Identification of 17 HrpX-Regulated Proteins Including Two Novel Type III Effectors, XOC_3956 and XOC_1550, in Xanthomonas oryzae pv. oryzicola

The function of some hypothetical proteins, possibly regulated by key hrp regulators, in the pathogenicity of phytopathogenic bacteria remains largely unknown. In the present study, in silicon microarray data demonstrated that the expression of 17 HrpX-regulated protein (Xrp) genes of X. oryzae pv. oryzicola (Xoc), which causes bacterial leaf streak in rice, were either positively or negatively regulated by HrpX or/and HrpG. Bioinformatics analysis demonstrated that five Xrps possess a putative type III secretion (T3S) signal in the first 50 N-terminal amino acids, six xrp genes contain a PIP-box-like sequence (TTCGB-NX-TTCGB, 9≤X≤25) in the promoter regions, and two Xrps have both motifs. Twelve Xrps are widely conserved in Xanthomonas spp., whereas four are specific for X. oryzae (Xrp6) or Xoc (Xrp8, Xrp14 and Xrp17). In addition to the regulation by HrpG/HrpX, some of the 17 genes were also modulated by another hrp regulator HrpD6. Mutagenesis of these 17 genes indicated that five Xrps (Xrp1, Xrp2, Xrp5, Xrp8 and Xrp14) were required for full virulence and bacterial growth in planta. Immunoblotting assays and fusion with N-terminally truncated AvrXa10 indicated that Xrp3 and Xrp5 were secreted and translocated into rice cells through the type-III secretion system (T3S), suggesting they are novel T3S effectors. Our results suggest that Xoc exploits an orchestra of proteins that are regulated by HrpG, HrpX and HrpD6, and these proteins facilitate both infection and metabolism.


Introduction
Bacterial leaf streak (BLS) of rice, which is caused by Xanthomonas oryzae pv. oryzicola (Xoc), is a destructive plant disease in Asia. The pathogen infects rice through leaf stomata or wounds and colonizes intercellular spaces in the mesophyll, resulting in water-soaked interveinal lesions that develop into translucent streaks [1]. The infection routes and the symptoms caused by Xoc differ from those incited by the closely-related pathogen, X. oryzae pv. oryzae (Xoo). Xoo enters rice leaves through hydathodes or wounds, propagates in the intercellular spaces of the underlying epidermis, and then spreads throughout the plant in the xylem, where it presumably interacts with xylem parenchyma cells [2,3,4]. The Xoc-rice pathosystem is an important working model to elucidate how pathogens evade the plant host immune system [2,5]. The complete genome sequence and comparative and functional genomic studies of three Xoo strains like KACC10331 [6], PXO99 A [7], MAFF311018 [8] and Xoc strain BLS256 [9] have furthered our understanding of Xanthomonas-rice interactions. However, it remains unclear whether the numerous hypothetical proteins, maybe possibly regulated by HrpX (Xrps) annotated in X. oryzae, are involved in virulence.
The expression of genes coding for the T3S and effectors is generally plant-inducible and regulated by a key hrp regulatory factor, HrpX [12,14,23,24]. HrpX is an AraC-type transcriptional regulator that controls the expression of genes in the HrpX regulon by binding the PIP (plant-inducible promoter)-box; this is a conserved cis-element with the consensus TTCGB-N 15 -TTCGB ('B' refers to any base except adenine) [25][26][27][28]. The PIP-box is normally followed by a 210 box that is located at 30-32 bp further downstream [29]. T3SEs also contain secretion signals in the first 50 N-terminal amino acids, which are characterized by one or more of the following: $20% Ser and Pro [22,26,30]; more than five Ser residues [13,31,32]; an aliphatic amino acid (Ile, Leu, or Val) or Pro at the third or fourth position; and a lack of negatively charged amino acids within the first 12 residues [33]. However, it is important to note that genes in the HrpX regulon may not be T3SEs; for example, HrpX-regulated proteins from Xoo recently identified by 2D-difference gel electrophoresis (2-DIGE) did not function as T3SEs [34]. The transcription and translocation of HrpX regulon candidates have been examined using several reporter systems, such as calmodulin-dependent adenylate cyclase (Cya) of Bordetella pertussis [22,30,31,35], gusA [36][37][38][39][40] and avirulence proteins (e.g., AvrBs1 and AvrXa10) lacking the T3S signal sequence [24,38,41].
The expression of HrpX is regulated by HrpG, which belongs to the OmpR family of two-component signal transduction response regulators [42,43]. HrpG regulates the expression of hrpX and hrpA operon and also controls the expression of several proteins that function as cell wall degrading enzymes (CWDEs), which are secreted by the type II secretion system (T2SS) [14,44,45]. Recently, a novel hrp regulator, HrpD6, was identified and shown to be regulated by HrpG and HrpX; HrpD6 regulates the expression of hpa2, hpa1, hpaB, hrcC, and hrcT [12]. However, it remains unclear whether Xrps in Xanthomonas are regulated by HrpG or HrpD6.
In this study, bioinformatic and genetic approaches were used to characterize 17 Xrp-coding genes from in silicon data. Different transcriptional profiles of these genes in the wild-type strain Xoc RS105, hrpG (RDhrpG), and hrpX (RDhrpX) mutants [12] were compared. Two Xrp proteins, XOC_3956 and XOC_1550, were identified as new T3SEs in Xoc.

