The coding sequence of OsERF922 (Os01g54890) was amplified with primers E922F (5′-ACAGGATCCCCGCATGTCTCTCTCCTTG-3′) and E922R (5′-TTACTCGAGTATCACGCACTCACCGTCGCT-3′) from a rice IR72 cDNA library and ligated into a T-vector to give pUCmT-OsERF922. To construct plasmids for overexpression, the encoding region of OsERF922 was fused at its N-terminal with the sequence encoding the Flag-tag and put under the control of a maize ubiquitin promoter, generating pCo-Ubi:Flag-OsERF922. To knock down OsERF922 transcription, a fragment of OsERF922 gene (from 1 to 203 bp) was constructed in a form to generate a self-complementary hairpin RNA structure, which was driven by the cauliflower mosaic virus (CaMV35S) promoter (pCaMV35S:dsOsERF922).

For analysis of OsERF922 localization, the full-length coding sequence of OsERF922 was amplified with primers E922F and E922R and was fused in frame to the N-terminus of the green fluorescent protein (GFP) gene to generate the CaMV35S:OsERF922-GFP construct. All of the vectors constructed were verified by sequencing and introduced into rice calli of cultivar Zhonghua 17 (japonica, Zh17) by the Agrobacterium-mediated transformation method. More than 10 independent transgenic lines were obtained for each construct. The overexpressing plasmid pCo-Ubi:Flag-OsERF922 was also transformed into tobacco (Nicotiana tabacum cv. Xanthin nc) using the similar Agrobacterium-mediated method. The transgenic lines at the T₂ generation were screened by planting seeds on 1/2 MS medium (containing 0.5% agar) supplemented with 40 mg l⁻¹ hygromycin and the antibiotic resistant seedlings were used in the following experiments unless otherwise indicated.

Seeds of Oryza sativa Zh17 were surface sterilized, imbibed in water at 37 °C for 2 d before germination and seedlings were grown in vermiculite supplemented with mineral nutrient solution in a greenhouse around 28 °C under natural sunlight. Three-week-old rice seedlings were used for abiotic stress treatments, including spraying with 100 μM ABA (dissolved in 10 mM 4-morpholineethanosulphonic acid (MES) buffer, pH 6.0), or the submerging roots in the nutrient solution containing 150 mM NaCl.

For analysis of salt tolerance, the transgenic and wild-type (WT) plants were grown on 1/2 MS medium with 150 mM NaCl (containing 0.5% agar) under a 16/8 light/dark cycle at 28 °C for 3 weeks. These plants were then transferred to paddy pots for recovery at normal growth conditions. The plants were evaluated 8 d later.

Inoculation of rice blast fungi, M. oryzae, was conducted as described by ^{Wang et al. (2007)}. Briefly, spores were harvested and suspended in 0.1% (w/v) gelatin to 10⁵ conidia ml⁻¹. Three-week-old rice plants were inoculated with the spore solution by spraying. The inoculated plants were maintained in the growth chamber at 24 °C in darkness for 24 h, then under growth conditions of a 16/8 light/dark cycle with 95% humidity. For analysis of gene expression, rice plants were inoculated with an avirulent M. oryzae strain P131 or a virulent strain P140. Leaves were sampled at designated times and used for RNA isolation.

To examine the capability of disease resistance, 3-week-old transgenic and the WT plants were inoculated with the virulent strain P140 as mentioned above. Some leaves from each line were sampled for analysis of gene expression 6 h post inoculation and disease severity was evaluated by a method of image analysis described by ^{Fukuoka et al. (2009)}. The area of lesions formed 7 d after the infection was counted on the third leaves (15 cm in length) of 20 plants for each line.

Subcellular localization of OsERF922 in rice was examined by using the CaMV35S:OsERF922-GFP transgenic seedlings and the CaMV35S:GFP transgenic plants were used as the control. 4′-6-Diamidino-2-phenylindole (DAPI) staining was used as a nuclear marker. Fluorescence signals were detected using a confocal microscope (Eclipse TE2000, Nikon).

