Transcription factor WRKY46 regulates osmotic stress responses and stomatal movement independently in Arabidopsis

Authors


Summary

Drought and salt stress severely inhibit plant growth and development; however, the regulatory mechanisms of plants in response to these stresses are not fully understood. Here we report that the expression of a WRKY transcription factor WRKY46 is rapidly induced by drought, salt and oxidative stresses. T–DNA insertion of WRKY46 leads to more sensitivity to drought and salt stress, whereas overexpression of WRKY46 (OV46) results in hypersensitivity in soil-grown plants, with a higher water loss rate, but with increased tolerance on the sealed agar plates. Stomatal closing in the OV46 line is insensitive to ABA because of a reduced accumulation of reactive oxygen species (ROS) in the guard cells. We further find that WRKY46 is expressed in guard cells, where its expression is not affected by dehydration, and is involved in light-dependent stomatal opening. Microarray analysis reveals that WRKY46 regulates a set of genes involved in cellular osmoprotection and redox homeostasis under dehydration stress, which is confirmed by ROS and malondialdehyde (MDA) levels in stressed seedlings. Moreover, WRKY46 modulates light-dependent starch metabolism in guard cells via regulating QUA-QUINE STARCH (QQS) gene expression. Taken together, we demonstrate that WRKY46 plays dual roles in regulating plant responses to drought and salt stress and light-dependent stomatal opening in guard cells.

Introduction

Drought and salt stress are major problems that are pervasive and cause economic damage in agricultural production (Epstein et al., 1980; Boyer, 1982). The adaptive responses of plants to these stresses can be conceptually grouped into: responses to direct effects caused by drought and salt, such as osmotic adjustment or ion homeostasis, and responses to indirect effects, including basic defenses in stress damage control and repair, detoxification, as well as growth regulation (Zhu, 2002). The phytohormone abscisic acid (ABA) plays a key role in response to these stresses, mainly through promoting stomatal closure in guard cells and regulating the expression of many genes that may function in dehydration tolerance in both vegetative tissues and seeds (Leung and Giraudat, 1998; Finkelstein et al., 2002; Himmelbach et al., 2003). So far, the signal transduction of the osmotic response has been widely investigated, especially the recently proposed new model of ABA signaling, from ABA perception to the activation of downstream gene expression via specific protein phosphatases, kinases, transcription factors (TFs) and other signaling components (Yamaguchi-Shinozaki and Shinozaki, 2006; Fujii and Zhu, 2009; Fujii et al., 2009; Cutler et al., 2010; Fujita et al., 2011; Soon et al., 2012; Umezawa et al., 2013).

A great number of TFs act in an ABA-dependent or ABA-independent manner in response to dehydration and salt stress (Zhu, 2002; Yamaguchi-Shinozaki and Shinozaki, 2006). The ABA-dependent positive regulators include the bZIP-type ABFs/AREBs (ABRE-binding or ABRE-responsive factors), which recognize the ABA-responsive elements (ABREs) in the promoters of ABA-inducible genes (Choi et al., 2000; Uno et al., 2000; Jakoby et al., 2002), and several other classes, such as AP2/ERF, MYB, NAC and bHLH (also reviewed by Fujita et al., 2011), whereas the key ABA-independent regulators are the DREB (dehydration-responsive element binding/cold-binding factor) proteins, which activate the expression of target genes responsible for osmoprotection and metabolism (Yamaguchi-Shinozaki and Shinozaki, 2006; Lata and Prasad, 2011).

Besides osmotic stress caused by drought and salt, oxidative damage is another well-known stress arising from any disturbance in the balance between the production and scavenging of reactive oxygen species (ROS; Mittler et al., 2004). The ROS scavenging mechanisms have been proven to protect plants against salt, drought or osmotic stresses (Miller et al., 2010). Several ROS-scavenging enzymes were demonstrated to improve photosynthesis under hyperosmotic conditions (Eltayeb et al., 2007; Lu et al., 2007; Tseng et al., 2007). Also, there are many ROS-response regulators that regulate a large set of genes involved in antioxidant mechanisms, such as the protein kinase MKK1, inositol polyphosphate 5–phosphatase7, or the TFs like Zat12 and ERF98 (Teige et al., 2004; Davletova et al., 2005; Kaye et al., 2011; Zhang et al., 2012). Nevertheless, the regulatory mechanisms of ROS homeostasis are largely unknown. On the other hand, ROS also acts as an important signal molecule involved in stomatal closure in the guard cells (Pei et al., 2000; Zhang et al., 2001), indicating its dual role in plant response to stimuli.

Stomatal pores are formed by pairs of guard cells and serve as major gateways for both CO2 influx into plants from the atmosphere and the transpirational water loss of plants (Kim et al., 2010). Regulation of stomatal movement is critical for plant growth and adaptation to environmental stresses. The opening and closing of stomatal pores is mediated by turgor and volume changes in guard cells (Schroeder et al., 2008). ABA promotes stomatal closing under dehydration stress, whereas stomatal opening is induced by light, including blue and red light (Shimazaki et al., 2007). Blue light activates the plasma membrane H+–ATPase, hyperpolarizes the membrane potential with simultaneous apoplast acidification and drives K+ uptake through voltage-gated K+ channels, which further results in a decrease of water potential and subsequent water uptake in guard cells (Kinoshita and Shimazaki, 1999; Briggs and Christie, 2002; Shimazaki et al., 2007). The elevated turgor then increases guard cell volume, and widens the stomatal aperture. Apart from the movement of ions, starch metabolism in the guard cell might also be involved. As the accumulation of positively charged K+ ions in guard cells must be compensated by anions, malate2−, the main anion accumulated in guard cells in most plant species during stomatal opening, is synthesized (Shimazaki et al., 2007). Malate2− can be produced through the degradation of starch in guard cells under blue light (Vavasseur and Raghavendra, 2005). How the starch is then metabolized to produce malate in guard cells remains unclear.