Microarray design
An oligonucleotide microarray was designed at the Shanghai Biotechnology Corporation (Shanghai, China). Each slide contained six arrays, and each array contained approximately 15,000 spots (our probes were represented in triplicate). For Xoc BLS256, the genome sequence was also available from the NCBI database as accession AAQN01000001(GI:94721269). Up to five candidate probes per target (sense orientation) were designed with the Agilent eArray web tool, using temperature-matching methodology, a preferred probe melting temperature of 80uC, no 39bias, and a target length of 60 bp. Shorter probes were extended to 60 bp using the Agilent linker.

RNA isolation and microarray execution
Xoc strain RS105 and the hrpG and hrpX mutants (RDhrpG and RDhrpX, respectively) [12] were cultured overnight in NB broth at 28uC in a shaking incubator and collected the following day via centrifugation at 5,000 rpm. Each strain was washed once in rice suspension cells [49] before resuspension in rice cells at OD 600 = 0.6. Strains were then incubated for 16 h at 60 rpm at 25uC. RNA was extracted from 1 ml of co-cultivated cells using TRIzolH Reagent (Invitrogen, Shanghai, China) as described by the manufacturer. All RNA was quantified using an Eppendorf BioSpectrometer kinetic (Eppendorf, Shanghai, China) and checked for quality using an RNA 6000 Nano Kit and a 2100 Bioanalyzer (Agilent Technologies). Fluorescent labeling of total RNA was performed as described previously [49] using the following array design on a single 6615 k format slide: 1, WT rep 1 (Cy3) and RDhrpG rep 1 (Cy5); 2, RDhrpG rep 2 (Cy3) and WT rep 2 (Cy5); 3, WT rep 3 (Cy3) and RDhrpG rep 3 (Cy5); 4, WT rep 1 (Cy3) and RDhrpX rep 1 (Cy5); 5, RDhrpX rep 2 (Cy3) and WT rep 2 (Cy5); 6, WT rep 3 (Cy3) and RDhrpX rep 3 (Cy5). This design incorporated a dye-swap and balanced labeling of all samples. Levels and efficiencies of labeling were estimated using a spectrophotometer. Microarray hybridization, washing and scanning were performed in the JHI Sequencing and Microarray Facility as described previously [50]. Microarray images were imported into Agilent Feature Extraction (FE) (v.9.5.3) software and aligned with the appropriate array grid template file (021826_D_F_20081029). Intensity data and quality control (QC) metrics were extracted using the recommended FE protocol (GE2-v5_95_Feb07). Entire FE datasets for each array were loaded into GeneSpring (v.7.3) software for further analysis.

Microarray analysis
Data were normalized using default settings for two-channel arrays and transformed to account for dye-swaps. Data from each array were normalized using the Lowess algorithm to minimize differences in dye incorporation efficiency. Unreliable data flagged as absent in all replicate samples by the FE software were discarded. Gene lists with significant change were generated from combined replicate datasets for each pares, RDhrpG/RS105 and RDhrpX/RS105, using volcano plot filtering if the ratio is lower than 0.55 or higher than 1.75 with the P value less than 0.05 (Student's t test).

