Partially Reciprocal Replacement of FlrA and FlrC in Regulation of Shewanella oneidensis Flagellar Biosynthesis

ABSTRACT

In some bacteria with a polar flagellum, an established regulatory hierarchy controlling stepwise assembly of the organelle consists of four regulators: FlrA, σ⁵⁴, FlrBC, and σ²⁸. Because all of these regulators mediate the expression of multiple targets, they are essential to the assembly of a functional flagellum and therefore to motility. However, this is not the case for the gammaproteobacterium Shewanella oneidensis: cells lacking FlrB, FlrC, or both remain flagellated and motile. In this study, we unravel the underlying mechanism, showing that FlrA and FlrC are partially substitutable for each other in regulating flagellar assembly. While both regulators are bacterial enhancer binding proteins (bEBPs) for σ⁵⁴, FlrA differs from FlrC in its independence of σ⁵⁴ for its own transcription and its inability to activate the flagellin gene flaA. These differences largely account for the distinct phenotypes resulting from the loss or overproduction of FlrA and FlrC.

IMPORTANCE The assembly of a polar flagellum in bacteria has been characterized as relying on four regulators, FlrA, σ⁵⁴, FlrBC, and σ²⁸, in a hierarchical manner. They all are essential to the process and therefore to motility, except in S. oneidensis, in which FlrB, FlrC, or both together are not essential. Here we show that FlrA and FlrC, as bEBPs, are partially reciprocal in functionality in this species. As a consequence, the presence of one allows flagellar assembly and motility in the other's absence. Despite this, there are significant differences in the physiological roles played by these two regulators: FlrA is the master regulator of flagellar assembly, whereas FlrC fine-tunes motility. These intriguing observations open up a new avenue to further exploration of the regulation of flagellar assembly.

INTRODUCTION

Many bacteria are motile because of a flagellum (or flagella), a complex molecular machine that functions by rotation to push or pull the cell and that plays a critical role in adhesion, biofilm formation, and host invasion (1). Although core flagellar components mostly originate through successive duplication and are thus fairly conserved (2), flagellation patterns, characteristic for each species, are remarkably diverse (3). These include the monotrichous (a single polar flagellum, as in Vibrio cholerae), lophotrichous (a tuft of flagella at one end of the cell, as in Helicobacter pylori), amphitrichous (a flagellum or a tuft of flagella on each pole of the cell, as in Campylobacter jejuni), and peritrichous (flagella at many locations around the cell surface, as in Escherichia coli) patterns.

More than 50 gene products are involved in the assembly of a functional flagellum; this is a highly conserved process and is tightly controlled at the transcriptional level (4). In most, if not all, monotrichous bacteria (with a single polar flagellum), the expression of flagellar genes is mediated by a four-tiered transcriptional regulatory hierarchy, as typified in Vibrio cholerae and Pseudomonas aeruginosa (5) (Fig. 1A). In brief, at the top (class I) is the master regulator FlrA (FleQ in P. aeruginosa), whose expression depends on the σ⁷⁰ factor (the housekeeping sigma factor, encoded by rpoD) (6). In the presence of the σ⁵⁴ factor (originally identified as the sigma factor for nitrogen-controlled genes; encoded by rpoN), FlrA activates the transcription of class II genes, which encode components of the basal body-switch-export apparatus as well as the two-component system (TCS) FlrBC (FleSR in P. aeruginosa) and the σ²⁸ factor (the flagellum-specific sigma factor, encoded by fliA) (7). Transcription of class III genes, encoding the remaining basal body components, hook, and some flagellins, requires σ⁵⁴ and the DNA-binding regulator FlrC in its phosphorylated form (5). Both FlrA and FlrC function as activators and are absolutely required for σ⁵⁴-dependent transcription as bacterial enhancer binding proteins (bEBPs) (8). Class IV genes, whose products include the remaining flagellins, motor subunits, and some chemotactic proteins, are directed by σ²⁸. In V. cholerae and P. aeruginosa, all four regulators, σ⁵⁴, FlrA (FleQ), FlrBC (FleSR), and σ²⁸, are essential for motility (7, 9–11).

FIG 1.

Transcription hierarchy of flagellar genes in bacteria with a single polar flagellum. (A) V. cholerae-based diagram derived from previous reports mentioned in the text with updated information (17). dep., dependent. (B) Proposed flagellar regulatory hierarchy in S. oneidensis. The hierarchy is based on the expression data and predicted binding sites for σ⁵⁴ and σ²⁸. Operons in lightface lack binding sites for corresponding sigma factors. In both models, flagellar genes (∼5) independent of σ⁵⁴ and σ²⁸ are omitted.

Shewanella oneidensis is a facultative gammaproteobacterium renowned for its ability to reduce diverse organic and inorganic compounds as terminal electron acceptors (12). In recent years, this bacterium has emerged as a research model for broad bacterial physiology, including flagellar assembly (13–21). At the genus level, Shewanella and Vibrio are the most closely related bacteria and share extensive regions of similar gene order (synteny), suggesting a wide range of horizontal gene transfer between them (22). Although S. oneidensis σ⁵⁴, FlrA, and σ²⁸ seemingly function similarly to their V. cholerae counterparts, the TCS FlrBC is not essential for flagellar assembly (17). Loss of either FlrB or FlrC has no noticeable impact on flagellar physiology. However, the depletion of both promotes motility, a phenomenon attributable, at least in part, to the altered production of two flagellins encoded in the genome, FlaA and FlaB, whose ratio is inversely correlated with motility (17). Although these two flagellins are highly homologous (sequence identity, 92%), a few amino acid residues that differ between them predominantly account for a substantial difference in their locomotion activity (14, 18). Since flaA transcription depends on FlrBC, the flrBC mutant produces flagellar filaments consisting of FlaB predominantly, leading to enhanced motility.

When overproduced, FlrC, but not FlrB or both, results in a multiflagellum phenotype (similar to the peritrichous pattern), leading to a ∼50% reduction in motility (17). Moreover, this effect becomes more drastic when FlrC is produced in its unphosphorylated form, contrasting with the finding that V. cholerae FlrC is active only when it is phosphorylated (23, 24). In addition, multiflagellum phenotypes have also been observed with S. oneidensis flhG and flhFG mutants: the former mutant has a lophotrichous pattern, whereas the latter develops flagella in a peritrichous pattern (19). In polar flagellates, FlhF and FlhG, a multiple-domain GTPase and a MinD-like ATPase, respectively, function as a cognate pair to regulate flagellar localization and number (25).

