Introduction

Erwinia amylovora, a devastating vascular pathogen of rosaceous plants, is the causal agent of fire blight disease, which causes severe economic losses to apple and pear growers (Zhao 2014). E. amylovora, belonging to the Enterobacteriaceae family, is closely related to many important human pathogens, such as Escherichia coli, Yersinia pestis, and Salmonella enterica (Oh and Beer 2005; Zhao and Qi 2011). The exopolysaccharide (EPS) amylovoran and type III secretion system (T3SS) have been demonstrated to be two major virulence factors in E. amylovora (Ancona et al. 2014; Khan et al. 2012; Li et al. 2014; Zhao 2014). Amylovoran is a large heteropolymer composed of repeating units of galactose, glucose, and glucuronic acid (Nimtz et al. 1996) and its biosynthesis is encoded by the 12-gene ams operon, from amsG to amsL (Bernhard et al. 1993). Mutants abolished in amylovoran biosynthesis are non-pathogenic on host plants and amylovoran plays significant roles in biofilm formation and survival of the pathogen (Bellemann and Geider 1992; Kelm et al. 1997; Koczan et al. 2009; Wang et al. 2009; Zhao et al. 2009a, b). In E. amylovora, the Rcs phosphorelay system positively regulates amylovoran biosynthesis and is essential for bacterial virulence (Wang et al. 2009). The goals of the current study were to investigate the role of the conserved aspartate and lysine residues of RcsB in amylovoran biosynthesis, virulence, and motility in E. amylovora; the role of phosphorylation on the kinase or phosphatase activity of RcsC; and to determine whether the acetylation site of RcsB is required for DNA binding.

In bacteria, two-component signal transduction (TCST) systems serve as basic stimulus–response coupling mechanisms to allow organisms to sense and respond to changes in many disparate environmental conditions. Among many TCSTs found in bacteria, the Rcs phosphorelay system is a unique, highly complex, and widely studied TCST system found only in members of the Enterobacteriaceae family (Clarke 2010; Huang et al. 2006; Majdalani and Gottesman 2005). As an atypical TCST, the Rcs system is composed of three core proteins, RcsC, RcsD and RcsB, as well as several auxiliary proteins, including RcsA and RcsF (Clarke 2010; Huang et al. 2006; Majdalani and Gottesman 2005). The response regulator RcsB and the membrane-localized hybrid sensor kinase RcsC represent the classical members of bacterial TCST, whereas the membrane-bound sensor RcsD appears to be inactive as it lacks conserved residues in the active sites (Schmoe et al. 2011). The auxiliary protein RcsA is similar to RcsB, but without the conserved aspartate (D) residue in its N-terminal phosphor receiver (PR) domain (Bernhard et al. 1990; Pristovsek et al. 2003).

The signal transduction mechanism of the Rcs system is an unusual multiple step phosphorelay, i.e. His–Asp–His–Asp, and the phosphorylation state of Rcs proteins modulates their interaction (Fig. 1). Upon sensing environmental stimuli, RcsC autophosphorylates on a conserved histidine (H479 in E. coli) residue in its histidine kinase (HK) domain, and the phosphoryl group is then transferred to an aspartate (D875) residue in the RcsC PR domain, and further transmitted to an aspartate (D56) in the PR domain of RcsB via a unique C-terminal histidine-containing phosphotransmitter (Hpt) domain of RcsD (Majdalani and Gottesman 2005). Phosphorylated RcsB could form RcsB–RcsB homodimers or interact with RcsA to form RcsAB heterodimers, which then bind to an “RcsAB box” (TaAGaatatTCctA) to regulate gene expression, including the promoter of the ams operon involved in amylovoran biosynthesis (Pristovsek et al. 2003; Wehland et al. 1999; Wehland and Bernhard 2000).

Fig. 1
figure 1

The Rcs phosphorelay system in Erwinia amylovora. A simplified model for the Rcs phosphorelay system in E. amylovora. The RcsC, RcsD, and RcsB are proposed to form a complex through the α–β-loop (ABL) domain (Schmoe et al. 2011). RcsC is auto-phosphorylated (H481) by ATP upon perceiving environmental signals. The phosphoryl group is then transferred to an aspartate (D876) residue, which further transmitted to an aspartate (D56) in the PR domain of RcsB via phosphotransmitter (Hpt) domain of RcsD (Majdalani and Gottesman 2005). When phosphorylated, RcsB interacts with RcsA to form either RcsA–RcsB heterodimer or RcsB–RcsB homodimer, which then binds to an “RcsAB box” to regulate EPS and flagellar gene expression (Pristovsek et al. 2003). OM outer membrane, IM inner membrane, PM periplasm, CM cytoplasm, H histidine kinase domain, D1/D2 phosphor receiver domain, Hpt histidine phosphor transfer domain, ABL α–β-Loop domain, EPS exopolysaccharide amylovoran. H481, D876, and D56: position of the amino acids

