Present address: Department of Microbiology, Harvard Medical School, Boston, MA, USA.
The conserved flaF gene has a critical role in coupling flagellin translation and assembly in Caulobacter crescentus
Article first published online: 5 JUL 2005
Volume 57, Issue 4, pages 1127–1142, August 2005
How to Cite
Llewellyn, M., Dutton, R. J., Easter, J., O'Donnol, D. and Gober, J. W. (2005), The conserved flaF gene has a critical role in coupling flagellin translation and assembly in Caulobacter crescentus. Molecular Microbiology, 57: 1127–1142. doi: 10.1111/j.1365-2958.2005.04745.x
- Issue published online: 5 JUL 2005
- Article first published online: 5 JUL 2005
- Accepted 25 May, 2005.
The expression of the flagellin proteins in Caulobacter crescentus is regulated by the progression of flagellar assembly both at the transcriptional and post-transcriptional levels. An early basal body structure is required for the transcription of flagellin genes, whereas the ensuing assembly of a hook structure is required for flagellin protein synthesis. Previous experiments have shown that the negative regulatory protein, FlbT, operates this second post-transcriptional checkpoint by associating with the 5′ untranslated region (UTR) of the fljK flagellin transcript, inhibiting translation and destabilizing the mRNA. In this paper we examine the role of flaF in flagellar biogenesis. The flaF gene, which is conserved in several speices of flagellated α-proteobacteria, is required for motility and flagellin protein synthesis. A deletion of flbT in a ΔflaF strain restored flagellin protein expression, but not motility, indicating that FlaF functions in filament assembly. Mutant strains with a deletion in flaF had no detectable fljK mRNA, the levels of which were restored by an additional mutation in flbT. Assay of fljK gene expression using transcription and translation reporter fusions indicated that FlaF was essential for the translation of fljK mRNA. FlaF protein levels were under cell cycle control, peaking during the period of flagellin expression and filament assembly, whereas FlbT was present throughout the cell cycle. These results suggest that FlbT and FlaF activities oppose one another in the regulation of flagellin expression in response to both the progression of flagellar assembly and the cell cycle.
The dimorphic bacterium Caulobacter crescentus synthesizes a single polar flagellum once every cell division cycle. The ensuing cell division gives rise to distinct cell types: a motile, flagellum-bearing swarmer cell and a cell with a polar stalk appendage (reviewed in England and Gober, 2001; Ausmees and Jacobs-Wagner, 2003; Quardokus and Brun, 2003; Ryan and Shapiro, 2003). The progeny swarmer and stalked cells exhibit different programmes of gene expression and DNA replication. In the newly formed stalked cell, DNA replication initiates immediately, whereas replication is repressed in the swarmer cell type for a distinct period when the cells are grown in culture. After this distinct period, the flagellum is shed, DNA replication initiates and a stalk is formed at the former site of the flagellum.
Flagellar biogenesis in C. crescentus requires up to 50 genes, most of which are expressed at a defined time in the cell cycle (reviewed in Gober and England, 2000; England and Gober, 2001). Two distinct regulatory checkpoints control the expression of flagellar genes in response to the progression of flagellar assembly, resulting in genes being expressed in the approximate order that their products are assembled into the nascent structure. This trans-acting regulatory hierarchy of gene expression commences shortly after the initiation of DNA replication in the newly formed stalked cell (Fig. 1). The first flagellar genes (class II) to be expressed in the cell cycle are those encoding the cytoplasmic membrane MS-ring, the flagellar switch and a flagellum-specific type III secretion (TTS) system. The assembly of this early flagellar substructure is required for the transcription of the remaining flagellar structural genes (class III and class IV) (Newton et al., 1989; Xu et al., 1989; Ramakrishnan et al., 1994; Mangan et al., 1995). This assembly checkpoint operates by regulating the activity of the global transcriptional activator FlbD (Mangan et al., 1995; Muir et al., 2001; Muir and Gober, 2002). In the absence of a class II-encoded flagellar structure, FlbD is unable to activate the transcription of class III and class IV flagellar genes. Flagellar assembly influences FlbD activity through a conserved regulatory partner protein called FliX (Muir et al., 2001; Muir and Gober, 2002; 2004). FliX physically interacts with FlbD and represses transcriptional activation in the absence of a class II flagellar gene-encoded structure (Muir and Gober, 2004).
The FliX/FlbD-regulated class III flagellar genes encode components of the distal rod, the outer membrane rings and the hook structure. Comparable to the assembly checkpoint coupling the early class II-encoded MS-ring/switch/TTS system complex, the complete assembly of the hook–basal body structure is required for the expression of the flagellin genes (class IV), which comprise the filament (Mangan et al., 1995; Anderson and Newton, 1997). Two major flagellin-encoding genes, fljK and fljL, have been shown to be transcribed in the absence of a completed hook structure; however, the cells do not accumulate significant levels of flagellin protein. Therefore, this second flagellar assembly checkpoint in C. crescentus functions post-transcriptionally to regulate flagellin expression. The flbT gene has been demonstrated to have a critical role in the operation of this assembly checkpoint (Mangan et al., 1999; Anderson and Gober, 2000). Strains with mutations in flbT, a gene located within a cluster of flagellar genes in C. crescentus, have been shown to accumulate high levels of flagellin protein (Schoenlein and Ely, 1989; Schoenlein et al., 1992). The introduction of a flbT mutation into strains that cannot assemble a flagellar hook structure resulted in the complete restoration of flagellin protein expression, suggesting that FlbT functioned as a negative regulator of flagellin expression in the absence of a fully assembled hook (Mangan et al., 1999). Hook mutants of Caulobacter possess reduced levels, and stability, of flagellin mRNA. In contrast, these mutant strains containing an additional mutation in flbT, exhibit remarkable flagellin mRNA stability with half-lives exceeding one-third of the cell cycle (Mangan et al., 1999; Anderson and Gober, 2000). This observation suggests that FlbT-mediated negative regulation results in enhanced mRNA degradation. Experiments using translational fusions of the fljK flagellin to a reporter lacZ gene have shown that FlbT can exert its repressive effect on gene expression with fusions containing as little as 14 codons of fljK mRNA plus an additional 63 nucleotide 5′ untranslated region (5′ UTR). This segment of fljK mRNA is predicted to fold into two different secondary structures, including one that would occlude the ribosome binding site via base pairing (Anderson and Gober, 2000). Using cell extracts, FlbT has been shown to bind to a loop region of the fljK message that may be important in stabilizing a secondary structure that would not favour translation. It is hypothesized that FlbT binds to flagellin mRNA in the absence of a completed flagellar hook structure and inhibits translation, and this results in destabilization of flagellin mRNA (Anderson and Gober, 2000). It is currently not known how FlbT ‘senses’ the completion of the external hook structure, thus permitting the translation of flagellin mRNA.
