A key antisense sRNA modulates the oxidative stress response and virulence in Xanthomonas oryzae pv. oryzicola

Pathogens integrate multiple environmental signals to navigate the host and control the expression of virulence genes. In this process, small regulatory noncoding RNAs (sRNAs) may function in gene expression as post-transcriptional regulators. In this study, the sRNA Xonc3711 functioned in the response of the rice pathogen, Xanthomonas oryzae pv. oryzicola (Xoc), to oxidative stress. Xonc3711 repressed production of the DNA-binding protein Xoc_3982 by binding to the xoc_3982 mRNA within the coding region. Mutational analysis showed that regulation required an antisense interaction between Xonc3711 and xoc_3982 mRNA, and RNase E was needed for degradation of the xoc_3982 transcript. Deletion of Xonc3711 resulted in a lower tolerance to oxidative stress due to the repression of flagella-associated genes and reduced biofilm formation. Furthermore, ChIP-seq and electrophoretic mobility shift assays showed that Xoc_3982 repressed the transcription of effector xopC2, which contributes to virulence in Xoc BLS256. This study describes how sRNA Xonc3711 modulates multiple traits in Xoc via signals perceived from the external environment.

Introduction Bacterial pathogens can adapt to stressful conditions by altering the activity and number of transcriptional regulators [1][2][3]. For example, regulatory proteins may be modulated at the transcriptional level or subjected to post-translational modifications such as phosphorylation and glycosylation [4,5]. In addition, global regulators of gene expression can be modulated by small regulatory RNA (sRNA) molecules that target mRNA at the post-transcriptional level via base pairing; this ultimately controls gene expression of the target and can impact virulence [6]. In prokaryotes, the sRNAs that base pair with target mRNAs can be further assigned into two subgroups: cis-and trans-encoded sRNAs [7]. The cis-encoded RNAs are transcribed from the complementary strands of their target; this group is often encoded by phages, plasmids and transposons and includes sRNAs that are classified as riboswitches [7]. The transencoded sRNAs have been extensively studied in prokaryotes; these sRNAs are transcribed from genomic loci that are physically separate from their target genes. The trans-encoded sRNAs generally mediate translation or stability of target mRNAs by partial or discontinuous base pairing [7]. sRNAs can regulate target genes positively or negatively. For example, positive regulation of the target gene may occur when sRNAs base pair with the target mRNA, which can unmask the ribosome binding site (RBS) in the target and promote its translation. Alternatively, sRNAs can negatively regulate their targets by inhibiting translation and/or stimulating degradation via ribonuclease RNase E [8]. The interaction of sRNA with target mRNA generally requires the RNA chaperone Hfq, which binds sRNAs, facilitates sRNA-mRNA base pairing, and directly binds and regulates translation of certain mRNAs [9].
One of the earliest cellular reactions to pathogen invasion and recognition is the generation of reactive oxygen species (ROS) by the host; this includes the superoxide anion (O 2 -) and its dismutation product, hydrogen peroxide (H 2 O 2 ) [10]. sRNAs can regulate pathogen metabolism by targeting a wide range of virulence factors and stress-response proteins to evade immune defenses and colonize their host. Bacterial sRNAs play major roles in stress tolerance both inside and outside the host cell and promote survival during suboptimal conditions [11,12]. For example, the sRNA RsaC modulates the oxidative stress response of Staphylococcus aureus during manganese starvation by repressing the translation of the Mn-containing enzyme SodA [13]. The sRNA DicF promotes the expression of genes in the type III secretion system (T3SS) in Escherichia coli O157:H7 under oxygen-limited conditions [14]. The sRNA OxyS integrates the oxidative stress response with other cellular responses to help protect E. coli from oxidative damage [15]. The gram-negative plant pathogen, Xanthomonas oryzae pv. oryzicola (Xoc), causes bacterial leaf streak in rice and is an important organism for studying plant-microbe interactions. Many regulatory genes have been characterized in Xoc, especially genes mediating pathogenicity and recognition of host plants. However, relatively few studies have documented the importance of sRNAs and sRNA-mediated regulation in Xanthomonas spp. In X. campestris pv. campestris, transcription of sRNA-Xcc1 was shown to be modulated by the T3SS regulators, HrpG and HrpX, indicating that sRNA-Xcc1 may have a regulatory role in virulence [16]. In the related pathogen X. campestris pv. vesicatoria, sRNA sX13 showed potential regulatory roles in motility and transcriptional regulation of virulence genes [17]. In X. oryzae pv. oryzae (Xoo), which is closely related to Xoc, a recent study identified sRNAs trans217 and trans3287 as virulence-associated sRNAs that are required for pathogenicity in susceptible rice plants and for the elicitation of the hypersensitive response in nonhost plants. The authors suggested that these sRNAs directly regulate the T3SS in Xoo [18]. In a prior study [19], eight sRNAs were functionally characterized in Xoo; among these, sRNA-Xoo1 was of special interest because it was conserved in other Xanthomonas spp., and its expression was Hfq-dependent. Analysis of a sRNA-Xoo1 mutant revealed down-regulated levels of superoxide dismutase, which suggests a potential regulatory role in oxidative stress.
Our lab is interested in the role of post-transcriptional RNA regulation and editing in Xoc, especially with respect to pathogenicity, motility, biofilm formation and adaptation to oxidative stress [20]. This study focuses on sRNA Xonc3711, which is the Xoc homolog of sRNA-Xoo1; as mentioned above, sRNA-Xoo1 was responsive to oxidative stress [19]. In the current study, we show that Xonc3711 plays an extensive role in modulating Xoc transcription during oxidative stress and biosynthesis of flagella. sRNA Xonc3711 interacts with the mRNA of xoc_3982, which encodes a DNA-binding protein. Furthermore, Xoc_3982 binds to the promoter region of the T3SS effector xopC2 to modulate the virulence of Xoc BLS256. Our results confirm a role for sRNA Xonc3711 in regulating multiple systems in Xoc.