DNA manipulation and plasmid construction
DNA manipulation was performed following standard procedures [46]. Biparental conjugal transfer of plasmids from E. coli to Xoc was performed as described previously [51]. PCR amplification was performed with primers (Table S2 in File S1) and genomic DNA of Xoc RS105; the genome sequence of Xoc BLS256 was used as a reference (http://cmr.jcvi.org/cgi-bin/CMR/ GenomePage.cgi?org = Xoc). All plasmids constructs were confirmed by restriction enzyme digestion and sequencing.
To construct transcriptional fusions of xrp genes with the promoterless gusA (b-glucuronidase) gene, a 500-bp region upstream of each xrp ORF was PCR-amplified with primer sets pxrpX-F/pxrpX-R (X refers to 17 xrp genes; 1 to 17) ( Table S2 in File S1). The PCR products were fused in-frame with gusA in pUFR034GUS (Table S1 in File S1), resulting in pxrpXGUS (X refers to 17 xrp genes; 1 to 17). Recombinant plasmids were introduced into the wild-type strain RS105, and the hrpG, hrpX and hrpD6 mutants (see reference 12 for RDhrpG, RDhrpX, and RDhrpD6) (Table S1 in File S1) using biparental conjugation as described above.
To generate transcriptional fusions of xrp genes with c-Myc tags, fragments containing native promoters and corresponding xrp ORFs were PCR-amplified with the xrpX-F/xrpX-R primer pairs (X refers to 17 xrp genes, Table S2 in File S1) using RS105 genomic DNA as the template. After sequencing for confirmation, PCR products were cloned in pUFR034Myc in-frame at suitable enzyme sites, generating pXrpXMyc (X refers 1 to 17 xrp genes) ( Table S1 in File S1). The recombinant plasmids were introduced into wild-type RS105 and the hrcV mutant (RDhrcV) by the biparental conjugation method.
Chimeric fusions were also generated between a variant of AvrXa10 lacking 28 amino acids at the N-terminus and xrp3 and xrp5. The truncated form of AvrXa10 was designated Avr-Xa10D28 and has been described previously [26]. The sequences obtained from xrp3 and xrp5 included approximately 500 bp upstream of the translational start site and the first 150 nucleotides. These regions of xrp3 and xrp5 were obtained using primer pairs xrp3-N-F/xrp3-N-R and xrp5-N-F/xrp5-N-R, respectively (Table S2 in File S1). The PCR products were ligated into pBlueAvrXa10D28 at SacI and PstI sites. The constructs were then cloned into pHMI at the SacI site, resulting in pXrp3AvrXa10D28 and pXrp5AvrXa10D28 (Table S1 in File S1). Recombinant plasmids were introduced into Xoo PXO99 A and PDhrcU by electroporation.
YFP was used as a reporter to investigate the subcellular localization of selected T3SEs. The complete ORFs encoding xrp3 and xrp5 without stop codons were PCR-amplified from genomic DNA of Xoc RS105 with xrp3-Y-F/xrp3-Y-R and xrp5-Y-F/xrp5-Y-R, respectively (Table S2 in File S1). The products were cloned inframe with yfp in pA7-YFP [52], resulting in pXrp3YFP and pXrp5YFP (Table S1 in File S1), respectively. These two recombinants were then transferred into Arabidopsis (Ecotype Col-0) mesophyll protoplasts by PEG-calcium fusion as described previously [53].

Quantitative real-time PCR (qRT-PCR)
The expression of selected xrp genes was assayed by qRT-PCR with corresponding primer pairs (Table S2 in File S1). All primers were designed with Beacon Designer 7 software. RNA was extracted from Xanthomonas strains as described previously [47] using TRIzolH Reagent (Takara, Dalian, China) and the manufacturer's recommendations. cDNA synthesis was conducted with AMV random primers purchased from Takara. Prior to synthesis of the first-strand cDNA, total RNAs were digested with RNase-free DNase I (TaKaRa) to remove potential traces of genomic DNAs. qRT-PCR was performed on the Applied Biosystems 7500 qRT-PCR System using SYBR Premix Ex-Taq TM (Takara). PCR conditions included the following parameters: denaturation at 95uC for 30 s; 40 cycles at 95uC, 5 s; and 60uC, 34 s. Experiments were performed at least three times in triplicate. Internal controls included gyrB (XOC_0006, DNA gyrase B subunit) and rpoD (XOC_2329, RNA polymerase sigma factor 70) [28,54].

b-glucuronidase (GUS) activity assays
For GUS activity assays, Xoc strains were preincubated in 5 ml NB broth at 28uC for 16-20 h until the OD 600 = 0.6. An aliquot (100 ml) of this culture was transferred into 5 ml fresh NB broth. Bacterial cells were collected, washed twice, and resuspended in XOM3 to an OD 600 = 2.0. After incubation at 28uC for 6 h, 1 ml of sonic buffer (40 mM Tris-HCl, pH 7.0, 20 mM b-mercaptoethanol, 10 mM EDTA, and 2% Triton X-100) was added into 1 ml of the bacterial culture. This mixture was frozen in liquid nitrogen and then thawed in at 37uC for 5 min. This procedure was repeated five times, and the mixture was centrifuged (12,000 rpm) at 4uC for 15 min. Then, 90 ml 4-methylumbelliferone-b-glucuronide (4-MUG) (Sigma, Shanghai, China) [55] was added into 10 ml supernatant and incubated at 37uC for 1 h. The reaction was terminated by adding 1 ml of 2 M Na 2 CO 3 . GUS activity was measured at 415 nm with the Modulus TM Single Tube Multifunction Tester (YuanPingHao, Beijing, China) [56]. One unit was defined as 1 nM of 4-methyl-umbelliferone (4-MU) produced per min per OD 600 of bacterial cells as described [56].
Three independent experiments were performed, and similar results were obtained.

