INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The life cycle of the aquatic bacterium Caulobacter crescentus provides a model system for the analysis of programmed developmental events. Each cell division produces two morphologically dissimilar progeny cells, a motile swarmer cell and a sessile stalked cell. Flagellum biogenesis is initiated in the predivisional cell and results in the synthesis of a basal body, hook, and flagellar filament at one pole of the developing swarmer cell (reviewed in reference 24). This process involves at least 50 genes (19). Most of the flagellar genes have been organized into a regulatory hierarchy that includes four classes of genes (reviewed in reference 54). In this hierarchy, the expression of genes in an earlier class is required for expression of the genes in subsequent classes. Furthermore, the genes that encode the structural components of the flagellum are transcribed in the order that their gene products are assembled into the structure. For instance, expression of the class II genes is required for expression of the class III hook and basal body genes. Similarly, expression of the class II and III genes and assembly of their products are required for expression of the class IV flagellin genes. One class II gene, rpoN, encodes sigma 54, the sigma factor that binds to class III and class IV promoters (11). In addition, integration host factor and transcriptional activators are required for expression of class III and class IV genes (7, 25, 26, 73). Synthesis of the flagellin subunits encoded by the class IV genes is also subject to posttranscriptional control mechanisms (3, 41, 42). Recently, Quon et al. (53) have shown that mutations in the class I ctrA gene result in altered expression of class II flagellar genes.

The flagellar filament is composed of three distinct flagellin monomers. A 29-kDa flagellin is initially assembled at the hook-proximal portion of the filament (18). Subsequently, the 27- and 25-kDa flagellins are synthesized and assembled consecutively (18, 71). The 25-kDa flagellin constitutes the distal two-thirds of the filament (18). In addition to these structural genes, the expression of the flagellar genes flmA (formerly flaA), flmD (flaR), flmE (flaZ), flmG (flbA), and flmH (flaG) is required for the synthesis of normal flagellin proteins (30). Strains containing mutations in any of these genes have a normal basal body and hook structure but fail to assemble a flagellar filament (30). Mutations in the flmA, flmD, flmG, and flmH genes result in the production of a novel 22-kDa flagellin protein. In addition, the production of the other flagellins is severely decreased (30, 57, 61). Mutations in the flmE gene resulted in the production of flagellins that migrate slightly faster than the wild-type flagellins on sodium dodecyl sulfate (SDS)-polyacrylamide gels. Thus, flagellins from the flmE mutant have apparent molecular masses of 26 and 24 kDa instead of 27 and 25 kDa (30). The 22- and 24-kDa proteins are also produced in flbT mutants when the flagellins are overproduced (57). Recently, we have shown that degradation of the fljK mRNA is regulated by the flbT gene product (42).

To analyze the role of the flm genes in the production of these flagellin proteins, we have cloned and determined the nucleotide sequences of the flmAB, flmCD, flmEF, and flmGH operons. Homology searches revealed that the deduced amino acid sequences of the flmA, flmB, flmC, and flmD genes are similar to sequences of proteins involved in capsular and spore coat polysaccharide biosynthesis. Thus, the flm genes may be involved in glycosylation. To study the regulation of the flm operons, we have constructed fusions of the flmA, flmC, flmE, and flmG promoters to the cat or lacZ reporter gene. We demonstrate that expression of the flmA, flmE, and flmG genes is altered by a mutation in the ctrA gene, a class I flagellar gene. In addition, we show that the flmA, flmE, and flmG genes are expressed primarily in predivisional cells, indicating that these putative operons are temporally regulated. In contrast, the flmC gene is expressed throughout the cell cycle, with a twofold increase in the predivisional cell. These results, along with studies of the flagellar regulatory hierarchy, suggest that the flm genes belong to a new class of flagellar genes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Tables 1 and 2, respectively. All C. crescentus strains used are derivatives of the wild-type strain CB15. C. crescentus strains were grown at 33°C in peptone-yeast extract (PYE) medium or in defined minimal medium M2 (28). Escherichia coli strains were grown at 37°C in Luria-Bertani (LB) or defined E medium supplemented with 1.0 mM proline (44). Complementation of motility defects was demonstrated in semisolid medium as previously described (59). Plasmids were transferred to the recipient C. crescentus flm mutants by conjugation from their E. coli host S17-1. Antibiotics were added, when appropriate, at the following concentrations: ampicillin, 50 µg/ml; sulfanilamide, 350 µg/ml in E medium; tetracycline, 15 µg/ml in LB and 1 µg/ml in PYE medium; and kanamycin, 50 µg/ml.

