and concluding with the filament. A critical aspect of assembly is that gene expression is coupled to assembly. When cells transition from a nonflagellated to a flagellated state, gene expression is sequential, reflecting the manner in which the flagellum is made. A key mechanism for establishing this temporal hierarchy is the σ28-FlgM checkpoint, which couples the expression of late flagellar (Pclass3) genes to the completion of the hook-basal body. In this work, we investigated the role of FliZ in coupling middle flagellar (Pclass2) gene expression to assembly in Salmonella enterica serovar Typhimurium. We demonstrate that FliZ is an FlhD4C2-dependent activator of Pclass2/middle gene expression. Our results suggest that FliZ regulates the concentration of FlhD4C2 posttranslationally. We also demonstrate that FliZ functions independently of the flagellum-specific sigma factor σ28 and the filament-cap chaperone/FlhD4C2 inhibitor FliT. Furthermore, we show that the previously described ability of σ28 to activate Pclass2/middle gene expression is, in fact, due to FliZ, as both are expressed from the same overlapping Pclass2 and Pclass3 promoters at the fliAZY locus. We conclude by discussing the role of FliZ regulation with respect to flagellar biosynthesis based on our characterization of gene expression and FliZ's role in swimming and swarming motility.
The bacterial flagellum is a rotary motor that enables cells to swim in liquid environments and drift along surfaces (5, 22). In Salmonella enterica serovar Typhimurium, over 50 genes divided among at least 17 operons are involved in motility (9). These genes encode not only the flagellar subunits and chemotaxis proteins but also a number of regulators that synchronize gene expression with the assembly process.
The flagellum consists of three structural elements: the basal body, the hook, and the filament (42). The basal body, embedded in the membrane, anchors the flagellum to the cell. It also houses the rotary motor necessary for swimming and the type III secretion apparatus, which is involved in assembly. The hook, a flexible joint, transmits torque produced by the motor to the filament, a rigid helical structure approximately 5 to 15 μm in length that functions as the propeller (61). In S. enterica serovar Typhimurium, there are approximately four to six flagella per cell (32). When the motors spin counterclockwise, the filaments form a helical bundle that propels the cell forward in a corkscrew-like manner.
Flagellar assembly proceeds in a sequential manner beginning at the base along the inner membrane and concluding with the filament (43). The type III secretion apparatus, located at the cytoplasmic interface of the flagellum, delivers the majority of the protein subunits through a central channel within the growing flagellar structure. The process concludes with the nucleation and elongation of the flagellar filament, driven by the secretion of flagellin monomers into the hollow interior of the filament and subsequent incorporation at the distal tip (3, 23, 47, 60).
A critical feature of flagellar biogenesis is that gene expression is coupled to assembly. Upon initiation, where cells transition from a nonflagellated to a flagellated state, gene expression proceeds in a sequential manner: first, genes encoding the basal body and hook proteins are expressed, and then, only after these structures are assembled, the late genes encoding the filament, motor, and chemotaxis proteins are expressed (29, 31). If hook or basal body assembly is unsuccessful, then the late genes are not expressed. This checkpoint enables cells to coordinate assembly and is the main regulatory mechanism observed during initiation. The way that cells enforce this checkpoint is to use late protein secretion as a proxy signal for hook-basal body (HBB) completion (24).
The flagellar promoters can be divided into three classes (9). A single Pclass1 promoter controls the expression of the flhDC master operon involved in initiating assembly. This promoter integrates environmental signals through the combinatorial action of multiple global transcriptional regulators, thus allowing cells to determine whether or not to be motile (51, 58). When motility is induced, the FlhD4C2 complex activates the Pclass2 promoters (25, 55). These promoters control the expression of the genes encoding the HBB and two regulatory proteins, σ28 and FlgM. The σ28 alternate sigma factor, also known as FliA, is required for activating the Pclass3 promoters, which control the expression of the late genes (50). However, prior to HBB completion, FlgM binds to σ28 and prevents it from activating the Pclass3 promoters (6, 7, 17). This inhibition, however, is relieved when the HBB is assembled, as the completed structure can secrete FlgM along with other late proteins involved in assembly (24). Thus, the cell is able to use protein secretion as a cue for HBB completion.
In addition to FlhD4C2, σ28, and FlgM, the flagellar proteins FlgN, FliT, and FliZ have been shown to regulate the assembly process. FlgN, the secretion chaperone for the hook associate proteins FlgK and FlgL, enhances translation of FlgM from class 3 transcripts (2, 4, 15, 31). FliT, the secretion chaperone for the filament cap protein FliD, binds the FlhD4C2 complex and prevents it from activating Pclass2 promoters (37, 57). FliZ, encoded in the fliAZY operon, is a positive activator of Pclass2/middle gene expression (37, 48). This protein also regulates the expression of genes encoding the Salmonella pathogenicity island 1 (SPI1) needle complex, which is involved in the invasion of intestinal epithelial cells (27, 40). In addition to FliZ, σ28 has also been proposed to activate Pclass2 promoters (28). However, unlike the case for the σ28-FlgM checkpoint involved in coupling Pclass3/late gene expression with HBB completion, the roles of the regulatory circuits dictated by these other proteins are still unclear.
