Strains and plasmids used in this study were listed in Table 1. A. avenae subsp. avenae strains were grown in Luria-Bertani (LB) broth or LB with 1.5% (w/v) agar at 30°C. LB agar without sucrose and LB agar containing 10% (w/v) sucrose were used for deletion mutagenesis (Li et al., 2011). The bacterial optical density (OD600) was determined with a spectrophotometer (Perkin Elmer Lambda35 UV/VIS). When required, antibiotics were added at the following concentrations: ampicillin (Amp), 100 μg ml^-1; chloromycetin (Chl), 3.4 μg ml^-1; kanamycin (Km), 50 μg ml^-1; and rifampicin (Rif), 100 μg ml^-1.

The cultivar of rice plant used in this study was ZheYou#1 (susceptible to A. avenae subsp. avenae). Experiments were performed in greenhouse. For inoculating plants, strains were grown in LB broth for 48 h, diluted in ddH₂O, and adjusted to OD600 of 0.6 (1 × 10⁸ CFU ml^-1; Liu et al., 2012). Leaf-clip inoculation was carried out on 6 weeks-old rice plants. Lesion length was measured at 14 days post-inoculation (Yu et al., 2011).

Each gene of interest was compared against sequences in the NCBI nr database using BLASTp (Altschul et al., 1997) followed by extracting the sequence of each specie with highest similarity for further study. We considered two genes as homologs when E-value <10^-5 and when the alignment similarity was higher than 40% with more than 80% coverage (Nogueira et al., 2009). ClustalW was used for sequences alignment (Thompson et al., 1994), and the conserved region of each alignment was trimmed with Gblocks (Castresana, 2000) under stringent settings described previously (Ciccarelli et al., 2006). Maximum Likelihood (Fang et al., 1999) phylogenies were evaluated and built by PhyML (Guindon and Gascuel, 2003) using a JTT model and a gamma distribution with eight rate categories. We performed 1,000 bootstraps to gain branch support values.

To investigate the role of tetR in A. avenae subsp. avenae, an in-frame deletion mutation by double cross-over events of this gene (RS_1091) was constructed as described previously using the suicide vector pKMS1 (Li et al., 2011; Zou et al., 2011) by homologous recombination on the background of wild-type strain RS-1. Briefly, two fragments flanking the start and stop codons of target gene, tetR1 (250 bp) and tetR2 (186 bp), were amplified from the wide-type genomic DNA (gDNA) with primer pairs tetR1-F/tetR1-R and tetR2-F/tetR2-R (Table S1), respectively. The two fragments were digested with BamHI and HindIII and ligated into the vector pKMS1 at the corresponding sites, resulting in pKMS-tetR. This recombinant plasmid was transferred into Escherichia coli S17-1 λ pir (Simon et al., 1983) and then introduced into A. avenae subsp. avenae by filter mating (Smith and Guild, 1980). The single colonies that emerged on LB plates containing kanamycin and rifampicin at 30°C for 2 days were then transferred to LB broth medium followed by incubating at 30°C and 200 rpm for 16 h. Afterward, the bacterial culture was plated to LB agar containing 10% (w/v) sucrose for the second cross-over through sacB and sucrose-positive selection. After sucrose resistant colonies were patched onto LB and LB plus kanamycin plates, respectively, individual colony that grew normally on LB plates, but did not grow on LB containing kanamycin, was considered potential deletion mutants where double exchange homologous recombination events occurred at the two tetR fragments, resulting in a 267 bp deletion. One of the mutants, named RS-tetR, was subsequently verified by PCR and sequencing before using for further study.

In order to complement the RS-tetR strain, the 1062 bp fragment containing full-length of tetR gene and 300 bp of its upstream region was amplified by PCR. The primers were designed based on RS-1 genome and listed in Table S1. The PCR product was cloned into pGEM-T Easy vector, verified by sequencing, and then cloned into pRADK (Gao et al., 2005). Complementation vector was introduced into mutant cells by electroporation and complementary strain was selected by resistance to Chl and Amp.

