The G-protein FlhF has a role in polar flagellar placement and general stress response induction in Pseudomonas putida

The flhF gene is part of a large operon

To determine the identity of the gene affected in MK107, the wild-type (MK1; Table 1) genomic DNA corresponding to the transposon insertion region of MK107 was sequenced (Experimental procedures). Figure 1, which is derived from our own data (submitted to the GenBank database under accession no. AF183382), that of Ditty et al. (1998) and of The Institute of Genomic Research (TlGR; K. E. Nelson, personal communication), indicates that the affected gene, flhF, resides in a large operon, which is concerned with flagellar synthesis, chemotaxis and motility. No obvious transcriptional terminator was seen within the region shown.

The transposon had inserted after the 42nd N-terminal nucleotide of the flhF gene. Based on the predicted amino acid sequence, flhF codes for a 437-amino-acid protein (48 kDa molecular weight), which is hydrophilic, devoid of an N-terminal signal sequence, a helix–turn–helix motif, as well as transmembrane regions. The FlhF protein contains the three consensus sequences typical of GTP-binding proteins and is homologous to the signal recognition particle (SRP) pathway family of proteins (Dever et al., 1987; Carpenter et al., 1992; Wolin, 1994; De Gier et al., 1997). Among the Escherichia coli SRP proteins, FlhF resembles the receptor protein, FtsY, more than the signal recognition particle protein, Ffh. It has an overall 22% identity and 22% similarity to FtsY, with 32% identity and 50% similarity in the GTP-binding C-terminus region (data not shown).

The flhF transposon does not have a polar effect

Given that the flhF gene is part of a large operon, it seemed prudent to ensure that the phenotype of MK107 results solely from the disruption of this gene. Reverse transcriptase–polymerase chain reaction (RT–PCR) and Western analyses were used to determine whether open reading frames (ORFs) downstream of the MK107 transposon-insertion site were expressed. RT–PCR products showed expression of all the ORFs whose transcription we looked for (C-terminus region of flhF, orfC and fliA;Fig. 1) in both MK107 and MK1; Western analysis confirmed that FliA was expressed to the same extent in both strains (data not shown).

MK107 shows altered motility and flagellar arrangement, which are restored by a wild-type flhF allele

As flhF is located in an operon concerned with cell motility and has been implicated in flagellar biogenesis, we investigated the motility of MK107. It failed to spread on motility plates, in contrast to the wild type (Fig. 2A and B, insets), and phase-contrast microscopy revealed that, although actively motile, it lacked directional movement. Electron micrographs showed that, unlike the wild-type polar location, the flagella in MK107 were randomly distributed on the cell surface (Fig. 2A and B); the cell shown in Fig. 2B has flagella only on one side, but other cells (Fig. 2E) showed a random distribution. MK107 also possessed on average more flagella per cell than MK1: 3.2 ± 0.5 versus 2 ± 0.3.

That the flhF gene was responsible for the altered motility and flagellar distribution of MK107 is confirmed by the fact that motility agar-spreading, directional motility, as well as polar flagellar location were restored in MK202 (i.e. strain MK107 containing a random insert of a single copy of the wild-type flhF allele on its chromosome; Table 1; Fig. 2C and inset). The average flagellar number also decreased in this strain to resemble the wild type, i.e. 2.6 ± 1 flagella per cell.

FlhF overproduction results in a large number of polar flagella

To determine the effect of overproduction of normal FlhF in strain MK107, we cloned the flhF gene in the pMMB67EH plasmid under the control of the tac promoter, constructing pSP5 and strain MK200 (i.e. MK107 containing pSP5; Table 1). Not only did MK200 regain polar flagella, it showed a marked increase in flagella per cell: 12 ± 2 (Fig. 2D). pMMB67EH, from which pSP5 is derived, generates five to eight copies per cell, and Western analysis showed an ≈ sixfold increase in FlhF levels in MK200 compared with MK201 (i.e. MK107 containing the pMMB67EH without the flhF gene; Table 1; data not shown). The results confirm that the FlhF protein directly or indirectly determines the polar flagellar location and show, in addition, that this protein is the limiting factor in determining the number of flagella per cell. Strain MK200 appeared to regain some directional motility, as judged by phase-contrast microscopy. Nevertheless, it did not spread on motility agar plates (Fig. 2D, inset). The control strain MK201 showed a flagellar arrangement and spread pattern similar to that of MK107 data not shown.