Mutant construction and complementation studies
To generate non-polar mutations in the 17 xrp genes, two fragments flanking each gene were PCR-amplified from Xoc RS105 genomic DNA with primer pairs xrpXI-F/xrpXI-R and xrpXII-F/xrpXII-R (X refers to the 17 xrp genes; 1 to 17, Table S2 in File S1). After verification by sequence analysis, the two fragments flanking each gene were fused and ligated to the suicide vector pKMS1 [57] at selected restriction sites (Table S2 in File S1), producing plasmids pKMSDxrpX (X refers 1 to 17, Table S1 in File S1). Mutations in the 17 xrp genes were generated by twostep homologous recombination as described previously [57]. Deletions in the xrp genes were confirmed by PCR-amplification with the primer pairs, xrpXI-F/xrpXII-R (Table S2 in File S1); the results confirmed that the PCR products were smaller than those obtained in RS105 (data not shown). For complementation studies, constructs designated pXrpXMyc (X refers 1 to 17, Table  S1 in File S1) were introduced into the corresponding xrp mutants by biparental conjugation. Mutants containing xrp genes in trans were verified by colony-PCR and named CRDxrpX (X refers 1 to 17, Table S1 in File S1).

Bacterial virulence and growth in planta
Rice cultivar Shanyou 63 (two weeks old) was used to evaluate the virulence of Xoc RS105 and derivatives. Bacterial cells were adjusted to 1610 8 cfu/ml and infiltrated into newly-expanded leaves with a needleless syringe at three locations per leaf. Three leaf discs (0.5 cm in diameter) were excised with a cork borer from each infiltrated area. After sterilization in 70% ethanol and 30% hypochlorite, the discs were macerated using a sterile mortar and pestle in 1 ml of distilled water, diluted and plated to determine cfu/cm 2 . Serial dilutions were spotted in triplicate on NA plates with appropriate antibiotics. The plates were incubated at 28uC for 3-4 days until single colonies could be counted. The bacterial population (cfu/cm 2 of leaf area) was then estimated, and the standard deviation was calculated using colony counts from three triplicate spots of three samples obtained at each time point. All rice cultivars were grown in a greenhouse maintained at 25uC with a 12-h photoperiod. Experiments were repeated at least three times.
Near-isogenic lines of rice cultivar IRBB10 were used to assay the pathogenicity of Xoo PXO99 A and derivatives as described previously [47]. Plant responses were scored 14 dpi for lesion lengths. Experiments were repeated at least three times.

Type III secretion assays
The pXrpXMyc plasmids were transformed into the wild-type RS105 and RDhrcV for detection of secreted proteins. Xoc strains were preincubated in NB medium, washed twice, and suspended at OD 600 = 2.0 with sterilized water. An aliquot (40 ml) of the bacterial suspension was inoculated into 1 ml of XOM3 medium (pH 6.0), adjusted to OD 600 = 1.0, and incubated at 28uC for 6 h with the appropriate antibiotics. Supernatant fractions were separated using a 0.22 mm filter, and the supernatant fraction (50 ml) was reduced to 5 ml by vacuum evaporation. Proteins were precipitated with 12.5% trichloroacetic acid at 4uC for 16 h, centrifuged at 30006 g for 15 min, and then washed briefly with acetone and air-dried. The protein pellet was resuspended in SDS buffer containing dithiothreitol (DTT) [58]. Proteins were separated on 10% SDS-PAGE gels and transferred to membranes for immunoblotting using primary antibody anti-c-Myc (Huaan, Hangzhou, China). Primary antibodies were recognized by anti-rabbit secondary antibodies (Huaan) and visualized by autoradiography with the Western-Light chemiluminescence system (Transgene, Beijing, China). Experiments were repeated at least twice and Hpa1 was used as a positive control [59].

