The bacterial strains used in this study are listed in Table 1. Plasmids used in this study are listed in Table 2. Escherichia coli strains were grown on Luria Bertani (LB) Miller agar and in LB Miller broth, while A. hydrophila strains were grown on Columbia Blood Agar (Oxoid, Basingstoke, UK) and in LB Miller broth. A. hydrophila strains were incubated at 30°C while E. coli strains were incubated at 37°C. Kanamycin [Km] (50 μg/ml), streptomycin [Sm] (50 μg/ml), ampicillin [Amp] (100 μg/ml), rifampicin [Rif] (50 μg/ml), chloramphenicol [Cm] (50 μg/ml) and gentamycin [Gm] (50 μg/ml) were added when required. Primers used in this study are listed in Supplementary Table S1.

General DNA manipulation methods involving DNA restriction endonuclease, T4 DNA ligase and alkaline phosphatase were used as recommended by the supplier (New England Biolabs) to construct plasmids in this study, including pKNG101 constructs, pET28a constructs, pMAL-c5x constructs, pKAGb-2(-) constructs, pKT25 constructs, pKNT25 constructs, pUT18 constructs and pUT18C constructs. PCR amplification was performed using Platinum Pfx DNA polymerase (Invitrogen) while PCR screening was performed using Taq DNA polymerase (New England Biolabs).

Conjugal transfer of pUT-mini-Tn5Km from E. coli S17-1λpir to A. hydrophila AH-3R was performed using a filter mating technique. Bacterial conjugation was allowed to proceed for 6–8 h at 30°C on sterile nitrocellulose filters (0.45 mm pore size) placed onto a Blood agar plate. Serial dilutions of the mating mix were plated on LB agar supplemented with Rif and Km, the latter added in order to select for mini-Tn5Km. All mutants were then analyzed for the presence of the transposon by Southern hybridisations of PstI chromosomal DNA digestions, as described before (Gryllos et al., 2001). As no PstI restriction sites were present in the transposon, variable size bands larger than the transposon were observed for each mutant. To determine the site of the transposon insertion, chromosomal DNA was isolated and direct genomic sequencing was employed using 24-mer primers were designed to the O and the I end of transposon Tn5 and a 99 cycle polymerase sequencing reaction

In order to facilitate a knockout in the exsA, exsC or lafK genes, a selectable marker was inserted in the middle of the gene. Mutants were created by the insertion of the Tn5 derived kanamycin resistance cartridge (nptII) from pUC4-KIXX (Amersham). This cartridge contains an outward reading promoter that drives the transcription of downstream genes when inserted in the correct orientation. Although the cartridge insertions ensure transcription of downstream genes, the regulation and expression levels of these genes could be altered. For each mutant the 1.4 kb SmaI digested kanamycin resistance cartridge was inserted into a convenient site within the middle of the gene. If a convenient site was not present, one was created by PCR. Constructs containing the mutated genes were ligated into the suicide vector pKNG101 and transferred into Aeromonas by conjugation to allow allelic exchange as described before (Tabei et al., 2009). After conjugation, a double crossover event was selected by picking colonies that were streptomycin sensitive and kanamycin resistant. The potential mutants were checked by PCR screening using the appropriate amplification primers followed by sequencing (provided by Core Genomic Facility, University of Sheffield).