Results
sRNA Xonc3711 targets xoc_3982 mRNA Liang et al. previously reported that the small RNA designated sRNA-Xoo1 was conserved in Xoc strain BLS256 [19]. The homologue of sRNA-Xoo1 in Xoc BL526 maps adjacent to Xoc_3711, which encodes a hypothetical protein (S1B Fig). Due to the proximity of the sRNA to Xoc_3711, it was named Xonc3711, with the 'nc' indicating that it is non-coding RNA. In preliminary experiments, the expression of xonc3711 was significantly upregulated in the presence of 0.1 mM H 2 0 2 (S1C Fig), indicating a potential role in oxidative stress. The secondary structure of Xonc3711 was predicted using software available at https://sfold.wadsworth.org (S1A Fig).
Previous results with sRNA-Xoo1 indicated that expression or stability of this small RNA was dependent on the RNA chaperone, Hfq [19]. Thus, we used electrophoretic mobility shift assays (EMSA) to evaluate whether Xonc3711 and Hfq interacted in vitro. EMSA clearly indicated a strong interaction between biotinylated Xonc3711 and Hfq ( Fig 1A). To evaluate whether Hfq impacted Xonc3711 transcription, expression was compared in the wild-type BL526 (WT), a hfq deletion mutant (Δhfq), and a Xonc3711 deletion mutant (ΔXonc3711) ( Fig  1B). There was a substantial decrease in Xonc3711 transcription in the Δhfq mutant, indicating that Hfq has a role in the expression of the sRNA Xonc3711.
In silico searches were performed to identify Xonc3711 targets with the CopraRNA algorithm using Xonc3711 sequence as the query and the Xoc BLS256 genome as the target [21]. Using this approach, Xonc3711 was predicted to target Xoc_3982, a putative DNA-binding protein. To evaluate whether Xonc3711 and Xoc_3982 interact, the expression of xoc_3982 was assessed in Xoc BL256 (WT), a strain overexpressing Xonc3711 (Xonc3711 OE ), and the ΔXonc3711 mutant ( Fig 1C). Expression of xoc_3982 in the Xonc3711 OE strain was much lower than expression in the ΔXonc3711 mutant, which suggests that Xonc3711 may promote xoc_3982 mRNA degradation. Expression of xoc_3982 was then compared in WT and ΔXonc3711 mutant strains with and without overexpression of Xonc3711; a deletion mutant in xoc_3982 was included as a control. Western blot analysis showed that Xoc_3982 protein levels were elevated in ΔXonc3711, and overexpression of Xonc3711 caused a reduction in Xoc_3982 protein levels ( Fig 1D). Northern blot results correlated with the western analyses and showed elevated expression of xoc_3982 transcripts in the ΔXonc3711 mutant ( Fig 1D).

