A Major Role of the MEKK1–MKK1/2–MPK4 Pathway in ROS Signalling

Abstract

Over the last few years, it has become evident that reactive oxygen species (ROS) signalling plays an important role in various physiological responses, including pathogen defense and stomatal opening/closure. On the other hand, ROS overproduction is detrimental for proper plant growth and development, indicating that the regulation of an appropriate redox balance is essential for plants. ROS homeostasis in plants involves the mitogen-activated protein kinase (MAPK) pathway consisting of the MAPK kinase kinase MEKK1 and the MAPK MPK4. Phenotypic and molecular analysis revealed that the MAPK kinases MKK1 and MKK2 are part of a cascade, regulating ROS and salicylic acid (SA) accumulation. Gene expression analysis shows that of 32 transcription factors reported to be highly responsive to multiple ROS-inducing conditions, 20 are regulated by the MEKK1, predominantly via the MEKK1–MKK1/2–MPK4 pathway. However, MEKK1 also functions on other as yet unknown pathways and part of the MEKK1-dependent MPK4 responses are regulated independently of MKK1 and MKK2. Overall, this analysis emphasizes the central role of this MAPK cascade in oxidative stress signalling, but also indicates the high level of complexity revealed by this signalling network

INTRODUCTION

Mitogen-activated protein kinase (MAPK) cascades are modules minimally consisting of a MAPK kinase kinase (MAPKKK/MEKK), a MAPK kinase (MAPKK/MKK), and MAPK. Upon activation of the MAPKKK, the signal is transduced via phosphorylation-mediated activation of the downstream MAPKK, which, in turn, phosphorylates and thereby activates its downstream MAPK. The Arabidopsis genome contains more than 60 MAPKKKs, 20 MAPKs, and 10 MAPKs, which can combine—depending on the environmental stimulus or developmental stage—into different MAPK modules (MAPK Group, 2002; Nakagami et al., 2005). MAPK cascades are key players in ROS signalling (Nakagami et al, 2005; Pitzschke and Hirt, 2006). Several studies have shown that ROS are not only the trigger, but also the consequence of activation of MAPK signalling pathways (Kovtun et al., 2000; Ren et al., 2002; Yoshioka et al., 2003; Rentel et al., 2004; Nakagami et al., 2004).

For about two million years, molecular oxygen arising from photosynthetic processes has become pivotal to almost all organisms. Reactive oxygen species (ROS), the partially reduced or activated derivatives of oxygen (H₂O₂, HO•, ¹O₂, O₂⁻) are the highly reactive by-products of aerobic metabolism. A sophisticated ROS network comprising antioxidative enzymes, antioxidants, and ROS-producing enzymes allows a plant to tightly control ROS levels. ROS, arising as toxic by-products of various chemical reactions, can lead to oxidative damage and destruction of cells. For this purpose, plants have developed efficient strategies for targeted production of ROS as well as sophisticated scavenging mechanisms for regulating ROS levels (Hoeberichts and Woltering, 2003; Mittler et al., 2004; Pitzschke and Hirt, 2006). The characterization of the MAPKKK MEKK1 as a regulator of redox homeostasis and its role as upstream regulator of the MAPKKs MKK1 and MKK2 and the downstream MAPK MPK4 opened new possibilities to understand the role of this MAPK cascade in ROS signalling. In this work, we present the molecular analysis of this pathway, with a particular focus on its function in ROS-regulated gene expression.

RESULTS

ROS-Dependent Gene Regulation

Previous studies have indicated that, depending on the type of ROS (hydrogen peroxide, superoxide, or singlet oxygen) or its sub-cellular production site (plastidic, cytosolic, peroxisomal, or apoplastic), a different physiological, biochemical, and molecular response is provoked. The specificity of ROS-driven transcript expression had been assessed in a recent study comparing transcriptome data generated from ROS-related microarray experiments (Gadjev et al., 2006). A comparison of datasets obtained by exogenous application of oxidative stress-causing agents (methyl viologen, Alternaria alternata toxin, 3-aminotriazole, and ozone) and from the fluorescent (flu) mutant and transgenic plants, in which the activity of individual antioxidant enzymes (catalase, cytosolic ascorbate peroxidase, and copper/zinc superoxide dismutase) was perturbed, led to the identification of marker transcripts specifically regulated by hydrogen peroxide, superoxide, or singlet oxygen. Also, several transcripts have been identified as general oxidative stress response markers based on their responsiveness to several of the ROS-generating stresses.

