ravA-ravR and ravS control bacterial virulence, flagellar development and swimming motility

The genomic organization of ravS, ravR and ravA and the putative structures of their products are shown in Fig 1A. Western blotting revealed that RavS is located in the membrane and cytoplasm, and that RavR is a cytoplasmic protein (Fig 1B). Although RavA does not contain a recognisable transmembrane helix, repeated experiments showed that RavA is also located in both the membrane and cytoplasm (Fig 1B), which suggests that the HK has a membrane-anchored peptide or that the protein contains an unrecognised transmembrane region. Semi-quantitative western blotting demonstrated that the estimated ratio of RavA:RavR:RavS is approximately 10:29:17, amounting to approximately 1890, 5510 and 3230 molecules, respectively, in a bacterial cell grown under the tested conditions (Fig 1C and S1 Fig). The estimated concentrations of RavA, RavR and RavS in a cell were 2.05 ± 0.53, 5.99 ± 0.46, and 3.51 ± 0.27 μM, respectively (average bacterial cell volume was defined as 1.53 ± 0.23 μM³). This HK:RR ratio is remarkably higher than the general trend observed that RR molecules are often several dozen times more than the HK [39].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. ravA, ravR and ravS differentially regulate bacterial virulence and swimming.

(a) Genomic organisation of ravARS genes and putative secondary structures of their protein products. Protein secondary structures were predicted by the SMART program. TrM: transmembrane domain; PAS: Per-ARNT-Sim domain; DHp: dimerisation and histidine phosphotransfer domain; CA: catalytic and ATP binding domain; REC: receiver domain. GGDEF and EAL domains are also shown. (b) Subcellular localisation of RavA, RavR and RavS. Western blotting was used to detect the proteins in different subcellular fractions. The known membrane-bound RpfC and cytoplasmic HPPK were detected as controls. (c) The ratio of RavA, RavR and RavS proteins in bacterial cells. Semi-quantitative western blotting was used to estimate the levels of these proteins (n = 3). (b) and (c), each experiment was repeated three times. (d–e) Mutation of ravA and ravR rather than ravS attenuated bacterial virulence significantly. (d) Bacterial virulence against host plant Brassica oleracea cv Zhonggan 11. Strains were inoculated onto plant leaves by scissor cutting. The lesion length was recorded 10 d after inoculation. The negative control was an inoculation of 10 mM MgCl₂. EV: empty vector. (e) Quantification of the lesion length in (d). Average lengths and standard deviations are shown. Asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05, n = 30 inoculation sites). (f) Extracellular polysaccharide (EPS) production of bacterial strains. EPS production was measured as the dry weight of EPS vs. the dry weight of bacterial cells. Asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05, n = 3). (g–h) Swimming motility of bacterial strains. (g) Bacterial strains were inoculated in NYG plates containing 0.15% agar and grown at 28°C for 28 h. (h) Average diameters of the migration zones of (g). Asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05, n = 10).

https://doi.org/10.1371/journal.ppat.1007952.g001

To further investigate the roles of ravA, ravR and ravS in controlling physiological processes, we performed phenotypic profiling of the in-frame deletion mutants of ΔravA, ΔravR and ΔravS. These mutants have phenotypes similar to those of the wild-type (WT) strain in growth in rich NYG medium, secretion of extracellular enzymes and resistance to oxidative stress (S2 Fig). However, these mutants exhibited remarkable but differing phenotypic changes in bacterial virulence and motility. Deletion of ravA and ravR substantially decreased bacterial virulence against the host plant Brassica oleracea cv. Zhonggan 11 (Fig 1D and 1E). Genetic complementation of the two genes, whose expression levels were controlled by their native promoters, fully or partially restored virulence to levels of the WT strain (Fig 1D and 1E). Although deletion of ravS has no impact on bacterial virulence, a genetic complementary strain (ΔravS-ravS) showed significant attenuation in virulence (Fig 1D and 1E), which might be caused by overexpression of ravS in a medium-copy, broad-host vector (pHM2). Accordingly, the production of extracellular polysaccharides (EPS), a major virulence determinate of X. campestris pv. campestris, decreased significantly to 43% and 42% of the levels of the WT strain in the ΔravA and ΔravR mutants, respectively, whereas the EPS levels were similar to the WT strain in their genetic complementary strains (Fig 1F). EPS production remained nearly unchanged in the ravS null mutant but decreased significantly in the ravS complementary strain (31% compared with that of the WT strain, (Fig 1F)). These results indicate that ravA and ravR are positive regulators of bacterial virulence but overexpression of ravS decreased bacterial virulence.

As shown in Fig 1G and 1H, compared with the WT, deletions of ravA and ravR increased swimming motility by approximately two-fold, but deletion of ravS almost completely eradicated the swimming capability of the mutant. Genetic complementation successfully restored the observed deficiencies to levels similar to that of WT strain (Fig 1G and 1H). Since bacterial swimming motility is controlled mainly by flagella, bacterial flagella were observed by transmission electron microscopy (TEM). As measured by the flagellar cell ratio (Fig 2A and 2B), ravA (flagellar cell ratio = 71.6%) and ravR (77.1%) mutant strains had a higher frequency of flagella in cells and significantly longer flagella than observed in the WT strain (32.5%). The average flagellar lengths of ΔravA and ΔravR mutant strains and the WT strain were 4.60, 4.84 and 2.82 μm (Fig 2C), respectively. However, inactivation of ravS resulted in complete loss of flagella (Fig 2A–2C). Genetic complementation of ravA, ravR or ravS in the corresponding mutant significantly suppressed these deficiencies in flagellar morphology.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. ravA, ravR and ravS differentially regulate the development of bacterial flagella.

(a) Morphology of bacterial flagella. Bacterial flagella were observed by transmission electron microscopy after negative staining. A representative image of each strain is shown (n > 30). Upper panel: bacterial population profile; lower panel: morphology of a single bacterium. (b) Ratio of bacteria with flagella. For each strain, cells with flagella were counted for three biological replicates with each containing at least 100 cells. (c) The average flagellar length of bacterial strains. The length was measured using the AutoCAD software. Three biological replicates were taken with each comprising at least 30 measurements. (d) The fliC mRNA level in bacterial strains. The level of fliC mRNA was measured by qRT-PCR. Amplification of cDNA from tmRNA was used as an internal control. The experiment was repeated three times and the result of a representative experiment is shown. In (b–d), standard deviations are provided; asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05).

https://doi.org/10.1371/journal.ppat.1007952.g002

ravA-ravR-ravS is located in the vicinity of gene clusters that encode flagella-associated proteins and may regulate the expression of these genes (S3 Fig). We selected eight genes to examine their transcription levels in various strains and found that fliC (encoding flagellin) is a representative gene that is controlled by ravA-ravR-ravS (S3A Fig). Deletion of ravA and ravR significantly increased the mRNA levels of fliC to 363% and 402% of the levels of the WT strain, respectively (Fig 2D). In contrast, mutation of ravS caused a significant decrease in the fliC transcription level to 5.6% of the level of the WT strain, while genetic complementation of ravS increased fliC mRNA to 253% of the level of the WT strain (Fig 2D). Based on these results, in controlling the development of bacterial flagella and fliC transcription, both ravA and ravR act as negative regulators but ravS is a positive regulator.

