σ²⁸-Dependent Transcription in Salmonella enterica Is Independent of Flagellar Shearing

Abstract

The FlgM anti-σ²⁸ factor is secreted in response to flagellar hook-basal body completion to allow σ²⁸-dependent transcription of genes needed late in flagellar assembly, such as the flagellin structural gene, fliC. A long-standing hypothesis was that one role of FlgM secretion was to allow rapid expression of flagellin in response to shearing. We tested this hypothesis by following FlgM secretion and fliC transcription in response to flagellar shearing. Experiments showed that the level of FlgM inside the cell was unchanged after shearing whereas the extracellular FlgM levels increased in the growth medium as time passed. Identical results were obtained with cells that were not exposed to shear forces: internal FlgM levels remained constant while external FlgM levels rose with time at rates similar to those for the sheared culture. Consistent with this find, FlgM/σ²⁸-dependent class 3 gene expression was unaffected by flagellar shearing but was affected by the growth phase of the cell. Regardless of exposure to shear forces, flagellar class 3 transcription rose sharply and then declined. These results demonstrate that flagellar regrowth following shearing is independent of FlgM secretion.

The flagellar organelle is a complex structure that, upon rotation, allows a bacterium to respond to chemical gradients within its environment. In addition to its role in motility, the flagellum plays an important role in colonization on surfaces and in the host-bacterium interaction, thus contributing to bacterial virulence (21). Salmonella enterica serovar Typhimurium possesses about 6 to 10 peritrichous flagella per cell (17). An individual flagellum is composed of three structural parts: the basal body, the hook, and the filament (3). The basal body is embedded in the membranes and acts as a molecular rotary motor, utilizing the proton motive force to drive flagellar rotation and propel the bacterium through a liquid environment and across surfaces. The hook is a flexible coupling between the external filament and the basal body, while the filament is a long (up to 10 μm), rigid, helical structure that extends from the cell surface (18). The pathway of assembly of the various flagellar components proceeds from the cytoplasm to the tip of the growing organelle, with the inner part of the basal body assembled first, followed by the hook and later by the filament. The structural subunits are added to the tip of the elongating filament (8, 11).

The flagellar genes are organized into a transcriptional hierarchy of three promoter classes (5). One striking feature of the flagellar regulatory system is the coupling of the sequential expression of the flagellar operons with the assembly process of flagellar structures. This mechanism is modulated by the complex activity of the transcription factor σ²⁸ (FliA) and its negative regulator, the anti-σ²⁸ factor FlgM. Regulation of intracellular (IC) FlgM levels, in response to flagellar biogenesis, results in the temporal regulation of σ²⁸-dependent transcription (13). In addition to these transcriptional controls, layers of posttranscriptional regulations are believed to complete the assembly process.

Research over the last few years has focused on the purpose of the σ²⁸-FlgM regulatory mechanism. The flgM gene is expressed from both class 2 and class 3 promoters (9). Class 2 FlgM has been demonstrated to be an internal checkpoint in the regulation of flagellar gene expression. It was shown that during initiation of flagellar biosynthesis, class 2 FlgM (which represents 20% of the total FlgM) is produced in the cytoplasm (13), binds to σ²⁸ (FliA), and prevents its association with RNA polymerase, thereby inhibiting flagellin gene transcription (20). However, once the hook-basal body (HBB) structure is assembled, FlgM is secreted outside the cells through the flagellar export apparatus (10, 14). Thus, FlgM protein senses the developmental state of the flagellum by being itself a substrate for secretion through the flagellum-specific type III secretion pathway. The onset of FlgM secretion would result in a decrease in the ratio of FlgM to σ²⁸ in the cell. Free σ²⁸ would then be available to associate with RNA polymerase and allow class 3 transcription (10). The flgM gene transcribed primarily from its class 3 promoter would be continuously secreted outside the cell through the tip of the elongating filament (12). The significance of the FlgM/σ²⁸ regulatory mechanism is not clear since cells deleted for the flgM gene are motile; however, they have about twice the number of flagella per cell (16), which would represent a significant energy cost. One hypothesis was that secretion of FlgM provided a mechanism to regulate flagellar length. Considering that the flagellar filaments grow to a length of approximately 15 μm (11), as the filament elongates FlgM secretion would begin to slow as a result of hindered diffusion of the protein subunits traveling through the elongating filament. Another hypothesized role for FlgM secretion was to regenerate flagellar filament following exposure to shear forces strong enough to shear off flagella. Given that FlgM is continuously secreted through the growing filament and that the rate of secretion decreases with filament length, shearing would result in an immediate high rate of FlgM secretion and an initial drop in its intracellular level. The reduced FlgM level resulting from the increased secretion that occurs immediately after filament shearing should significantly elevate class 3 gene expression and filament regeneration. The work reported here was designed to test this hypothesis.