Post-transcriptional regulation of xoc_3982
Hfq-dependent sRNAs activate or repress mRNA targets by several methods [6]. One regulatory mechanism includes base-pairing between the sRNA with the coding sequence (CDS) of the target mRNA, which inhibits translation [22]. We predicted that nucleotides 14-59 of the xoc_3982 mRNA target sequence would be complementary with the Xonc3711 seed region (Fig 2A). To test this hypothesis, the start codon of xoc_3982 (region +3 relative to the GUG; Fig 2B) was translationally fused to green fluorescent protein (GFP), resulting in construct X +3 (Fig 2B). Xonc3711 failed to regulate the X+3 reporter since both RNA and Xoc_3982 protein levels remained unchanged with this fusion (Fig 2C, lanes 1 and 2). When Xonc3711 was paired with the X+1242 reporter fusion, RNA levels remained unchanged, but Xoc3982 protein levels were repressed relative to the controls (lanes 3 and 4); this result supported our prediction that Xonc3711 targeted a region within the xoc_3982 CDS. Furthermore, biotinylated Xonc3711 interacted with full-length xoc_3982 mRNA in gel shift assays (Fig 2D, lanes 2 and Northern blot analysis of Xonc3711 expression in wild-type Xoc BLS256, ΔHfq, ΔXonc3711 and ΔXoc_3982 strains grown to OD 600 = 1.0 in NB. 5S rRNA was used as a loading control, and Image J was used to calculate expression levels. The intensity of the band in the first lane (WT) was normalized to a value of 100. (c) qRT-PCR analysis of xoc_3982 expression in Xoc BLS256 overexpressing Xonc3711 (Xonc3711 OE ), wild-type BLS256 (WT) and the ΔXonc3711 mutant. Expression levels of target genes were calculated relative to rpoD using the ΔΔCT method, where CT is the threshold cycle. Four independent biological replicates were carried out in this study (Wilcoxon-Mann-Whitney test). (d) Northern (upper two panels) and western (lower two panels) blot analysis of Xoc_3982 mRNA and protein levels, respectively, in BLS256 (WT), ΔXonc3711, and Δ3982 strains. In lanes labeled with (+), Xonc3711 was overexpressed from the pHM1::Xonc3711 construct. Expression of 5S rRNA and levels of RNAP were used as loading controls for northern and western blots, respectively. Values above each band represent band intensity and were calculated using Image J software. Band intensity in the first lane was normalized as 100. ND, not detected. https://doi.org/10.1371/journal.ppat.1009762.g001

PLOS PATHOGENS
Antisense sRNA modulates the virulence in Xanthomonas oryzae pv. oryzicola 3). These findings indicate that xoc_3982 is the target of sRNA Xonc3711, which regulates xoc_3982 mRNA after transcription by base pairing with the xoc_3982 CDS.

Short CDS pairing is essential for xoc_3982 repression
We validated the Xonc3711-Xoc3982 interaction in vivo with compensatory point mutations. In mutant Xonc3711 � , nucleotides UGC were mutated to CAA, whereas nucleotides ACG were mutated to GTT in xoc_3982 � (Fig 2A). As predicted, Xonc3711 repressed xoc_3982 expression relative to the deletion mutant (Fig 3, lanes 1 and 2); however, Xonc3711 � was impaired in its ability to repress xoc_3982 relative to the WT (Fig 3A, lane 3). Expression of xoc_3982 � was only slightly reduced in the wild-type containing Xonc3711 (Fig 3A, lane 3); however, a high level of xoc_3982 repression was observed when the compensatory mutations in Xonc3711 � and xoc_3982 � interacted ( Fig 3A, lane 6). Collectively, these results suggest that Xonc3711 pairs with the xoc_3982 mRNA CDS to inhibit xoc_3982 expression.