The WRKY domain contains proteins that constitute one of the largest TF families in plants (Eulgem et al., 2000). WRKY proteins always contain one or two domains composed of the conserved amino acid sequence WRKYGQK, together with a zinc-finger-like motif (Ulker and Somssich, 2004). With the domains, they can activate or repress transcription through directly binding to the W–box that has a core sequence (T)(T)TGAC(C/T) in promoters of the target genes (Eulgem et al., 2000). In past years, many studies have revealed that the WRKY TFs form complex webs to modulate a great number of processes in plants, including senescence, seed development, seed dormancy and germination, and stress responses, particularly in response to biotic stresses (Ulker and Somssich, 2004; Rushton et al., 2010). Recent studies proved the role of WRKY TFs in response to abiotic stresses, especially to drought or high salinity (Rushton et al., 2010; Ren et al., 2010; Tao et al., 2011; Jiang et al., 2012; Vanderauwera et al., 2012).

AtWRKY46 was previously reported to participate in basal resistance against the pathogen Pseudomonas syringae (Hu et al., 2012), as well as in the regulation of aluminum (Al)-induced malate secretion in Arabidopsis (Ding et al., 2013). In the present study, we found that WRKY46 was rapidly induced by drought, salt and hydrogen peroxide. We demonstrated that WRKY46 regulates a set of genes in cellular osmoprotection and oxidative detoxification under drought and salt stress. On the other hand, WRKY46 was also involved in stomatal opening via modulating starch metabolism in guard cells.

Results

WRKY46 is an early response factor induced by drought, salt and hydrogen peroxide

In a primary experiment, we isolated five WRKY genes that were significantly and quickly responsive to oxidative stress induced by cadmium from 33 WRKY members through RT-PCR (Figure S1a). WRKY46 could respond rapidly to cadmium stress, even after just 5 min, and its transcripts reached the highest level after 30 min (Figure S1b). To investigate whether it is also responsive to other abiotic stresses, we used real-time qPCR to examine transcript abundance in mock-treated plants and those subjected to a variety of stresses for 30 min (Figure 1a). We found that WRKY46 was more responsive to hydrogen peroxide, high salinity and drought, as well as salicylic acid (SA), compared with cadmium, wounding and other treatments, and was somewhat downregulated by heat and methyl jasmonate (MeJA) treatments. The induction by SA has been reported previously (Hu et al., 2012). Results indicated that WRKY46 plays a role in response to abiotic stresses in addition to biotic stresses. A detailed time-course analysis showed that WRKY46 could rapidly respond to hydrogen peroxide, drought and salt, and that its transcriptional level reached a maximum in 30 min (Figure 1b). In contrast, WRKY46 was weakly induced by ABA, suggesting that it responded to drought and salt stress possibly in an ABA-independent manner. To confirm whether the protein level of WRKY46 is corresponding to its mRNA abundance under stress, we generated the transgenic line (wrky46 background) expressing WRKY46 fused with a GFP tag under the native promoter (1.65 kb). A western blot assay demonstrated that WRKY46 protein was significantly induced in an hour under drought, salt and hydrogen peroxide treatment (Figure 1c), confirming that it is an early factor responding to drought, salt and oxidative stress.

Figure 1.

Expression of WRKY46 in response to abiotic stresses. (a) qRT-PCR analysis for WRKY46 gene expression under multiple treatments: 2–week-old wild-type (Col–0) seedlings with or without the indicated treatments for 30 min; values are means ± SEs (n = 3). (b) Time-course expression of WRKY46 in response to abiotic stresses. Two-week-old Col–0 seedlings were grown with or without treatment for the indicated time. Values are means ± SEs (n = 3). (c) Accumulation of WRKY46-GFP fused protein under drought, salt and hydrogen peroxide treatments. Anti-GFP was used to detect the abundance of WRKY46 protein. Ponceau S indicates Ponceau staining of the loaded proteins. Three independent repeats were performed with similar results.

WRKY46 T–DNA insertion mutants are sensitive to drought and salt stress

To analyze the function of WRKY46 in stress responses, the T–DNA inserted mutant wrky46 (SAIL_1230_H01) that carries the T–DNA insertion in the last exon of the WRKY46 gene was used (Figure S2). Three-day-old wild-type (Col–0) and wrky46 seedlings were tested for tolerance to osmotic and salt stress. It was evident that wrky46 mutants had more inhibition on shoot growth on the medium containing 100 mm NaCl or 200 mm mannitol, compared with Col–0, but showed no significant difference in root growth with Col–0 (Figure 2a). The corresponding biomass was consistent with the phenotypes (Figure 2b). With the increase of NaCl concentration to 150 mm, wrky46 mutants showed more severe chlorosis and lower survival rates than Col–0 (Figure S3a,b). These indicate that the wrky46 mutant is less tolerant to osmotic and salt stress. The native promoter-complemented lines of wrky46 exhibited, in varying degrees, more tolerance to osmotic and salt treatment than did wrky46 mutants (Figures 2d and S3c–e), owing to the varying expression of WRKY46 in the different lines (Figures 2c and S2), indicating that the transformation functionally complemented the WRKY46 mutation. Besides, another T–DNA inserted line, wrky46–2 (SALK_134310C), was also phenotypically similar to the above wrky46 line when exposed to drought and salt stresses (Figure S4). We further tested the growth of Col–0 and wrky46 plants, as well as the complemented lines, under drought and salt treatment in soil culture, and obtained similar results to those obtained with agar medium (Figure 2e). These results suggest that WRKY46 is required for tolerance to drought and salt stress.

