WRKY transcription factors (TFs) are key regulators of many plant processes, including the responses to biotic and abiotic stresses, senescence, seed dormancy and seed germination. For over 15 years, limited evidence has been available suggesting that WRKY TFs may play roles in regulating plant responses to the phytohormone abscisic acid (ABA), notably some WRKY TFs are ABA-inducible repressors of seed germination. However, the roles of WRKY TFs in other aspects of ABA signalling, and the mechanisms involved, have remained unclear. Recent significant progress in ABA research has now placed specific WRKY TFs firmly in ABA-responsive signalling pathways, where they act at multiple levels. In Arabidopsis, WRKY TFs appear to act downstream of at least two ABA receptors: the cytoplasmic PYR/PYL/RCAR-protein phosphatase 2C-ABA complex and the chloroplast envelope–located ABAR–ABA complex. In vivo and in vitro promoter-binding studies show that the target genes for WRKY TFs that are involved in ABA signalling include well-known ABA-responsive genes such as ABF2, ABF4, ABI4, ABI5, MYB2, DREB1a, DREB2a and RAB18. Additional well-characterized stress-inducible genes such as RD29A and COR47 are also found in signalling pathways downstream of WRKY TFs. These new insights also reveal that some WRKY TFs are positive regulators of ABA-mediated stomatal closure and hence drought responses. Conversely, many WRKY TFs are negative regulators of seed germination, and controlling seed germination appears a common function of a subset of WRKY TFs in flowering plants. Taken together, these new data demonstrate that WRKY TFs are key nodes in ABA-responsive signalling networks.
WRKY transcription factors (TFs) have mostly been studied with regard to their roles in regulating plant responses to pathogens. They are key regulators, both positive and negative, of the two partly interconnected branches of plant innate immunity: microbe/pathogen-associated molecular pattern-triggered immunity (MTI/PTI) and effector-triggered immunity (ETI) (Rushton et al., 2010). More recently, it has become clear that WRKY TFs also play key roles in responses to abiotic stresses such as cold and high temperature, water stress, high CO2 levels, high ozone concentrations and salt stress (Rushton et al., 2010). Many of these stress responses are regulated by the plant hormone abscisic acid (ABA). Two of the first identified WRKY TFs (AfWRKY1/ABF1 and AfWRKY2/ABF2) were implicated in the regulation of gene expression during seed germination (Rushton et al., 1995), a process that is regulated jointly by the plant hormones gibberellin (GA) and ABA. Although notable progress has been made determining the involvement of WRKY TFs in seed germination (Zhang et al., 2004, 2009; Zou et al., 2004, 2008; Xie et al., 2005, 2006), the role of WRKY TFs in ABA signalling in general has remained obscure. This situation has now changed with the recent publication of several reports that place specific WRKY TFs into signalling networks that respond to ABA (Jiang and Yu, 2009; Ren et al., 2010; Shang et al., 2010). Some of these WRKY TFs appear to regulate the expression of other TFs (for example, bZIPs, MYBs and ERFs) that were already known to be regulators of ABA responses. These new insights into the mechanisms of ABA signalling have important consequences for the manipulation of processes such as drought responses and seed germination in crop plants.
Abscisic acid signalling
Abscisic acid was identified in the 1960s and is a plant hormone that coordinates responses to stresses such as drought, extreme temperature and high salinity, as well as regulating nonstress responses including seed maturation, seed germination and bud dormancy (Shang et al., 2010; Umezawa et al., 2010). ABA functions through a complex web of signalling networks, and many parts of these pathways have been identified in recent years (Umezawa et al., 2010). These signalling components include TFs of various classes, E3 ligases, phospholipases C/D, G proteins, receptor-like kinases and other classes of protein kinases and phosphatases (Shang et al., 2010; Umezawa et al., 2010). However, plant scientists laboured for decades to understand how plant cells sense ABA (Pennisi, 2009), and the elusive ABA receptors remained to be found. Starting in 2006, several potential ABA receptors were proposed, although a role as a true ABA receptor remains, in many cases, controversial. The first reported ABA receptor was an RNA-binding protein called FCA (Razem et al., 2006), although the data associated with the report proved to be unreliable (Risk et al., 2008). In the same year, a second potential ABA receptor was proposed (Shen et al., 2006). The protein was the magnesium-protoporphyrin IX chelatase large subunit (Mg-chelatase H subunit [CHLH]/putative ABA receptor [ABAR]). ABAR is a chloroplast/plastid protein and has multiple functions in plant cells. Doubts have also been expressed about the role of ABAR as an ABA receptor as knockout mutations in the barley homologue, XanF, appear to have no effect on ABA signalling (Muller and Hansson, 2009). This lack of a detectable phenotype may, however, be a consequence of the presence of duplicated or multiple ABA receptors in plants. Nevertheless, recent additional work on ABAR provides not only more data in support of the protein as an ABA receptor but also a mechanism of action with a near-complete signalling pathway (Shang et al., 2010). ABAR spans the chloroplast envelope, and the cytosolic C-terminus interacts with a group of WRKY TFs (AtWRKY40, AtWRKY18 and AtWRKY60) that function as negative regulators of ABA signalling. Other reports of possible ABA receptors have centred on plasma membrane–located G protein-coupled receptors. Two have been implicated in ABA responses. First, an unconventional G protein-coupled receptor GCR2 (Liu et al., 2007) and secondly, GTG1 and GTG2, members of a novel class of G protein-coupled receptor (Pandey et al., 2009). The role of both proteins as ABA receptors, again, remains controversial (Johnston et al., 2007; Pennisi, 2009).