In this paper, we investigate the role of FliZ in regulating flagellar assembly. We demonstrate that FliZ is an FlhD4C2-dependent activator of Pclass2/middle gene expression. Furthermore, our data suggest that FliZ regulates FlhD4C2 levels posttranslationally. We demonstrate that σ28 is unable to activate Pclass2 promoters on its own or in conjunction with FlhD4C2. Rather, our data suggest that the effect of σ28 on Pclass2/middle gene expression is through FliZ. Based on these results, we speculate that FliZ couples Pclass2/middle gene expression to FlgM secretion, committing cells to aggressively building multiple flagella once the first few are complete. This model suggests that assembly is far more complex than the sequential one proposed and also suggests that cells tune both Pclass2/middle and Pclass3/late gene expression in response to FlgM secretion rates.
All culture experiments were performed in Luria-Bertani (LB) broth at 37°C unless noted otherwise. Motility agar contained 0.3% Bacto agar, 1% tryptone, and 0.8% NaCl. Swarming plates contained either LB broth with 0.6% Difco agar and 0.5% glucose (34) or 0.6% Bacto agar, 1% tryptone, 0.8% NaCl, and 0.02% Tween 80 (49). Antibiotics were used 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 the growth of strains containing plasmid pKD46 were performed at 30°C as previously described (10). Loss of pKD46 from strains was achieved by growth at 42°C after electroporation of the PCR products. 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.
Bacterial strains and plasmids are described in Tables Tables11 and and2,2, respectively. All S. enterica serovar Typhimurium strains are isogenic derivatives of strain 14028 (American Type Culture Collection). The generalized transducing phage of S. enterica serovar Typhimurium P22 HT105/1 int-201 was used in all transductional crosses (11).
Standard “scarred” FLP recombination target (FRT) mutants were produced as previously described using either pKD3 or pKD4 as the PCR template (10). The ΔfliZ mutant was made using primers SS007F (CAC GTT TCA CCA ACA CGA CTC TGC TAC ATC TTA TGC TTT TGT GTA GGC TGG AGC TGC TTC) and SS007R (TGA TCG CAC CCG AAA AGT GCC GCA CAA CGT ATA GAC TAC CCA TAT GAA TAT CCT CCT TAG). The ΔfliA mutant was made using primers SS013F (ATA CGT TGT GCG GCA CTT TTC GGG TGC GAT CAT GCG CGA CGT GTA GGC TGG AGC TGC TTC) and SS013R (TCT GTA GAA ACG GAT AAT CAT GCC GAT AAC TCA TTT AAC GCA TAT GAA TAT CCT CCT TAG). The ΔflhDC mutant was made using primers SS058F (GCT GCT ATG CAT TTG ACC TTT TTG CTT CTT TTA CCG GGC CCA TAT GAA TAT CCT CCT TAG) and SS058R (TAC AGC CTG ATG AGG CGA CCA CGG CTG AGC TGT GTT TCG CGT GTA GGC TGG AGC TGC TTC). The ΔfliT mutant was made using primers SS078F (AAA GCA TCT TTC CAG GAG TCT CGT TAA TGA CCT CAA CCG TGT GTA GGC TGG AGC TGC TTC) and SS078R (TCT GGA GTA TGG AAG AAT TTT CAT ACG AGA CGG GAA AAT ACA TAT GAA TAT CCT CCT TAG). The ΔfliAZ mutant was made using primers SS013F and SS007R. Deletions of the SPI1 regulator rtsA (STM4315) and flhDC repressor rtsB (STM4314) were made using primers SS120F (CTG AAG ATG ATA TCC AGA GTT GCC TTG CCT ACC ACT CTA CGT GTA GGC TGG AGC TGC TTC) and SS120R (ATA TAG CTA TAT ATT AGT TTT CTT TTG AAA TAT TTT TCA GCA TAT GAA TAT CCT CCT TAG). All mutations were checked using primers that had 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.
The ΔPflhDC::tetRA strain was made by chromosomal replacement of the flhDC promoter with the tetRA element from transposon Tn10 using λ-Red recombination in cells carrying pKD46 as described by Datsenko and Wanner (10). The tetRA element was PCR amplified from TH8094 as described by Karlinsey (30) using the primers SS100F (AAA CAA AAA AGA ATT TGG TGT TGA CGT ACC CCT ATT CAG CAG AGT AGG GAA CTG CCA) and SS100R (GTG CGA CGT AGC CGC ACC CCG TGA TGT CGC CGG GAA GGC CCT AAG CAC TTG TCT CCT G). This arrangement put flhDC expression under the control of the tetA promoter. In the absence of tetracycline, the tetA promoter is repressed by TetR. In the presence of tetracycline, TetR repression is relieved and flhDC is transcribed from the tetA promoter.
Strains with the FlhC-3× FLAG tag epitope for Western blotting were made by amplifying the 3× FLAG tag and kanamycin resistance marker from the template pSUB11 (54) using the primers SS125F (TAT TCC ACA ACT GCT GGA TGA ACA GAT CGA ACA GGC TGT TGA CTA CAA AGA CCA TGA CGG) and SS125R (GGG CAA AAA AAA GCA GCG GTA CGT CGT TAC CGC TGC TGG ACA TAT GAA TAT CCT CCT TAG). The PCR product was then electroporated into wild-type cells carrying pKD46, and cells were then selected for kanamycin resistance. This arrangement resulted in expression of FlhC protein with a 3× FLAG tag at its carboxy terminus.
