Xanthomonas oryzae pv. oryzae Type III Effector XopN Targets OsVOZ2 and a Putative Thiamine Synthase as a Virulence Factor in Rice

Xanthomonas oryzae pv. oryzae (Xoo) is spread systemically through the xylem tissue and causes bacterial blight in rice. We evaluated the roles of Xanthomonas outer proteins (Xop) in the Xoo strain KXO85 in a Japonica-type rice cultivar, Dongjin. Five xop gene knockout mutants (xopQ KXO85, xopX KXO85, xopP1 KXO85, xopP2 KXO85, and xopN KXO85) were generated by EZ-Tn5 mutagenesis, and their virulence was assessed in 3-month-old rice leaves. Among these mutants, the xopN KXO85 mutant appeared to be less virulent than the wild-type KXO85; however, the difference was not statistically significant. In contrast, the xopN KXO85 mutant exhibited significantly less virulence in flag leaves after flowering than the wild-type KXO85. These observations indicate that the roles of Xop in Xoo virulence are dependent on leaf stage. We chose the xopN gene for further characterization because the xopN KXO85 mutant showed the greatest influence on virulence. We confirmed that XopNKXO85 is translocated into rice cells, and its gene expression is positively regulated by HrpX. Two rice proteins, OsVOZ2 and a putative thiamine synthase (OsXNP), were identified as targets of XopNKXO85 by yeast two-hybrid screening. Interactions between XopNKXO85 and OsVOZ2 and OsXNP were further confirmed in planta by bimolecular fluorescence complementation and in vivo pull-down assays. To investigate the roles of OsVOZ2 in interactions between rice and Xoo, we evaluated the virulence of the wild-type KXO85 and xopN KXO85 mutant in the OsVOZ2 mutant line PFG_3A-07565 of Dongjin. The wild-type KXO85 and xopN KXO85 mutant were significantly less virulent in the mutant rice line. These results indicate that XopNKXO85 and OsVOZ2 play important roles both individually and together for Xoo virulence in rice.


Introduction
Xanthomonas oryzae pv. oryzae (Xoo) causes bacterial leaf blight, which is one of the most serious diseases in rice (Oryza sativa L.). This bacterium invades the xylem of rice leaves through hydathodes or wounds. The strain of Xoo KXO85 (KACC10331) was isolated from diseased rice leaves in Korea, and its whole genome sequence was published in 2005 [1].
T3SS-dependent plant bacterial effectors are important for bacterial growth, colonization, virulence, and race specificity in their host plants [18][19][20][21][22][23]. However, the biochemical functions of most T3SS-dependent plant bacterial effectors in their hosts have not been well characterized. Xanthomonas outer proteins (Xop) are known as non-TAL bacterial effector proteins that are delivered to the plant cell via Hrp T3SS. The major roles of non-TAL bacterial effectors involve modulation of signaling in the plant defense response [11,24]. For example, XopX Xcv from X. campestris pv. vesicatoria (Xcv) affects the virulence of Xcv on pepper (Capsicum annuum) and tomato (Lycopersicum esculentum) and targets basic innate immunity in plants [24]. XopD Xcv is a small ubiquitin-like modifier (SUMO) protease in Xcv that promotes bacterial growth in tomato and slows leaf chlorosis and necrosis in tomato at late stages of infection [25]. Another T3SS-dependent non-TAL effector, XopN Xcc , plays important roles in colonization and virulence of X. campestris pv. campestris (Xcc) in their hosts [26]. XopN is highly conserved among Xanthomonas species [27]. In addition, XopN Xcv may suppress pathogen-associated molecular pattern (PAMP)-triggered immunity in tomato [28].
In Xoo, considerable efforts have been made to characterize functional roles of TAL effectors in various strains [32,33]. The contribution of each TAL effector protein to Xoo virulence varies; some are critical for virulence, while others have relatively moderate roles [32,34]. However, the roles of non-TAL effectors in Xoo virulence have been poorly investigated. When 18 non-TAL effectors were evaluated for virulence in the Philippine strain PXO99 A , deletion of both copies of xopZ PXO99 conferred significant reduction of virulence, whereas the other non-TAL effectors showed little influence on virulence in 4week-old rice leaves [35]. Disease severity of Xoo in susceptible cultivars varies depending on leaf stage [36][37][38][39]. This led us to assess the virulence of each xop mutant at the adult stage in the field with the expectation of more distinct and different disease response outcomes compared to virulence assay results at the young leaf stage. Here, we report the contribution of XopN KXO85 to Xoo virulence in the Korean strain KXO85 at flag leaf stage in the field, identification of targets of XopN KXO85 in rice, and their important roles for Xoo virulence.