In addition, because type III secretion (T3SS) genes (hrp) are critical in virulence of X. campestris, we selected three hrp cluster genes (hrpD, hrpF and hrpG) to measure their expression levels. As shown in S3C Fig, the mRNA levels of hrpF did not have remarkable change in the ravR and ravA mutants. However, the amounts of hrpG and hrpD transcripts were significantly decreased comparing to the WT strain. Therefore, we further analysed the expression level of hrpG because HrpG is an OmpR-family response regulator to control the transcription of an AraC-family transcription factor HrpX, and the latter binds to the plant inducible promoters (PIP) of some hrp gene clusters to modulate the expressions of these T3SS genes [30, 40]. In the ravR and ravA mutants, the hrpG transcripts significantly decreased to the 5.2% and 2.2% levels of the WT strain, while genetic complementation of the two genes restored the hrpG expression to 111.1% and 75.6% levels of the WT strain (S3D Fig). This result reveals that ravR and ravA positively modulate the transcription of hrpG.

ravS is genetically downstream of ravA–ravR in regulating bacterial swimming

Epistatic analysis was performed to dissect the regulatory relationships among ravR and ravA or ravS. Although deletion of ravA or ravR significantly decreased the bacterial virulence and production of EPS (Fig 1), the double mutants (ΔravA-ΔravS or ΔravR-ΔravS) exhibited virulence levels similar to those of the WT strain (Fig 3A and 3B and S4A Fig). Similarly, inactivation of ravS effectively suppressed the increased swimming motility, flagellar cell ratio and flagellar length that were caused by ravA or ravR deletion (Fig 3C–3E, and S4B and S4C Fig). The fliC expression level of the ravA or ravR mutant was increased significantly (Fig 2D). However, in the double mutants, fliC mRNA levels decreased significantly to the level of ΔravS (Fig 3F). Furthermore, two double mutants on ΔravA and ΔravR backgrounds were constructed with the conserved phosphorylation residue (His⁵⁰⁰) of RavS substituted by Ala (ΔravA-ravS^H500A and ΔravR-ravS^H500A). The phenotypic changes in the single mutant ravS^H500A were similar to those observed for the ΔravS mutant, and in the double mutants, analyses revealed that the ravS^H500A point mutation effectively suppressed all the tested phenotypic deficiencies that were caused by the ravA or ravR deletion (Fig 3A–3F and S4A–S4C Fig). Collectively, these genetic results suggest that ravS is downstream of ravR and ravA in regulation. The point mutant ravS^H500A, in which RavS is constitutively dephosphorylated, exhibits reduced swimming capacity, which suggests that a high level of phosphorylated RavS (RavS~P) positively controls bacterial swimming but negatively modulates virulence. Furthermore, the suppressive function of ravS depends completely on the function of the conserved His⁵⁰⁰ residue, which suggests that RavA and RavR may negatively regulate the RavS~P level in the epistasis relationship.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Epistatic analysis revealed that ravS is downstream of ravA-ravR in regulation.

(a–f) Deletion of ravS or the ravS^H500A point mutation effectively suppressed the phenotypic deficiencies of X. campestris pv. campestris caused by ravA or ravR mutations. (a) Bacterial virulence measured by the lesion length on leaves. Bacterial strains were inoculated onto plant leaves of Brassica oleracea cv Zhonggan 11. The lesion length was recorded 10 d after inoculation (n = 30). (b) Production of extracellular polysaccharides (EPS). EPS production was quantified as the ratio of dry weight of EPS vs. the dry weight of bacterial cells (n = 3). (c) Bacterial swimming motility measured by the diameters of swimming zones. Bacterial strains were inoculated in NYG plates containing 0.15% agar and grown at 28°C for 28 h. The average diameters of the migration zones were measured (n = 10). (d) Ratio of bacterial cells with flagella. For each strain, cells with flagella were counted (n = 100). (e) The average flagellar length of bacterial strains (n = 30). (f) fliC mRNA levels in bacterial strains. The level of fliC mRNA was measured by qRT-PCR. Amplification of cDNA from tmRNA was used as an internal control. The experiment was repeated three times and the result of a representative experiment is shown. In (a–f), standard deviations are provided; asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05).

https://doi.org/10.1371/journal.ppat.1007952.g003

The EAL domain of RavR positively regulates RavS-mediated bacterial swimming in an Asp⁴⁹⁶-dependent manner

Intriguingly, the RR RavR is upstream of RavS (HK) in regulation. The protein has an EAL domain and a GGDEF domain that associated with c-di-GMP turnover. Using ³²P-labelled c-di-GMP produced by a tDGC cyclase (S5A Fig) [41], we confirmed that the EAL domain of RavR can form a homodimer in vitro and has phosphodiesterase activity to degrade c-di-GMP into pGpG, which depends on the E-A-L motif of the domain (Fig 4A, S5B Fig and S5C Fig). The GGDEF signature motif of RavR is degenerate as “GSDEM” and did not have diguanylate cyclase activity to synthesize c-di-GMP from GTP (S5B Fig), suggest that RavR does not encode cyclase activity. Phosphorylation of RavR by RavA did not affect its phosphodiesterase activity (S5D Fig and S5E Fig), consistent with the result that inactivation of two HK genes (ravS or ravA) did not remarkably affect the intracellular concentration of c-di-GMP (S5F Fig). Additionally, deletion of the coding sequences of the EAL domain or ravR gene increased the intracellular c-di-GMP to 185.9% and 150.0% of the level of the WT strain, respectively (Fig 4B). These results suggest RavR mainly acts as a phosphodiesterase.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The EAL domain counteracts the regulation of phosphorylation residue Asp⁴⁹⁶ of RavR.

(a) RavR and its EAL domain have phosphodiesterase activities to degrade c-di-GMP. ³²P-labelled c-di-GMP was synthesised by tDGC cyclase in vitro. None or 5 μM purified proteins were added to the PDE reaction mixture with 1 μM ³²P-labelled c-di-GMP, and reactions were incubated at 28°C for 30 min. Samples were analysed on thin-layer chromatography plates. (b) Deletion of coding sequences of the EAL domain and ravR increased intracellular c-di-GMP concentrations. Cellular c-di-GMP was extracted from 5 mL mid-log phase bacterial cells (OD₆₀₀ = 0.8) and quantified by HPLC-MS/MS. (a, b) Each experiment was conducted three times. (c, d) Bacterial virulence assay. (c) Bacterial strains were inoculated onto plant leaves of Brassica oleracea cv Zhonggan 11. (d) The lesion length was recorded 10 d after inoculation (n = 30). (e) Production of extracellular polysaccharides (EPS). Bacterial strains were grown in TGM medium at 28°C for 72 h before EPS quantification, which was calculated as the dry weight of EPS vs. the dry weight of bacterial cells (n = 3). (f, g) Bacterial swimming motility. (f) Bacterial strains were inoculated in NYG plates containing 0.15% agar and grown at 28°C for 28 h. (g) Average diameters of the migration zones were measured (n = 10). (h–j) Morphology of bacterial flagella. (h) Bacterial flagella were observed by TEM. A representative image of each strain is shown. Upper panel: bacterial population profile; lower panel: a single bacterium. (i) Ratio of bacterial cells with flagella. For each strain, cells with flagella were counted (n = 100). (j) The average flagellar length of bacterial strains (n = 30). (k) fliC mRNA level in bacterial strains. The level of fliC mRNA was measured by qRT-PCR. Amplification of cDNA from tmRNA was used as an internal control. The experiment was repeated three times and the result of a representative experiment is shown. Standard deviations are provided; asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05).

https://doi.org/10.1371/journal.ppat.1007952.g004

We further genetically analysed the interaction between the phosphor-receiving domain (REC) and the EAL domain of RavR and their regulatory relationship in controlling ravS. Similar to the null mutant of ravR (ΔravR), point mutation of the conserved Asp⁴⁹⁶ residue within the REC domain of RavR decreased bacterial virulence and EPS production (Fig 4C–4E) but increased bacterial swimming (Fig 4F and 4G), flagellar cell ratio (Fig 4H and 4I), flagellar length (Fig 4J) and the transcriptional level of fliC (Fig 4K), which suggests that phosphorylation residue Asp⁴⁹⁶ of RavR is indispensable for regulation.