In our continuing effort to address the physiological role of FlrC in S. oneidensis, we obtained a spontaneous mutant of a ΔflrA strain that recovered some motility. The mutation turned out to be an insertion of an ISSod1 element upstream of the flrC gene, resulting in FlrC overproduction. Further analyses revealed that FlrC in excess activates most of the class II and class III flagellar genes, prompting the ΔflrA strain to assemble multiple flagella. FlrA and FlrC differ from each other in that the latter, but not the former, can function as an activator for the flagellin gene flaA. Moreover, we unraveled the mechanism for the multiflagellum phenotype resulting from loss of FlhG: enhanced production and activity of FlrA. These intriguing observations open up a new avenue for further exploration of the regulation of flagellar assembly.

RESULTS

Identification of FlrC as a functional replacement of FlrA.

By chance, we obtained a spontaneous mutant from an S. oneidensis flrA-null mutant (the ΔflrA^sup strain, with a suppressor mutation of ΔflrA) that regains the ability—albeit a rather limited ability—to move on semisolid LB agar plates (Fig. 2A). Since there are multiple types of cell movement, we first determined whether the observed motility was flagellum dependent. The S. oneidensis flagellar filament comprises two flagellins, FlaA and FlaB; a flagellin-free mutant (FFM) (ΔflaA ΔflaB) is consequently aflagellated and nonmotile (14, 18). The removal of both flagellin genes rendered the ΔflrA^sup strain completely nonmotile (Fig. 2A), supporting the notion that the suppression relies on the flagellar device. Additionally, we extracted filaments from the ΔflrA^sup, ΔflrA, and FFM strains and quantified flagellin. As shown in Fig. 2B, the ΔflrA^sup strain produced considerable amounts of flagellin proteins, whereas in the FFM and ΔflrA strains, they were below the detection limit. These data, collectively, indicate that the ΔflrA^sup strain possesses one or more flagella, to which the motility is attributable.

FIG 2.

Motility phenotypes of S. oneidensis ΔflrA and ΔflrA^sup strains. (A) (Top) Shown are the wild type (WT) and its isogenic mutants (MT) spotted onto the same 0.25% soft agar plates and incubated for 16 h. The ΔflrA mutant was previously validated by genetic complementation and was further confirmed by complementation by controlled flrA expression (presented in Fig. 4). The flagellin-free strain FFM (ΔflaA ΔflaB) was used as the negative control. (Bottom) Relative motility was determined by normalizing to WT motility. (B) Analysis of flagellin levels. Flagellin proteins from similar numbers of cells for each strain indicated were extracted and were subjected to SDS-PAGE. The relative abundance (RA) of flagellins was calculated based on signal intensities estimated by ImageJ, with the WT value set at 100%. (C) Insertion of an ISSod1 element (blue letters) before flrC. The insertion resulted in a σ⁷⁰-dependent promoter; −35 and −10 boxes are underlined. Lowercase letters represent the start code (ATG) of flrC and the stop code (TGA) of flrB. (D) Activities of P_flrC and P_flrC(M) in WT and ΔflrA strains. DNA fragments of ∼300 bp centered on the predicted promoter were generated by PCR and were placed in front of the full-length E. coli lacZ gene on integrative plasmid pHGEI01. The resulting vectors, verified by sequencing, were introduced into S. oneidensis strains for the integration and subsequent removal of the antibiotic marker. Mid-log-phase cells of the indicated strains carrying integrated reporter systems were collected and assayed for β-galactosidase activity. Vec, empty vector (nonpromoter control). The micrographs and gel show representative data; graphical data and values below the gel in panel B are means ± SEM from at least three independent experiments.

This unexpected phenotype implies that the loss of FlrA can be partially complemented by other factors, presumably proteins. To identify suppressor mutations, we sequenced the major region for flagellar genes, encoding most of the structural components and regulatory proteins (see Fig. S1 in the supplemental material). The only mutation found was the insertion of an ISSod1 element near the 3′ end of the flrB gene (immediately upstream of the flrC gene) (Fig. 2C), a scenario reported previously for S. oneidensis (26). In that case, the ISSod1 insertion generates a strong hybrid promoter for the gene after the insertion, resulting in overexpression. Given that the FlrB loss has no detectable effect on S. oneidensis motility, but FlrC in excess does (17), we reasoned that the insertion results in FlrC overproduction. A bioinformatics analysis of the insertion region suggested the presence of a σ⁷⁰-dependent promoter [P_flrC(M) (where “M” stands for “mutant”) rather than P_flrC(WT) (where “WT” stands for “wild type”)], consisting of a putative −35 box within ISSod1 and a recognizable −10 box within the flrC gene (Fig. 2C). In contrast, the analysis failed to identify a promoter from the native sequence, in agreement with our previous finding that flrC is transcribed from a σ⁵⁴-dependent promoter for the flrB-flrC operon (17). To test whether P_flrC(M) functions, ∼300-bp fragments from both the wild-type and ΔflrA^sup strains for P_flrC(M) and P_flrC(WT), respectively, were amplified and were fused transcriptionally to E. coli lacZ within the integrative plasmid pHGEI01. The resulting vectors were introduced into relevant S. oneidensis strains and were integrated into the chromosome, and subsequently the antibiotic marker was eliminated (27, 28). In wild-type cells grown to the mid-log phase (optical density at 600 nm [OD₆₀₀], ∼0.4), the activities of P_flrC(WT) and P_flrC(M) were assessed by β-galactosidase assays (Fig. 2D). The activity of P_flrC(WT) was extremely low (<20 Miller units), not significantly different from that of the control without a promoter. In contrast, P_flrC(M) displayed substantially increased activity, >1,000 Miller units. This hybrid promoter was comparably robust in the ΔflrA strain, indicating that it is independent of FlrA. These data were verified by direct analysis of the abundances of flrC transcripts in the wild-type, ΔflrA, and ΔflrA^sup strains by quantitative real-time PCR (qRT-PCR) (see Fig. S2 in the supplemental material). As expected, flrB and flrC transcript levels were comparable to each other in both the wild-type and ΔflrA backgrounds, but flrC transcript levels were drastically higher in the ΔflrA^sup background. Thus, we conclude that the insertion has generated a promoter for the flrC gene, leading to flrC expression, which may partially complement the loss of FlrA.

FlrC in excess can partially restore motility in the ΔflrA strain but not in the ΔrpoN or ΔfliA strain.

