Assembling Flagella in Salmonella Mutant Strains Producing a Type III Export Apparatus without FliO

Abstract

The type III export apparatus of the Salmonella flagellum consists of six transmembrane proteins (FlhA, FlhB, FliO, FliP, FliQ, and FliR) and three soluble proteins (FliH, FliI, and FliJ). Deletion of the fliO gene creates a mutant strain that is poorly motile; however, suppressor mutations in the fliP gene can partially rescue motility. To further understand the mechanism of suppression of a fliO deletion mutation, we isolated new suppressor mutant strains with partially rescued motility. Whole-genome sequence analysis of these strains found a missense mutation that localized to the clpP gene [clpP(V20F)], which encodes the ClpP subunit of the ClpXP protease, and a synonymous mutation that localized to the fliA gene [fliA(+36T→C)], which encodes the flagellar sigma factor, σ²⁸. Combining these suppressor mutations with mutations in the fliP gene additively rescued motility and biosynthesis of the flagella in fliO deletion mutant strains. Motility was also rescued by an flgM deletion mutation or by plasmids carrying either the flhDC or fliA gene. The fliA(+36T→C) mutation increased mRNA translation of a fliA′-lacZ gene fusion, and immunoblot analysis revealed that the mutation increased levels of σ²⁸. Quantitative real-time reverse transcriptase PCR showed that either the clpP(V20F) or fliA(+36T→C) mutation rescued expression of class 3 flagellar and chemotaxis genes; still, the suppressor mutations in the fliP gene had a greater effect on bypassing the loss of fliO function. This suggests that the function of FliO is closely associated with regulation of FliP during assembly of the flagellum.

INTRODUCTION

The bacterial flagellum enables cells to swim through liquid environments and to swarm and spread on surfaces (1, 2). The best-characterized flagellum belongs to a member of the Gammaproteobacteria, Salmonella enterica serovar Typhimurium. The flagellum consists of three parts: the basal body, the hook, and the filament (1). Synthesis of the flagellum is coupled with regulation of expression of flagellar genes. The flagellar regulon of Salmonella is organized into a three-tier hierarchy of class 1, class 2, and class 3 flagellar genes (3). Class 1 genes belong to the flhDC operon, which encodes the FlhD₄FlhC₂ transcription factor. Class 2 genes require FlhD₄FlhC₂ for expression and include genes that encode the components of the hook-basal body complex. They also include the fliAZY operon: fliA encodes the flagellar gene sigma factor, σ²⁸, fliZ encodes an activator of class 2 flagellar gene expression, and fliY is a nonflagellar gene with its own promoter (4, 5). The class 3 genes encode proteins that make the filament, motor, and chemotaxis proteins. Expression of class 3 genes requires RNA polymerase and σ²⁸. Some flagellar genes (including the fliAZY operon) are expressed from both class 2 and class 3 promoters (3–5).

Within the center of the flagellar basal body pore is a type III export apparatus that exports the external building blocks of the flagellum. A homologous type III export apparatus is found within the type III secretion injectisomes that are produced by some pathogenic Gram-negative bacteria, including S. enterica (6). The injectisomes export effector proteins from the cytoplasm of the bacterium and inject them into eukaryotic cells during infection. The type III export apparatus uses energy from ATP hydrolysis and the proton motive force across the cytoplasmic membrane for protein export (7, 8). The flagellar export apparatus is made of six transmembrane proteins (FlhA, FlhB, FliO, FliP, FliQ, and FliR) and three soluble proteins (FliH, FliI, and FliJ) (1, 6). FlhA, FlhB, FliP, FliQ, and FliR are highly conserved and essential for export apparatus function (1, 6). FlhA (75 kDa) has a 40-kDa C-terminal cytoplasmic domain, and FlhB (42 kDa) has a 19-kDa C-terminal cytoplasmic domain that functions in export substrate docking (9). FliI (49 kDa) is an ATPase that together with FliH (26 kDa) and FliJ (17 kDa) forms an ATPase ring complex at the export gate that also plays an important role in substrate recognition (9). The functions of FliO (13 kDa), FliP (25 kDa), FliQ (10 kDa), and FliR (29 kDa) are unknown. FliP is made with an N-terminal signal peptide that is required for insertion of FliP into the cytoplasmic membrane and afterwards is cleaved to produce mature FliP (10). Mature FliP and FliR copurify with the basal body (11).

FliO is a bitopic membrane protein that has an 11-kDa cytoplasmic domain and is less conserved than the other export apparatus proteins. A homolog of FliO exists for the flagella of many species of proteobacteria (12, 13), but a homolog has not been identified for the flagella of some distantly related species, such as Aquifex aeolicus, and a homolog is absent from the injectisomes (14). In a previous study, we made an in-frame fliO gene deletion mutant strain of S. enterica, and it was poorly motile (15). However, it produced suppressor mutant strains with somewhat rescued motility. We characterized two suppressor mutant strains and found suppressor mutations by using chain-termination dideoxynucleotide sequencing of genes encoding proteins that make the type III export apparatus. For each strain, a suppressor mutation was localized to the fliP gene. One suppressor mutant strain bore a fliP(F190L) missense mutation. We determined that the other suppressor mutant strain bore at least two suppressor mutations, a fliP(R143H) missense mutation, and another, unknown mutation. The unknown mutation was confirmed because a ΔfliO fliP(R143H) double-mutant strain made with bacteriophage lambda Red homologous recombination was less motile than the suppressor mutant strain. We have now identified the unknown mutation by using comparative genome sequence analysis. In addition, we investigated the source of rescued motility of other suppressor mutant strains and found that the newly identified mutations affect regulation of expression of flagellar genes.

MATERIALS AND METHODS

Growth of bacterial strains.

The S. enterica strains and the plasmids used in this study are listed in Table 1 (16–23). Escherichia coli NovaBlue (Novagen, EMD Millipore) was used for DNA manipulation. Bacterial strains were routinely grown aerobically in Luria-Bertani (LB) medium (24) with shaking (250 rpm) at 37°C. Ampicillin was used at 100 μg ml⁻¹ to maintain plasmids in S. enterica and at 50 μg ml⁻¹ to maintain plasmids in E. coli. Kanamycin was used at 50 μg ml⁻¹. Tetracycline was used at 15 μg ml⁻¹. All chemicals were obtained from Sigma-Aldrich or Wako (Japan), unless otherwise indicated.

TABLE 1.