Molecular techniques. General cloning procedures were carried out as described by Sambrook et al. (55). C. crescentus chromosomal DNA was isolated as previously described by Malakooti and Ely (40). All enzymes used in the manipulations of DNA were used according to the specifications of the manufacturer. Transformation of E. coli was carried out as described by Sambrook et al. (55). Transformation of C. crescentus was carried out by electroporation according to the procedure of Gilchrist and Smit (23). The nucleotide sequence of both strands of DNA containing the gene of interest was determined from either double-stranded templates or single-stranded templates by the dideoxy-chain termination method (56) using a Sequenase version 2.0 kit (U.S. Biochemical, Cleveland, Ohio). Nucleotide and amino acid sequence analyses were performed with the Wisconsin Package of the Genetics Computer Group (Madison, Wis.) (16).

Construction of flmA::cat, flmC::cat, flmE::cat, and flmG::cat gene fusions. For the flmCD, flmEF, and flmGH operons, a promoterless chloramphenicol acetyltransferase (CAT) cartridge contained in a 0.8-kb HindIII fragment was inserted into the unique HindIII sites of pGL39, pGL11, and pGL24, respectively. In the case of the flmAB operon, a 1.7-kb BamHI-SalI fragment containing the flmA promoter region was cloned into pGL30 that contained the cat gene inserted in the orientation opposite that of the lacZ promoter. The resulting plasmids, pGL20, pGL21, pGL22, and pGL41, were introduced in C. crescentus LS107 by electroporation. Transformants were selected for ampicillin resistance, causing the integration of the nonreplicating plasmid into the chromosome. Single-crossover recombinants were identified by Southern blot analysis (data not shown), resulting in strains SC3971, SC3973, SC3975, and SC4016, containing the chromosomal fusions flmA::cat, flmC::cat, flmE::cat, and flmG::cat, respectively.

CAT and -galactosidase assays. Strains carrying a transcriptional fusion (plasmid borne or integrated into the chromosome by homologous recombination) were grown to the exponential growth phase (125 Klett units or A₆₀₀ of 0.5) in PYE medium supplemented with appropriate antibiotics. Cell extracts were prepared by sonic disruption of the cells in 0.1× TE (10 mM Tris-0.1 mM EDTA [pH 8.0]). The cell debris was removed by centrifugation, and the supernatant was assayed to determine the protein concentration (9). CAT activity was assayed by using [³H]acetyl coenzyme A according to the directions of the manufacturer (NEN, Boston, Mass.). Assays of -galactosidase activity were performed as previously described (44). Totals of 5 and 12 µg of protein were assayed for -galactosidase and CAT activities, respectively.

Cell synchronization and immunoprecipitation. Strains containing flmC::cat, flmE::cat, and flmG::cat chromosomal integrated fusions or flmA::lacZ plasmid (pSCW1967)-borne fusion were grown in M2 medium and synchronized by differential centrifugation (6). Swarmer cells were allowed to proceed synchronously through the cell cycle at 30°C. Samples were removed at specific times and pulse-labeled for 10 min with 10 µCi of Tran³⁵S-label (ICN, Costa Mesa, Calif.). Cell extracts were prepared and immunoprecipitated as described by Gomes and Shapiro (27), using antibody to CAT or -galactosidase protein. Flagellin immunoprecipitation was used as a positive control for cell cycle-dependent expression. The cell cycle was also monitored by light microscopy. The immunoprecipitated proteins were resolved by SDS-10% polyacrylamide gel electrophoresis and visualized by autoradiography.