We have shown previously that transcription of the flrBC operon is dependent on both FlrA and σ⁵⁴ and that the operon promoter is rather weak (approximately 50 Miller units) in cells grown aerobically under normal conditions (17). Apparently, the insertion of the ISSod1 element results in substantially enhanced expression of the flrC gene. This gained further support from controlled expression of the flrC gene by using an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter, P_tac. The motility of the ΔflrA strain carrying P_tac-flrC increased with increasing IPTG concentrations (Fig. 3A). When the IPTG concentration was at 0.5 mM or higher, a motility level comparable to that of the ΔflrA^sup strain was achieved. According to our previous calibration, IPTG at 0.5 mM confers an activity of approximately 1,000 Miller units on P_tac within plasmid pHGE-Ptac (29). This is in excellent agreement with the activity of P_flrC(M) in the ΔflrA^sup strain. To simplify description, 0.5 mM IPTG was used for FlrC overproduction throughout the rest of this study unless otherwise noted. It is worth mentioning that overproduced FlrC negatively regulated motility in the wild type, a phenomenon reported previously (17).

FIG 3.

Influence of FlrC at various levels on flagellar biosynthesis and motility. (A) Influence of FlrC at various levels on motility. S. oneidensis wild-type (WT) and ΔflrA strains producing various levels of FlrC from the IPTG-inducible promoter P_tac were assayed for motility in comparison with the WT carrying an empty vector, whose motility was set to 100%. (B) SDS-PAGE analysis of flagellins isolated from the indicated strains. Flagellins from the same volume of mid-log-phase cultures (adjusted to the same optical density) were extracted and were separated by SDS-PAGE. The band intensity was estimated by using ImageJ. The relative abundance (RA) of flagellin in mutant strains was calculated by normalizing to the abundance in the WT, which was set at 100% for presentation. Relative motility (RM) was determined by normalizing to WT motility, which was set at 100%. Overproduction of FlrC was achieved with 0.5 mM IPTG. (C) Influence of overproduced FlrC on the activities of the indicated promoters. Mid-log-phase cultures were induced by 0.5 mM IPTG for 2 h, collected, and assayed for β-galactosidase. The gel shows representative data; graphical data and values below the gel in panel B are means ± SEM from at least three independent experiments.

In monotrichous bacteria, in addition to FlrA and FlrBC, there are two other regulators: σ⁵⁴, required for the transcription of class II and III operons, and σ²⁸, required for the transcription of class IV operons (30). We thus examined whether the effects of excessive FlrC are independent of σ⁵⁴ or σ²⁸. Although the S. oneidensis σ⁵⁴- and σ²⁸-deficient strains (the ΔrpoN and ΔfliA mutants, respectively) are completely nonmotile, they differ from each other in flagellation: the former is aflagellated, but the latter develops severely shortened filaments in about 30% of the population (14, 15) (see Fig. S3 in the supplemental material). FlrC overproduction had no detectable influence on the motilities of the ΔrpoN and ΔfliA strains (Fig. 3B; see also Fig. S4 in the supplemental material). However, the nonmotile phenotypes of these two strains clearly resulted from different mechanisms. In the presence of excess FlrC, the ΔrpoN strain remained aflagellated, but the ΔfliA strain increased flagellin production drastically (even exceeding that in the wild type), leading to flagellar filaments indistinguishable from those of the wild type (Fig. S3).

To test whether FlrC in excess could override the requirement of σ²⁸ for the transcription of class IV genes, we compared the activities of promoters for the flaB, pomAB, and flgMN operons with integrative lacZ reporters (17). As shown in Fig. 3C, in the absence of σ²⁸, all of these genes were not expressed, regardless of FlrC levels; however, when σ²⁸ was present, FlrC in excess substantially enhanced the expression of these operons. These data indicate that both σ⁵⁴ and σ²⁸ are prerequisites for overproduced FlrC to regulate flagellar assembly and motility.

FlrA and FlrC in excess result in multiflagellum phenotypes.

Although FlrC can take on the role of FlrA to some extent, the protein produced at any level fails to fully restore motility to wild-type levels. This may be simply a result of ectopic FlrC production. To test this, the effects of controlled FlrA production on the motility of the wild-type and ΔflrA strains were examined (Fig. 4A). In the wild type, overproduced FlrA driven by P_tac, similarly to FlrC (Fig. 3A), negatively regulated motility. Nevertheless, FlrA was apparently more robust, in that it was able to reduce motility to 37% of that without overexpression when produced with 0.5 mM IPTG, contrasting with a reduction to 55% of motility without overexpression observed with FlrC. (Unless otherwise noted, motility levels reported below are given relative to the motility of the wild type.) In the ΔflrA strain, motility can be restored to wild-type levels with FlrA produced by 0.05 mM IPTG, while FlrA production enhanced by higher levels of IPTG resulted in inhibition (motility, ∼37%) as strong as that observed in the wild type but much less severe than that observed from FlrC overproduction (∼16%) (17). These data indicate that the ectopic production of FlrC is unlikely to be accountable for the failure of FlrC to fully restore the motility of the ΔflrA strain and rather suggest a difference between the physiological activities of FlrA and FlrC in excess. Additionally, given the substantial difference in motility between flrA⁺ (wild-type) and flrA-null (ΔflrA) strains caused by FlrC overproduction, it is reasonable to predict that FlrA alleviates the effects of overproduced FlrC.

FIG 4.

Influence of FlrA at various levels on flagellar biosynthesis and motility. (A) Influence of FlrA at various levels on motility. S. oneidensis wild-type (WT) and ΔflrA strains producing various levels of FlrA from the IPTG-inducible promoter P_tac were assayed for motility in comparison with the WT strain carrying an empty vector, whose motility was set at 100%. (B) TEM visualization of the S. oneidensis ΔflrA strain overexpressing FlrC or FlrA. (C) Influence of FlrA at various levels on the numbers of flagellar filaments. Swimming cells were scraped from the leading edges of each swarm, stained for flagellar filaments, and visualized under a phase-contrast microscope. For each strain, 100 cells were analyzed. (D) SDS-PAGE analysis of isolated flagellins from the indicated strains. Micrographs and gels show representative data. Graphical data and values are means ± SEM from at least three independent experiments.