The genome of C. crescentus contains two clusters of flagellin-encoding genes (Schoenlein et al., 1992; Ely et al., 2000). One of these, the α-cluster, contains genes encoding three flagellins of differing sequences and molecular mass (Fig. 2A). The fljK gene encodes a 25 kDa flagellin that is essential for motility (Minnich et al., 1988). Upstream of the fljK gene is fljL, encoding a 27 kDa flagellin that has been shown to be important, but not essential for motility. Lying downstream of fljK is the fljP (formerly fljJ) gene encoding a 29 kDa flagellin that, although presumed to be contained within the flagellar filament, is not essential for wild-type motility. This region of the genome also contains genes of unknown function that are essential for motility, including flaEY, genes encoding flagellin modifying enzymes, and flbT (Schoenlein and Ely, 1989; Schoenlein et al., 1989; 1992). Downstream of flbT lies the flaF gene, predicted to encode an 11.5 kDa polypeptide of unknown function (Fig. 2A). Previous genetic experiments analysing point mutations and utilizing a plasmid complementation analysis of larger deletions in this α-flagellin region had indicated that flaF was required for motility and for wild-type levels of flagellin proteins (Johnson and Ely, 1979; Johnson et al., 1983; Schoenlein and Ely, 1989; Schoenlein et al., 1992).
In this report, we investigate the post-translational regulation of the Caulobacter fljK flagellin gene. Mutant strains with a deletion in flaF were non-motile and failed to express flagellin genes. The introduction of a flbT deletion into a ΔflaF strain restored flagellin expression but did not confer motility, suggesting that flaF is required for flagellin secretion or assembly of the filament structure. The lack of flagellin expression in the ΔflaF strain was the result of a marked decrease in the abundance of flagellin mRNA, possibly the result of a block in translation. The cell cycle pattern of FlaF protein abundance indicates that FlaF functions during filament assembly in swarmer cells.
The flaF gene is required for flagellin synthesis and motility
Previous experiments have demonstrated that the C. crescentus flaF gene was essential for motility and flagellin synthesis (Johnson and Ely, 1979; Schoenlein and Ely, 1989; Schoenlein et al., 1989; 1992). As the flaF gene lies directly downstream of the regulatory gene flbT, we wanted to explore the role of flaF in regulating flagellin synthesis, as well as its possible influence on FlbT activity. As the previously characterized strains with flaF mutations were either point mutations or deletions that extended into adjacent genes in the region of the α-flagellin cluster (Johnson and Ely, 1979; Johnson et al., 1983; Schoenlein and Ely, 1989; Schoenlein et al., 1992), we created a strain containing only a deletion in the flaF coding sequence (see Experimental procedures). When wild-type cells were inoculated into semi-solid motility agar, a diffuse ring of growth was observed, indicative of motile and chemotaxis-competent cells that have swum away from the point of inoculation (Fig. 2B). When the ΔflaF cells were inoculated into this medium they formed a compact colony indicative of a motility defect (Fig. 2B). Furthermore, phase-contrast light microscopy of this strain revealed cells of normal cellular morphology that exhibited a complete lack of motility (data not shown). Additionally, immunoblot analysis using anti-FlaF antisera revealed an absence of the FlaF protein in the ΔflaF strain (data not shown). The motility defect of the ΔflaF strain could be complemented by a fragment of C. crescentus genomic DNA containing the entire flbT and flaF coding sequences plus 199 bp of DNA upstream of the flbT initiation codon (Fig. 2A) present in trans on a multicopy plasmid (pML1) (Fig. 2B). Deletion of the DNA upstream of flbT resulted in a plasmid (pML2) (Fig. 2A) that was unable to complement the motility defect in the ΔflaF strain, suggesting the flbT and flaF constitute an operon (Fig. 2C). Next we tested whether a deletion in the regulatory gene, flbT, would affect the motility phenotype of the ΔflaF strain. In order to create a mutant strain containing deletions in both flbT and flaF, we performed inverse polymerase chain reaction (PCR) on a plasmid containing the flbTflaF operon, essentially as described above, instead using primers that resulted in deleting DNA from codon number 91 in flbT to codon 13 in the flaF coding region (Fig. 2A). The wild-type DNA in this region of the chromosome was then replaced with the deleted DNA by allelic exchange. This mutant strain failed to produce both FlbT and FlaF when assayed by immunoblot, confirming that the genes encoding both of these proteins were non-functional (data not shown). The presence of the additional deletion in flbT in the ΔflaF strain, could not restore motility as assayed both by growth in motility agar (Fig. 2B) and by light microscopy (data not shown). Note that strains containing a single mutation in flbT (i.e. flbT650) are motile (Schoenlein and Ely, 1989), but exhibit a swarm size that is significantly smaller than that generated by wild-type cells (Fig. 2B). As was the case with the ΔflaF strain, the presence of wild-type flbT and flaF in trans on a multicopy plasmid (pML1) restored the motility defect of the ΔflaF/ΔflbT strain (Fig. 2B).
We next tested the effect of a flaF deletion on the levels of flagellin protein using immunoblot analysis (Fig. 2D). As reported previously for strains bearing point mutations in flaF (Johnson and Ely, 1979; Schoenlein and Ely, 1989; Schoenlein et al., 1992) both the 27 kDa and 25 kDa flagellins were undetectable in the ΔflaF strain. Flagellin protein levels were restored in this strain by the complementing multicopy plasmid, pML1 (Fig. 2D). (Note that this antisera preparation does not effectively recognize the 29 kDa flagellin in an immunoblot assay.) Previous experiments have demonstrated that a mutation in flbT could restore flagellin protein levels in mutant strains that exhibit flagellar assembly defects (Mangan et al., 1999; Anderson and Gober, 2000). Likewise, in the ΔflaF/ΔflbT double mutant strain, flagellin protein levels were completely restored (Fig. 2D). Interestingly, this strain, similar to mutant strains containing a single mutation in the flbT gene, produced relatively large amounts of a protein band of apparent lower molecular mass that reacted with the anti-flagellin antibody (labelled as an asterisk in Fig. 2D). This approximately 22 kDa polypeptide has been shown to be derived from the 25 kDa flagellins (Ely et al., 2000) and thus may be viewed as degradation product that is markedly abundant in strains deficient in flbT. The presence of wild-type copy of flbT on the multicopy pML1 plasmid resulted in a wild-type complement of flagellin produced by the ΔflaF/ΔflbT strain (Fig. 2D).
