All strains used in this study are listed in Table 1. DH5α was used in the production of proteins needed for AI-2 biosynthesis and cloning 21 . V. harveyi BB170 was used in the bioluminescence bioassay as a reporter strain 22 . E. coli strains were routinely grown in Luria-Bertani (LB) (Bacto) broth or on agar plates at 37°C. V. harveyi was grown in LB or AB medium 23 at 30°C, also under normal atmospheric conditions. H. pylori strains were routinely grown and maintained on Columbia blood agar plates (No.2, with 5% [v/v] horse blood; Oxoid) or grown in Brucella broth (BB) (Bacto) containing 7% (v/v) fetal bovine serum (Gibco). H. pylori J99 was incubated at 37 °C for 24 h to 72 h as required in a MG500 VAIN-cabinet (Don Whitley Scientific) in an atmosphere of 5% CO₂, 86% N₂, and 6% O₂ (all v/v). For motility experiments the method of Wand et al. 24 was used to achieve motile cultures for analysis, see below. Antibiotics were used at the following concentrations: ampicillin at 100 μg/ml, kanamycin at 30 μg/ml.

Preparation of plasmid DNA, DNA ligation, gel electrophoresis and transformation of E. coli strains were performed in accordance with standard methods 25 . All PCRs were performed with Taq DNA polymerase (Roche Diagnostics, Lewes, UK). TA cloning was carried out using the pGEM-T vector system (Promega, Madison, WI). Plasmid DNA was extracted using the QIAquick spin miniprep kit (QIAGEN, UK). DNA fragments were purified from agarose gel using a QIAquick gel extraction kit (QIAquick, UK) according to the manufacturer's instruction. H. pylori genomic DNA was isolated as described previously 26 . DNA sequencing was conducted using standard fluorescent dye terminator chemistries, and analysis performed using the Applied Biosystems 3730 DNA Analyzer system (Geneservice, Cambridge, UK, Applied Biosystems Inc, Foster City, CA.). Results were analysed using the Bioedit software suite 27 .

H. pylori J99 wild-type was transformed with the plasmid pGEMTluxSXN396 containing a km-sacB construct encoding kanamycin resistance (Km^r) and (5%) sucrose sensitivity (Suc^s) 17 . Disruption of the chromosomal luxS gene was accomplished by natural transformation, allelic exchange, and screening for kanamycin-resistance as previously described 15 , resulting in the J99 ΔluxS mutant strain. The presence of the km-sacB cassette was verified by amplifying fragments of H. pylori chromosomal DNA using primers luxS-F/luxS-R (forward, 5'>GTG GCT TTA GCG GGA TGT TTT<3'; reverse, 5'>GCGA ACA AAT CCC CGC TG<3') and DNA sequencing. The J99 ΔluxS was then transformed with plasmid pGEMTluxS (encoding wild-type luxS), and transformants in which km-sacB had been replaced with the introduced original luxS locus were selected for sucrose resistance on medium containing 5% sucrose and screened for kanamycin sensitivity. The presence of the original luxS gene was verified by amplifying fragments on H. pylori chromosomal DNA using primers luxS-F/luxS-R and DNA sequencing.

Bacterial broth cultures were started from a blood agar plate culture, diluted to an OD_{600 nm} of 0.05 in fresh BB medium, and grown at 37°C in a VAIN-cabinet with shaking. OD_{600 nm} measurements were taken at the 6 h, 24 h, 48 h and 72 h time points, and at the same time cell suspensions were harvested and filtered through a 0.2 μm pore size filter. The AI-2 activity in cell free supernatants (CFS) was tested as previously described using the V. harveyi reporter strain BB170 9 22 . Briefly, an overnight V. harveyi culture was diluted 1:2500 in fresh AB medium 23 . CFS samples were diluted 1:10 in the AB medium containing BB170 into the 96 well bioluminescence plates to give a final volume of 200 μl and were incubated at 30°C. The bioluminescence and optical density were determined at 30 min intervals for at least 8 h using a luminometer (Anthos Labtech LUCY 1.0). AI-2 activity alterations in bioluminescence were expressed as induction (n-fold) over the negative control.