Green fluorescent protein (GFP) transcriptional fusions were made by amplifying the promoter region for the respective gene using PCR and then cloning the PCR product into the pPROBE-gfp[tagless] vector (46). The flgA transcriptional fusion was made using primers SS008F (CAT AGG TAC CTT ACT GGC CTA ATG TCA GCG) and SS008R (GGG AGA ATT CTG TGA AGC AAG CAT CAA CGC). The flgB transcription fusion was made using primers SS033F (GCT GGT ACC CGC GCA AAC TGA CGG CAT CC) and SS033R (GGG GAA TTC AGC CTG TCG AGC ATA TCT CC). The flhB promoter was made using primers SS068F (GGG GGT ACC GAC GGC GCA GGC GGC GGA AC) and SS068R (GGG GAA TTC TCG TCG TCG CTC TCT TCT GC). The fliE promoter was made using primers SS069F (GGG AAG CTT GCG GGC GCT TTT TAC CGG CC) and SS069R (GGC GGT ACC CAA TCC CCT GTA TTG CTG CC). The flgK promoter was made using primers SS076F (GGG GGT ACC ACC AAC TGA AAG CGA AAG CGG GC) and SS076R (GCG GAA TTC GGC GTG ATT AAT CAA GCT GG). The fliD promoter was made using primers SS071F (TTT GGT ACC GTC AGA CCT TTG ATG TTC GC) and SS071R (GTC GAA TTC ATG ATG AAA TTG AAG CCA TG). The fliC promoter was made using primers SS030F (ATC GGT ACC AGT GGT GCT GGA CGC CAC GG) and SS030R (ATC GAA TTC TTT GTA TTA ATG ACT TGT GC). The flhD promoter was made using primers SS077F (CCC GGT ACC TTT GTT CAA TCG GAT AAT CC) and SS077R (CCC GAA TTC GAC ATT GTG ACG TAT AAC GC). The PCR fragments were digested with KpnI and EcoRI (sequences are underlined) and then cloned into the multiple cloning site of the pPROBE-gfp[tagless] vector. In the case of the fliE promoter, the fragment was cloned in pPROBE-gfp[tagless] using HindIII and KpnI, as there is an internal EcoRI site. All constructs were sequenced prior to transformations into the wild-type and mutant strains.
In order to construct the luciferase reporters, PCR was used to amplify the Photorhabdus luminescens luxCDABE operon from the plasmid pRG19 (18) using primers SS082F (GGC GAA TTC CTT TAT AAG GAG GAA AAA CAT ATG ACT AAA AAA ATT TCA TT) and SS082R (AAA TCT AGA TTA TCA ACT ATT AAA TGC TTG GT). The PCR fragment was then cloned into the multiple cloning site of the plasmid pPROTet.E. The pPROTet.E plasmid was then digested with the enzymes EcoRI and AvrII, in order to include the TI terminator, and then subcloned into the EcoRI and NheI sites of pPROBE-gfp[tagless], yielding pPROBE-luxCDABE. Both AvrII and NheI yield compatible digested fragments. luxCDABE transcriptional fusions were then made by cloning the promoter fragments described above into the KpnI and EcoRI (and HindIII and KpnI for fliE) restrictions sites of pPROBE-luxCDABE.
The expression plasmids for fliA, fliZ, and flhDC were made by cloning the respective gene(s) into the EcoRI and HindIII or PstI sites (sequences are underlined) of pPROTet.E under the control of the strong promoter PLtetO-1 (41), resulting in the plasmids pFliA, pFliZ, and pFlhDC, respectively. The plasmid pFliA was made using primers SS053F (AGT GGT ACC TGC CGA TAA CTC ATT TAA CG) and SS053R (GGT AAG CTT CGC GAC CTA TAA CTT ACC CA). The plasmid pFliZ was made using primers SS046F (ATG GAA TTC GCC GCA CAA CGT ATA GAC TA) and SS046R (GGC AAG CTT TTA ATA TAT ATC AGA ACT GG). The plasmid pFlhDC was made using primers LC186 (ATA GAA TTC GTG CGG CTA CGT CGC ACA AA) and LC198 (ATA CTG CAG CGG TTA AAC AGC CTG TTC G). In the absence of TetR, the PLtetO-1 promoter is constitutively active. In order regulate the promoter activity, the tetR gene from transposon Tn10 was cloned using primers SS052F (GGT CTG CAG TGT CAA CAA AAA TTA GGA AT) and SS052R (GCT GCG GCC GC CGG AAA AAG GTT ATG CTG CT) into the PstI and NotI sites of pPROTet.E. In the absence of the inducer anhydrotetracycline (aTc), PLtetO-1 expression is inhibited due to TetR. However, in the presence of aTc at 200 ng/ml, PLtetO-1 expression is activated as TetR is inhibited.
FliZ was put under control of its native promoter by cloning the fliAZ operon with its promoter using primers SS123F (GGG CTC GAG AGT TTT CGC GCC CAA ATA CC) and SS123R (AAT AAG CTT TTA ATA TAT ATC AGA ACT GG). The PCR product and the plasmid pPROTET.E were digested with XhoI and HindIII, resulting in plasmid pFliAZ with the fliAZ genes under control of their native promoter. To remove the fliA gene, the plasmid pFliAZ was amplified using primers SS124F (CGT CTC TAT AGA CTA CCA GGA GTT CTC) and SS124R (CGT CTC GCT ATG ATA AAC AGC CCT GCG TTA A). The PCR product was then digested with BsmBI and ligated onto itself such the fliZ gene now was solely under control of the native PfliAZ promoter. The resulting plasmid was called pFliZ-native.