In order to monitor the growth rate, the growth curve assays were performed as described by Tu et al. (2012) with minor modification. 1 ml overnight culture, which was grown at 30°C with 200 rpm agitation in LB broth, was added into 99 ml fresh LB broth to OD600 of 0.05. The concentration of commercial 30% H₂O₂ (10 mol L^-1) solution and 100 mmol L^-1 paraquat (N,N′-dimethyl-4,4′-bipyridinium dichloride, C₁₂H₁₄Cl₂N₂, 257.16 g mol^-1) solution (2.57 g solid paraquat diluted with 100 ml ddH₂O) were used. The reactive oxygen compound (H₂O₂ or paraquat) was added at mid-log phase (OD600=0.3) to the indicated final concentrations. After comparing with the untreated bacterial growth curve, the sensitivity and tolerance of cells to the reagents was determined. This experiment was repeated three times independently.

Protein extraction was performed as described by Tu et al. (2012) with some modifications. Briefly, 2 ml overnight grown cells were added to 100 ml fresh LB broth to continue growing at 30°C with 200 rpm agitation. Paraquat or H₂O₂ was added to cells at mid-log phase (OD600=0.3) to the indicated concentration that perturbed the growth without causing cell death according to the survival assay (40/60 mM H₂O₂ or 0.4/0.8 mM paraquat, RS-1/RS-tetR). Cells were harvested by centrifugation (5,000 g, 10 min, 4°C) after exposure to the oxidative compounds for 60 min. Then the cells were treated with lysozyme (1 mg/mL) in the phosphate buffer (pH 8.0), disrupted by ultrasound, and centrifuged at 12,000 g for 15 min at 4°C in a Sorvall centrifuge to remove cell debris. A second-time centrifugation step (20,000 g, 30 min, 4°C) was required by by Sorvall centrifuge (Sorvall WX100, Thermo Scientific, USA). The protein content was quantified using the enhanced BCA Protein Assay Kit (Beyotime, China).

Protein extracts were subject to 1D SDS-PAGE gel composed of 5% acrylamide for stacking gel and 10% for running gel (Schägger and Von Jagow, 1987) with a mini-gel apparatus (VE-180 vertical electrophoresis bath, Tanon, China). Subsequently, the separated protein bands in the SDS-PAGE gel were visualized by silver staining.

The peptides released from trypsin digestion for LC-MS/MS analyzing were prepared in two biological replicates. LC was performed using a Dionex Ultimate 3000 nano-LC system. Firstly, the tryptic peptides were acidified by 2% acetonitrile with 0.025% trifluoroacetic acid before loading onto the Dionex Acclaim PepMap 100, C18 trap column (20 mm × 100 μm, 5 μm, 100 Å) at the flow rate of 10 μl min^-1. Then, the Dionex Acclaim PepMap 100, C18 analytical column (150 mm × 75 μm, 3 μm, 100 Å) were applied to divide the enriched tryptic peptides by gradient elution. The Bruker amaZon electron transfer reaction (ETD) ion trap system coupled with nano source expanded the capability to identify the trapped tryptic peptides under 300–1400 m/z and 50–2200 m/z scan range for MS and MS/MS, respectively.

The MASCOT LC-MS/MS ion search algorithm (Matrix Sciences) was applied to evaluate the LC/MS spectra and sequence similarity of resulting peptides to A. avenae subsp. avenae RS-1 was compared with Acidovorax species accessible on NCBI. To further identify the proteins, the cross correlation scores (X corr; Lin et al., 2007) of singly-, doubly- and triply-charged peptides were fixed greater than 1.8, 2.5 and 3.5, respectively. Then, a list of peptide sequences with the highest X corr values was identified. The hydrophobic nature of proteins was accessed through evaluating the grand average of hydropathicity (GRAVY) score of peptides by ProtParam ExPASy. Furthermore, proteins were annotated by RAST automatic pipeline to analyze their function.