Recently, PYR/PYL/RCAR proteins, members of the START domain superfamily, were reported to function as cytosolic ABA receptors by directly inhibiting type 2C protein phosphatases (Ma et al., 2009; Park et al., 2009). Shortly after these first reports, six independent groups simultaneously defined the structural and functional mechanisms by which ABA is sensed (Fujii et al., 2009; Melcher et al., 2009; Miyazono et al., 2009; Nishimura et al., 2009; Santiago et al., 2009; Yin et al., 2009). In the absence of ABA, a phosphatase PP2C acts as a constitutive negative regulator of a family of kinases (SnRK2) whose autophosphorylation is required for kinase activity towards downstream targets. The binding of ABA by the PYR/PYL/RCAR receptor facilitates binding of the receptor to PP2C and this represses PP2C activity. This sequestration of PP2C in the ABAR-receptor complex allows autoactivation of the SnRK2 kinase which then phosphorylates downstream TFs leading to the transcriptional activation of ABA-responsive genes (Sheard and Zheng, 2009). This elegant mechanism of ABA perception and signal transduction likely represents a major component of ABA signalling, although other receptors such as ABAR are probably additional features of a complex signalling web with both multiple inputs and multiple outputs.
WRKY transcription factors
WRKY proteins comprise one of the largest families of TFs found in plants (Rushton et al., 2010). As with most TFs, the defining feature, or signature, of WRKY proteins is their DNA-binding domain. This is called the WRKY domain after the almost invariant WRKY amino acid sequence (Rushton et al., 1996, 2010; Eulgem et al., 2000). The WRKY domain is about 60 residues in length and has two components. At the N-terminal end is the WRKY amino acid signature, and this is followed by a zinc finger structure at the C-terminus. The amino acid sequence of the zinc finger in the WRKY domain is CX4-7CX22-23HXH/C, and the exact amino acid sequence of the finger reflects the subfamily of WRKY genes to which the protein belongs. The bipartite nature of the WRKY domain is underlined by the observation that in many subfamilies of WRKY genes (groups I, IIc, IId, IIe and III), the two component parts of the WRKY domain are separated by an intron (Eulgem et al., 2000). Some of the important outstanding questions about the WRKY domain were answered by an NMR solution structure of a WRKY domain (Yamasaki et al., 2005) followed 2 years later by a crystal structure determination (Duan et al., 2007). Both the solution structure and the crystal structure revealed that the WRKY domain consists of a four-stranded β-sheet, with a zinc-binding pocket formed by the conserved Cys/His residues. The WRKYGQK residues form the most N-terminal β-strand and appear to enter the major DNA groove and form contacts with an approximately 6-bp region of the DNA. This 6-bp region of interaction is consistent with the length of the W box, (Yamasaki et al., 2005), which is the core binding site for most WRKY proteins (Rushton et al., 1996, 2010; Eulgem et al., 2000).
Outside of the WRKY domain, WRKY proteins contain characteristic features of TFs such as nuclear localization signals, activation/repression domains and domains associated with protein–protein interactions such as leucine zippers (Eulgem et al., 2000; Rushton et al., 2010). Although some of these domains are conserved between members within a subfamily, it is only the WRKY domain itself that is shared by all WRKY TFs. The availability of the complete genome sequence of several flowering plant species has facilitated the classification of WRKY TFs into seven major subfamilies called groups I, IIa, IIb, IIc, IId, IIe and III (Eulgem et al., 2000). Phylogenetic analyses have then more accurately divided the WRKY family into groups I, IIa + b, IIc, IId + e and III with the group II genes not being monophyletic. Group IIa and IIb genes form two closely related clades, as do group IId and IIe genes (Zhang and Wang, 2005; Rushton et al., 2008, 2010).
Fifteen years of research have confirmed that the conservation of the WRKY domain in WRKY proteins is mirrored by a remarkable conservation of its cognate binding site, the W box (TTGACC/T) (Rushton et al., 1996, 2010; Eulgem et al., 2000). Both bioinformatic-based and functional studies of plant promoters have found clusters of W boxes in stress-inducible promoters and, in some cases, multiple W boxes appear to have a synergistic effect on transcription (Eulgem et al., 1999). Almost all WRKY TFs bind preferentially to W boxes, and this raises the question as to how they show specificity for the promoters of their target genes (Rushton et al., 2010). Data concerning the binding-site specificity of WRKY proteins are surprisingly rare, but Ciolkowski et al. (2008) showed that although the W box core is required, adjacent sequences also play a role in determining binding-site preference (Ciolkowski et al., 2008). The binding of WRKY proteins to W boxes is a feature of both biotic and abiotic stress responses. The presence of functional W boxes in the promoters of abiotic stress–inducible genes has recently been clearly demonstrated in Arabidopsis using both chromatin immunoprecipitation (ChIP) and gel retardation assays (Shang et al., 2010).