Three different types of flhC translational fusions were made. In the first, the fusion was made to the otherwise intact flhDC operon under the control of its native promoter. In this construct, the lacZ gene was fused in frame to the first 30 base pairs of the flhC gene. In order to the build this fusion, the region 812 bases upstream of the flhD start codon and 30 base pairs into the flhC gene was amplified using primers SS128F (GGG GCA TGC GGC GAC AAG AAT ATG GGT GT) and SS128RIII (GGG GGT CTC ATC ATA GCT TCC TGA ACA ATG CTT T) and then digested using SphI and BsaI. The lacZ gene was amplified using SS127F (GGG GGT CTC TAT GAC CAT GAT TAC GGA TTC) and LC096F (ACT TAA CGG CTG ACA TGG) using pAH125 as the template and digested with BsaI and NheI (20). The plasmid pPROBE-gfp[tagless] was digested with SphI and NheI and ligated with the flhDC and lacZ DNA fragments, thus replacing the GFP gene with the flhDC-lacZ translational fusion. This plasmid was called PflhDC flhDC-lacZ.
In the second translational fusion, the lacZ gene was fused 30 bases into the flhC gene. The resulting fusion was then placed under the control of the constitutively active PLtetO-1 promoter. The flhC gene fragment, consisting of 20 bases upstream of the flhC start codon and 30 base pairs into the flhC gene, was amplified using primers SS129FI (GGG GAA TTC CGA TAC GGC GCG TAA GAA AA) and SS128RIII and the resulting product digested using EcoRI and BsaI. The lacZ gene was amplified using SS127F and 127R and digested with BsaI and NheI. The digested products were then ligated with pPROTet.E digested with EcoRI and AvrII. The resulting flhC-lacZ translational fusion plasmid was called PLtetO-1 flhC-lacZ.
The third translational fusion was made by placing both the flhD gene and the flhC-lacZ fusion under the control of the constitutively active PLtetO-1 promoter. The region, 20 bases upstream of the flhD start codon and 30 bases into the flhC gene, was amplified using primers SS129FII (GGG GAA TTC AAT AAA GTT GGT TAT TCT GG) and SS128RIII and digested with EcoRI and BsaI. The lacZ gene was amplified using primers SS127F and SS127R and digested with BsaI and NheI. The digested products were then ligated with pPROTet.E digested with EcoRI and AvrII. The plasmid was called PLtetO-1 flhDC-lacZ. We also made an equivalent set of plasmids where the first 90 bases of the flhC gene were fused to lacZ. However, the results were the same and, therefore, are not reported.
End point and dynamic measurements of the reporter systems (GFP and luxCDABE) were made using a Tecan Safire2 microplate reader. For fluorescence end point measurements, 1 ml of culture was grown for 12 h at 37°C and then diluted 1:1,000 into fresh medium and grown for 6 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 absorbance to account for cell density. For time course measurements using the luciferase reporters, a protocol slightly modified from the one developed by Kalir and coworkers was employed (29). One milliliter of culture was grown overnight at 37°C and then diluted by a factor of 1,000 into fresh medium and grown to an OD600 of 0.2. One hundred microliters of culture was then transferred to a 96-well microplate that was sealed with a Breath-Easy membrane to allow for aeration while also minimizing evaporation. The temperature was maintained at 30°C, and luminescence and OD600 readings were taken every 10 min in the microplate reader, with shaking in the interval between subsequent readings. The kinetic experiments were performed at 30°C in order to obtain better resolution and temporal ordering between Pclass2 and Pclass3 activation. Identical experiments were also performed at 37°C with similar results except that the separation between the two promoter classes was far less pronounced. Fluorescent measurements of swarming cells were performed by gently scraping the swarm with an o-loop and then resuspending the recovered cells in phosphate-buffered saline (PBS). The cells were then diluted to an OD600 of 0.1 in PBS prior to making the fluorescence measurement. All experiments were done in triplicate, and average values with standard deviations are reported.
Cells from overnight cultures were subcultured 1:1,000 in fresh medium and grown at 37°C for 6 h. Prior to lysis, OD600 measurements were taken to ensure that there were equivalent numbers of cells between samples. To lyse the cells, cultures were spun down, resuspended 3:1 in 4× sodium dodecyl sulfate solubilizer, and boiled at 95°C for 10 minutes. Lysates were run on a 4 to 20% Tris-HCl precast gel (Bio-Rad) for 50 min at 150 V. Transfer to the membrane was done using Immobilon transfer membranes with 0.2-μm pore size (Millipore). 3× FLAG-tagged FlhC was detected with an anti-FLAG M2 monoclonal antibody (Sigma) and an anti-mouse horseradish peroxidase-conjugated antibody (Jackson Laboratories) using the ECL Plus Western blotting detection system (Amersham).
In order to quantify the relative protein levels, the membrane was scanned using a STORM 840 PhosphorImager (Amersham) and then analyzed using the LabWorks software package (UVP). The relative amount of protein in each lane was estimated by measuring the integrated OD for each band. All measurements were performed in triplicate.
Cells were grown in LB medium at 37°C and harvested after 6 h of growth. Quantification of β-galactosidase activity was performed as previously described by Miller (45). All measurements were performed in triplicate.