Nucleotide sequence accession numbers. The DNA sequences of flmAB, flmCD-flmEF, and flmGH operons described in this report have been assigned GenBank accession no. U27301, U27302, and U28867, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Isolation and characterization of flmC, flmD, flmE, and flmF genes. A cosmid, pLSG1, containing about 25 kb of C. crescentus chromosomal DNA was identified by complementation of the motility defect of strains SC1030 (flmE::Tn5), SC1127 (flmD::Tn5), and SC3090 (flmD::Tn5) (58). Plasmid pLSG1 could also complement the SC175 (flmE) and SC305 (flmD) mutants. Deletion analysis of pLSG1 revealed that a 14-kb EcoRI fragment retained in pSCC2 was able to restore motility to the flmD and flmE Tn5 insertion mutants. To define the locations of the flmD and flmE genes in pSCC2, DNA fragments were deleted or subcloned into plasmids that can replicate in C. crescentus (Fig. 1). Plasmids pSCC33, pGL2, and pGL38 could restore motility to the rec⁺ strains SC305, SC1127, and SC3090, but they were unable to complement the motility defect of the recombination-deficient strain SC3898 (recA526 flmD). These results indicate that the restored motility in the rec⁺ strains resulted from a recombinational event. Since pGL17 and pGL19 both complemented strain SC3898, and pGL18 and pSCC33 both failed to complement SC3898, we deduced that the flmD gene is contained within a fragment beginning 460 bp to the right of the SacIb site and ending at the HpaI site. By contrast, pSCC33 was able to complement both SC175 (flmE) and SC3899 (flmE recA), indicating that the flmE gene is present entirely within the 3.7-kb SacI fragment. Since both pGL18 and pGL38 were able to complement the recA flmE double-mutant strain SC3899, and since pGL19 failed to complement SC3899, we concluded that the flmE gene is located in a 0.8-kb fragment beginning 410 bp to the right of the HpaI site and ending at the NcoI site. The locations of the flmD and flmE genes were confirmed by Southern analysis of chromosomal DNA from Tn5 insertion mutants. Strains SC1127 and SC3090 contained a flmD::Tn5 insertion located in the 1.2-kb SalI fragment, and strain SC1030 had a flmE::Tn5 insertion located in the 0.8-kb SalI fragment (data not shown).

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Analysis of the flmCD and flmEF regions. Shown is a restriction map of plasmid pSCC2, which contains a 14-kb EcoRI DNA fragment of the C. crescentus chromosome. Subclones from this region in plasmid pRK2L1 or R300B were tested for the ability to complement the motility defect of strains SC305 (flmD148), SC175 (flmE102), SC3898 (flmC148 recA526 zzz::Tn5), and SC3899 (flmE102 recA526 zzz:Tn5). +, complementation of the motility defect; , failure to complement. Solid arrows represent predicted open reading frames and direction of transcription. Abbreviations: A, SacI; B, BamHI; E, EcoRI; N, NcoI; P, HpaI; S, SalI.