Full size image

The Rcs system is found to be involved in the regulation of a wide array of phenotypes, including EPS, swarming motility, antibiotic resistance, biofilm formation, and virulence (Castelli and Garcia Vescovi 2011; Erickson and Detweiler 2006; Ferrieres and Clarke 2003; Francez-Charlot et al. 2003; Hinchliffe et al. 2008; Wang et al. 2007). These regulations by the Rcs system require the phosphorylation state of RcsB to promote its interaction with its target genes. In E. coli, RcsB phosphorylation is required for interaction with RcsA, thus repressing flhD expression and motility (Francez-Charlot et al. 2003; Venkatesh et al. 2010). Phosphorylated RcsB interacts with GadE to activate acid resistance pathways (Krin et al. 2010). In Salmonella typhimurium, phosphorylated RcsB inhibits biofilm formation (Latasa et al. 2012). In contrast, the BglJ–RcsB interaction de-represses expression of the bgl operon in E. coli regardless of RcsB phosphorylation status (Venkatesh et al. 2010).

In addition, RcsB can be acetylated by protein acetyltransferase (Pat) at conserved lysine residue (K180) in the function domain; while protein deacetylase CobB can revert RcsB K180 acetylation (Hu et al. 2013; Thao et al. 2010). Single amino acid substitutions to K180 that mimic acetylation (K180Q) or deacetylation (K180R, K180A) of RcsB abolish its repression of the flhDC genes, and, therefore, could not rescue the hypermotile phenotype of the rcsB mutant in E. coli (Thao et al. 2010); whereas RcsB was hyper-acetylated in the cobB mutant and transcription of the small RNA rprA, which requires phosphorylated RcsB for expression, was decreased (Hu et al. 2013). Furthermore, it has been reported that RcsC has dual functions, either as a kinase or as a phosphatase (Clarke et al. 2002). A single amino acid substitution, A904V, leads to a loss of function rcsC137 mutation, in which the expression of the cps and rprA genes was constitutively activated, thus abolishing RcsC phosphatase activity (Brill et al. 1988; Majdalani et al. 2005). However, an H479 mutation in the RcsC HK domain or the entire PR domain of RcsC, but not mutated D875 in the RcsC PR domain, could complement the rcsC137 mutation, indicating that the RcsC HK domain is not necessary for its phosphatase activity and other amino acids may be involved in the phosphatase activity of RcsC (Clarke et al. 2002). Domain structures of the RcsBC PR domains in E. coli have been recently revealed (Rogov et al. 2004, 2006; Schmoe et al. 2011), which provide a basis for structure–function studies. However, the role of conserved phosphorylation and acetylation residues of RcsB as well as RcsC has not been dissected in E. amylovora. Our results from site-directed mutagenesis and complementation studies clearly demonstrated that conserved aspartate and lysine residues of RcsB are required for amylovoran biosynthesis, virulence, and DNA binding in E. amylovora, and suggested that phosphor transfer from PR domain of RcsC is critical for its phosphatase activity.

Materials and methods

Bacterial strains and growth conditions

Bacterial strains and plasmids used in this study are listed in Table S1. LB broth was used for routine growth of E. amylovora and E. coli strains. When required, antibiotics were used at the following concentrations: 50 µg ml−1 kanamycin, 100 µg ml−1 ampicillin and 10 µg ml−1 chloramphenicol. For amylovoran production assays, bacteria were grown in MBMA medium (3 g KH2PO4, 7 g K2HPO4, 1 g [NH4]2SO4, 2 ml glycerol, 0.5 g citric acid, 0.03 g MgSO4) supplemented with 1 % sorbitol (Bellemann et al. 1994).

Generation of single, double and quadruple mutants by λ-red recombinase cloning

E. amylovora mutant strains were generated using λ phage recombinase method as described previously (Datsenko and Wanner 2000; Zhao et al. 2009a). Briefly, overnight cultures of E. amylovora strains harboring pKD46 were inoculated in LB broth containing 0.1 % arabinose and grown to exponential phase (OD600 = 0.8). Cells were harvested, made electrocompetent and stored at −80 °C. These cells were electroporated with recombination fragments of a cat or kan gene with its own promoter flanked by a 50-nucleotide homology from the mutagenesis target genes, obtained by PCR amplification from pKD32 or pKD13 plasmids. To confirm rcsA, rcsAB, rcsBC and rcsABCD mutations, PCR amplifications from internal cat or kan primers to the external region of the target genes were performed. The coding region of the rcsA, rcsAB, rcsBC and rcsABCD genes was absent from the corresponding mutant strains except for the first and last 50 nucleotides.