After protein purification as mentioned above, 1–2.5 μg protein was used to measure superoxide dismutase (SOD) or catalase activities. Precisely, SOD activity was determined with a SOD assay Kit-WST (Beyotime, China). The Catalase Assay Kit (Beyotime, China) was used to detect catalase activity. Briefly, for SOD activity assay, the samples were treated by WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)- 2H-tetrazolium, monosodium salt] and incubated at 37°C for 30 min. Afterward, the absorption maximum at 450 nm was measured by Thermo Multiskan EX Micro plate Photometer (Thermo Fisher Scientific, Waltham, MA, USA). SOD activity was calculated by standard curve according to the instruction of manufacturer in kit. For catalase assay, the samples were treated with 10 μl of 250 mM H₂O₂ and incubated 5 min at room temperature, and the remaining H₂O₂ (not decomposed by catalase) was coupled with a substrate to generate N-4-antipyryl-3-chloro-5-sulfonate-p-benzoquinonemonoimine, which has an absorption maximum at 520 nm and was quantified spectrophotometrically. Catalase activity was then calculated by standard curve according to the instruction of manufacturer in kit. These experiments were repeated three times independently.

Biofilm formation assays were performed using the crystal violet (CV) assay method (Peeters et al., 2008). Bacteria were grown overnight with approximately 200 rpm agitation and diluted at 1:50 into fresh LB media. 100 μl of 10⁸ CFU ml^-1 mutant or wild-type bacterial suspension was added to one well of 96-well microtitre plate with sterile ddH₂O serving as blanks. Cells were allowed to form biofilm at 30°C for 48 h without agitation. After the planktonic organisms were removed and each well in the plate was rinsed and air-dried, 125 μl of 0.1% (w/v) CV solution was used to stain biofilm and 150 μl of 33% acetic acid was added to release the bond CV. The optical density was measured at 590 nm using a Thermo Multiskan EX Micro plate Photometer (Thermo Fisher Scientific, Waltham, MA, USA). This experiment was repeated three times independently with 12 replicates each.

The bacterial culture (OD600=1.0) was utilized for RNA extraction after exposure to the oxidative stressor (40/60 mM H₂O₂ or 0.4/0.8 mM paraquat, RS-1/RS-tetR) for 10 min. Bacterial total RNA was extracted from RS-1 and RS-tetR strain respectively as described by the manual of RNeasy Protect Bacteria Mini Kit (QIAGEN) and then used for generating first strand complementary DNA (cDNA) as described in the protocol of the Takara PrimeScript RT reagent Kit with gDNA Eraser (Takara). Briefly, 1 ml of bacterial cells were mixed with 2 ml of RNAprotect Bacteria Reagent before incubating for 5 min at room temperature. A pellet was obtained after centrifugation and was then treated by TE buffer (10 mM Tris⋅Cl, 1 mM EDTA; pH 8.0) containing 1 mg/ml lysozyme at room temperature for 5 min. After detecting the RNA quantity by agarose gel electrophoresis and the quality by Nanodrop ND1000 spectrophotometer V 3.5.2 (NanoDrop Technologies, Wilmington, DE, USA), 1 μg of the resulting total RNA was used at 42°C for 2 min to eliminate gDNA by gDNA Eraser and buffer before obtaining cDNA. Then the reverse transcription reaction was accomplished by incubating at 37°C for 15 min and then 85°C for 5 s in the presence of random RT primers. An ABI PRISM 7500 Real-Time PCR System and SYBR green fluorescence chemistry (SYBR Green I Master Kit, Roche Diagnostics) were employed to perform quantitative real-time RT-PCR amplification and analysis. Real-time PCR amplification was generated with 30 s of initial denaturation at 95°C, followed by 40 cycles of 5 s at 95°C, 30 s at 60°C for amplification. The reference genes (16S rRNA of both RS-1 and RS-tetR) were used to normalize the measurements between samples. Data analyses were based on change-in-cycling-threshold method (2^-ΔΔCt; Livak and Schmittgen, 2001). RNA extraction was carried out with two replicates per strain and qPCR was performed with three replicates.