These results, consistent with those reported previously (Schoenlein and Ely, 1989; Schoenlein et al., 1992), indicated that flaF was required for the synthesis of both the 27 kDa and the 25 kDa flagellins, and that the lack of these in the ΔflaF mutant strain was likely resulting in a motility defect. Interestingly, the presence of an additional mutation in flbT restored flagellin synthesis, but could not restore motility (Fig. 2), suggesting that flaF is required for either proper flagellum assembly or flagellar rotation. In order to distinguish between these two possibilities, we examined the morphology of the flagellar structure in the strain containing a deletion in flaF using electron microscopy (Fig. 3A). The ΔflaF mutant cells possessed a partially assembled flagellar structure containing an intact hook but lacking the flagellar filament (Fig. 3A). This phenotype is similar to that observed in several other C. crescentus non-motile strains containing mutations in other unlinked flagellar genes (Johnson et al., 1979). The absence of a filament in the ΔflaF strain was not surprising as these cells do not contain detectable levels of flagellin protein. We also performed electron microscopy on cells containing the double ΔflaF/ΔflbT deletion. Although this strain produced abundant levels of flagellin protein, there were no observable flagellar filaments indicating that flaF, in addition to being required for flagellin synthesis, is also required for filament assembly (Fig. 3B). One remarkable property of flbT mutant strains is the frequent occurrence of cells possessing an intact flagellum near the base of the stalk (Driks et al., 1990) (Fig. 3C). Interestingly, in cells containing the double ΔflaF/ΔflbT deletion, a partial flagellar structure containing an intact hook but no filament was frequently observed at the base of the stalk (Fig. 3B).
Effect of flaF on flagellin protein and mRNA stability
An interesting characteristic of the ΔflaF mutant strain is the absence of detectable flagellin protein in cells that contain an intact hook structure. Previous epistasis experiments have established that the expression of flagellin-encoding genes follows the completion of the hook structure (Mangan et al., 1995; Anderson and Newton, 1997). Consequently, mutations that block hook assembly abolish flagellin gene expression. The data presented here suggest that the flaF gene product is required for either flagellin expression or filament assembly following synthesis of the hook. An analysis of the predicted primary amino acid sequence of FlaF for signal sequences and transmembrane segments using the PSORT algorithm (Nakai and Kanehisa, 1991) (http://psort.nibb.ac.jp/), indicated that FlaF was likely to reside in the cytoplasm. One plausible cytoplasmic function for FlaF in flagellar assembly may be to act as a secretory chaperone for flagellin protein. The secretory chaperones of TTS systems, including those of the flagellum, are proposed to bind secreted substrates, either protecting them from degradation and/or polymerization prior to secretion (reviewed in Parsot et al., 2003). Thus, the phenotype of the C. crescentus flaF mutant strain is consistent with a deficiency in a secretion chaperone, exhibiting low cytoplasmic levels of flagellin and a defect in flagellar assembly. In order to test the idea that FlaF may function as secretory chaperone for flagellin we compared the stability of flagellin protein in the ΔflaF strain with that of wild-type cells in a pulse-chase labelling experiment. Protein samples were collected over time following the chase, and immunoprecipitated flagellin was subjected to SDS-PAGE. Three flagellins of distinct molecular mass, 29, 27 and 25 kDa, were detectable in this experiment when extracts were prepared from labelled wild-type cells (Fig. 4A). The major 25 kDa flagellin was by far the most abundant and exhibited a half-life of approximately 10 min. The 29 and 27 kDa flagellins were present in quantities too low to determine an accurate stability measurement. We were unable to assay the stability of the 25 kDa flagellin in the ΔflaF mutant strain, as these cells failed to synthesize detectable amounts of this protein (Fig. 4B). As was reported in previous experiments, the ΔflaF mutant strain exhibited an increased rate of synthesis of the minor FljP 29 kDa flagellin (Schoenlein and Ely, 1989). This result suggests that flagellar assembly, and perhaps FlaF, negatively regulates the synthesis of FljP. As we were unable to determine the stability of the 25 kDa flagellin in the ΔflaF mutant strain, we performed the same experiment on cells containing the double ΔflaF/ΔflbT mutation, which contain abundant steady-state levels of flagellin protein (see Fig. 2C). In cells containing a single mutation in flbT there was no significant difference in the stability of the 25 kDa flagellin compared with wild-type cells; however, the 22 kDa flagellin degradation product was present in detectable quantities (Fig. 4C). The ΔflaF/ΔflbT cells exhibited a rate of 25 kDa flagellin synthesis that was approximately half of that of the flbT mutant strain; however, the stability of this flagellin was comparable to that observed with wild-type cells (Fig. 4D). This result indicates that FlaF may be required for maximal rates of 25 kDa flagellin synthesis, but is not required for flagellin protein stability. Based on these results we propose that FlaF probably does not function as a secretion chaperone for flagellin protein.
We next examined the effect of flaF on flagellin mRNA stability as previous experiments have shown that mutant strains that are unable to secret flagellin (i.e. those deficient in hook assembly) possess a markedly decreased stability of flagellin mRNA (Mangan et al., 1999; Anderson and Gober, 2000). In order to accomplish this, RNA synthesis was inhibited by the addition of rifampicin and RNA samples were obtained at this time (t = 0 min) and following 10 min of incubation. As reported previously, the 25 kDa flagellin fljK mRNA is a relatively long-lived transcript with a half-life of approximately 10 min (Fig. 4E) (Mangan et al., 1999; Anderson and Gober, 2000). (Note that mRNA half-life values were estimated from the data in Fig. 4E). In contrast, fljK mRNA was not detectable in the ΔflaF mutant strain (Fig. 4E). FlbT has been shown to function as a negative, post-transcriptional regulator of flagellin expression in the absence of a completed hook structure (Mangan et al., 1999; Anderson and Gober, 2000). We wished to determine whether FlbT was also functioning to negatively regulate fljK mRNA levels in the ΔflaF mutant strain. Cells with mutations in flbT exhibit an increase in the steady-state levels of fljK mRNA, as well as a marked increase in message stability (see Fig. 4E). The presence of a flbT mutation in flaF-deficient cells restored the levels of fljK mRNA to greater than those of wild-type cells, but less than those of flbT mutant cells containing a wild-type copy of flaF (Fig. 4E). Additionally, the stability of fljK mRNA in the ΔflaF/ΔflbT cells resembled that of wild-type cells. These results suggest that FlaF has a critical role in maintaining the stability and possibly the rate of synthesis of fljK mRNA.