Construction of rcsB and rcsC mutant alleles

To introduce amino acid mutations in RcsB, the complementation plasmid pWDP3 (Wang et al. 2009), containing the rcsB–rcsD genes together with their regulatory region, was subjected to site-directed mutagenesis using the QuickChange XL kit (Stratagene). Plasmids containing the rcsB variants RcsBD56E (GAC–GAA), RcsBD56N (GAC–AAC), RcsBK180Q (AAA–CAG), RcsBK180A (AAA–GCA), RcsBD56E, K180Q and RcsBD56E, K180A were generated. The final plasmids were introduced into rcsB, rcsAB, rcsBC, and rcsABCD strains by electroporation. Transformants were selected on LB plates with appropriate antibiotics. Similarly, mutant alleles of RcsC were generated using a pWDP2 plasmid (Wang et al. 2009).

CPC assay to determine amylovoran concentration

Amylovoran concentration in mutant strains expressing RcsB alleles was quantitatively determined by the cetylpyrimidinium chloride (CPC) method as previously described (Bellemann et al. 1994). Briefly, overnight cultures of bacterial strains were harvested by centrifugation and washed three times with PBS. Five milliliters MBMA medium supplemented with 1 % sorbitol was inoculated to a final OD600 of 0.2 and incubated at 28 °C with shaking. After 24 h, 1 ml of each culture was centrifuged at 7,000 rpm for 10 min and 50 µl of CPC at 50 mg ml−1 was added to the supernatant. After 10 min of incubation, amylovoran concentration was determined by measuring OD600 turbidity of the suspension. Amylovoran concentration was normalized for a cell density of 1.0. Each experiment was performed in triplicate and repeated at least three times.

RNA isolation

Bacterial strains grown overnight in LB media with appropriate antibiotics were harvested by centrifugation and washed twice with PBS before inoculating 5 ml of MBMA medium supplemented with 1 % sorbitol. After 18 h incubation, 2 ml of RNA protect reagent (Qiagen) was added to 1 ml of bacterial cell culture. Cells were harvested by centrifugation and RNA was extracted using an RNeasy® mini kit (Qiagen) according to the manufacturer’s instructions. DNase I treatment was performed in column before elution and RNA was quantified using Nano-drop ND100 spectrophotometer.

RT-qPCR

One microgram of total RNA was reverse transcribed using Superscript III reverse transcriptase (Invitrogen) following the manufacturer’s instructions (Wang et al. 2012). Real-time PCR was performed using the ABI 7300 System (Applied Biosystems). Power SYBR® Green PCR master mix (Applied Biosystems) was used to detect gene expression of selected genes with primers designed using Primer3 software. RT-qPCR amplifications were carried out at 50 °C for 2 min, 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Dissociation curve analysis was performed after the program was completed to confirm amplification specificity. Relative gene expression levels were calculated with the ∆∆Ct method using the 16S ribosomal RNA as an internal control.

Virulence assays on apple trees and immature pear fruits

Virulence assays were performed as described previously (Wang et al. 2011). Briefly, overnight cultures of E. amylovora WT and mutant strains were harvested by centrifugation and suspended in 0.5 X PBS. Cell suspensions were adjusted to OD600 = 0.1 in PBS and inoculated onto seven actively growing “Gala” apple shoots by pricking the tip with a sterile needle and pipetting 5 µl of bacterial inoculum. Symptom development was recorded after 7 days post-inoculation and the experiment was performed at least two times. Immature pear fruits were pricked once with a sterile needle and inoculated with 2 µl of 100 X dilution of pathogen suspension at OD600 = 0.1. Disease symptoms were recorded after 8 days post-inoculation. The experiment was performed in triplicate at least three times.

Swarming motility assay

Bacterial strains were grown overnight on LB with appropriate antibiotics, harvested by centrifugation and washed three times with PBS. Bacterial suspensions were adjusted to an OD600 = 1.0, and 5 µl drop was placed onto the center of swarming agar plates (10 g tryptone, 5 g NaCl, 3 g agar per liter) as previously described (Hildebrand et al. 2006; Zhao et al. 2009b). Plates were incubated at 28 °C and swarming diameters were measured after 48 h post-inoculation. The experiments were performed in triplicate and repeated at least three times.

Construction of rcsB and rcsA in overexpression plasmids for protein purification

The rcsB gene was amplified from E. amylovora WT strain using 5′ and 3′ primers containing KpnI and NotI restriction sites, respectively. The rcsA gene was amplified from E. amylovora using primers containing KpnI and HindIII restriction sites. PCR products were digested with restriction enzymes and ligated into pKLD66 plasmid (Rocco et al. 2008) digested with the same enzymes. The presence of the rcsB and rcsA gene was verified by restriction digestion and sequencing. RcsB and RcsA overexpressing from pKLD66 plasmids produce proteins fused to an N-terminal His6–maltose-binding protein (MBP) tag (Rocco et al. 2008). The final plasmids were designated pRcsB6 and pRcsA6, respectively.