Plate motility assay of H. pylori was performed in Brucella broth medium (BD Biosciences), supplemented with 7% (v/v) fetal bovine serum (Gibco), 0.35%-0.45% (w/v) agar (No.1, Oxoid) and the indicator, 40 μg/ml triphenyl tetrazolium chloride (Sigma, UK). Inclusion of this indicator made it easier to see the small recombinant colonies. Plates were seeded with 5 μl H. pylori liquid culture (forming a circle with 3 mm diameter) standardised to an OD_{600 nm} of 1.0 and were incubated at 37°C for up to 7 days under the conditions described above. The motility halos were recorded using a digital camera and the area of each halo was measured using a GS-800 Calibrated Densitometer (Biorad).

Motility analysis was also carried out by direct observation under phase-contrast microscopy using a Nikon Eclipse E600 after cells were grown in co-culture conditions as used by Wand et al. 24 . Briefly, co-cultures of H. pylori-human gastric adenocarcinoma (AGS) cells were prepared (described below). After 24 h, 10 μl culture was placed onto a microscope slide and covered with a coverslip and freely-motile H. pylori cells were analysed under the microscope.

The liquid chemically defined media were prepared as previously described 15 28 . 60 ml of sterile chemically defined media were added to 40 ml of molten 1% Oxoid No. 1 agar base to make 0.4% semi-solid chemically defined agar. Cysteine supplemented plates (CSP) were made by adding cysteine to the molten agar, shortly before it set. The final concentration of cysteine was 1.0 mM, which was non-limiting for H. pylori growth. The centre of each plate was seeded with one-day incubated H. pylori cells and was incubated for 5 days under the conditions described above. The motility halos were recorded using a digital camera and the area of each halo was measured using a GS-800 Calibrated Densitometer (Biorad).

AI-2 was synthesised enzymatically as described previously using purified proteins LuxS _{E. coli} and Pfs _{E. coli} 8 . For complementation of the ΔluxS _Hp motility phenotype, soft motility agar plates (0.4% w/v) were made as previously described. Bioluminescence activity of the AI-2 product was quantified using the V. harveyi bioassay and compared to CFS from H. pylori wild-type broth culture standardised to an OD_{600 nm} of 1.0 at the time point in the growth curve that maximal AI-2 activity was measured. 1/400 diluted in vitro synthesised AI-2 product shows the same level of bioluminescence as seen in the H. pylori wild-type CFS in the V. harveyi bioassay. Therefore, in the complementation experiment AI-2 was added to motility plates to a final concentration of 0.25% (v/v). 24 h H. pylori cultures were seeded individually onto the centre of each motility plate and incubated for 5 days. The area of outward migration was recorded with a digital camera and measured using a GS-800 Calibrated Densitometer (Biorad).

All chemicals were obtained from Gibco, UK. AGS cells were grown in nutrient mixture Ham's F-12 supplemented with L-glutamine (200 mM) and fetal bovine serum (Gibco) (10% v/v) in a 37°C incubator containing 5% CO₂. After cells had grown to confluency, a 1 in 5 or 1 in 8 dilution was added to a 75 cm² flask containing fresh media mix and incubated in the same conditions as before to allow cells to re-grow to confluency.

AGS cells were counted using the trypan (0.35% v/v) blue dye method. Cells were seeded at a density of 1 × 10⁵ cells/ml into 6 well plates and grown to 80% confluency. The cell-media mix was removed and replaced with 2 ml fresh F-12 media. Plates were inoculated with 24 h H. pylori liquid cultures standardised to an OD_{600 nm} of 0.1 and incubated for one day in a microaerobic environment. Bacterial cells were then analysed using a phase-contrast Nikon Eclipse E600 microscope and electron microscopy.