Salmonella strains and plasmids used for this study

Strain or plasmid	Genotype and/or description	Source and/or referencea
Salmonella enterica serovar Typhimurium strainsb
SJW1103	Wild type for motility and chemotaxis; LT2 (fljB-off) Δ(hin-fljAB)	16
JR501	Strain for converting plasmids to Salmonella compatibility; restriction negative, modification positive	17
TT13206	LT7 phoN51::Tn10-11(Tetr)	18; SGSC3718
SJW1368	Δ(cheW-flhD); nonmotile and without flagella	19
CB184	Δ(fliO-fliP)22251::FRT	15
CB187	ΔfliO22252::FRT	15
CB190	ΔfliO22252 fliA22300(+36T→C); suppressor mutant strain
CB191	ΔfliO22252 fliP22254(R143H) clpP96(V20F); suppressor mutant strain	15
CB228	ΔfliO22252; suppressor mutant strain
CB229	ΔfliO22252; suppressor mutant strain
CB230	ΔfliO22252; suppressor mutant strain
CB281	ΔfliO22252 fliP22254(R143H)	15
CB282	ΔfliO22252 fliP22255(F190L)	15
CB382	ΔfliO22252 ΔclpP100::tetRA
CB383	ΔfliO22252 fliP22254(R143H) ΔclpP100::tetRA
CB389	ΔfliO22252 ΔclpA97::FKF
CB408	ΔfliO22252 Δ(clpP-clpX)98::FKF
CB432	ΔfliO22252 ΔclpA97::FRT
CB433	ΔfliO22252 Δ(clpP-clpX)98::FRT
CB443	ΔfliO22252 clpP96(V20F)
CB447	ΔfliO22252 fliP22254(R143H) clpP96(V20F)
CB476	ΔfliA22301::tetRA
CB478	ΔfliO22252 ΔfliA22301::tetRA
CB484	ΔfliO22252 ΔclpX99::FKF
CB502	ΔfliO22252 fliA22300(+36T→C)
CB506	ΔfliO22252 ΔclpX99::FRT
CB558	fliA22300(+36T→C)
CB588	fliO22252 fliP22254(R143H) ΔfliA22301::tetRA
CB589	fliO22252 fliP22255(F190L) ΔfliA22301::tetRA
CB590	fliO22252 fliP22254(R143H) fliA22300(+36T→C)
CB591	fliO22252 fliP22255(F190L) fliA22300(+36T→C)
CB593	ΔfliO22252 ΔfliP22305::tetRA
CB594	ΔfliO22252 ΔfliP22305::tetRA fliA22300(+36T→C)
CB595	ΔfliO22252 ΔfliA22305::tetRA clpP96(V20F)
CB597	ΔfliO22252 fliP22254(R143H) fliP22255(F190L)
CB598	ΔfliO22252 fliP22254(R143H) fliP22255(F190L) fliA22300(+36T→C)
CB599	ΔfliO22252 fliA22300(+36T→C) clpP96(V20F)
CB818	ΔflgM22443::FKF
CB820	ΔfliO22252 ΔflgM22443::FKF
CB821	ΔfliO22252 ΔflgM22443::FRT
Plasmidsc
pKD46	Bacteriophage λ Red expression plasmid; Ampr	20
pKD13	Carries an FRT-flanked kanamycin resistance cassette; Ampr Kmr	20
pKD4	Carries an FRT-flanked kanamycin resistance cassette; Ampr Kmr	20
pCP20	Flp expression plasmid; Ampr Camr	21
pTrc99A-FF4	Expression vector; Ampr	10
pET-22b(+)	pT7 expression vector, optional C-terminal His tag; Ampr	Novagen
pFZY1	Transcriptional lacZ fusion vector; Ampr	22; NIG
pMC1403	Translational lacZ fusion vector; Ampr	23; NIG
pCB20	pTrc99A-FF4 carrying fliO	15
pCB211	pTrc99A-FF4 carrying fliP	15
pCB259	pTrc99A-FF4 carrying fliO and fliP(R143H)	15
pCB320	pTrc99A-FF4 carrying fliP and fliO
pCB322	pTrc99A-FF4 carrying fliP(R143H)
pCB323	pTrc99A-FF4 carrying fliP(F190L)
pCB331	pET-22b(+) carrying fliP codons 106 to 185 and codons for a C-terminal 6×His tag
pCB568	pTrc99A-FF4 carrying fliA
pCB569	pTrc99A-FF4 carrying fliZ
pCB572d	pFZY1 fliA (−116 to +400)
pCB573d	pFZY1 fliA (−116 to +400), synonymous mutation (+36T→C)
pCB574d	pMC1403 fliA (−118 to +399)
pCB575d	pMC1403 fliA (−118 to +399), synonymous mutation (+36T→C)
pCB576d	pMC1403 fliA (−118 to +399), P2 negative (−33GCC−31→−33CGT−31)
pCB577d	pMC1403 fliA (−118 to +399), P2 negative (−33GCC−31→−33CGT−31), synonymous mutation (+36T→C)
pCB578d	pMC1403 fliA (−58 to +399)
pCB579d	pMC1403 fliA (−58 to +399), synonymous mutation (+36T→C)
pCB587	pTrc99A-FF4 carrying flhDC
pCB596	pTrc99A-FF4 carrying fliO and fliP(R143H, F190L)
pCB659	pTrc99A-FF4 carrying fliC
pCB660	pTrc99A-FF4 carrying fliS
pCB661	pTrc99A-FF4 carrying fliS and fliC

^aAll strains and plasmids were made in this study unless indicated otherwise. SGSC, Salmonella Genetic Stock Center, University of Calgary, Canada; NIG, National Institute of Genetics, Japan.

^bFRT refers to the flippase enzyme recognition target site. FKF is an FRT-flanked kanamycin resistance cassette.

^cFlp refers to the recombinase enzyme.

^dThe DNA regions inserted into the plasmids were numbered relative to the translation initiation codon of the fliA gene, which was considered position 1.

Construction of strains and plasmids.

Replacement of wild-type genes on the chromosome with mutant alleles was done with the bacteriophage lambda Red recombination method of Karlinsey (25). Chromosomal gene deletions were made with the lambda Red recombination method of Datsenko and Wanner (20). Full details of strain and plasmid construction are given in the supplemental material. Oligonucleotide primers used for strain and plasmid construction are listed in Table S1 in the supplemental material and were synthesized by Life Technologies. DNA manipulations were performed according to standard procedures (24). Site-directed mutagenesis was performed using a QuikChange Lightning site-directed mutagenesis kit (Agilent Technologies).

Isolation of suppressor mutant strains and motility assays.

Soft tryptone agar plates (0.35% [wt/vol] agar) were used in motility assays at 30°C (26). Antibiotic was added to maintain plasmids as required. To produce suppressor mutant strains from the fliO deletion mutant strain (CB187), soft tryptone agar was inoculated with a 10-μl streak of an overnight culture. Suppressor mutants arose after about 24 h. Swimming rates were determined by measuring the increase in swim-ring diameter (mm) at intervals (every 1 to 4 h). The statistical significance of the difference in swimming rate between samples was determined with two-tailed Student's t test.