Figure 2.

The WRKY46 T–DNA insertion mutant is sensitive to drought and salt stress. (a) Growth of wild-type and wrky46 seedlings on agar medium with or without 100 mm NaCl and 200 mm mannitol, for 2 weeks. Scale bars: 1 cm. (b) Relative root and shoot biomass of Col–0 and wrky46 seedlings on agar medium under the indicated treatments, compared with normal conditions; FW, fresh weight. Values are means ± SEs (n = 4, *< 0.01). (c) The identification of wrky46 mutant and wrky46 complementation lines. WRKY46 expression was detected in Col–0, wrky46 mutant and wrky46 complementation lines under control and drought treatment for 30 min. Values are means ± SEs (n = 3). (d, e) Relative shoot biomass of Col–0, wrky46 and wrky46 complemented lines on agar medium (c) or in soil culture (d) under the indicated treatments, compared with normal conditions; DW, dry weight. Values are means ± SEs (n = 4, *< 0.01).

Overexpression of WRKY46 leads to hypersensitivity to drought and salt stress in soil culture, but not on agar plates

To further investigate the role of WRKY46 in dehydration tolerance, two WRKY46 overexpressor lines (OV46 and OV46–2) driven by CaMV35 promoter were used (Figure 3g). The wild type, the wrky46 mutant and the OV46 plants grew similarly under normal conditions (Figure 3a,b). Surprisingly, the OV46 line was hypersensitivitive (as compared with Col–0 and the wrky46 mutant) when exposed to dehydration and high salinity for 10 and 7 days in soil culture, respectively (Figure 3a–c). After drought treatment, plants were re-watered and cultured for one more week. The OV46 line displayed a significantly lower survival rate than Col–0 (Figure 3a,c). This might result from a more rapid loss of water in OV46 plants than in Col–0 plants (Figure 3d). By contrast, the OV46 line showed more resistance to NaCl and mannitol on agar plates, compared with wild-type seedlings (Figures 3e and S5). All these data indicate that different biological mechanisms might be involved in the OV46 line grown on agar plates in a closed environment (plates were sealed with Parafilm) and grown in soil culture with an open environment.

Figure 3.

Phenotypes of WRKY46-overexpressing lines under stress treatment. (a) Growth of different genotypes (as indicated) under normal and dehydration conditions for 10 days. The photographs were taken after the plants were re-watered for another 7 days. (b) Growth of indicated genotypes under normal or high-salinity treatment for 1 week. (c) Survival rates of indicated genotypes upon re-watering for 1 week after dehydration treatment. Three independent repeats were performed, each with about 40 plants. Values are means ± SEs (n = 3, *< 0.01). (d) Water loss from 0.5 g of detached leaves, from Col–0, wrky46 and the OV46 line, was measured at different times with triple replicates. Values are means ± SEs (n = 3, *< 0.01, **< 0.001). (e) Growth of Col–0 and the OV46 line on agar medium containing 200 mm mannitol or 100 mm NaCl. (f) Relative shoot biomass of different genotypes grown on agar medium with the indicated treatments, compared with normal conditions. Values are means ± SEs (n = 4, *< 0.01). (g) Identification of the OV4646 line. WRKY46 expression was detected in Col–0 and WRKY46-overexpressing lines under control and drought treatment for 30 min. Scale bars: (a, e) 1 cm; (b) 2 cm.

ABA-induced stomatal closing is impaired in the OV46 line

As water loss mainly depends on stomatal regulation, we found that the OV46 line showed relatively greater stomatal apertures than Col–0 when exposed to light (Figure 4a,b). When treated with ABA, the stomata of the OV4646 line opened more widely than Col–0 (Figure 4a,b), whereas there was no significant difference between Col–0 and wrky46 (Figure 4b). This implies that stomatal closure is affected in the OV46 line; however, no significant difference in stomatal density and guard cell size was observed between OV46 and Col–0 (Figure 4a,c), indicating that the more rapid water loss observed in the OV46 line should mainly be ascribed to impaired stomatal closure.

Figure 4.

Response of WRKY46-overexpressing plants to ABA-mediated stomatal closure. (a, b) Epidermal peels of indicated genotypes were treated with or without ABA for 2 h after stomatal pre-opening under light for 3 h, and the stomatal aperture was measured by microscope. Values are means ± SEs (n = 50, *< 0.01). (c) Stomatal distribution in abaxial epidermis of leaves from different genotypes (as indicated). Values are means ± SEs (n = 10). (d) Two-week-old Col–0 and OV46 seedlings, with or without dehydration treatment, were used to detect the ABA content by ELISA. Values are means ± SEs (n = 3). (e) ROS accumulation in guard cells of Col–0 and OV46 line, indicated with the fluorescence dye 2′,7′–dichlorodihydrofluorescein diacetate (H2DCF-DA). (f) Quantification of ROS levels in guard cells of indicated genotypes. The Fluorescent intensity in Col–0 before ABA treatment was taken as 100%. Values are means ± SEs (n = 30, *< 0.05). Scale bar: 5 μm.