H. pylori cells were pre-grown as described above for motility analysis. 15 μl of culture was allowed to settle on a carbon formvar grid (Agar Scientific) for 1 min. The suspension was removed and the grid washed by addition of 15 μl of Phosphate Buffered Saline (PBS) for an additional minute. This was removed and the cells stained with 0.5% Phospho-tungstic acid (PTA) pH 7.0 for 1 min. Grids were examined and pictures taken using a JEOL JEM1010 Transmission Electron Microscope. We quantified changes, rounding to the nearest 5% and quote means ± SD. Essentially, three groups of H. pylori cell samples prepared on different dates were examined. Each group of samples contained wild-type, ΔluxS and ΔluxS ⁺ cells treated and not treated with DPD. For each group, 100 H. pylori cells from each culture sample were examined.

Cysteine from Sigma products was dissolved in distilled water according to the manufacturer's recommendation. Synthetic DPD was purchased from Omm Scientific Inc. DPD (AI-2) activity was quantified with the bioluminescence bioassay and compared to wild-type H. pylori grown to an OD_{600 nm} of 1.0, at which maximal AI-2 activity was obtained. To test for complementation of motility, DPD (at a physiological concentration of 150 μM) and non-limiting cysteine (1.0 mM) were added individually to bacteria-AGS cell co-cultures. DPD was added after 10 h of incubation and once again after 18 h of incubation. Cysteine was added from the beginning of incubation. Bacterial motility and cells were observed and visualized by phase-contrast microscope and EM, respectively. For gene transcription studies, DPD (150 μM) and cysteine (1.0 mM) were also added (in the same way) individually to H. pylori liquid cultures of different genotypes. After 24 h, RNA was extracted and the transcript levels of genes of interest were measured.

H. pylori wild-type, its ΔluxS _Hp mutant, the complemented ΔluxS _Hp ⁺ mutant and controls (H. pylori wild-type 17874 29 , and derived mutants ΔflaA (a kind gift from Paul O'Toole) and ΔflgE 30 ) were grown in Brucella broth at 37°C for up to 24 h, at which point high levels of AI-2 activity were detected. To exclude global differences in protein production between strains, we corrected our loading for numbers of bacteria rather than for total protein levels. To do this, 24 h liquid (Brucella broth) culture of each strain was adjusted to OD_{600 nm} of 1.0. A 500 μl cell sample of each strain was then centrifuged at 5500 rpm for 1 min. Culture supernatants were removed and cell pellets were fully resuspended in 1 ml sterile PBS. 100 μl protein sample was collected. The same volume of 2 × sample buffer was added and boiled for 10 min. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequent immunoblotting were carried out as described previously under standard conditions 25 . The gel contained 10% acrylamide. 4 μl protein stock from each strain sample was loaded into each well of the SDS-PAGE gel. For immunoblotting, proteins were transferred from SDS-PAGE gels to nitrocellulose paper by the methanol Tris-glycine system described by Towbin et al. 31 . To see whether similar amounts of protein were loaded using our methodology, membranes were inspected following Ponceau red staining prior to immunoblotting; protein levels appeared similar on each membrane by inspection. The blots were incubated with rabbit polyclonal antibodies against H. pylori flagellin and hook protein (a generous gift from Paul O'Toole) 32 . Bound antibodies were detected using secondary anti-rabbit IgG alkaline phosphatase conjugate antibody (Sigma, UK). The blots were developed using the BCIP/NBT substrate system (Dako, UK). The quantitative scan of the protein bands was performed using a GS-800 Calibrated Densitometer (Biorad). The reflective density (RD) of each protein band was measured using the Quantity One 4.6.5 software (Biorad).