Mutagenesis of five xop genes in the Korean Xoo strain KXO85
Five xop genes, xopQ KXO85 (XOO4466), xopX KXO85 (XOO4287), xopP1 KXO85 (XOO3425), xopP2 KXO85 (XOO3426), and xopN KXO85 (XOO0343) (Table S1), were characterized among 18 xop homologs in the strain KXO85 (www.xanthomonas.org/t3e.html), which showed significant homology with reported xop genes. EZ-Tn5 insertion mutants of xopQ KXO85 , xopX KXO85 , xopP1 KXO85 , xopP2 KXO85 , and xopN KXO85 ( Figure S1) were generated in the strain KXO85, and then the virulence of each xop gene knockout mutant was evaluated in 3-month-old leaves of the Japonica-type rice cultivar Dongjin. Mutations in the xopQ KXO85 , xopX KXO85 , xopP1 KXO85 , xopP2 KXO85 , or xopN KXO85 gene did not significantly affect virulence ( Figure  1A). When the xopN KXO85 mutant was inoculated into the flag leaves of Dongjin in the field, the mutant was significantly less virulent than the wild-type KXO85 ( Figure 1B). Virulence the xopN KXO85 mutant carrying each wild-type xop gene in a multicopy plasmid was recovered to the wild-type level ( Figure  1B). These observations indicate that xopN KXO85 exhibits important roles for virulence of Xoo. Therefore, we chose xopN KXO85 for further characterization. The bacterial population of the xopN KXO85 mutant was reduced up to 21 days after inoculation of flag leaves compared to the growth of wild-type strain KXO85 in Dongjin ( Figure 1C).

Expression of xopN KXO85 is regulated by HrpX KXO85
As expression of hrp and xop genes in Xcv and other xanthomonads is controlled by two regulatory genes, hrpG and hrpX, we examined whether xopN KXO85 is regulated by HrpX KXO85 in Xoo KXO85. Expression of xopN KXO85 was below the limit of detection as assessed by quantitative real-time polymerase chain reactions (PCR) in the wild-type KXO85 or in the hrpX KXO85 mutant strain in rich PSB medium ( Figure S2A). In the hrp-inducing medium XOM2, xopN KXO85 expression in the wild-type KXO85 was approximately 3-fold higher than that in the hrpX KXO85 mutant ( Figure S2A). We found a conserved cisregulatory element plant-inducible promoter (PIP) box (TTCGG-N 15 -TTCTG) in the region from -263 to -239 upstream of the start codon of xopN KXO85 ( Figure S2B). These results indicate that xopN KXO85 belongs to the HrpX KXO85 regulon in Xoo KXO85.

XopN KXO85 is a T3SS-dependent effector translocated into plant cells in the strain KXO85
To investigate whether XopN KXO85 is translocated into plant cells in a T3SS-dependent manner, we conducted a XopN KXO85 translocation assay using the XopN-Cya fusion protein in the wild-type strain KXO85 and the T3SS-deficient mutant KXO85 hrpB5 KXO85 ::EZ-Tn5 in rice ( Figure S3A). The level of cAMP increased in the wild-type strain KXO85, whereas no change in cAMP level was detected in the T3SS-deficient mutant KXO85 hrpB5 KXO85 ::EZ-Tn5 ( Figure S3B). This indicates that XopN KXO85 is translocated into rice cells in a T3SS-dependent manner.