Isolation and characterization of the flmA, flmB, flmG, and flmH genes. The flmA and flmB genes (formerly designated flaA) were identified by complementation of strains SC229 (flmA104) and SC1128 (flmA::Tn5) by the cosmid clone pLSG6 that contained about 20 kb of C. crescentus chromosomal DNA (58). Pulsed-field gel electrophoresis and Southern analysis of chromosomal DNA from SC1128 revealed that the flmA and flmB genes were present on an 18-kb EcoRI fragment of chromosomal DNA (data not shown). Cosmid pLSG6 contained two EcoRI sites, one in the vector and a second in the cloned C. crescentus DNA. A subclone of the 5-kb EcoRI fragment of pLSG6, pSCC7, could complement the motility defect of both strains SC229 and SC1128. Deletion analysis of pSCC7 revealed that a 3.5-kb BamHI-EcoRI fragment (pNC1341) could fully complement strain SC229 (Fig. 2A). Analysis of additional subclones (pNC1355 and pNC1356) demonstrated that the DNA on both sides of the central SalI site was required for flmA complementation (Fig. 2A). Nucleotide sequence analysis of the entire 3.5-kb BamHI-EcoRI fragment (GenBank accession no. U27301) revealed two potential open reading frames with overlapping termination and initiation codons similar to those found in the flmCD and flmEF operons. The first open reading frame was designated flmA since it spanned the SalI site required for complementation.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   (A) Analysis of the flmAB region. Genetic organization of plasmid pSCC7 harboring the 5.0-kb EcoRI fragment is represented. Solid arrows represent open reading frames and direction of transcription. The ability to complement strain SC229 (flmA104) is shown. (B) Organization of the flmGH region. Solid arrows represent open reading frames and direction of transcription. + and  denote the ability and inability, respectively to swarm in a semisolid medium. Abbreviations: B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P, HpaI; S, SalI.

Database comparisons. The deduced amino acid sequences of the eight Flm proteins were compared to entries in the GenBank database. As shown in Table 3, FlmA, FlmB, FlmC, and FlmD show significant levels of identity (23 to 41% identity) with proteins involved in capsular, lipopolysaccharide (LPS), and spore coat polysaccharide biosynthesis from Bacillus subtilis, Methanococcus jannaschii, and other bacteria. Furthermore, FlmC also shows homology with the CMP-KDO syn- thetase (3-deoxy-manno-octulosonate cytidylyltransferase) involved in LPS biosynthesis in E. coli (8), Chlamydia trachomatis (67), and Haemophilus influenzae (21). In Helicobacter pylori, a FlmA homolog, FlaA1, has been sequenced and found to show 61% identity over a 321-amino-acid overlap (GenBank accession no. AE00595) (68). Since flm mutants produce flagellins with altered migration in SDS-polyacrylamide gels (30), these results suggest that FlmA, FlmB, FlmC, and FlmD could be involved in the glycosylation of flagellin monomers or other proteins involved in flagellin biogenesis.

Effect of flagellar mutations on the expression of flmA, flmC, flmE, and flmG fused to cat. Quon et al. (53) have proposed that CtrA binds directly to promoters containing the (TTAA-N₇-TTAAC) consensus site to activate the flagellar regulatory hierarchy, to prevent replication of DNA, and to control DNA methylation and cell division. To test whether CtrA regulates the expression of the flmAB, flmCD, flmEF, and flmGH operons, plasmids pGL48 (flmC::cat), pGL49 (flmE::cat), pGL50 (flmA::cat), pGL53 (flmG::lacZ), pCS91 (rsaA::lacZ), and pWZ162 (fliQ::lacZ) were mated into strain LS2195, which contains a temperature-sensitive ctrA401 mutation. The resulting constructs were grown in PYE medium under the permissive condition (28°C) and then shifted to the restrictive condition (37°C) for 6 h. Cell extracts were prepared, and CAT and -galactosidase activities were assayed (Table 4). As previously reported (53), expression of the fliQ::lacZ decreased about twofold in the ctrA401 background at the restrictive temperature, and the crystalline surface array protein promoter, rsaA::lacZ, was relatively unaffected by the ctrA401 mutation. However, the level of transcription of the rsaA::lacZ gene fusion showed a 2- to 2.5-fold increase at 37°C in both LS107 and LS2195 backgrounds (data not shown), suggesting that its expression is heat induced. More importantly, expression of the flmA::cat, flmE::cat, and flmG::lacZ gene fusion products was significantly (2.4- to 3.6-fold) reduced in the strain LS2195 (ctrA401) at the restrictive temperature. In contrast, flmC expression was relatively unaffected by the ctrA401 mutation at either temperature. These results suggest that CtrA positively regulates the flmA, flmE, and flmG promoters either directly or indirectly.

