Protease susceptibility of the Caulobacter crescentus flagellar hook-basal body: a possible mechanism of flagellar ejection during cell differentiation

The freshwater bacterium Caulobacter crescentus has a unique life cycle, which includes the transition between two microscopically distinguishable cell forms: a motile and chemotactically active swarmer (SW) cell with a polar flagellum, and a sessile, surface-attached stalked (ST) cell with a polar membranous extrusion, the stalk. The SW cell releases its single flagellum from the cell body during its differentiation into a ST cell (Aldridge & Jenal, 1999; Shapiro & Maizel, 1973). Since the release of the flagellum is synchronized with cell differentiation and the cell cycle, an active and controlled ejection mechanism should exist (Aldridge & Jenal, 1999). The ejected flagella are composed of the filament, hook and rod substructures, with the break in the structure reportedly occurring at the junction between the proximal and a distal portion of the rod (Shapiro & Maizel, 1973; Stallmeyer et al., 1989). How this break occurs and how this process is regulated during development is as yet unknown.

The flagellar structure is anchored in the cytoplasmic membrane via the MS ring structure (DePamphilis & Adler, 1971). In Salmonella typhimurium the FliF protein is the sole component of the MS ring complex (Ueno et al., 1992). Although both terminal regions of FliF, which form the cytoplasmic domain, are cleaved by a tryptic digest, the major part of the basal bodies is considerably stable against this treatment in vitro (Ueno et al., 1994). Jenal & Shapiro (1996) showed by immunoblot and pulsechase assays that the C. crescentus FliF protein is proteolytically turned over during the SW-to-ST cell differentiation, suggesting that degradation of the FliF anchor protein is coupled to flagellar ejection.

To understand how this proteolytic digestion of FliF and consequently the destruction of the MS ring complex occurs, we developed a method for purification of intact flagella from C. crescentus and examined the physico-chemical properties of the basal structure. We found that the C. crescentus MS ring structure, in contrast to the same structure isolated from enteric bacteria, is very susceptible to protease treatment. This is in line with the idea that the physico-chemical properties of the C. crescentus HBB structure are adjusted to the developmentally regulated process of flagellar ejection.

Strains, media and growth conditions.
C. crescentus wild-type strain NA1000, a pleD mutant (strain UJ248) and a pleC mutant (strain UJ506) were used. Cells were cultured in PYE medium (1 % peptone, 1 % yeast extract, pH adjusted to 7). Both reagents were of the first grade and purchased from Wako Chemical. Cells were cultivated at 30 °C for 12 h by gentle shaking and harvested at around OD₆₆₀ 1·0.

Reagents.
All reagents used were special grade. Trypsin and chymotrypsin were purchased from Wako Pure Chemical Industries and Pronase E was from Sigma.

Observations of cell types.
Cells were analysed in a concave well on a hole-slide glass (Toshinriko, Oosaka), observed with a phase-contrast microscope (Olympus CH-2), and recorded on VTR through a CCD camera (Panasonic BL200) attached to the microscope. Sixty to one hundred cells in each scene of 1020 s duration were counted, and the ratio of cell types (SW cells : ST cells : PD cells) was calculated.

Purification method.
The hookbasal body (HBB) complexes of C. crescentus were purified by a modification of the original protocol used for S. typhimurium (Aizawa et al., 1985). The modified protocol was as follows (see Results and Discussion for details).

Spheroplasts were formed by adding lysozyme and 0·5 mM EDTA, and lysed with 1 % AM-3130N (Nihon Surfactant Kogyo K. K.). AM-3130N is a non-ionic detergent and its major constituent is myristic acid betaine. Chromosomal DNA present in the lysate was digested by adding a small amount of DNase. Bundles of flagella formed after addition of 2 % polyethylene glycol (PEG) were collected by low-speed centrifugation, and resuspended in 0·2 M KCl/KOH buffer (pH 12). The sample was resuspended in TET buffer (10 mM Tris/HCl pH 8, 1 mM EDTA and 0·1 % Triton X-100) for further analysis.