FlaF is required for the translation of the fljK 25 kDa flagellin
We next wanted to determine the cause of the lack of fljK mRNA in ΔflaF mutant cells. Translational lacZ reporter fusions were introduced into the ΔflaF mutant and the measured β-galactosidase was compared with that assayed in wild-type, as well as flbT and ΔflaF/ΔflbT mutant strains. Additionally, we tested the effect of deletions of the fljK 5′ UTR, and/or coding region, on lacZ reporter gene expression in the wild-type and flagellar mutant strains (Fig. 5A). The expression of a fljK::lacZ protein fusion (fljK::lacZ) was markedly reduced in the ΔflaF strain generating β-galactosidase activity that was 10% of that assayed in wild-type cells (Fig. 5B). In C. crescentus strains possessing mutations in flagellar structural genes, fljK protein fusions are also not efficiently expressed, and previous experiments have shown that this is a consequence of negative regulation by FlbT (Mangan et al., 1999; Anderson and Gober, 2000). FlbT also functions as a negative regulator of fljK::lacZ expression in wild-type cells, as in this case, the reporter gene generated β-galactosidase activity over twice that of wild-type cells (Fig. 5B). Combining the flbT deletion with the ΔflaF mutation restored fljK::lacZ expression to levels greater than wild-type (6681 versus 4566 units), but not as high as those measured in the strain containing a single flbT mutation.
When a flbT mutation was present alone, or combined with a flagellar structural gene mutation, the resulting increase in fljK expression could be observed using reporter fusions that contained an intact 5′ UTR (Mangan et al., 1999). This 67 nucleotide 5′ UTR along with the first 14 codons of the fljK transcript is predicted to form two different secondary mRNA structures that have been proposed to influence the efficiency of translation (Anderson and Gober, 2000). We tested the effect of a flaF mutation on several fljK::lacZ reporter fusions containing either deletions in the coding sequence or the 5′ UTR (Fig. 5A). We found that although the presence of the flbT mutation in a flaF mutant restored the expression of fusions containing either the first 14 codons of fljK (fljK1) or no fljK coding region (fljK2) (viz. containing only the 5′ UTR plus the ATG), the levels of β-galactosidase activity generated were significantly lower than the corresponding reporters in either the flbT mutant strain or wild-type cells (Fig. 5B). In contrast, in the ΔflaF/ΔflbT double mutant strain there was no expression of a fusion with a deletion of the entire 5′ UTR but containing an intact fljK coding sequence (fljK3) (Fig. 5B). This result is similar to previous experiments showing that the 5′ UTR is important for efficient translation (Mangan et al., 1999).
FlaF and FlbT influence the transcription of the fljK 25 kDa flagellin
A lack of transcription could be one possible mechanism to account for the absence of fljK::lacZ expression in the ΔflaF mutant strain. In order to determine whether this was occurring, we introduced a fljK–lacZ transcription reporter fusion into wild-type cells, as well as into the flbT650, ΔflaF and ΔflaF/ΔflbT mutant strains. This fusion, fljK–lacZ/290 (Fig. 6A), contained the fljK promoter plus the entire 5′ UTR and the first 23 codons of fljK fused upstream of a promoterless lacZ gene. In this particular reporter plasmid (placZ/290) there lies approximately 190 bp of vector DNA upstream of the 5′ end of lacZ (Gober and Shapiro, 1992). Importantly, the presence of this intervening DNA permits the translation of lacZ even when translation of the upstream DNA is inhibited (i.e. in class III flagellar genes). In previous experiments, we have found that this reporter permits the assay of transcription even when the naturally occurring flagellin mRNA is unstable as a consequence of inhibition of translation (Anderson and Gober, 2000). The expression of this fljK–lacZ transcription fusion in the ΔflaF mutant strain generated slightly (25%) greater β-galactosidase activity than in wild-type cells (6219 versus 4810 units), indicating that flaF was not required for transcription of the 25 kDa flagellin fljK gene (Fig. 6A). Thus, the marked decrease in both flagellin protein levels and the expression of fljK::lacZ protein fusions in the ΔflaF mutant strain is likely to be attributable to a failure to efficiently translate fljK mRNA. Interestingly, in a flbT mutant, the fljK–lacZ transcription fusion (fljK–lacZ/290) generated approximately ninefold lower β-galactosidase activity than that measured in wild-type cells, suggesting that FlbT influences promoter activity (Fig. 6A). In support of this idea, fljK promoter activity was also comparably reduced in the double mutant ΔflaF/ΔflbT strain (Fig. 6A).
As previous results have indicated that FlbT exerts its effect on fljK translation through an interaction with the 5′ UTR and coding sequence of the mRNA, we determined whether these same sequences were required for FlbT to influence fljK promoter activity. In order to accomplish this we constructed a lacZ transcriptional reporter fusion that was missing the entire 5′ UTR and fljK coding sequences (fljKp–lacZ/290) (Fig. 6B). The β-galactosidase activity generated from this fusion was not significantly different in strains deficient in flbT (i.e. flbT650 and ΔflaF/ΔflbT) from that assayed in wild-type cells (Fig. 6B). These results indicate that FlbT may function as positive regulator of fljK transcription and requires an intact 5′ UTR, as well as some number of fljK codons, to exert its maximal effect.
Cell cycle regulation of flbT and flaF expression
In C. crescentus, the flagellar filament is assembled in predivisional cells just prior to cell division. Following cell division, in newly formed swarmer cells the filament continues to lengthen, supplied by both an abundant pool of 25 kDa flagellin and the continuous translation of the relatively stable fljK mRNA (Milhausen and Agabian, 1983). As both FlbT and FlaF have a significant influence on flagellin expression, we determined their cell cycle pattern of expression throughout the cell cycle using synchronized populations of C. crescentus cells. In order to assay the cell cycle transcription of flbT and flaF, swarmer cells containing a flbT–lacZ transcription reporter fusion were isolated, suspended in fresh culture medium, and the population was permitted to proceed synchronously through the cell cycle. At various times, proteins were pulse-labelled and β-galactosidase was immunoprecipitated. The expression of the flbT–lacZ reporter fusion was under cell cycle control with expression occurring in swarmer cells, a decline in promoter activity upon differentiation into stalked cells, followed by a return of transcriptional activity in predivisional cells (Fig. 7A). The cell cycle pattern of FlaF and flagellin synthesis paralleled the peak in flbT promoter activity (Fig. 7A). (Note that we were unable to immunoprecipitate sufficient quantities of labelled FlbT in order to assay the cell cycle pattern of FlbT synthesis.)