RNA was isolated from H. pylori cells grown in BB medium for 24 h. Cultures were treated with RNA protection reagent (QIAGEN, UK) and RNA was extracted using RNeasy mini kit (QIAGEN, UK). Contaminating genomic DNA was removed using a DNA free kit (Ambion). Synthesis of cDNA was performed using Ominiscript RT kit (QIAGEN, UK) and random hexamers (Roche, Germany). Quantitation of transcripts of selected genes of interest was accomplished by quantitative reverse transcription-PCRs (qRT-PCRs) using Rotor-gene 3000. Primers utilised in RT-PCRs are listed in Table 2. All RT-PCR reaction mixtures contained 12.5 μl of SYBR Green Mix (QIAGEN, UK), 5 μl of gene specific primers, 2 μl cDNA template (cDNA was diluted 10-fold prior to adding into the RT-PCR reactions) and RNase free water to a final volume of 25 μl. The amplification program was 95°C for 15 min, followed by 35 cycles of 95°C for 15 sec, 56°C for 60 sec, and 72°C for 30 sec. All samples, including the controls (16 S rRNA and no-template), were run in triplicate. Transcript levels of each gene were normalised to the 16 S rRNA in each sample. The relative quantity of transcription of each gene was obtained using Pfaffl's analytical methodology.

Statistical analysis was by Student's t test.

Our previous study has demonstrated that luxS _Hp, mccA _Hp and mccB _Hp genes comprise a reverse transulphuration pathway in H. pylori, which is the sole cysteine biosynthesis pathway 15 . We then wanted to determine whether these mutants in a motile strain of H. pylori, J99, would be useful in differentiating whether H. pylori motility was affected by luxS associated AI-2 production or by cysteine provision. Firstly, we needed to establish whether mutations in mcc _Hp genes in our candidate motile strain J99 changed expression of luxS _Hp and AI-2 biosynthesis. To do this, H. pylori J99 wild-type and derived ΔmccA _Hp, ΔmccB _Hp, and ΔluxS _Hp mutants were grown in Brucella broth containing serum (10% v/v). Once they reached logarithmic growth phase, AI-2 activity in the culture supernatant was measured using the V. harveyi AI-2 bioassay previously described 4 8 . As expected, the wild-type produced AI-2 in a growth dependent manner, with AI-2 accumulating during the late logarithmic phase, and reaching maximal levels in the stationary phase. During stationary phase, AI-2 levels decreased and were almost undetectable by 72 h. Similar data were obtained with ΔmccA _Hp and ΔmccB _Hp mutants, despite the fact that the ΔmccB _Hp mutant grew slightly less well than the other mutants and the wild-type. The ΔluxS _Hp mutant, unlike the wild-type and the other two mutants, yielded almost undetectable levels of bioluminescence at each time point, indicating that the production of AI-2 is luxS _Hp-dependent and that insertion of a kanamycin cassette (aphA3) into mccA _Hp and mccB _Hp did not affect expression of the downstream gene luxS _Hp (Figure. 1A).

To investigate whether motility of H. pylori was affected by cysteine biosynthesis, we first compared the motility of H. pylori wild-type with ΔluxS _Hp, ΔmccA _Hp and ΔmccB _Hp mutants. To do this, a 24 h liquid culture of each strain was spotted onto each quarter of a semi-solid agar plate and incubated for up to 7 days. The resulting motility halo areas were quantified after 3, 5 and 7 days of incubation. Halo areas that surrounded the wild-type, ΔmccA _Hp and ΔmccB _Hp strains kept increasing during continuous incubation, although the ΔmccA _Hp strain was slightly delayed in comparison to the others. After 7 days of culture, the ΔluxS _Hp mutant remained almost non-motile and produced a significantly (p < 0.001) reduced motility halo compared to wild-type, ΔmccA _Hp and ΔmccB _Hp strains in 3 independent repeat experiments (Figure. 1B). After 7 days, the wild-type, ΔmccA _Hp and ΔmccB _Hp mutants produced halos of (mean ± SD) 8.5 ± 0.6 mm, n = 4; 5.6 ± 0.9 mm, n = 4; and 7.8 ± 0.6 mm, n = 4 increases in diameter, respectively, all significantly greater than the ΔluxS _Hp mutant which produced a halo size of 1.1 ± 0.1 mm, n = 4. These results revealed that the reduction in motility was likely a result peculiar to luxS _Hp mutation rather than due to disruption of cysteine biosynthesis.