Temporal regulation of the flmA, flmC, flmE, and flmG genes. To determine whether expression of the flmA, flmC, flmE, and flmG genes is temporally regulated, strains containing a chromosomally inserted transcriptional cat fusion or a plasmid-borne lacZ fusion were synchronized and analyzed throughout the cell cycle. Expression of the flmA, flmE, and flmG operons occurred primarily in predivisional cells (Fig. 3). Transcription of both the flmA and flmG genes is very low or absent in swarmer cells and during the stalk-to-predivisional cell transition (0 to 0.5 cell division unit). It reaches a peak of expression in late predivisional cells (0.8 cell division unit) that coincides with the completion of filament assembly and the appearance of motility. The pattern of flmE expression is similar, but it shows a periodicity analogous to that observed for the 25-kDa flagellin where transcription continues in swarmer cells. In contrast, transcription of the flmC gene occurred throughout the cell cycle, with only twofold increase in predivisional cells (Fig. 3).

View larger version (39K): [in this window] [in a new window]   FIG. 3.   Cell cycle expression of the flmA, flmC, flmE, and flmG operons. Synchronized populations of Caulobacter strains SC3971, SC3973, SC4016, and SC4250 were pulse-labeled with [35S]methionine at 15-min intervals during the cell cycle. (A) Immunoprecipitation of labeled proteins with CAT, -galactosidase, or flagellin antibodies. A cartoon showing progress through the cell cycle is shown at the top. The cell cycle-dependent expression of the flagellin genes is shown as a control. (B) Quantification of these data by using a Alpha Innotech photodocumentation system. Percentage of maximal expression of each sample is shown as a function of cell division units. One cell division unit is equivalent to a generation time of 180 min. Closed squares represent 25-kDa flagellin expression (recovered from SC3973 cells carrying the flmE::cat fusion); open squares represent expression of the flmA::lacZ, flmC::cat, flmE::cat, or flmG::cat fusion.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The regulation of flagellum biogenesis is complex, and many details remain to be elucidated. The results presented above demonstrate that the flm genes represent a new class of flagellar genes. DNA sequence analysis revealed that the flmAB, flmCD, and flmEF genes are structurally organized as operons. In each operon, the termination codon of the first gene overlaps the initiation codon of the second gene. DNA sequence analysis of the flmG and flmH genes confirmed that they also are organized in an operon as reported by Schoenlein et al. (59). However, in this case, the two coding regions are separated by 152 bp. The close spacing of the genes in the flmAB, flmCD, and flmEF operons suggests that translation of these operons may be governed by a translational coupling mechanism. In E. coli, there are many examples of translational coupling where the interruption of translation of the first gene causes a severe decrease in the expression of the translationally coupled distal gene (1, 52, 77). For translational coupling to occur, the efficient expression of the distal genes would be dependent on both translation of the first gene and termination of this translation in close proximity to the start codon for the second gene. Thus, translational coupling could be a mechanism to ensure equimolar synthesis of both proteins.

