The bacterial strains and plasmids used in this study are listed in Supplementary Table S1. Xoo strains were grown in either nutrient broth (NB) medium or hrp-inducing medium XOM2 (Xiao et al., 2007) at 28°C. The antibiotics used in this study were spectinomycin (50 μg mL^-1), kanamycin (25 μg mL^-1) and ampicillin (100 μg mL^-1).

The PhaR gene knockout construct was made by overlap PCR to produce a chimeric DNA including 892 bp upstream flanked DNA of phaR gene, 1.4 kb kanamycin cassette from vector pKD13 and 729 bp downstream flanked arm of phaR gene. This chimeric DNA was then cloned into pMD19 through T-A cloning. The resulting plasmid was introduced into PXO99^A cells by electroporation (Song and Yang, 2010). The single exchange mutant was screened on kanamycin and ampicillin-containing media. The second exchange occurred simultaneously during the culturing in a liquid medium without any antibiotics. The clones that could grow on NA medium supplemented with kanamycin and ampicillin-sensitive were chosen as phaR gene deficient mutant candidates for further PCR characterization.

To complement the phaR mutant, the cosmid pHMPhaR containing the coding sequence of phaR gene was transformed into PXO99ΔPhaR by the freeze-thaw method (Dityatkin et al., 1972) to make the complementary transformant C-ΔphaR. Primers used for construction are listed in Supplementary Table S2.

For motility assay, bacterial strains were grown in NB medium for 2 days. Bacterial cells were then harvested by centrifugation, re-suspended in sterilized distilled water, and adjusted to an optical density (OD₆₀₀) of 1.0. 1 μL of bacterial suspension was then dropped onto the plates containing semisolid medium (containing per liter 5 g Bacto Peptone, 1 g yeast extract, 3 g beef extract, 10 g sucrose and 0.03 g agar). Plates were incubated at 28°C for 3 days. The diameters of the swimming zones indicated the ability of bacterial movement. The experiment was repeated four times with three replicates for each time.

Bacterium was cultured in NB overnight at 28°C. An aliquot of 20 mL culture was taken out and added with KCl to a final concentration of 1%. Then two volumes of absolute ethanol were added to precipitate EPS at -20°C overnight. EPS was collected by centrifuging at 12000 rpm for 20 min, and the pellet was dried at 55°C overnight before weighed. The bacterial concentrations of the culture used for EPS extraction were measured by diluting serially and plating in triplicate on NA plates to count the colony numbers. The production of EPS was evaluated in terms of weight (g) per cfu. The experiment was repeated three times.

One-month-old rice cv. IR 24 seedlings were used for virulence assays. The top two fully expanded leaves were inoculated with Xoo strains at OD₆₀₀ of 0.5 via the leaf-clipping method. Lesion lengths were measured 15 days after inoculation, and the inoculated leaves were then pooled and cut into small pieces and ground in sterile water (with 1 ml sterile water per leaf) using a sterilized mortar and pestle. The resulting suspension was diluted serially and plated in triplicate on NA agar with appropriate antibiotics. Counts were taken and converted into CFU per leaf.

One-month-old Nicotiana benthamiana was used for HR assays. A bacterial suspension at OD₆₀₀ of 0.2 was infiltrated into leaf tissue using a needleless syringe. Symptoms were recorded 24 h after infiltration. The assays for micro-HR response, active oxygen burst, and ionic conductivity of tobacco leaf were performed 12 h after infiltration as previously described (Peng et al., 2003; Fu et al., 2007). For in situ detection of H₂O₂ at indicated time points, tobacco leaves were detached and vacuum infiltrated in a 3,3′-diaminobenzidene (DAB) solution (1 mg/ml, pH 3.8) for 10 min, and then incubated at 25°C for 7–8 h. The deposits formed were visualized under microscopy after discoloring leaves in boiled absolute ethanol. Micro cell death was monitored by staining with lactophenol–trypan blue and destaining in chloral hydrate (2.5 mg/ml) as described by Peng et al. (2003).

Bacteria strains were grown either in nutrient broth (NB) medium or in hrp-inducing medium XOM2 overnight at 28°C (Xiao et al., 2007). 1 mL culture was collected for RNA isolation using SV Total RNA Isolation System (Promega, Madison, WI, United States). Reverse transcription was carried out using PrimeScript RT Master Perfect Real Time Kit (TakaRa Bio. Inc., Dalian). Quantitative real-time PCR (qRT-PCR) was performed using SYBR Premix Ex TaqTM Kit (TaKaRa Bio. Inc., Dalian). 16s rDNA was used as internal reference gene. The primer sets used are listed in Supplementary Table S2.

Phylogenetic analysis of the aligned full-length sequence of phaR gene was conducted with MEGA 3.1 using neighbor joining and the Poisson correction model (Saitou and Nei, 1987; Kumar et al., 2004). Pairwise deletion was used for handling of sequence gaps. Members of the PhaR effector family used for phylogenesis analysis are indicated as their bacterial origins. The responding locus tags and bacterial origins are listed in Supplementary Table S3. The sequences used for phylogenetic tree generation are listed in Supplementary Table S4.