RNase E is required for Xonc3711-dependent degradation of xoc_3982
Ribonuclease RNase E is a critical enzyme in sRNA processing and turnover [23]. To better understand the sRNA-dependent degradation of xoc_3982 mRNA, we examined the role of

PLOS PATHOGENS
RNase E in Xonc3711-xoc_3982 decay. The C-terminus of RNase E forms a scaffold and is involved in RNA degradation [24]; thus a C-terminal deletion in RNase E was constructed in strain BLS256 and was designated ΔRNaseEC (S1 Table). A second mutation was generated in this mutant by deleting Xonc3711, resulting in the double mutant ΔRNaseECΔXonc3711. The ΔRNaseEC strain showed significant upregulation in Xoc_3982::GFP protein levels when compared with the WT (Fig 3B, lanes 1 and 2), and Xoc_3982::GFP protein levels were only slightly higher in the double mutant strain (lane 4). These results indicate that RNase E contributes to the Xonc3711-mediated degradation of xoc_3982.

Xonc3711 mutant shows decreased tolerance to oxidative stress
Xonc3711 expression was measured in Xoc BLS256 at 0, 7, 15, and 45 min after exposure to 0.1 mM H 2 O 2 (S1C Fig). Xonc3711 transcript levels were highest at the 15 min time point, indicating that Xonc3711 expression was induced by oxidative stress. To further investigate the potential role of Xonc3711 in oxidative stress, growth of selected strains was compared in the presence and absence of 0.1 mM H 2 O 2 in NB medium (Fig 4). Strains grown in NB without H 2 O 2 showed similar growth patterns ( Fig 4A); however, a delayed lag phase of approximately 8 h was observed in the ΔXonc3711 and ΔXoc_3982 strains grown in NB supplemented with 0.1 mM H 2 O 2 when compared with the WT (Fig 4B). Pairwise comparisons of OD values for each strain and growth condition were analyzed using the Kolmogorov-Smirnov test against the values obtained for the WT. Strain Xonc3711 OE showed a high tolerance to 0.1 mM H 2 O 2 (P < 0.01), whereas ΔXonc3711 showed a significantly reduced tolerance to oxidative stress relative to the WT (P < 0.01, Fig 4B). The ΔXoc_3982 mutant was also impaired in oxidative stress tolerance relative to the WT (P < 0.01, Fig 4B). These results suggest that the sRNA Xonc3711 interacts with the DNA-binding protein Xoc_3982 to help Xoc BLS256 adapt to oxidative stress. Upper three panels show northern blots using Xonc3711 or Xonc3711 � , xoc_3982, and 5S rRNA as probes. The lower two panels show Xoc_3982 GFP or Xoc_3982 �GFP protein levels as determined by immunoblotting with mouse anti-GFP antisera; RNAP was used as a loading control. (b) RNase E is essential for xoc_3982 repression by Xonc3711. Upper two panels show northern blot analysis of Xonc3711 expression in the ΔRNaseEC mutant, the wild-type BL256, and the ΔXonc3711 and ΔRNaseECΔXonc3711 mutants. The lower two panels show western blot analysis of Xoc_3982 GFP production. GFP fusion proteins were detected using anti-GFP antisera. Values above each band represent band intensity and were calculated using Image J software. Band intensity in the first lane was normalized as 100. ND, not detected. https://doi.org/10.1371/journal.ppat.1009762.g003

Xonc3711 impacts flagella structure and reduces biofilm formation
RNA-seq was used to compare WT and ΔXonc3711 to further understand the involvement of sRNA Xonc3711 in oxidative stress tolerance. Prior to comparing RNA-seq profiles, reproducibility was evaluated in two replicate experiments using pairwise linear correlation analysis. The correlation coefficients (r) between the two replicate experiments were 0.999 and 0.997, indicating reproducibility of the RNA-seq data under the experimental conditions. Based on a stringent FDR (<0.01) as a cutoff, a large number of genes were downregulated in ΔXonc3711, including genes involved in flagella assembly, basal body formation, flagella motor, and T3SSrelated genes (S2 Fig). Multiple genes involved in flagella synthesis and assembly, including fliC, fliF, fliM, flgA, flhA, and flhB, were downregulated in ΔXonc3711 as compared to the WT ( Fig 4D). Interestingly, the expression of fliC, which encodes the flagellar filament structural protein, was approximately 8-fold lower in ΔXonc3711 as compared to the WT (Fig 4D). In vitro growth curves revealed that the fliC mutant was less tolerant to H 2 O 2 than the wild-type, and growth of the fliC mutant was similar to ΔXonc3711 during oxidative stress (Fig 4B). These results indicated that Xonc3711 has an impact on the structure of flagella. The effect of Xonc3711 on flagella was assessed by comparing the ultrastructure of selected strains via highresolution transmission electron microscopy (TEM). The mutants ΔXonc3711 and ΔXoc_3982 produced fewer flagella than the WT (Fig 5A, 5B and 5D). Ten fields of view were randomly selected from the WT and ΔXonc3711 and used to compile statistical differences in flagellar length (Fig 5A and 5B). The results showed that flagella in ΔXonc3711 were