In S. oneidensis, excessive FlrC results in a multiflagellum phenotype similar to a peritrichous pattern (17). To test whether the ΔflrA strain overproducing FlrC has a similar flagellation pattern, we used transmission electron microscopy (TEM). Indeed, FlrC overproduction enabled the ΔflrA strain to assemble multiple flagella located randomly around the cell (Fig. 4B). As expected, the same phenotype was observed with FlrA overproduction. Despite this, a significant difference between the strains was evident, as revealed by quantification of the flagellar filaments of 100 individual cells growing on the soft agar plates (Fig. 4C). Control strains exhibited expected results: 69 wild-type cells were flagellated with exclusively a single polar flagellum, and in the absence of FlrA, all cells were aflagellated. For a wild-type population overproducing FlrC, cells with multiple (≥3) flagella dominated (48 cells) and the number of aflagellated cells was reduced to 17. In the flrA deletion background, FlrC overproduction resulted in 43, 23, and 6 cells with one, two, and multiple (≥3) flagella, respectively. Overproduction of FlrA had a more drastic impact on flagellar numbers: as many as 36 cells had multiple flagella.

The multiflagellum phenotypes of these strains were confirmed by flagellin levels (Fig. 4D; also Fig. S4 in the supplemental material). In line with flagellar numbers, excess FlrC induced the wild type to produce more flagellin proteins than the ΔflrA strain. In parallel, FlrA was more effective than FlrC at promoting flagellar assembly when overproduced. However, multiflagellation is not a major reason for the extremely low motility of the ΔflrA strain overexpressing FlrC, because it has much fewer flagella than the wild type with excess FlrC. Hence, other factors contribute.

Enhanced FlrA activity underlies the multiflagellum phenotype resulting from the deletion of flhG or flhFG.

In addition to strains overproducing FlrA and FlrC, two S. oneidensis mutants, the ΔflhG and ΔflhFG mutants, develop flagella in lophotrichous and peritrichous patterns (19) (Fig. S3 in the supplemental material). We wondered if there is a common mechanism that accounts for the phenotype in all these strains. The flhFG operon was removed from the ΔflrA strain, and the motility of the resultant mutant was assayed. In contrast to the ΔflhFG strain, the ΔflrA ΔflhFG strain is neither motile nor flagellated (Fig. 5A; also Fig. S4), showing that the effect of the FlhFG loss relies on FlrA. With a small amount of FlrC produced from the P_tac promoter without IPTG (P_tac within pHGE-Ptac is slightly leaky [17, 31]), the ΔflrA ΔflhFG strain was able to assemble a single randomly located flagellum (Fig. S3) and recovered motility to about 16% of wild-type motility (Fig. 5A; also Fig. S4). In the presence of overproduced FlrA or FlrC with 0.5 mM IPTG (Fig. S5), the ΔflrA ΔflhFG strain assembled multiple flagella and regained a slight ability to move, comparable to that of the ΔflhFG strain. These data indicate that the multiflagellum phenotype resulting from excessive FlrA or FlrC is independent of FlhFG.

FIG 5.

Influence of FlhFG on flagellar biosynthesis and motility. (A) SDS-PAGE analysis of isolated flagellins from strains of interest. Numbers in parentheses represent millimolar concentrations of IPTG. (B) Activities of P_flrA and P_flrBC. Shown are promoter activities in the indicated strains carrying an empty vector (vec) or overproducing FlhFG due to induction by 0.5 mM IPTG. Asterisks indicate statistically significant differences (**, P < 0.01; ***, P < 0.001). The gel shows representative data. Graphical data and values below the gel are means ± SEM from at least three independent experiments.

Studies of P. aeruginosa FleN, an ortholog of FlhG, have established that FleN inhibits FleQ (the FlrA ortholog) by direct interaction (32–34). For Shewanella spp., a similar mechanism for the antagonistic effect of FlhG against FlrA was proposed recently (35). Despite this, given that overproduction of FlrA or FlrC results in the same phenotype, we reasoned that FlhG and FlhFG may influence the expression of FlrA and FlrC. Ptac-flhG and Ptac-flhFG were introduced into the wild-type, ΔflhF, ΔflhG, ΔflhFG, ΔflrA ΔflhG, and ΔflrA ΔflhFG strains carrying integrated lacZ reporters. As shown in Fig. 5B, the removal of FlhF (ΔflhF) did not affect the expression of flrA or flrBC. In contrast, in the absence of either FlhG alone (ΔflhG) or FlhFG together (ΔflhFG), the flrA and flrBC promoters were significantly upregulated over expression in the wild type. While overproduction of FlhG or FlhFG (0.5 mM IPTG) hardly influenced the activities of the flrA and flrBC promoters in the wild-type background, it decreased them to wild-type levels in the ΔflhG and ΔflhFG backgrounds. In the ΔflrA ΔflhG and ΔflrA ΔflhFG strains, similar results for the activity of the flrA promoter but not for that of the flrBC promoter, which was barely detectable, were obtained. Intriguingly, although depletion of FlhG alone and depletion of FlhFG together consistently upregulated the expression of flrA and flrBC, their impacts were not identical. The stronger influence of the FlhFG double loss implies that FlhF is also a factor contributing to the regulation of flrA expression. Taken together, these data suggest that in the ΔflhG and ΔflhFG strains, FlrA overproduction, likely along with increased FlrA activity due to the lack of the antagonistic effect of FlhG, underlies the multiflagellum phenotype.

FlrC in excess alters the flagellin ratio.

In S. oneidensis, a key factor that dictates motility is the ratio of two flagellins, FlaA and FlaB (18). This is because, with respect to motility, FlaB is predominant but the effect of FlaA is negligible (14). The consequence is that motility increases with the ratio of FlaB to FlaA; for example, an flrBC-deficient strain, whose filaments are dominated by FlaB, has enhanced motility (17). To test whether flagellin ratios are altered under conditions of FlrA and FlrC overproduction, flagellar filaments were extracted from relevant strains and were analyzed by liquid chromatography-tandem mass spectrometry (LC–MS-MS) as described previously (18). We found that the ratio of the two flagellins, based on averaged intensities of unique signature peptides for each flagellin in the ΔflrA strain overproducing FlrC or FlrA, was significantly affected by FlrC but not by FlrA (Fig. 6A). While the ratio of FlaA to FlaB in the wild type was about 43:57, overproduction of FlrC increased the FlaA percentage significantly, to approximately 52% and 66% in the wild-type and ΔflrA strains, respectively.

FIG 6.

FlrC, but not FlrA, influences flaA expression. (A) The ratio of FlaA to FlaB increased in the presence of excess FlrC. Flagellins from the same volume of mid-log-phase cultures adjusted to the same optical density were extracted, digested with trypsin, and analyzed by LC–MS-MS to determine the relative abundances of FlaA and FlaB. (B) SDS-PAGE analysis of isolated flagellins from strains of interest. (C) Activities of P_flaA and P_flaB. Shown are promoter activities in the indicated strains overproducing FlrA or FlrC due to induction by 0.5 mM IPTG. Asterisks indicate statistically significant differences (*, P < 0.05; **, P < 0.01). The gel shows representative data. Graphical data and values below the gel are means ± SEM from at least three independent experiments.