WRKY proteins can activate or repress transcription, and some WRKY TFs appear to possess both functions (Rushton et al., 2010). The mechanisms of activation and repression require further elucidation; however, an increasing number of proteins have been shown to interact with WRKY TFs (Rushton et al., 2010). These include both proteins that direct epigenetic changes, such as histone deacetylases, and signalling components, such as MAP kinases, MAP kinase kinases, 14-3-3 proteins and calmodulin (Rushton et al., 2010). The discovery of interacting partners facilitates the reconstruction of signalling pathways that contain WRKY proteins and identifies both inputs and outputs for these TFs.
The importance of WRKY TFs in plant stress signalling is illustrated by two recent reports concerning interacting partners of the group IIa proteins HvWRKY1, HvWRKY2, AtWRKY18, AtWRKY40 and AtWRKY60 (Shen et al., 2007; Shang et al., 2010). In barley, ETI to barley powdery mildew requires the recognition of the fungal avirulence AVR10 effector by the resistance protein MLA. This occurs in the cytoplasm and leads to a subsequent association of the MLA resistance protein with HvWRKY1 and HvWRKY2 inside the nucleus (Shen et al., 2007). This association of WRKY TFs with resistance proteins is exciting, as it reveals a direct signalling mechanism involving two well-characterized components of plant stress responses. Another interaction occurs between the Arabidopsis group IIa WRKY proteins, AtWRKY18, AtWRKY40 and AtWRKY60, and the chloroplast/plastid-localized ABA receptor, ABAR (Shang et al., 2010). These data reveal direct interactions between WRKY TFs and important components of plant-signalling webs such as resistance proteins and receptors and underlines the importance of WRKY TFs in these signalling networks.
WRKY transcription factors in ABA-signalling networks
Several recent publications have not only placed WRKY TFs in ABA-induced signalling networks but have also provided novel insights into both the ABA receptors that lie upstream of the WRKY TFs and the target genes that are downstream. New data concerning the ABA receptor, ABAR, provide more information in support of the protein as an ABA receptor and also a mechanism of action that includes at least three group IIa WRKY TFs (Shang et al., 2010). In an extensive set of studies using ABAR-GFP fusion proteins and immunodetection, it was shown that ABAR spans the chloroplast envelope and that both the N- and C-terminal portions of the protein are exposed to the cytoplasm. Previous work had shown that it is the C-terminal part of ABAR that binds ABA (Wu et al., 2009a), and this C-terminus (amino acids 692–1381) was used as a bait in a yeast two-hybrid screen. These experiments led to the identification of AtWRKY40 as a protein that interacts with the C-terminus of ABAR. The other two Arabidopsis group IIa proteins, AtWRKY18 and AtWRKY60, also interact with ABAR, albeit with a lower affinity. The interaction of ABAR and AtWRKY40 was confirmed by both coimmunoprecipitation and luciferase complementation imaging (Shang et al., 2010). This interaction is stimulated by ABA, and ABA also recruits AtWRKY40 from the nucleus to the cytoplasm. This suggests a mechanism of ABA signalling that operates by the removal of AtWRKY40 from the nucleus (Figure 1). Further evidence provided a clearer insight into this mechanism. Knockout mutants of AtWRKY18, AtWRKY40 and AtWRKY60 all show ABA-hypersensitive phenotypes in ABA-induced postgermination growth arrest, and inhibition of seed germination and the mutant analyses suggest that the three WRKY TFs cooperate to negatively regulate ABA signalling. The ABA-induced movement of AtWRKY40 from the nucleus to the cytoplasm therefore represents a de-repression of ABA signalling and reveals the first stages in a novel mechanism where ABA induces gene expression (Figure 1).