To rule out the possibility that second site mutations in the ΔluxS _Hp mutant were inhibiting motility, genetic complementation was performed to create the ΔluxS _Hp ⁺ strain (see Materials and Methods). The non-motile ΔflhB mutant was used as a negative control 24 . 24 h cultures of wild-type, ΔluxS _Hp, ΔluxS _Hp ⁺ and ΔflhB _Hp strains grown in Brucella broth were individually spotted onto normal motility plates. After 5 days of incubation, the mean halo diameter of the ΔluxS _Hp ⁺ strain was 6.9 ± 0.2 mm, n = 4, which was slightly larger than that of the wild-type (4.7 ± 0.7 mm, n = 4). The ΔluxS _Hp and ΔflhB _Hp mutants showed non-motile phenotypes (Figure. 2A).

To examine whether AI-2 can influence the motility of H. pylori, we assessed the motility of the wild-type, ΔluxS _Hp and ΔflhB mutants on AI-2 supplemented plates (ASP). The ASP was prepared using 0.4% soft agar containing in vitro synthesised AI-2 (0.25% v/v). The buffer control plate (BCP) was also produced using 0.4% soft agar into which was added the buffer control solution (0.25% v/v) produced in parallel to in vitro AI-2 synthesis (buffer containing no AI-2). After 5 days of incubation, the halo size of the wild-type on ASP increased by 11.2 ± 0.7 mm, n = 4, compared with a 5.4 ± 0.2 mm, n = 4 increase on the non-supplemented plate (compare Figure. 2A or the right panel of Figure. 2B with the left panel of Figure. 2B). Whilst the ΔluxS _Hp mutant was non-motile on the BCP, the halo increased by 4.6 ± 0.4 mm, n = 4 on ASP (Figure 2B). The control strain ΔflhB _Hp mutant remained non-motile on the ASP (Figure. 2B).

Having established an influence on motility for one of the chemicals reliant on LuxS_Hp function (AI-2), we sought to establish whether another (cysteine) would have a similar influence. Our previous studies revealed that exogenous cysteine rescues growth defects of mutants unable to complete cysteine biosynthesis via the RTSP of H. pylori (ΔluxS _Hp, ΔmccA _Hp and ΔmccB _Hp mutants) in chemically defined broth 15 . Chemical complementation of motility was thus performed using chemically defined plates supplemented with 1.0 mM cysteine. Methionine was added to these plates as the sulphur source since all known H. pylori strains are methionine auxotrophs. After 5 days of incubation, wild-type H. pylori and ΔmccA _Hp and ΔmccB _Hp mutants formed motility halos of 4.9 ± 0.3 mm, n = 4; 3.6 ± 0.6 mm, n = 4; and 4.3 ± 0.9 mm, n = 4 increases in diameter, respectively. The ΔluxS _Hp mutant remained non-motile (Figure. 2C).

Taken together, these data indicate that the motility defect of the ΔluxS _Hp mutant was restored either genetically or chemically with AI-2, but not with exogenous cysteine. This suggests that luxS and AI-2 play a role in enhancing bacterial motility, rather than an intact cysteine biosynthesis pathway, implying a likely role of luxS _Hp in signalling.

Motility plates effectively indicate motility phenotypes of the population, but do not give any indication of the structure of the motility organelles (flagella), or the motility pattern of individual cells. To characterise the phenotypes underlying the decreased ability of the ΔluxS _Hp mutant to swarm in soft agar, we examined motility of individual bacterial cells using phase-contrast microscopy and also the flagellar morphology of the cells using electron microscopy. Cells tested included wild-type, ΔluxS _Hp and ΔluxS _Hp ⁺, all grown in the presence and absence of DPD and cysteine. All cells were grown in co-culture with human gastric adenocarcinoma (AGS) cells for 24 h before testing, as previous experiments in our laboratory have shown that this gives highly reproducible results in H. pylori motility experiments.