We also assayed the steady-state levels of both FlaF and FlbT by immunoblot analysis of synchronized populations of cells at various points during the cell cycle. FlaF protein levels changed throughout the course of the cell cycle (Fig. 7B), reflecting its cell cycle pattern of protein synthesis. FlaF was relatively abundant in swarmer cells and declined to low levels when the swarmer cells differentiated into stalked cells. FlaF protein levels increased rather abruptly, and then peaked in late predivisional cells (Fig. 7B). This cell cycle pattern of FlaF protein abundance parallels the temporal pattern of flagellar filament assembly. The cell cycle pattern of FlbT protein levels was notably different than that of FlaF and the flagellins (Fig. 7B). Unlike these proteins, the cellular levels of FlbT did not significantly fluctuate throughout the cell cycle. Thus, FlaF exhibits a temporal pattern of protein abundance, the timing of which corresponds to its proscribed function in positively regulating flagellin expression and filament assembly. The persistence of FlbT throughout the cell cycle suggests that its activity is either regulated in a temporal fashion, or antagonized by a factor that promotes translation such as FlaF.
The temporal assembly of a polar flagellum once every C. crescentus cell cycle demands that regulatory networks simultaneously co-ordinate the progression of several morphogenetic events. In this paper, we have investigated the novel flaF gene, and its role in the temporal and cell-type-specific expression of flagellins. We have found that flaF mutants are non-motile, do not possess a filament structure, and similar to previously reported experiments (Johnson and Ely, 1979; Schoenlein and Ely, 1989; Schoenlein et al., 1989; 1992), synthesize little or no 25 kDa flagellin. The 25 kDa flagellin gene, fljK, is transcribed in the absence of FlaF but the mRNA is not translated, possibly leading to enhanced degradation of the message. The defect in fljK translation is probably a result of negative regulation by FlbT as a double ΔflaF/ΔflbT mutant strain synthesized wild-type levels of flagellin protein.
In a flaF mutant strain there was a marked decrease in the abundance of flagellin mRNA. Flagellin mRNA could be restored to levels approaching those found in wild-type cells by the introduction of an additional mutation in the regulatory gene flbT. FlbT has been shown to operate a regulatory checkpoint in Caulobacter linking the assembly of a completed hook structure to the expression of flagellins (Mangan et al., 1999; Anderson and Gober, 2000). In strains containing mutations in any class III flagellar structural gene, the genes encoding the flagellins are transcribed but the mRNA is not translated. Mutations in flbT can restore flagellin protein expression in class III mutant strains, demonstrating that it functions as a negative post-transcriptional regulator of flagellin expression. FlbT in cell extracts has been shown to bind to the 25 kDa flagellin mRNA in vitro (Anderson and Gober, 2000). Computational analysis of the fljK mRNA secondary structure including its 63 nt 5′ UTR predicts that it could adopt two equally energetically favourable conformations. One of these conformers would be predicted to favour translation, whereas the other would obscure the ribosome binding site via base pairing, and thus inhibit translation. It has been proposed that in the absence of a completed flagellar hook structure, FlbT binds to the 5′ UTR of the fljK transcript, stabilizes the structural conformation that disfavours translation, and this results in mRNA degradation (Anderson and Gober, 2000). It is not known how the completion of the hook structure influences the activity of FlbT, but is speculated to involve another protein that may bind in complex with FlbT to the fljK mRNA. In support of this idea is the observation that FlbT binding to the fljK transcript could only be demonstrated using cell extracts and not purified FlbT protein (Anderson and Gober, 2000).
Several of the experiments presented here suggest that FlaF may oppose FlbT-mediated repression by positively regulating the translation of fljK mRNA. Cells deficient in flaF can assemble a hook structure and yet are unable to translate flagellin mRNA. Therefore, FlaF functions at the transition between the completion of the hook structure and assembly of the flagellar filament. FlaF does not function solely in a regulatory role, however, as the ΔflaF/ΔflbT mutant strain, which was capable of synthesizing flagellin protein, did not possess a flagellar filament. Thus, FlaF functions as an activator of flagellin translation and, additionally, may be required for secretion or filament assembly. Based on these, and earlier findings, we propose that FlbT and FlaF operate a regulatory checkpoint coupling hook assembly to flagellin translation.
In addition to regulating fljK translation, FlbT also has a significant effect on fljK promoter activity. Interestingly, strains containing mutations in class III flagellar structural genes and flaF exhibit an increase in fljK promoter activity (Mangan et al., 1999; Anderson and Gober, 2000). This effect is not a result of an increased activity of the transcription factor, FlbD, as other FlbD-dependent promoters exhibit wild-type levels of activity in class III mutant strains (Mangan et al., 1995). Earlier experiments (Anderson and Gober, 2000), and those presented here, indicate that FlbT is required for the observed increase in fljK promoter activity in class III mutant strains. Mutational analysis of the 5′ UTR combined with mRNA binding studies using FlbT, showed that mutations that interfered with either FlbT binding or the predicted secondary structure of the fljK mRNA resulted in lessening the influence of FlbT on promoter activity (Anderson and Gober, 2000). These observations suggest that the positive influence of FlbT on fljK transcription requires the 5′ UTR. In support of this idea, we show here that a fljK–lacZ transcription reporter that is lacking transcribed fljK DNA (i.e. 5′ UTR and leader peptide) is unaffected by the absence of FlbT. We hypothesize that the 5′ structure of the fljK transcript may be functioning to terminate transcription. In this scenario, FlbT upon binding to the transcript would shift the structure into one favouring transcription, thus acting as an antitermination factor. The regulatory implications are that when FlbT is less active, under conditions where translation and flagellin secretion are favoured, there is a negative effect on fljK transcription. Therefore, a dynamic interplay exists between structural assembly, transcription, translation and flagellin secretion.