For confirmation, the flagellin genes flaA and flaB were deleted from the ΔflrA strain separately and together, resulting in ΔflaA ΔflrA, ΔflaB ΔflrA, and ΔflaAB ΔflrA mutants. SDS-PAGE analysis of flagellar filaments extracted from these mutants overproducing FlrC revealed that FlaA exceeded FlaB in abundance significantly, making up >60% of total flagellins (Fig. 6B; also Fig. S4 in the supplemental material). In line with the finding that FlaB is superior to FlaA for motility, with excess FlrC, the ΔflaA ΔflrA strain showed motility much stronger than that of the ΔflrA strain, and the ΔflaB ΔflrA strain was hardly able to move (Fig. 6B; also Fig. S4). Notably, both FlaB and FlaA were produced in increased quantities in the presence of excess FlrC. Indeed, with integrated lacZ reporters, we found that the activities of the flaA and flaB promoters were elevated in the ΔflrA strain overexpressing either flrC or flrA (Fig. 6C). Nevertheless, a significant difference in expression between flaA and flaB was observed in cells overexpressing flrC but not flrA, supporting the notion that excess FlrC differentially regulates the expression of these two flagellin genes. Taking all the data together, we conclude that the flagellin component of the filament is in part accountable for the motility difference resulting from excess FlrA and FlrC.

FlrC, but not FlrA, directly mediates flaA expression.

Both FlrA and FlrC are characterized as bacterial enhancer binding proteins (bEBPs), which function as activators absolutely required for σ⁵⁴-dependent transcription (8). The flaA promoter is σ⁵⁴ dependent; it is inactive in the absence of σ⁵⁴ (18). However, this promoter is not among the σ⁵⁴-dependent promoters identified by a highly specific weight-matrix-based screening method, implying that it may be distinct from the canonical σ⁵⁴-dependent promoters, at least to some extent (17). We reasoned that this difference may underlie the substantial difference in flaA expression resulting from excess FlrA and FlrC.

To test direct interaction between these two regulators and the flaA promoter, we expressed and purified His-tagged FlrA and FlrC proteins in E. coli for electrophoretic mobility shift assay (EMSA) analysis (Fig. 7A). Since FlrC in its unphosphorylated form has higher activity in regulation (17), both wild-type and D54A mutant FlrC proteins (FlrC^WT and FlrC^D54A) were prepared. The flhFG promoter (∼350 bp covering the promoter region), which possesses a canonical σ⁵⁴-binding site (TGGAACAGATGTTGC), was used as a positive control (17). Significant binding to the flhFG DNA probe occurred for both FlrA and FlrC^D54A, but barely for FlrC^WT-P (FlrC^WT phosphorylated with carbomoyl phosphate) (Fig. 7B). Binding to the flaA fragment was observed with purified FlrC^D54A, which is specific, because FlrC^D54A did not interact with the flaB promoter, a member of class IV (Fig. 7C). In contrast, neither FlrA nor FlrC^WT-P could bind.

FIG 7.

FlrC, but not FlrA, interacts with the flaA promoter. (A) Expression and purification of FlrA, FlrC^WT, and FlrC^D54A. His-tagged proteins were expressed and purified from E. coli as described in Materials and Methods and were analyzed by 12% SDS-PAGE (whole gels are shown in Fig. S6 in the supplemental material). (B) Interaction of flhFG promoter DNA with S. oneidensis His-tagged FlrA, FlrC^WT, and FlrC^D54A. The probe was prepared by PCR with ³³P-end-labeled primers. EMSA was performed with 2 nM ³³P-end-labeled probes and various amounts of proteins. Nonspecific competitor DNA (0.2 μg poly dI·dC) was added to the first four lanes, and a specific competitor (10 μM unlabeled flhFG probe) was added to the last lanes. FlrC^WT-P represents FlrC^WT phosphorylated with carbamoyl phosphate. (C) The binding assay was performed in the presence of 0.5, 1, or 2 μM proteins and 2 nM radiolabeled promoter DNA. A 0.2-μg/μl concentration of poly(dI·dC) was used in all these binding reactions to block nonspecific interactions. Shown are representative data from at least three independent experiments.

To determine interactions in vivo, we employed a bacterial one-hybrid (B1H) system as described previously (17, 36). The B1H system allows reporter cells to grow if there is direct interaction between a vector containing a “bait” (DNA) and a vector containing a “target” (DNA-binding regulator) (37). DNA fragments used in EMSA were cloned into pBXcmT, which was paired with pTRG carrying flrA or flrC genes encoding wild-type and D54A mutant proteins for cotransformation. The results revealed interactions from protein-promoter pairs FlrA-flhFG, FlrC^WT-flhFG, FlrC^D54A-flhFG, FlrC^WT-flaA, and FlrC^D54A-flaA. No interaction was detected from FlrA-flaA, FlrC^WT-flaB, or FlrC^D54A-flaB (Table 1). In this assay, both FlrC^WT and FlrC^D54A proteins were able to interact with projected DNA sequences. This is reasonable, because the S. oneidensis FlrC^WT produced is unlikely to be phosphorylated, since E. coli lacks an ortholog of FlrB. Taking these results together with the EMSA data, we conclude that FlrC, but not FlrA, interacts with the flaA promoter, while both recognize canonical σ⁵⁴-dependent promoters.

TABLE 1.

Bacterial one-hybrid (assay of FlrA, FlrC^WT, and FlrC^D54A

Bait vector	Target vector	No. of colonies on:			Interaction
		Nonselective platesa	Selective platesb	Confirmation platesc
PflhFG		214	0		No
	FlrA	257	0		No
PflhFG	FlrA	171	123	123	Yes
PflhFG	FlrCWT	208	71	62	Yes
PflhFG	FlrCD54A	234	99	98	Yes
PflaA	FlrA	331	6	1	No
PflaA	FlrCWT	266	104	98	Yes
PflaA	FlrCD54A	185	106	104	Yes
PflaB	FlrA	209	6	0	No
PflaB	FlrCWT	238	7	1	No
PflaB	FlrCD54A	272	2		No

^aContaining M9 agar, 25 μg/ml chloramphenicol, and 12.5 μg/ml tetracycline.

^bNonselective plates with 5 mM 3-AT.

^cSelective plates with 12.5 μg/ml streptomycin.