Identification of XopN KXO85 targets in rice by yeast twohybrid screening
To identify XopN KXO85 target proteins in rice, we carried out yeast two-hybrid screening using GAL4-XopN as a bait protein and a rice cDNA library constructed in the prey vector in the Saccharomyces cerevisiae strain MaV203. We found two possible candidates: Oryza sativa vascular plant one zinc finger protein 2 (OsVOZ2: NP_001056041, Os05g0515700) and O. sativa XopN KXO85 binding protein (OsXNP: NP_001059841, Os07g0529600) ( Figure 2A and Table S2). The OsVOZ2 gene is 3,630 bp in length consisting of four exons and three introns and encodes a protein of 69,901 Da. OsVOZ2 is a homolog of Arabidopsis thaliana vascular plant  one zinc finger protein 2 (AtVOZ2; At2g42400) that has a conserved zinc finger domain ( Figure S5 and Figure S6). The OsXNP gene is 1,489 bp in length with two exons and one intron and possibly encodes a putative protein of 37,224 Da that has significant homology with thiamine biosynthetic enzyme in Saccharum hybrid cultivar GT28 (Table S2). XopN KXO85 , OsVOZ2, and OsXNP were expressed in yeast as confirmed by immunoblot using anti-GAL4BD and anti-GAL4AD antibodies ( Figure S4).

Visualization of the interactions of OsVOZ2 and OsXNP with XopN KXO85
A bimolecular fluorescence complementation (BiFC) assay was performed to examine the interactions between XopN KXO85 and OsVOZ2 and XopN KXO85 and OsXNP in planta. The coding sequences of xopN KXO85 , OsVOZ2, OsXNP, and OsVOZ1 were cloned into pDEST-SCYNE(R) GW and pDEST-SCYCE(R) GW using the Gateway recombination system to yield pSCYNE(R)-XopN, pSCYCE(R)-OsVOZ2, pSCYCE(R)-OsXNP, and pSCYCE(R)-OsVOZ1, respectively (Table S3). When Agrobacterium cells carrying both plasmids were infiltrated into N. benthamiana leaves, the super cyan fluorescent protein (SCFP3A) signal was detected in the cytoplasm of the N. benthamiana cells ( Figure 2C). As a positive control, we used the Cnx6 and Cnx7 interaction model to form a complex of molybdopterin synthase in A. thaliana using pEXP-SCYNE(R)-Cnx7 and pEXP-SCYCE(R)-Cnx6 [40]. These results indicate that XopN KXO85 interacts with OsVOZ2 and OsXNP in the cytoplasm of N. benthamiana cells. However, XopN KXO85 does not interact with OsVOZ1 in N. benthamiana cells ( Figure 2C).

Subcellular localization of XopN KXO85 , OsVOZ2, and OsXNP
To determine their subcellular localizations, XopN KXO85 , OsVOZ2, and OSXNP were tagged with GFP at their C-termini in p2GWF7-XopN, p2GWF7-OsVOZ2, and p2GWF7-OsXNP, respectively (Table S3). In transient expression assays using maize mesophyll protoplasts, GFP signals from XopN-GFP and OsXNP-GFP were mostly detected in the cytoplasm, whereas those from OsVOZ2-GFP were detected in both the cytoplasm and the nucleus compared to the nuclear marker OsABF1-RFP ( Figure 3). These data indicate that XopN KXO85 and OsXNP are localized in the cytoplasm, whereas OsVOZ2 is localized in a nuclear and cytoplasm ( Figure 3).