Reports describing the regulatory function of FliZ are inconsistent. Kutsukake et al. (37) demonstrated that FliZ positively regulates flagellar gene expression. In contrast, Frye and colleagues (16) were unable to detect a significant change in gene expression in a ΔfliZ mutant compared to the wild type using microarray analysis, though we note that the strains used in these two studies were different. To better understand the putative role of FliZ as a positive regulator of flagellar gene expression, we measured gene expression from a subset of flagellar promoters in the wild type (strain 14028), an isogenic ΔfliZ mutant, and the ΔfliZ mutant constitutively expressing fliZ from the PLtetO-1 promoter on a plasmid (41). GFP transcriptional fusions were employed as an indirect measure of promoter activities (46). In the ΔfliZ mutant, both Pclass2 and Pclass3 activities were reduced slightly less than twofold relative to wild-type activities (Fig. (Fig.1).1). However, in the ΔfliZ mutant constitutively expressing fliZ, both Pclass2 and Pclass3 activities were increased at least twofold relative to wild-type levels. These results indicate that FliZ positively regulates both Pclass2 and Pclass3 activities, consistent with the observations of Kutsukake et al. (37).
We next tested how deleting fliZ affected the dynamics of the Pclass2 flgA and Pclass3 fliC promoters. In these experiments, we used transcriptional reporters involving the luciferase operon, luxCDABE, from Photorhabdus luminescens (56). One advantage of using bacterial luciferase instead of GFP is that it is far more sensitive to changes in gene expression dynamics, particularly at low levels of expression (19). In agreement with our static fluorescence experiments, we observed that deleting fliZ reduced the maximal expression level for both the Pclass2 flgA reporter and the Pclass3 fliC reporter relative to the wild-type levels (Fig. (Fig.2).2). However, the dynamics were the same in the two strains; the only difference was in the peak response.
Previously, a number of researchers have observed that σ28 is a positive activator of Pclass2 activity (28, 36, 39). In particular, Kalir and coworkers demonstrated that overexpressing fliA leads to increased Pclass2 activity (28). As FliZ is also a positive regulator of Pclass2 activity, this means that two positive factors are encoded in the same operon. Therefore, any changes to fliA expression will result in a reciprocal change in fliZ expression and vice versa, as the fliAZY operon is under the control of both Pclass2 and Pclass3 promoters (26, 48). In previous studies of σ28 activity, assays have always been performed with strains in which the fliZ gene and fliAZY promoter were intact. We therefore asked whether the observations made by Kalir and coworkers and others are a direct result of σ28 or an indirect effect due to FliZ. To isolate the effects of σ28 and FliZ on Pclass2 promoters, we expressed fliZ or fliA (σ28) and tetR from the PLtetO-1 promoter on a plasmid in a ΔfliAZ mutant. In the absence of aTc, TetR negatively regulates the PLtetO-1 promoter and, as a consequence, FliZ or FliA is weakly expressed. However, in the presence of aTc, TetR no longer represses the PLtetO-1 promoter and FliZ or FliA is strongly expressed. In the case of fliZ, we observed approximately a twofold increase in Pclass2 activity upon induction with aTc (Fig. (Fig.3).3). However, in the case of fliA, we observed no effect on Pclass2 activity. As a control, we measured Pclass3 activity in the ΔfliAZ mutant strain. As expected, the plasmid expressing fliZ upon induction was unable to activate Pclass3 promoters, whereas the plasmid expressing fliA upon induction could. We conclude, therefore, that FliZ is a positive activator of Pclass2 activity that functions independently of σ28. Furthermore, our results suggest that the previously observed effect of σ28 on Pclass2 activity was actually via FliZ, for as FliZ is under the control of both Pclass2 and Pclass3 promoters, overexpressing σ28 increases FliZ expression, which thus leads to increased Pclass2 activity.
As a result of our constitutive expression experiments, we compared Pclass2 activity in ΔfliZ, ΔfliA, and ΔfliAZ mutants and the wild type. Interestingly, we observed a greater decrease in the ΔfliA and ΔfliAZ mutants than in the ΔfliZ mutants (Fig. (Fig.4A).4A). FlhD4C2 is known to negatively regulate its own expression (35). In addition, FliT counteracts this negative regulation (J. D. Brown et al., submitted for publication), presumably by binding FlhD4C2 and preventing it from inhibiting flhDC expression (37). As fliT, located in the fliDST operon, is under the control of both Pclass2 and Pclass3 promoters, its expression is responsive to fliA (σ28) expression (59). Therefore, in a ΔfliA mutant, fliT expression is reduced (5,394 ± 567 RFU/OD [wild type] versus 3,109 ± 224 RFU/OD [ΔfliA mutant]). As FliT is thought to prevent FlhD4C2 from inhibiting its own expression, we hypothesize that reduced fliT expression reduces flhDC expression. The net result is reduced class 2 gene expression. We therefore compared Pclass2 activity in ΔfliZ and ΔfliAZ mutants to that in ΔfliZ ΔfliT and ΔfliAZ ΔfliT mutants. Consistent with our hypothesis, there was no further decrease in Pclass2 activity when ΔfliT was introduced into the ΔfliZ, ΔfliA, and ΔfliAZ backgrounds (Fig. (Fig.4A).4A). Note that Pclass2 activity is slightly elevated in the ΔfliA ΔfliT relative to the ΔfliZ ΔfliT and ΔfliAZ ΔfliT mutants, as fliZ is still being expressed.
To confirm that the observed changes in Pclass2 activity are a result of the modulation of FlhD4C2 autoregulation by FliT, we replaced the flhDC promoter with the tetRA element from Tn10 (ΔPflhDC::tetRA). This arrangement constitutively expresses flhDC from its chromosomal locus in the presence of tetracycline. Using this construct, we observed that the decrease in Pclass2 activity was similar for the ΔfliZ, ΔfliA, and ΔfliAZ mutants in the ΔPflhDC::tetRA background (Fig. (Fig.4B).4B). Likewise, there was no appreciable difference between the ΔfliZ and ΔfliAZ mutants in the ΔPflhDC::tetRA ΔfliT background. Therefore, we conclude that the decrease in Pclass2 activity observed in the ΔfliA and ΔfliAZ mutants is due to a combination of reduced fliZ and flhDC expression, the latter being a direct result of the concomitant reduction in fliT expression. However, in the absence of FlhD4C2 autoregulation, the decrease is due solely to FliZ.