MK200 shows greatly enhanced protein secretion

The flagellar channel has recently been implicated in cellular protein secretion (Young et al., 1999). As the flagellar content and arrangement differs in the above P. putida strains, and FlhF overproduction upregulates flagellar number, these strains could differ in cellular protein secretion. MK1 secreted the least amount and number of proteins (Fig. 3). The secretion was increased somewhat in MK107, while MK202 resembled the wild type in this respect. MK200 showed vastly increased secretion both quantitatively and qualitatively. A T-broth control (Experimental procedures) indicated that the protein bands in lanes 1–4 were not derived from the medium. There was no evidence of cell lysis in any of the cultures, and viable counts continued to increase throughout incubation. This notion is further supported by the fact that Western blots failed to detect the cytoplasmic protein σ^s in the culture supernatants.

Western blots did reveal the presence of the periplasmic protein, β-lactamase, in the culture supernatant of MK200, but not in that of MK201, suggesting that the increased flagellar number enhanced the outer membrane leakiness in the former strain. As the amount of FlhF and flagellar number and arrangement in MK201 is similar to that in MK107, it can be assumed that the latter strain is also not leaky to a periplasmic protein such as β-lactamase. While the secretion of β-lactamase by MK200 implies outer membrane leakiness in the increased protein secretion by this strain (Fig. 3), it remains possible that the increased activity of the type III secretion pathway also has a role; this pathway is made up of homologues of components of the flagellar apparatus (Macnab, 1996).

MK107 is unable to develop SGSR, which is restored by a wild-type flhF allele

As stated above, the flhF gene was shown to have a role in starvation survival of P. putida. Starvation genes fall into two classes: those concerned with starvation survival alone (Fraley et al., 1998); and those that confer SGSR and affect a broad spectrum of starvation protein synthesis (Lange and Hengge-Aronis, 1991; McCann et al., 1991). We therefore determined whether flhF disruption affected SGSR. Starved MK 107 cells were less resistant to heat and H₂O₂, as well as to ethanol compared with MK1 (Table 2). However, there was no difference in the resistance to these stresses of exponential phase cells of the two strains (data not shown). That the disruption of flhF had a role in the compromised SGSR of MK107 was confirmed by the fact that the strain MK201 (Table 1) resembled the wild type in SGSR (Table 2). MK200 (which overproduces FlhF) possessed greater SGSR (10–20%) than MK107, but less than that of MK1 (data not shown).

MK107 is impaired in starvation protein synthesis

As SGSR development requires starvation protein synthesis (Matin et al., 1999), the above results suggest that flhF mutation interferes with this synthesis. A comparison of exponential and stationary phase two-dimensional gel maps (Experimental procedures) showed that the synthesis of at least 50 starvation proteins is altered in MK107; selected regions of the gels (Fig. 4A and B) illustrate the point. Some 30 polypeptides in these regions have higher levels in the starved wild-type cells than in the mutant [e.g. 1–5, 10–12 cluster, 15–20 (Fig. 4A), and 1–6, 8, 10, 13 (Fig. 4B)], although the reverse is true for other polypeptides [e.g. spots 7, and 8 (Fig. 4A) and spot 9 (Fig. 4B)]. The top part of each figure is the polypeptide synthesis map of exponential phase MK1 cells in the corresponding region of the gel; the exponential phase polypeptide synthesis pattern of MK107 cells in this region was very similar.

However, in other regions of the gels, the exponential phase MK1 and MK107 protein synthesis pattern differed markedly (Fig. 5A and B), affecting over 70 polypeptides in the regions shown. Most of the numbered polypeptides are higher in MK1 than in MK107 in the region shown in Fig. 5A, while the opposite was the case with several polypeptides shown in Fig. 5B; overall, most of the affected polypeptides showed a higher synthesis rate in the wild type. As stated above, the flhF gene is expressed at a significant level in the exponential phase cells; these results show that it has a direct or indirect effect on the synthesis of exponential phase proteins. Restoration of the flhF wild-type allele to MK107 tended to restore the wild-type protein synthesis pattern (data not shown).

σ^s levels increase to a lesser extent in starving MK107 than in MK1

Increase in σ^s concentration is responsible for the enhanced synthesis of starvation proteins and SGSR development in several bacteria, including P. putida (Ramos-Gonzalez and Molin, 1998; Matin et al., 1999). σ^s levels were comparable during exponential growth in MK1 and MK107. At the onset of starvation, the levels increased in both, but less in MK107, the difference becoming pronounced after 1 h of starvation (Fig. 6). This difference persisted for the rest of the experiment, i.e. up to 5 h of starvation (data not shown).

Bacterial strains, plasmids, and growth media

All strains and plasmids used in this study are listed in Table 1. Luria–Bertani (LB), tryptone and M9 minimal media were prepared as described previously (Sambrook et al., 1989). The following antibiotics were used at the indicated concentrations (in µg ml⁻¹): for E. coli: ampicillin (100), kanamycin (30), chloramphenicol (20); for P. putida: kanamycin (50), rifampicin (150), carbenicillin (4000), streptomycin (400), spectinomycin (50). IPTG (1 mM) was added to media when needed for the induction of the tac promoter.