Isothermal assembly was used to assemble multiple DNA fragments into one plasmid vector in a single reaction (Gibson et al., 2009). ISA required three DNA fragments and one digested plasmid for the assembly, in which exsD was amplified by PCR into two fragments using primers exsD F1 forward/reverse and exsD F2 forward/reverse (Table 3), each containing adapter sequence overlapping with the plasmid or the kanamycin cassette. The kanamycin cassette was amplified by PCR from plasmid pJMK30 using Pfx DNA polymerase and primers Kan forward/reverse (Table 3), while the plasmid pGEM-3Zf(+) was digested with the restriction enzyme HincII. Each fragment assembled in this reaction should have an approximately 30 bp adapter sequence overlapping each other. Then the fragments, which were equi-molar when added, were mixed together with ISA Buffer (Gibson et al., 2009) and three enzymes: T5 exonuclease, Phusion polymerase and Taq ligase. This mixture was incubated at 50°C overnight and transformed into E. coli DH5α to select for ampicillin and kanamycin resistance, generating pGEMexsDKm^R. The pGEMexsEKm^R plasmid construct was assembled using the same method. Both constructs were checked by nucleotide sequencing. The exsD::Km^R fragment was then amplified from the pGEMexsDKm^R by PCR using Pfx DNA polymerase with exsD_pGEM forward and exsD_pGEM reverse primers (Table 3). A similar PCR amplification was carried out to obtain the exsE::Km^R fragment (exsE_pGEM forward and exsE_pGEM reverse primers) on the plasmid construct pGEMexsEKm^R. Both PCR fragments of exsD::Km^R and exsE::Km^R had blunt ends and were ligated into SmaI-digested suicide plasmid pKNG101. After ligation, the samples were transformed into competent cells of E. coli CC118-λpir to select for kanamycin and streptomycin resistant colonies, containing pKNGexsDKm^R or pKNGexsEKm^R constructs. Once the suicide plasmid constructs were prepared, they were transformed into the E. coli donor cell S17-1-λpir and then conjugated into A. hydrophila AH3R to allow allelic exchange. A double crossover event was selected by picking up colonies that were streptomycin sensitive and kanamycin resistant. The chromosomal DNA of these potential exsD or exsE mutants were then extracted and used as templates for PCR screening using Taq DNA polymerase with exsD F1 forward and exsD F2 reverse primers, followed by sequencing to confirm the knockout. The same PCR screening and sequencing were carried out on potential exsE mutant using exsE F1 forward and exsE F2 reverse primers.

Swarming motility of A. hydrophila strains were assessed on swarming agar plates (0.5% NaCl, 0.6% Difco Nutrient Broth, and 0.6% Eiken agar). A 1ml volume of the bacterial strains that were pre-cultured in broth media were centrifuged at 15,000 × g for 1min. A sterile toothpick was used to transfer a small amount (less than a colony size) of the pellet into the center of swarming agar plates. The plates were incubated face-up at 30°C overnight. Swarming motility was measured by examining the migration of bacteria across the agar (swarming diameters) from the center toward the periphery of the plate (Kirov et al., 2004). Bacterial cells required for lateral flagella transcriptional fusion assays were grown at 30°C overnight and the edge of the bacterial swarm was resuspended in sterile PBS for the β-galactosidase assay.

The promoter of interest was fused to the promoter-less lacZ on the reporter plasmids pKAGb-2(-), followed by conjugation into A. hydrophila mutant and wild type strains as described elsewhere (Tabei et al., 2009). Activity of the promoters were measured as a function of β-galactosidase activity. A. hydrophila cultures were grown in triplicate to an OD600 of 0.5 – 0.8 and were then chilled on ice for 15 min. Duplicate assays were performed at 30°C on 200 μl of cells for each culture in a total volume of 1ml following permeabilization of the cells with chloroform sodium-dodecyl sulfate (Miller, 1972). Values are presented in Miller units (MU).

The E. coli protein expression strain (BL21Star or NEB ER2523 Express) containing the plasmid of interest (pET28a or pMAL-c5x constructs respectively) was incubated overnight in 10 ml LB broth with appropriate antibiotics and 1% (w/v) filter sterilized glucose. On the next day, 1 ml of the culture was transferred into 1 L of LB broth with appropriate antibiotics and 1% (w/v) filter-sterilized glucose. It was then incubated at 37°C for approximately 2 h until the OD₆₀₀ was around 0.2. At this point, 0.3 mM – 1 mM IPTG was added to induce the protein expression. After 2–3 h incubation, the cells were harvested by centrifugation at 8,000 × g for 15 min at 4°C. The pellet was resuspended in Binding Buffer [10 mM Imidazole, 500 mM NaCl, 5% (v/v) Glycerol and 20 mM Phosphate Buffer] for Histidine-tagged proteins or in Column Buffer (20 mM Tris-HCl pH7.4, 200 mM NaCl, 1 mM EDTA, 1 mM sodium azide and 10 mM β-mercaptoethanol) for MBP-tagged proteins and frozen overnight at -20°C. The pellet was then sonicated using Jencons Vibracell at 20 kHz (20% amplitude) for 8x 20 s with 1 min intervals on ice. The sonicated sample was then centrifuged at 30,000 × g for 30 min at 4°C to separate soluble and insoluble proteins. After sonication, the supernatant that contained soluble proteins was transferred into a fresh universal tube while the pellet was stored at -20°C. Histidine-tagged proteins (pET28a constructs) were purified using 1 ml HisTrap HP Column (GE Healthcare) as recommended by the supplier. While MBP-tagged proteins (pMAL-c5x constructs) were purified using 1ml MBPTrap HP columns (GE Healthcare) according to the pMAL Protein Fusion and Purification System manual (New England BioLabs). Purification to approximately 95% was assessed by SDS-PAGE.