PLOS PATHOGENS
significantly shorter than the WT (Fig 5E). A fliC deletion mutation (ΔfliC) was used as a control and was devoid of flagella as expected (Fig 5C). Interestingly, the phenotype of the Xoc_3982 mutant with respect to flagella was similar to ΔXonc3711; in other words, the ΔXoc_3982 mutant produced fewer, shorter flagella than the WT (Fig 5D and 5E). These results suggest that mutations in both Xonc3711 and xoc_3982 impact flagella synthesis and structure.
Previous studies have demonstrated that flagella-driven motility facilitates the formation of biofilms [25], which contribute to oxidative stress tolerance [26,27]. Our results indicate that Xonc3711 contributes to both oxidative stress and motility; thus, we measured biofilm formation using a confocal laser scanning microscope (CLSM) and 3D serial layer scanning. Mutant ΔXonc3711 was impaired in its ability to adhere to glass surfaces and showed reduced fluorescence when compared to the WT (Fig 6 and S1 Video); these results confirm a relationship between sRNA Xonc3711 and Xoc motility and biofilm formation.

Xonc3711 overexpression contributes to virulence
Bacterial biofilms are generally more resistant to antimicrobial agents and host defense systems than individual cells; furthermore, bacterial biofilms may exhibit stronger virulence than cells in a planktonic state [28]. To evaluate the potential contribution of Xonc3711 to virulence, leaves of six-week-old rice cv. Yuanfengzao were inoculated with the WT, ΔXonc3711, ΔXoc_3982, and Xonc3711 OE (Fig 7A). At 14 d post-inoculation, lesions induced by Xon-c3711 OE were significantly larger than those induced by the WT and ΔXonc3711 (Fig 7B). Interestingly, the ΔXoc_3982 mutant showed elevated virulence and produced slightly larger lesions than the WT and ΔXonc3711. Thus, our results suggest that Xonc3711 contributes to virulence in in Xoc BLS256.

Xoc_3982 directly regulates the effector encoded by xopC2
The Xoc_3982 protein was analyzed at the NCBI Conserved Domain Database (https://www. ncbi.nlm.nih.gov/cdd), and the results indicated that Xoc_3982 was a potential DNA-binding protein with relatedness to DNA modification/repair proteins in the radical SAM family (S3 Fig). The xoc_3982::GFP fusion X+3 (pKMS1::X+3, S1 Table) was introduced into Xoc BLS256 and used in a ChIP-seq assay to identify genes regulated by Xoc_3982. The results showed that potential targets of Xoc_3982 included hemF, xdhC, xpsE, xopC2, and mutM; these genes contained a conserved sequence, 5'-CGCTTTT-3', which was identified by MEME analysis as a putative Xoc_3982 binding site (Fig 8A). We were particularly interested in xopC2, which encodes a T3SS effector that has been identified in a number of xanthomonads, including Xoc BLS256 [29][30][31][32] and was shown to function in the virulence of X. axonopodis pv. punicae [33]. RNA-seq data indicated that the expression of xopC2 was downregulated in the ΔXonc3711 mutant as compared to the wild-type BL526. The putative Xoc_3982 binding site was located in the xopC2 promoter at -334 to -327 with respect to the translational start site. EMSA

PLOS PATHOGENS
confirmed that Xoc_3982 interacted with the xopC2 promoter, and the interaction was disrupted when the xopC2 promoter was mutated (Fig 8B). The relative expression of xopC2 was significantly higher in the ΔXoc_3982 mutant than the WT, which suggests that Xoc_3982 is a negative regulator of xopC2 (Fig 8C). Furthermore, the ΔXopC2 strain was downregulated in virulence when compared to the WT (Fig 7), which indicated that xopC2 contribute to the virulence in Xoc BLS256.