Screening of T3SE candidates from Xoc
Genome-wide identification of bacterial virulence genes has been greatly facilitated by the availability of microarrays [60,61]. Considering the critical fact that HrpG and HrpX are two key regulators for pathogenicity determinants in Xanthomonads [14,26,28,29,43,44], the expression profiles of genome widely annotated hypothetical proteins in Xoc RS105 and hrpG and hrpX mutants were only evaluated based on the annotated sequence of Xoc BLS256 (AAQN01000001.1 (GI:94721269)). The three Xoc strains (RS105, RDhrpG, RDhrpX) were co-cultivated with rice cells for 16 h at 25uC, and bacterial RNAs were extracted and hybridized with the BLS256 genechip. The expression levels of hypothetical protein genes in the hrpG and hrpX mutants were compared with the wild-type RS105 and taken as the candidates if the ratio was lower than 0.55 or higher than 1.75 with the P value less than 0.05 (Table S3 in File S1). For a comparison, we checked the expression of hrp genes and known T3SE genes (Table S3 in File S1) and found that the expression patterns of these genes in silicon were consistent with those that were experimentally explored previously [12,23,44,56,61] Totally, there were 247 hypothetical protein genes significantly (P#0.05, t test) altered their expression levels in the hrpG or/and hrpX mutants, comparing to the wild-type RS105 (Table S3 in File S1), indicating that these in silicon data are helpful to find HrpG/HrpX-regulated proteins (Xrp).
To reveal whether there is(are) any T3SE protein(s) in these Xrp candidates, three criteria are considered: (i) candidate genes are homologous to known T3SEs in other phytopathogenic bacteria [16,22]; (ii) expression of candidate genes is HrpX-dependent with the presence of a PIP-box in the promoter [29,38], and (iii) the candidate protein product contains targeting signals at the Ntermini that are indicative of T3S secretion [22,47]. For these, sequences in the xrp promoter regions and in the N-termini of Xrp translational products (identical correspondingly to these in Xoc BLS256 after sequencing confirmation) were analyzed and the protein IDs were changed into Table 1 and Table 2 according to the new version of the genome sequence of Xoc BLS256 (NZ_AAQN01000001.1, GI:353459993). In Xrp1, Xrp3, Xrp4,,Xrp5, Xrp7, Xrp8, Xrp9, Xrp11 and Xrp14, the total number of Ser and Pro residues in the first 50 N-terminal amino acids constitutes more than or equal to eight (Table 1). Xrp1, Xrp3, Xrp4, Xrp5, Xrp6, Xrp8 and Xrp11 contain at least five Ser residues at the N-terminus (Table 1). Promoter analysis showed that there are PIP-box-like sequences upstream of xrp1, xrp2, xrp8, xrp9, xrp13, xrp14, xrp15 and xrp16 ORFs (Table 1). Interestingly, these PIP-box-like sequences, which contained 9-25 nucleotides between TTCGB motifs, differ from the typical consensus PIP-box (TTCGB-N 15 -TTCGB). To have a better understanding, three candidates (Xrp10, Xrp12 and Xrp17) without either of two properties abovementioned were selected from 247 hypothetical protein genes (Table S3 in File S1) as the comparison.
The in silicon data showed that the expression of xrp1, xrp2, xrp9, xrp13, xrp14 was significantly (t test, P#0.05) positively regulated by both HrpG and HrpX. the expression of xrp8 was positively regulated only by HrpX, the expression of xrp3, xrp4, xrp11 and xrp12 was positively regulated only by HrpG, the expression of xrp7 xrp15, and xrp17 was negatively regulated by HrpX, and the expression of xrp5, xrp6 and xrp16 was significantly negative regulated only by HrpG with respect to expression levels in the wild-type RS105 (Fig. 1A, Table S3 in File S1). These results suggest that mutations in hrpG or hrpX may alter the expression of these 17 xrp genes during bacterial infection in rice.