Proteins were separated by SDS-PAGE and transferred onto a nitrocellulose membrane (0.45 μm pore size). The membrane was then blocked in 20 ml of PBS + 5% (w/v) dry skimmed milk powder for 1 h. After blocking, the membrane was washed 10 min in 20 ml PBS Buffer with 0.1% (v/v) Tween20 (PBST). Then 10 μl (1:2000) of primary antibody (anti-penta-His [Qiagen] for His-tagged proteins and anti-MBP [New England BioLabs] for MBP-tagged proteins) was added to 20 ml of PBST + 5% milk, in which the membrane was soaked for 1 h. After primary antibody binding, the membrane was washed three times 10 min in 20 ml PBST Buffer. Then the membrane was probed with 20 ml PBST + 5% milk with 5 μl HRP-conjugated secondary antibody (1:4000) for 1 h. After secondary antibody binding, the membrane was washed three times 10 min in 20 ml PBST Buffer again and was developed by Pierce ECL Western Blotting and Chemidoc XRS+ System (Bio-Rad).

Far-Western blotting was based on western blotting and used to investigate protein–protein interactions. In Far-Western Blots, the target proteins were probed with a putative interacting protein, which had a different tag to the target protein. The interactions between two proteins could then be detected by Western Blot using antibodies against the tag on the putative interacting protein. After the transfer of proteins and blocking, the nitrocellulose membrane was probed with a non-antibody protein with a different fusion tag to the loaded proteins. The membrane was probed for 1 h followed by three times 10 min wash with PBST Buffer. Then the membrane was then incubated with the primary antibody against the tag on the probing protein for 1 h followed by three times 10 min wash with PBST Buffer. This was followed by a conjugated secondary antibody against the primary antibody and following washing the membrane was developed as above.

The BACTH Assay was carried out using the Euromedex BACTH System Kit. The target genes were first cloned into the BACTH plasmids (pKT25, pKNT25, pUT18 and pUT18C) in frame with either T25 or T18 fragments of the cyaA gene at the 5′ or 3′ end of the gene of interest (Table 2), to allow co-expression of fusion proteins. Then one of the T25 derived plasmid constructs was co-transformed with a T18 derived plasmid construct into E. coli BTH101 reporter cells and incubated on MacConkey/maltose agar [4% (w/v) MacConkey agar base (Difco^TM), 1% (w/v) maltose, 0.5mM IPTG and appropriate antibiotics) for 2 nights at 30°C.

When the two proteins of interest interacted with each other, heterodimerization of the fusion proteins allowed the complementation of the T25 and T18 fragments to form a catalytic domain of adenylate cyclase (CyaA), thus cAMP was synthesized. Then the mal operon was activated by cAMP/CAP complex. Therefore the maltose metabolism pathway was switched on in the E. coli BTH101 reporter strain, which allowed the fermentation of maltose and the production of acid that turned the pH indicator in MacConkey agar pink. Therefore positive interaction colonies were pink/red in color whereas negative colonies were white.

In order to isolate mutants involved in swarming motility that were encoded outside the lateral flagella gene cluster, random transposon mutagenesis was performed. Over 2000 mutants generated by this method were qualitatively screened for reduced or increased swarming on swarm plates. Several mutant strains consistently exhibited reduced swarming when compared to the wild type, however, their swimming ability in motility agar remained the same as the wild type. In two of the mutants (A5 and A25) the transposon had inserted into 3′ end of the exsD gene of the aeromonad type III secretion system (T3SS). As the aeromonad T3SS is very similar to that found in P. aeruginosa (Vilches et al., 2004) the aeromonad exsD gene encodes a protein that is the anti-activator of ExsA the major transcriptional regulator of the T3SS.

The regulator ExsA usually works as an activator of T3SS although there is some evidence that it can act as a repressor of other systems such flagella. In order to test whether the mutations in the exsD gene are allowing ExsA to act directly or indirectly as a repressor we created individual mutations in the putative regulator and anti-activator genes of the aeromonad T3SS system, namely exsA, exsD, exsC, and exsE.