The tetR gene was PCR-amplified from A. avenae subsp. avenae RS-1 genome with primers that introduced a BamHI site overlapping the translation initiation codon and a SalI site downstream of the stop codon, respectively (Table S1). A 763 bp tetR-containing DNA fragment was cut with BamHI and SalI and subcloned into the pGEX6P-1 expression vector (Novagen, USA) with the same restriction enzymes, yielding pGEX6P-tetR. Subsequently, pGEX6P-tetR was transformed into E. coli BL21(DE3) and was grown in LB medium containing 100 μg ml^-1 ampicillin at 30°C to OD600=0.5. Afterward, the culture was grown for an additional 3 h after isopropylthiogalactoside (IPTG) was added to a final concentration of 1 mM. The cells were then harvested by centrifugation at 8,000 g, 4°C for 10 min, and the pellet was resuspended in 10 ml of lysis buffer (70 mM HEPES, 20 mM imidazole, 650 mM NaCl, 0.5 mM β-mercaptoethanol, 10% glycerol, pH 8; Balhana et al., 2013). Cells were disrupted on ice by ultrasonic treatment for 20 min with 8 s rest period every 16 s and the supernatant was recovered by removing the cellular debris through centrifugation (12,000 g, 20 min, 4°C). Subsequently, TetR protein with GST-tag was separated and eluted by GST-Tag Bind resin (Sangon Biotech, China) followed by the manufacturer and the purity was confirmed by SDS-PAGE. The concentration of the purified protein was determined using the enhanced BCA Protein Assay Kit (Beyotime, China).

A 292 bp fragment comprising the intergenic region between tetR and pqiA’ was PCR amplified from A. avenae subsp. avenae RS-1 gDNA and the 3′-end of the fragment was labeled by biotin to be used as probe, according to the procedures of electrophoretic mobility shift assay (EMSA) Probe Biotin Labeling Kit (Beyotime, China). A 150 bp DNA fragment from the downstream of pqiB gene was worked as non-specific DNA control. Binding reactions were carried out by incubating varying concentrations of protein with 0.05 pmol of labeled probe in a total reaction volume of 20 μl. The EMSA were performed using a Light-Shift chemiluminescence EMSA kit (Beyotime, China) following the manufacturer’s instructions and the membrane was washed and detected following chemiluminescent method using Streptavidin-HRP conjugates and BeyoECL Plus reagents. ECL signals were captured and visualized by exposing the membrane to Hyperfilm^TM (GE Healthcare).

When needed, the data of quantitative assays were analyzed by SPSS 16.0 (SPSS Inc., Chicago, IL, USA) using an analysis of variance test and the mean values were compared with the least significant differences test.

Clusters of Orthologous Groups (COGs) enrichment analysis was carried out based on the Hypergeometric Test,

In which, N means the number of genes in the RS-1 genome, M means the number of genes filling into one COG category in the RS-1 genome, n means the number of differential proteins in tetR mutant LC-MS/MS result and I means the number of genes filling into one COG category in tetR mutant LC-MS/MS result. P < 0.01 was used as the cutoff to define the significance.

Horizontal gene transfer is an important mechanism driving the evolution of microbial genomes (Gogarten and Townsend, 2005). By comparing with A. avenae subsp. avenae ATCC 19860 and A. citrulli AAC00-1, two most closely related genomes deposited in public database, we found that the upstream and downstream regions of this cluster were highly conserved (Figure 1), indicating a novel gene cluster in RS-1. This cluster encodes four genes including pqiAA’B and tetR gene. The tetR gene is located upstream from the paraquat-inducible genes (pqiAA’B) in tandem. In the phylogenetic trees generated using the four genes in this cluster, RS-1 always grouped together with different bacterial species, with Pseudomonas as the out-group (Figure S1). These gene trees were highly incongruent with the bacterial species tree. Based on these results, we hypothesize that this gene cluster may be horizontally transferred from a Pseudomonas ancestor.