In the present paper, we investigate the effect of flagellar shearing on Salmonella motility and on the intracellular and extracellular (EC) levels of FlgM. Unexpectedly, the shearing of flagella had no effect on FlgM levels or on class 3 fliC gene transcription. Another unexpected find was that flagella sheared from stationary-phase cells do not regrow, indicating that the capacity for flagellar growth is shut down in stationary-phase cells. However, a number of short flagella were visible 60 min after shearing on stationary-phase cells, suggesting that regrowth in stationary phase is not inhibited but occurs at a greatly reduced rate.

MATERIALS AND METHODS

Bacterial strains and growth medium conditions.

The following strains were used in this study: TH3740 [Δhin-5717::FRT(fliC^on) P_flhDC5451::Tn10dTc(del-25) ΔflgM5301], TH6232 [Δhin-5717::FRT(fliC^on)], TH8223 [Δhin-5717::FRT(fliC^on) P_flhDC5451::Tn10dTc(del-25) fliA5225(H14D)], TH8223 [P_flhDC5451::Tn10dTc(del-25) Δhin-5717::FRT fliA*5225(H14D)], and TH6232 lysogenic for a P22 lysogen that carries a P_fliC-lac reporter construct (6). Strains were grown in Luria-Bertani (LB) medium with aeration (7). Motility assays were performed on motility agar plates that consisted of 10 g of tryptone, 5 g of NaCl, and 0.3% (wt/vol) agar per liter (pH 7.4). Plates were incubated at 37°C.

Mechanical deflagellation.

Cultures (80 ml) of S. enterica serovar Typhimurium TH6232 and TH8223 were grown in LB broth at 37°C with shaking until they reached an optical density at 600 nm (OD₆₀₀) of ∼0.6 to 0.7 for log-phase cells and an OD₆₀₀ of ∼2 for stationary-phase cells. Cells were washed in 0.85% saline solution and resuspended in the same amount of fresh LB or the spent media from a ΔflgM strain (TH3740) grown in LB. Centrifugation was performed at a speed of 2,000 × g for 10 min at 4°C to minimize unwanted flagellar shearing. These conditions were chosen after a series of control experiments to determine which cell-pelleting conditions minimized flagellar shearing by centrifugation. Each cell culture was divided into two equal portions, one of which was not blended and used as a control. Flagella were sheared off mechanically by using an Omni homogenizer in a 25-ml-capacity cup. A 10-ml portion was blended three times for 10 s, separated by 5-s-pause intervals. Blender treatment was performed at 4°C as previously described (24). Samples were collected before and at 0, 5, 10, 20, 30, and 60 min after blending. To determine the effects of shearing on class 3 gene expression, strains were grown in 80 ml of LB broth culture to mid-log phase (OD₆₀₀, ∼0.6 to 0.7), gently pelleted, and resuspended in the same volumes (80 ml) of fresh LB. The cultures were then divided in two flasks (30 ml each); one was subject to blending (see above), and the other was used as an unblended control. Samples were collected before and at 0, 5, 10, 20, 30, and 60 min postshearing.

Agglutination test.

A 500-μl portion of blended cells was washed from detached flagella in saline solution and resuspended in 40 μl of LB. On a glass microscope slide, 1 drop of Salmonella H antiserum 1 complex (Difco) was added to 8 μl of cell culture. Unblended portions were washed as well since washing treatment did not diminish the ability of unblended samples to agglutinate. Agglutination was visible by eye and demonstrated the presence of FliC after blending, which determined whether the filament was completely (no agglutination) or partially (agglutination) sheared off.

β-Galactosidase assays.

β-Galactosidase assays were performed in triplicate as previously described (19). Cells were grown to 100 net Klett units (OD₆₀₀, ∼0.6) and permeabilized with chloroform and sodium dodecyl sulfate.