Regardless of the exact role of ABAR in ABA perception, it is nevertheless clear that the three Arabidopsis group IIa WRKY TFs play important roles in ABA signalling (Chen et al., 2010; Shang et al., 2010). The expression of a large number of known ABA-responsive genes is altered in AtWRKY40 or AtWRKY40/AtWRKY18 knockout lines. These genes include ABF4, ABI1, ABI2, ABI4, ABI5, DREB1A, DREB2A, MYB2, PYL2/RCAR13, PYL2/RCAR11, RAB18, PYL2/RCAR9, PYL2/RCAR7, SnRK2.2 and SnRK2.3 (Shang et al., 2010). ChIP experiments show that AtWRKY40 directly targets a number of these genes as the protein binds in vivo to W box-containing fragments of the promoters of the ABI4, ABI5, ABF4, MYB2, DREB1A and RAB18 genes (Table 1) (Shang et al., 2010). These data place AtWRKY40, AtWRKY18 and AtWRKY60 upstream of other known ABA-responsive TFs such as the AP2/ERF genes DREB1A and ABI4, the MYB gene MYB2 and the bZIP genes ABI5 and ABF4. ABI5 controls seed germination and postgermination growth and is one of the most important and genetically well-characterized ABA-signalling regulators (Finkelstein and Lynch, 2000; Finkelstein et al., 2002). Both cotransfection experiments using the ABI5 promoter and AtWRKY40 and the analysis of ABI5 expression in AtWRKY40 knockout lines suggest that AtWRKY40 directly represses ABI5 expression (Shang et al., 2010). This observation that these WRKY TFs regulate known ABA-responsive TFs is strong evidence that they are early nodes in the ABA-signalling web.
Table 1. Abscisic acid-signalling components that are direct targets of AtWRKY40
Type of gene
ChIP, chromatin immunoprecipitation.
AP2/ERF transcription factor
Binds to the promoter in vivo (ChIP). Yeast one-hybrid analyses Gel shifts
bZIP transcription factor
Binds to the promoter in vivo (ChIP) Yeast one-hybrid analyses Gel shifts
bZIP transcription factor
Binds to the promoter in vivo (ChIP) Yeast one-hybrid analyses Gel shifts
MYB transcription factor
Binds to the promoter in vivo (ChIP) Yeast one-hybrid analyses Gel shifts
AP2/ERF transcription factor
Binds to the promoter in vivo (ChIP)
AP2/ERF transcription factor
Binds to the promoter in vivo (ChIP)
Binds to the promoter in vivo (ChIP)
WRKY transcription factor
Gel shifts Cotransfection (together with AtWRKY18)
The observation that WRKY TFs regulate the expression of ABF genes necessitates a cautionary note here on the naming of genes. The name ABF has been used several times to describe TFs. Human ABF-1 is a bHLH transcription factor (Mitchell et al., 2000). Yeast ABF-1 is a trans-acting factor involved in the regulation of transcription and in DNA replication (Rhode et al., 1989). More importantly, the names ABF1 and ABF2 had already been used to describe plant TFs before a subfamily of bZIP genes were called ABFs (Choi et al., 2000). ABF1 and ABF2 from wild oat were among the first WRKY genes isolated, and ABF2 was the first group IIa gene described (Rushton et al., 1995). Both ABF1 and ABF2 have been implicated in GA and ABA signalling (Rushton et al., 1995). Wild oat ABF2 appears to be an orthologue of Arabidopsis AtWRKY40 and may therefore be a negative regulator of wild oat ABF genes (bZIPs) in a similar way to its Arabidopsis counterpart. The potential for confusion is clear. We propose the names AfWRKY1/ABF1 and AfWRKY2/ABF2 to avoid any confusion caused by the multiple use of the name ABF.
An additional recent report also demonstrated that AtWRKY40, AtWRKY18 and AtWRKY60 are involved in plant responses to ABA and abiotic stress (Chen et al., 2010) and also provides some additional information about the complexity of the regulation of ABA responses by these three WRKY proteins. Through analysis of single, double and triple mutants and overexpression lines, it appears that AtWRKY40 does indeed negatively regulate ABA responses during seed germination and postgermination growth. By contrast, it was suggested that AtWRKY18 and AtWRKY60 have a positive effect on ABA responses and also enhance plant sensitivity to salt and osmotic stress. Both AtWRKY40 and AtWRKY18 are rapidly induced by ABA, whereas induction of AtWRKY60 is slower (Chen et al., 2010). It appears that AtWRKY60 might be a direct target of AtWRKY40 and AtWRKY18 because induction of AtWRKY60 is almost completely abolished in wrky18 and wrky40 mutants, and both AtWRKY40 and AtWRKY18 proteins recognize a cluster of W box sequences in the AtWRKY60 promoter. The authors suggest that a AtWRKY18/AtWRKY40 heterocomplex may regulate the expression of the AtWRKY60 gene and that homo- and heterodimer complexes of these three group IIa WRKY proteins then regulate ABA responses (Chen et al., 2010). This cross-regulation at the transcriptional level and the involvement of homo- and heterodimer WRKY complexes add an extra level of complexity to the ABA-signalling network.