Total proteins were extracted from leaves of 7-d-old seedlings by grinding with small plastic pestles in the extraction buffer (100 mM HEPES, pH 7.5, 5 mM EDTA, 5 mM EGTA, 10 mM Na₃VO₄, 10 mM NaF, 50 mM β-glycerophosphate, 10 mM dithiothreitol, 1 mM phenylmethylsulphonyl fluoride, 5 μg ml⁻¹ leupeptin, 5 μg ml⁻¹ aprotinin, 5% glycerol). After centrifugation at 13,000 g for 15 min, the supernatants were transferred into clean tubes, quickly frozen in liquid nitrogen, and stored at –80 °C until use. Protein concentration was determined by using a protein assay kit (Bio-Rad) with bovine serum albumin as a standard. Protein samples (2 μg) were separated on a 10% SDS-PAGE and analysed by protein gel blot. Mouse anti-Flag antibody (Sigma-Aldrich) with 1:4000 dilution was used as the primary antibody and goat anti-mouse peroxidase-conjugated AffiniPure Goat Anti-Mouse IgG (H+L) with 1:8000 dilution was used as the secondary antibody. The membranes were visualized using a Super-Signal West Femto Trial Kit (Themor) following the manufacturer’s instructions.

A reporter plasmid was constructed by replacing the full CaMV35S promoter in pBI121 using a fragment harbouring four repeats of the GCC-box and the CaMV35S minimal promoter (from –46 to +10), kindly provided by Prof. Huang (^{Zhang et al., 2004}). Agrobacterium cells harbouring the reporter plasmid were harvested by centrifugation and resuspended in 10 mM MES, pH 5.6, 10 mM MgCl₂, and 0.2 mM acetosyringone to an optical density (600 nm) of 0.7. After 4 h incubation at 28 °C, the bacteria were infiltrated into the leaves of 10-d-old pCo-Ubi:Flag-OsERF922 transgenic and WT tobacco plants using a needle-free syringe. The tissues around the infiltration sites were punched off into microcentrifuge tubes 48 h post treatment. GUS activity was determined as described using 4-methylumbelliferone as the standard (^{Jefferson, 1987}).

DNase-treated RNA (5 μg) was reverse transcribed in a total volume of 50 μl using reverse transcriptase (Takara). Real-time quantitative PCR (qPCR) was performed using 20 ng of total RNA in a 20 μl reaction volume using SYBR Green PCR MasterMix (Takara) on a StepOne Quantitative PCR system (Applied Biosystems) following the manufacturer’s protocols. Primer sequences used for the qPCR analysis are shown in Supplementary Table S1 (available at JXB online).

For estimation of endogenous ABA levels of rice seedling, samples were snap-frozen with liquid nitrogen and lyophilized. About 30 mg of dried sample was ground to powder with liquid nitrogen and extracted at 4 °C at 12 h with 1 ml of solution A (10% MeOH, 1% acetic acid) containing 50 ng of each D6-ABA (ICON Isotopes) as internal standards. The total extract was centrifuged (13,000 g, 10 min) and the pellet was re-extracted once with 1 ml of solution A. The combined supernatant was dried with N₂ gas, and resultant residue was taken up in 50 μl solution A for LC-MS analysis (^{Fan et al., 2009}).

Samples (10 μl) were then analysed by HPLC-electrospray ionization/MS-MS using an Agilent 1100 HPLC coupled to an Applied Biosystems Esquire 6000 (Bruker). Chromatographic separation was carried out on a Phenomenex Luna 3 μm C18 150 × 2.0 mm column at 35 °C. The solvent gradient used was 100% solvent A (H₂O/CH₃CN/CHOOH 94.9:5:0.1) to 100% solvent B (H₂O/CH₃CN/CHOOH 5:94.9:0.1) over 20 min. Solvent B was held at 100% for 5 min then the solvent returned to 100% A for 10 min equilibration prior to the next injection. The solvent flow rate was 200 μl min⁻¹. Analysis of the compounds was based on appropriate multiple reaction monitoring of ion pairs for labelled and endogenous ABA.