Surprisingly, inactivation of RavR phosphodiesterase activity by an in-frame deletion of the coding sequence of its EAL domain (ravR^ΔEAL) caused phenotypic changes that were completely different from those of the ravR null mutant (ΔravR) or ravR^D496A point mutant. The virulence level and EPS production of ravR^ΔEAL were unaffected (Fig 4C–4E) rather than reduced. The swimming motility of the ravR^ΔEAL mutant was reduced significantly (Fig 4F and 4G) rather than showing a dramatic increase that was observed for ΔravR or the ravR^D496A mutant. TEM analysis revealed the absence of flagellum in the bacterial cells of the ravR^ΔEAL mutant; thus, its flagellar cell ratio was zero (Fig 4H–4J). Additionally, the fliC mRNA level of the ravR^ΔEAL mutant was reduced significantly (Fig 4K). We then constructed a double mutant of ravR^ΔEAL(D496A) to investigate the epistatic relationship between the EAL domain and Asp⁴⁹⁶. ravR^ΔEAL(D496A) phenocopied the ravR^D496A mutant displaying a decrease in virulence and EPS production but increases in swimming capability, flagellar length, flagellar cell ratio and fliC transcription level (Fig 4A–4K). This genetic evidence suggests that RavR has two opposite functions in regulating bacterial swimming. Full-length RavR negatively controls swimming and flagellar development in an Asp⁴⁹⁶-dependent manner, whereas its EAL domain positively controls swimming and flagellar development. Additionally, epistatic analysis revealed that the phosphorylation of Asp⁴⁹⁶ potentially downstream of the PDE activity of EAL domain in regulating bacterial swimming. The biochemical relationship will be dissected in the following experiments.

Based on the above results, the genetic code of the conserved His⁵⁰⁰ residue of the downstream HK gene ravS was mutated on the backgrounds of ravR^D496A and ravR^ΔEAL(D496A) to construct triple mutants. As predicted, the pattern of phenotypic changes in these three triple mutants was suppressed by the mutation and similar to that of ΔravS or ravS^H500A (Fig 4A–4K).

Both RavA and RavS phosphorylate RavR

The phosphorylation process and regulation of RavA-RavR-RavS have not been investigated previously. In vitro phosphorylation assays using autoradiography and Phos-tag gel analysis revealed that full-length RavA autophosphorylates its conserved RavA^His164 residue because the RavA^H564A substitution completely inactivated RavA autophosphorylation (S6A Fig, S6B Fig and S6D Fig). The addition of RavR or RavR^ΔEAL into the reaction mixture remarkably decreased the RavA phosphorylation level (RavA~P), although no recognisable signal of phosphorylated RavR (RavR~P or RavR^ΔEAL~P) was detected in autoradiography (S6A Fig, S6B Fig and S6D Fig). Recombinant RavR with a conserved Asp⁴⁹⁶ substitution did not cause a decrease in the RavA~P level (S6A Fig, S6B Fig and S6D Fig), which suggests that RavA transfers the phosphoryl group onto Asp⁴⁹⁶ of RavR and that RavR~P is highly unstable. Unlike isotope-labelling method, Phos-tag gel analysis successfully detected a RavR~P band in the RavA-RavR phosphotransfer reaction under an extremely high ATP concentration (2 mM, S6D Fig, lane 5), providing an evidence to demonstrate that RavA phosphorylates RavR.

For RavS, repeated efforts failed to detect any signal representing the autophosphorylated RavS-containing cytosolic region (RavS^ΔTrM, S6C Fig). However, a truncated RavS that lacks the N-terminal transmembrane helix and the two PAS domains (RavS^ΔN) exhibited strong autokinase activity in the presence of ATP (Fig 5A, lane 1 and S6C Fig). The addition of RavR^ΔEAL to the reaction mixture slightly decreased the level of RavS~P after 30 min of co-incubation (Fig 5A, lane 2, S6E Fig, lane 3). No signals representing RavR~P were observed by the isotope-labelling and Phos-tag gel analysis (Fig 5A, lane 3, S6E Fig and S6F Fig) and substitution of the conserved Asp⁴⁹⁶ of RavR (RavR^ΔEAL(D496A)) eliminated its activity to decrease the RavS~P level. Together, these biochemical analyses demonstrated that RavA and RavS are bona fide HKs. RavA can phosphorylate RavR at residue Asp⁴⁹⁶, and RavR~P is highly unstable.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. c-di-GMP specifically and significantly enhances RavS-RavR phosphotransfer.

(a) c-di-GMP enhances RavS-RavR phosphotransfer in vitro. A total of 5 μM RavS^ΔN was autophosphorylated for 15 min by the addition of 10 μCi [γ-³²P]ATP. A total of 15 μM RavR^ΔEAL or RavR^ΔEAL(D496A) was added and incubated for 30 min at 25°C. A total of 100 μM c-di-GMP was added if necessary. The reactions were terminated with 5× SDS loading buffer before autoradiography. The experiment was repeated three times. (b) Time course analyses of the impact of c-di-GMP on the velocity of RavS-RavR phosphotransfer. The autophosphorylation and phosphotransfer reactions were set as mentioned in (a). Various concentrations of c-di-GMP were added to the reactions and isotopic signals (band intensities) of RavS~P were measured after autoradiography. (c–d) c-di-GMP triggered an increase in RavS-RavR phosphotransfer in a dose-dependent manner. (c) Various concentrations of c-di-GMP (0, 5, 10, 15, 30, 50, 75 and 100 μM) were added to the RavS-RavR phosphotransfer reactions. (d) The phosphorylated level of RavS^ΔN~P was quantified by ImageJ and the EC₅₀ value was estimated by Origin 8.5 software. In (b) and (d), the intensity of each data point is the average from three experimental repeats. (e) c-di-GMP triggered increase in RavS-RavR phosphotransfer is highly specific. Other nucleotides, including GTP, GMP, cGMP and c-di-AMP, did not affect the velocity of the RavS-RavR phosphotransfer. (f) RavS specifically interacts with fl-c-di-GMP. The interaction between RavS^ΔTrM and various nucleotides, including fl-c-di-GMP, fl-cGMP and fl-c-di-AMP, were measured by microscale thermophoresis (MST). A total of 20 nM fluorescein-labelled nucleotides and increasing concentrations of proteins were incubated and measured by MST. The dissociation constant (K_d) was quantified to estimate the binding affinity. Each experiment was repeated three times.

https://doi.org/10.1371/journal.ppat.1007952.g005

c-di-GMP specifically and significantly enhances RavS-RavR phosphotransfer

Since ravS is downstream of ravR during the regulation of bacterial swimming and RavR is a functional phosphodiesterase that degrades c-di-GMP, we reasoned that c-di-GMP modulates the activity of RavS in controlling bacterial swimming. We analysed the impact of c-di-GMP on the enzymatic activities of RavS to test this hypothesis.