The regulatory activities of FlbT and FlaF are not restricted to the major 25 kDa and 27 kDa flagellins. Previous experiments have also shown that the rate of 29 kDa flagellin (fljP, formerly fljJ) protein synthesis increased significantly in strains containing point mutations in flaF and decreased in flbT mutant cells (Schoenlein and Ely, 1989). Likewise, the experiments presented here show that fljP is regulated by the opposing activities of FlbT and FlaF. In the experiments presented here, FljP protein synthesis occurred at rate that was fivefold greater in a flaF mutant than in wild-type cells. In strains containing a flbT mutation, FljP protein was not detectable. Early studies have also demonstrated that the 29 kDa flagellin exhibits an enhanced rate of synthesis in class II and class III flagellar mutant strains and, thus, is considered to be regulated independently from the flagellar trans-acting regulatory hierarchy (Champer et al., 1987; Schoenlein and Ely, 1989). Therefore, fljP expression is regulated by FlaF and FlbT in a fashion that is the mirror opposite of 25 kDa flagellin regulation. It will be interesting to determine whether FlaF/FlbT-mediated regulation of fljP occurs at the transcriptional and/or translational level, as well as its significance to flagellar biogenesis.
This hook–flagellin assembly checkpoint is likely to operate similarly in other bacteria belonging to the α-proteobacteria as both flbT and flaF homologues are present in the completed genomes of all flagellated organisms in this group of bacteria (Fig. 8). Thus, FlaF and FlbT constitute an evolutionarily conserved regulatory pair that controls flagellin synthesis in response to the progression of flagellar assembly in a fashion that is distinct from those mechanisms operating the analogous checkpoint in enteric bacteria (i.e. FlgM/σ28). Interestingly, the subset of polarly flagellated α-proteobacteria contain an additional conserved flagellar assembly checkpoint linking the completion of the MS-ring/switch/TTS system complex to the expression of later flagellar genes. This checkpoint is operated by another conserved regulatory pair, FliX and FlbD, and in contrast to FlbT/FlaF-mediated regulation, controls transcription in response to early flagellar assembly events (Mangan et al., 1995; Muir et al., 2001; Muir and Gober, 2002; 2004).
We found that flbT was transcribed under cell cycle control; however, the steady-state levels of FlbT protein did not change during the cell cycle. This is similar to case of other regulatory genes such as fliX, flbD, parA and parB, all of which exhibit a peak of transcription in predivisional cells, yet have steady-state levels of their encoded proteins that do not change during the course of the cell cycle (Wingrove et al., 1993; Mohl and Gober, 1997; Muir and Gober, 2002). In the case of FlaF, the steady-state levels and the rate of synthesis throughout the cell cycle paralleled that of flbT promoter activity. The pattern of FlaF abundance, appearing in predivisional and swarmer cells, is similar to that described for a number of C. crescentus flagellar proteins, including flagellins, hook and the MS-ring subunits (reviewed in Gober and England, 2000).
FlaF and FlbT are required to regulate flagellin synthesis at two distinct times in the C. crescentus cell cycle, and in actuality, in two distinct cell types. In predivisional cells, prior to hook assembly, FlbT would function to repress fljK mRNA translation. Derepression, requiring FlaF, would occur in this cell type following the completion of the hook structure, and may possibly be regulated by hook structural intermediates as described above. Late in the C. crescentus cell cycle, in predivisional cells, fljK transcription is restricted to the swarmer cell compartment, as a consequence of the swarmer pole-specific activation of the transcription factor, FlbD (Wingrove et al., 1993; Wingrove and Gober, 1994; Muir and Gober, 2002). Polar transcription of this flagellin gene results in progeny swarmer cells containing a supply of fljK mRNA; this mRNA continues to be translated, and thus the filament continues to lengthen for what is thought to be the lifetime of the swarmer cell (Milhausen and Agabian, 1983). Upon differentiation into a stalked cell, flagellin translation ceases, which is an event that requires FlbT (Mangan et al., 1999). In FlbT mutants, the mRNA of fljK is so long-lived that flagellin protein synthesis continues in stalked and early predivisional cells (Mangan et al., 1999). The cell cycle experiments presented here now reveal a possible mechanism to account for this developmental regulation of flagellin protein synthesis. Although FlbT protein levels do not change during the cell cycle, the peak in the cellular levels of FlaF parallel the time in the cell cycle when flagellins are translated and assembled into the filament structure. Specifically, FlaF levels are relatively high in swarmer cells, and then sharply decrease in stalked cells, when filament assembly ceases. The loss of FlaF in stalked cells, combined with the persistence of the negative regulator, FlbT, would result in the repression of flagellin translation. Thus, C. crescentus regulates flagellin translation by two distinct cell-type-dependent mechanisms, both of which involve FlaF and FlbT. In predivisional cells, FlaF and FlbT are regulated by the progression of flagellar hook assembly. At the transition from a swarmer cell type to a stalked cell, when flagellar hook assembly is long since completed, flagellin translation is regulated, we speculate, by the programmed proteolysis of FlaF.
Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this work are listed in Table 1. Plasmids were introduced into C. crescentus by bacterial conjugation using E. coli S17-1 (Simon et al., 1983) as a donor, and C. crescentus strains were grown at 31°C in peptone-yeast extract (PYE) (Poindexter, 1964) either alone or supplemented with one or more of the following: tetracycline (2.0 µg ml−1), kanamycin (20 µg ml−1), naladixic acid (20 µg ml−1). The E. coli strains were grown at 37°C in LB medium (Miller, 1972) either alone or supplemented with one or more of the following: kanamycin (50 µg ml−1), tetracycline (12.5 µg ml−1). In order to visualize cells using transmission electron microscopy, late logarithmic-phase cultures were grown in PYE medium. The cells were directly applied to carbon-coated grids, allowed to adhere and then were stained with 1% uranyl acetate for approximately 45 s.