Recent mutant analysis has presented direct evidence that WRKY TFs are components of other parts of ABA-induced signalling networks (Jiang and Yu, 2009; Ren et al., 2010). It is relatively rare to identify mutants in WRKY genes that have detectable phenotypes, largely because of functional redundancy (Eulgem and Somssich, 2007; Rushton et al., 2010). Nevertheless, two recent reports provide details of WRKY knockout mutants that are hypersensitive to ABA responses during seed germination and postgermination growth. The ABA-hypersensitive mutant, abo3, was found to be caused by a T-DNA insertion in AtWRKY63 (At1g66600). The abo3 mutant was hypersensitive to ABA in both seedling establishment and seedling growth. Conversely, stomatal closure was less sensitive to ABA, and the mutant was therefore also less drought tolerant than the wild type (Ren et al., 2010). AtWRKY63 is a member of the group III subfamily of WRKY TFs and is not only a different type of WRKY protein from AtWRKY40, AtWRKY18 and AtWRKY60 but it is also to be found in a different part of the ABA-signalling web (Figure 2). The abo3 mutation impaired the expression of ABF2 and downstream genes such as RD29A and COR47, but the levels of ABF3, DREB2A, RD22 and KIN1 did not differ (Ren et al., 2010). Additionally, the transcriptional induction of AtWRKY63/ABO3 by ABA was impaired in abi1, abi2 and abi5 mutant lines. This places the AtWRKY63 gene downstream of ABI1, ABI2 and ABI5 but upstream of ABF2, RD29A and COR47 (Figure 2). Interestingly, AtWRKY40 appears to act upstream of the bZIP transcription factor ABI5 (Shang et al., 2010), whereas AtWRKY63 acts downstream of it in seed germination and postgermination growth (Figure 2) (Ren et al., 2010; Shang et al., 2010). Taken together, this suggests that ABA induces a cascade of transcription factor activation with AtWRKY40 repressing ABI5 gene expression in the absence of ABA. Upon ABA perception by ABAR, PYR/PYL/RCAR or other receptors, de-repression of ABI5 leads to activation of AtWRKY63 at the transcriptional level. AtWRKY63 then activates downstream genes such as RD29A and COR47. As each of these TFs appears to have multiple target genes, this pathway is therefore only a small module within a complex web of gene activation and repression that ABA perception sets in motion.
Another knockout mutation that affects seed germination and postgermination growth was reported in AtWRKY2 (Jiang and Yu, 2009). The knockout mutant has a similar phenotype to the abo3/AtWRKY63 mutation except that there appears to be no effect on stomatal closure. wrky2 knockout mutants displayed delayed or decreased expression of ABI5 and ABI3 and increased or prolonged expression of Em1 and Em6. Analysis of AtWRKY2 expression levels in ABA-insensitive and ABA-deficient mutants indicated that ABA-induced AtWRKY2 accumulation during germination and postgermination early growth requires ABI5, ABI3, ABA2 and ABA3 (Jiang and Yu, 2009). This suggests a feedback loop involving ABI3 and ABI5 in AtWRKY2 gene expression and also places AtWRKY2 upstream of both genes in the initial ABA response. Taken together, these analyses suggest a role for AtWRKY2, AtWRKY40, AtWRKY18 and AtWRKY60 upstream of ABI5, whereas AtWRKY63 acts downstream of it.
WRKY transcription factors, abscisic acid, drought responses and stomatal opening
Drought stress is one of the major causes of crop loss (Cominelli and Tonelli, 2010; Nakashima and Yamaguchi-Shinozaki, 2010), and improving plant responses to drought stress is a major aim of plant biotechnology. Drought stress induces the accumulation of ABA, which leads to stomatal closure. Closing of the stomata helps to maintain the water status of cells within the plant under water deficit conditions by reducing water loss as a result of transpiration (Schroeder et al., 2001). The abo3 mutation impairs ABA-induced stomatal closure, and the plants are therefore more sensitive to drought stress than wild-type plants, suggesting that ABO3/AtWRKY63 functions in ABA-mediated drought stress response pathways (Ren et al., 2010). This positive role of ABO3/AtWRKY63 in ABA-mediated stomatal closure contrasts with its negative role in seed germination and root growth. Interestingly, it has also been proposed that WRKY TFs are part of a mechanism of stomatal closure that is induced by pathogens (Schulze-Lefert and Robatzek, 2006). An elegant set of experiments showed that stomata close upon detection of potential microbial pathogens to prevent infection of the plant (Melotto et al., 2006). In the continual battle between plants and their pathogens, pathogenic bacteria have evolved strategies to suppress this stomatal closure mechanism and it was suggested that WRKY TFs play a role in this suppression by bacteria (Schulze-Lefert and Robatzek, 2006). The possibility that WRKY TFs are involved in stomatal closing as a response to both biotic and abiotic stress is an area that requires more research but more evidence is slowly appearing. Ectopic overexpression of the ABA-inducible OsWRKY45 gene in Arabidopsis conferred a number of properties to the transgenic plants, including enhanced disease resistance, enhanced tolerance to salt and drought stress, decreased sensitivity to ABA during seed germination and postgermination growth and enhanced induction of stress-related genes (Qiu and Yu, 2009). Importantly, under water stress conditions, the OsWRKY45-overexpressing plants had a lower rate of water loss than control plants and this appeared to be because these plants had a greater number of closed stomata. The reduced transpiration rate because of greater stomatal closure allowed the OsWRKY45-overexpressing plants to maintain a more favourable water balance and resulted in greater drought tolerance. There are other recent reports of WRKY genes regulating water stress responses. Rice lines overexpressing OsWRKY11 showed significant desiccation tolerance and induction of raffinose synthase and galactinol synthase genes (Wu et al., 2009b). Induction of these genes and accumulation of raffinose as an osmoprotectant are both well-characterized responses to water stress and this suggests a regulatory role for OsWRKY11 in these processes. More evidence that WRKY genes regulate galactinol synthase induction has come from the resurrection plant Boea hygrometrica (Wang et al., 2009). The BhGolS1 gene is inducible by both dehydration and ABA. The BhGolS1 promoter contains four W boxes, and ChIP showed that it is bound in vivo by the early dehydration and ABA-inducible BhWRKY1 (Wang et al., 2009). These data provide direct evidence linking a dehydration-inducible WRKY TF with a downstream target gene that plays an important role in drought responses.