In the absence or presence of c-di-GMP (Fig 5A, lane 1 and lane 4), the autokinase activity of RavS^ΔN was unchanged, which suggests that c-di-GMP does not affect RavS^ΔN autophosphorylation. However, when RavR^ΔEAL was added to the RavS-RavR phosphotransfer reaction, the presence of c-di-GMP in the mixture remarkably increased RavS-RavR phosphotransfer because the RavS^ΔN~P level immediately decreased when compared with the level in the absence of c-di-GMP (Fig 5A, lane 5 vs. lane 2, and Fig 5B). The time (T₅₀) that signal intensity of the RavS~P decreased to the half of the initiation stage (before adding RavR^ΔEAL) was estimated: without c-di-GMP, the T₅₀ value of RavS~P is 38.2 min. When 15 μM c-di-GMP is in presence, the T₅₀ value of RavS~P remarkably decreased to 16.9 min. Since some HKs or RRs have phosphatase activity to dephosphorylate HKs, it is possible that c-di-GMP positively regulates RavS or RavR phosphatase activity to decrease the RavS~P level. However, addition of c-di-GMP alone did not affect RavS^ΔN~P levels (Fig 5A, lane 4 vs. lane 1, and Fig 5B) and the addition of an inactive RavR^ΔEAL(D496A) did not have any impact on RavS phosphorylation levels, regardless of the presence or absence of c-di-GMP (Fig 5A, lanes 3 and 6). These results indicate that c-di-GMP enhanced significantly RavS phosphotransferase activity towards RavR. This c-di-GMP-triggered kinetic preference effectively decreases the RavS~P level.

Along with the elevation of c-di-GMP concentrations, the RavS-RavR phosphotransfer, which was measured by the intensity of the RavS~P signal, gradually increased. This suggests that the impact of c-di-GMP on RavS autokinase activity is dose-dependent (Fig 5C). The estimated value of the median effect concentration (EC₅₀) of c-di-GMP was approximately 15.58 μM (Fig 5D). Furthermore, the regulation of c-di-GMP on the RavS-RavR phosphotransfer to decrease the RavS~P level is highly specific because that other nucleotides and derivatives, including GTP, GMP, cGMP and c-di-AMP, did not have any impact on RavS phosphotransferase activity (Fig 5E).

Arg⁶⁵⁶ within the CA region of RavS is essential for c-di-GMP binding

Next, microscale thermophoresis (MST) was used to determine whether c-di-GMP binds directly to RavS. 2′-Fluo-AHC-c-di-GMP (fl-c-di-GMP) was used in the assay since the analogue contains a carboxyfluorescein that has excitation and emission wavelengths of 497 nm and 520 nm, respectively, which can be detected directly by a MST instrument [42]. As shown in Fig 5F, fl-c-di-GMP directly bound a truncated, soluble RavS protein lacking the transmembrane region (RavS^Δ). The dissociation constant was 6.06 ± 0.92 μM, which is in the range of physiological concentrations of c-di-GMP in bacterial cells [22]. As controls, 2′-Fluo-AHC-cGMP (fl-cGMP) or 2′-Fluo-AHC-c-di-AMP (fl-c-di-AMP) bound RavS^ΔTrM with weaker affinity (fl-cGMP, 64.4 ± 9.43 μM) or completely failed to bind the protein (fl-c-di-AMP, Fig 5F). Furthermore, to determine the c-di-GMP binding region on RavS, we expressed and purified various RavS fragments (S7A Fig). MST assays revealed that fl-c-di-GMP bound RavS with the DHp-CA or CA region, whereas this second messenger did not bind the PAS-A, PAS-B and DHp regions of RavS (Fig 6A and 6B). Of note, a lower binding affinity of c-di-GMP with the CA domain than the DHp-CA domain was observed, which suggests that the DHp domain indirectly affects the interaction (Fig 6B). MST analysis also revealed that fl-c-di-GMP did not bind to RavR^ΔEAL, RavA and an unrelated HK VgrS since the K_d values were greater than 80 μM (S7B Fig), which did not have physiological significance. Therefore, these results revealed that c-di-GMP specifically binds to RavS and the CA domain of RavS is the possible interacting region of c-di-GMP.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Arg⁶⁵⁶ in the CA domain of RavS is a key residue for c-di-GMP binding.

(a, b) c-di-GMP binds to the DHp-CA domain of RavS. PAS-A, PAS-B and DHp-CA in (a), and DHp and CA domains of RavS in (b), were purified and the interaction with c-di-GMP was measured by MST. Each assay was repeated three times. Standard deviations are shown. (c) Molecular docking model of the interaction between c-di-GMP and DHp-CA of RavS. c-di-GMP is shown as a ball-and-stick model and DHp-CA is shown in ribbon representation. (d) Schematic view of the predicted docking site between c-di-GMP and DHp-CA. Potential hydrogen bonds are indicated as blue dashed lines. Autodock software was used to predict the interaction. (e) Effect of amino acid substitutions on RavS–c-di-GMP binding. Ten recombinant proteins of the DHp-CA region, each containing a substitution of a putative essential residue involved in c-di-GMP binding, were used in the MST assay to quantify the K_d value of RavS–c-di-GMP binding. fl-c-di-GMP was used in the MST assay. Each experiment was repeated three times. (f) Substitution of Arg⁶⁵⁶ of RavS resulted in the loss of RavS-RavR phosphotransfer upon stimulation by c-di-GMP. A total of 5 μM RavS^ΔN or RavS^ΔN(R656A) was phosphorylated by 10 μCi [γ-³²P]ATP for 15 min in the absence or presence of 15 or 100 μM c-di-GMP, and then 15 μM RavR^ΔEAL was added and incubated for 30 min at 25°C. The reactions were terminated with 5× SDS loading buffer and the products were separated by 12% SDS-PAGE, exposed to a phosphor screen and analysed by Typhoon FLA7000. The experiment was repeated three times.

https://doi.org/10.1371/journal.ppat.1007952.g006

Similar to most HKs, the three-dimensional (3D) structure of RavS has not been elucidated. As an alternative, molecular docking was used to predict the structural interaction of the RavS–c-di-GMP complex. We predicted the DHp-CA (residues E464–R697) structure of RavS by homology modelling with the crystal structures of other HKs, such as CpxA-HDC (PDB ID: 4BIU), DivL (4EW8) and VraS (4GT8), as templates. As shown in Fig 6C, the putative structure of DHp-CA is composed of seven α-helices, four β-strands and random crimps. The active centre includes the helices of α3, α4, α5 and α6 and strands β3 and β4, and is highly similar to other DHp-CA domains of HKs. The 3D structure of c-di-GMP was extracted from the crystal structure of the c-di-GMP-VCA0042 complex (PDB ID: 2RDE) [43]. Autodock 4.0 software was used for molecular docking. The results showed that helices α4 and α7 and strands β3 and β4 of the RavS DHp-CA domain formed a c-di-GMP binding pocket and 10 residues in the vicinity of the pocket potentially interact with c-di-GMP (Fig 6C and 6D). Among these residues, Ser⁶³⁶, Asp⁶³⁸, Asp⁶⁴⁰, Ile⁶⁵³, Glu⁶⁵⁷ and Leu⁶⁶⁶ putatively interact with c-di-GMP via hydrophobic interactions and van der Waals forces, whereas Asp⁶³⁷, Gly⁶³⁹, Leu⁶⁵² and Arg⁶⁵⁶ form hydrogen bonds with c-di-GMP (Fig 6D). In contrast to the other residues, the amino hydrogens of Arg⁶⁵⁶ and the phosphate group oxygens of c-di-GMP formed two hydrogen bonds with bond lengths of 2.3 Å and 2.5 Å, respectively (Fig 6D). These results suggest that c-di-GMP potentially interacts with the CA region of RavS and that Arg⁶⁵⁶ is a key residue in the interaction.