For analysis of Na⁺ and K⁺ contents, the transgenic and WT plants were grown on 1/2 MS medium with 100 mM NaCl (containing 0.5% agar) under a 16/8 light/dark cycle at 28 °C for 3 weeks. Leaves were sampled and used for analysis of Na⁺ and K⁺ contents with a flame photometer (2655-00 Digital Flame Analyzer, Cole- Parmer Instrument Company, Chicago, USA), as described by ^{Song et al. (2005)}.

Previous studies demonstrated that tobacco OPBP1, an ERF transcription factor, is involved in disease resistance and salt tolerance (^{Guo et al., 2004}). Multiple database searches and DNA sequence alignment analyses showed OsERF922 is one of the genes with high similarity to OPBP1 (Supplementary Fig. S1). OsERF922 has an ERF/AP2 domain, which contains the 14th Ala and 19th Asp residues, two key amino acids contributing to GCC-box binding in many ERFs (^{Sakuma et al., 2002}). In addition to the ERF/AP2 domain, OsERF922 protein includes a relatively acidic and Ser-rich (56–81 aa) region, which might act as a transcription activation domain.

Many ERF members have been shown to take part in hormonal signal transduction in response to biotic and abiotic stresses (^{Zhang et al., 2004}; ^{Liang et al., 2008}; ^{Pré et al., 2008}). To investigate whether OsERF922 is involved in rice defence responses, the current study used qPCR to analyse the expression patterns of OsERF922 during both the compatible and incompatible interactions between rice and rice blast fungi. Following infection with an avirulent M. oryzae strain, P131, transcription of the OsERF922 gene was induced as early as 1 h post inoculation (hpi), and its level peaked at 3 hpi with a 14-fold increase in mRNA accumulation compared with that at 0 h and decreased sharply to near basal level at 12 h (Fig. 1A). Expression of OsERF922 was also induced by inoculation with a virulent P140 strain, but the induction was slower and weaker than during the incompatible interaction (Fig. 1A). The results suggested that OsERF922 might be involved in early defence responses against pathogens. Further, this study examined the expression levels of OsERF922 in 3-week-old rice seedlings under abiotic stresses and hormone treatments. The mRNA accumulation of OsERF922 was increased remarkably after treatment with ABA (100 μM) for 1 h, reached to a maximal level after 6 h, and was maintained at a relatively high level throughout the rest of the experimental period (Fig. 1C). Similarly, expression of OsERF922 was induced about 5-fold by the NaCl treatment (Fig. 1C). Induction of PR1a (Fig. 1B) and Rab16a (Fig. 1D) revealed the efficacious pathogen and chemical treatments, respectively. However, no significant changes of OsERF922 expression were observed during methyl jasmonate or ET treatments (data not shown). The regulation of OsERF922 expression by various stresses suggests that it might play multiple roles in response to environmental stimuli.

To examine the interaction of OsERF922 with the GCC box in vivo, a transient expression experiment was performed. A DNA fragment harbouring four GCC-element repeats was fused to the minimal CaMV35S promoter (–46 bp) and the GUS gene and used as a reporter plasmid (^{Zhang et al., 2004}) (Fig. 2A). Tobacco plants overexpressing OsERF922 under the control of a maize ubiquitin promoter were used as effectors. The 4 × GCC-GUS reporter construct was introduced into the leaves of the homozygous transgenics of T₃ progenies by means of Agrobacterium-mediated infiltration method (^{Yang et al., 2000}). The GUS activity was increased about 5–10-fold in the six transgenic lines, when compared with the WT control, demonstrating that OsERF922 could bind to the GCC-box in planta and act as a transcriptional activator (Fig. 2B).

The deduced OsERF922 protein has no obvious nuclear localization signal, as predicted by PSORT analysis (^{Nakai and Kanehisa, 1992}). To determine the subcellular localization of OsERF922 protein in rice, the construct of a CaMV35S-controlled OsERF922-GFP chimeric gene (CaMV35S:OsERF922-GFP) and the control CaMV35S:GFP vector were stably transformed into rice. Inner epidermal cells of the transgenic T₁ progenies were observed by microscopy technique for fluorescence signal. Localization of the OsERF922-GFP fusion protein was visualized mainly in the nuclei (Fig. 3). These results suggest that OsERF922 could target to the nucleus.