Discussion
In this report, sRNA Xonc3711 was shown to control expression of xoc_3982, which encodes a DNA-binding protein in Xoc BLS256. Xoc_3982 repressed the expression of the T3SS effector encoded by xopC2, which suggests that Xoc_3982 functions as a transcriptional repressor. It is important to note that sRNAs can positively or negatively modulate transcriptional regulators [34]; for example, the sRNAs CsrB and CsrC in E. coli sequester the translational repressor CsrA, which impacts biofilm formation [35]. Although the precise mechanisms are unclear, sRNA Xonc3711 modulates multiple traits in Xoc including the formation of flagella and biofilms; this suggests that Xonc3711 regulates genes that interact with the external environment.
Target sites of small RNAs are often present in the 5' UTR of the target gene; however, exceptions exist and bacterial sRNAs have been identified that lack obvious binding sites in the 5' UTR of the target gene [36,37]. For example, the Salmonella typhimurium sRNA MicC targets the ompD mRNA within its CDS [38]. Similarly, we show that Xonc3711 targets the xoc_3982 mRNA within the CDS (Fig 2B and 2C). Mutational analysis showed that the regulation of xoc_3982 is direct and requires an antisense interaction between Xonc3711 and xoc_3982 mRNA (Fig 2A); this was confirmed by EMSA (Fig 2D). Another important feature of sRNA Xonc3711 is an A/U-rich motif that could bind the RNA chaperone Hfq for

PLOS PATHOGENS
stabilization and base-pairing [39]. Although the precise nucleotides in Xonc3711 that interact with Hfq were not identified in this study, Xonc3711 and Hfq interacted in gel shift assays (Fig 1A).
Seed-borne pathogens like Xoc are exposed to reactive oxygen in the natural environment and inside the plant host during the defense response [10]. Liang et al. [19] previously reported that the expression of over 20 genes, including superoxide dismutase, was regulated in the sRNA-Xoo1 mutant; these findings indicated that sRNA-Xoo1 is likely involved in oxidative stress tolerance in Xoo PXO99. To deal with oxidative stress, bacterial pathogens often deploy enzymes that either tolerate or scavenge ROS. We recently demonstrated that adenosine-toinosine (A-to-I) RNA editing in Xoc increased tolerance to H 2 0 2 [20]. A-to-I editing in the target fliC caused structural changes in flagella that increased biofilm formation and ultimately improved ROS tolerance [20]. A number of studies have shown that sRNAs can also regulate tolerance to oxidative stress in prokaryotes [22]. In the present study, we used a genetic approach to show that sRNA Xonc3711 contributes to ROS tolerance in Xoc BLS256.
RNA-seq showed that multiple flagella-related genes were downregulated in ΔXonc3711 (Figs 4D and S2); however, our analysis failed to identify a Xonc3711 target that was involved in flagella biosynthesis or regulation. Flagellar-driven motility is critical for biofilm development in many pathogens [25], and biofilm formation is associated with increased adhesion of bacteria to surfaces and improved stress resistance [40,41]. We measured biofilm formation by TEM, and discovered that adherence of ΔXonc3711 to glass surfaces was severely inhibited as compared to the WT (Fig 6 and S1 Video). Thus, it seems likely that the reduced biofilm formation by the ΔXonc3711 mutant resulted in decreased tolerance to oxidative stress.
Exopolysaccharides, degradative enzymes, and toxins all contribute to virulence in X. oryzae [42][43][44]. Most phytopathogenic xanthomonads secrete effector proteins via the T3SS to suppress the defense response. The effectors that are designated as Xanthomonas outer proteins (Xops), are known to be key factors required for bacterial growth and colonization in distinct eukaryotic host [45]. In this study, we also established an important role for Xonc3711 in Xoc virulence and demonstrated that the DNA-binding protein xoc_3982, the target of Xonc3711, negatively regulates xopC2 expression (Fig 8C). A genetic approach was then used to show that xopC2 contributes to lesion size in Xoc BL526 (Fig 7B). Efforts are underway to identify additional genes regulated by Xoc_3982 to fully understand its role in bacterial metabolism and virulence.
Effector proteins encoded by hrp (hypersensitive reaction and pathogenicity) gene clusters are important virulence factors in pathogens. HrpX, a key regulator of hrp genes, regulates the expression of effector genes at a conserved plant-inducible promoter (PIP)-box in the effector promoter region [46]. The PIP-box is a conserved cis-element and its sequence, TTCGB-N 15 -TTCGB (B stands for any base except A), is generally located about 30 bp upstream of the effector gene start codon [46]. ChIP-seq data revealed the potential Xoc_3982 binding site as 5'-CGCTTTT-3' (region -327 to -334 with respect to the translational start site of xopC2); however, the xopC2 promoter region does not contain a PIP-box. In this regard, xopC2 is similar to other xop genes that lack PIP boxes but maintain regulation by HrpX [47]. We did not confirm a role for HrpX in xopC2 regulation; however, this has been reported for xopC in X. campestris pv. vesicatoria [47].
This study provides insight into RNA-mediated regulation of environmental signaling in bacterial physiology and pathogenesis (Fig 9). Xonc3711 base pairs within the xoc_3982 CDS to inhibit translation, which is relatively rare for sRNAs [48,49]. Xonc3711 contributes to biofilm formation and improves oxidative stress tolerance in Xoc BLS256. Based on ChIP-seq data, the DNA-binding protein Xoc_3982 was found to bind to the promoter region of xopC2, a T3SS effector that has been implicated in virulence in some xanthomonads [33]. The identification of other Xonc3711 targets will be helpful in understanding the biological circuitry regulated by sRNAs in phytopathogenic Xanthomonas spp.