To experimentally verify the above prediction, we expressed and purified 10 recombinant DHp-CA proteins in which each of the predicted c-di-GMP-interacting amino acids were separately substituted to Ala. MST assays revealed that the recombinant proteins DHpCA^D637A, DHp-CA^G639A, DHp-CA^D640A, DHp-CA^I653A, DHp-CA^E657A and DHp-CA^L666A exhibited no significant change in c-di-GMP binding affinity. In contrast, DHp-CA^S636A, DHp-CA^D638A and DHp-CA^L652A exhibited 1–2-fold increases in K_d values when compared with that of the positive control (DHp-CA protein), which suggests that these substitutions disrupt the protein-nucleotide interaction. Notably, the K_d value of the c-di-GMP–DHp-CA^R656A interaction was 62.4 ± 3.19 μM, which showed a 10-fold decrease in binding affinity (Fig 6E). Consistent with this result, an in vitro phosphotransfer assay revealed that c-di-GMP failed to stimulate the phosphotransferase activity of RavS^ΔN(R656A)~P towards RavR^ΔEAL and the RavS^ΔN(R656A)~P level did not change noticeably or decreased slightly in the presence of c-di-GMP (15 or 100 μM, Fig 6F). Thermal shift assay revealed that the DHp-CA^R656A is as stable as DHp-CA protein (S8A Fig), and RavS^ΔN and RavS^ΔN(R656A) have similar autokinase activity (S8B Fig), suggesting that the Arg⁶⁵⁶ substitution did not remarkably affect the structure of the recombinant protein. Collectively, these results demonstrate that Arg⁶⁵⁶ of RavS is a key residue in c-di-GMP binding. Substitution of Arg⁶⁵⁶ leads to dissociation between c-di-GMP and RavS such that the RavS-RavR phosphotransfer is not accelerated. Under these circumstances, the majority of RavS remains phosphorylated.

RavS^R656A substitution causes phenotypic changes that mimic constitutively activated RavS

Based on the above results, we constructed a ravS^R656A point mutant to determine whether the dissociation of c-di-GMP from RavS affects bacterial phenotypes. Western blotting revealed that RavS is stable in this mutant (S8C Fig). As shown in Fig 7A and 7B, the virulence of the ravS^R656A mutant was attenuated substantially and the production of EPS by this mutant had also decreased significantly (Fig 7C). The swimming zone diameter of the ravS^R656A mutant was 2.38 ± 0.10 cm, which was much larger than that of the WT strain (1.16 ± 0.08 cm) (Fig 7D and 7E). Therefore, we measured the parameters of bacterial flagella and found that the ravS^R656A mutant had longer flagella (4.79 ± 1.31 μm, Fig 7F and 7H). The ravS^R656A mutant flagellar cell ratio was two-times the level of the WT strain (Fig 7G) and the fliC mRNA level had increased by 292% when compared with that of the WT strain (Fig 7I). Therefore, the pattern of phenotypic changes of the ravS^R656A point mutant is similar to that of the ravS overexpression strain (ΔravS-ravS; S9A–S6F Figs) but is quite different from that of the ravS deletion or ravS^H500A mutant whose virulence and EPS production were unaffected, whereas the tested parameters of swimming motility were reduced significantly (Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. ravS^R656A suppresses a phenotypic deficiency caused by ravR^ΔEAL in a RavS His⁵⁰⁰-dependent manner.

(a, b) Bacterial virulence. Bacterial strains were inoculated onto plant leaves of Brassica oleracea cv. Zhonggan 11. The lesion length was recorded 10 d after inoculation (n = 30). (c) Production of extracellular polysaccharides (EPS). EPS quantification was calculated as the dry weight of EPS vs. the dry weight of bacterial cells (n = 3). (d, e) Bacterial swimming motility. (d) Bacterial strains were inoculated in NYG plates containing 0.15% agar and grown at 28°C for 28 h. (e) Average diameters of the migration zones were measured (n = 10). (f) Morphology of bacterial flagella. Bacterial flagella were observed by TEM after negative staining. A representative image of each strain is shown. Upper panel: bacterial population profile; lower panel: single bacterium. (g) Ratio of bacterial cells with flagella. For each strain, cells with flagella were counted (n = 100). (h) The average flagellar length of bacterial strains (n = 30). (i) fliC mRNA levels in bacterial strains. The amount of fliC mRNA was measured by qRT-PCR. Amplification of cDNA from tmRNA was used as an internal control. The experiment was repeated three times and the result of a representative is shown. In the figure, standard deviations are shown; asterisk: significant difference, as tested by Student’s t-test (P ≤ 0.05). Three biological replicates were performed.

https://doi.org/10.1371/journal.ppat.1007952.g007

Thus, these results suggest that the ravS^R656A substitution, which eliminated the binding capability of c-di-GMP, mimics the constitutive phosphorylated state of RavS. Subsequently, the conserved His⁵⁰⁰ residue was mutated on the genetic background of ravS^R656A. Compared with the ravS^R656A mutant, the double mutant ravS^R656A-H500A exhibited restored virulence and EPS production levels (Fig 7A–7C). The swimming motility (Fig 7D and 7E), flagellar cell ratio (Fig 7F and 7G), flagellar length (Fig 7H) and fliC mRNA levels (Fig 7I) of the double mutant were all significantly reduced or undetectable. Therefore, point mutation of His⁵⁰⁰ completely suppressed the mutational effects caused by the Arg⁶⁵⁶ point mutation. Together with the biochemical results demonstrating that c-di-GMP did not bind ravS^R656A and that the phosphorylation level of RavS^R656A is insensitive to c-di-GMP inhibition, these findings suggest that c-di-GMP associates with the Arg⁶⁵⁶ residue to negatively regulate the RavS phosphotransferase activity towards RavR.

RavR regulates RavS~P levels through a c-di-GMP-enhanced phosphate sink

The aforementioned genetic and biochemical results prompted the following hypothesis: a high level of RavS~P positively regulates bacterial swimming but is detrimental to virulence; however, dephosphorylation of RavS reduces the expression of fliC and is negative in modulating bacterial swimming. Since c-di-GMP binds to RavS to inhibit the phosphotransferase activity of RavS towards RavR, RavR is likely to be a special phosphate sink that controls the RavS~P level through opposite mechanisms: 1) the REC domain of RavR receives the phosphoryl group from RavS~P, which decreases the RavS~P level and then negatively regulates bacterial swimming. Additionally, c-di-GMP inhibits bacterial swimming by further enhancing the RavS-RavR phosphotransfer; and 2) the EAL domain of RavR degrades c-di-GMP and then inhibits the c-di-GMP-triggered RavS-RavR phosphotransfer. This process prefers to maintain the RavS~P level and positively regulates bacterial swimming. This hypothesis is true when constitutive dissociation by genetic manipulation of the binding event between c-di-GMP and RavS results in a high RavS~P level. As a consequence, this process will positively modulate bacterial swimming independent of the c-di-GMP concentration in cells.

To test this hypothesis, we point mutated Arg⁶⁵⁶ of ravS to Ala to construct a double mutant on the ravR^ΔEAL genetic background (ravR^ΔEALravS^R656A). The RavR phosphodiesterase activity of this mutant was genetically inactivated and recombinant RavS did not bind c-di-GMP because Arg⁶⁵⁶ is essential for the protein-[c-di-GMP] interaction. As shown in Fig 7, the point mutation of ravS^R656A effectively suppressed the phenotypic effects (i.e., virulence and EPS production) caused by the EAL domain deletion (Fig 7A–7C) and bacterial swimming was restored completely (1.48 ± 0.08 cm; Fig 7D and 7E). The number of flagella and flagellar cell ratio were increased significantly when compared with the level of the WT strain (Fig 7F–7H). The fliC mRNA level of the ravR^ΔEALravS^R656A mutant was also significantly increased from 9.4% (ravR^ΔEAL) to 199% of the level of the WT strain (Fig 7I). Similarly, overexpression of ravS on the ravR^ΔEAL ΔravS background (ravR^ΔEALΔravS-OE-ravS; S9A Fig), which mimics constitutive phosphorylation of RavS, also effectively suppressed the mutational effects of the EAL domain deletion on bacterial swimming (S6 Fig).