Homology searches of the deduced amino acid sequences revealed that FlmA, FlmB, FlmC, and FlmD have significant levels of identity with proteins involved in capsular, LPS, and spore coat polysaccharide biosynthesis from B. subtilis, M. jannaschii, and other bacteria. However, since FlmC also shared homology to the CMP-KDO synthetase from E. coli, C. trachomatis, and H. influenzae, these results suggested that these proteins could be involved in LPS biosynthesis. To test this hypothesis, we measured the KDO synthetase enzyme activity in flmA, flmD, flmE, flmG, and flmH mutants. Each mutant had wild-type levels of KDO synthetase activity and appeared to have wild-type LPS profiles (37). Thus, it does not appear that mutations in the flm genes affect LPS biosynthesis. The other Flm (FlmEFGH) proteins show homology to proteins involved in glycosylation, methylation, and/or acetylation in several bacteria (Table 3). Mutations in flmA, flmD, flmE, flmG, and flmH genes result in the production of a 22-kDa flagellin. Furthermore, we have shown that the 22-kDa protein results from a modification or a breakdown product of the 25-kDa flagellin proteins (20). Glycosylation of flagellin proteins has been reported for Campylobacter (17), Spirochaeta aurantia (10), some archaea (35, 63, 72), and Azospirillum brasilense (46). Azospirillum contains flmAB homologs, and a mutation in one of these genes prevents assembly of the flagellar filament (45). In addition, Wieland et al. (72) have suggested that in halobacteria, glycosylation of the flagellins was necessary for proper incorporation of the flagella into the cell envelope and that overproduction of flagellins resulted in subunits with lower molecular weights. In Caulobacter, the 22-kDa flagellin is present in a flbT mutant that overproduces flagellins (57). Flagellins also can be modified by methylation (2, 12, 36, 38), phosphorylation (31), and sulfation (36, 72). Strains containing mutations in flmA, flmD, flmE, flmG, and flmH genes have a normal basal body and hook structure but fail to assemble a flagellar filament (30). Therefore, our current hypothesis is that modification of the flagellin subunits or some other flagellar proteins by glycosylation, acetylation, and methylation is required for proper assembly of flagellin subunits into the filament. Clearly, this hypothesis has important implications for the structure and mechanism of assembly of the flagellar filament. Recently, we have determined the nucleotide sequences of five of the six flagellin genes in C. crescentus (20). Analysis of deduced amino acids indicated that there is a discrepancy between the calculated molecular weight and the actual mass determined by mass spectroscopy (37).

It has been demonstrated in E. coli (32), Salmonella typhimurium (34), and C. crescentus (14, 48, 74) that a cascade of positive and negative transcriptional control regulates the temporal expression of flagellar genes. Previously, the flmAB, flmCD, flmEF, and flmGH operons had been placed in class III in the flagellar gene regulatory hierarchy (48). However, the experiments presented in this report demonstrate that the flm operons represent a new class of flagellar genes. First, we have shown that none of the flm operons require the RNA polymerase sigma factor 54 for transcription, indicating that they are not class III or IV genes (Table 5). Second, we have shown that flmAB, flmEF, and flmGH are positively regulated by CtrA (Table 4), a transcriptional response regulator that controls class II flagellar genes (53). However, in contrast to class II flagellar genes, mutations in flmAB, flmCD, flmEF, and flmGH operons do not cause defects in cell division. In addition, previous studies (3, 48, 74) have shown that the flmA, flmD, flmE, and flmG genes do not regulate transcription of genes from class II (fliF and flhA), class III (flgE, flgK, and flbG), or class IV (fljK and fljL). Taken together, we conclude that the flmAB, flmCD, flmEF, and flmGH operons belong to a new class of flagellar genes.

It has been shown that synthesis of the flagellin subunits encoded by the class IV genes is subject to posttranscriptional control mechanisms (3, 41, 42). Anderson and Newton (3) showed that both fljK::lacZ transcriptional and translational fusions were expressed at nearly wild-type levels in strains carrying mutations in flmA, flmD, or flmH. Nevertheless, immunoprecipitation experiments measuring short (30-s or 1-min) pulses of flagellin protein synthesis demonstrated that mutations in these genes do result in reduced levels of flagellin synthesis (30). Our current hypothesis is that this reduced level of flagellin synthesis may be due to a feedback mechanism involving unassembled flagellin subunits rather than any direct action involving the flm gene products. Furthermore, since we have shown that the FlbT product regulates flagellin synthesis by altering mRNA stability (42), it is likely that the effects of flm mutations on flagellin gene expression involve mRNA stability as well.