Whole-genome sequencing.

A Genome Analyzer IIx (Illumina, San Diego, CA) instrument was used for parallel sequencing of the 4.9-Mbp whole genome of the wild-type strain SJW1103, the fliO deletion mutant strain CB187, and five suppressor mutant strains (CB190, CB191, CB228, CB229, and CB230) with rescued motility produced by CB187. A multiplexed 76-bp, single-end-read genomic DNA sequencing run was used. A total of 530 Mbp of data was obtained for each genome. Library samples were prepared according to Illumina's instructions for multiplexing. Processing of raw sequence data from the Genome Analyzer IIx was performed using Illumina Genome Analyzer Pipeline software. The genome sequence of S. enterica serovar Typhimurium LT2 (NC_003197.1; Typhimurium.fasta) was used as the reference platform. Sequence data were then analyzed to identify fixed mutations by using inGAP software (27) and independently by mapping with the Burrows-Wheeler Aligner (http://bio-bwa.sourceforge.net/), formatting with SAMtools (http://samtools.sourceforge.net/), and single nucleotide polymorphism calling with SAMtools and BCFtools.

Immunoblot analyses.

Immunoblotting was performed similarly to a previously described method (28). Colonies of S. enterica were inoculated into 10 ml LB medium with an appropriate antibiotic and grown with shaking at 37°C to the required growth phase, or colonies of bacteria transformed with plasmids were inoculated into 10 ml LB medium with antibiotic and grown until an optical density at 600 nm (OD₆₀₀) of 0.8, and then protein expression was induced for 2 h with the addition of 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). Cells (normalized to 0.5 ml culture broth for an OD₆₀₀ of 1.0) were harvested by centrifugation at 16,100 × g for 5 min. Pellets were suspended in 40 μl sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (100 mM Tris-HCl [pH 6.8], 2% SDS, 1 mM β-mercaptoethanol, 7 M urea, and 0.1% bromophenol blue) and incubated at 95°C for 15 min. Equivalent amounts of whole-cell lysates (4 to 10 μl) were separated by SDS-PAGE, and proteins were transferred to polyvinylidene difluoride (PVDF) membranes. Bound antibodies were detected using the WesternBreeze chromogenic immunodetection system (Life Technologies). Band densitometry was performed using a ChemiDoc XRS+ system with Image Lab software (Bio-Rad). The statistical significance of the difference in band densitometry between samples was determined with two-tailed Student's t test.

Rabbit sera containing polyclonal antibodies were used at the following dilutions: σ²⁸, 1:1,000; FliS, 1:2,000; FliO, 1:10,000; and FliP, 1:10 (after purification). To produce antiserum containing polyclonal antibodies for FliP, a FliP_(106–185)-polyhistidine tag construct was purified using nickel affinity chromatography under denaturing conditions (29). However, further purification of FliP antibodies from the antiserum was required, by adsorption of nonspecific antibodies (30). Briefly, 1.5 mg protein from whole-cell lysate of strain CB184 [Δ(fliO-fliP)] was transferred to a PVDF membrane and incubated overnight at 4°C with a 1:100 dilution of FliP antiserum in primary antibody wash solution. This was then used as the purified FliP antibody stock.

FlgE and FliC export assays.

Export assays were performed as described previously (31). Briefly, three independent cultures of each S. enterica strain were grown at 37°C in LB medium with an appropriate antibiotic to late exponential phase (OD₆₀₀ of 1.2). Culture broth (1.5 ml) was harvested by centrifugation at 16,100 × g for 5 min to obtain cell pellet and culture supernatant fractions. Proteins secreted into 1.4 ml of the culture supernatant were precipitated with 10% trichloroacetic acid and suspended in 40 μl Tris–SDS-PAGE loading buffer. Cell pellets were then suspended in 40 μl SDS-PAGE loading buffer. Samples were heated at 95°C for 15 min. Proteins were separated by SDS-PAGE, using 12.5% (wt/vol) polyacrylamide gels, and transferred to a PVDF membrane for immunoblot analyses. The following amounts of sample were loaded onto each polyacrylamide gel: supernatant for FliC detection, 5 μl; supernatant for FlgE detection, 20 μl; pellet for FliC detection, 0.5 μl; and pellet for FlgE detection, 3 μl. Rabbit antiserum containing polyclonal antibodies for either FlgE (1:1,000 dilution for the supernatant fractions and 1:10,000 dilution for the cell pellets) or FliC (1:20,000 dilution) was used. Bound antibodies were detected using the WesternBreeze chromogenic immunodetection system (Life Technologies). Band densitometry was performed using a ChemiDoc XRS+ system with Image Lab software (Bio-Rad).

Microscopy of cells.

Motility of cells grown to exponential phase (OD₆₀₀ of 0.8) in 10 ml LB medium at 37°C was examined by phase-contrast microscopy. A BX53 system microscope (Olympus Corporation, Japan) was used with a DP72 digital camera at a magnification of ×1,000. Whole cells were also examined by transmission electron microscopy (12). Cells were harvested by gentle centrifugation (300 × g) and suspended in molecular biology-grade water. Cells were then immobilized on Mextaform HF-34 200-mesh carbon-coated copper grids and stained with 1% uranyl acetate (pH 5). Grids were examined using a JEM-1230R transmission electron microscope (Jeol, Ltd., Japan) at 100 keV. The statistical significance of the difference in the distribution of flagellar numbers between samples was determined with a two-tailed Mann-Whitney U test.

β-Galactosidase assays.

β-Galactosidase activity was assayed by measuring the hydrolysis of o-nitrophenyl-β-d-galactosidase (ONPG) by the method of Miller (32) for cells grown to exponential phase (OD₆₀₀ of 0.65) in LB medium with the appropriate antibiotic at 37°C. The statistical significance of the difference in enzyme activity between samples was determined with two-tailed Student's t test.

RNA isolation and real-time qRT-PCR.