Interactions between XopN KXO85 and OsVOZ2 are important for Xoo virulence in rice
To determine whether OsVOZ2 and its interactions with XopN KXO85 are critical for Xoo virulence, the OsVOZ2 knockout mutant line PFG_3A-07565 from the rice T-DNA Insertion Sequence Database (http://signal.salk.edu/cgi-bin/RiceGE) [41] was inoculated with wild-type KXO85. In the mutant line PFG_3A-07565, T-DNA is inserted 929 nucleotides downstream from the translational start site of OsVOZ2. RT-PCR analysis detected OsVOZ2 transcript in wild-type Dongjin but not in the OsVOZ2 mutant line PFG_3A-07565 ( Figure 4A), which confirmed knockout mutation in OsVOZ2. Wild-type KXO85 and xopN KXO85 mutant strains were inoculated into wildtype Dongjin and the OsVOZ2 mutant line, and the xopN KXO85 mutant was shown to exhibit reduced virulence in the wild-type Dongjin. However, both strains showed significantly reduced disease severity in the OsVOZ2 mutant line compared to the wild-type Dongjin ( Figure 4B). The xopN KXO85 mutant was less virulent in the OsVOZ2 mutant line than the wild-type KXO85 ( Figure 4B and 4C). The population of xopN KXO85 mutant was smaller than that of wild-type KXO85 in the OsVOZ2 mutant line ( Figure 4D). These results indicate that XopN KXO85 is a virulence factor and that its interactions with OsVOZ2 are critical for Xoo virulence in rice.