Firstly to investigate the role of the genes in the aeromonad T3SS, the promoter regions of the T3SS operons (Vilches et al., 2004, 2009) were isolated and fused to the promoter-less lacZ gene in the transcriptional reporter plasmid pKAGb-2(-). The promoter activities of a selection of the promoter constructs were then investigated in the A. hydrophila wild type and exsA, exsC, exsD and exsE mutant strains. The individual reporter plasmids were introduced into the A. hydrophila strains and the promoter activities measured. The activity of promoter PascN in the A. hydrophila wild type was 808 MU, this was reduced in the exsA mutant, in which the promoter activity was down to 15 MU (Figure 1A). In the exsD mutant, the PascN promoter activity increased to 1551 MU, which was an increase from the activities in both wild type and exsA mutant strains (p < 0.001). The PascN promoter activity was repressed, but not completely shut down, to 444 MU in the A. hydrophila exsC mutant (Figure 1A). While in the A. hydrophila exsE mutant strain, the promoter activity was further increased to 2059 MU, which was a significant increase compared to the activities in the A. hydrophila exsA and exsC mutants (p < 0.001).

The promoter PexsD showed a similar trend of activities as for the PascN promoter. In the A. hydrophila wild type strain PexsD activity was 494 MU and it was significantly reduced to 121 MU when exsA gene was knocked out (p < 0.001) (Figure 1B). In the A. hydrophila exsD mutant, the promoter activity of PexsD increased to 744 MU, which was higher than the promoter activity measured in the wild type strain (p < 0.05). While in the exsC mutant the promoter activity was reduced to 413 MU, which was close to the wild type value but significantly lower than in the exsD mutant strain (p < 0.001). When exsE was absent, the promoter activity of PexsD reached 1109 MU, which was a significant increase compared to the activities in all other backgrounds (p < 0.01) (Figure 1B). The patterns of the promoter activities in the different mutant backgrounds suggest a modulating regulatory cascade similar to that of P. aeruginosa (Diaz et al., 2011).

The promoter PexsA, which was responsible for the expression of the putative T3SS master regulator ExsA in A. hydrophila, the promoter activity was much higher than the other promoters of the T3SS. In the wild type the PexsA promoter activity was 10,066 MU. Unlike the other promoters of the T3SS, all of which were repressed when exsA was knocked out, the promoter activity of PexsA was not reduced but demonstrated a significant increase in activity in the exsA mutant to 15,546 MU (p < 0.001) (Figure 2). More surprisingly, the promoter activity of PexsA decreased to 5,030 MU in the exsD mutant background, which was significantly lower than the promoter activities of both the wild type and the exsA mutant strains (p < 0.001). In the exsC mutant, the promoter activity of PexsA increased to 12,826 MU, which was significantly higher than both the activities in the wild type (p < 0.05) and in the exsD mutant strains (p < 0.001), but not significantly different from the promoter activity in the exsA mutant. The promoter activity of PexsA in the exsE mutant was 7,860 MU, which was significantly lower than both the activities in the exsA mutant (p < 0.001) and in the exsD mutant strains (p < 0.01) (Figure 2).

Due to the unexpected results of the PexsA promoter activities in the different A. hydrophila backgrounds, another approach was required to investigate the regulation of the T3SS. Thereby, E. coli was utilized as a heterologous host to re-constitute T3SS regulation by co-transforming two plasmids into this cell background. The reporter plasmid pKAGb-2(-) containing the aeromonad T3SS promoters fused to lacZ, and the gene encoding ExsA the putative master regulator of the A. hydrophila T3SS, was cloned into the broad host range pBBR1MCS-5 plasmid. Both plasmids were co-transformed into E. coli and assayed for β-galactosidase activity.

Each of the T3SS promoter activities was measured in the presence or absence of exsA in E. coli cells by co-transforming the reporter plasmid with pBBR5exsA or empty pBBR1MCS-5 respectively (Figure 3). All of the T3SS promoters, except for promoter PexsA, had very low activity in E. coli, whether or not exsA was present in-trans. When the promoter activity of PexsA was measured with empty pBBR1MCS-5, the β-galactosidase activity was approximately 2,806 MU, but when measured with pBBR5exsA construct, the promoter activity of PexsA had decreased to approximately 1,962 MU (p < 0.001) (Figure 3). This agreed with our findings of PexsA promoter activities in A. hydrophila that the master regulator ExsA negatively regulates its own promoter.