Cells were grown to exponential phase (OD₆₀₀ of 0.65) at 37°C in LB medium. Total RNA was isolated using a RiboPure Bacteria kit (Life Technologies) and was treated with DNase I for 30 min at 37°C to eliminate genomic DNA contamination. RNA quality was determined using an Agilent 2100 bioanalyzer. The RNA was used as the template in first-strand cDNA synthesis using a SuperScript Vilo cDNA synthesis kit, which includes SuperScript III reverse transcriptase (RT) and random primers (Life Technologies). Real-time quantitative RT-PCR (qRT-PCR) mixtures were set up using SYBR Select master mix (Life Technologies). The primers listed in Table S1 in the supplemental material were used at 200 nM in the qRT-PCR mixtures. Reactions and data analysis were performed using a model 7500 real-time PCR system (Applied Biosystems). Three cultures were prepared for each strain (biological triplicates), and the level of target from each culture was determined twice (technical duplicates). Results were analyzed by the relative quantification (ΔΔC_T) method (33). Relative quantities of target mRNAs for each sample were determined by normalization of fluorescence signals for the target mRNAs to those of two endogenous controls (gapA and gyrB mRNAs) (34) and by comparison of the samples to the reference sample (SJW1103; wild type).

RESULTS

Novel suppressor mutations that bypass a loss of fliO function.

An S. enterica mutant strain bearing a nonpolar deletion mutation of the fliO gene is poorly motile but produces suppressor mutant strains with rescued motility about 24 h after inoculation into soft tryptone agar. We previously described mutations in the fliP gene [fliP(R143H) and fliP(F190L)] that bypass a loss of fliO function (15), and here we describe novel mutations that also bypass a loss of fliO function. Genomic DNAs were prepared and sequenced for SJW1103 (wild type), CB187 (the fliO deletion mutant strain), and five suppressor mutant strains: CB190, CB191, CB228, CB229, and CB230. Suppressor mutations were localized by comparative sequence analysis of the genomes (Table 2). CB191 bore the fliP(R143H) mutation and an additional missense mutation in the clpP gene, at codon 20 [clpP(V20F)], encoding the ClpP peptidase subunit. The ClpP protein is made with a leader peptide that is processed, so the mutation produces a V6F substitution in the mature protein. CB190 bore a synonymous mutation within the coding region of the fliA gene [fliA(+36T→C)], encoding the flagellar gene-specific sigma factor σ²⁸. The mutation changed codon 12, encoding aspartate, from 5′-GAT-3′ to 5′-GAC-3′. The clpP and fliA genes regulate flagellar gene expression: the ClpXP protease regulates proteolytic turnover of the flagellar regulon master regulator FlhD₄FlhC₂ (35), and σ²⁸ is required for expression of class 3 and some class 2 flagellar genes (3). No additional mutations were found for CB228, CB229, or CB230. These suppressor mutant strains were unstable and, after repeated subculturing, gradually lost the rescued motility.

TABLE 2.

Randomly selected mutations that rescued motility for strains bearing a fliO deletion mutation

Strain	Gene	Protein name	Mutationa	Amino acid substitutionb
CB190	fliA	σ28, the sigma factor for flagellar genes	+36T→C	None; codon 12 for aspartate was changed from 5′-GAT-3′ to 5′-GAC-3′
CB191	clpP	ATP-dependent Clp protease proteolytic subunit	+58G→T	V20F (V6F)
	fliP	FliP	+428G→A	R143H (R122H)
CB227	fliP	FliP	+568T→C	F190L (F169L)

^aThe position of the mutation is relative to the start codon of translation. Mutations in the fliP gene of CB191 and CB227 were identified previously (15). Mutations in the fliA and clpP genes were found by whole-genome sequencing and confirmed by chain-termination dideoxynucleotide sequencing.

^bClpP and FliP are synthesized with short N-terminal leader peptides of 14 and 21 amino acid residues, respectively, which are cleaved to generate the mature proteins. Positions of the substitutions in the mature proteins are shown in parentheses.

Motility of strains bearing a fliO deletion mutation was rescued additively, either by combining the suppressor mutations or by increased expression of flagellar gene regulators.

To understand the effect of the suppressor mutations, a number of strains were made by using the Karlinsey method (25) of bacteriophage lambda Red homologous recombination. The motility (swimming rate) of the strains in soft tryptone agar was then determined (Table 3). The mean swimming rate of the fliO deletion mutant strain (0.7 ± 0.1 mm h⁻¹) was 10-fold less than that of the wild-type strain (7.0 ± 0.1 mm h⁻¹). Four suppressor mutations that bypass a loss of fliO function [clpP(V20F), fliA(+36T→C), fliP(R143H), and fliP(F190L)] were separately introduced into strains together with a fliO deletion mutation, producing double-mutant strains. All double-mutant strains were significantly (P < 0.05) more motile than the strain bearing the fliO deletion mutation alone. However, the double-mutant strains were all much less motile than the wild type. The fliP(F190L) suppressor mutation was most effective at bypassing a loss of fliO function. The ΔfliO fliP(F190L) double-mutant strain was >4-fold more motile than the fliO deletion mutant strain. A triple-mutant strain [ΔfliO fliP(R143H, F190L)] was made and was 6-fold more motile than the fliO deletion mutant strain. These results suggest that the suppressor mutations in the fliP, clpP, and fliA genes act in independent pathways to bypass the loss of FliO.

TABLE 3.

Swimming rates of strains bearing a fliO deletion mutation with other mutations or harboring plasmids carrying genes encoding flagellar proteins

Strain	Genotype or description	Swimming rate (mm h−1)a
SJW1103	Wild type	7.0 ± 0.1
CB187	ΔfliO	0.7 ± 0.1
CB443	ΔfliO clpP(V20F)	1.7 ± 0.0*
CB502	ΔfliO fliA(+36T→C)	1.6 ± 0.0*
CB281	ΔfliO fliP(R143H)	1.0 ± 0.1*
CB282	ΔfliO fliP(F190L)	3.0 ± 0.6*
CB447	ΔfliO fliP(R143H) clpP(V20F)	3.0 ± 0.1*
CB599	ΔfliO fliA(+36T→C) clpP(V20F)	1.9 ± 0.1*
CB590	ΔfliO fliP(R143H) fliA(+36T→C)	2.2 ± 0.1*
CB591	ΔfliO fliP(F190L) fliA(+36T→C)	5.5 ± 0.3*
CB597	ΔfliO fliP(R143H, F190L)	4.5 ± 0.2*
CB598	ΔfliO fliP(R143H, F190L) fliA(+36T→C)	6.7 ± 0.1*
CB432	ΔfliO ΔclpA	0.6 ± 0.0
CB433	ΔfliO Δ(clpP-clpX)	2.3 ± 0.1*
CB506	ΔfliO ΔclpX	1.7 ± 0.1*
CB821	ΔfliO ΔflgM	1.5 ± 0.1*
CB558	fliA(+36T→C)	7.7 ± 0.1
SJW1368	Δ(cheW-flhD)	0.0 ± 0.0
SJW1103	Wild type + pTrc99A-FF4 (empty plasmid vector)	6.3 ± 0.1
CB187	ΔfliO mutant + pTrc99A-FF4 (empty plasmid vector)	0.6 ± 0.0
CB187	ΔfliO mutant + pCB587 (flhDC)	1.2 ± 0.0*
CB187	ΔfliO mutant + pCB568 (fliA for σ28)	1.3 ± 0.0*
CB187	ΔfliO mutant + pCB569 (fliZ)	0.7 ± 0.1
CB187	ΔfliO mutant + pCB659 (fliC)	0.6 ± 0.0
CB187	ΔfliO mutant + pCB660 (fliS)	0.7 ± 0.1
CB187	ΔfliO mutant + pCB661 (fliS and fliC)	0.6 ± 0.0