The MEKK1–MKK1/2–MPK4 Pathway Is a Central Regulator of ROS Metabolism

A wide range of environmental stimuli, including bacterial and fungal elicitors as well as diverse abiotic stresses, can initiate MAPK cascades. They can be perceived by as yet mostly unknown receptors that then transduce the signal to the MAPK cascade. However, secondary signals that are produced by the challenged plant can be involved also. Examples are the plant-derived peptide systemin, which is formed upon wounding (McGurl et al., 1992), and the plant hormone salicylic acid (SA), which is synthesised in a stress-dependent manner and essential for many biotic stress responses (Koornneef and Pieterse, 2008). As previous studies have shown, stress, SA, ROS, and MAPK cascades are strongly interconnected (Nakagami et al., 2005; Pitzschke and Hirt, 2006).

MEKK1 was one of the first MAPKKKs to be characterized in Arabidopsis; the gene is transcriptionally up-regulated in response to touch, cold, and salt stress (Mizoguchi et al., 1996; Covic et al., 1999). MEKK1 has been implicated in the biotic and abiotic activation of the MAPKs MPK3, MPK4, and MPK6 (Asai et al., 2002; Teige et al., 2004). Moreover, MEKK1 can be activated by H₂O₂ through post-translational modifications including proteasome-mediated MEKK1 protein stabilization. Mekk1 homozygous knockout plants show a severe dwarfism, accumulate high amounts of ROS, and develop local lesions reminiscent of programmed cell death (PCD) (Ichimura et al., 2006; Nakagami et al., 2006). Transcription of a number of genes encoding redox-regulatory enzymes is strongly mis-expressed in mekk1 knockout plants (Ichimura et al., 2006; Nakagami et al., 2006). Whereas homozygous mpk4 mutants display a similar lethal phenotype as mekk1 mutants, no such effects are observed in mpk3 or mpk6 mutant plants. Similarly, altered expression of redox-regulatory genes and the age-dependent lethal phenotype were only found in mpk4, but not in mpk3 mutants (Nakagami et al., 2006).

Because MEKK1 is required for H₂O₂-induced activation of MPK4, but not of MPK3 or MPK6, a major role in ROS homeostasis was ascribed to MEKK1 and its downstream target MPK4. Programmed cell death often is associated with the concurrent accumulation of ROS and salicylic acid (SA). Indeed, mekk1 and mpk4 mutants were found to contain highly elevated SA levels and to constitutively express SA-dependent stress genes (Ichimura et al., 2006; Nakagami et al., 2006). The dwarfed phenotype of mekk1 and mpk4 plants can be partially rescued by the expression of the bacterial SA-degrading NahG enzyme (Petersen et al., 2000; Suarez-Rodriguez et al., 2007). These data suggest that the mutant phenotypes of mekk1 and mpk4 plants are predominantly due to elevated SA levels. However, SA and ROS do not always seem to go together with cell death. The activation-tagged bud1 mutant plants, which overexpress the MAPKK MKK7, have elevated SA levels and exhibit constitutive pathogenesis-related (PR) gene expression but no spontaneous lesions (Zhang et al., 2007), indicating that ROS generation can be uncoupled from SA signalling.

Interestingly, expression of a kinase-impaired version of MEKK1(K361M) also largely rescued the dwarfism, PR gene expression, and the callose deposition phenotype of mekk1 mutant plants (Suarez-Rodriguez et al., 2007). mekk1 mutant plants are impaired in flagellin (flg22)-mediated activation of MPK4, but not of MPK3 or MPK6 and ectopic expression of MEKK1(K361M) leads to wild-type profiles of MAPK activation in the mekk1 mutants. Although it cannot be excluded that residual kinase activity may reside in the MEKK1(K361M) protein, it appears that the kinase activity of MEKK1 may be dispensable for at least some of its in-planta functions (Suarez-Rodriguez et al., 2007).