Other data suggest that manipulation of WRKY TF expression may lead to improved drought responses through changes, not in stomata but in root architecture. Heterologous overexpression of the rice WRKY gene OsWRKY08 in Arabidopsis improved osmotic tolerance. Although not overexpressed in rice, these data are of interest as they suggest a physiological basis for the increased drought tolerance. 35S:OsWRKY08 transgenic Arabidopsis plants have increased lateral root number and primary root length during root development (Song et al., 2009).
Gibberellin, abscisic acid and WRKY transcription factors in seed germination
By far, the best studied role of WRKY TFs in regulating ABA responses is their role as ABA-inducible repressors of seed germination. This has a history that goes back to the discovery of AfWRKY1/ABF1 and AfWRKY2/ABF2 over 15 years ago (Rushton et al., 1995). ABA is integral to establishing and maintaining seed dormancy. In this capacity, ABA and GA participate in antagonistic crosstalk; while GA initiates germination, ABA prevents this initiation (Finkelstein et al., 2008). In cereals, upon imbibing, the embryos of nondormant seeds produce GA. This GA is transported to the aleurone, a thin layer of cells surrounding the endosperm. Upon arrival in aleurone cells, GA interacts with the receptor, GID1 (Murase et al., 2008). Through a series of signal transduction events (Ueguchi-Tanaka et al., 2007; Murase et al., 2008), a number of TFs, including GAMYB (Gubler et al., 2002), initiate the processes involved in germination. These processes include the induction of α-amylase production, allowing this enzyme to be secreted into the endosperm, where it mobilizes starch reserves to fuel embryo growth (Gubler et al., 1997) (Figure 3a). ABA inhibits this pathway through a number of mechanisms, including the repression of GAMYB expression (Gomez-Cadenas et al., 2001), and possibly through the inhibition of GA biosynthesis (Gubler et al., 2005).
Several WRKY TFs have been shown to be involved in this suppression of germination by ABA. In rice, OsWRKY45 and OsWRKY24 inhibit transcription from the promoter of the ABA-responsive gene, HVA22, under ABA treatment (Xie et al., 2005). In contrast, another two, OsWRKY72 and OsWRKY77, enhance transcription from the HVA22 promoter by ABA. Similar experiments in barley aleurone cells, using a reporter construct containing the GA-inducible Amy32bα-amylase promoter (Lanahan et al., 1992) showed that OsWRKY71, a homologue of wild oat AfWRKY2/ABF2 (Rushton et al., 1995), antagonizes the activation of the reporter construct by GA and GAMYB (Zhang et al., 2004). Further studies indicate that OsWRKY71 works with another WRKY family member, OsWRKY51, to mediate crosstalk between GA and ABA in the control of α-amylase production (Zhang et al., 2004; Xie et al., 2006). Promoter-binding studies suggest that OsWRKY71, but not OsWRKY51, binds to the O2S element (containing two W boxes) of the Amy32bα-amylase promoter. Interestingly, bimolecular fluorescence assays demonstrate that the two proteins interact in the nuclei of barley aleurone cells, (Xie et al., 2006) and interaction of the two TFs increases the binding of OsWRKY71 to the W box. The current model of ABA and GA action sees the OsWRKY51 and OsWRKY71 TFs form a heterotetramer that binds the Amy32b promoter and prevents the activation of the promoter by OsGAMYB. The OsWRKY71/51 heterotetramer, along with other proteins in a repression complex, appears to play a role in preventing early release of the dormancy induced by ABA. Another interesting aspect of this mechanism is that GA promotes the degradation of OsWRKY71 (but not OsWRKY51), resulting in disintegration of the repression complex, expression of genes such as those encoding α-amylases and hence seed germination (Zhang et al., 2004; Xie et al., 2006).