Expression patterns of xrp genes controlled by HrpG, HrpX and HrpD6
To confirm the expression profiles of these 17 xrp genes in RS105, RDhrpG, and RDhrpX (Fig. 1A), we used quantitative realtime PCR (q-RT PCR) to determine whether the expression was HrpG or HrpX-dependent. Primers (Table S2 in File S1) selected for this experiment were based on the genome of BLS256 (NZ_AAQN01000001.1, GI:353459993)(10). Since HrpD6 is a newly identified hrp regulatory factor, we also investigated whether the expression of these 17 xrp genes was altered in the mutant RDhrpD6. Controls included Hpa1, which contains a T3S signal at the N-terminus and the PIP-box promoter [24,59] and XOC_0618, which is homologous to XopX [22]. Hpa1 and XOC_0618 were positively regulated by HrpG and HrpX, but negatively regulated by HrpD6 (Fig. S1 in File S1); this observation is consistent with previous results [12]. As shown in Fig. 1B, the expression of xrp1, xrp2, xrp9, xrp13 and xrp14 was significantly positively regulated by both HrpG and HrpX, which is consistent with our in silicon data (Fig. 1A, Table S3 in File S1); the expression of xrp3 and xrp8 was decreased in the hrpX mutant, relative to the wild-type and hrpG mutant; the expression of xrp4, xrp5, xrp10, xrp11 and xrp12 was decreased in RDhrpX, but increased not only in RDhrpG but also in RDhrpD6; and the expression of xrp6, xrp7, xrp15 xrp16 and xrp17 was significantly increased in the hrpX mutant. The qRT-PCR results (Fig. 1A) were generally consistent with the expression in silicon (Fig. 1). In contrast, the expression of xrp1, xrp2, xrp3, xrp4, xrp5, xrp6, xrp7, xrp10, xrp11, and xrp15 was significantly increased in the hrpD6 mutant (Fig. 1B).
To further investigate the expression of these 17 xrp genes, we fused the putative promoter regions (500 bp upstream of the translational start site) to a promoterless gusA gene. The promoters were PCR-amplified from genomic DNA of strain RS105 using the primers listed in Table S2 in File S1. After sequencing these PCR products from RS105, no obvious difference was found in corresponding regions in BLS256 genome (data not shown). The pxrp-gusA fusions were then transferred into the wild-type RS105, the hrpG, hrpX and hrpD6 mutants, respectively, then incubated in XOM3 (a hrp-inducing medium) [49] at 28uC for 6 h. The GUS expression patterns (Fig. 1C) detected by these 17 xrp promoters was similar to the regulation observed by genechip and qRT-PCR analysis (Figs. 1A, B). However, there were no obvious differences in GUS activities expressed from the xrp4, xrp13, xrp14 and xrp16 putative promoters in RS105, and the hrpG, hrpX and hrpD6 mutants (Fig. 1C), possibly because that these four genes localize within adjacent operons where the tested promoters are not real for GUS activity detection (data not shown). xrp1, xrp2, xrp5, xrp8 and xrp14 genes are required for Xoc virulence to rice The expression patterns of these 17 xrp genes in the hrpG, hrpX and hrpD6 mutant backgrounds prompted us to determine whether they are involved in Xoc virulence. Each xrp gene was PCRamplified using primers derived from the BLS256 genome (Table  S2 in File S1). Sequence and BLAST analysis of xrp PCR products from RS105 showed no differences in these xrp genes with the corresponding ORFs in BLS256 ( Table 2). All these Xrp proteins of RS105 showed 100% identity to the homologs of BLS256 ( Table 2). Homologs of Xrp2 were not identified in X. axonopodis pv. citri strain 306 (Xac 306) and X. campestris pv. campestris (Xcc) ATCC33913. Xrp6 was only identified in Xoc BLS256 and Xoo PXO99 A ; Xrp8 had no orthologs in Xoo strains but in other three Xanthomonas species, and Xrp12 was not present in Xoo PXO99 A . Xrp14 was identified in Xoc BLS256, but only had lower similarity (29%) in Xac 306; Xrp13 and Xrp15 had no homologous counterparts in Xcc; and Xrp17 was a putative ORF upstream of HrpF only in Xoc BLS256 genome (designed HrpFB herein) ( Table 2). It should be mentioned here that the ORFs and putative products for xrp1, xrp7, xrp12, xrp13, xrp14, xrp15 and xrp16 were re-annotated in the new version of Xoc BLS256 genome sequence (NZ_AAQN01000001.1, GI:353459993) ( Table 2).
In order to explore the putative role of the xrp genes in virulence, we generated in-frame deletion mutants of the 17 xrp genes (Table S1 in File S1) in Xoc RS105 as described previously [57]. The left and right fragments flanking the xrp genes were PCR-amplified with primers listed in Table S2 in File S1 using RS105 genomic DNA as the template, fused, transferred into the suicide vector pKMS1 and generated deletion mutants correspondingly (Table S1 in File S1). These 17 nonpolar deletion mutants were inoculated into leaves of rice cv. Shanyou 63, which is susceptible to Xoc. The lesion lengths generated by xrp1, xrp2, xrp5, xrp8 and xrp14 mutants were significantly shorter than those produced by Xoc RS105 (Fig. 2A), and these five mutants were impaired in their ability to grow in rice leaves compared to the wild-type (Fig. 2B-G). Virulence and bacterial growth in planta were restored to wild-type levels by the introduction of corresponding xrp genes in trans (Fig. 2B-G). The above data indicate that xrp1, xrp2, xrp5, xrp8 and xrp14 genes are required by Xoc for bacterial virulence and growth in rice.