ROS-Related Transcriptome Analysis of mekk1, mkk1/2, and mpk4 Mutants

Previous studies have shown MEKK1 to be an activator of two highly homologous MAPKKs (MKK1 and MKK2), which function upstream of the MAPKs MPK4 and MPK6 (Teige et al., 2004). All of these components have been implicated in biotic, abiotic, and ROS signalling (Nakagami et al., 2005). Moreover, Arabidopsis mutants lacking mekk1, mkk1/2, or mpk4 show striking similarities in many aspects, including a dwarfed phenotype, formation of spontaneous lesions as well as accumulation of ROS and SA.

In order to gain a global view on the molecular processes affected in these mutants, we performed a full genome transcriptome analysis. In agreement with the results from Qiu et al. (2008), a strong overlap in the transcriptome patterns of mekk1, mkk1/2, and mpk4 plants confirmed that MEKK1, MKK1, MKK2, and MPK4 are components of the same MAPK module (Figures 1 and 2).

Figure 1.

Overlap of Transcriptome Patterns of mekk1, mpk4, and mkk1/2 Mutants.

Within a given dataset, the number of genes with reported induction (SA+) or repression (SA–) by salicylic acid (Schenk et al., 2000) as well as the distribution of general ROS-responsive transcription factors (TF) (Gadjev et al., 2006) are shown. See text for details.

Figure 2.

Regulation of Gene Expression by MEKK1 via MKK1/MKK2 and MPK4-Dependent and -Independent Pathways.

The numbers of genes (red: up–; green: down-regulated in the respective mutants) controlled via the individual pathways are indicated.

Regulation of Common Stress-Responsive Genes by MEKK1–MKK1/2–MPK4

Recently, a genome-wide bioinformatical analysis of microarray data has established a list of 197 genes with responsiveness to cold, osmotic stress, wounding as well as to biotic stress (Ma and Bohnert, 2007). These common stress-induced genes also include genes encoding MAPK components: MPK5, MKK9, and MAPKKK14. Also ACS6, encoding the rate-limiting enzyme of ethylene biosynthesis and a substrate for MPK6 (Liu and Zhang, 2004) as well as six ERF/AP2 transcription factors (AtERF) were among the common stress genes. Therefore, ethylene signalling-mediated engagement of a subset of the MAPK family as a component of the common stress response has been proposed (Ma and Bohnert, 2007).

Of the 197 common stress-responsive genes reported by Ma and Bohnert (2007), 54 are differentially expressed (all up-regulated!) in mekk1, mkk1/mkk2, and/or mpk4 knockout plants (Table 1). Transcription factor-encoding genes are clearly overrepresented. Almost all of those common stress genes with enhanced transcript levels in mpk4 and/or mkk1/2 are also up-regulated in mekk1: 19 genes are common to all three mutants, 10 are specific to mekk1 and mkk1/2, 34 are specific to mekk1 and mpk4 mutants. These findings strongly suggest that in the context of stress signalling, the signal mediated to MKK1/2 and/or MPK4 almost exclusively has been transduced via MEKK1.

Table 1.

List of Genes Differentially Expressed in mekk1, mkk1/2, and/or mpk4 Overlapping with the Dataset of 197 Genes Responsive to Multiple Stresses (Ma and Bohnert, 2007).

Many of these genes are responsive to ROS (right column). Genes encoding transcription factors are highlighted.