Immunoblotting of FlgE and FlgM.

Immunoblot analyses were performed in triplicate as previously described (12). Samples were prepared using 16% Tricine sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to a nitrocellulose filter. Filters were probed using either anti-FlgE or anti-FlgM polyclonal antibody (purified in our laboratory), followed by a secondary antibody conjugated with horseradish peroxidase (Bio-Rad). The peroxidase activity was visualized with an enhanced-chemiluminescence ECL Plus kit (Amersham Biosciences). Protein quantification was performed with the program ImageQuant for Macintosh, v1.2 (Molecular Dynamics). Protein level was calculated as the percentage of total protein at time t divided by the amount of protein at time zero and multiplied by 100.

Flagellar immunostaining.

Each well of a multitest slide (ICN Biomedicals, Inc.) was treated for 15 min at room temperature (RT) with 10 μl of a 0.1% poly-l-lysine solution (Sigma). Slides were washed two times with deionized water and allowed to dry. Fixation of bacterial culture was performed by using 100 μl of fresh paraformaldehyde-glutaraldehyde (500 μl of 16% paraformaldehyde-1 μl of 25% glutaraldehyde) fixative solution (Electron Microscopy Sciences), 20 μl of 1 M sodium phosphate buffer (pH 7.4), and 500 μl of bacterial culture. The suspension culture was gently mixed, and a volume of 10 μl was applied to each well of a poly-l-lysine-coated slide using a wide-bore Eppendorf tip to avoid flagellar shearing by pipetting. Slides were incubated for 20 min at RT, washed three times with phosphate-buffered saline (PBS; 10 mM NaPO₄ [pH 7.4], 150 mM NaCl, 3 mM KCl), and air dried. Each well was treated with 10 μl of 2% bovine serum albumin (BSA)-PBS (BSA fraction V; Sigma) for 10 min before 10 μl of the anti-FliC primary antibody (Difco) was added at 1:100 dilutions. Slides were placed on a wetted Whatman 3MM filter paper in a petri dish (wrapped with Parafilm to prevent evaporation) and incubated overnight at 4°C. Each well was then washed 10 times with 10 μl of PBS and 1 time with 10 μl of 2% BSA-PBS. The secondary anti-rabbit antibody conjugated to fluorescein isothiocyanate (FITC; Sigma) was added at 1:100 dilutions. After 3 hours of incubation at RT in the dark, cells were washed eight times with PBS and equilibrated for 10 min with 1 drop of Slow Fade equilibration buffer (Molecular Probes). At this step, 1 μl of 1 mg/ml FM4-64 (Molecular Probes) and 2 μl of 2 μg/ml DAPI (4′,6′-diamidino-2-phenylindole; Molecular Probes) were added to stain the bacterial membrane and the bacterial DNA, respectively. After a quick washing with Slow Fade equilibration buffer, 1 drop of Slow Fade glycerol (Molecular Probes) was placed into each well before a fitting slide with a 24- by 60-mm cover glass slip. Excess of Slow Fade glycerol was carefully aspirated, avoiding coverslip movements. When necessary, slides were stored in the dark at −20°C. Samples were examined using a Zeiss Axioscope II fluorescence microscope with Chroma filters 41028 for FITC, 31058 for FM4-64, and 31041 for DAPI and analyzed with the accompanying Zeiss image analysis programs.

RESULTS

Deflagellation by blender treatment results in removal of the distal parts of long flagellar filaments.