To introduce amino acid mutations in RcsB, plasmid pRcsB6 was subjected to site-directed mutagenesis using the QuickChange XL kit (Stratagene) as described above. Plasmids containing the RcsB variants RcsBD56E (GAC–GAA), RcsBD56N (GAC–AAC), RcsBK180Q (AAA–CAG), and RcsBD56E, K180Q were generated. The presence of the mutation was confirmed by sequencing.

Protein purification

RcsB and its mutant alleles as well as RcsA overexpression plasmids were transformed into an E. coli BL21 (DE3∆pat) strain and overnight cultures were inoculated into 500 ml of LB broth containing 150 µg ml−1 ampicillin. Cultures were grown at 28 °C for 2–3 h before inducing with IPTG (0.1 mM) and continued growth at 15 °C overnight. Cells were harvested by centrifugation, washed once with cell wash buffer (50 mM MOPS, 150 mN NaCl), resuspended 1:10 ratio w/v in wash buffer and frozen at −80 °C. Thawed cell suspensions were treated with lysozyme (250 µg mL−1) for 30 min and lysed by sonication. Cell lysates were centrifuged and Ni–NTA agarose resin (Qiagen) was added to the supernatants. Columns were washed with equilibration buffer (50 mM MOPS, 300 mM NaCl, 60 mM imidazole) and proteins were eluted with equilibration buffer containing 0.5 M imidazole. Samples were dialyzed against buffer (20 mM MOPS, 1 mM DTT). Protein concentrations were determined by Qubit protein assays (Life Sci).

Electrophoresis mobility shift assay (EMSA)

The light shift chemiluminescent EMSA kit (Pierce) was used for protein–DNA binding assays. Complementary oligonucleotides comprising the RcsAB box from the ams and flhD promoter region of E. amylovora (Wehland et al. 1999) were 3′ biotinylated and annealed before use. Reaction volumes of 10 µl with 20 fmol of labeled oligonucleotides were incubated with increasing equimolar amounts of RcsB and RcsA in 1X binding buffer with 50 ng/µL Poly(dI•dC), 0.5 mM MgCl2, 0.1 % Nonidet P-40, 0.05 mg mL−1 BSA, and 5 % glycerol. Reactions were incubated at room temperature for 25 min, 2.5 µL of 5X loading buffer was added, and the reactions were resolved into a 6 % native polyacrylamide gel in 0.5 % TBE buffer. Resolved binding reactions were transferred to a positively charged nylon membrane. Detection by chemiluminescence was performed according to the manufacturer’s instructions, and the membrane was exposed to X-ray film.

Results

RcsB phosphorylation mimicry activates amylovoran production, amylovoran biosynthesis gene expression, and restores virulence of the rcsB mutant

RcsB, along with RcsA, positively regulates amylovoran biosynthesis in E. amylovora (Figs. 1, 2a; Wang et al. 2009, 2011, 2012). To investigate the role of RcsB phosphorylation in amylovoran biosynthesis, the rcsB mutant was complemented with RcsB alleles containing amino acid substitutions, including D56E and D56N. D56E (aspartate to glutamic acid) mimics the phosphorylated state of RcsB, whereas D56N (aspartate to asparagine) mimics the inactive form of RcsB. In addition, substitutions of acetylation site K180 of RcsB to glutamine (K180Q), which mimics the acetylated RcsB, and to alanine (K180A) were also generated. As reported previously (Wang et al. 2009), amylovoran production was recovered in the rcsB mutant complemented with plasmid containing the native rcsB gene (Fig. 2a). Interestingly, amylovoran production was about 18 and seven times higher in RcsBD56E than in the WT and the rcsB mutant complemented with the native rcsB gene, respectively. However, amylovoran production was not restored in the rcsB mutant complemented with RcsB containing other mutations, including D56N, K180Q, K180A, D56E/K180Q and D56E/K180Q (Fig. 2a).