Extrapolation from other related bacterial T3SS indicate the master regulator of the T3SS in the A. hydrophila is ExsA (Diaz et al., 2011). This protein is able to self-interact for cooperative activation of the T3SS promoters as well as interacting with the anti-activator ExsD, while not interacting with ExsC or ExsE (Diaz et al., 2011). It is not known if a regulatory cascade similar to that of P. aeruginosa occurs in A. hydrophila. Therefore a bacterial two-hybrid approach was utilized to investigate the potential protein–protein interactions between the aeromonad Exs proteins. Each exs gene was inserted into all four bacterial two-hybrid plasmid vectors. All combinations are shown in Table 3. These plasmid combinations were used to co-transform the reporter strain E. coli BTH101. ExsA–ExsA interactions were demonstrated with co-transformation of pKT25-exsA with either pUT18-exsA or pUT18C-exsA, and with pKNT25-exsA with pUT18-exsA or pUT18C-exsA (Figure 4A; Table 3). Each of the exsA-fused BACTH plasmid constructs showed positive interactions when co-transformed with each of the exsD-fused BACTH plasmid constructs (Figure 4B). However, ExsA showed no interaction with ExsC or ExsE (Figures 4C,D; Table 3).

ExsD which had been shown to interact with ExsA, showed no evidence of ExsD–ExsD self-interaction. However, ExsD was shown to interact with ExsC in every combination, while no evidence of ExsD–ExsE interaction was found (Table 3).

ExsC did not interact with ExsA in any combination. Strong ExsC–ExsD interactions were shown in every case, while weak interactions were observed for ExsC–ExsC self-interaction due to weaker colouration than seen with ExsC–ExsD or ExsC–ExsE. Very strong ExsC–ExsE interaction was observed in every combination of exsC-fused BACTH plasmid constructs with exsE-fused BACTH plasmids (Table 3).

ExsE was shown to interact with ExsC only, the exsE-fused BACTH plasmids showed no positive interaction with either ExsA or ExsD. As demonstrated in the ExsC interactions, ExsE–ExsC interactions were very strong in every case. A weak ExsE–ExsE self-interaction was observed (Table 3).

To confirm the interaction results from the bacterial two-hybrid assay, the Exs proteins were further investigated by Far western blot. In order to facilitate this each protein was expressed and purified as both His-tagged and MBP-fused versions.

The purified four His-tagged Exs proteins were loaded on an SDS-PAGE gel (Figure 5A) and blotted together with an MBP-fused ExsC as a positive control. When probed with the MBP5 kit control protein (without any Exs fusion) only the positive control MBP-ExsC fusion protein (and degradation products) was detected, demonstrating that the His-tagged Exs proteins were unable to interact with the MBP5 negative control (Figure 5B).

Only the _HisExsD protein was detected in the Far-Western Blot when probed with MBP-ExsA (Figure 5C) indicating the interaction between ExsA and ExsD. No interaction with ExsC or ExsE was seen. Furthermore, no self-interaction with ExsA was observed (Figure 5C), contrary to the findings of the BACTH assay.

When MBP-ExsD was used to probe the His-tagged Exs proteins, no bands were observed except for the positive control (data not shown). Therefore, the Far-Western Blot was carried out in reverse, using _HisExsD to probe the MBP-fused Exs proteins. Only the MBP-ExsC protein was detected (and the positive control), showing the interaction between ExsD and ExsC (Figure 5D).

When MBP-ExsC was used as a probe, both the _HisExsD and _HisExsE proteins were detected (Figure 5E). There were two bands of different sizes detected for _HisExsD, one was the size of the _HisExsD (∼33 kDa) while the other one was possibly the size of dimerized _HisExsD (∼66 kDa). This demonstrates ExsC–ExsD interaction and ExsC–ExsE interaction. No interaction was seen for ExsA or self-interaction with ExsC (Figure 5E).

The _HisExsC and the _HisExsE proteins were detected when MBP-ExsE was used as a probe (Figure 5F) demonstrating interactions between ExsE and ExsC as well as the ExsE–ExsE self-interaction.