The mechanical treatment of blending cells has been chosen as an efficient, nonlethal method of removing flagella from Salmonella. Even prolonged blending of S. enterica serovar Typhimurium cells does not reduce cell viability (24). Blender treatment was performed on both log-phase (OD₆₀₀, ∼0.6 to 0.7) and stationary-phase (OD₆₀₀, ∼2) cultures as described in Materials and Methods. Loss of motility checked by light microscopy was close to 100%, suggesting that this method resulted in a complete loss of bacterial motility and presumably deflagellation of the bacteria. However, the blended cells were tested for H agglutinability and exhibited a weak agglutination reaction compared to that of the controls. As a positive control, we used a sample of unblended cells, which showed a typical, strong agglutination reaction. As a negative control, a nonflagellated strain (TH3740 in the absence of the inducer tetracycline [Tc]) did not agglutinate with the H antiserum. The presence of some agglutination following shearing suggested the presence of bacteria with short flagella, which were a result of either incomplete shearing or short flagella that were in the process of growing at the time of shearing but not long enough to be sensitive to the shear forces. To demonstrate the presence of short flagellar filaments at time zero after blending treatment, we exposed sheared cells to anti-FliC antibody and an FITC-conjugated secondary antibody. To facilitate the visualization of individual cells, we also added the live-membrane stain FM4-64 and DAPI to stain the DNA. Cells were examined immediately after deflagellation by fluorescence microscopy. Bacteria from log-phase blended cultures showed an average of one to two very short flagella per cell (Fig. 1A) at time zero after flagellar shearing, while only a few bacteria from the stationary-phase blended culture showed this phenotype at the same time point (Fig. 1B). Considering that cells grown in stationary phase decrease synthesis of new flagella, we suspected that for log-phase cells, most of the observed short flagella were nascent flagella able to resist shearing. These findings suggested that the probability that a flagellum is sheared after blending treatment is related to its length.

FIG. 1.

Flagellar-filament visualization with FITC-conjugated anti-FliC antibody. (A) Immediately after shearing (t = 0), cells from a log-phase culture showed an average of one to two very short flagella. (B) Immediately after shearing (t = 0), only a few cells from a stationary-phase culture showed short flagella; one cell having this phenotype is shown on the left. (C) As a positive control, we used a sample of unblended cells (picture on the left). As a negative control, a nonflagellated strain was used (picture on the right).

Deflagellation by blending treatment did not alter the capacity of the bacterium to produce flagella, as shown by the proportions of motile bacteria seen at different time points after shearing. Motility was monitored by light microscopy of deflagellated log-phase cells of strain TH6232 and found to rapidly increase from time zero to 60 min after shearing. Cells showed first a rolling motion 10 to 20 min after flagellar shearing, indicating the presence of short, rotating filaments, and at 30 min, 40% of the population were swimming. At 30 min after shearing, the number of swimming cells in the cultures appeared slightly less than that in the unblended culture. Based on these data, it was necessary to elucidate how the blending treatment removed flagella. Breakage occurred either throughout the structure or at its base. In particular, we looked at the flagellar HBB integrity, due to the finding that a filament does not regrow on broken flagella but new filaments are synthesized instead (1). As shown in Fig. 2, we did not see any detectable FlgE (hook) protein in the growth medium at any time point after flagellar shearing; moreover, cellular levels of FlgE were constant with time, with no relevant difference between unblended and blended cells. This result suggested that the hook structure was not broken by the blending treatment and was potentially available for regeneration of sheared flagella, although it was possible that our assay was not sensitive enough to detect small amounts of FlgE protein present among relatively vast amounts of flagellin. The absence of FlgE protein in the media was consistent with the hypothesis that the exposure to shear force breaks flagella at the distal end rather than at the base.

FIG. 2.

Detection of cellular and supernatant FlgE hook subunit protein. Western blots of IC and EC extracts made from cells of strain TH6232 were prepared using anti-FlgE antibody. Supernatant Western blots were prepared using anti-FlgE antibody and protein present in the growth medium from the same cell culture. Cells were washed two times in saline and resuspended in fresh LB broth. The absence of FlgE in the supernatant suggests that the hook subunit is not broken by blending treatment. According to Aizawa and Kubori, if a flagellum is broken somewhere on the basal body, a filament does not regrow; instead, a new filament is synthesized (1). The arrows indicate the positions of FlgE in the gel. Pre, preblending.

Effect of flagellar shearing on FlgM secretion in log-phase and stationary-phase cells.

One goal of this study was to determine what specific role FlgM secretion played in the flagellar assembly process. One hypothesis was that FlgM secretion could play a role in flagellar-filament regeneration after bacterial deflagellation. As flagella grow, substrate secretion slows due to hindered diffusion of secreted substrates, including FlgM, through the flagellum. The shearing of flagella could result in an initial burst of high-level FlgM secretion (and a concomitant drop in intracellular FlgM level) coupled to immediate high-level class 3 flagellar gene transcription (as a result of released σ²⁸).