Discussion
There has been some confusion regarding the roles of Xop of Xoo because previous studies have used different Xoo strains. The Xoo PXO99 A strain has 18 non-TAL bacterial effectors [35]. Among these, XopZ PXO99 acts as a virulence factor in the Xoo PXO99 A strain and suppresses plant basal defense mechanisms [35]. XopR MAFF311018 was reported as a virulence factor in rice and inhibits the plant basal defense in A. thaliana [42]. Nine non-TAL effectors have been identified in the Chinese strain 13751, among which XopR 13751 has been shown to affect virulence in Xoo [43]. In the present study, we chose XopN KXO85 to evaluate functional roles in the KXO85 strain and confirmed that it is secreted in an Hrp T3SSdependent manner, translocated into the plant cytoplasm, and that its gene expression is regulated by HrpX KXO85 , as reported previously for other Xoo strains [13]. Among the Xop homologs in KXO85, we found that XopN KXO85 is the most critical for Xoo virulence in the Korean strain KXO85. This result is similar to other reports indicating that xopN Xcv and xopN Xcc mutants show reduced virulence [26,28].
It is worth noting that different Xop effectors from different Xoo strains have been reported to be major Xops involved in Xoo virulence. Differences in genetic backgrounds of Xoo strains and rice cultivars used for virulence assays may explain why different research groups have reported different Xops as major virulence factors. For example, in one study, a mutation in the xopN homolog in Xoo PXO99 A did not alter disease severity in rice cultivar IR24 that was grown in a growth chamber for 4 weeks [35]. However, it should also be noted that differences in environmental conditions and various rice leaf stages used for inoculation of different Xoo strains may also result in different outcomes in virulence assays. In previous studies on Xoo PXO99 A and Chinese strain 13751, relatively young rice leaves were used for virulence assays in a growth chamber or a greenhouse [35,43], whereas we used flag leaves grown in a paddy field during the regular rice growing season. Disease severity induced by Xoo depends on rice leaf stage [36][37][38][39]. These observations correspond well with previous reports that the response to Xoo in rice depends on the age of the host [36]. Environmental conditions for growing rice and virulence assays are additional factors that may affect disease severity. It will be of interest to determine whether the xopN PXO99 mutant of Xoo PXO99 A shows differences in virulence assays when the mutant is inoculated into rice flag leaves.
Identification of target proteins of bacterial effectors in their hosts provides a basis for understanding effector functions and their roles in pathogenesis and host defense. XopN Xcv targets a tomato atypical receptor-like kinase1 (TARK1) and four tomato 14-3-3 isoforms (TFT1, TFT3, TFT5, and TFT6) to affect the defense signal mechanism [28]. In Xoo, the Xoo1488 of Xoo MAFF311018 inhibits OsRLCK185 phosphorylation and the downstream MAPK signaling [31]. Therefore, we postulated that XopN KXO85 may interact with known kinases that are involved in signal transduction pathways in rice. However, unlike OsRLCK185 in rice, we found no kinase homologs as XopN KXO85 targets but rather two previously unknown rice proteins, OsVOZ2 and OsXNP, were identified based on yeast two-hybrid analysis, pull-down, and BiFC assays.
The AtVOZs were first identified as novel transcription factors in A. thaliana [44]. AtVOZs interact with phytochrome B and accelerate flowering time in A. thaliana [45]. In the nuclei of A. thaliana cells, AtVOZ2 is controlled by light quality in a phytochrome-dependent manner [45]. In addition, AtVOZs are involved in controlling many stress reactions and changing the expression of various stress-related genes, such as those related to drought or freezing responses and pathogens [46]. The genome of the wild-type rice Dongjin has an OsVOZ2 homolog, OsVOZ1, which is also an ortholog of AtVOZ2 and has conserved zinc finger amino acid residues [44]. OsVOZ1 and OsVOZ2 share 60.4% identity ( Figure S6). Due to the high degrees of similarity between OsVOZ1 and OsVOZ2, we performed BiFC analysis to determine whether OsVOZ1 is a target protein of XopN KXO85 . However, there was no evidence of an interaction between XopN KXO85 and OsVOZ1 ( Figure 2C).
AtVOZ2 interacts with five proteins in A. thaliana: phytochrome B (PHY B, At2g18790), guanine nucleotidebinding protein alpha-1 subunit (GP ALPHA1, At2g26300), guanine nucleotide-binding protein subunit beta (AGB1,At4g34460), pirin (PRN, At3g59220), and a hypothetical protein (At4g26410) [45,47]. The most apparent AtVOZ2dependent phenotype is regulation of flowering period in A. thaliana after it interacts with phytochrome B [45]. However, it appears that OsVOZ2 is not involved in determining rice flowering time because we found no noticeable differences in flowering time between wild-type Dongjin and the OsVOZ2 mutant rice line. Other than our findings indicating that OsVOZ2 is a target of XopN KXO85 and is involved in Xoo virulence, no other functions have yet been reported in rice.
Another target of XopN KXO85 is a putative thiamine synthase, OsXNP, which is present as a single-copy gene in rice. The thiamine synthase gene is related to pathogen-induced defense-responsive protein 8 in Indica rice cultivars. Treatment with thiamine induces callose deposition and hydrogen peroxide accumulation and triggers systemic acquired resistance and transient expression of pathogenesis-related genes against pathogen invasion in rice and several other plants [48,49]. These phenomena are consistent with the observation that thiamine plays important roles in host defense mechanisms against pathogen infection. Therefore, we propose that XopN KXO85 interacts with a putative thiamine synthase to hinder thiamine biosynthesis, thereby decreasing the defense of rice against Xoo infection. The target proteins of XopN KXO85 in rice are completely different from the previously reported targets of XopN Xcv . These observations indicate that XopN plays a common role as a virulence factor in Xcv, Xcc, and Xoo but functions in different ways in monocots and dicots, reflecting the different pathogen response mechanisms that arose during the coevolution of pathogens and their hosts.
In addition to roles of XopN KXO85 as a virulence factor, OsVOZ2 is also important for Xoo virulence because wild-type KXO85 failed to successfully infect OsVOZ2 mutant rice. This suggests that interactions between XopN KXO85 and OsVOZ2 in rice increases susceptibility to Xoo infection. That is, Xoo produces XopN KXO85 as an effector molecule and utilizes the host protein OsVOZ2 for successful infection and increased virulence. Although the functions of OsVOZ2 are not fully understood in the interactions between Xoo and rice, it is evident that OsVOZ2 is a key factor in Xoo virulence in rice.

Ethics Statement
No specific permits were required for these kinds of field studies. This field is owned by the University Farm, College of Agriculture and life Sciences, Seoul National University. This university farm is located in Suwon, which is approximately 40 kilometers south of the main campus of Seoul National University in Seoul, Republic of Korea. The location is not privately-owned or protected in any way. The field studies did not involve endangered or protected species.