As the results presented here suggest the presence of a T3SS regulatory cascade of Exs proteins in A. hydrophila, further investigation of the effect of the exs mutations on swarming motility in A. hydrophila were carried out. Different strains of A. hydrophila including the exsA mutant, exsC mutant, exsD mutant, exsE mutant, and a lafK mutant were plated on the swarming agar together with the wild type. The wild type strain was able to swarm on the surface of the 0.6% (w/v) semi-solid agar, but when the major regulator of the Aeromonas lateral flagella system was knocked out in the lafK mutant, swarming motility was completely lost. Moreover, the swarming motility was not affected in the exsA mutant or exsC mutant as the mutant strains were able to swarm as much as in wild type strain. However, when the exsD gene or exsE gene were mutated, the swarming motility was reduced.

The phenotypes of the swarming assays were quantified by measuring the swarming diameters of each strain (Figure 6). The average swarming diameter of the A. hydrophila wild type strain was 6.9 cm while it was significantly decreased to 1.1 cm in the lafK mutant strain (p < 0.001). The swarming motility of the exsA and exsC mutant strains was not significantly different from the wild type, with an average swarming diameter of 5.7 and 6.3 cm respectively (p > 0.05). Swarming motility was repressed in exsD and exsE mutant strains when compared to the wild type, and the exsA and exsC mutant strains (p < 0.001). The average swarming diameters of exsD mutant and exsE mutant strains were 2.7 and 2.6 cm respectively, although reduced these were also significantly different from the lafK mutant (p < 0.05) (Figure 6).

In order to investigate the effect of the mutations on the aeromonad lateral flagella system, several of the lateral flagella promoter regions (Wilhelms et al., 2013) were fused to lacZ in the promoter probe transcriptional fusion vector pKAGb-2(-) and introduced into A. hydrophila strains. The activities of the lateral flagella promoters were measured in bacteria grown on swarming agar.

Among all putative promoters, PfliM, PflgB and PlafA were chosen, one from each class of the lateral flagella transcriptional hierarchy (Wilhelms et al., 2013). To investigate the potential cross-talk between the lateral flagella system and the T3SS, promoter activities were measured in different mutant backgrounds.

The PfliM promoter activity in the wild type was 247 MU this decreased to 22 MU in the lafK mutant (p < 0.001). The promoter activity of PfliM was 215 MU in the exsA mutant but decreased to 88 MU in the exsD mutant strain (p < 0.001). Similarly, in the exsC mutant strain, the promoter activity of PfliM was 225 MU and decreased to 143 MU in the exsE mutant strain (p < 0.001). There was no significant difference of the promoter activities among the exsA mutant, the exsC mutant and the wild type strains (p > 0.05). However, the promoter activity in the exsD mutant was significantly higher than in the lafK mutant while it was significantly lower than in the exsE mutant (p < 0.001) (Figure 7A).

The PflgB promoter activity was 970 MU in the wild type and decreased to 20 MU in the lafK mutant (p < 0.001). The promoter activity was 684 MU in the exsA mutant strain, which was not significantly different with the promoter activities in the wild type and the exsC mutant strain (p > 0.05) while it was significantly higher than the promoter activities in the lafK mutant, exsD mutant and exsE mutant strains (p < 0.001). The promoter activity of PflgB was 105 MU in the exsD mutant while it increased to 665 MU in the exsC mutant strain (p < 0.001). PflgB promoter activity was 435 MU in the exsE mutant, which was significantly lower than in the wild type, the exsA mutant and the exsC mutant strains but significantly higher than in the lafK mutant and exsD mutant strains (p < 0.001) (Figure 7B).

The pattern of PlafA promoter activity was similar to PfliM and PflgB. In the wild type, PlafA was 976 MU that decreased to 22 MU in the lafK mutant strain (Figure 7C). PlafA promoter activity in the exsA mutant was 771 MU, which was significantly higher than in the lafK mutant (p < 0.001), the exsD mutant (p < 0.001) and the exsE mutant (p < 0.01). The promoter activity in the exsD mutant was 240 MU, which was approximately fourfold lower than in the wild type strain and threefold lower than in the exsA mutant strain. PlafA activity was 734 MU in the exsC mutant strain, which was not significantly different with the promoter activity in the exsA mutant (p > 0.05). Unlike PfliM and PflgB, the promoter activity of PlafA in the exsE mutant strain was 638 MU, which was not different with the promoter activity measured in the exsC mutant (p > 0.05) but significantly lower than the exsA mutant (p < 0.01). The promoter activity of PlafA in the exsE mutant was also significantly higher than in the exsD mutant (p < 0.001) (Figure 7C).