SDS-PAGE and Western blotting.
SDS-PAGE was carried out using the mini-gel kit from Bio-Rad. Acrylamide concentrations of gels were 10 %, 12·5 % or 15 % depending on the molecular range of the proteins of interest in each experiment. Gels were stained by silver. Molecular mass standard protein markers were obtained from Bio-Rad. To confirm the degradation of FliF by trypsin, Western blotting was carried out, using anti-C. crescentus FliF antibody (Jenal & Shapiro, 1996).

Electron microscopy.
Samples were negatively stained with 2 % phosphotungstic acid (pH 7·0) and observed with a JEM-1200EXII electron microscope (JEOL). Micrographs were taken at an accelerating voltage of 80 kV.

Flagellar ejection during cell cycle
Three different forms of C. crescentus cells can be distinguished morphologically: the swarmer (SW) cell, the stalked (ST) cell, and the predivisional (PD) cell, which combines both cell types. PD cells are defined by a clearly visible constriction (pinching) near the middle of the cell, which indicates ongoing cytokinesis. Elongating stalked cells have not initiated cell division and lack a constriction. Flagella are assembled in the PD cell and are ejected into the supernatant during the SW-to-ST cell differentiation. When cells of the wild-type strain were cultured at 30 °C by gentle shaking, the ratio of these cell types was roughly 1 : 1 : 1 at the mid-exponential phase of growth. This is consistent with earlier measurements of the C. crescentus cell cycle phases (Judd et al., 2003; Stephens et al., 1996).

Flagella ejected into the medium during the SW-to-ST cell transition were collected by PEG precipitation from an overnight culture. Electron microscopy revealed that the ejected structure consisted of the filament, the hook and part of the rod (Fig. 1). The length of the sheared rod was about 18 nm (Table 1). Acid treatment (pH 4·4) of these isolated flagellar structures gave rise to a complex consisting of the hook and rod structure. This complex consists of four major protein bands visible in SDS gels, 70, 68, 33 and 28 kDa; these were identified by N-terminal amino acid sequencing as FlgK (hook-associated protein HAP1), FlgE (hook), ribosomal protein L2 and FlgG (distal rod), respectively (Table 2).

(144K): Fig. 1. (a) Electron micrograph of ejected flagella from the supernatant of overnight C. crescentus cultures. (b. c) Close-up views of the hookrod junction (arrows) of the ejected structure. The samples were negatively stained with 2 % phosphotungstic acid (pH 7·0). Bars, 100 nm.

Table 1. Mean lengths of rod structures isolated from C. crescentus

Table 2. Protein components of purified flagellar HBB of C. crescentus

These results support earlier observations indicating that a complex composed of filament, hook and rod is released into the culture fluid and that the break point of the structure can be localized to the MS ringrod junction (Shapiro & Maizel, 1973; Stallmeyer et al., 1989). This leaves two possibilities for the release of flagella in C. crescentus: either by a breakage between the MS ring complex and the rod, or by destruction of the MS ring complex itself.

Purification method for C. crescentus flagellar HBBs
To examine the physico-chemical properties of the C. crescentus basal body structure, we had to develop a purification method that allowed the isolation of intact flagella in high enough quantities. This is a general problem encountered with mono-flagellated bacteria, and was aggravated by the fact that due to the intrinsic properties of the C. crescentus developmental cycle, more than half of the population is non-flagellated. This problem was overcome by using two regulatory mutants that fail to eject the flagellum during cell differentiation: the first has a mutation in the pleC gene, which encodes a polar sensor histidine kinase (strain UJ248; Aldridge et al., 2003); the second has a deletion in pleD, which encodes a response regulator involved in polar development (strain UJ506; Paul et al., 2004). The basal body structure and composition isolated from both mutants was compared with that of flagella isolated from wild-type cells. There were no discernible differences between them, arguing that the failure of these mutants to eject the flagellum is not caused by an altered structure of the HBB itself.

The original protocol used for S. typhimurium was modified as follows.