Transposon insertion and marker-exchange mutagenesis
All recombinant DNA techniques were performed according to standard methods [50]. To generate the xopN KXO85 mutant, the approximately 3-kb BamHI fragment carrying the xopN KXO85 gene from BAC clone G2 (Table S3) of Xoo KXO85 was cloned into pML122. EZ-Tn5<TET-1> was inserted into the coding region of xopN KXO85 in pML122 by in vitro transposition according to the supplier's instructions (Epicentre) yielding pXopN::EZ-Tn5 (Table  S3). pXopN::EZ-Tn5 was electroporated into Xoo KXO85, and the transformed cells were cultured on PSA medium containing tetracycline. The markerexchanged mutant Xoo KXO85 xopN KXO85 ::EZ-Tn5 was isolated and confirmed by Southern hybridization. Transposon insertion and marker-exchange mutagenesis of the other xop genes (xopQ KXO85 , xopX KXO85 , xopP2 KXO85 , and xopP1 KXO85 ) were performed by the same strategy as described above to generate the xopN KXO85 mutant in Xoo KXO85.

Virulence assay
Rice plants of cultivar Dongjin were grown in a paddy field. The OsVOZ2 mutant rice seeds (PFG_3A-07565; T 0 seed) were affirmed by the rice T-DNA Insertion Sequence Database (http://signal.salk.edu/cgi-bin/RiceGE) [41]. The homozygous T 7 transgenic mutant line of the OsVOZ2 mutant rice was obtained and confirmed by RT-PCR analysis. Overnight cultures of Xoo cells were adjusted to approximately 1.8×10 8 CFU/mL and inoculated into 3-month-old leaves or fully expanded flag leaves by the scissor clip method [51]. Symptoms were scored by measuring lesion lengths 21 days after inoculation. The growth of Xoo cells in plants was determined as described previously [16].

Quantitative real time RT-PCR analysis
The bacterial strains used were cultured in liquid medium XOM2 [52] or PSB for 24 h. Total RNA was isolated from the wild-type strain KXO85 and KXO85 hrpX KXO85 ::EZ-Tn5 using an RNeasy kit (Qiagen) according to the manufacturer's instructions. A total of 1 µg RNA was reverse transcribed into cDNA using M-MLV reverse transcriptase (Promega) for 1 h at 42°C. RT-PCR products from samples were analyzed on agarose gels and the bacterial 16s rRNA was used as a standard. Quantitative real-time RT-PCR (qRT-PCR) was performed using the cDNA and gene-specific primers (Table  S4). The transcription levels were determined by Power SYBR Green PCR Master Mix on an Applied Biosystems 7500 Real-Time PCR System (Applied Biosystems). The thermal cycling parameters were: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s and 60°C for 1 min. Expression of 16S rRNA was used to normalize the expression values in each sample, and relative expression values were determined against the average value of wild-type strain KXO85 using the comparative Ct method.

Adenylate cyclase assays
To generate the xopN-cya gene fusion protein, the xopN KXO85 gene was cloned into the XbaI and XhoI sites of pMLTC to generate pMCXopN (Table S3) followed by transformation into Xoo KXO85 and KXO85 hrpB5 KXO85 ::EZ-Tn5. For the assay of adenylate cyclase activity in rice leaf tissues, rice leaves were hand-inoculated with bacterial suspension using a needleless syringe. After 12 h, samples were frozen with liquid nitrogen and homogenized in assay buffer supplied with the cAMP Biotrak Enzyme Immunoassay System (GE Healthcare). The level of cAMP in leaf samples was measured by the cAMP Biotrak Enzyme Immunoassay System according to the manufacturer's directions.