DISCUSSION

S. oneidensis has emerged as a research model for broad bacterial physiology in recent years. With respect to the flagellar system, extensive studies have been carried out, revealing a number of surprising findings that provide important insights into the assembly of the locomotive organelle and its regulation. These include the biology of two stator systems (13, 20, 21), flagellins (14, 16, 18, 38), flagellar number and location (19), flagellum and protein secretion (15), and flagellar regulation (17). Additionally, studies of dual flagellar systems in other Shewanella species have elucidated the role played by these systems in acquiring a fitness gain in various environments (39–42).

While the majority of these explorations identify novel flagellar components/systems and decipher the molecular mechanisms by which they function, others focus on flagellar proteins whose physiological roles are already established and widely accepted, regulators in particular. In monotrichous bacteria, prior to our work, all four regulators, FlrA (FleQ in P. aeruginosa), σ⁵⁴, FlrBC (FleSR in P. aeruginosa), and σ²⁸, were found to be essential for flagellar assembly and therefore motility (9, 10). In S. oneidensis, however, while FlrA, σ⁵⁴, and σ²⁸ remain essential, the absence of FlrC has no detectable influence on flagellar assembly and motility (14, 17). This is also true of its cognate partner, the sensor kinase FlrB. Nevertheless, loss of the entire TCS increases motility, largely by reducing the ratio of flagellin FlaA to FlaB (17). On the other hand, overproduction of FlrC but not of FlrB or FlrBC together causes a multiflagellum phenotype.

In this study, we endeavored to unravel mechanisms for the nonessentiality of FlrC and the multiflagellum phenotype resulting from its overproduction in S. oneidensis. Based on the data presented, we propose a flagellar regulatory hierarchy deriving from the previous model (17) (Fig. 1B). In this modified model, σ⁷⁰-dependent FlrA remains the top-tier regulator that controls the expression of class II genes. However, there is only one operon in class II, flrBC, contrasting with multiple ones in the model proposed before (17). FlrC activates the transcription of flaA and class III genes, which comprise all the σ⁵⁴-dependent genes except for flrBC and flaA. As a result, when FlrA is absent, FlrC ensures the transcription of class III genes such that a functional flagellum (or flagella) can be assembled. This explains why FlrC could function as a replacement for FlrA. However, it should be noted that FlrC could not be naturally produced in the absence of FlrA, because σ⁵⁴-dependent transcription of flrC requires FlrA as a bEBP. In parallel, FlrA, bypassing FlrB, FlrC, or FlrBC, is able to directly activate all σ⁵⁴-dependent class III operons but flaA. Because of this, a flagellum can be assembled in the absence of the FlrBC TCS.

Although the finding that FlrA and FlrC are mutually substitutable to some extent is unprecedented, it is not completely beyond imagination. σ⁵⁴-dependent transcription absolutely requires the presence of a bEBP, which couples the energy generated from ATP hydrolysis to the isomerization of the RNA polymerase (RNAP)-σ⁵⁴ closed complex (8). Both FlrA and FlrC are characterized as bEBPs, having similar domain organizations, an N-terminal receiver domain, a central ATPase associated with diverse cellular activities (AAA⁺) domain, and a C-terminal DNA-binding domain (8). In line with this feature, we observed that both proteins are able to activate the transcription of the σ⁵⁴-dependent operon flhFG. Nevertheless, the impacts of overproduced FlrA and FlrC on flagellar biosynthesis are not identical: the multiflagellum phenotypes resulting from their overproduction are more drastic with FlrA than with FlrC. We envision that FlrA is superior to FlrC in binding to σ⁵⁴-dependent promoters; however, this merits further investigation. It should be stressed that FlrC of S. oneidensis is active in its unphosphorylated form, contrasting with its V. cholerae counterpart (23). This is important because FlrC may not be promptly phosphorylated by FlrB when it is in excess.

FlrC, but not FlrA, activates the transcription of the flagellin gene flaA. By altering the ratio of FlaA to FlaB, enhanced production of FlaA results in reduced motility (17). This phenomenon does not occur in cells overproducing FlrA, whose motility defect therefore is mainly due to multiple flagella and thus is much less severe. Although FlrA and FlrC are classical bEBPs, they bear different features. Most σ⁵⁴-dependent activators, such as FlrA, typically bind to the sites upstream of the RNAP-σ⁵⁴ binding site, but FlrC binds the elements located downstream (24, 43). This difference, along with the fact that the σ⁵⁴-dependent promoter of flaA is not sufficiently conserved, probably prevents FlrA from binding.

Studies of V. cholerae FlrB and FlrC have concluded that these two proteins constitute a canonical TCS (23, 24, 44). In S. oneidensis, the absence of either FlrB or FlrC does not cause any detectable phenotype, but loss of the entire TCS increases motility, largely by reducing the ratio of FlaA to FlaB (17). Since FlrC is active only in its unphosphorylated form, the negligible effect of the FlrB loss implies that there is a quantity threshold for FlrC to elicit a physiologically significant impact. Meanwhile, why the depletion of FlrC alone fails to negatively mediate the expression of the flaA gene remains elusive. Unlike most histidine kinase homologs, FlrB and P. aeruginosa FleS are cytoplasmically located, and to date, signals that trigger the autophosphorylation of FlrB remain unidentified. Given these unexpected puzzles, this TCS represents one of the urgent challenges in the field of bacterial motility.

In this study, we have unraveled mechanisms underlying multiflagellum phenotypes. We have previously observed two different forms of multiflagellum phenotype: the lophotrichous pattern resulting from the deletion of flhG and the peritrichous pattern resulting from the deletion of flhFG (17, 19). The difference relies on FlhF, a polarly positioned multidomain GTPase that determines flagellar location (19). In its absence, cells develop either a single randomly located flagellum or flagella in a peritrichous pattern if a condition for multiflagellum phenotype is provided. The number of flagella is controlled by MinD-like ATPase FlhG (33); the lack of FlhG, either alone or together with FlhF, results in a multiflagellum phenotype. Interestingly, although FlhF is present, excess FlrA and FlrC prompt cells to assemble flagella in a peritrichous pattern, as revealed here and previously (17). One possible explanation is that the resultant overexpression of flagellar genes exceeds the quantity of free FlhF that is required for positioning at poles.