Phase-contrast microscopy revealed that > 40% of wild-type and ΔluxS _Hp ⁺ cells were motile; whereas less than 2% of ΔluxS _Hp cells were motile. When grown with exogenous DPD, motile cells again made up > 40% of the population for wild-type and ΔluxS _Hp ⁺ cells, but now also made up > 40% of the population for ΔluxS _Hp cells. Cultures of the ΔluxS _Hp grown with exogenous cysteine consistently contained less than 2% motile cells. To exclude the possibility that the restoration of motility of ΔluxS _Hp cells was due to an effect of DPD on AGS cells rather than on H. pylori, we set up a control sample in which the wild-type and ΔluxS _Hp mutant were co-cultured individually with AGS cells that had been treated with DPD overnight. DPD was washed off with the media before co-culturing. As expected, both wild-type and ΔluxS _Hp cells in these control cultures showed very similar motility phenotypes to those co-cultured with normal AGS cells, indicating that DPD is a functional signalling molecule to H. pylori cells rather than it working through affecting eukaryotic cells. Moreover, the approximate speed of motile ΔluxS _Hp cells was visibly lower compared to the wild-type, ΔluxS ⁺ and all cell samples plus DPD.

Electron microscopic images (Figure. 3) showed that all samples tested (wild-type, ΔluxS _Hp and ΔluxS _Hp ⁺, grown in the presence or absence of DPD) produced a flagellar filament of some kind in the majority of bacterial cells, but those of the ΔluxS _Hp strain were consistently short and usually fewer in number. In our experiments, nearly all of the wild-type cells tested had flagella (95% ± 3%, n = 3) and most of these had multiple flagella, which were usually at one pole and typically 3-4 in number (90% ± 3%, n = 3) (Figure. 3A). In contrast, fewer ΔluxS _Hp cells tested had flagella (70% ± 5%, n = 3) and these were typically shorter and also fewer in number (30% ± 5%, n = 3 of cells had only one or two short flagella (Figure. 3B)). The complemented ΔluxS _Hp ⁺ cells were similar to wild-type, with nearly all cells possessing 3-4 normal long flagella at least one pole (95% ± 3%, n = 3) (Figure. 3C). Addition of DPD to ΔluxS _Hp cells also converted them to a wild-type morphology, with the vast majority producing 3-4 wild-type length flagella usually present at a single pole (95% ± 3%, n = 3) (Figure. 3E). Addition of DPD to wild-type cells had little significant effect with nearly all remaining flagellate as before (95% ± 3%, n = 3) although more cells were seen with a flagellum at both poles (Figure. 3D). Addition of DPD to the ΔluxS _Hp ⁺ cells had a similar effect, with more cells with flagella at both poles (Figure. 3F).

The reduced number and length of flagella in ΔluxS _Hp cells observed by electron microscopy could emanate from a number of different changes in the proteome. As previous work had suggested possible involvement of major flagella proteins, we investigated these first by immunoblotting whole cell lysates. Cell lysates were adjusted so that protein from equivalent numbers of bacteria was loaded (see Materials and Methods), and probed with anti-flagellin (FlaA and FlaB) and anti-FlgE (hook protein) antiserum (Figure. 4). In practice, FlaB levels were very similar between all wild-type and mutant strains and were not shown to vary in our subsequent transcription analysis. Our main aim here was to compare ratios of flagella proteins between wild-types and mutants, so we expressed results of other flagella proteins (FlaA and FlgE) relative to FlaB levels within each strain. H. pylori wild-type 17874, and derived mutants (ΔflaA and ΔflgE) were used as positive and negative controls, respectively. In our experiments, four repetitions were included, when the reflective density (RD) of each protein band was measured using Quantity One 4.6.5 software (Biorad).