The burst of flagellin gene (fliC or fljB) transcription would allow the cells to resynthesize filaments in response to filament loss from shearing. To test this idea, we monitored the secretion of FlgM after flagellar shearing. Cells were grown as described in Materials and Methods, and cultures were washed two times in cold saline to eliminate the FlgM protein already secreted in the media during cell growth and resuspended in fresh LB for log-phase TH6232 cultures and ΔflgM-spent media for stationary-phase TH8223 cultures. In a log-phase culture of TH6232, FlgM was detected in the growth medium and in the cell pellets of both flagellated (control) and deflagellated cells (Fig. 3). In particular, intracellular FlgM levels remained constant with time while extracellular FlgM levels started to rise 10 min after flagellar shearing and continued to the final measured time point (t = 30). However, no significant difference in FlgM levels was observed between blended and unblended cells.

FIG. 3.

Quantification of FlgM protein inside (IC) and outside (EC) the cells after flagellar shearing. Immunoblot analysis of TH6223 was performed in triplicate using log-phase cells washed in saline and resuspended in fresh LB; quantification was performed with ImageQuant for Macintosh, v1.2 (Molecular Dynamics). Pre, preblending.

To monitor FlgM secretion in blended stationary-phase cultures, we used TH8223 that has the flhDC operon expressed from a tetracycline-inducible promoter, P_tetA. These cells can produce flagella only in the presence of tetracycline (13). Without Tc inducer, cells do not produce flagella. When cells were grown in Tc-containing medium and Tc was removed, the elongation of existing filaments ceased. This strain also carried a fliA* mutation, H14D, which increased the stability of σ²⁸ (4) and therefore allowed the transcription of class 3 genes after switching off the flhDC master operon. When a stationary-phase culture of TH8223 was grown in LB with Tc and then resuspended in ΔflgM-spent media without the inducer, IC-FlgM levels remained constant with time while EC-FlgM levels start to rise only 30 min after flagellar shearing (Fig. 4). Also in this case, FlgM secretion rates were similar for blended and unblended cells. These results suggested that FlgM secretion played no significant role in regrowth of flagellar filaments.

FIG. 4.

Detection of IC and EC FlgM protein. (A) IC Western blots were prepared using anti-FlgM antibody and extract made from stationary-phase cells of strain TH8223. EC Western blots were prepared using anti-FlgM antibody and protein present in the growth medium. Cells were washed two times in saline and resuspended in ΔflgM-spent media without inducer; samples were taken at different time points after flagellar shearing. Pre, preblending. (B) Data from panel A presented in graph format. FlgM levels are presented relative to that of preblended IC FlgM, set at 100.

Immunofluorescence microscopy revealed that the average number of flagella on blended cells was less than that on the unblended controls.

Analysis by light microscopy demonstrated that all motility was lost immediately after blending treatment. Log-phase cells showed rolling motion 10 min after shearing, indicating that short filaments had grown and were rotating. By 30 min postshearing, 40% of the population was swimming. However, this percentage of swimming cells was less than that observed before deflagellation and less than that for the unblended control cells, suggesting that many flagella did not regrow. This observation was supported by the finding that blended samples always showed less agglutination than the corresponding unblended controls. Moreover, in a blended stationary-phase culture of TH8223 resuspended in ΔflgM-spent media or in LB without inducer, no swimming bacteria appeared 60 min after flagellar shearing.

To determine whether loss of swimming behavior was related to a deficit in flagellar regeneration and a subsequent reduction in average flagellar number, we followed flagellar-filament regeneration after blending treatment by immunofluorescence microscopy. Blended cells and controls were stained by using the dyes FM4-64 for the bacterial membrane, DAPI for the bacterial DNA, and anti-FliC primary antibody and FITC-conjugated secondary antibody for the flagellar filament FliC (22, 23). Due to the possibility that flagella might break as a result of slide preparation, the number of flagella per cell is an estimate of many independent microscopic observations.

In a log-phase culture, cells monitored at time zero (immediately after flagellar shearing) averaged one to two short flagella (Fig. 5, panel 2, t = 0), a result consistent with the weak agglutination reaction seen in the H agglutinability experiment. On subsequent incubations (Fig. 5, panel 2, t values of 5, 10, and 20), a rapid increase in flagellar-filament length was detected, and a gradual increase in flagellar number (4 ± 1 flagella per cell) was seen at the 30-min time point (Fig. 5, panel 2, t = 30). The average flagellar number always remained lower than before deflagellation. Unblended cells showed an average of 6 ± 2 flagella per cell (Fig. 5, panel 1). At the 60-min time point (Fig. 5, panel 2, t = 60), cultures were approaching stationary-phase growth and appeared less flagellated than in log phase, an observation consistent with previous results (2).