For relative gene expression assays, Student’s t-test at the 99% confidence level was performed. One-way analysis of variance (ANOVA) statistical analyses were performed on the motility assay, EPS production assay and virulence assay. The Tukey honest significant difference test was used for post-ANOVA pairwise tests for significance, set at 5% (P < 0.5).

Based on the genomic sequence annotations of Xoo strain PXO99, five PHA metabolism related genes were found (Figure 1A), which include a beta-ketoacyl-ACP reductase (KACPR, PXO_RS09075) gene, a polyhydroxyalkanoate synthesis repressor (phaR, PXO_RS09070) gene, a poly(R)-hydroxyalkanoic acid synthase subunit PhaC (phaC, PXO_ RS11155) gene, a PHA synthase subunit (phaE, PXO_RS11160,) gene, and PHB depolymerase (phaZ, PXO_RS14020) gene. Among these, phaC and phaE were adjacent. To investigate the relationship of phaR with other related genes, we qualified the transcriptional expression levels of KACPP, phaC, phaZ, and phaE genes in the phaR mutant and the wildtype. Considering phaR gene is a stress responsible gene, we selected two time points, the first at 12 h representing the early stage of growth with the equal growth rates of mutant and wildtype strains, the other at 22 h representing the later stage with differential growth rates of mutant and wildtype strains. The results showed that mutagenesis of phaR gene affected the expression of other PHA metabolism related genes (Figures 1B,C). At both stages, the most significantly affected was the expression of phaZ, responsible for the degradation of PHB granule. The relative expression levels of phaZ for phaR mutant were 13.9 and 65.8 folds of those in the wildtype at 12 and 22 h after inoculation, respectively. The expression of phaC and phaE were also induced in the mutant, with 5.9 and 5.7 folds of induction for phaC at 12 and 22 h, respectively, and 4.7 and 2.1 folds of induction for phaE at 12 and 22 h, respectively. These results indicate that the bacteria may produce poly(3-hydroxybutyrate) (PHB), a type of PHA, and that phaR gene of PXO99 negatively regulated the degradation of PHB, which was more apparent in the late stage of growth with insufficient oxygen and nutrients.

The Xoo phaR gene is commonly found in plant pathogenic Xanthomonas species, but the gene sequences from different strains are polymorphic. The phylogenetic relationship was established based on phaR gene nucleotide sequences of 43 strains of Xanthomonas (Figure 2). As shown in Figure 2, the phylogenomic relationship displayed host, tissue, and geographic specificity. There are two main phylogenetic groups. One consists of strains of Xoo and X. oryzae pv. oryzicola (Xoc), which infect monodicot. The other group, with the exception of X. axonopodis pv. dieffenbachiae LMG 695 (Xad LMG695), a pathogen of monocot Ariod plants (Robène et al., 2016), consists of dicot plant parasites, including X. axonopodis pv. phaseoli (XapCFBP6982, pathogen of bean (Gent et al., 2005), X. perforans (Xp91-118, pathogen of Solanacearum plant (Itako et al., 2015), X. euvesicatoria LMG930 (Xe LMG930, pathogen of pepper Vancheva et al., 2018), X. campestris pv. vesicatoria (Xcv85-10, pathogen of pepper and tomato; Jones et al., 2004; Quelas et al., 2016), X. axonopodis pv. citrumelo F1 (XacF1, pathogen of citrus Jalan et al., 2011), X. citri pv. vignicola strain CFBP7111 (XcivCFBP7111, pathogen of cowpea Ruh et al., 2017a), X. fuscans subsp. aurantifolii strain 1566 (Xfa1566, pathogen of citrus Rigano et al., 2007), X. phaseoli pv. phaseoli (XppCFBP6988, pathogen of common bean Ruh et al., 2017b), X. fuscans subsp. fuscans str. 4834-R (Xff4834-R, pathogen of common bean Aritua et al., 2015). Within the first group, 11 Xoc strains infecting the mesophyll tissue formed a clade, and had identical sequences. Interestingly, 22 Xoo strains were divided into two clades, one for 9 African strains with identical sequences, the other for 13 Asian strains originated from Japan, Korea and the Philippines having almost identical sequences except one sequence from PXO99 strain with a 1 bp difference.

To examine the effects of phaR mutagenesis on the morphology of PXO99^A, we spotted the wildtype PXO99^A, the mutant PXO99ΔphaR and the complemented strain C-ΔphaR on NA plates. 3 days later, PXO99ΔphaR developed dry colonies with reduced slime on the surface, while PXO99^A and C-ΔphaR had normal mucoid colonies (Figure 3A). This indicated that phaR gene might regulate the growth rate of the bacteria or affect the production of EPS.