The phytohormone ABA induces stomatal closure (Schroeder et al., 2008); however, there was no significant difference in the endogenous ABA level between OV46 and Col–0 (Figure 4d), indicating that the impaired stomatal closure of OV46 is independent of ABA. As H2O2 promotes leaf stomatal closure acting downstream of ABA (Pei et al., 2000), H2O2 accumulation in guard cells was measured by a fluorescence dye, 2′,7′–dichlorodihydrofluorescein diacetate (H2DCF-DA). H2O2 accumulation was less in OV46 than in Col–0 (Figure 4e,f), suggesting that a reduction of H2O2 in guard cells probably underlies the impaired stomatal closure in the OV46 line.

WRKY46 is expressed in guard cells and involved in stomatal opening

As drought can rapidly induce the expression of WRKY46, it is self-contradictory for plants to increase WRKY46 transcript levels in guard cells, because overexpression of WRKY46 results in a reduction of H2O2 in guard cells, and subsequently causes impaired stomatal closure, but plants need quick control of such stomatal movement when exposed to dehydration stress. Thus, we isolated both mesophyll cell and guard cell protoplasts to examine whether WRKY46 is expressed in guard cells. The genes At4 g26530 and KAT1, predominantly expressed in mesophyll cells and guard cells, respectively, were used as controls. The real-time qPCR revealed that WRKY46 had a higher transcriptional level in guard cells than that in mesophyll cells (Figure 5a). We further generated GUS reporter lines using WRKY46 native promoter and confirmed its expression in guard cells (Figure 5b). Moreover, WRKY46 is also expressed in hypocotyls, and in the vascular tissues of cotyledon and root in seedlings (Figure 5b); however, GUS expression was significantly induced by dehydration in leaves and roots of plants, but not in guard cells (Figure 5c). Further GUS quantification assays in roots, shoots, epidermal strips and isolated guard cells confirmed these results (Figure 5d). These suggested that different regulatory mechanisms of WRKY46 may exist between guard cells and other WRKY46-expressing tissues.

Figure 5.

Expression of WRKY46 in guard cells. (a) Comparison of WRKY46 transcript level in wild-type guard cell and mesophyll cell protoplasts. The guard cell-specific gene KAT1 and mesophyll cell-specific gene At4 g26530 were used as controls; GCP, guard cell protoplast; MCP, mesophyll cell protoplast. Values are means ± SEs (n = 3). (b) GUS staining of WRKY46 localization in guard cell (i), seedling (ii), hypocotyl (iii) and root (iv). Scale bars: (i) 10 μm; (ii, iii) 200 μm; (iv) 100 μm. (c) GUS staining of WRKY46 in leaves (v and viii), roots (vi and ix) and guard cells (vii and x), with or without dehydration treatment (CK, control check; v–vii; treatment, viii–x). Scale bars: (v and viii) 2 mm; (vi and ix) 100 μm; (vii and x) 10 μm. (d) GUS activity measured in protein extracts from different tissues, with or without dehydration treatment. Activity units are given in nmol methyl-umbelliferone (μg protein)−1 min−1. Values are means ± SEs (n = 4, *< 0.01).

As the stomata of the OV46 line opened more widely than Col–0 when exposed to light (Figure 4a,b), we speculated that WRKY46 is involved in light-regulated stomatal opening. Expression analysis under dark–light cycles revealed that WRKY46 could be significantly induced by light (Figure 6a), indicating that it is a light-responsive factor. The OV46 line exhibited wider stomatal apertures in both dark and light conditions, whereas there was no significant difference between Col–0 and wrky46 mutant plants (Figure 6b). Furthermore, OV46 showed higher stomatal conductance in both dark and light conditions, and wrky46 had lower stomatal conductance than Col–0 after 1 h of light, but no significant difference was observed between wrky46 and Col–0 in the dark and after more than 1 h of light (Figure 6c). These results suggest that WRKY46 functions in light-regulated stomatal opening.

Figure 6.

WRKY46 is involved in light-dependent stomatal opening. (a) Expression of WRKY46 from wild-type leaves under a 12–h dark/12–h light cycle. D1–D12 indicate the hours in darkness; L1–L9 indicate the hours in light. Values are means ± SEs (n = 3). (b) Effect of light on stomatal opening in Col–0, wrky46 and the OV46 line. Epidermal peels were performed under dark for 2 h, and then exposed to light for 3 h. The stomatal aperture was measured by microscope. Values are means ± SEs (n = 50, *< 0.01). (c) The stomatal conductance of Col–0, wrky46 and the OV46 line under both dark and light conditions. Values are means ± SEs (n = 15, *P < 0.01, **P < 0.001).