FIG. 5.

Flagellar-filament regeneration over time on log-phase cells. Flagellar filaments were visualized at different time points after blending treatment of log-phase cells (OD₆₀₀, ∼0.6). Cell membranes and bacterial DNA were stained with the dyes FM4-64 and DAPI, respectively. Flagellar filaments were detected by using the anti-FliC primary antibody and the secondary antibody conjugated with the fluorescence molecule FITC. (Panel 1) Unblended cells. Individual cells showed an average of 6 ± 2 flagella of different lengths (see time points 0, 5, and 10). (Panel 2) Blended cells. Cells monitored at time zero, immediately after shearing, showed an average of one to two short flagella per cell. Flagellar length rapidly increased on subsequent incubations at 5 and 10 minutes after shearing, while a gradual increase in flagellar number was detected at 30 and 60 minutes after shearing.

When a stationary-phase culture of TH6232 or TH8223, deflagellated by blending, was reincubated in LB or ΔflgM-spent media without Tc inducer, only a small proportion of flagellated bacteria appeared 60 min after flagellar shearing (Fig. 6). Moreover, flagellar filaments were very short compared to those on the controls at the same time point. These cells were not able to synthesize new flagella due to the lack of Tc-dependent flhDC induction; therefore, flagellar filaments that grew presumably did so from basal bodies formed in earlier generations. Considering that the flagellum increases in length for a limited period of time until it slows to a length of more then 10 μm (1), it might be that fully grown flagella (old flagella) were incapable of regeneration once broken off. When bacteria are deflagellated from a stationary-phase culture, they fail to regenerate most of their flagellar filaments. This is likely due to degradation of most of the flagellar class 3 mRNA and late-secretion substrates in stationary phase.

FIG. 6.

Flagellar-filament visualization of stationary-phase cells. (Panel 1) Unblended cells. Cell membranes were stained with the FM4-64 dye. Flagellar filaments were detected on stationary-phase cells (OD₆₀₀, ∼2) by using the anti-FliC primary antibody and the secondary antibody conjugated with the fluorescence molecule FITC. (Panel 2) Blended cells. Cells were stained similarly to the unblended controls, except that the chromosomal DNA was also stained with DAPI. Pictures were taken immediately after blending treatment (t = 0) and 60 min later (t = 60).

Deflagellation did not affect class 3 fliC transcription.

The unexpected finding that FlgM secretion was not affected by shearing predicts that induction of flagellin class 3 gene expression will also be unaffected by flagellar shearing. The effect of shearing on fliC gene transcription was tested using β-galactosidase assays of an FlgM/σ²⁸-dependent fliC-lac reporter construct. As shown in Fig. 7 and consistent with the FlgM secretion assays, the β-galactosidase profiles were the same for unblended and blended cells. The expression of the fliC gene was stable from the preblended state to 20 min after shearing. After 20 min, β-galactosidase increased to a peak level at 30 minutes after shearing, after which expression declined. These results suggest that shearing did not affect FlgM/σ²⁸-dependent gene expression. The β-galactosidase levels were stable during log-phase growth (OD₆₅₀, ∼0.7 to 0.9). Remarkably, as soon as cells reached an OD₆₅₀ of 1.2 to 1.3, fliC transcription significantly increased to a peak level that declined later as cell cultures continued to grow (OD₆₅₀, ∼1.6 to 1.8). This result was consistent with the observation that by 30 min postshearing, cells appeared to have more flagella (an average of 6 ± 2 flagella per cell) and were longer than at earlier time points (Fig. 5). These data suggest that the amount of flagellin produced by the cell is related to flagellar number and length during post-log-phase growth.

FIG. 7.

β-Galactosidase assays for the TH6232 ataA::[P22 P_fliC-lac] construct. Log-phase, flagellated cells carrying a P_fliC-lac reporter construct were assayed for β-galactosidase activity before and at different time points after shearing. PRE, preblending.