Xrp3 and Xrp5 are novel T3SEs
In the complementation studies mentioned above, a c-Myc tag was fused in frame at the C-termini of the 17 Xrp proteins (see Table S1 in File S1), which facilitated subsequent expression studies. The c-Myc-tagged constructs were introduced into Xoc RS105 and RDhrcV [62] and incubated in hrp-inducing medium XOM3 [48] for 6 h. The supernatants (SN) and total extracts (TE) of bacterial cells were used to investigate whether the Xrps were expressed by Western blotting using a c-Myc specific polyclonal antibody (Huaan, Hangzhou, China). With the exception of Xrp4, Xrp6, Xrp9, Xrp10 and Xrp13, the other 12 Xrp proteins were detectable in TEs of both RS105 and RDhrcV (representative data are shown in Fig. 3). However, Xrp3 and Xrp5, like Hpa1, were detected in SNs of the wild-type, but not in the T3S mutant (Fig. 3). Xrp1 and Xrp2 were detected in the SN fractions of both the wildtype and the T3S mutant, whereas Xrp14 were not (Fig. 3). These data indicate that Xrp3 and Xrp5 are secreted via the T3S, but Xrp1 and Xrp2 may be secreted via other systems. Xanthomonas T3SEs containing up to 20% Ser/Pro residues or at least five Ser residues in their N-terminal sequences are deemed to contain a T3S signal [22,33]. One strategy for determination of T3SE secretion via the T3S is to fuse the N-terminus of a candidate effector with an Avr protein that lacks the secretion signal, and investigate whether the fused effector can be translocated into plant cells [24,38]. The first 50 N-terminal amino acid residues of Xrp3 and Xrp5 contain a total of 10 and 8 Ser/Pro residues, respectively (Fig. 4A, B), indicating that Xrp3 and Xrp5 may be translocated into host cells. To further investigate this possibility, the N-terminal 50 amino acids of Xrp3 and Xrp5 were fused with a truncated AvrXa10 that lacked 28 amino acid residues at the N-terminus (avrXa10D). The chimeric proteins Xrp3avrXa10D and Xrp5avrXa10D were constructed using native xrp promoters (see Materials and Methods), resulting in pPXrp3avrXa10D and pPXrp5avrXa10D, respectively (Table S1 in File S1, Fig. 4A, B). The wild-type RS105 harbouring avrXa10 does not trigger an HR in rice line IRBB10, which carries the cognate R gene Xa10 [24]; thus, we utilized Xoo strain PXO99 A to investigate whether the fused proteins can trigger an HR in IRBB10. The T3S mutant PDhrcU was used as a negative control. PXO99 A containing pPXrp3avrXa10D or pPXrp5avrXa10D produced an HR in IRBB10, as did PXO99 A (pavrXa10) (Fig. 4C). However, HR was not elicited by PXO99 A containing pavrXa10D or the empty vector pUFR034 (Fig. 4C). As predicted, the T3S mutant PDhrcU did not elicit symptoms in cultivar IRBB10 (Fig. 4C). We then used immunoblotting to determine whether the translational fusions present in pPXrp3avr-Xa10D and pPXrp5avrXa10D were expressed in the tested strains. As expected, the chimeric proteins PXrp3avrXa10D and PXrp5avrXa10, like the wild-type AvrXa10 and the N-terminal truncated AvrXa10D, were detectable in the TEs of both wild-type PXO99 A and PDhrcU, but not in SN fraction of PDhrcU (Fig. 4D). The above data indicate that the N-terminal portions of Xrp3 and Xrp5 enable the N-terminal truncated AvrXa10 to be secreted through the T3S and translocated into rice cells for HR induction.

Subcellular localization of Xrp3 and Xrp5
The localization of Xrp3 and Xrp5 was examined by utilizing YFP-tagged Xrp3 and Xrp5 proteins. The protocol (see Materials and Methods) utilizes mesophyll protoplasts of Arabidopsis and PEG-calcium-mediated transfection to deliver pXrp3-YFP and pXrp5-YFP (Table S1 in File S1) into plant cells; GUS-NLS-YFP and YFP were used as a reference. Fluorescence confocal microscopy indicated that YFP-tagged Xrp3 proteins are localized throughout the cell, but Xrp5-YFP is targeted to the plasma membrane and cytoplasm. The latter result was clearly different from the YFP control, which was partially retained in the nucleus (Fig. 5). In general, our results suggest that Xrp3-YFP is localized throughout the cell (cytoplasm, nucleus, plasma membrane) and Xrp5-YFP preferentially targets the cytoplasm.