It appears that this repression of seed germination by a subset of WRKY TFs is a conserved feature of flowering plants, as barley HvWRKY38 appears to play a similar role to its rice orthologue OsWRKY71 (Zou et al., 2007). In barley, additional components of the transcriptional network have been uncovered including the DOF transcription factor HvBPBF that acts as a negative regulator of GA response in aleurone cells (Mena et al., 2002). HvWRKY38 is capable of forming a homodimer as well as interacting with HvBPBF to form a heterodimer. Another DOF protein, SAD, interacts with HvGAMYB to positively regulate GA responses in aleurone cells (Diaz et al., 2005). Cobombardment studies with the transcriptional activators HvGAMYB and SAD and the transcriptional repressors HvWRKY38 and HvBPBF suggest a possible mechanism of competition between these four TFs (Figure 3). Acting as a homodimer, HvWRKY38 blocks induction of Amy32b by HvGAMYB or SAD, although coexpression of these transcriptional activators overcomes the repression by HvWRKY38. Conversely, combination of HvWRKY38 and HvBPBF strengthens the ability of HvWRKY38 to block the induction of Amy32b expression even in the presence of both HvGAMYB and SAD. The interaction between repressors and activators of Amy32b suggest regulatory control by a repressosome, which includes proteins such as HvWRKY38 and HvBPBF, and an enhanceosome, which includes proteins such as HvGAMYB and SAD (Rushton et al., 2010). The Amy32b promoter also contains an AMY box, which can be bound by the repressor HvMCB1 or the activator HvMYBS3 (Rubio-Somoza et al., 2006), and these proteins may also play a part in the repressosome and enhanceosome, respectively (Figure 3). In addition, the HRT zinc finger protein can bind the GA response element and has been found to act as a repressor (Raventós et al., 1998), and the RAMY zinc finger protein can bind the W box element (Peng et al., 2004). Addition of these proteins to the repressosome and enhanceosome model produces a possible mechanism whereby Amy32b expression is controlled by the balance between four enhanceosome proteins, RAMY, SAD, HvGAMYB and HvMYBS3, and four repressosome proteins, HvWRKY38, HvBPBF, HRT and HvMCB1 (Figure 3). Use of such a mechanism might allow the fine control of genes involved in seed germination and postgermination growth.
WRKY TFs from dicots that play roles in regulating ABA-mediated seed dormancy and germination are increasingly being reported. As already mentioned, ABA induces the expression of AtWRKY2 in wild-type plants, and the ABA-signalling mediators ABI3 and ABI5 are required for its induction by ABA (Jiang and Yu, 2009). The seeds of AtWRKY2 knockout plants show an increased delay in germination in the presence of ABA, compared with wild-type plants (Jiang and Yu, 2009). This indicates that AtWRKY2 acts as a negative regulator of the ABA-mediated maintenance of dormancy. Together with the data presented previously concerning AtWRKY40, AtWRKY18, AtWRKY60 and AtWRKY63, it is clear that WRKY TFs constitute multiple nodes in signalling webs that regulate seed germination in Arabidopsis (Table 2).
Table 2. Arabidopsis WRKY genes and their positions in abscisic acid-signalling networks
Apparent signalling network position
Seed germination Seedling establishment
Downstream of ABI5, ABI3, ABA2 and ABA3
Seed germination Postgermination growth
Upstream of ABF4, ABI1, ABI2, ABI4, ABI5, DREB1A, DREB2A, MYB2, PYL2/RCAR13, PYL2/RCAR11, RAB18, PYL2/RCAR9, PYL2/RCAR7, SnRK2.2 and SnRK2.3 May function as a heterodimer with AtWRKY40 to activate AtWRKY60
Seed germination Postgermination growth
Directly targets ABI4, ABI5, ABF4, MYB2, DREB1A and RAB18 Also upstream of ABI1, ABI2, ABI4, DREB1A, DREB2A, PYL2/RCAR13, PYL2/RCAR11, RAB18, PYL2/RCAR9, PYL2/RCAR7, SnRK2.2 and SnRK2.3
Seed germination Postgermination growth
May target some of the same genes as AtWRKY18 and AtWRKY40 Possibly activated by a AtWRKY18/AtWRKY40 heterodimer
Downstream of ABI1, ABI2, ABI3 and ABI5 Upstream of ABF2, COR47 and RD29A
This widespread connection between WRKY TFs and seed germination is underlined by the observation that it is not just a single subgroup of WRKY proteins that appear to regulate this process, but rather that WRKY TFs from across the superfamily are involved. For example, AtWRKY2 is a group I WRKY protein, whereas AtWRKY63 is a member of group III, and AtWRKY40, AtWRKY18 and AtWRKY60 are found in group IIa. The widespread occurrence of WRKY TFs that regulate seed germination suggests that this is a common feature of seed germination in flowering plants.