^aSwimming rates in soft tryptone agar (0.35% [wt/vol] agar) were determined independently at least four times for each strain by measuring the increase in the diameter of the swim ring over time. Means and standard deviations are shown. An asterisk indicates that the swimming rate of a strain with a fliO deletion mutation either with an additional mutation(s) or harboring a plasmid is significantly different (P < 0.05) from the swimming rate of the mutant strain with only the fliO deletion mutation. The swimming rates of the strains were compared using Student's t test.

The suppressor mutant strain CB191 was a ΔfliO fliP(R143H) clpP(V20F) triple-mutant strain. Strain CB447 was made to have the same genotype by lambda Red homologous recombination to confirm the results of the genome sequence analysis. CB191 and CB447 were equally motile (data not shown). They proved to be significantly more motile than either the ΔfliO fliP(R143H) or ΔfliO clpP(V20F) double-mutant strain (Table 3). Since a cooperative effect was observed for the fliP(R143H) mutation with either the fliP(F190L) or clpP(V20F) mutation, we looked to see if other suppressor mutations were also additive (Table 3). This was done to see whether it would be possible to completely bypass the function of fliO. The effect of the fliA(+36T→C) allele in combination with either the clpP(V20F), fliP(R143H), or fliP(F190L) allele was additive. Impressively, a ΔfliO fliP(R143H, F190L) fliA(+36T→C) quadruple-mutant strain was almost as motile (6.7 ± 0.1 mm h⁻¹) as the wild type. Surprisingly, a strain bearing the fliA(+36T→C) mutation alone was slightly yet significantly (P < 0.05) more motile (7.7 ± 0.1 mm h⁻¹) than the wild type. These results suggest that it is almost possible to compensate for the loss of fliO function by combining the suppressor mutations in the fliP, fliA, and clpP genes and that the fliA(+36T→C) allele is more active than the native fliA gene.

The ClpP peptidase subunit has two AAA⁺ ATPase binding partners, ClpA and ClpX, which direct the specificity of the protease complexes ClpAP and ClpXP, respectively (36). The ClpXP protease, but not the ClpAP protease, is known to regulate the cellular level of FlhD₄FlhC₂ (35). Accordingly, strains bearing clpA, clpP-clpX, or clpX gene deletion mutations together with the fliO deletion mutation were constructed (Table 3). The fliO clpA double-deletion mutant strain was similarly motile to the fliO deletion mutant strain. The clpX or clpP-clpX deletion mutation increased the motility of strains bearing a fliO deletion mutation similarly to the case with the clpP(V20F) mutation. This suggests that inactivation of the ClpXP protease, but not the ClpAP protease, can rescue motility of cells bearing a fliO deletion mutation. The flgM gene encodes a negative regulator of σ²⁸ (37). A fliO flgM double-deletion mutant strain was significantly more motile (P < 0.05) than a fliO deletion mutant strain. This suggests that increased activity of σ²⁸ can rescue motility of cells bearing a fliO deletion mutation.

The clpP and fliA genes encode regulators of flagellar gene expression. Consequently, we investigated whether increased expression of the flagellar transcription factor FlhD₄FlhC₂, FliZ, or σ²⁸ would rescue motility of the fliO deletion mutant strain. Also, we investigated whether increased expression (and supply) of flagellin (encoded by fliC) or the chaperone protein for flagellin (encoded by fliS) separately or together would bypass the loss of fliO function. Plasmids carrying the flhDC, fliA, fliZ, fliC, fliS, or fliS and fliC genes were made. Synthesis of FlhD₄FlhC₂, σ²⁸, FliZ, or FliS by strains harboring these plasmids was confirmed by SDS-PAGE and Coomassie brilliant blue staining or, in the case of FliS, by immunoblot analysis (data not shown). Motility of the fliO deletion mutant strain in soft tryptone agar was rescued 2-fold (P < 0.05) by plasmids carrying the flhDC or fliA gene but not by those carrying the other genes (Table 3). These results suggest that increased expression of all flagellar genes or increased expression of class 3 flagellar and chemotaxis genes can partially bypass the loss of fliO function but that an increased supply of flagellin is not sufficient.

We examined the swimming of cells of the fliO deletion mutant strain and strains bearing additional mutations by using phase-contrast microscopy. Results are presented in Movie S1 and Movie S2 in the supplemental material. Motility of the cells in liquid medium and population heterogeneity can be observed. A small number of cells of the fliO deletion mutant strain were motile, but most exhibited only Brownian motion. The proportion of motile cells was slightly increased by the fliA(+36T→C) or fliP(R143H) suppressor mutation. However, the fliP(F190L) mutation had a greater effect, and most cells of the ΔfliO fliP(F190L) double-mutant strain were motile. Almost all cells of the ΔfliO fliP(R143H, F190L) triple-mutant strain and ΔfliO fliP(R143H, F190L) fliA(+36T→C) quadruple-mutant strain were motile, and motility was close to that of the wild type. These results suggest that cells with a fliO deletion mutation produce functional flagella with a low probability and that the probability of producing flagella is increased additively by the suppressor mutations.

Suppressor mutations rescued biosynthesis of flagella for strains bearing a fliO deletion mutation.

Using immunoblot analysis of flagellin (FliC) and hook (FlgE) proteins made within cells and exported into the culture supernatant, we investigated the biosynthesis of the flagella (Fig. 1 and Fig. 2). For the fliO deletion mutant strain, FliC and FlgE were exported into the culture supernatant at less than 10% of the wild-type levels. The clpP(V20F), fliA(+36T→C), fliP(R143H), and fliP(F190L) suppressor mutations all partially rescued export of the flagellar proteins (Fig. 1). The fliP(F190L) mutation had the greatest effect, which was consistent with earlier observations of motility. The ΔfliO fliP(R143H, F190L) triple-mutant strain and ΔfliO fliP(R143H, F190L) fliA(+36T→C) quadruple-mutant strain exported more FliC and FlgE than the double-mutant strains. The quadruple-mutant strain made and exported these proteins at about two-thirds of the wild-type levels. The strain bearing only a fliA(+36T→C) mutation made and exported FlgE and FliC normally. These results suggest that cells bearing a fliO deletion mutation produced few flagella but that biosynthesis of flagella was rescued by suppressor mutations [fliP(R143H), fliP(F190L), clpP(V20F), and fliA(+36T→C)] that increased the formation of hook-basal body complexes or that activated expression of class 3 flagellar and chemotaxis genes.