We found that all strains tested produced FlaB at approximately the same level (Figure. 4). The reflective density of the FlaB bands of the wild-type, ΔluxS _Hp mutant and the complemented ΔluxS _Hp ⁺ mutant were (means ± SD) 0.210 ± 2.0E-03 RD, n = 4; 0.204 ± 5.8E-04 RD, n = 4; and 0.207 ± 5.8E-04 RD, n = 4, respectively. We expressed all other results (FlaA and FlgE) relative to FlaB in each strain. Mutagenesis of LuxS_Hp reduced the expression of FlaA relative to FlaB (from mean 1.60 in the wild-type to 1.23 in the ΔluxS _Hp mutant, p < 0.01), and complementation increased the ratio back to wild-type levels (mean 1.70 in the ΔluxS _Hp ⁺ mutant, p < 0.01 compared with the ΔluxS _Hp mutant). Next, we examined FlgE expression, and a similar trend was found (wild-type FlgE:FlaB ratio mean 0.74; ΔluxS _Hp mutant 0.51; complemented ΔluxS _Hp ⁺ mutant 0.77; p < 0.01 for differences between ΔluxS _Hp mutant and wild-type and complemented stains). These data show that FlaA and FlgE synthesis was reduced relative to FlaB in the ΔluxS _Hp mutant and these changes were restored by genetic complementation.

Previous reports have provided evidence that luxS _Hp-dependent QS may occur to modulate motility via transcriptional regulation of flaA or flhA 20 . We utilised quantitative RT-PCR (qRT-PCR) to screen for alterations in transcription of these and other genes involved in flagellar assembly to extend our understanding of the regulatory mechanisms that might be involved. To exclude an effect of cysteine biosynthesis, exogenous addition of cysteine was also undertaken. The concentration of cysteine was non-limiting to H. pylori growth. 16 S rRNA transcription was used for normalization and ureA served as a non-flagella linked gene control (Figure. 5D).

The flagellar genes tested included several from different regulatory hierarchy positions in flagellar synthesis 33 : class 1 genes flhA (encodes flagellar regulator component), motA and motB (encode flagellar motor proteins); class 2 genes flaB (encodes hook-proximal minor flagellin) and flgE (enodes flagellar hook protein); and class 3 gene flaA (encodes major flagellin). fliI (encodes membrane-associated export ATPase of the flagellar basal body) was also examined (Figure. 5).

For class 1 genes tested, flhA showed a consistent pattern of 1.75 fold reduced transcription (p < 0.001), and both motA and motB showed a consistent pattern of 2 fold (p < 0.001) reduced transcription in the ΔluxS _Hp mutant compared to the wild-type (Figure. 5A). For class 2 genes tested, flgE was 1.5 fold (p < 0.001) down-regulated in the ΔluxS _Hp mutant; while flaB did not exhibit any significant change. flaA was the only class 3 gene tested, which was 3.5 fold (p < 0.001) down-regulated in the ΔluxS _Hp mutant compared to the wild-type (Figure. 5B). Additionally, the transcript of fliI was also significantly (1.5 fold, p < 0.001) decreased in the mutant (Figure. 5C).

The reduced transcription of flhA, motA, motB, flgE, flaA and fliI was restored genetically by the complementation of the mutant with the wild-type luxS _Hp gene. Also, 150 μM DPD was sufficient to restore the transcription of these genes in the ΔluxS _Hp mutant to levels similar to the wild-type (Figure. 5E). Although Figure 5E shows that 50 μM and 150 μM DPD induced almost the same level of bioluminescence as the wild-type, we chose to use 150 μM DPD in the complementation experiment because this concentration was shown to be more reproducible (it has the smaller error bar). In wild-type cells, addition of DPD markedly increased transcription of motA, motB, flaA and flaB, whilst flhA, flgE and fliI only showed a marginal increase. Exogenous addition of cysteine to the ΔluxS _Hp mutant did not significantly increase transcription of any of the genes studied; suggesting that addition of cysteine was not able to restore the transcription of flagellar genes (data not shown). Consistent with the analysis of protein levels, these RT-PCR data indicate that luxS _Hp disruption has a greater effect upon transcription of flaA than of flaB. Taken together, these data suggest that the effect of LuxS in cysteine metabolism does not regulate expression of flagellar genes, and that the effects on flagellar gene transcription are likely through AI-2 production.