To reveal the biological function of OsERF922 in plants, full-length cDNA of OsERF922 was fused downstream of the maize ubiquitin promoter, generating a Ubi:OsERF922 construct with a Flag-tag at the N-terminus for rice transformation. For OsERF922 knockdown, a hairpin structure (CaMV35S:dsOsERF922) was used for RNA interference. More than 15 independent transgenic lines were obtained for each construct, none of which showed a significant difference in plant morphological features (such as root length, plant height, and numbers of spikelets) compared with the WT plants under normal growth conditions (data not shown). To determine the level of Flag-OsERF922 in the transgenic lines, total proteins were isolated from six randomly selected transgenic plants and analysed by Western blotting using the anti-Flag antibody. The results showed high levels of the fusion proteins in all lines tested, although the levels differ between lines (Fig. 4A).

To examine whether the OsERF922 gene is involved in disease resistance, the transgenic plants were inoculated with the virulent fungal pathogen M. oryzae P140. As illustrated in Fig. 4I, lesions formed by pathogen infection were remarkably increased in the OsERF922-overexpressing plants of the T₂ progeny compared to the WT plants, whereas the OsERF922 RNAi plants displayed a significant decrease of lesion formation. The differences were further evaluated by quantification of the lesion areas (Fig. 4J), which again indicates that knockdown of OsERF922 expression enhanced disease resistance against the rice blast pathogen.

To examine whether the altered disease resistance phenotypes of the overexpressing lines and the RNAi lines are associated with altered defence gene expression, this study analysed transcript levels of several defence-related genes with or without pathogen inoculation using the qPCR method. The level of the OsERF922 transcript was increased more than 18-fold in the overexpressing plants and decreased 4-fold or more in the RNAi plants compared with the WT plants (Fig. 4B). As expected, expression of OsERF922 was induced in WT plants 6 h post M. oryzae inoculation, whereas its level did not change significantly in the OsERF922-overexpressing linesm which is most likely due to the high level of constitutive expression of the transgene. The transcript level remained low in the knockdown plants following pathogen infection, suggesting efficient suppression of OsERF922 in the RNAi lines. Among the defence-related genes, expression of the PR genes, including PR1a, PR1b, and PR10 were reduced in the overexpressing plants but enhanced in the RNAi plants in comparison with these in WT plants (Fig. 4C–E). Increases in expression of these genes were observed in all of the plants after inoculation with the rice blast pathogen. However, the transcript levels of the PR1a, PR1b, and PR10 genes were increased much more in the knockdown lines than those in the control plants. In contrast, the induction of these genes by the pathogen infection was compromised in the OsERF922-overexpressing lines.

Phenylalanine ammonia lyase (PAL) catalyses the first committed step of phenolic biosynthesis in plants, including SA. Analysis of PAL expression revealed slight suppression in the overexpressing lines and an increase of 5-fold or more in the RNAi lines compared with the control plants (Fig. 4F). Stronger induction of PAL expression was observed in the RNAi plants during blast fungal infection, whereas its transcript level was not increased significantly in the overexpressing plants. Diterpene cyclases, ent-copalyl diphosphate (ent-CDP) synthase (OsCPS2/OsCyc2), and syn-CDP synthase (OsCPS4/OsCyc1) catalyse the conversion of geranylgeranyl diphosphate (GGDP) to ent-CDP and syn-CDP, respectively, and ent-CDP leads to synthesis of phytocassanes A to E and oryzalexins A to F and syn-CDP leads to synthesis momilactones A and B and oryzalexin S (^{Kanno et al., 2006}). These antimicrobial compounds, known as phytoalexins or allelochemicals, are synthesized in the rice leaf in response to infection with the blast pathogen or exposure to UV irradiation (^{VanEtten et al., 1994}; ^{Kato-Noguchi and Ino, 2003}). Similarly to the expression pattern of the PAL gene, accumulation of OsCPS2 mRNA was increased in the RNAi lines and enhanced with the inoculation with blast fungi (Fig. 4G). However, expression of the OsCPS4 gene was suppressed more than 20-fold in the overexpressing progenies and the level was much lower than the control plants even upon the infection with M. oryzae (Fig. 4H). These results suggest that OsERF922 is involved in disease resistance as a negative regulator by suppressing defence-related genes.