To test if phaR regulates bacterial growth, we examined the growth curves of phaR mutant both in rich (NB) and minimal (XOM2) nutrient medium. The results showed that, in both media, the growth rate of PXO99ΔphaR was reduced greatly compared with that of the wildtype (Figures 3B,C). When inoculated with the same starting concentrations in liquid NB and XOM2 medium, the wildtype PXO99^A, mutant PXO99ΔphaR and the complementary strain PXO99ΔphaR/phaR all grew at the similar rate for the first few hours, and then started to differ about 12 h after inoculation. PXO99ΔphaR kept propagating steadily but at much lower level, with the population size about 15.9 and 14.1% of the wildtype at 32 h (in NB) and 60 h (in XOM2), respectively (Figures 3B,C). The complement strain grew similarly as the wildtype in both media.

Since PXO99ΔphaR displayed the dry colony phenotype, we next tested if the EPS production was affected in the mutant. To do this, we extracted and compared the EPS mass from the 46 h NB culture supernatant of PXO99^A and PXO99ΔphaR. With the same volume of culture supernatant, significantly less EPS pellet was precipitated by ethanol from the phaR mutant than the wildtype. Since the growth of the mutant was reduced compared to the wildtype and the complement strain, we measured the bacterial numbers and determined the EPS production per cell, in order to deduce if the reduced EPS production was caused by a reduced amount of bacteria or by the reduced EPS production efficiency. The results showed that EPS productivity of the mutant was greatly impaired and was about 22% of the wildtype (Figure 3D). This suggests that phaR positively regulates EPS production in PXO99^A.

Since phaR mutagenesis affected the bacterial growth in media, esp. in XOM2, a medium mimicking the host nutrient state in vitro, we hypothesized that it might also affect the bacterial propagation of Xoo in rice. To test this, we inoculated rice cv. IR24 with PXO99^A, PXO99ΔphaR, and PXO99ΔphaR/ΔphaR by the leaf-clipping method. 14 days after inoculation, the average lesion length caused by PXO99ΔphaR inoculation was about 4 cm, which was about one third of that caused by wildtype strain PXO99^A (Figures 4A,B). The bacterial multiplication in the inoculated leaves was measured by counting the colonies grown on media. Compared with PXO99^A, the population size of PXO99ΔphaR in rice was reduced about 68% (Figure 4C). Complementation of the mutant with phaR gene restored the phenotype similar to the wildtype level. This data showed that the growth of PXO99ΔphaR is drastically restrained due to the loss of the phaR gene, which then reduces the overall virulence of Xoo in rice.

The fact that the phaR mutant failed to travel to long distance to cause leaf lesion suggested that phaR might be involved in bacterial movement. To test this hypothesis, we evaluated the swimming motility of Xoo strains in semisolid motility agar. Unexpectedly, the results showed that PXO99ΔphaR was hypermotile, producing swimming haloes 1.36-fold in diameter of those of the wildtype PXO99^A. The complementary strain restored motility to wildtype level (Figure 5). Thus, phaR might negatively affects motility in PXO99^A.

In Xoo, pathogenesis on host plants and HR on non-host plants are correlated properties regulated by the hrp genes. Because PXO99ΔphaR displayed attenuated virulence in the host plant rice, we next tested if the HR on non-host tobacco would be changed. To do this, a bacterial suspension was infiltrated into tobacco leaves to compare the ability of different strains to induce HR. 24 h after infiltration, PXO99ΔphaR induced an intensive HR and produced a larger necrosis area than that caused by PXO99^A (Figure 6A).

To further measure the HR reaction strength, two landmark events of HR, the oxidative burst and micro-HR cell death were examined at 12 hpi through directly staining the biochemical component H₂O₂ with DAB or the dead cells with the trypan blue. DAB is oxidized by hydrogen peroxide to give a dark-brown color, and trypan blue selectively stains dead tissues or cells. The DAB staining results showed that PXO99ΔphaR induced extensive and intensive oxidative burst (dark brown spots) around the infection site (Figure 6B). Trypan blue staining of the leaf disks showed PXO99ΔphaR inoculation induced obvious cell death (blue spots), indicating micro-HR occurred around the infiltration site; while for PXO99^A treatment, no obvious cell death was found (Figure 6D). Loss of phaR gene also increased the ability of PXO99^A to induce cell membrane permeability of tobacco (Figure 6C). The reintroduction of the phaR gene restored the mutant’s phenotypes similar to those of the wildtype strain (Figure 6). These results indicated that the phaR gene negatively regulated HR induction in tobacco.

In Xoo, HR induction in non-host plants is associated with hrp genes, esp. the hpa1 gene which encodes HR inducing harpin protein. To evaluate if the increased induction of HR by PXO99ΔphaR in tobacco is related to the changes of harpin expression, we analyzed the expression levels of hpa1 gene in different strains under hrp inducing medium via quantitative real time PCR (qPCR). The results showed that when cultured in XOM2 medium, which mimics the nutrient state of the plant, the expression levels of hpa1 in both mutant and wildtype strains increased significantly compared with those in nutrient rich medium NA broth. Moreover, compared with wildtype, the expression of hpa1 increased significantly in PXO99ΔphaR both in NA and XOM2 broth media (Figure 7). These results confirmed that hpa1 gene is induced by plant and suggested that phaR gene negatively regulated the expression of hpa1 in Xoo.