Discussion
In this report, two new T3SEs, Xrp3 (XOC_3956) and Xrp5 (XOC_1550), were identified and shown to be HrpX-dependent based on transcriptional data in silicon (Fig. 1A). These two newlyidentified T3SEs can be added to the extensive list of Xanthomonas T3SEs at http://www.Xanthomonas.org/t3e.html [16]. Interestingly, xrp3 and xrp5 are highly conserved in Xanthomonads rather than other plant pathogenic bacteria whose genomes have been completely sequenced (Table 2). Xrp5, but not Xrp3, was important for a full level of virulence in rice (Fig. 2). This is consistent with a previous study that many NTALEs but XopZ in Xoo do not impact bacterial virulence in rice [63].
The in silicon and qRT-PCR data (Figs. 1A, B) consistently showed that the expression of 17 xrp genes was regulated by HrpX and/or HrpG at the transcriptional level. However, GUS activity detection showed that the expression of xrp4 (XOC_3955) and xrp16 (XOC_2828) did not display any obvious differences in regulation in WT and mutant strains (Fig. 1C), being due to that they are members of their adjacent operons (data not shown). Giving that the expression of xrp1 (XOC_1601), xrp2 (XOC_4584), xrp3 (XOC_3956), xrp4 (XOC_3955), xrp5 (XOC_1550), xrp6 (XOC_3440), xrp7 (XOC_4583), xrp10 (XOC_0560), xrp11 (XOC_3130) and xrp15 (XOC_2829) was negatively regulated by HrpD6 but either positively regulated by HrpX and/or HrpG or negatively regulated by HrpX (Fig. 1B), HrpG-, HrpX-and HrpD6-mediated regulation in xanthomonads is very complicated to follow the concept that hrpX expression is controlled by HrpG [14,44] and HrpX regulates the expression of hrpD6 [12]. In general, one criterion for identification of HrpX regulon genes is the PIP-box [26,27,43,64]. However, PIP-box-like sequences are absent in the xrp3 and xrp5 upstream regions (Table 1). This is consistent with reports that the expression of some HrpXregulated genes that lack PIP-box promoters is modulated directly by HrpX and indirectly by HrpG [22]. For example, there is no PIP-box in the promoter regions of hpaJ, XCV_0869 or XCV_3406 genes, but their expression is HrpX-dependent [26]. Thus, there may be an alternative regulatory system in Xanthomonas where HrpX activates genes independently via an unknown regulator(s), like HrpD6.
It should be emphasized that the location of xrp6 (XOC_3440) and xrp17 (hrpFB) is within the hrp-hrc-hpa cluster. xrp6 is a putative ORF between hrpG and hrpX, and xrp17 is hrpFB upstream of hrpF. These two ORFs have not been annotated in other Xanthomonas genomes (Table 1). Our data suggest that xrp6 and xrp17 are not involved in Xoc virulence ( Fig. 2A); however, the expression of these two ORFs was negatively regulated by HrpX, but not affected by HrpG (Fig. 1B), and the gene product of xrp17 was not detectable in the SN of the WT and mutant strains (data not shown). It remains unclear whether Xrp6 or Xrp17 function in the regulation of the hrp-hrc-hpa cluster.
Xrp2 (XOC_4584) is conserved in Xoc, Xoo and Xcv, but not in Xcc or Xac (Table 2); database searches provided no functional clues regarding Xrp2 function. The expression of Xrp2 is HrpGand HrpX-dependent, but upregulated by HrpD6, like the expression of Xrp1 (Fig. 1). Thus, in addition to regulating hrp gene expression [12], HrpD6 may also regulate other virulence factors that are not secreted via the T3S. We found that Xrp8 (XOC_2462) is highly conserved in Xoc, Xcv, Xac and Xcc, but has no homolog in Xoo strains PXO99 A , MAFF311018 or KACC10331 (Table 2); thus, this virulence factor (Fig. 2F) may potentially be required for full virulence of Xcv, Xac and Xcc in plants.
Xrp14 (XOC_0084) is present in Xoc BLS256 and Xac 306 ( Table 2) and encodes a putative gene belonging to the cytochrome P450 family [9]. The P450 family proteins use heme iron to oxidize molecules, often making them more water-soluble [72]. It is interesting to note that there are three Ser and seven Pro residues at the N-terminus of Xrp14 (Table 1), indicating that Xrp14 is a potential T3SE. However, the protein is undetectable in the SN of the wild-type strain and T3S mutant (Fig. 3), suggesting that the involvement of Xrp14 in Xoc virulence is worthy of further investigation.
In Xanthomonas spp., T3SEs have been identified based on sequence similarities identified from genome sequence data [16,22], avirulence reporter fusion assays [38], Cya-fusion approaches [22] and functional assays of T3S-dependent expression and secretion [73]. It has been demonstrated that the fusion expression of an N-terminally truncated avr gene (avrXa10) with the first 50 N-terminal amino acids of a T3SE is an important tool for identifying Xoc T3SEs by utilizing the Xoo-rice pathosystem [24,47,74]. However, it may be important to utilize the highly sensitive Cya assay to identify Xoc T3SEs [22,33,75]. In R. solanacearum, 72 T3SEs have been characterized [76]. To date, approximately 26 NTALEs and 29 TALEs have been confirmed  or predicted in Xoc BLS256 (http://www.xanthomonas.org/t3e. html), excluding HrpE3 [74], and Xrp3 (XOC_3956) and Xrp5 (XOC_1550). Considering the long period of co-evolution of Xoc and rice, we speculate that the number of T3SEs in Xoc or Xoo may possibly be more than the currently identified. Hence, identifying T3SEs is still an important endeavor in Xoc where HrpX-regulated proteins could potentially function as T3SEs.

Supporting Information
File S1 File containing supporting Figure S1 and Tables S1-S3.  The subcellular localization of Xrp3 and Xrp5 was examined by fusing these proteins with yellow fluorescence protein (YFP) and evaluating their expression in Arabidopsis mesophyll cells. Images were generated by fluorescence at 525-550 nm (YFP, yellow) and 610-700 nm (chloroplasts, red) using excitation at 514 nm. Expression was driven by the CaMV 35S promoter. PEG-calcium-mediated transfection was used to deliver DNA into protoplasts, and photos were taken 12 h after transformation. Controls included YFP, which localizes in both cytoplasm and nucleus, and GUS-NLS-YFP, which localizes in the nucleus. Bars correspond to 10 mm. The experiment procedure was performed as described previously [54] and repeated three times with similar results. doi:10.1371/journal.pone.0093205.g005