Induction of OsERF922 expression by NaCl and ABA treatments prompted this study to test the tolerance of the transgenic rice plants to NaCl. The transgenic seeds of T₂ progeny and WT plants were germinated on MS medium and the seedlings were then transplanted on MS medium containing 150 mM NaCl. After growing on the NaCl medium for 3 weeks, leaves of the OsERF922-overexpressing lines became scorched particularly at the top of the leaves (Fig. 5A). The symptom was less severe in the RNAi and WT plants. After these plants were transferred to paddy pots for recovery (Fig. 5B), nearly half of the plants overexpressing OsERF922 died during the 8-d recovery period (Fig. 5E). Those plants that survived also showed significant retardation of growth and less biomass than that of the WT plants (Fig. 5C and D).

Salt tolerance in plants is the consequence of several physiological processes, such Na⁺ uptake, Na⁺/K⁺ balance, and ion compartmentalization. Damage of leaves caused by salinity is attributed to accumulation of Na⁺ in the shoot by transport of Na⁺ from the roots experiencing high external salt concentration (^{Lin et al., 2004}). Thus, the current study measured the accumulation of Na⁺ and K⁺ in the rice shoots. The levels of K⁺ were lower in OsERF922-overexpressing lines than those in the WT and RNAi plants on 1/2 MS media, whereas the difference of Na⁺ accumulation was not obvious among the plants tested (Fig. 6A). During NaCl treatment, an increase of Na⁺ content was observed, with significantly higher accumulation in OsERF922-overexpressing lines than the WT and RNAi plants. On the other hand, the content of K⁺ was suppressed during the salinity treatment (Fig. 6B), in agreement with competition between Na⁺ and K⁺ uptake in the shoots (^{Lin et al., 2004}). However, the level of K⁺ in the overexpressing plants was even lower than those in the WT and OsERF922 RNAi lines (Fig. 6B). The results indicate that the enhanced salt sensitivity is possibly the consequence of inhibition in K⁺ accumulation in OsERF922-overexpressing plants.

In addition to its well-established role in response to abiotic stress, ABA is also involved in defence responses. Therefore, this study measured ABA levels and expression of ABA biosynthesis-related genes. The amount of ABA was about 2-fold more in OsERF922-overexpressing line S6 and 1.4-fold in line S59 than that in the WT plant (Fig. 7A). A decrease in ABA accumulation was observed in the OsERF922 RNAi lines: in the D1 line, the ABA level in the D1 and D30 lines was about one-sixth and one-third of that in WT, respectively (Fig. 7A). 9-cis-Epoxycarotenoid dioxygenase (NCED) catalyses the conversion of 9-cis-epoxycartenoids to xanthoxin, a rate-limiting step in ABA biosynthesis (^{Schwartz et al., 2003}). The expression of OsNCED genes was examined in the transgenic and control plants by qPCR. OsNCED3 and 4 genes were both upregulated in the OsERF922-overexpressing plants and OsNCED4 was suppressed in the RNAi plants, in comparison with the WT plants (Fig. 7B and C). Further, this study examined the expression of Rab16a, an ABA-responsive marker gene (^{Yamaguchi-Shinozaki et al., 1990}). A much higher level of Rab16a transcripts was observed in the OsERF922-overexpressing plants as compared to the WT, whereas the expression of Rab16a was decreased in the RNAi lines (Fig. 7D). The increase of ABA accumulation was consistent with the enhanced expression of the ABA-biosynthesis and -responsive genes in the overexpressing plants.