Strains, plasmids and primers
The bacterial strains and plasmids used in this study are described in S1 Table. Primers used for the construction of mutant strains, plasmids and DNA templates are provided in S2 Table.
Seeds of rice cv. Yuanfengzao were obtained from the International Rice Research Institute and cultivated at Shanghai Jiao Tong University as described.

Construction of deletion, point and overexpression mutants
Bacterial mutant strains were generated as described by Baba with minor modifications [50]. Two fragments flanking the target gene were amplified from the chromosomal DNA of Xoc BLS256 using Pfu polymerase (TransGen Biotech, Beijing, China) and the primers described in S2 Table. The PCR products were digested, subcloned into the suicide vector pKMS1 [51], and introduced into bacteria by electroporation (Bio-Rad Laboratories Inc., Hercules, CA, USA) with kanamycin selection. A single transformant with kanamycin resistance was selected, cultured for 8 h in NB, and inoculated as 10-fold dilutions to NA with 15% sucrose to obtain sucrose-insensitive clones. For site-directed mutagenesis, plasmids were modified with the Fast Mutagenesis System (Transgen Biotech, Beijing, China) to obtain clones containing point mutations (Xonc3711 � , 3982 � , xopC2 � ; S1 Table).
To obtain the Xonc3711 overexpression mutant (Xonc3711 OE ), the full-length corresponding gene was amplified, and the fragment was cloned into pHM1 with the lac promoter. The recombinant plasmid was transferred into WT by electroporation, and transformants were screened on NA plates supplemented with spectinomycin.

Bacterial growth and gene expression in response to oxidative stress
The optical density of bacterial solutions was measured with a Bioscreen C (Labsystem, Helsinki, Finland). Individual wells of a microtiter plate containing 99 μL of NB or LB broth with or without 0.1 mM H 2 O 2 were inoculated with 1 μL of overnight suspensions of Xoc (1 × 10 9 CFU/mL). OD values at 420-580 nm were obtained at 15 min intervals over a 48 h period with constant agitation at 28˚C. Viable cell counts in the presence and absence of H 2 O 2 were determined as described previously [20]. All experiments were performed in quadruplicate, and the Kolmogorov-Smirnov test was used to evaluate significance.
Assays for resistance to H 2 O 2 were performed as described previously [52]. Briefly, Xoc strains were cultured to the mid-log phase (OD 600 = 1.0~1.2) and exposed to 0.1 mM H 2 O 2 at 28˚C; aliquots were removed at 0, 7, 15, and 45 min and pelleted by centrifugation at 4˚C. Pellets were washed twice in cold PBS, and the total RNA was immediately extracted using the RNeasy Protect Bacteria Mini Kit (Qiagen) as recommended. Two biological replicates were used in this experiment.