Fig. 2
figure 2

Effect of amino acid substitutions of RcsB on amylovoran biosynthesis and amylovoran gene expression. a Amylovoran production of Erwinia amylovora WT, the rcsB mutant strain and the rcsB mutant complemented with various RcsB mutant alleles, including RcsBD56E, RcsBD56N, RcsBK180Q, RcsBK180A, RcsBD56EK180Q and RcsBD56EK180A. Bacterial strains were grown in MBMA media supplemented with 1 % sorbitol for 24 h at 28 °C with shaking. Amylovoran concentrations was measured by CPC method and normalized to a cell density of 1. Data points represent the means of three replicates ± standard deviations. OD600 = optical density at 600 nm. b Relative expression of selected genes in MAMA medium by qRT-PCR of WT, the rcsB mutant and its complementation strain harboring various RcsB mutant alleles, including RcsBD56E, RcsBD56N, RcsBK180Q and RcsBD56E, K180Q, grown in MBMA medium supplemented with 1 % sorbitol for 18 h. The 16S rRNA gene was used as endogenous control. Fold changes were the result of the mean of three replicates. Each experiment was performed at least two times with similar results. Error bars indicated standard deviations

Full size image

To confirm that amylovoran production is correlated with the expression of amylovoran biosynthetic genes, expression of amsG, rcsA, and two RcsB-regulated genes (Eam_0255 and Eam_2938) (Wang et al. 2012) was determined by qRT-PCR (Fig. 2b). Expression of amsG, rcsA, Eam_0255 and Eam_2938 was up-regulated about 25-, 11-, 11- and 8-fold, respectively, in the rcsB mutant complemented with RcsBD56E; and down-regulated in the rcsB mutant complemented with all other alleles, except RcsBD56E, K180Q, in which expression was similar to the WT (Fig. 2b). These results indicate that phosphorylation of RcsB is critical for amylovoran production and expression of amylovoran biosynthesis genes.

Furthermore, a virulence assay was conducted for the rcsB mutant complemented with the mutated RcsB alleles on both immature pear fruits and apple shoots (Fig. 3a; Table 1). As previously described, virulence of the rcsB mutant could be complemented by expressing the native rcsB gene in trans (Fig. 3a; Table 1) (Wang et al. 2009). As expected, virulence of the rcsB mutant was completely restored by RcsBD56E on immature pear fruits (Fig. 3a), but was significantly reduced on apple shoots (Table 1). On the other hand, virulence of the rcsB mutant was not restored by expressing all other mutated RcsB alleles (D56N, K180Q, K180A, D56E/K180Q, and D56E/K180A) (Fig. 3a; Table 1). These results further indicate that phosphorylation of RcsB is essential for virulence of E. amylovora, but may have detrimental side effects if the phosphorylation state of RcsB is sustained (such as RcsBD56E).

Fig. 3
figure 3

Effect of amino acid substitution of RcsB on virulence. a Virulence assay of E. amylovora WT, the rcsB mutant strain and the rcsB mutant complemented with various RcsB mutant alleles, including RcsBD56E, RcsBD56N, RcsBK180Q, RcsBK180A, RcsBD56E, K180Q and RcsBD56E, K180A on immature pear fruits. b and c Virulence assay of E. amylovora WT, the rcsA, rcsC, rcsAB, rcsBC and rcsABCD mutant strains and the rcsAB, rcsBC and rcsABCD complemented with RcsBD56E, and RcsBD56N. Immature pears were surface sterilized, pricked with a sterile needle and inoculated with 2 µl of bacterial suspensions. Symptoms were recorded at 8 days post-inoculation (dpi)

Full size image
Table 1 Comparison of disease severity and motility in the rcsB mutant expressing RcsB alleles
Full size table

RcsB phosphorylation mimicry bypasses RcsD and RcsC to restore virulence and activates amylovoran production

To further dissect the role of RcsB phosphorylation in E. amylovora virulence, several deletion mutants, including rcsA, rcsBC, rcsAB, and rcsABCD, were generated. The double and quadruple mutants were non-pathogenic and did not produce amylovoran, as reported for rcs single mutants (Wang et al. 2009), except rcsA, which caused reduced disease and produced less amylovoran (Bernhard et al. 1990) (Figs. 3b, c; 4a); whereas the rcsC mutant produced more amylovoran, but was still non-pathogenic (Wang et al. 2009). First, virulence was fully restored in the rcsAB mutant complemented with RcsBD56E, partially with native RcsB and not with RcsBB56N (Fig. 3b), indicating that phosphorylation of RcsB is required for virulence, and also suggesting that RcsA contributes to virulence in E. amylovora. On the other hand, virulence was not restored in the rcsBC and rcsABCD mutants complemented with both RcsB and RcsBD56N, but virulence was partially restored in both mutants complemented with RcsBD56E (Fig. 3b, c), indicating that RcsC is required for full virulence and further suggesting that phosphorylation of RcsB is essential for virulence of E. amylovora, but may again have detrimental side effects if the phosphorylation state of RcsB remains. These results further suggest that dephosphorylation of RcsB by RcsC phosphatase activity may promote virulence, and that RcsA enhances the effect of RcsB for full virulence.