FIG 1.

Export of flagellar proteins is rescued by extragenic suppressor mutations in strains bearing a fliO deletion mutation. The strains included were SJW1103 (wild type), CB558 [fliA(+36T→C)], SJW1368 (ΔcheW-flhD), CB187 (ΔfliO), CB443 [ΔfliO clpP(V20F)], CB502 [ΔfliO fliA(+36T→C)], CB281 [ΔfliO fliP(R143H)], CB282 [ΔfliO fliP(F190L)], CB597 [ΔfliO fliP(R143H, F190L)], and CB598 [ΔfliO fliP(R143H, F190L) fliA(+36T→C)]. (A and B) Immunoblot analyses of flagellin (FliC; 52 kDa) levels for cultures of the strains fractionated into supernatant (A) and cell pellet (B) fractions. (C and D) Immunoblot analyses of hook protein (FlgE; 42 kDa) levels in culture supernatant (C) and cell pellet (D) fractions. Polyclonal antibodies for FliC or FlgE were used. The numbers below each lane indicate the relative band densities (%) compared to those of the wild type.

FIG 2.

Export of flagellar proteins is rescued by plasmids carrying genes encoding flagellar regulators in strains bearing a fliO deletion mutation. The strains included were SJW1103 (wild type), SJW1368 (ΔcheW-flhD), and CB187 (ΔfliO). The strains harbored the following plasmids carrying the indicated genes: pTrc99A-FF4 (empty plasmid vector), pCB587 (flhDC), pCB568 (fliA), pCB569 (fliZ), pCB659 (fliC), pCB660 (fliS), and pCB661 (fliS plus fliC). (A and B) Immunoblot analyses of flagellin (FliC; 52 kDa) levels for cultures of the strains fractionated into supernatant (A) and cell pellet (B) fractions. (C and D) Immunoblot analyses of hook protein (FlgE; 42 kDa) in culture supernatant (C) and cell pellet (D) fractions. Polyclonal antibodies for FliC or FlgE were used. The numbers below each lane indicate the relative band densities (%) compared to those of the wild type.

Expression of flhDC, fliA, or fliS plus fliC in the fliO deletion mutant strain increased FliC export to a much greater extent than expression of fliZ, fliS alone, or fliC alone (Fig. 2). However, expression of fliS with fliC did not increase motility for cells bearing a fliO deletion mutation, but expression of flhDC or fliA did (Table 3). Expression of flhDC in the fliO deletion mutant strain greatly increased FlgE export in comparison to the expression of other genes, although there was a small increase of FlgE export caused by fliA or fliZ gene expression (Fig. 2C). These results suggest that for cells bearing a fliO deletion mutation, expression of flhDC markedly increased the formation of hook-basal body complexes and expression of fliA activated expression of class 3 flagellar and chemotaxis genes.

We also examined the biosynthesis of flagella for several strains by using transmission electron microscopy (Fig. 3). Most cells of the fliO deletion mutant strain lacked flagella, but some made one or two flagella. Flagella of the fliO deletion mutant strain were about 3% as abundant as those of the wild type. The proportion of cells producing one or two flagella was increased about 2-fold (P < 0.05) for the ΔfliO fliA(+36T→C) and ΔfliO clpP(V20F) double-mutant strains compared to the fliO deletion mutant strain. However, the ΔfliO fliP(R143H) double-mutant strain did not produce significantly more flagella than the fliO deletion mutant strain. The ΔfliO fliP(F190L) double-mutant strain produced increased numbers of flagella per cell relative to the fliO deletion mutant strain, and flagella were about 18% as abundant as the case for the wild type. The effects of the suppressor mutations were additive: flagella of the ΔfliO fliP(R143H, F190L) triple-mutant strain were 31% as abundant as those of the wild type, and flagella of the ΔfliO fliA(+36T→C) fliP(R143H, F190L) quadruple-mutant strain were 58% as abundant as those of the wild type.

FIG 3.

Biosynthesis of flagella is rescued by extragenic suppressor mutations in strains bearing a fliO deletion mutation. Negatively stained transmission electron micrograph images are shown for the following strains: SJW1103 (wild type) (A), CB187 (ΔfliO) (B), CB443 [ΔfliO clpP(V20F)] (C), CB502 [ΔfliO fliA(+36T→C)] (D), CB281 [ΔfliO fliP(R143H)] (E), CB282 [ΔfliO fliP(F190L)] (F), CB597 [ΔfliO fliP(R143H, F190L)] (G), and CB598 [ΔfliO fliP(R143H, F190L) fliA(+36T→C)] (H). Representative cells are shown for each strain. Lengths of flagella varied from cell to cell. Bars, 1 μm. Histograms to the right of each image display the number of cells (y axis) producing the indicated numbers of flagella per cell (x axis). The means and standard errors of the numbers of flagella per cell are shown. N, number of cells inspected for each strain. Asterisks indicate cases where the numbers of flagella were significantly (P < 0.05) higher for strains bearing a fliO deletion mutation with suppressor mutations than for the mutant strain with the fliO deletion mutation only. The data were analyzed by the Mann-Whitney U test.

FliP R143H and F190L mutants were made normally.

It was important to know whether the fliP(R143H) or fliP(F190L) suppressor mutation affected FliP synthesis. To investigate FliP synthesis, an antiserum containing polyclonal antibodies against FliP was made. Strains harboring plasmids carrying either fliP and fliO, fliP only, fliP(R143H), or fliP(F190L) were grown to late exponential phase and harvested. Immunoblot analysis of lysed cells with the FliP antiserum determined that similar amounts of wild-type FliP and FliP R143H and F190L mutant proteins were made and that the expression level of FliP was unaffected by FliO (see Fig. S1 in the supplemental material).

The fliA(+36T→C) synonymous mutation increased expression and translation of fliAZY mRNA.

The fliA gene is part of the fliAZY operon, which is under expression control of class 2 and class 3 promoters. The P1 mRNA is for σ²⁸ and FliZ translation, and the P2 mRNA is for FliZ translation (5). To understand the mechanism of suppression by the fliA(+36T→C) allele, it was important to determine whether the synonymous mutation increased expression of the fliAZY operon and/or translation of σ²⁸ from either P1 or P2 mRNA.