Spheroplast formation.
EDTA is necessary for lysozyme to digest the peptidoglycan layer of Gram-negative bacteria and to form spheroplasts. While spheroplast formation was easily achieved above 0·5 mM EDTA, an excess of chromosomal DNA was released under these conditions and could not be removed efficiently by endogenous DNase. Therefore, we maintained the EDTA concentration at 0·5 mM during spheroplast formation.

Cell lysis.
For C. crescentus cells, the non-ionic detergent Triton X-100 did not effectively remove membranous material. We surveyed a collection of detergents (kindly provided by Nihon Surfactant Kogyo K. K., Utsunomiya) and found that the non-ionic detergent AM3060N was best suited to solubilize most of the membranous remnants, leaving the flagella intact.

CsCl density gradient.
CsCl density-gradient centrifugation is a final step for further purification of intact flagella to separate the basal structure from membranous contaminants for S. typhimurium, Bacillus subtilis and Rhodobacter sphaeroides (Aizawa et al., 1985; Kubori et al., 1997; Kobayashi et al., 2003). However, CsCl effectively destroyed the basal structure of C. crescentus flagella. Therefore, we omitted this step. Instead, we employed polyethylene glycol (PEG); bundles of flagella formed after addition of 2 % PEG were collected by low-speed centrifugation.

Depolymerization of filaments.
To purify HBBs at a high yield, it is necessary to remove flagellar filaments, which constitute 99 % of the material of intact flagella. In conventional methods, acidic buffers (pH 23) were employed for this purpose. C. crescentus HBBs were fragile to acid and started to disintegrate at pH 4. We found that at pH 4·5, filaments dissociated into flagellin monomers, leaving the HBB structure intact.

Analysis of purified HBBs
The overall structure of C. crescentus HBBs purified as described above was similar to the structure of S. typhimuriun HBBs, with four rings and a central penetrating rod connected to the curved hook structure (Fig. 2). Although C. crescentus HBBs have been reported to contain an extra ring between the inner MS ring and the distal PL ring complexes, called the E ring (Stallmeyer et al., 1989), we were unable to detect such an extra ring structure in our preparations. There are several possible reasons for this discrepancy. First, the E ring could have detached from the HBBs in our preparation. Second, the E ring reported by Stallmeyer et al. (1989) could be an artifact of negative staining. It is important to note that so far a gene encoding an E ring structure has not been identified in C. crescentus (Wu & Newton, 1997).

(151K): Fig. 2. (a) Electron micrograph of purified HBBs from C. crescentus. (b) Close-up views of the basal body part. The samples were negatively stained with 2 % phosphotungstic acid (pH 7·0). Bars, 100 nm.

The protein components of purified HBBs were analysed by SDS-PAGE. We chose major bands that were constantly observed from preparation to preparation; there were nine, with approximate molecular masses of 70, 68, 66, 42, 33, 30, 29, 28 and 25 kDa (Fig. 3). The protein bands were blotted onto PDF membrane and analysed by N-terminal amino acid sequencing (Table 2). Even after the filament was removed by heat treatment, trace amounts of flagellins remained in the sample. The C. crescentus filament is composed of multiple flagellins (Ely et al., 2000). We identified FljL as the 30 kDa protein on the SDS gel, and a mixture of FljK and FljJ formed the 29 kDa protein band (Fig. 3a). The remaining bands on the gel shown in Fig. 3(a) were identified as follows: 66 kDa, FliF/MS ring; 42 kDa, FlgH/L ring; 25 kDa, FlgF/proximal rod; and 33 kDa, ribosomal protein L2, which is known to co-purify with flagella.

(28K): Fig. 3. (a) Protein components of C. crescentus HBBs as analysed by SDS-PAGE. The purified HBB preparation gave rise to typically nine major bands in SDS gels (12·5 %): 70, 68, 66, 42, 33, 30, 29, 28 and 25 kDa (lane 2). The protein bands were blotted onto PDF membrane, and were analysed by N-terminal amino acid sequencing. Identified proteins are labelled on the right. The protein composition of S. typhimurium HBBs is shown for comparison (lane 1). (b) Time-course of the trypsin degradation of C. crescentus HBBs as detected by Western blotting using anti-FliF antibody. Lanes: 1, protein marker; 26, samples treated with no trypsin (2), and at 2 min (3), 10 min (4), 30 min (5) and 60 min (6) after trypsin was added.