Fig. 4
figure 4

Effect of amino acid substitutions of RcsB and RcsC on amylovoran production and amylovoran biosynthesis gene expression in vitro. a Amylovoran production of E. amylovora WT, the rcsA, rcsB, rcsC, rcsAB, rcsBC and rcsABCD mutant strains and the rcsAB, rcsBC and rcsABCD mutants complemented with RcsBD56E, and RcsBD56N. Bacterial strains were grown in MBMA media supplemented with 1 % sorbitol for 24 h at 28 °C with shaking. Amylovoran concentrations were measured by CPC method and normalized to a cell density of 1. Data points represented the means of three replicates ±standard deviations. OD600 = optical density at 600 nm. b Relative expression of selected genes in MAMA medium by qRT-PCR of the rcsC mutant and its complementation strain harboring RcsC, RcsCH481A, RcsCD876N, RcsCT904A, and RcsCA905M; grown in MBMA medium supplemented with 1 % sorbitol for 18 h. The 16S rRNA gene was used as endogenous control. Fold changes were the result of the mean of three replicates. Each experiment was performed at least two times with similar results. Error bars indicated standard deviations

Full size image

We also determined amylovoran production in the double and quadruple mutants complemented with RcsB, RcsBD56E and RcsBD56N (Fig. 4a). As expected, the rcsBC, rcsAB, and rcsABCD mutant strains complemented with RcsBD56N could not produce any amylovoran (Fig. 4a). However, amylovoran was produced much higher when the rcsBC and rcsABCD mutants were complemented with RcsBD56E and RcsB; whereas only RcsBD56E, but not RcsB, could complement the rcsAB mutant to produce amylovoran (Fig. 4a). On the other hand, rcsAB (RcsBD56E) produced similar amount of amylovoran when compared to rcsABCD (RcsBD56E) and rcsBC (RcsB), whereas the highest level of amylovoran was produced by rcsBC (RcsBD56E) (Fig. 4a). These results again suggest that in vitro amylovoran production is not directly correlated with virulence since the rcsA mutant does not produce amylovoran, but remains virulent; whereas the rcsC and rcsBC (RcsB) produced higher amount of amylovoran, but remained non-pathogenic (Wang et al. 2009; Fig. 3b). In addition, these results further suggest that RcsB phosphorylation is required for interaction with RcsA, which enhances amylovoran production. However, in the rcsABCD mutant, expression of RcsB alone could lead to amylovoran production, suggesting that RcsC phosphatase activity is dominant under in vitro conditions to control amylovoran production.

Mutations of T904 and A905, but not phosphorylation mimicry of RcsC D876, constitutively activate the Rcs system

To further examine the role of RcsC, H481(histidine) in the HK domain and D876 (aspartate) in the PR domain of RcsC were mutated by site-directed mutagenesis. We then compared amylovoran production, gene expression and virulence in the rcsC mutant complemented with mutated RcsC alleles (Fig. 4b; Table 2). Amylovoran production in the rcsC mutant expressing RcsCH481A was similar to that complemented with the native RcsC, but three times higher than that of the rcsC mutant complemented with RcsC(D876E or D876N) (Table 2). Expression of rcsA, amsG, Eam_0255 and Eam_2938 was well correlated with amylovoran production data (Fig. 4B). However, all these mutant alleles could not complement the rcsC mutant to cause disease (Table 2), suggesting that although D876E mimics the phosphorylation state of RcsC, phosphor transfer is required for activating the Rcs system.

Table 2 Comparison of disease severity and amylovoran production in the rcsC mutant expressing RcsC alleles
Full size table

In E. coli, a point mutation rcsC137 (A904V) is a recessive loss of function mutation, which leads to constitutive expression of the cps genes (Clarke et al. 2002). When the corresponding A905 in E. amylovora RcsC was mutated, amylovoran production and expression of rcsA, amsG, Eam_0255 and Eam_2938 in the rcsC mutant complemented with RcsCA905M were all higher (Table 2; Fig. 4b). Similar results were obtained for the rcsC mutant complemented with RcsCT904A, a conserved threonine residue (Schmoe et al. 2011) (Table 2; Fig. 4b). However, when a second mutation of H481A or D876N was introduced, amylovoran production was significantly reduced (Table 2). Although A905M and T904A as well as double mutations in RcsC could not complement the rcsC mutant to cause disease in apple shoots, RcsC A905M and T904A mutant alleles still could restore the rcsC mutant to cause disease in pears (data not shown). These results suggest that both A905 and T904 are required for the phosphatase activity of RcsC, which can be rescued in vitro by mutations in active HK and PR domains.