Transcriptional fusions of the fliAZY operon promoter region to the lacZ gene, carried by a single-copy promoter probe vector, have been studied previously (4). In this study, similar transcriptional fusions were made for the fliAZY operon promoter region with or without the fliA(+36T→C) mutation. β-Galactosidase activity was measured in various genetic backgrounds (Table 4). A 16% increase (P < 0.05) in mean enzyme activity was measured within the fliO deletion mutant strain for the transcriptional fusion with the fliA(+36T→C) mutation compared to the transcriptional fusion to the native fliAZY operon promoter region. Also, the activity of operon fusions with or without the synonymous mutation was increased 1.5- to 2-fold (P < 0.05) within the ΔfliO clpP(V20F) or ΔfliO Δ(clpP-clpX) double-mutant strain compared to the activity within the fliO deletion mutant strain. These results suggest that the fliA(+36T→C) mutation produced a small increase in mRNA stability or fliAZY expression, and for cells bearing a fliO deletion mutation, fliAZY expression was increased by inactivation of the ClpXP protease.

TABLE 4.

Activity of fliAZY operon and fliA gene fusions to the lacZ gene

Fusion and strain	LacZ activity (Miller units) of the strain harboring the indicated plasmida		Statistical significance (P value)b
Transcriptional fusion of the fliAZY operon promoter region to the lacZ gene	pCB572, native	pCB573, +36T→C mutation
Wild type	350 ± 30	410 ± 80	0.01
ΔfliO mutant	395 ± 90	460 ± 120	0.03
ΔfliO clpP(V20F) mutant	770 ± 140	940 ± 250	0.01
ΔfliO Δ(clpP-clpX) mutant	600 ± 130	720 ± 110	0.01
Δ(cheW-flhD) mutant	0 ± 0	0 ± 0	NS
Translational fusion of the fliA gene to the lacZ gene controlled by promoter 1	pCB576, native	pCB577, +36T→C mutation
Wild type	1,150 ± 90	2,200 ± 120	<0.01
ΔfliO mutant	1,320 ± 250	3,160 ± 90	<0.01
Δ(cheW-flhD) mutant	3 ± 0	4 ± 0	<0.01
Translational fusion of the fliA gene to the lacZ gene controlled by promoter 2	pCB578, native	pCB579, +36T→C mutation
Wild type	31 ± 1	51 ± 6	<0.01
ΔfliO mutant	31 ± 1	37 ± 1	<0.01
Δ(cheW-flhD) mutant	23 ± 0	24 ± 5	NS

^aCells were grown to exponential phase (OD₆₀₀ of 0.65) in LB medium with the appropriate antibiotic at 37°C. Measurements were made in duplicate for at least three independent cultures of each strain. Means and standard deviations are shown.

^bStatistical significance indicates the level of significance for differences determined by Student's t test for the enzyme activity for a fusion plasmid with the native fliAZY operon promoter region compared to the corresponding fusion plasmid with the fliAZY operon promoter region bearing the fliA(+36T→C) mutation. NS, not significant.

Translational fusions of the fliA gene to the lacZ gene under expression control of either the P1 or P2 promoter in a multicopy promoter probe vector were studied previously (5). We made similar translational fusions with and without the fliA(+36T→C) synonymous mutation and measured their activities in various strains (Table 4). The activity of β-galactosidase for a fliA′-lacZ translational fusion expressed from the P1 promoter was high and increased about 2-fold (P < 0.05) with the synonymous mutation. This was observed when the fusions were expressed within the wild-type strain or within the fliO deletion mutant strain. Activity from the P2 promoter remained low, as expected. The P2 promoter controls translation of FliZ but not σ²⁸. However, β-galactosidase activity for a fliA′-lacZ translational fusion expressed from the P2 promoter was increased from 20 to 60% (P < 0.05) by the synonymous mutation. These results suggest that the most important effect of the fliA(+36T→C) mutation was to increase σ²⁸ translation from fliAZY mRNA transcribed from the P1 promoter.

Expression of class 3 flagellar genes was rescued by the clpP(V20F) or fliA(+36T→C) mutation for strains bearing a fliO gene deletion.

ClpXP and σ²⁸ are regulators of flagellar gene expression. In order to understand the consequences of the clpP(V20F) and fliA(+36T→C) suppressor mutations, we examined their effects on flagellar gene expression. We analyzed synthesis of σ²⁸ by immunoblotting lysates prepared from the wild-type strain, the fliO deletion mutant strain, the ΔfliO clpP(V20F) double-mutant strain, and the ΔfliO fliA(+36T→C) double-mutant strain (Fig. 4A and B). Surprisingly, the amounts of σ²⁸ for the fliO deletion mutant strain and the ΔfliO clpP(V20F) double-mutant strain were not significantly different. However, σ²⁸ was synthesized about 1.8-fold more (P < 0.05) by the ΔfliO fliA(+36T→C) double-mutant strain than by the other strains.

FIG 4.

Expression of class 3 flagellar genes is rescued by clpP(V20F) or fliA(+36T→C) mutation in strains bearing a fliO deletion mutation. (A) Synthesis of σ²⁸ (27 kDa) was analyzed by immunoblotting using polyclonal antibodies for σ²⁸. The following strains grown to mid-exponential phase (OD₆₀₀ of 0.65) were included: SJW1103 (wild type), SJW1368 [Δ(cheW-flhD)], CB187 (ΔfliO), CB443 [ΔfliO clpP(V20F)], and CB502 [ΔfliO fliA(+36T→C)]. A typical immunoblot is shown. (B) The amount of σ²⁸ relative to the wild-type level was determined by densitometry analysis of immunoblots for three cultures of each strain. Error bars represent standard deviations. The asterisk indicates that the amount of σ²⁸ was significantly (P < 0.05) increased for the fliO fliA(+36T→C) double-mutant strain compared to the fliO deletion mutant strain. (C) Comparison of mRNA levels of flagellar genes by real-time quantitative RT-PCR. For each strain, three cultures were grown to mid-exponential phase (OD₆₀₀ of 0.65) for total RNA purification. The comparative threshold cycle (C_T) method for relative quantification was used to analyze the results. Error bars represent relative quantification minimum and maximum levels (range of 95% confidence). Asterisks above bars indicate that the mRNA level was significantly reduced (P < 0.05) for the fliO deletion mutant strain compared to the wild type or that the mRNA level was significantly increased for the ΔfliO clpP(V20F) or ΔfliO fliA(+36T→C) double-mutant strain compared to the fliO deletion mutant strain.

mRNA levels for genes of the flagellar regulon were profiled for the four strains by real-time qRT-PCR (Fig. 4C). The flhC gene (of the flhDC operon) is under expression control of a class 1 promoter. The fliA and fliZ genes and the flgM gene (of the flgAMN operon) are under expression control of class 2 and class 3 promoters. The motAB genes (of the motAB-cheAW operon), the cheY gene (of the tar-cheRBYZ operon), and the fliC gene are under expression control of class 3 promoters. Expression of the flhC, fliA, fliZ, and flgM genes was similar for the four strains. Compared to the wild-type levels, mRNA levels of the cheY, motA, and fliC genes were significantly (P < 0.05) reduced for the fliO deletion mutant strain. Levels of mRNA for these class 3 genes were significantly increased for the ΔfliO clpP(V20F) and ΔfliO fliA(+36T→C) double-mutant strains in comparison to the fliO deletion mutant strain, although not compared to the wild type.