Visualization of biofilms and flagella
Biofilm production by Xoc was visualized using GFP-labeled strains as described previously [20,53]. Protocols used for observing biofilms by confocal microscopy have been described [20]. Images, surface topographies and 3D architectures were processed with the Leica Application Suite X (v. 3.4.2.18368).
TEM was used to detect the formation of flagella by Xoc strains. Samples were mounted on carbon-coated grids for 1 min, washed with deionized water and negatively stained with 3% (w/v) phosphotungstic acid for 30 s. A Talos F200 transmission electron microscope (Thermo Fisher Scientific, USA) was used to acquire images at 120 kV.

mRNA purification and cDNA synthesis
Samples of total RNA (10 μg) were treated with the MICROBExpress Bacterial mRNA Enrichment kit (Ambion) and RiboMinus Transcriptome Isolation Kit (Bacteria) (Invitrogen) as recommended by the manufacturers' instructions. Total RNA samples were resuspended in 15 μL of RNase-free water, chemically fragmented to 200-250 bp and used to generate cDNA with Magic 1st cDNA Synthesis Kit (Magic-Bio, China) as described previously [52].

RNA sequencing and analysis
The Illumina Paired End Sample Prep kit was used to create a RNA-Seq library as described [52]. After removing low quality reads and adaptors, RNA-Seq reads were aligned to the corresponding Xoc BLS256 genome using Tophat 2.0.7 [54]as described previously [20]. Differentially expressed genes (FDR value < 0.01) were selected for further analysis. Heatmaps were generated using Cluster 3.0 and Treeview 1.1.6 based on reads per kb of transcript per million mapped reads (RPKM) values [55,56].

In vitro synthesis and labeling of RNA
Xoc_3982 mRNA and Xonc3711 sRNA were prepared using 5 μg of DNA that was generated by PCR with primers F/R-T7Xoc_3982 and F/R-T7Xonc3711 (S2 Table) and the Megascript T7 Transcription Kit (Ambion, Austin, TX, USA). The MAXIscript T7 In Vitro Transcription Kit (ThermoFisher, USA) was used to synthesize RNA from DNA templates; and the RNA transcripts were purified with the MEGAclear Transcription Clean-Up Kit (ThermoFisher, USA). A biotinylated nucleotide was added to the 3' termini of the synthesized RNA molecules using the Pierce RNA 3' End Biotinylation Kit (ThermoFisher, USA) as recommended by the manufacturer.

Electrophoretic mobility shift assays
The Hfq and Xoc_3982 proteins were expressed and purified using the intein-based Impact Kit (New England Biolabs, USA) as described [9]. Binding reactions were conducted in 10 μl volumes with the LightShift Chemiluminescent RNA EMSA Kit (ThermoFisher, USA); reactions were incubated at 37˚C for 20 min, and 5 μl of loading buffer (50% glycerol) was then added. The interaction of Hfq and Xonc3711 sRNA was conducted in 1× binding buffer with 3'-biotinylated Xonc3711 sRNA. The interaction of sRNA Xonc3711-with Xoc_3982 mRNA was investigated using EMSA as described previously [9]. Samples were separated in 5% nondenaturing polyacrylamide gels in 0.5× TBE at 4˚C and visualized by phosphoimaging on a ChemiScope 3000 mini (CLiNX, Shanghai, China).

Quantitative real-time PCR
qRT-PCR was conducted as described previously [20]. Gene expression was normalized relative to rpoD using the ΔΔCT method, where CT is the threshold cycle. Four independent biological replicates were included and analyzed using the Wilcoxon-Mann-Whitney test.

Plant inoculations
Virulence assays were conducted with Xoc suspensions (OD 600 = 0.8), which were inoculated to six-week-old seedlings of rice cv. Yuanfengzao with needleless syringes. Lesion lengths were measured 14 d after inoculation. Twelve or more leaves were inoculated and evaluated for each Xoc strain.