WRKY transcription factors that mediate ABA responses also form parts of other signalling networks
One of the most important recent insights into the role of WRKY TFs is the realization that a single WRKY protein may be involved in regulating several seemingly disparate processes (Rushton et al., 2010). We have already encountered the Arabidopsis group IIa WRKY proteins, AtWRKY18, AtWRKY40 and AtWRKY60 with respect to their roles in mediating ABA responses, but these TFs also play other important roles in Arabidopsis. AtWRKY18 and AtWRKY40 play a major and partly redundant role in PAMP-triggered basal defence (Pandey et al., 2010), where they negatively affect pre-invasion host defence. Using ChIP, direct in vivo interactions of WRKY40 with promoter regions of the regulatory gene EDS1, the AP2-type transcription factor gene RRTF1 and JAZ8, a member of the JA-signalling repressor gene family, were recently demonstrated (Pandey et al., 2010). The data suggest that WRKY18/40 negatively modulate the expression of positive regulators of defence such as CYP71A13, EDS1 and PAD4, but positively modulate the expression of some key JA-signalling genes. Taken together, these data suggest that AtWRKY18, AtWRKY40 and AtWRKY60 represent nodes in Arabidopsis signalling networks that take inputs from numerous stimuli and that they are involved in mediating responses to numerous phytohormones including salicylic acid, jasmonic acid, ABA and GA. These roles in multiple signalling pathways may in turn partly explain the pleiotropic effects commonly seen when TF genes are overexpressed. In the past, these pleiotropic effects have often been attributed to the binding of TFs to promoters of genes that are not normally targets because of the increased concentration of these DNA-binding proteins. This explanation may not always hold true.
The involvement of WRKY TFs that regulate ABA responses in regulating other processes can also be seen in monocots, where the barley gene HvWRKY1/38 is involved in regulating cold and drought responses (Marèet al., 2004) while also being a repressor of basal defence that directly interacts with the MLA resistance protein (Shen et al., 2007). These data clearly show not only that these WRKY genes can be regulators of several different processes such as biotic stress responses, abiotic stress responses and seed germination but may also partly explain the mechanisms of crosstalk between ABA signalling and other signalling pathways. A WRKY gene that mediates ABA responses can be a node or hub in signalling that takes inputs from other phytohormones and stimuli.
New insights reveal that WRKY TFs are key nodes in ABA-responsive signalling networks
Until very recently, the role of WRKY TFs in ABA responses was obscure, and their possible roles were normally overlooked in reviews into ABA signalling (Hirayama and Shinozaki, 2007; Seki et al., 2007; Agarwal and Jha, 2010; Urano et al., 2010). The recent significant advances in the study of ABA signalling discussed in this review show that WRKY TFs are key nodes in ABA-responsive signalling networks. These ABA-regulated plant processes include seed germination and dormancy, postgermination growth and also the opening and closing of stomata. The real significance of the recent work is that it places specific WRKY TFs into signalling networks that respond to ABA. At the molecular level, WRKY TFs act downstream of at least two ABA receptors: the cytoplasmic PYR/PYL/RCAR-protein phosphatase 2C-ABA complex and the chloroplast envelope–located ABAR–ABA complex. WRKY genes also operate at multiple levels in the ABA-signalling networks. For example, in Arabidopsis, ABA induces a cascade of transcription factor activation. In the initial absence of ABA, AtWRKY40 represses expression of the bZIP transcription factor ABI5. Upon ABA perception by ABAR, PYR/PYL/RCAR or other receptors, de-repression of ABI5 leads to activation of AtWRKY63. AtWRKY63 then activates downstream genes such as RD29A and COR47. This part of the network of ABA-inducible gene activation is therefore characterized by a WRKY-bZIP-WRKY module, with WRKY TFs active at multiple levels (Figure 2b). These new WRKY-containing signalling modules can now be investigated as parts of the larger signalling networks.
A recent burst of activity in WRKY TF research has clearly demonstrated that WRKY TFs are components of ABA signalling at several different levels. Some, such as AtWRKY40, are early components of signalling pathways and repress ABA responses through a novel mechanism that involves the ABA receptor ABAR. Others, such as AtWRKY63, are further downstream and target known response genes such as RD29A and COR47. These new insights also show that some WRKY TFs represent major hubs in plant signalling as they take input signals from multiple stimuli. This has major implications for the use of WRKY genes in crop improvement. On the one hand, it may make the manipulation of a single plant process difficult, but on the other, it may be possible to improve several different stress responses (for example, both biotic and abiotic stresses) using a single gene. The future of WRKY research will certainly hold some interesting surprises.
We thank Senthil Subramanian for helpful comments on the manuscript. This project was supported by National Research Initiative grants 2008-35100-04519 and 2008-35100-05969 from the USDA National Institute of Food and Agriculture to PJR and QJS and grant 2007-35304-18297 to QJS. Research in the Rushton laboratory is also supported by The United Soybean Board, The Consortium for Plant Biotechnology Research, The South Dakota Soybean Research and Promotion Council and The North Central Soybean Research Program. Funding for X.Chen was provided by the US Army Environmental Quality research program at ERDC. The authors appreciate kind discussions with, and suggestions from, Dr. Edward J. Perkins, Dr. Kurt A. Gust and Dr. Victor F. Medina of the Environmental Laboratory, the U.S. Army Engineer Research and Development Center.