WRKY46 regulates a group of genes involved in cellular osmoprotection and redox homeostasis

Expression profile analysis using the Affymetrix Arabidopsis ATH1 Genome Array was performed to identify WRKY46 target genes. Two-week-old wild-type and OV46 seedlings, with or without dehydration treatment for 1 h, were used in a Genechip experiment. We found that among ~24 000 Arabidopsis genes, the transcription levels of 297 genes were altered significantly (with more than a twofold change; P < 0.05) in OV46 compared with Col–0 without treatment (Table S1). Half of these genes were upregulated (146 genes), whereas the others were downregulated (151 genes), in OV46. When under dehydration treatment, there were 314 genes significantly altered, 212 and 102 of which were up- and downregulated, respectively, in OV46 (Table S2). Among the upregulated genes, however, there were few involved in ABA signaling, consistent with the expression of WRKY46 in response to drought and salt stress independently of ABA (Figure 1b). In spite of this, a group of LEA (late embryogenesis abundant protein) genes that function in osmotic regulation and/or protection of cellular structure under dehydration conditions (Ingram and Bartels, 1996) were significantly altered in OV46 (Figure 8), indicating that WRKY46 plays a role in cellular osmoprotection. Using the e–Northern Expression Browser tool (Toufighi et al., 2005), we found that among these upregulated genes (146 and 212) 94 and 105 genes, respectively, were significantly induced (by threefold or more) in at least one time point during osmotic, salinity or drought stresses (Figure 7a). In particular, more than 45% of these 94 and 105 genes were also responsive to oxidative stress (Figure 7a). These data suggested that WRKY46 might function in the regulation of cellular redox and ROS homeostasis to control the oxidative damage caused by drought and salt stress. Because WRKY46 is responsive to SA and biotic stress (Hu et al., 2012), we found that most of WRKY46-regulated genes were involved in biotic stress responses (genes upregulated by threefold or more; Figure 7b). Nevertheless, there was a large overlap between the genes in response to abiotic stress (osmotic, salt and drought stresses) and genes in response to biotic stress among the upregulated genes, and half of the overlapping genes were responsive to oxidative stress (Figure 7b), indicating that a common regulatory mechanism conferred by WRKY46 is presented in response to biotic and abiotic stresses, especially in the regulation of redox homeostasis.

Figure 7.

Venn diagrams showing the overlap between WRKY46 upregulated genes in response to different stresses. (a) Comparison of WRKY46-upregulated genes under normal and dehydration conditions in response to osmotic, salt, drought and oxidative stresses, by using the e–Northern Expression Browser tool; Osmo, osmotic stress; Salt, salt stress; Drou, drought stress; Oxid, oxidative stress; Dehy, dehydration treatment. (b) Comparison of WRKY46-upregulated genes under normal and dehydration conditions in response to abiotic, biotic and oxidative stresses. Abiot, abiotic stress (osmotic, salt and drought stress); Biot, biotic stress; Oxid, oxidative stress.

Genes encoding LEA proteins and antioxidative enzymes, such as MDHAR, GSTF14, TRX5 or Peroxidase54 (Noctor and Foyer, 1998; Schurmann and Jacquot, 2000; Laloi et al., 2004; Passardi et al., 2005), were further selected for qPCR confirmation (Figure 8a). The time-course expression analysis showed significant differences in these genes in transcriptional level among the wrky46 mutant, Col–0 and the OV4646 line, indicating that these genes are indeed regulated by WRKY46. These results also confirmed the validity of the chip experiment. Promoter analysis revealed that W–box, the WRKY TF binding site, was greatly enriched in the promoters of these genes (Figure S6). Further analysis using chromatin immunoprecipitation (ChIP)-based qPCR demonstrated that WRKY46 regulated these genes by directly binding to their promoters under dehydration stress (Figure 8b).

Figure 8.

WRKY46 regulates a set of genes involved in cellular osmoprotection and redox homeostasis. (a) Expression of WRKY46 downstream genes in 2–week-old Col–0, wrky46 and OV46 seedlings under drought treatment for different hours. Values are means ± SEs (n = 3). (b) WRKY46 directly binds to the promoters of its downstream genes under normal or drought conditions by ChIP-qPCR. STOP1 was used as a negative control. Values are means ± SEs (n = 3).

To confirm the role of WRKY46 in the regulation of ROS homeostasis, 3–day-old wrky46, Col–0 and OV46 seedlings were stained with the fluorescent dye H2DCF-DA to detect ROS content. All the seedlings showed no obvious fluorescence signal under normal conditions (Figure S7a,b), whereas with NaCl treatment ROS accumulation in cotyledons was decreased in the order wrky46 > Col–0 > OV46 (Figure S7a,b), indicating that WRKY46 plays an important role in controlling ROS homeostasis under salt stress. Levels of malondialdehyde (MDA) also showed no obvious difference among three genotypes under normal conditions, but after dehydration treatment followed the same order as for ROS (Figure S7c). These results further demonstrated that WRKY46 is able to regulate ROS homeostasis and reduce the oxidative damage caused by drought and salt stress.

WRKY46 regulates starch metabolism in guard cells

As WRKY46 is involved in stomatal movement, we analysed potential target genes from the microarray data and only found the QQS (qua-quine starch) gene, which encodes a PUF (protein with unknown functions) with no known sequence homologs or predicted structural motifs, to be highly expressed in guard cells (Figure S8). QQS was reported to regulate starch metabolism (Li et al., 2009). The qPCR using specific controls confirmed its preferential expression in guard cells (Figure 9a). Moreover, QQS was also responsive to light (Figure 9b) and indeed regulated by WRKY46, with lower and higher expression levels in wrky46 and OV46, respectively, compared with that in Col–0 (Figure 9c). Furthermore ChIP-qPCR revealed that WRKY46 was enriched in the QQS promoter (Figure 9d), suggesting a direct regulation between WRKY46 and QQS.

Figure 9.