Furthermore, the suppression of the ravS^R656A mutation and ravR^ΔEAL deletion completely depends on the conserved His⁵⁰⁰ site of RavS. The triple mutant of ravR^ΔEALravS^R656A(H500A) exhibited performance similar to that of ravR^ΔEAL in all of the tested parameters associated with swimming or flagella development (Fig 7A–7I). Collectively, these genetic analyses support the hypothesis regarding the regulation of RavR in controlling the RavS~P level. Asp⁴⁶⁹ within the REC domain of RavR receives the phosphoryl group from RavS to reduce the RavS~P level, whereas the EAL domain of RavR degrades c-di-GMP, which binds to RavS and promotes the specificity between RavS-RavR by broadening the phosphotransfer flux. Dissociation of RavS–c-di-GMP leads to enhanced bacterial swimming, which depends on the high phosphorylation level of RavS.

Bacterial strains, plasmids and culture conditions

All bacterial strains and recombinant vectors used in this work are listed in S1 Table. The WT strain of X. campestris pv. campestris ATCC 33913 and recombinant strains were grown at 28°C in rich NYG medium (5 g L⁻¹ tryptone, 3 g L⁻¹ yeast extract and 20 g L⁻¹ glycerol, pH 7.0). We used 210 medium (5 g L⁻¹ sucrose, 8 g L^–1 casein enzymatic hydrolysate, 4 g L⁻¹ yeast extract, 3 g L⁻¹ K₂HPO4 and 0.3 g L⁻¹ MgSO₄·7H₂O, pH 7.0) to prepare for electro-component cell production and the TGM medium (10 g L⁻¹ tryptone, 5 g L⁻¹ yeast extract, 20 g L⁻¹ glycerol, 10 g L⁻¹ glucose, 0.7 g L^–1 K₂HPO₄, and 0.25 g L⁻¹ MgSO₄.7H₂O) for the EPS production assay. Escherichia coli (E. coli) DH5α cells were used as a host for the construction of all recombinant vectors. The E. coli BL21(DE3) strain was used as a host for expressing recombinant proteins with the pET30a, pET22b or pET28b vectors (Novagen, USA), and the E. coli TB1 strain was used as a host for expressing recombinant proteins with the pMAL-p2X vector (Novagen). Appropriate antibiotics were added to the media when needed at the following concentrations: kanamycin (50 μg ml^–1), spectinomycin (150 μg ml^–1) and ampicillin (100 μg ml^–1). Electro-competent cells were prepared by washing bacterial cells thoroughly three times with ice-cold glycerol (10%). The transformation conditions of bacterial cells were set at 1.8 kV cm^–1, 25 μF and 200 Ω and conducted in a Bio-Rad Pulser XCell (Bio-Rad, USA).

Construction of mutant and genetic complementation

General molecular biology experiments, including PCR, DNA ligation, enzyme restriction and western blotting, followed the protocols from Molecular Cloning unless otherwise mentioned. All in-frame deletion (markerless) mutants were constructed using the suicide vector pK18mobsacB by a homologous, double crossover method. Briefly, the 5′ and 3′ genomic sequences of a targeted region were amplified using the primers listed in S2 Table, and correct PCR products were ligated into pK18mobsacB. Point mutants were constructed from the corresponding deletion mutants using pK18mobsacB with the point mutated genes. The recombinant pK18mobsacB vector was electroporated into X. campestris pv. campestris competent cells to generate single-crossover mutants by selection on NYG plates containing kanamycin. Next, single-crossover mutants were cultured in NYG medium for 1–2 h and then grown on NYG plates containing 10% sucrose to select second-round homologous crossovers. Candidate bacterial mutants were verified by PCR and sequencing. To genetically complement mutants, a full-length gene with a native promoter was amplified using primers listed in S2 Table and ligated into the broad-host vector pHM2. Then, the pHM2 recombinant vector was used to complement the mutated genes. The recombinant medium-copy, broad-host-range vector pBBR1MCS2 that carries full-length sequences of the genes of interest (under the control of the Plac promoter) was used to construct overexpression strains.

Semi-quantitative western blotting

Briefly, bacterial culture grown to OD_{600 nm} = 0.5 and 0.1 ml or 1.0 ml cell cultures were transferred to into two new tubes, A and B, respectively. The tube B were centrifuged for 10 min at 4°C at 5,000 g. Cell pellets in tube B were resuspended in 80 μl ddH₂O and 20 μl 5×sample loading buffer and the mixtures were boiled for 15 min before being subjected to12% (w/v) SDS-PAGE gels for electrophoresis analysis. Proteins were then electrotransferred onto a Fluoro Trans polyvinylidene fluoride membrane (PALL, USA) following the manufacturer's protocol and were detected with immunoblotting by the antiserum against RavS, RavA or RavR and HRP-conjugate goat anti-rabbit IgG secondary antibody (Abmart, China) at a dilution of 1:5,000. The protein bands were visualized by Enhanced ECL Chemiluminescent Substrate Kit (Yeasen, China). The densities of protein bands were determined by Image J. The amounts of individual protein bands were calculated from the standard curves derived from a set of purified and quantified RavS, RavA or RavR protein standards run on the same blot. Protein levels were normalised to bacterial cells, which were counted in a serially diluted bacterial culture from tube A that was grown on NYG plates for 48 hours. To estimate the concentration of a protein in a cell, morphology of 30 bacterial cells were selected and their volumes were calculated as cylinders. Each experiment was performed with three biological replicates.

Virulence assays

Plant inoculation and virulence assays were conducted using six-week-old cabbage cultivar Brassica oleracea cv. Zhonggan 11 as host plants. The WT strain of X. campestris pv. campestris ATCC 33913 and sterile 10 mM MgCl₂ were used as positive and negative controls, respectively. All bacterial strains were cultured overnight in NYG medium containing appropriate antibiotics. Cells were collected and washed with 10 mM MgCl₂ and the concentrations were adjusted to OD₆₀₀ = 0.1 before inoculation using sterile scissors. After inoculation, the plants were kept in a greenhouse at 25–30°C and relative humidity >80%. The lesion length was measured 10–12 d after inoculation. For each strain, at least 30 cutting sites were made to evaluate the virulence level.

EPS production assays

An assay of EPS production was conducted according to a previous study with small modifications [37, 54]. Bacterial strains were cultured at 28°C in NYG medium until the OD₆₀₀ = 0.8. Then, 200 μL subculture was inoculated into 20 mL TGM medium and cultured at 28°C for 72 h before measurement. To determine the production of EPS, the supernatants of 12 mL bacterial cultures were separated by centrifugation at 25,000 g for 30 min. A total of 24 mL ethanol and 1.2 mL saturated KCl were added into the supernatants and the mixtures were kept at 4°C for 4 h. The precipitated EPS was pelleted by centrifugation at 25,000 g for 30 min and washed twice with 12 mL 95% ethanol. The bacterial cells and EPS were dried at 60°C overnight before determination of their dry weights. EPS production was quantified by the ratio of the weight of EPS vs. the dry weight of bacterial cells.