In previous reports, TetR family members are particularly abundant in microbes and play important roles in bacterial adaptability or fitness when they are exposed to adverse environmental challenges, such as oxidative stress conditions (Ramos et al., 2005). Thus, in order to understand the role of the TetR protein in RS-1, we decided to investigate its molecular function by deleting tetR gene in RS-1 strain (Figure S2).

The growth rate of the tetR mutant strain RS-tetR decreased significantly compared with the wild-type as well as the complemented strain RS-tetR-comp (Figure S3). Wild-type and RS-tetR-comp strains entered the log phase 1 h after the culture started and the OD600 reached up to 2.2 while the RS-tetR strain entered the log phase much later and its OD600 only reached 1.7.

In order to investigate the tolerance capability of A. avenae subsp. avenae strains to oxidative stress, we carried out a comparative analysis of the growth kinetics under different concentrations of paraquat or H₂O₂. Generally, paraquat or H₂O₂ strongly inhibited the growth of both mutant and wild-type strains. Importantly, the mutant strain showed higher resistance than wild-type strain to these reactive oxygen reagents (H₂O₂ or paraquat) when added at mid-log phase (OD600=0.3; as the arrow shown in Figure S4). Six different concentrations of paraquat (0.0–1.0 mM) or H₂O₂ (0–100 mM) were used to test their effects. At 0.4 mM paraquat or 40 mM H₂O₂, RS-1 and RS-tetR-comp were completely inhibited, showing no further growth (Figures S4A,B,E,F), whereas the RS-tetR was little affected and continued to grow. At 0.8 mM of paraquat or 60 mM of H₂O₂, growth of RS-tetR strain was inhibited (Figures S4C,D), whereas RS-1 and RS-tetR-comp cells started to die. Moreover, the results from LB plate assay were consistent with those of growth curves. The RS-tetR grew well on LB plates containing 0.8 mM paraquat or 60 mM H₂O₂, while RS-1 and RS-tetR-comp was inhibited (Figure 2). These data suggest that TetR is involved in negative regulation of cell growth and adaption under oxidative stress.

The morphological appearance and thickness of biofilm showed obvious differences after 48 h of adhesion at 30°C without agitation. Particularly, RS-tetR strain formed a thicker biofilm compared with RS-1 and RS-tetR-comp strains. When quantified, RS-tetR yielded readings twice as high as RS-1 and RS-tetR-comp (Figure 3). The quantitative data confirmed that the mutant strain exhibited significantly (P < 0.05) higher biofilm-forming capability compared to the other two strains. However, the virulence did not show significant difference comparing among wild-type, mutant and complementation strains by observation of symptoms and analysis of lesion lengths 14 days post-inoculation (Figure S5).

Superoxide dismutase and catalase in cytoplasm of bacterial cells play a pivotal role in protection against superoxide and peroxide stress (Henriques et al., 1998; Inaoka et al., 1999). They are highly efficient enzymes and likely to be transiently induced (Hassett and Cohen, 1989). Our results in Figure 4 showed that their enzyme activities were increased in response to oxidative stress compared to the untreated control (Figure 4). More importantly, the RS-tetR mutant strain showed markedly higher levels of catalase and SOD activities compared to wild-type and complemented strains (Figure 4). The quantitative analysis of SOD activity showed that the absence of TetR led to ∼2.3-fold increase in enzyme activity in paraquat treatment and ∼2.2-fold in the H₂O₂ treatment. The quantitative catalase activity analysis yielded similar results: deletion of tetR resulted in ∼2.7-fold and ∼2.1-fold increase in H₂O₂-treated and paraquat-treated samples, respectively. These results raised the question as to whether TetR can also influence expression of genes encoding other oxidative stress detoxifying enzymes, such as sodAB, katA, ahpCF and hemAXCDBL (Zuber, 2009). We therefore carried out LC-MS and quantitative PCR analysis to address this issue.