WRKY46 regulates light-dependent stomatal opening via modulating starch metabolism in guard cells. (a) Comparison of QQS transcript level in wild-type guard cell and mesophyll cell protoplasts preparations. KAT1 and At4 g26530 were used as controls. Values are means ± SEs (n = 3). (b) Expression of QQS under a 12–h dark/12–h light cycle. Values are means ± SEs (n = 3). (c) Expression of QQS in Col–0, wrky46 and OV46 seedlings. Values are means ± SEs (n = 3). (d) WRKY46 directly binds to QQS promoter in vivo by ChIP-qPCR analysis. The 340–bp length of QQS promoter is indicated, and two columns indicate TGAC core sequences present in the promoter; the black line indicates a promoter region detected by qPCR; TSS, transcriptional start site. STOP1 was used as a negative control. Data without bars indicate not detectable. Values are means ± SEs (n = 3). (e,f) Comparison of starch content (f) and malate content (g) in the guard cells of Col–0, wrky46 and the OV46 line under dark and light conditions. Values are means ± SEs (n = 3, *< 0.01). (g) Epidermal peels from Col–0, wrky46 and the OV46 line were performed under dark for 2 h, and then exposed to blue light for 2 h. The stomatal aperture was measured by microscope. Values are means ± SEs (n = 50, *< 0.05).

Starch metabolism plays an important role in stomatal movement, as the main anion of malate2− in guard cells accumulated from light-dependent starch degradation during stomatal opening (Vavasseur and Raghavendra, 2005). It has been demonstrated that QQS modulated starch metabolism, and that the downregulation of QQS resulted in elevated starch content in Arabidopsis leaves (Li et al., 2009). We then measured starch and malate contents in guard cells of Col–0, wrky46, wrky46 complemented line and OV46 under dark and light conditions. OV46 contained significantly less starch under both dark and light conditions, whereas wrky46 had a higher starch content than Col–0 under light (Figure 9e). There was no significant difference between Col–0 and the complemented line (Figure 9e). Inversely, OV46 accumulated more malate in guard cells, whereas wrky46 had less malate than Col–0 and the complemented line under light (Figure 9f). Moreover, we also measured sucrose content in guard cells, as sucrose plays an important role in maintaining stomatal apertures via osmotic regulation, and one of its sources may be the degradation of starch in guard cells (Reddy and Das, 1986; Tallman and Zeiger, 1988), but there was no significant difference among them. Because blue light can greatly promote starch degradation and malate formation in guard cells (Shimazaki et al., 2007), we further compared the stomatal aperture of Col–0, wrky46 and OV46 exposed to blue light under a background of red light for 2 h. The stomata of Col–0 opened further than those of wrky46, but less than those of OV46 (Figure 9 g). These results suggest that the altered accumulation of starch and malate in guard cells may interpret the role of WRKY46 in stomatal opening.

Discussion

WRKY46 functions in abiotic and biotic responses

Multiple treatments revealed that WRKY46 was more responsive to dehydration, high salinity and hydrogen peroxide than other abiotic treatments (Figure 1a). Both the mRNA and protein of WRKY46 accumulated rapidly under drought, salt and hydrogen peroxide treatment (Figure 1b,c), indicating it is an early response factor when plants are exposed to these stresses. By contrast, the protein level of WRKY46 was not very similar to its transcriptional level under time course treatment, as the high protein abundance was maintained for several hours, whereas the transcriptional level declined from the highest level after 30 min (Figure 1b,c), implying that WRKY46 was further regulated at the post-translational level and that protein stability was tightly controlled under stress treatment.

Phenotypical analysis demonstrated that WRKY46 T–DNA insertion mutants were less tolerant to dehydration and salt stress (Figures 2 and S3 and S4); however, the OV46 line showed decreased tolerance under these stresses in soil culture (Figure 3a–c), but was more tolerant to the treatments on agar plates (Figures 3e,f and S5), where the stresses were simulated by the addition of NaCl and mannitol into the agar medium. Because WRKY46 plays a role in stomatal regulation (Figures 4 and 6), the inverse phenotypes of OV46 were possibly ascribed to the different stress conditions: in soil culture, the OV46 plants show a higher water loss rate than Col–0 (Figure 3d), whereas in airtight agar plates (sealed with Parafilm), the transpiration of seedlings is almost negligible (Verslues et al., 2006).

Microarray analysis revealed that more than half of WRKY46 upregulated genes were responsive to osmotic, salt or drought stresses (Figure 7a), consistent with the phenotypes of wrky46 mutants in response to these stresses. Importantly, about half of those genes were also responsive to oxidative stress (Figure 7a), suggesting that WRKY46 probably functions in the regulation of cellular redox and ROS homeostasis to control oxidative damage caused by drought and salt stress. This was further physiologically demonstrated, as the wrky46 mutant and OV46 line accumulated altered ROS and MDA content under stress treatment (Figure S7). Moreover, WRKY46 also upregulated a group of LEA genes that play important roles in dehydration tolerance (Figure 8; Ingram and Bartels, 1996), indicating a role of WRKY46 in the modulation of osmoprotection.