Proteolysis of purified HBB
To determine the stability of C. crescentus HBBs, we examined their susceptibility to several proteases. When purified flagella (Fig. 4a) were incubated with 100 µg trypsin (or chymotrypsin) ml¹ or with 10 µg Pronase E ml¹ at 30 °C for 1 h, the HBBs lost both the MS and the PL ring complexes, leaving the rod exposed (Fig. 4b). The rods uncovered after proteolysis appeared as short as those observed in spontaneously released flagella. The length of exposed rods was 18 nm or less, probably depending on the degree of digestion.

(151K): Fig. 4. Electron micrographs of HBBs treated with different proteases. (a) Intact flagella, (bd) after treatment with 100 µg trypsin ml1 for 1 h (b), 10 µg trypsin ml1 for 30 min (c) or 1 µg Pronase E ml1 for 30 min (d). The samples were negatively stained with 2 % phosphotungstic acid (pH 7·0). Bars, 100 nm.

Partial degradation of HBBs exposed the intact rod structure. When incubated with 10 µg trypsin (or chymotrypsin) ml¹ at 30 °C for 30 min, the HBBs selectively lost their MS ring complexes, leaving the proximal rod exposed below the PL ring complex (Fig. 4c). At concentrations between 10 µg ml¹ and 100 µg ml¹, the appearance of degraded HBBs remained unaltered. When the degraded HBB sample was analysed by SDS-PAGE followed by Western blotting, the 66 kDa band of FliF disappeared and concomitantly bands at 58, 48, 47 and 35 kDa appeared as degradation products of the full-length FliF (Fig. 3b). All the other bands remained unchanged (data not shown). This indicates that the C. crescentus MS ring structure might be intrinsically susceptible to proteolytic attack. Alternatively, the fact that C. crescentus FliF (536 aa) has a higher number of lysine residues (29 K) than FliF from Salmonella (19 K/559 aa) could explain the greater susceptibility of FliF from C. crescentus to tryptic digestion.

When flagella were incubated with 1 µg Pronase E ml¹ at 30 °C for 30 min, both the MS and PL ring structures were degraded. The remaining rods were apparently longer than those of ejected flagella (Fig. 4d). The mean length of rods of ejected flagella was 18 nm, while rods of Pronase E-treated flagella had a mean length of 31 nm. The distance between the S and P ring of intact flagella was 10 nm, while the thickness of the P and L ring complex was 14 nm (Fig. 5, Table 1), indicating that a few nanometres of the rod might penetrate into the MS ring complex. From this we conclude that the rod exposed after partial digestion by Pronase E is intact. Interestingly, the naked rod appeared as a uniform rod with no trace of the E ring or a narrow gap along the axis, indicating that the E ring is not functionally linked to flagellar ejection.

(13K): Fig. 5. Schematic representation of a complete HBB and an ejected substructure. The length of the rod of ejected flagella is 18 nm (a), while that of Pronase E-treated flagella is 31 nm. The gap between the S and P ring of intact flagella is 11 nm, while the thickness of the P and L ring complex is 14 nm (b).

Conclusions
Purified C. crescentus HBBs are physico-chemically more fragile to high CsCl concentration, acidic pH or protease treatment than HBBs of S. typhimurium. Effects of CsCl and acid on C. crescentus HBBs can be excluded as the direct cause of flagellar ejection. Although no specific proteases for HBB digestion have yet been identified, our data together with previous observations consistently suggest that during cell differentiation ejection of the flagellar structure could occur by controlled proteolysis of one or several components located at the base of the HBB structure. We are grateful to Takuya Gotoh for his technical help, and to Tomohisa Hoshino and Show-Taro of DAGA Graphics for figures. This work was supported by grants from the Ministry of Education, Science, Sports, and Culture of Japan to S.-I. A. and by Swiss National Science Foundation fellowship 31-59050.99 to U. J.