Both RcsB and RcsBD56E bind to the amsG promoter

To assess whether RcsB and its mutant alleles bind to the amsG promoter, electrophoretic mobility shift assays (EMSAs) were performed with the amsG promoter as described previously (Lehti et al. 2012). As expected, RcsA itself did not bind to the amsG promoter, whereas RcsB and RcsBD56E bound to the ams promoter in the absence of RcsA (Fig. 5a, b) and we observed several shifted bands, which could be resulted from formation of dimers or tetramers. Furthermore, addition of equal amount of RcsA with RcsBD56E, but not RcsB, resulted in an increase in the molecular weight (MW) of shifted DNA (Fig. 5a, b), indicating that RcsA may only interact with phosphorylated RcsB. However, D56N, K180Q and D56E/K180Q of RcsB mutant alleles resulted in loss of binding of RcsB to the labeled DNA (Fig. 5c, d), suggesting that both D56 and K180 are required for binding and activation of the ams promoter.

Fig. 5
figure 5

Effect of amino acid substitutions of RcsB on its interaction with the amsG promoter. Complementary 30-bp oligonucleotides were 3′ labeled with biotin, reconstituted to double stranded and analyzed for binding of RcsB and its amino acid derivatives in EMSA assays. a RcsB; b RcsBD56E; c RcsBD56N; and d RcsBK180Q and RcsBD56E, K180Q. Numbers indicate the pmol of proteins added; large block arrows indicate free labeled DNA; small block arrows indicate band shifts

Full size image

Both RcsB and RcsBD56E bind to the flhDC promoter

In E. coli, the rcsB mutant is hypermotile and the acetylation mimic substitution K180Q abolishes RcsB DNA-binding ability of the flhDC promoter (Thao et al. 2010). In contrast, overexpression of RcsB represses flhDC expression in E. coli (Francez-Charlot et al. 2003). In E. amylovora, the rcsB mutant exhibited an irregular swarming pattern and decreased mobility as previously described (Wang et al. 2009; Zhao et al. 2009b). Swarming motility was performed for the rcsB mutant harboring various RcsB alleles (Table 1). Results showed that motility was slightly reduced for most RcsB alleles, with an exception for RcsBD56E mutation, which resulted in significantly reduced motility (Table 1). This significant reduction of motility can be restored by a second mutation of K180A or K180Q (Table 1). These results suggest that both phosphorylation and K180 are required for suppression of motility, possibly through binding to the flhDC promoter and suppressing flhDC gene expression (Francez-Charlot et al. 2003). EMSA results showed that RcsB or RcsBD56E alone bound to the flhDC promoter; whereas K180Q and D56E/K180Q substitutions resulted in loss of binding of RcsB to the flhDC promoter (Fig. 6). Interestingly, addition of equal amount of RcsA with either RcsBD56E or RcsB did not result in an increase in the MW of shifted DNA (Fig. 6), indicating that RcsA may not be required for binding to the flhD promoter. These results further indicated that D56 and K180 are required for DNA binding.

Fig. 6
figure 6

Effect of amino acid substitutions of RcsB on its interaction with the flhD promoter. Complementary 30 bp oligonucleotides were 3′ labeled with biotin and reconstituted to double stranded and analyzed for binding of RcsB and its amino acid derivatives in EMSA assays. a RcsB; b RcsBD56E; c RcsBK180Q; and d RcsBD56E, K180Q. Numbers indicate the pmol of proteins added; large block arrows indicate free labeled DNA; small block arrows indicate band shifts

Full size image

Discussion

TCSTs represent paradigms for gene regulation in prokaryotes and many studies have established the significant role of the Rcs system in the virulence of many enterobacterial systems (Huang et al. 2006; Majdalani and Gottesman 2005). Due to its complexity and existence in only members of Enterobacteriaceae, the Rcs system becomes a focus in understanding the molecular mechanisms of gene regulation. Recent biochemical, structural and computational studies have demonstrated and/or predicted many conserved features, and thus have provided new insight into structure-based regulatory mechanisms of the Rcs system (Rogov et al. 2004, 2006; Schmoe et al. 2011). In this study, we provided evidence that D56 residue in the PR domain and K180 residue in the function domain of RcsB are required for amylovoran biosynthesis, virulence and DNA binding in E. amylovora. We also demonstrated that, without other components of the Rcs system, the RcsB phosphorylation mimic (D56E) alone, but not RcsB, could restore virulence and amylovoran production. However, the effect of this constitutive activation of the Rcs system by the RcsB phosphorylation mimic may be detrimental as virulence was significantly reduced. In addition, we also confirmed that mutation of the A905 residue in the PR domain of RcsC led to constitutive activation of the Rcs system and further identified that mutation of T904 had similar effect. These results demonstrated that RcsB phosphorylation and dephosphorylation, and phosphor transfer from RcsC are essential for the function of this important signal transduction system.