The above results also allowed us to test whether FliZ and FliT interact with one another. One potential mechanism from this hypothesis is that FliZ binds to FliT and prevents it from inhibiting FlhD4C2. To test the hypothesis that FliZ may operate through FliT, we compared Pclass2 activity in ΔfliZ, ΔfliT, and ΔfliZ ΔfliT mutants in the PflhDC::tetRA background to control for flhDC autoregulation (Fig. (Fig.4B).4B). If FliZ operates through FliT, then the ΔfliZ ΔfliT mutant should behave similar to the ΔfliT mutant. Contrary to this prediction, we observed roughly a 50% decrease in Pclass2 activity in the ΔfliZ ΔfliT mutant relative to the wild type. The decrease is slightly less than the decrease observed in the ΔfliZ mutant. As a comparison, we observed roughly a 25% increase in Pclass2 activity in the ΔfliT mutant relative to the wild type. We note that similar experiments were performed by Kutsukake et al. (37), and our results are consistent with theirs. Therefore, we conclude that FliZ and FliT operate independently, though we cannot discount the possibility that their actions are competitive.
FlhD4C2 is the master transcriptional regulator in the flagellar system. In the case of Pclass3 promoters, σ28 can activate these promoters independent of FlhD4C2. In the case of Pclass2 promoters, σ70 can activate these promoters only when aided by FlhD4C2. As FliZ has a positive effect on Pclass2 activity, one possibility is that it can interact with σ70 to activate Pclass2 promoters independent of FlhD4C2. Using the aTc-inducible expression system, we observed that FliZ is unable to increase the activity of Pclass2 promoters in the ΔflhDC mutant (results not shown). As a control, we also expressed flhDC from the PLtetO-1 promoter using the aTc-inducible expression system. In this case, the addition of aTc resulted in the strong activation of Pclass2 promoters (flgA, 488 ± 107 RFU/OD [uninduced] versus 12,742 ± 1,301 RFU/OD [induced]). These results demonstrate that FliZ activation of Pclass2 requires FlhD4C2.
In addition to being a flagellar regulator, FliZ is known to also enhance the expression of SPI1 genes (27, 40). Likewise, a number of SPI1 transcriptional regulators such as HilA and RtsB are known to affect flagellar gene expression (12, 13, 53). Due to this cross talk, we wanted to determine whether FliZ's effect on flagellar gene expression was through its effect on SPI1 gene expression. To test this possibility, we compared Pclass2 and Pclass3 activity in ΔSPI1 ΔrtsAB and ΔSPI1 ΔrtsAB ΔfliZ mutants. In the ΔSPI1 ΔrtsAB mutant, flagellar gene expression was similar to that in the wild type. However, in the ΔSPI1 ΔrtsAB ΔfliZ mutant, Pclass2 and Pclass3 activity was reduced approximately 50% compared to that in the ΔSPI1 ΔrtsAB mutant, similar to the decrease observed in the ΔfliZ mutant relative to the wild type (Fig. (Fig.5).5). Therefore, we conclude that the regulations of flagellar gene expression by FliZ and SPI1 are independent of each other.
The results described above suggest that FliZ acts through FlhD4C2. One possibility is that FliZ regulates flhDC transcription, resulting in an indirect effect on Pclass2 activity due to changes in FlhD4C2 expression. However, our ΔPflhDC::tetRA data suggest regulation at a posttranscriptional level. To confirm that FliZ acts posttranscriptionally, we measured the expression of the Pclass1 flhDC promoter in the wild type and a ΔfliZ mutant. As predicted, only a slight change in flhDC expression in the ΔfliZ mutant (8,661 ± 184 RFU/OD) relative to the wild type (10,481 ± 159 RFU/OD) was observed. This suggests the possibility that FliZ affects flhDC translation. To test this model, we fused a 3× FLAG tag to flhC at its native chromosomal locus in a manner similar to that described by Kato and coworkers (33). Note that this strain is still motile, as it forms rings of similar size as those of the wild type on motility plates (results not shown). Using this construct, we observed approximately a 50% decrease (87.0 ± 9.2 absorbance units [AU] [wild type] versus 47.0 ± 2.1 AU [ΔfliZ mutant]) in the amount of FlhC-3×FLAG protein in a ΔfliZ mutant relative to the wild type (Fig. (Fig.6).6). We also performed similar experiments in a ΔPflhDC::tetRA background and observed a similar decrease in FlhC-3×FLAG protein in the ΔfliZ mutant (111.0 ± 28.8 AU [wild type]) versus 51.5 ± 1.6 AU [ΔfliZ mutant]). Based on these results, we conclude that FliZ operates primarily at the level of the FlhD4C2 protein.