Localization of the MK107 transposon-insertion site homologue in the wild-type MK1

The transposon-insertion site in MK107 was cloned previously as a 14.3 kb genomic insert in pMK103 (Kim et al., 1995). A 5.4 kb EcoRI/HindIII region of this insert was cloned to yield pSP1 (Table 1; Fig. 1) and labelled with digoxygenin (DIG; Boehringer Mannheim) to probe a SuperCos1 chromosomal DNA MK1 cosmid library, prepared according to the Stratagene protocol. Cosmid DNA from the positive colonies was digested with EcoRI enzyme. Four of the digests each generated a 9.5 kb fragment, which hybridized with the probe; one of these (from pSP2; Table 1) was cloned in pUC18 (generating pSP3) and sequenced. Also sequenced was the EcoRI/HindIII fragment of the genomic insert in pSP1.

Single-copy complementation of the mutated gene in MK107

Random insertion of the 9.5 kb EcoRI genomic fragment of pSP3 [containing the flhF gene and contiguous sequence (Fig. 1; see Results)] into the MK107 chromosome was attained as described previously (DeLorenzo et al., 1990). The fragment was first flanked with NotI sites by cloning in p18NotI (yielding pSP9) and then cloned in pUT/mini-Tn5, resulting in pSP10 (Table 1). pSP10 was transformed into MK107 via E. coli CC118 λpir strain, as described previously (Kim et al., 1995). Exconjugants were selected on LB plates containing rifampicin, kanamycin, streptomycin and spectinomycin. In several of these, plate-spreading motility was restored. MK202, which exhibited a marked restoration, was selected for further study; Southern hybridization confirmed random insertion of the EcoRI fragment into the MK107 chromosome. Some of the flhF flanking sequences were included in the complementing fragment to ensure expression in vivo. It has been shown that these influence its transcription; moreover, the upstream NtrC binding site (Fig. 1) could regulate its σ⁵⁴ promoter (Kim et al., 1995).

Multicopy complementation of the flhF gene

The flhF gene was cloned using PCR, with pSP3 as template; the primers were complementary to appropriate sequences (accession no. AF183382) and were designed to flank the PCR product with EcoRI and HindIII restriction sites. The PCR product was cloned into pMMB67EH via pSP4. This yielded pSP5, which was introduced into MK107, generating strain MK200. Sequencing ensured that the PCR procedure did not introduce mutations. The flhF gene in pSP5 is controlled by the tac promoter. The control strain (MK201) consisted of MK107 containing the empty plasmid pMMB67EH.

Stress tests

These were carried out essentially as described previously (Groat et al., 1986; Jenkins et al., 1988). Briefly, 24 h-starved cell suspensions (≈ 5 × 10⁸ cells ml⁻¹) were used. For heat challenge, the culture (2 ml) was placed in a preheated 49°C heat block; samples were removed at various intervals and processed for viability determination by spreading serial dilutions on LB plates. For ethanol (12%) and oxidative (2 mM H₂O₂) shocks, the cell suspension was mixed with the indicated agent, and viability was determined periodically.

DNA extraction and sequencing

Total, plasmid and cosmid DNA were isolated as described (Birnboim and Doly, 1979; Del Sal et al., 1988). Large-scale preparations were made using a Qiagen midi kit. DNA bands from agarose gels were isolated by QiaexII gel extraction kit. Sequencing was performed after generating nested deletions, subcloning of restriction fragments or using ordered primers. DNA strider and blast programs were used to identify the open reading frames (ORFs) and the Tfsitescan-dynamicPlus server to analyse regulatory sequences. The helix–turn–helix motif was searched as described by Dodd and Egan (1990).

RNA extraction and RT–PCR

Total RNA was prepared using the Qiagen RNeasy Total RNA kit. RT–PCR reactions were performed using the Promega Access RT–PCR system. Total RNA (1 µg) was digested with RNase-free DNase to remove DNA quantitatively. The following primers (50 pmol) were used: for flhF (forward) 5′-atcggcgcccaggagcag-3′; (reverse) 5′-agctactttgcacacgcttcatggtc-3; for orfC: (forward) 5′-agcttggccgcagggtca-3′; (reverse) 5′-gaggttgtcgccgatttc-3′; for fliA: (forward) 5′-aatatgacgccagcaaag-3′; (reverse) 5′-cgcgcaggttcgacatgg-3′. Negative controls omitted the AMV-reverse transcriptase from the reaction mixture. The PCR products were separated in 2% agarose/TAE gel and visualized by UV transillumination after ethidium bromide staining.