Yeast two-hybrid assay
A Gal4-based system with Gateway technology (Invitrogen) was used for a yeast two-hybrid assay. The xopN KXO85 gene was amplified by PCR using Xoo KXO85 genomic DNA as a template. The PCR primers (Table S5) were flanked with the attB1 and attB2 sites required for the Gateway cloning system. The PCR product was cloned into pDONR222 by BP recombination to generate the entry clone. Subsequently, the xopN KXO85 gene was transferred to the yeast destination bait plasmid pDEST32 by LR recombination resulting in pD32XopN (Table S3). To construct a Dongjin cDNA library, cDNA of approximately 0.5-3 kb was cloned into pDONR222 and subsequently into the prey plasmid pDEST22 by LR recombination yielding pD22Lib (Table S3). pD32XopN contains the DNA-binding domain of Gal4 and the leucine selection marker gene LEU2. pD22Lib contains the GAL4 transcription activation domain and the tryptophan selection marker gene TRP1. All constructs were checked by restriction enzyme analysis and confirmed by DNA sequencing. pD32XopN (bait) and pD22Lib (prey) were co-transformed into yeast strain MaV203 according to the manufacturer's protocol (Invitrogen). The transformants were cultured on synthetic complete (SC) medium lacking leucine (-Leu) and tryptophan (-Trp). After 72 h, colonies were picked and mixed with 100 µL of sterile water, and 10 µL of the cell suspension was spotted onto selection plates to screen for expression of the three reporter genes (HIS3, URA3, and lacZ). Growth of the yeast transformants was assessed on SC-Leu-Trp-His supplemented with 0-50 mM 3-amino-1,2,4-triazole (3AT) as a histidine inhibitor and SC-Leu-Trp-Ura. A change in the blue color of the transformants was monitored in the presence of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). To check for autoactivation of the reporter genes, pD32XopN, pD22OsVOZ2, and pD22OsXNP were combined with pDEST32 or pDEST22 and tested for autoactivation activity. We used the controls provided by Invitrogen: S (strong control: pEXP TM 32/Krev1 + pEXP TM 22/RalGDS-wt), W (weak control: pEXP TM 32/Krev1 + pEXP TM 22/RalGDS-m1), and A (absent control: pEXP TM 32/Krev1 + pEXP TM 22/RalGDS-m2). Protein expression was confirmed by immunoblotting using anti-GAL4BD (Clontech) and anti-GAL4AD (Clontech) antibodies. Signals were visualized using an Immun-Star WesternC Kit (Bio-Rad).

Agrobacterium-mediated transient expression
Agrobacterium infiltration into N. benthamiana leaves was performed as described previously [28]. Cells of A. tumefaciens strain C58C1 (pCH32) [53] were cultured at 28°C for 2 days on LB agar medium containing 50 µg/mL kanamycin and 2.5 µg/mL tetracycline. The recombinant agrobacteria were grown in 10 mL LB liquid medium supplemented with appropriate antibiotics at 28°C and then harvested by centrifugation. The cell pellet was resuspended in buffer (10 mM MES, pH 5.6, 10 mM MgCl 2 , and 150 µM acetosyringone), adjusted to a final OD 600 of 0.6, and then incubated for 3 h at room temperature before inoculation. Cells were hand-infiltrated onto N. benthamiana leaves using a needleless 1 mL syringe. Inoculated plants were incubated at 26°C in a growth chamber for 1 to 2 days.

Localization of OsVOZ2, OsXNP, and XopN KXO85
The OsVOZ2, OsXNP, and xopN KXO85 genes in pENTR D TOPO were cloned into the destination vector p2GWF7 to create a C-terminal GFP fusion [54] using the Gateway LR recombinase (Invitrogen). The constructs were introduced into maize mesophyll protoplasts by polyethylene glycol-calciummediated transformation [55,56]. The protoplasts were examined after incubation for 12-24 h. OsABF1-RFP was used as a nuclear marker [57].

Statistical analysis
JMP ® 10 software (SAS Institute) was used for statistical analysis. Statistical significance was determined by a one-way or a two-way ANOVA with Tukey HSD post-test.  Figure S4. Self-activation test and Western blot analysis showing expression of the yeast plasmid constructs in yeast two-hybrid screening. A. Self-activation tests were conducted using pD32XopN + pDEST22, pDEST32 + pD22OsVOZ2, pDEST32 + pD22OsXNP, and pDEST32 + pDEST22. B. Total proteins were extracted from the indicated yeast strains. Anti-GAL4BD and anti-GAL4AD antibodies were used for immunoblotting. M, size marker; 1, pD32XopN and pD22OsVOZ2; 2, pD32XopN and pD22OsXNP; and 3, pDEST32 and pDEST22. The expected molecular weights of the proteins were as follows: GAL4BD-XopN KXO85