|Strains or plasmids||Genotype or description||Reference or source|
|LS107||syn-1000 bla-6||Stephens et al. (1997)|
|SC276||flbT650||Johnson and Ely (1979)|
|JG550||flbT650 pfljK::lacZ||Mangan et al. (1999)|
|JG554||syn-1000 bla-6 pfljK::lacZ||This work|
|JG555||syn-1000 bla-6 pfljK1::lacZ||This work|
|JG556||syn-1000 bla-6 pfljK2::lacZ||This work|
|JG557||syn-1000 bla-6 pfljK3::lacZ||This work|
|JG558||flbT650 pfljK1::lacZ||This work|
|JG559||flbT650 pfljK2::lacZ||This work|
|JG560||flbT650 pfljK3::lacZ||This work|
|JG1071||syn-1000 bla-6 ΔflaF||This work|
|JG1072||syn-1000 bla-6 ΔflaF ΔflbT||This work|
|JG1073||syn-1000 bla-6 ΔflaF pML1||This work|
|JG1074||syn-1000 bla-6ΔflaF pfljK::lacZ||This work|
|JG1075||syn-1000 bla-6ΔflaF pfljK1::lacZ||This work|
|JG1076||syn-1000 bla-6ΔflaF pfljK2::lacZ||This work|
|JG1077||syn-1000 bla-6ΔflaF pfljK3::lacZ||This work|
|JG1078||syn-1000 bla-6ΔflaF ΔflbT pfljK::lacZ||This work|
|JG1079||syn-1000 bla-6ΔflaF ΔflbT pfljK1::lacZ||This work|
|JG1080||syn-1000 bla-6ΔflaF ΔflbT pfljK2::lacZ||This work|
|JG1081||syn-1000 bla-6ΔflaF ΔflbT pfljK3::lacZ||This work|
|JG1082||syn-1000 bla-6 ΔflaF pML2||This work|
|BL21(DE3)||F-ompT hsdSB(rB– mB–; an E. coli B strain) with λ prophage carrying T7 RNA polymerase gene||Studier et al. (1990)|
|S17-1||Rp4-2, Tc::Mu, Km::Tn7||Simon et al. (1983)|
|pMR4||Broad host range vector, tetr||C. Mohr|
|pML1||1145 bp BamHI-PstI fragment containing the entire flbTflaF coding sequences including upstream promoter sequences in pMR4||This work|
|pML2||946 bp BamHI-PstI fragment containing only the entire flbTflaF coding sequences without upstream promoter sequences in pMR4||This work|
|pJBZ282||lacZ protein fusion vector||M. R. K. Alley|
|pfljK::lacZ||505 bp PstI-EcoRI fragment containing the fljK promoter and 23 codons inserted in frame to lacZ in pJBZ282||Mangan et al. (1999)|
|pfljK1::lacZ||497 bp BamHI-HindIII fragment containing the fljK promoter and codons 1–14 inserted in frame to lacZ in pJBZ282||Mangan et al. (1999)|
|pfljK2:::lacZ||458 bp BamHI-HindIII fragment containing the fljK promoter and the ATG inserted in frame to lacZ in pJBZ282||Mangan et al. (1999)|
|pfljK3::lacZ||471 bp BamHI-HindIII fragment containing the fljK promoter region with the upstream leader deleted and 23 codons inserted in frame to lacZ in pJBZ282||Mangan et al. (1999)|
|pfljK–lacZ/290||505 bp PstI-EcoRI fragment containing the fljK promoter region subcloned upstream of a promoterless lacZ reporter gene to lacZ in pJBZ282||Wingrove et al. (1993)|
|pfljKp–lacZ/290||535 bp PstI-HindIII PCR fragment containing the fljK promoter region with its 3′ terminus at the +1 transcription start site subcloned upstream of a promoterless lacZ reporter gene||This work|
In order to construct a C. crescentus strain containing a deletion in flaF the entire flbTflaF operon including the upstream promoter sequences [1145 bp BamHI-PstI fragment (restriction site introduced by PCR)] in Bluescript KS was subjected to inverse PCR with Pfu DNA polymerase using primers that upon amplification created a 39 bp deletion spanning, and including, codons 1–13 of flaF and introducing a novel Bgl II site. This deleted BamHI-PstI fragment was subcloned into the sacB-containing pNPTS139, and introduced into C. crescentus cells by conjugation. These cells were then subjected to allelic exchange with selection against the sacB marker (Schweizer, 1992). Integrants containing the ΔflaF plasmid were grown overnight in PYE medium and plated onto PYE containing 3% sucrose. Sucrose-resistant colonies were screened for kanamycin sensitivity and motility defects in PYE containing 0.3% agar. Approximately 40% of the kanamycin-sensitive colonies exhibited a motility defect. The presence of the flaF deletion was verified by PCR, Southern blot and complementation of the motility defect. The construction of a strain containing a deletion both in flbT and flaF employed a similar strategy. The deleted DNA fragment in pNPTS139, in this case contained a 216 bp deletion that spanned from codon 91 of flbT to codon 13 of flaF.
In order to overexpress his-tagged FlaF, the entire flaF open reading frame was subcloned in frame downstream of the his6-encoding sequence in pET21b (Novagen). The overexpressed his-tagged FlaF was purified by nickel affinity chromatography under denaturing conditions (Ausubel et al., 1989) and used to raise antisera in rabbits (Cocalico). All DNA manipulations were performed essentially as described (Ausubel et al., 1989).
Assay of gene and protein expression
In order to assay flagellin protein stability, cultures were grown to mid-logarithmic phase in M2 minimal media (Johnson and Ely, 1977). The cellular proteins were labelled with [35S]-Trans label (ICN) for 5 min; the label was chased with non-radioactive methionine; and samples (1 ml) were collected over time and subjected to centrifugation. The cell pellets were processed for immunoprecipitation using anti-flagellin antibodies as described previously (Muir and Gober, 2002). Immunoprecipitated labelled flagellins were subjected to SDS-PAGE and the dried gel was analysed using a phosphorimager.
The stability of fljK mRNA was assayed over time following the addition of rifampicin (200 µg ml−1; final concentration) to cultures grown to mid-logarithmic phase in PYE media. Total RNA was isolated using a commercially available kit (Qiagen) from samples incubated over time in the presence of rifampicin. fljK mRNA was detected using a primer extension assay with an end-labelled primer complementary to the RNA transcript and reverse transcriptase as described previously (Mangan et al., 1999). Labelled DNA product from this assay was subjected to denaturing electrophoresis, and the dried gel was analysed with a phosphorimager.
The cell cycle pattern of flagellar protein abundance was performed on synchronized cultures as described previously (Muir and Gober, 2002). Immunoblots were performed essentially as described in the study by Towbin et al. (1979) and were analysed using anti-FlbT, anti-flagellin (Anderson and Gober, 2000) and anti-FlaF antisera. Expression of the flagellar gene transcriptional reporters in unsynchronized cultures was measured as previously described (Mangan et al., 1995). All quantitative measurements of β-galactosidase activity on unsynchronized cultures were determined in triplicate, on three separately grown cultures. In order to assay flbT promoter activity, a 425 bp BamHI-EcoRI fragment from the flbT-flaF/KS clone described above was subcloned upstream of the promoterless lacZ gene in placZ/290 (Gober and Shapiro, 1992).
We are grateful to Martin Phillips of the UCLA-DOE Instrumentation Facility for help with the electron microscopy. We thank J. England for helpful discussions and critical reading of this manuscript. M.L. was supported by a USPHS predoctoral fellowship GM07104. This work was supported by Public Health Service Grant GM48417 from the National Institutes of Health.