Importantly, we found that the multiflagellum phenotype resulting from the deletion of flhG or flhFG is due to enhanced activity of FlrA. The association of FlhG with FlrA was initially established with the P. aeruginosa FlhG ortholog, FleN; FlhG and FleN share 35% sequence identity and are highly similar in structure (32–34). FleN exerts an antagonistic effect against FleQ through direct protein-protein interaction, to which both ATP hydrolysis and dimerization of FleN are essential (45). A recent study demonstrated that the regulatory activity of Shewanella FlrA is also antagonized by FlhG, suggesting that FlhG plays a similar role in FlrA biology (35). In addition to the direct inhibition, we showed that FlhG also has a repressing effect on the expression of the flrA gene. Intriguingly, such regulation appears to involve FlhF, because depletion of both FlhF and FlhG together results in a stronger impact on flrA expression. Since both FlhF and FlhG possess nucleotide hydrolyzing activity, we speculate that this feature may be a contributing factor. Efforts to test this notion are under way.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

All bacterial strains and plasmids used in this study are listed in Table 2. Information for primers used in this study is available upon request. E. coli and S. oneidensis were grown in lysogeny broth (LB; Difco, Detroit, MI) medium under aerobic conditions at 37 and 30°C for genetic manipulation. When appropriate, the growth medium was supplemented with the following chemicals at the following final concentrations: 2,6-diaminopimelic acid (DAP), 0.3 mM; ampicillin, 50 μg/ml; kanamycin, 50 μg/ml; gentamicin, 15 μg/ml; streptomycin, 100 μg/ml.

TABLE 2.

Strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
Strains
E. coli
DH5α	Host for cloning	Lab stock
WM3064	Donor strain for conjugation; ΔdapA	W. Metcalf, UIUC
XL1-Blue MRF′	Host for B1H vectors	Stratagene
BL-21	Host for expressing recombinant proteins	Novagen
S. oneidensis
MR-1	Wild type	Lab stock
FFM	ΔflaA ΔflaB; derived from MR-1	14
HG3210	ΔfliA; derived from MR-1	15
HG3211	ΔflhG; derived from MR-1	17
HG3212	ΔflhF; derived from MR-1	19
HG3212-1	ΔflhFG; derived from MR-1	19
HG3230-1	ΔflrBC; derived from MR-1	17
HG3232	ΔflrA; derived from MR-1	15
HG3237	ΔflaB; derived from MR-1	14
HG3238	ΔflaA; derived from MR-1	14
HG3961	ΔrpoN; derived from MR-1	15
ΔflrA ΔflaA strain	ΔflrA; derived from ΔflaA strain	This study
ΔflrA ΔflaB strain	ΔflrA; derived from ΔflaB strain	This study
ΔflrA ΔflhG strain	ΔflrA; derived from ΔflhG strain	This study
ΔflrA ΔflhFG strain	ΔflrA; derived from ΔflhFG strain	This study
ΔflrAS strain	Spontaneous mutant from ΔflrA strain	This study
FFM ΔflrAS	ΔflaA ΔflaB; derived from ΔflrAS strain	This study
ΔflrAS ΔflhFG strain	ΔflhFG; derived from ΔflrAS strain	This study
Plasmids
pHGM01	Apr Gmr Cmr; att-based suicide vector	46
pHGEI01	Integrative lacZ reporter vector	28
pBBR-Cre	Spr helper plasmid used with pHGC01	27
pHGE-Ptac	Kmr; IPTG-inducible Ptac expression vector	31
pET-28a(+)	His-tagged protein expression vector; Apr	Novagen
pBTcmT	B1H vector for DNA fragments	37
pTRG	B1H vector for regulators	Stratagene
pHGEI01-PflrC	Vector for measuring PflrC activity	This study
pHGEI01-PflrC(M)	Vector for measuring PflrC(M) activity	This study
pHGEI01-PflaB	Vector for measuring PflaB activity	This study
pHGEI01-PpomA	Vector for measuring PpomA activity	This study
pHGEI01-PflgM	Vector for measuring PflgM activity	This study
pHGE-Ptac-flrC	Inducible expression of flrC	17
pHGE-Ptac-flrA	Inducible expression of flrA	This study
pHGE-Ptac-flhFG	Inducible expression of flhFG	This study
pBTcmT-PflaA	pBTcmT carrying PflaA	17
pBTcmT-PflaB	pBTcmT carrying PflaB	17
pBTcmT-PflhFG	pBTcmT carrying PflhFG	This study
pTRG-flrC	pTRG expressing FlrC	17
pTRG-flrCD54A	pTRG expressing FlrCD54A	17
pTRG-flrA	pTRG expressing FlrA	This study
pET28(a)-flrA	pET28(a) expressing FlrA	This study
pET28(a)-flrC	pET28(a) expressing FlrC	This study
pET28(a)-flrCD54A	pET28(a) expressing FlrCD54A	This study

In-frame mutant construction and complementation.

In-frame deletion strains for S. oneidensis were constructed using the att-based fusion PCR method as described previously (46). In brief, two fragments flanking the gene of interest were amplified by PCR and were linked by a second round of PCR. The fusion fragments were introduced into plasmid pHGM1.0 by using Gateway BP clonase II enzyme mix (Invitrogen) according to the manufacturer's instructions, giving mutagenesis vectors in E. coli WM3064, which were subsequently transferred into S. oneidensis via conjugation. Integration of the mutagenesis constructs into the chromosome was selected by resistance to gentamicin and was confirmed by PCR. Verified transconjugants were grown in LB broth in the absence of NaCl and were plated on LB supplemented with 10% sucrose. Gentamicin-sensitive and sucrose-resistant colonies were screened by PCR for deletion of the target gene. Mutants were verified by sequencing the site for the intended mutation.

Genetic complementation of mutants was carried out with the IPTG-inducible gene expression system pHGE-Ptac (31). The gene of interest generated by PCR was introduced into the vector. After sequencing verification, the resulting vectors were transferred into the relevant strains via conjugation for complementation and/or expression.

Physiological characterization.

The growth of flagellar mutant strains constructed previously and in this study in LB was measured by recording the OD₆₀₀ of cultures under aerobic conditions in triplicate with the wild type as the reference. Motility testing (swimming) was performed with semisolid LB agar plates (0.25% [wt/vol] agar). Briefly, cultures of relevant strains at the mid-log-phase (OD₆₀₀, ∼0.4; the same afterwards) were adjusted to similar ODs with fresh LB broth, and 5 μl of each resulting culture was spotted onto a swimming plate by piercing it with a thin pipette tip. For comparison, the wild type was always included on each plate. Plates were incubated at room temperature, and photographs were taken 16 h later.

Microscopic analysis.