Swimming motility assays

NYG semi-soft agar (0.15% agar) motility plates were used to determine the swimming capability of bacterial strains. Semi-soft plates were prepared and allowed to cool and dry at room temperature for 3 h prior to inoculation. Strains were cultured at 28°C overnight and adjusted to OD₆₀₀ = 0.8. Then, 2.5 μL of the mixture was transferred to the motility plates and incubated at 28°C for 28 h. The plates were photographed and the diameter of the migration zone of bacteria was measured. Assays were performed with at least three biological replicates, each containing 10 repeats. Statistical analysis was performed using the two-tailed Student’s t-test.

Negative staining and transmission electron microscopy (TEM)

Bacterial strains were cultured at 28°C for 36 h. Negative staining and TEM observations were performed. Briefly, bacterial cells were acquired from NYG plates and gently suspended in 40 μL ddH₂O for 15 min. The bacterial turbid liquid was then spotted onto 400 mesh carbon-coated copper grids, which were glow discharged for 15 s immediately prior to use. Samples on the grids were negatively stained twice with 0.15% uranyl acetate before blotting with filter paper. Pictures were taken with a JEM-1400 electron microscope (JEOL, Japan) at an operating voltage of 80 kV.

Flagellated cell ratio statistics and flagellar length measurements

The number of flagellated cells (N_f) and non-flagellar cells (N_nf) in the TEM images were counted and the flagellated cell ratio = N_f/(N_f + N_nf) determined. Assays were performed with three biological replicates and the number of counted cells in each assay was larger than 100. Statistical analysis was performed using the two-tailed Student’s t-test.

To measure flagellar length, the tagged image file (TIF) format pictures of cells were transformed to the AutoCAD Drawing Database (DWG) format and then opened by auto Desk auto CAD 2014 software. Flagella were traced by the PLINE model and the relative lengths were calculated by the PROPERTIES model. The scale segment of each picture was measured in the same way to calculate the flagellar length. Assays were repeated with three biological replicates and the number of measured flagella in each assay was larger than 30.

qRT-PCR

qRT-PCR was used to measure the level of mRNA. Total RNA was extracted by TRIzol (Invitrogen, USA). The DNA contamination in total RNA samples was eliminated by RNAase-free DNase I (Ambion, USA). The first strand of cDNA was generated using random primers (Promega, USA) and Superscript III reverse transcriptase (Invitrogen). qRT-PCR was conducted using Maxima SYBR Green (Fermentas, USA) in a DNA Engine Option 2 System (Bio-Rad), according to the manufacturer’s instructions. Amplification of tmRNA was used as an internal control. Generally, a qRT-PCR experiment was repeated independently three times with three technical repeats of each sample. A representative of all the biological repeats was selected and reported.

Protein expression, purification and polyclonal antibodies

C-terminal, His₆-tagged recombinant proteins were produced by corresponding recombinant pET30a, pET22b or pET28b (Novagen) vectors that were transformed into the E. coli BL21(DE3) strain. His₆-tagged proteins were extracted and purified using affinity chromatography with Ni-NTA agarose beads (Novagen), according to the manufacturer′s instructions. N-terminal MBP-tagged recombinant proteins were produced by the corresponding recombinant pMAL-P2X (Novagen) vector that was transformed into the E. coli TB1 strain. If necessary, the TEV protease cleavage sequence was appended at the N-terminus of the target protein. The MBP fusion protein was purified and the elute buffer was exchanged to the reaction buffer (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 0.5 mM EDTA and 1 mM DTT), then incubated with TEV protease at a ratio of 1:100 for 8 h at 4°C. Cleaved MBP or uncleaved fusion protein was removed by MBP resin and the TEV protease was removed by Ni-NTA agarose beads. Purified target protein was obtained by size exclusion chromatography with a Superdex 75 10/300 GL column (GE Healthcare, Piscataway, NJ, USA). Purified proteins were concentrated using Centricon YM-10 columns (Millipore) and the elute buffer was changed to the storage buffer (50 mM Tris-HCl, pH 8.0, 0.5 mM EDTA, 50 mM NaCl and 5% glycerol) for further use or storage at −80°C. Protein preparations were examined for purity by SDS-PAGE and quantified by a Bradford assay (Bio-Rad). Polyclonal antiserums of RavR, RavS and RavA were prepared by immunising rabbits with approximately 3 mg purified, soluble proteins. The polyclonal antibodies of RpfC and HPPK were reported by our previous studies [32, 55].

Gel filtration chromatography

The purified EAL protein was concentrated by VivaSpin Turbo, 5K MWCO (Sartorius, German). A total of 15 μM protein and 300 μM c-di-GMP are incubated in phosphodiesterase reaction buffer (50 mM Tris-HCl pH 7.5, 250 mM NaCl, 25 mM KCl, 10 mM MgCl₂ and 2 mM DTT) at 4°C for 30 min and subsequently subjected to gel filtration with Fast Protein Liquid Chromatography AKTA Purifier 10 with Frac-900 (GE Healthcare, USA). The ATKA system was pre-equilibrated with phosphodiesterase reaction buffer at a flow rate of 0.5 ml/min and then applied to a Superdex 75 10/300 GL column to separate dimer and monomer. The elution profiles were collected at A₂₈₀ and confirmed by SDS-PAGE and Coomassie brilliant blue staining.

In vitro autophosphorylation and the phosphoryl transfer assay

An in vitro autophosphorylation assay was conducted as described in our previous study [32, 55]. Purified protein was incubated with 10 μM ATP containing 10 μCi [γ-³²P]ATP (PerkinElmer, USA) in the reaction buffer (50 mM Tris-HCl, pH 7.8, 25 mM NaCl, 25 mM KCl, 5 mM MgCl₂, 2 mM DTT) for the indicated time at 28°C. If necessary, c-di-GMP was added to the mixture at the same time as ATP. For the phosphoryl transfer assay, purified RR protein or its derivative was added into the reaction mixture containing the phosphorylated HK. The autophosphorylation or phosphoryl transfer reaction was terminated by adding 5× SDS-PAGE loading buffer. Phosphorylated proteins were separated by 12% SDS-PAGE. After electrophoresis, gels were exposed to a phosphor screen (GE Healthcare) and the autoradiographic signals were detected by a Typhoon FLA7000 (GE Healthcare).

Phos-tag acrylamide gel analysis

Phos-tag acrylamide gels were prepared according to the instructions described by the manufacturer with minor modifications. Phos-tag acrylamide running gels contained 8% or 12% (w/v) 29:1 acrylamide:N, N”-methylene-bis-acrylamide, 375 mM Tris pH 8.8, 0.1% (w/v) SDS. All the gels were copolymerized with 50 μM Phos-tag acrylamide and 100 μM MnCl₂. All stacking gels contained 5% (w/v) 29:1 acrylamide:N, N”-methylene-bis-acrylamide, 125 mM Tris pH 6.8, 0.1% (w/v) SDS. All Phostag acrylamide containing gels were run at 4°C for 3.5 hours under constant voltage (150 V). Coomassie brilliant blue staining was used to detect the phosphorylated proteins.