In order to investigate how many proteins can be influenced by TetR under oxidative conditions, we employed 1D SDS-PAGE to compare cytoplasmic proteins of RS-1 and RS-tetR, prepared as described in Section “Materials and Methods” (Figure 5). The resultant protein bands were further digested and peptides were then analyzed by LC-MS/MS subsequently.

The protein profiling data of LC-MS/MS from two biological replicates revealed a non-redundant list of proteins, after removing overlapping entries, with high (>98%) confidence. The results clearly suggest differential expression of these newly identified cytoplasmic proteins between wild-type and tetR mutant strains. In total, there are 111 proteins identified (Tables S2 and S3). Specifically, 15 proteins were only identified in RS-1 while 96 proteins were identified only in RS-tetR. These proteins are mainly associated with oxidative stress response, detoxification, and biofilm-formation. COG enrichment suggested that there were more proteins involved in energy production and conversion, cell motility, post-translational modification in the RS-tetR strain, compared to RS-1 (P < 0.01, hypergeometric distribution test); whereas the wild-type strain produced more proteins involved in coenzyme transport and metabolism, amino acid transport and metabolism, and cell division (Figure S6). Overall, these results provide an explanation for the observed physiological changes caused by TetR and support our main hypothesis that this gene cluster plays a key role in bacterial survival under oxidative stress.

The LC-MS/MS study exhibited a comprehensive profile of proteins whose coding genes are regulated by the TetR regulator under oxidative stress. For further confirmation, we selected 11 genes from the proteomic profile which were involved in oxidative stress response or biofilm-formation, or encoded detoxifying enzymes. We examined their gene expression with quantitative real-time PCR. The results in Figure 6 showed different fold change in gene expression regulated by tetR comparing mutant to wild-type strain. As expected, the expression of pqiA’AB was significantly up-regulated in the RS-tetR strain (6.2-, 6.5-, 7.8-fold, respectively). Similar results were obtained for genes sodA, ahpF and katA, which encode SOD, alkyl hydroperoxide reductase, and catalase, respectively (Figure 6). Up-regulated expression in these genes is expected to lead to increase in detoxifying enzyme activities, as shown in SOD and catalase activities assay (Figure 6). Moreover, the increased detoxifying enzyme activity can also contribute to the higher resistance to paraquat and H₂O₂. Flagella are important for bacterial colonization and biofilm formation (Conrad, 2012; Bogino et al., 2013; Ren et al., 2013). Quantitative real-time PCR showed that the gene expression level of fliL and flaB, two flagella-associated genes, was 1.4-fold and 1.8-fold higher in the absence of TetR. The clpAB gene family plays a role in synthesis of stable and native proteins, which are required for intracellular replication, stress tolerance, and biofilm formation of bacteria, in the presence of ATP (Wawrzynow et al., 2003; Frees et al., 2004). Thereby it was not surprising to see these genes up-regulated in the absence of the tetR regulator.

To investigate whether the repression of the genes on the gene cluster by TetR occurs via physical interaction between TetR and the promoter of the gene cluster, we performed EMSA using a 292 bp putative promoter fragment of the pqiA’-tetR intergenic region from A. avenae subsp. avenae RS-1 (Figure 7A). We purified the TetR protein with GST tag (about 48 kDa) and confirmed it by SDS-PAGE with silver staining (Figures S7A,B). After mixing the biotin-labeled probe with TetR, a band with substantial mobility shift corresponding to a TetR-DNA complex was detected (Figure 7B, Lane 3–5). When TetR concentration was increased, an increase in the shifted band was observed (Figure 7B, Lane 3–5). This result indicates that the binding of TetR to the DNA fragment is concentration dependent. As a control, we showed that there was no mobility shift of the probe when TetR was not added (Figure 7B, Lane 2). In addition, a strong binding shift appeared when unlabeled non-specific DNA was added (Figure 7B, Lane 1); while when unlabeled cold probe was added, it abolished the binding, demonstrating the binding sensitivity and specificity of TetR for this intergenic region (Figure 7B, Lane 6–7).