The above analysis indicates that FliZ's effect on FlhD4C2 is the result of either translational regulation or protein stability. We therefore explored whether FliZ directly affects flhC translation by utilizing translational reporter fusions. We constructed three different flhC-lacZ translation fusions on plasmids and then measured β-galactosidase activity in the wild type and a ΔfliZ mutant. In the first, which we called PflhDC flhDC-lacZ, the lacZ gene was fused to the flhDC operon under the control of its native promoter. In the second, which we called PLtetO-1 flhC-lacZ, the lacZ gene was fused to flhC under the control of the constitutive PLtetO-1 promoter. In the third, which we called PLtetO-1 flhDC-lacZ, the lacZ gene was fused to the flhDC operon again under the control of the constitutive PLtetO-1 promoter. In all three cases, the fusions were made on plasmids (see Materials and Methods for construction details). Comparing the results using these three translational fusions in the wild type and the ΔfliZ mutant, we observed no significant change in β-galactosidase activity (PflhDC flhDC-lacZ, 1,668 ± 30 [wild type] versus 1,424 ± 27 [ΔfliZ mutant]; PLtetO-1 flhC-lacZ, 1,239 ± 37.1 [wild type] versus 1,144 ± 208 [ΔfliZ mutant]; PLtetO-1 flhDC-lacZ, 1,552 ± 101.9 [wild type] versus 1,467 ± 28.2 [ΔfliZ mutant]). Based on these results, we conclude that FliZ is a posttranslational regulator of FlhD4C2 protein levels.
The goal of the flagellar regulatory network is to make sure that bacterial cells committed to moving have sufficient flagella to efficiently reach their desired destination. Therefore, to understand the role that FliZ regulation plays in motility, we tested the ability of ΔfliZ mutants to swim through liquids and swarm on surfaces. In the case of liquids, we observed that the ΔfliZ mutant formed smaller rings than the wild type on motility plates (Fig. (Fig.7A).7A). While there is a reduction in the ability to form rings, consistent with reduced flagellar gene expression, the ΔfliZ mutant is still motile. These results indicate that FliZ is not essential for swimming motility, though deleting it causes reduced swimming efficiency as determined by motility plates. Complementing this mutant with fliZ constitutively expressed from a plasmid, however, did not restore ring size. In addition, the ring morphology changes, with a diffuse ring phenotype observed in the ΔfliZ pFliZ strain compared to the sharp rings observed in the ΔfliZ mutant and the wild type. In contrast, we were able to partially restore the ring size when fliZ was expressed using its native fliAZY promoter on a plasmid, in the process recapitulating a similar result by Ikebe and coworkers (26). These results imply that the timing and regulation of fliZ expression play an important role in motility.
Apart from swimming in liquid viscous media, S. enterica serovar Typhimurium also employs flagella to swarm on solid surfaces (21). We therefore tested the swarming phenotype of a ΔfliZ mutant using two different medium recipes. In the first case, where we used LB medium with 0.6% agar and 0.5% glucose (34), we observed reduced swarm diameters in the ΔfliZ mutant relative to the wild type (Fig. (Fig.7B).7B). We were able to restore the swarm diameter by expressing fliZ from both a constitutive promoter and its native one on a plasmid. In the second case, where we used TB medium with 0.6% agar and 0.02% Tween 80 (49), we observed that the ΔfliZ mutant strain was unable to swarm whereas the wild type could (Fig. (Fig.7C).7C). When the ΔfliZ mutant was complemented with fliZ expressed from either a constitutive promoter or its native one on a plasmid, the swarming phenotype was partially restored. As a comparison, we note that FliT has no effect on swarming (results not shown).
We also measured Pclass2 and Pclass3 activity in swarming cells using the LB medium recipe. Consistent with our results for swimming cells, we observed a decrease in gene expression in the ΔfliZ mutant relative to the wild type (PflgA, 249 ± 38 [wild type] versus 98 ± 21 [ΔfliZ mutant]; PfliC, 284 ± 39 [wild type] versus 107 ± 18 [ΔfliZ mutant]). Similar results were also obtained in a ΔPflhDC::tetRA background (data not shown). The simplest interpretation regarding the swarm results is that the cells make insufficient flagella when FliZ is not expressed. Depending on the medium/environment, these changes can have either a minor or a profound effect. Finally, we note that we also measured swarming in S. enterica serovar Typhimurium LT2. In this strain, we observed only a minor decrease in the swarm diameter in the ΔfliZ mutant irrespective of the medium. Likewise, FliZ's effect on gene expression in LT2 was also less pronounced (results not shown).
We have shown that FliZ is an FlhD4C2-dependent activator of Pclass2 activity that acts, at least in part, at the level of FlhD4C2 protein. Deleting fliZ causes approximately a 50% decrease in Pclass2 activity and FlhC protein levels. Likewise, overexpressing FliZ roughly doubles Pclass2 activity. Furthermore, we have shown that FliZ functions independently of σ28 and FliT. In the case of σ28, we have shown that it does not directly activate Pclass2 promoters. Rather, the ability of σ28 to influence Pclass2 activity is indirect through FliZ, as the two are under the control of both Pclass2 and Pclass3 promoters. This indicates that fliZ expression is partly responsive to σ28 activity and, as a consequence, late protein secretion. Finally, an interesting finding was that FliZ had a strong effect on swarming motility for one specific medium recipe. In particular, we found that a ΔfliZ mutant was unable to swarm on TB medium supplemented with Tween 80. However, only a moderate reduction in swarming was observed when a more traditional recipe was used. These results potentially imply that the specific environment places differential demands on noncanonical flagellar regulators such FliZ.