Microscale thermophoresis assay

The interaction between c-di-GMP nucleotides and various proteins was measured by using MST with a fluorescein-labelled c-di-GMP (2′-Fluo-AHC-c-di-GMP, abbreviated as fl-c-di-GMP; 2′-Fluo-AHC-c-di-AMP, fl-c-di-AMP; 2′-Fluo-AHC-cGMP, fl-cGMP; Biolog, Germany). These chemicals contain carboxyfluorescein that has excitation and emission wavelengths of 497 nm and 520 nm, respectively, which can be detected directly by a MST instrument. The experiments were performed on a Monolith NT.115 device using standard treated capillaries (NanoTemper Technologies, Germany). The concentration of the particular protein varied from 0.036 to 150 μM with a 2-fold gradient and the concentration of fl-c-di-GMP was constant at 20 nM. Fluorescence intensity due to thermophoresis was recorded using the blue channel optics of the instrument (λ_ex = 470 ± 15 nm, λ_em = 520 ± 10 nm) during a 30 s period of infrared laser heating at 80% of the maximum laser power, followed by a 5 s cooling period. Measurements were performed in buffer containing 20 mM HEPES, pH 7.5, 150 mM NaCl, 2 mM MgCl₂, 2 mM DTT and 0.05% Tween 20. The KD Fit of NanoTemper Analysis software (ver. 1.5.41) was used for fitting the curve and calculation of the dissociation constant (K_d). Assays were repeated with at least three biological replicates in triplicate.

Thermal shift assay

Thermal shift assays were performed as previously described with minor modification[32]. Briefly, purified protein was added to the reaction to a final concentration of 5 μM in the MST buffer [20 mM HEPES, pH 7.5, 150 mM NaCl, 2 mM MgCl_2, 2 mM DTT, 0.05% Tween 20 and 1:500 dilution of SYPRO Orange Dye (Invitrogen, USA)]. A melt curve protocol was run on a Bio-Rad qPCR instrument (Bio-Rad, USA). The fluorescence was measured using the ROX reporter with a temperature gradient of 20–95°C in 1.0°C increments at 30 second intervals. Melt curve data were trimmed to three data points after maximum and the data were plotted with Boltzmann model to obtain the temperature midpoint of unfolding (T_mid) of the protein using Prism 5.0 software (GraphPad). Three biological replicates were assayed in triplicate and statistical significance was determined with two-tailed Student’s t-test.

Synthesis and purification of [³²P]c-di-GMP

[³²P]-labelled c-di-GMP was chemically synthesised using a labelled [³²P]GTP (3000 Ci/mmol, PerkinElmer, USA) and the purified His₆-tagged enzyme tDGC; a protein construct with a key residue mutation (R158A) of diguanylate cyclase from Thermotoga maritime [41]. Five micromolar tDGC and 20 μCi [³²P]GTP (mixed with 50 μM cold GTP) were added to a mixture of 20 μL reaction buffer (300 mM NaCl, 50 mM Tris-HCl, pH 7.5, 20 mM MgCl₂, and 2 mM DTT). After 3 h at 45°C, the reaction was terminated by heating at 98°C for 10 min. The precipitated protein was removed by centrifugation at 20,000 g for 5 min. Radioactive [³²P]c-di-GMP was tested by separation on a polyethyleneimine-cellulose plate (1:1.5 (v/v) saturated (NH₄)₂SO₄ and 1.5 M KH₂PO₄, pH 3.6). The mixture contained more than 95% [³²P]c-di-GMP without further purification.

In vivo c-di-GMP quantification

c-di-GMP extraction was performed as described previously with modifications [34]. Briefly, 8 mL bacterial culture grown to OD₆₀₀ = 0.8 was centrifuged at 5000 g and 4°C for 10 min. Cell pellets were resuspended in 2× 800 μL NYG medium and 0.2 and 1.0 mL of the cell resuspension was transferred to two tubes, A and B, respectively. The tubes were centrifuged at 5,000 g and 4°C for 10 min. The cell pellet in tube B was resuspended in 300 μL extraction solution (40% acetonitrile, 40% methanol and 20% water), incubated on ice for 15 min and lysed by a non-contact ultrasonication system (Bioruptor UCD-200, Diagenode, Belgium) for 10 min with an alternating 30 s power on and 30 s power off procedure. Samples were then centrifuged at 20,000 g and 4°C for 10 min and the supernatant was transferred into a 2.0 mL tube. Extraction was repeated twice with 200 μL extraction solution but heating at 95°C was omitted. The combined supernatant fluids of three extractions were completely evaporated. c-di-GMP powders were resuspended in 100 μL HPLC-grade ddH₂O and analysed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on an AB SCIEX QTRAP 4500 system (AB SCIEX, USA). A Synergi Hydro-RP 80A LC column (4 μM, 150 × 2 mm, Phenomenex, Torrance, CA, USA) was used for reversed-phase liquid chromatography. Solvent A was 0.1% acetic acid in 10 mM ammonium acetate and solvent B was 0.1% formic acid in methanol. The gradient used was as follows: time (t) = 0–4 min, 98% solvent A, 2% solvent B; t = 10–15 min, 5% solvent A, 95% solvent B. The injection volume was 5 μL and the flow rate for chromatography was 200 μL/min. The amount of c-di-GMP in samples was calculated with a standard curve generated from pure c-di-GMP (Sigma-Aldrich, St. Louis, MO, USA) suspended in ddH₂O (Biolog, Germany). c-di-GMP levels were normalised to bacterial cells, which were counted in a serially diluted bacterial culture in tube A that was grown on NYG plates for 48 h. Each c-di-GMP quantification experiment was performed with three biological replicates. The levels of c-di-GMP were compared with those of WT with the two-tailed Student’s t-test.

Diguanyl cyclase (DGC) and phosphodiesterase (PDE) activity assay

For the DGC activity assay, purified proteins were added to the reaction buffer (300 mM NaCl, 50 mM Tris-HCl, pH 7.5, 20 mM MgCl₂, 2 mM DTT) and 20 μCi [³²P]GTP (mixed with 50 μM cold GTP) was added. After 1 h at 28°C, the reaction was terminated by the addition of an equal volume of 0.5 M EDTA, pH 8.0. For the PDE activity assay, the purified RavR protein was added to the reaction buffer consisting of 50 mM Tris-HCl, pH 7.5, 250 mM NaCl, 25 mM KCl, 10 mM MgCl₂ and 2 mM DTT. Reactions were initiated by the addition of ~1 μM of [³²P]c-di-GMP substrate. Reactions were incubated at 28°C for 30 min before termination by adding an equal volume of 0.5 M EDTA, pH 8.0. For the DGC and PDE assays, reaction products were mixed with an equal volume of running buffer consisting of 1:1.5 (v/v) saturated (NH₄)₂SO₄ and 1.5 M KH₂PO₄, pH 3.6. Two microlitres of the reaction mixture was spotted and dried onto Cellulose PEI TLC plates (Selecto Scientific, USA). Plates were developed in running buffer, air-dried and exposed to a storage phosphor screen (GE Healthcare) and then the autoradiographic signals were recorded on a Typhoon FLA7000 (GE Healthcare).

Homology modelling and molecular docking analyses

The three-dimensional structure of the DHp-CA domain of RavS was generated by homology modelling methods. Homology modelling was carried out using Modeller 9.10 software. The known structures of CpxAHDC (4BIU), DivL (4EW8) and VraS (4GT8) were used for multiple template homology modelling.

The structure of c-di-GMP was extracted from the c-di-GMP-VCA0042 structure complex (PDB ID: 2RDE). The structure of DHp-CA was treated by adding hydrogen atoms, calculating the charge and combining nonpolar hydrogens. Docking calculations were carried out using Autodock 4.0 software. Affinity (grid) maps of 40 × 40 × 40 Å grid points were generated using the Autogrid program. After 100 calculations, the lowest energy conformation was chosen for c-di-GMP and protein interaction analysis.

Ethics Statement

There were no animal experiments in this study.