DISCUSSION

The shearing of flagellar filaments is believed to occur constantly in nature, but the mechanism regulating flagellar regrowth is not understood. The fact that flagella grow immediately after shearing (25) while it takes 30 min to generate a hook-basal body (13) suggests that flagella regenerate from the existing HBB structures left from sheared flagella. Stocker and Campbell first examined the removal of flagella by mechanical methods and their reappearance when little was known about flagellar structure, gene regulation, and biosynthesis (24). Four observations led us to examine a role for the anti-σ²⁸ factor FlgM in flagellar shearing: flagellar growth decreases exponentially with length (11), hindered diffusion is likely to be the limiting factor for rate of flagellar growth and final flagellar length (27), FlgM is secreted in response to HBB completion (13), and the filament cap gene, fliD, is expressed from class 2 and class 3 flagellar promoters (15). Intracellular production of the filament cap (FliD) subunits prior to HBB completion made sense with regard to efficiency of flagellum assembly. The FliD cap provides the polymerization site for flagellin monomers at the filament tip (28). Thus, the cap must be assembled prior to secretion of flagellin. If class 2 FliD is present at the time of HBB completion, then the switch in secretion substrate specificity from hook-type secretion substrates to late-type secretion substrates would allow FliD to be secreted along with FlgM prior to fliC or fljB transcription. Only five subunits of FliD are assembled onto a flagellum, compared to 20,000 subunits of flagellin (28). The presence of FliD in the cytoplasm prior to HBB completion would allow it to be secreted immediately upon HBB completion without competing with flagellin. The presence of a class 3 promoter for fliD might then be required to synthesize fliD in response to shearing since a new cap must be resynthesized before a filament can be regenerated.

We presumed that when flagella are at their shortest length, rate of FlgM secretion would be at its highest (11). This would provide a mechanism to respond immediately to loss of flagella by shearing, allowing a burst of class 3 σ²⁸-dependent gene expression to resynthesize lost filaments. We wanted to examine the effect of flagellar shearing on bacterial motility in the context of FlgM secretion coupled to regulation of class 3 σ²⁸-dependent gene expression. Specifically, we wanted to know whether initiation of class 3 transcription was required for this process and whether FlgM secretion was involved.

The suggested model proposes that rate of FlgM secretion would increase through shearing of flagella, resulting in an immediate drop in intracellular FlgM concomitant with an immediate increase in σ²⁸-dependent gene expression. We tested this model on both log-phase and stationary-phase cells. FlgM was detected in the growth medium and in the cell pellets of both deflagellated and flagellated cells (Fig. 3 and 4). We found that intracellular FlgM levels remained constant with time and FlgM rates were similar for blended and unblended cells. It is possible that since the flgM gene is also expressed from a class 3 promoter in addition to its class 2 promoter, it would self-regulate such that secretion of FlgM would result in the immediate resynthesis of FlgM. However, a small but transient increase in the σ²⁸/FlgM ratio could have occurred, resulting in a transient, increased σ²⁸-dependent transcription. Changes in intracellular FlgM levels might have been too small to be detected by Western blot analysis. In fact, we examined fliC-lac expression and saw no effect of flagellar shearing on β-galactosidase levels. Since β-galactosidase is a very stable protein, any burst in σ²⁸-dependent fliC-lac expression should have been observed, but none was. We did observe a burst of fliC-lac expression as cells entered stationary phase, but this was independent of shearing. This is consistent with a previous report indicating that flagellar class 3 gene expression shows a sharp increase as the cells enter stationary phase on swarm plates, followed by inhibition of gene expression (26). Inhibition of flagellar-gene expression during stationary phase would account for the inability to regrow flagella after shearing during stationary phase.

The observation that flagellar shearing did not affect σ²⁸-dependent fliC-lac expression leads us to conclude that the high steady-state level of σ²⁸-dependent gene expression in log-phase cells (TH6232) is adequate to respond to a need for increased FliC production. For stationary-phase cells, no swimming bacteria appeared 60 min after flagellar shearing. One possibility was that either flagellar mRNA or σ²⁸ was degraded in stationary-phase cells and there were no stored late-secretion substrates. Another possibility is that flagella reach a terminal-growth stage and that the flagellar-specific secretion apparatus ceases to function as a secretion apparatus. In this case, all flagellar regrowth after shearing would occur only on new or unfinished flagella.

.