We still do not know how FliZ increases FlhD4C2 protein levels. Our data suggest that FliZ does not directly increase protein translation rates but rather enhances the stability of FlhD4C2. Note that as we measured only FlhC protein levels and translation rates, it is not yet clear whether FliZ is also affecting FlhD. As for a possible mechanism, sequence analysis indicates that FliZ shares no similarity to any other known regulator. However, FliZ does possess a SAM-like phage integrase domain (PFAM PF02988) (14, 52). The presence of this domain suggests that FliZ is a DNA-binding protein. In the related insect pathogen Xenorhabdus nematophila, FliZ was recently shown to bind to the flhDC promoter and increase the rate of transcription (38). As the FliZ proteins from S. enterica serovar Typhimurium and X. nematophila share 56% sequence identify, these results suggest that FliZ is also a transcription factor in S. enterica serovar Typhimurium. However, transcriptional regulation of the flagellar genes in S. enterica serovar Typhimurium and X. nematophila is different despite the fact that the organisms share common regulators. For example, transcription of the fliAZ operon in X. nematophila is FlhD4C2 dependent, whereas the equivalent fliAZY operon is under independent control of both Pclass2 and Pclass3 promoters in S. enterica serovar Typhimurium (26, 48). Likewise, our data show that FliZ is not a transcriptional regulator of the flhDC operon in S. enterica serovar Typhimurium, whereas it is in X. nematophila. Furthermore, our data suggest that FliZ regulates the FlhD4C2 complex posttranslationally. Based on the results from X. nematophila regarding FliZ being a DNA-binding protein, the most likely mechanism in S. enterica serovar Typhimurium is that FliZ increases the expression of some protein that stabilizes FlhD4C2 or, alternatively, inhibits the expression of a protease such as ClpXP that degrades it. Furthermore, such promiscuity in FliZ DNA binding between the organisms is not altogether unreasonable, as the greatest difference between the two FliZ orthologs is at the C terminus, the region predicted to bind DNA based on the homology to the SAM-like phage integrase domain (38).
In the overall context of assembly, our data suggest the following model for FliZ-dependent regulation. Prior to completion of the HBB, FlhD4C2 can only partially activate Pclass2 promoters due to low protein levels arising from weak FliZ expression. However, when the first few HBBs are completed and FlgM is secreted at a sufficient rate such that σ28 is active, FliZ expression then increases. Note that FliZ is under the control of both Pclass2 and Pclass3 promoters, where the latter is the dominant promoter. Increased FliZ expression then leads to increased stability of FlhD4C2 and, as a consequence, further activation of the Pclass2 promoters. This mechanism suggests that Pclass2 activity is coupled to assembly. Unlike the case for Pclass3 promoters, where the coupling results in a strict binary checkpoint, Pclass2 activity is “tuned” in response to assembly. In other words, the cell increases Pclass2 activity once the first few HBBs are complete. In support of this model, we have recently shown that in HBB mutants Pclass2 expression rates are reduced, consistent with reduced FliZ regulation prior to HBB completion (Brown et al., submitted).
The current model for flagellar gene regulation focuses predominantly on initiation, when cells transition from a nonflagellated to a flagellated state (1, 9, 44). Key to regulating initiation is the σ28-FlgM checkpoint. Once cells already possess flagella, however, this checkpoint is no longer relevant, as FlgM is being continuously secreted from the cell. The fact that cells already possess flagella does not mean that the assembly process no longer needs to be regulated. Rather, different regulatory tasks and their associated mechanisms become necessary. In particular, cells need to ensure that they have sufficient flagella, not too many or too few. Furthermore, every time the cell divides, the progeny need to reinitiate the assembly process to ensure that they have sufficient flagella for efficient motility. This regulation means that cells need to continually monitor assembly and adjust gene expression accordingly. We speculate that FliZ is one element of this regulation, providing cells with a mechanism for increasing gene expression in response to assembly.
Note that FliT has the reciprocal effect of FliZ. Like fliZ expression, fliT expression is under the control of both Pclass2 and Pclass3 promoters, where again the latter is dominant. This means that fliT is maximally expressed when its Pclass3 promoter is active, concomitant with the completion of the first few HBBs. Unlike FliZ, however, FliT is a negative regulator of FlhD4C2 activity. Furthermore, FliD, the filament cap protein, regulates FliT activity. Prior to secretion, FliD binds FliT and is thought to prevent FliT from inhibiting FlhD4C2. Note that FliZ is indirectly regulated by late protein secretion, as its expression is controlled by σ28. As both FliZ and FliT are activated by the same signal, namely, late protein secretion, an open question is how these two antagonizing regulators operate. In fact, for FliT, both its expression (promoter activity) and function (loss of inhibition by FliD secretion) are responsive to protein secretion. One possibility is that these two regulators are dominant at different secretion rates, where by secretion rate we mean the total amount of protein secreted from a cell per unit time. In particular, we imagine that FliZ is dominant at low secretion rates and that FliT is dominant at high secretion rates, for example, in cases when the cell possesses too few or too many active HBBs, respectively. Such a mechanism would enable cells to regulate flagellum abundance if we assume that the secretion rate is proportional to number of flagella. In other words, this mechanism would enable cells to increase gene expression when there are too few flagella, due to the action of FliZ at low secretion rates, and decrease gene expression when there are too many, due to the action of FliT at high secretion rates.
We thank Vince Cannistraro, Joyce Karlinsey, James Slauch, and Hanna Walukiewicz for strains and technical advice. The pRG plasmids were a generous gift from R. Goodier and B. Ahmer. We also thank Linda Kenney for the generous gift of FlhD antibody used during our preliminary investigations.
This work was partially funded by BBSRC grant BB/D015855/1 (to P.D.A.) and National Science Foundation grant 0644744 (to C.V.R.). J.D.B. was partially supported by a Wellcome VIP award.
Published ahead of print on 9 May 2008.