The Rcs system regulates synthesis of EPS and virulence in several plant pathogenic bacteria, including Pantoea stewartii, Pectobacterium carotovorum, and E. amylovora (Andresen et al. 2010; Wang et al. 2009; Wehland et al. 1999). While mutation of the rcsB gene abolishes the ability to produce EPS, the rcsC mutation leads to higher levels of EPS production in vitro, but both mutations are non-pathogenic in planta (Bereswill and Geider 1997; Gottesman et al. 1985; Majdalani and Gottesman 2005; Wang et al. 2009). Our recent microarray studies provide differential gene expression data that could fully explain this discrepancy under in vitro and in vivo conditions (Wang et al. 2012). In this study, our results showed that the Rcs phosphorelay requirement for the regulation of amylovoran production in E. amylovora is dependent on the phosphorylation status of RcsB as well as its DNA-binding ability. We also observed that mutant strains expressing RcsBD56E overproduced amylovoran, but grew slower than WT and mutant strains (data not shown). Growth inhibition has also been reported in E. coli overproducing capsular polysaccharides (Brill et al. 1988). Thus, it is possible that constitutively activating the Rcs system may be very costly and be detrimental to its survival.

This phenomenon was further illustrated in the rcsB, rcsBC and rcsABCD mutants complemented by RcsBD56E phosphorylation mimic, in which amylovoran production was constitutively activated, but virulence was only partially restored on immature pear fruits. Moreover, virulence on apple shoots was further reduced in these mutants expressing RcsBD56E, suggesting that RcsB phosphorylation is required for virulence, but maintaining it without dephosphorylation may be detrimental to growth and survival of E. amylovora in vivo. Similarly, constitutive activation of the Rcs system through mutations of T904/A905 in the PR domain of RcsC led to overproduction of amylovoran, but not virulence, further suggesting that dephosphorylation of RcsB by the RcsC phosphatase activity is essential for virulence and also indicating that both T904 and A905 are required for the phosphatase activity of RcsC. In contrast, mutations of T904 and A905 could be complemented by both H481 (HK domain) and D876 (PR domain) mutations of RcsC, whereas in E. coli, only H479 mutation, but not D875 mutation could complement the rcsC137 mutation (A904V) (Clarke et al. 2002).

On the other hand, RcsB can be acetylated at a conserved K180 in the function domain, which may lead to instability or degradation of the RcsB (Hu et al. 2013; Thao et al. 2010). Acetylation mimic K180Q of RcsB could not rescue the hypermotile phenotype of the rcsB mutant in E. coli (Thao et al. 2010). In this study, our results demonstrated that acetylation mimic K180Q as well as D56E/K180Q double mutation of RcsB were unable to restore amylovoran biosynthesis, activate RcsB-regulated gene expression, and promote virulence in complementing the rcsB mutant. Moreover, although K180 is located in the DNA-binding domain of RcsB in E. amylovora, previous studies suggest that K180 did not make significant contact with DNA (Pristovsek et al. 2003). Our results showed that both K180Q and D56E/K180Q mutations of RcsB failed to bind to promoters of both amsG and flhD, indicating K180 in the DNA-binding site is required for DNA binding and also suggesting that acetylation may be an important post-translational modification in controlling the function of RcsB.

The negative regulatory effect of RcsB on swarming motility has been studied in E. coli and Salmonella enterica (Francez-Charlot et al. 2003; Wang et al. 2007). In both pathogens, the rcsB mutants are hypermotile (Fredericks et al. 2006; García-Calderón et al. 2007; Thao et al. 2010; Wang et al. 2007). In contrast, overexpression of RcsB represses flhDC expression in E. coli (Francez-Charlot et al. 2003). In E. amylovora, the rcsB mutant exhibited irregular swarming pattern and decreased mobility as previously described (Wang et al. 2009; Zhao et al. 2009b). However, flhDC promoter activity increased in the rcsB mutant, suggesting that the Rcs phosphorelay negatively regulates flagellar gene expressions in E. amylovora (Wang et al. 2009). In this study, the rcsB mutant complemented with RcsBD56E showed a strong decrease in motility as compared to WT and RcsB, confirming that the Rcs phosphorelay is also a negative regulator of swarming motility in E. amylovora. Similarly, acetylation mimic K180Q and D56E/K180Q double mutation of RcsB did not affect swarming motility when complemented the rcsB mutant, indicating that K180 is required for repressing swarming motility as in E. coli (Thao et al. 2010).

In summary, our results from site-directed mutagenesis and complementation studies clearly demonstrated that conserved aspartate and lysine residues of RcsB as well as RcsB phosphorylation are required for amylovoran biosynthesis, virulence, and DNA binding in E. amylovora. Our results also imply that phosphor transfer from RcsC and dephosphorylation of RcsB by RcsC through its phosphatase activity are essential for the function of RcsB and expression of RcsB-regulated genes. In addition, other post-translational modifications such as acetylation may also play a role in fine-tuning the function of this important signal transduction system in many enterobacterial pathogens and thus tightly control its activity to avoid overactivation, which may have detrimental effects on its survival in the harsh plant environment.