DISCUSSION

The FliO protein is required for efficient biosynthesis of the S. enterica flagellum. A fliO deletion mutant strain synthesizes flagella with a low probability (about 3% of the wild-type level). Previously, we showed that extragenic mutations in the fliP gene [fliP(R143H) or fliP(F190L)] partially bypass a loss of fliO function (15). Here we identified new mutations in regulators of flagellar genes [clpP(V20F) and fliA(+36T→C)] that also partially bypass a loss of fliO function. The suppressor mutations acted cooperatively with each other to promote flagellar assembly and motility. The ΔfliO fliP(R143H, F190L) fliA(+36T→C) quadruple-mutant strain assembled about 60% of wild-type flagellar numbers and was nearly fully motile.

After processing of the N-terminal domain, the clpP(V20F) allele produces a mature ClpP V6F mutant protein. Fortuitously, crystal structures of E. coli ClpP and a ClpP V6A mutant have been solved (36). The N-terminal domain of ClpP forms a β-hairpin conformation that was disrupted in the V6A mutant and prevented interaction with ATPase. Since the amino acid sequences of the mature ClpP proteins from E. coli and S. enterica can be aligned with 99% identity over their 193-amino-acid length, this suggests that the structure of the N-terminal domain of the ClpP V6F mutant protein is also disrupted. Deletion of the clpP-clpX genes or the clpX gene alone bypasses fliO function, which confirms that the clpP(V20F) allele is a loss-of-function mutation.

Our analysis of the β-galactosidase activity of fliAZY-lacZ operon fusions suggests that the fliA(+36T→C) synonymous mutation has a small effect on either gene expression or mRNA stability. However, analysis of the β-galactosidase activity for fliA′-lacZ gene fusions showed that the fliA(+36T→C) mutation increases mRNA translation 2-fold and that σ²⁸ levels correspondingly increase. Recently, Chevance et al. (38) showed the significance of synonymous mutations for in vivo translation speed and observed that the effects of synonymous mutations are not neutral. Surprisingly, the fliA(+36T→C) mutation even increased the motility of the strain that bore this mutation alone compared to that of the wild-type strain.

These results suggest that an overall increase in expression of flagellar genes can partially bypass the function of fliO and rescue flagellar assembly and function. It has been predicted that inactivating the ClpXP protease should increase the expression of class 2 and class 3 flagellar genes by reducing the proteolysis of FlhD₄FlhC₂ (35). A plasmid carrying the flhDC genes partly recovered motility of a fliO deletion mutant strain. In addition, we observed increased synthesis and export of hook protein for the fliO deletion mutant strain harboring a plasmid carrying the flhDC genes. This suggests that an increased production of hook-basal body complexes accompanied the partially recovered motility, as observed with increased flhDC expression (39). Motility and biosynthesis of flagella of a fliO deletion mutant strain were also rescued by increased synthesis of σ²⁸ from a plasmid carrying the fliA gene or by deletion of the flgM gene, encoding FlgM, the negative regulator of σ²⁸. On the other hand, motility of a fliO deletion mutant strain could not be complemented by a plasmid carrying the fliS and fliC genes, despite increased export of flagellin. The fliA(+36T→C) and clpP(V20F) mutations rescued expression of class 3 flagellar genes and chemotaxis genes, including fliC, motAB, and cheY. This suggests not only that an increased supply of flagellin is required to bypass a loss of fliO function but also that a general increase in expression of class 3 genes is required. Thus, without fliO, motility and flagellar biosynthesis can be increased by the synthesis of components of the flagellum made after the hook-basal body complex is completed.

Previously, other studies showed suppression of a defect in flagellar biosynthesis from mutations in various flagellar genes by increased levels of FlhD₄FlhC₂. Aldridge et al. (40) isolated suppressor mutant strains produced by an flgN mutant strain that bore loss-of-function mutations in the clpX and fliT genes. Also, Erhardt and Hughes (39) found that mutations that increased expression of the flhDC operon bypassed the requirement for the cytoplasmic C-ring of the flagellum. One category of mutation was the unstable duplication of the region of the chromosome carrying the flhDC operon, which increased expression of the flhDC operon. Such chromosomal duplications would not be detected in our analysis of single nucleotide polymorphisms between the genome sequences of the suppressor mutant strains. No mutations were identified for the suppressor mutant strains CB228, CB229, and CB230 isolated in this study, and the rescued motility was transient. Therefore, chromosomal duplications may have bypassed fliO function in these strains.

The fliP(F190L) mutation was the most effective suppressor mutation: it rescued the number of flagella per cell as well as the proportion of cells producing flagella. Moreover, the ΔfliO fliP(R143H, F190L) triple-mutant strain produced about one-third of the wild-type number of flagella. This suggests that the function of FliO is most associated with regulation of FliP. Polyclonal antibodies to FliP were produced in this study so that expression of native (untagged) FliP could be measured. The fliP(R143H) or fliP(F190L) mutation did not affect the level of expression of FliP, and the expression of FliP was not affected by FliO synthesis. This suggests that our previous conclusion that FliO regulates FliP stability is incorrect (15). The function of FliO therefore remains a matter of speculation. The fliP(R143H) or fliP(F190L) mutation does not produce amino acid substitutions in FliP that correspond to the amino acid residues present in FliP homologs found in the flagellum of A. aeolicus or in the injectisomes (15). One possibility for FliO function is that it acts as a coupling protein, promoting the association of FliP with other components of the export apparatus and basal body during assembly. FliO was among the flagellar proteins required for subcellular localization of an FlhA-cyan fluorescent protein (CFP) fusion protein, suggesting a role for FliO in basal body assembly (41). This may explain why increased expression of flagellar proteins can partially bypass a loss of fliO function by increasing the probability of protein-protein interactions. Future studies should address whether FliO and FliP interact directly, whether FliO is found within the basal body pore, and also which proteins bind to the cytoplasmic domain of FliO.