Both phase-contrast microscopes and transmission electron microscopes (TEMs) were used to visualize flagellar filaments of swimming cells scraped from the leading edges of each swarm. For flagellar counting, cells were stained for flagellar filaments and were visualized on a glass slide with a Motic BA310 phase-contrast microscope. Micrographs were captured with a Moticam 2306 charge-coupled-device camera and Motic Images Advanced software, version 3.2. For TEM, cells were collected, resuspended with phosphate-buffered saline (PBS), and fixed by the addition of glutaric dialdehyde at a final concentration of 2% as described previously (14). Samples were then applied to a carbon-coated grid and were stained with 2% phosphotungstic acid neutralized at pH 6.8 before viewing under a CM12 Philips TEM.

Flagellin purification.

Flagellar filament isolation and purification were performed essentially as described previously (16). In brief, a 250-ml bacterial batch culture was centrifuged at 5,000 × g for 10 min at 4°C. The cell pellet was resuspended in 10 ml PBS buffer, pH 7.0, and was vortexed for 20 min to shear off flagella. Cells were removed by centrifugation at 10,000 × g for 30 min at 4°C, and the supernatant containing flagellar filaments was filtered through a 0.45-μm-pore-size filter. The filtrate was centrifuged at 100,000 × g for 2 h, and the pellet containing concentrated flagellar filaments was resuspended in double-distilled water (ddH₂O). Throughout this study, protein concentrations of samples were determined by a Bradford assay using bicinchoninic acid (BCA) as a standard with a Pierce BCA protein assay kit (Thermo). Sample purity was checked by SDS-PAGE with Coomassie brilliant blue R-250 staining (Sigma, St. Louis, MO).

β-Galactosidase activity assay.

A β-galactosidase activity assay was performed to determine gene expression. DNA fragments of ∼400 bp covering sequences upstream of target genes were amplified and placed in front of the full-length E. coli lacZ gene on plasmid pHGEI01 (28). The resulting vector was verified by sequencing, then transformed into E. coli WM3064, and conjugated with relevant S. oneidensis strains. Cells of the mid-log phase were harvested by centrifugation, washed with PBS, and treated with lysis buffer (0.25 M Tris-HCl [pH 7.5], 0.5% Triton X-100) for 30 min. The resulting soluble protein was collected after centrifugation and was used for an enzyme assay by adding o-nitrophenyl-β-d-galactopyranoside (ONPG) (4 mg/ml). β-Galactosidase activity was determined by monitoring color development at 420 nm using a Synergy 2 Pro200 multidetection microplate reader (Tecan) and is presented as Miller units.

qRT-PCR.

Total RNA was isolated from S. oneidensis cells at mid-log phase using a combination of Trizol (Invitrogen) with the RNeasy minikit (Qiagen), and quantitative real-time PCR (qRT-PCR) analyses were carried out with an ABI 7300 96-well system (Applied Biosystems) as described previously (47). The expression of each gene was determined from three replicas in a single qRT-PCR experiment. The cycle threshold (C_T) values for each gene of interest were averaged and normalized against the C_T value of the arcA gene, whose abundance was constant under experimental conditions (48). The relative abundance (RA) of each gene was standardized to the C_T value of the arcA gene using the equation RA = 2^−ΔCT, yielding similar fold differences.

Expression and purification of FlrA and FlrC recombinant proteins.

E. coli BL21(DE3) and the pET28a plasmid were used for the production of His₆-tagged recombinant FlrA and FlrC variants as described previously (49). Soluble FlrA and FlrC variants were obtained following induction with 0.2 mM isopropyl-β-d-1-thiogalactopyranoside at 16°C overnight. Following lysis by French pressure cell (Constant Systems Ltd.) treatment and centrifugation, the clarified bacterial supernatant was loaded onto a nickel-ion affinity column (Qiagen). After the removal of contaminant proteins with a washing buffer containing 20 mM imidazole, the His-tagged FlhF variants were eluted in elution buffer containing 100 mM imidazole. The protein was concentrated by ultrafiltration (10-kDa cutoff) and exchanged into 20 mM Tris-HCl (pH 8.0) containing 150 mM NaCl. The peak fractions were analyzed by 12% SDS-PAGE.

EMSA.

FlrC phosphorylation was performed in a buffer containing 100 mM Tris-HCl (pH 7.0), 10 mM MgCl₂, 125 mM KCl, and 50 mM dilithium carbamoyl phosphate for 60 min at room temperature as described previously (50). The probes used for EMSA were prepared by PCR with ³³P-end-labeled primers. The binding reaction was performed with ∼25 fmol (∼2 nM) labeled probes and various amounts of protein in 12 μl binding buffer containing 100 mM Tris-HCl (pH 7.4), 20 mM KCl, 10 mM MgCl₂, 2 mM dithiothreitol (DTT), and 10% glycerol at 15°C for 60 min, and proteins were resolved on prerun 4.8% polyacrylamide native gels (50). The band shifts were visualized by autoradiography.

B1H assay.

A bacterial one-hybrid (B1H) system (Stratagene) was used to investigate DNA-protein interaction in vivo in E. coli cells (37). Plasmid constructs were created by cloning the bait DNA and target protein into the pBXcmT and pTRG vectors and were verified by sequencing as described before (36). The resultant plasmids were used to cotransform BacterioMatch II Validation Reporter competent cells on M9 salt agar plates containing 25 mg/ml chloramphenicol and 12.5 mg/ml tetracycline with or without 3-amino-1,2,4-triazole (3-AT). The system was validated previously with the pBXcmT-P_katB/pTRG-OxyR and pBXcmT-P_16S/pTRG-OxyR pairs (17). The plates were incubated for 24 h and were then moved to room temperature for an additional 16 h (the colonies indicating positive interaction usually appeared between 18 and 24 h). The positive interactions were confirmed by streaking colonies onto plates containing both 3-AT and streptomycin (12.5 mg/ml).

Immunoblotting assays.

Immunoblot analysis was performed essentially as described previously (50). Proteins separated by SDS-PAGE were electrophoretically transferred to a polyvinylidene difluoride (PVDF) membrane according to the manufacturer's instructions (Bio-Rad). The gels were blotted for 2 h at 60 V using a Criterion blotter (Bio-Rad). The blotting membrane was probed with rabbit monoclonal antibodies against green fluorescent protein (GFP). Horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Roche Diagnostics) was used as the secondary antibody (1:5,000), and the signal was detected using a chemiluminescence Western blotting kit (Roche Diagnostics) in accordance with the manufacturer's instructions. Images were visualized with a UVP imaging system.

Other analyses.

The relative intensities of specific protein signals were measured using ImageJ (51). Experimental values were subjected to statistical analyses and are presented as means ± SEM (standard errors of the mean). Student's t test was performed for pairwise comparisons of groups.