- 2000) FlbT, the posttranscriptional regulator of flagellin synthesis in Caulobacter crescentus, interacts with the 5′ UTR of flagellin mRNA. Mol Microbiol 38: 41–52. , and (
- 1997) Posttranscriptional regulation of Caulobacter flagellin genes by a late flagellum assembly checkpoint. J Bacteriol 179: 2281–2288. , and (
- 2003) Spatial and temporal control of differentiation and cell cycle progression in Caulobacter crescentus. Annu Rev Microbiol 57: 225–247. , and (
- 1989) Current Protocols in Molecular Biology. New York, NY: John Wiley and Sons. , , , , , , and (eds) (
- 1987) Cascade regulation of Caulobacter flagellar and chemotaxis genes. J Mol Biol 194: 71–80. , , and (
- 1990) A Caulobacter gene involved in polar morphogenesis. J Microbiol 172: 2113–2123. , , , , and (
- 2000) A family of six flagellin genes contributes to the Caulobacter crescentus flagellar filament. J Bacteriol 182: 5001–5004. , , , Jr, and (
- 2001) Cell cycle control of cell morphogenesis in Caulobacter. Curr Opin Microbiol 4: 674–680. , and (
- 2000) Regulation of flagellum biosynthesis and motility in Caulobacter. In Prokaryotic Development. Brun, Y.V., and Shimkets, L.J. (eds). Washington, DC: American Society for Microbiology, pp. 319–339. , and (
- 1992) A developmentally regulated Caulobacter flagellar promoter is activated by 3′ enhancer and IHF binding elements. Mol Biol Cell 3: 913–926. , and (
- 1977) Isolation of spontaneously derived mutants of C. crescentus. Genetics 86: 25–32. , and (
- 1979) Analysis of non-motile mutants of the dimorphic bacterium Caulobacter crescentus. J Bacteriol 137: 627–634. , and (
- 1979) Flagellar hook and basal complex of Caulobacter crescentus. J Bacteriol 138: 984–989. , , , and (
- 1983) Synthesis and assembly of flagellar components by Caulobacter crescentus motility mutants. J Bacteriol 1154: 1137–1144. , , and (
- 1995) A mutation that uncouples flagellum assembly from transcription alters the temporal pattern of flagellar gene expression in Caulobacter crescentus. J Bacteriol 177: 3176–3184. , , and (
- 1999) FlbT couples flagellum assembly to gene expression in Caulobacter crescentus. J Bacteriol 181: 6160–6170. , , , , , and (
- 1983) Caulobacter flagellin mRNA segregates asymmetrically at cell division. Nature 302: 630–632. , and (
- 1972) Assay of β-galactosidase. In Experiments in Molecular Genetics. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, pp. 352–355. (
- 1988) Role of the 25-, 27-, and 29-kilodalton flagellins in Caulobacter crescentus cell motility: method for construction of deletion and Tn5 insertion mutants by gene replacement. J Bacteriol 170: 3953–3960. , , , and (
- 1997) Cell cycle-dependent polar localization of chromosome partitioning proteins in Caulobacter crescentus. Cell 88: 675–684. , and (
- 2002) Mutations in FlbD that relieve flagellum assembly alter the temporal and spatial pattern of developmental transcription in Caulobacter crescentus. Mol Microbiol 43: 597–616. , and (
- 2004) Regulation of FlbD activity by flagellum assembly is accomplished through direct interaction with the trans-acting factor, FliX. Mol Microbiol 54: 715–730. , and (
- 2001) The Caulobacter crescentus flagellar gene, fliX, encodes a novel trans-acting factor that couples flagellar assembly to transcription. Mol Microbiol 39: 1623–1637. , , and (
- 1991) Expert system for predicting protein localization sites in Gram-negative bacteria. Proteins 11: 95–110. , and (
- 1989) Genetic switching in the flagellar gene hierarchy of Caulobacter requires negative as well as positive regulation of transcription. Proc Natl Acad Sci USA 8: 6651–6655. , , , , and (
- 2003) The various and varying roles of specific chaperones in type III secretion systems. Curr Opin Microbiol 6: 7–14. , , and (
- 1964) Biological properties and classification of the Caulobacter group. Bacteriol Rev 28: 231–295. (
- 2003) Cell cycle timing and developmental checkpoints in Caulobacter crescentus. Curr Opin Microbiol 6: 541–549. , and (
- 1994) Multiple structural proteins are required for both transcriptional activation and negative autoregulation of Caulobacter crescentus flagellar genes. J Bacteriol 176: 7587–7600. , , and (
- 2003) Temporal and spatial regulation in prokaryotic cell cycle progression and development. Annu Rev Biochem 72: 367–394. , and (
- 1989) Characterization of strains containing mutations in the contiguous flaF, flbT, or flbA-flaG transcription unit and identification of a novel fla phenotype in Caulobacter crescentus. J Bacteriol 171: 1554–1561. , and (
- 1989) Organization of the flaFG gene cluster and identification of two additional genes involved in flagellum biogenesis in Caulobacter crescentus. J Bacteriol 171: 1544–1553. , , and (
- 1992) The Caulobacter crescentus flaFG region regulates synthesis and assembly of flagellin proteins encoded by two genetically unlinked gene clusters. J Bacteriol 174: 6046–6053. , , , and (
- 1992) Allelic exchange in Pseudomonas aeruginosa using novel ColE1-type vectors and a family of cassettes containing a portable oriT and the counter-selectable Bacillus subtilis sacB marker. Mol Microbiol 6: 1195–11204. (
- 1983) A broad host range mobilization system for in vivo genetic engineering: transposon mutagenesis in Gram-negative bacteria. Biotechnology 1: 784–790. , , and (
- 1997) Identification of the fliI and fliJ components of the Caulobacter flagellar type III protein secretion system. J Bacteriol 179: 5355–5365. , , , , , and (
- 1990) Use of T7 RNA polymerase to direct expression of cloned genes. Meth Enzymol 185: 60–89. , , , and (
- 1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Natl Acad Sci USA 76: 4350–4354. , , and (
- 1994) A σ54 transcriptional activator also functions as a pole-specific repressor in Caulobacter. Genes Dev 8: 1839–1852. , and (
- 1993) Spatial and temporal phosphorylation of a transcriptional activator regulates pole-specific gene expression in Caulobacter. Genes Dev 7: 1979–1992. , , and (
- 1989) Negative transcriptional regulation in the Caulobacter flagellar hierarchy. Proc Natl Acad Sci USA 86: 6656–6660. , , and (