Because crops in the field are always exposed to a mix of biotic and abiotic stresses, many investigations have focused on the interaction of these different stresses (Fujita et al., 2006; Atkinson and Urwin, 2012). As controlling a wide range of downstream events, TFs provide one of the greatest opportunities for conferring multiple stress tolerance transgenically (Atkinson and Urwin, 2012). A number of TFs were reported in controlling both biotic and abiotic responses, and were used for the study of the interaction between different hormone signaling pathways. For example, MYC2 (also called JIN1), which was first isolated in ABA signaling and dehydration tolerance (Abe et al., 2003), is a positive regulator of jasmonic acid (JA) signaling, and is a key repressor of the SA pathway (Anderson et al., 2004; Laurie-Berry et al., 2006). It thus may act as a central regulator by which ABA controls the biotic stress signaling pathway (Asselbergh et al., 2008). As WRKY46 was also reported in SA signaling and biotic responses (Hu et al., 2012), it is interesting to elucidate how WRKY46 acts to influence both biotic and abiotic signaling responses. In fact, most of WRKY46 upregulated genes were responsive to biotic stress (Figure 7b). Interestingly, a large overlap was observed between the genes responding to abiotic stress (osmotic, salt and drought stresses) and genes responding to biotic stress among the upregulated genes, and half of the overlapping genes were responsive to oxidative stress (Figure 7b). This suggests that WRKY46-regulated biotic and abiotic responses share a common molecular mechanism during the specific signal transduction, especially in the modulation of cellular redox homeostasis. This widens our understanding on the coordination of plant responses to different biotic and abiotic stresses.

WRKY46 is involved in stomatal opening

Overexpression of WRKY46 resulted in hypersensitivity to drought and salt stress, and caused a higher water loss rate, which led us to suppose that WRKY46 is involved in stomatal regulation. ABA-induced stomatal closing was impaired in line OV46 (Figure 4a,b), and this was more or less ascribed to the decreased ROS accumulation in the OV46 guard cells (Figure 4e,f), suggesting that WRKY46 plays a role in the inhibition of stomatal closing. But the main function of WRKY46 in stomatal regulation seems not to only inhibit stomatal closing, because it is contradictory for plants to induce the expression of WRKY46 under dehydration conditions to inhibit stomatal closing.

As WRKY46 was expressed in guard cells (Figure 5a,b), the phenotypes of the OV46 line in the stomatal response did not result from the ectopic expression of WRKY46 in guard cells driven by the CaMV35 promoter. More importantly, dehydration treatment induced WRKY46 expression in other tissues, but not in guard cells (Figure 5c,d), suggesting a potentially unknown role of WRKY46 in guard cells. Further analysis demonstrated that WRKY46 was responsive to light (Figure 6a), and that the stomata of OV46 opened more widely, especially under light conditions (Figure 6b,c), indicating that WRKY46 was possibly involved in stomatal opening. Microarray and qPCR revealed that the QQS gene that regulates starch metabolism (Li et al., 2009) was significantly upregulated by WRKY46 (Figure  9c; Table S1). Moreover, QQS was expressed in guard cells and directly targeted by WRKY46 (Figure 9a,d). It has been demonstrated that light-dependent starch degradation plays an important role in stomatal opening by promoting the synthesis of sucrose and malate in guard cells (Vavasseur and Raghavendra, 2005; Shimazaki et al., 2007). We found that the OV46 line accumulated less starch but more malate in guard cells (Figure 9e,f), and subsequently promoted stomatal opening (Figure 6b). In contrast, wrky46 accumulated more starch and less malate in guard cells (Figure 9e,f). Although WRKY46 was reported to directly regulate the expression of ALMT1 (Ding et al., 2013), which encodes a plasma-localized malate transporter (Hoekenga et al., 2006), ALMT1 is specifically expressed in Arabidopsis roots rather than in shoots (Kobayashi et al., 2007), suggesting that the increased malate accumulation in OV46 guard cells is not a result of altered ALMT1 expression.

The time-course analysis of stomatal conductance revealed an obvious difference between wrky46 and Col–0 in the first hour after exposure to light, but no significant difference after 1 hour (Figure 6c), suggesting that the stomata of wrky46 opens slower than Col–0 when exposed to light. Because other anions such as Cl and inline image are also taken up by guard cells to compensate for positively charged K+ during stomatal opening (Guo et al., 2003; Shimazaki et al., 2007), the guard cells may then accumulate more of these anions to make up for the deficiency of malate in the wrky46 mutant, and subsequently open the stoma as wide as those in Col–0. This may be the reason why there was no obvious difference between wrky46 and Col–0 in stomatal apertures under prolonged light conditions (Figure 6b,c), as well as the rate of water loss in detached leaves (Figure 3d). Furthermore, we can observe a significant difference between wrky46 and Col–0 in stomatal aperture under blue light conditions (Figure 9g). This is because blue light can greatly promote starch degradation and malate formation, and cause rapid stomatal opening (Shimazaki et al., 2007), but the guard cells of the wrky46 mutant are impaired in blue light-dependent starch degradation. Such a phenotype of wrky46 is similar to a phosphoglucomutase-deficient mutant that has no starch accumulation in guard cells (Lasceve et al., 1997). All these results indicated that WRKY46 regulates light-dependent starch degradation during stomatal opening.