Motility assay

Motility plates were made from half-strength LB solidified with 0.4% Bacto agar. Inoculum from well-isolated colonies was stabbed in the centre of the plates and incubated overnight at 30°C. Similar results were obtained when 5 µl (A₆₆₀ of 2.4) of stationary phase cultures were spotted.

Electron microscopy

Formvar carbon-coated copper grids were placed on drops of diluted (A₆₆₀ of 0.5) late exponential phase culture and incubated for 3 min to allow cell adherence. After three washings in distilled water, the grids were negatively stained (2 min) in 1% uranyl acetate and washed again. To quantify flagella/cell, 200 cells of each strain were examined; flagella were counted only in well-isolated cells: 30 MK1, 73 MK107, 30 MK202 and 28 MK200. A Philips CM12 transmission electron microscope operating at 80 kV was used.

Analysis of secreted proteins

Cells were grown overnight in tryptone broth and pelleted by centrifugation (6000 × g for 20 min). The supernatants were filtered through 0.22 µm filters, and the volumes were normalized to correspond to an A₆₆₀ of 2 and treated with 10% trichloroacetic acid (TCA) to precipitate the proteins. After washing in acetone, the precipitate was dissolved in SDS sample buffer. Equal volumes were heated (95°C for 10 min), run on a 12% SDS–PAGE and visualized by silver staining (Blum et al., 1987). The control sample consisted of culture medium treated in the same way. For determining σ^s and β-lactamase proteins, the SDS–PAGE gels were immunoblotted and analysed using appropriate polyclonal antibodies; extracts of appropriate cells served as positive controls.

FlhF overproduction, purification and preparation of a polyclonal anti-FlhF antiserum

For FlhF overproduction, a DNA fragment containing the flhF gene was cloned into pET28a+ (Novagen), yielding pSP6 (Table 1), which was transformed into E. coli DH5α, generating AMS3000. The overproduced FlhF in strain AMS3000 formed inclusion bodies after induction with IPTG (37°C for 3 h). FlhF was purified according to the Novagen protocol and submitted for antibody production (Josman Laboratories) as described previously (Zgurskaya et al., 1997).

Immunoblotting

Protein (10 µg) of lysed cells was boiled and subjected to 12% SDS–PAGE with subsequent electroblotting and immunodetection, using nitrocellulose membranes (Bio-Rad) and anti-FlhF, -RpoS, -FliA or -β-lactamase polyclonal antiserum. The polyclonal antibody used against RpoS also reacts with an unknown protein of ≈ 55 kDa molecular weight that remains unchanged during growth or phase transitions; this served as an internal control (Zgurskaya et al., 1997). The membranes were developed with the ECL reagent system (Amersham protocol).

Two-dimensional gel electrophoresis

Two-dimensional gel analysis was performed by Kendrick Laboratories according to the method of O’Farrell (1975), essentially as described previously (Groat et al., 1986). Cells were labelled by 15 min incubation with a 40 µCi ml⁻¹ culture of ³⁵S-‘Met protein express labelling mix’ (specific activity 1175 Ci mmol⁻¹; NEN Life Science). After a 1 min chase with cold 1 mM methionine + cysteine, the cells were washed twice in prewarmed M9 medium, lysed and treated with DNase and RNase. An aliquot was added to BSA carrier solution and ice-cold 50% TCA, centrifuged and washed with 10% TCA. The pellet was dissolved in Soluene 350 (Kendrick Laboratories) and counted. Samples of 10⁷ d.p.m. were loaded on the gels for each culture. Mid-exponential (A₆₆₀ of 1.0; 0.3% glucose–M9 medium) or starving-phase cells were used; the latter were prepared by suspending mid-exponential phase cells in prewarmed glucose-free M9 medium and incubating for 40 min at 30°C (Givskov et al., 1994). Isoelectric focusing used 2.0% pH 4–8 ampholines (Hoefer Scientific Instruments) and 9600 V-h. After equilibration of the tube gels in SDS sample buffer, SDS slab gel electrophoresis (4 h at 12.5 mA/gel) was performed. Appropriate molecular weight markers were used. The gels were dried and autoradiographed using Kodak X-OMAT AR film. Autoradiographs were acquired on an AGFA flatbed scanner (ARCUS II) at 1.25-fold of the original size (300 dots per inch) through Adobe PhotoShop, version 5.5. Boxes and circles were placed on the scans using Adobe Illustrator version 8.0. Selected boxes were enlarged fourfold. Background was adjusted to equal levels in all scans.