In conclusion, we found that the WRKY transcription factor WRKY46 plays a dual role in stress responses and stomatal opening (Figure S9). WRKY46 is rapidly induced upon water stress, and it directly or indirectly regulates a number of genes involved in cellular osmoprotection and oxidative detoxification; the stress does not induce the expression of WRKY46 in guard cells, in which WRKY46 is regulated by light and modulates starch metabolism and ROS levels to control stomatal opening as well as inhibit their closing. It is interesting to note that in Solanum tuberosum (potato), the expression levels of NAD+-GAPDH, H+–ATPase, PEP carboxylase and the inward-rectifying K+ channel (KST1), all of which are involved in stomatal opening, were downregulated simultaneously in a guard cell-specific manner under drought stress (Kopka et al., 1997), indicating that different regulatory mechanisms are needed for plants in guard cells to adapt well to unfavorable environmental conditions. Although many genes involved in stomatal opening are downregulated by ABA and dehydration stress (Shimazaki et al., 2007), it may be necessary for plants to maintain the expression of some genes that take part in stomatal opening in guard cells, like WRKY46, to keep the stomatal aperture at a specific level, and subsequently balance CO2 influx and water loss under water-deficit conditions.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana plants were grown in an environmentally controlled growth chamber programmed for a 16–h light/8–h dark cycle, with daytime temperature of 24°C and a night temperature of 21°C. The T–DNA insertion mutant wrky46 (SAIL_1230_H01; Col–0 background) and wrky46-2 (SALK_134310C; Col–0) were obtained from the Arabidopsis Biological Reso-urce Center (ABRC, http://abrc.osu.edu). Pro46-WRKY46-GFP and Pro46-GUS were transformed into wrky46 and Col–0, respectively (Ding et al., 2013). The WRKY46 overexpressor line was obtained from Prof. Diqiu Yu (Chinese Academy of Sciences, Kunming, China). Plants were grown on a modified half-strength MS medium or solution (pH 5.6), as described previously (Ding et al., 2013).

Stress treatment

For gene expression analysis, 2–week-old seedlings exposed to different treatments were used. For phenotypical analysis, 3–day-old seedlings grown on half-strength MS medium (0.8% agar) were transferred to new agar plates containing different concentrations of NaCl and mannitol. For more details, please see Appendix S1.

Water loss assay and ABA content determination

For the water loss assay, rosette leaves of plants (about 0.5 g of leaves from the same stage of different plants) growing under normal conditions for 4 weeks were detached and weighed for the indicated times. For the ABA content assay, 0.5 g of 2–week-old seedlings with or without dehydration treatment were used. The measurement of ABA content was performed by using an enzyme-linked immunosorbent assay kit provided by Prof. BaoMin Wang (China Agricultural University, Beijing; see Appendix S1).

Stomatal movement assay

ABA-induced stomatal closing assays were performed as described previously, with slight modification (Pei et al., 1997; Appendix S1).

Determination of ROS accumulation and MDA content

For the ROS accumulation assay in guard cells, prepared epidermal peels with or without ABA treatment were loaded with 50 μm 2,7–dichlorofluorescin diacetate (H2DCF-DA; Sigma-Aldrich, http://www.sigmaaldrich.com) for 15 min, as described previously (Pei et al., 2000). For ROS accumulation and MDA content in seedlings, please see Appendix S1.

GUS staining and activity assay

For GUS staining and quantification, the transgenic plants with or without stress treatment were used as described previously (Ding et al., 2013; see Appendix S1).

Isolation of guard cell and mesophyll cell protoplasts

Guard cell protoplasts were isolated according to Pandey et al. (2002), and mesophyll cell protoplasts were isolated mainly according to Yoo et al. (2007). For more detail, please see Appendix S1.

Starch and malate determination in guard cells

Guard cell contents were extracted using the ‘oil well technique’ as previously described (Outlaw and Manchester, 1979). Total starch and sucrose content was quantified using a commercial starch assay kit (catalog no. E0207748; R–Biopharm AG, http://www.r-biopharm.com), according to instructions in the kit protocol. Malate concentrations were measured according to the enzymatic method described in Delhaize et al. (1993); please see Appendix S1.

Gene expression analysis

Real-time qPCR was performed as previously described (Ding et al., 2013). Each experiment was repeated independently three times. Transcript levels of each mRNA were determined and normalized with the level of UBQ10 and ACT2 mRNAs using the ΔCt method (Czechowski et al., 2005; Schmittgen and Livak, 2008). UBQ10 was used to detect the samples under stress treatment; ACT2 was used to detect samples under normal conditions (Czechowski et al., 2005). Gene-specific primers are listed in Table S3.

The Arabidopsis eFP Browser provided by BAR (The BioArray Resource for Arabidopsis Functional Genomics; http://bbc.botany.utoronto.ca) was employed to show the heat map of QQS expression in guard cells.

Microarray analysis

The microarray experiments were performed at ShanghaiBio Corporation. Total RNA was isolated using the RNase kit (Qiagen, http://www.qiagen.com) from three replicate samples of the 2–week-old wild-type and OV46 seedlings under normal and dehydration treatment for 1 h. For detailed experimental and data processing, please see Appendix S1.

Protein extraction and western blot assay

Proteins from 2–week-old transgenic seedlings expressing WRKY46 fused with a GFP tag driven by its native promoter under different stress treatments were extracted, and were further determined using the Bio-Rad (http://www.bio-rad.com) protein assay kit, with BSA as a standard. A western blot assay was performed by using the GFP multiclonal antibody anti-GFP (abcam, http://www.abcam.com). Please see Appendix S1.

ChIP-qPCR analysis

The EpiQuik Plant ChIP kit (Epigentek, http://www.epigentek.com) was used to perform the ChIP assays (Ding et al., 2013). For more details, please see Appendix S1.

Accession numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: WRKY46 (At2g46400), KAT1 (At5g46240), LEA18 (At2g35300), COR27 (At5g42900), TRX5 (At1g45145), GSTF14 (At1g49860), POX54 (At5g06730), MDHAR (At3g09940) and QQS (At3g30720). The microarray data are deposited at NCBI with access number GSE49418.

Acknowledgements

This research was supported by the Program for Innovative Research Team in Universities (IRT1185) and the Fundamental Research Funds for the Central Universities.

Ancillary