SA-Dependent Gene Expression in mekk1, mkk1/2, and mpk4 Mutants

The accumulation of SA and H₂O₂ in mekk1, mkk1/2, and mpk4 mutants is reflected by the strong overlap of genes differentially expressed in these mutants with known SA- or ROS-dependent genes. Of 441 and 254 SA-regulated genes (at least two-fold up or down, respectively) (Schenk et al., 2000), 142 have altered transcript levels in at least one of the mutants mekk1, mpk4, and/or mkk1/2. Interestingly, these genes represent all possible combinations with respect to enhanced or repressed expression in the mutants versus SA response (up/up, up/down/, down/up, and down/down) (Figure 1 and Table 2). These findings are indicative of an expression dynamics in a number of SA-regulated genes: short-term exposure (by exogenous treatment for 24 h; Schenk et al., 2000) results in enhanced expression, whereas long-term exposure (as in SA-accumulating mutants) leads to decreased expression, or vice versa. Thus, in the SA-accumulating mutants, negative feedback mechanisms counteracting the effects of constantly high SA levels might have been initiated over the prolonged period. Good candidates involved in this feedback regulation might be ROS: while short-term SA leads to an oxidative burst and induction of genes encoding products that further accelerate the SA-triggered processes, proteins arising from constitutively high SA-mediated gene induction might have accumulated to amounts exceeding a certain threshold, which then act as negative regulators of other SA-responsive genes. This feedback regulation might help the plant to prevent excessive induction of stress genes whose products strongly interfere with proper development, namely leading to even more severe phenotypes than those observed for mekk1, mkk1/2, and mpk4 mutants.

Table 2.

List of SA-Responsive Genes that Are Differentially Expressed in mekk1, mkk1/2, and/or mpk4 Mutants.

ROS-Responsive Gene Expression in mekk1, mkk1/2, and mpk4 Mutants

A comparison of the mekk1, mkk1/2, and mpk4 transcriptome with genes related to oxidative stress revealed a strong overlap. In the study of Gadjev et al. (2006), several transcription factor genes responsive to multiple oxidative stresses had been identified. These transcription factors are candidates that could be responsible for orchestrating the specific transcriptomic signatures triggered by different ROS. Using the ROS gene dataset (Gadjev et al., 2006), we investigated the abundance of genes responsive to individual or multiple ROS-generating stresses in the datasets of genes differentially expressed in the mekk1, mkk1/2, and mpk4 mutants (Tables 3A–3C).

Table 3A.

Common ROS-Responsive Genes that Are Differentially Expressed in mekk1, mkk1/2, and/or mpk4 Mutants.

Transcription factor-encoding genes are highlighted.

Table 3C.

Genes Responsive to Multiple or to a Given Type of ROS and their Altered Expression in mekk1, mkk1/2, and mpk4 Mutants.

	Response to multiple types of ROS		response to specific ROS type
	total	transcription factors	singlet oxygen	superoxide	H2O2
total number of genes identified by Gadjev et al., (2006)	103	32	314	194	193
misregulated in mekk1, mkk1/2 and/or mpk4 mutants	46	20	48	31	43
proportion	45%	63%	15%	15%	22%

Total number of responsive genes as identified by Gadjev et al. (2006) compared to their abundance in the differential transcriptomes of mekk1, mkk1/2, and/or mpk4 mutants. For details, see Tables 3A and 3B.

Table 3B.

Transcripts Specifically Responsive to Singlet Oxygen, Superoxide or H₂O₂ and their Altered Abundance in mekk1, mkk1/2, and/or mpk4 Mutants.

Of 701 genes specifically responsive to hydrogen peroxide (193), superoxide (194), or singlet oxygen (314), 120 have altered expression in at least one of the mutants mekk1, mkk1/2, and mpk4; and many are shared by all mutants. There is no apparent preference for hydrogen peroxide-, superoxide- or singlet oxygen-responsive genes in any of the mutants’ differential transcriptomes, indicating major disturbances in the balances of all ROS species in these mutants. This is in line with the finding that a large number of ROS scavenging enyzme-encoding genes have altered transcript levels in the mutants (see below). Of 103 genes commonly responsive to various types of ROS, 51 are differentially expressed in mekk1, mkk1/2, and/or mpk4. Among the 32 transcription factors showing a very strong expression response to oxidative stress in general (Gadjev et al., 2006), as many as 20 are differentially expressed (all up-regulated) in mekk1, mkk1/2, and/or mpk4 mutant plants (Tables 3A and 3C, and Figure 1). Whereas none of these 20 transcription factor genes is exclusively affected in mpk4 or mkk1/2, two are specific for mekk1, four are shared by mkk1/2 and mekk1, four are shared by mpk4 and mekk1, and 10 are common to all three mutants. These findings strongly support a role of MEKK1 as a major regulator of redox homeostasis, which is mainly controlled through the MEKK1–MKK1/2–MPK4 cascade.

ROS Scavenging Mechanisms Are Disturbed in mekk1, mkk1/2, and mpk4 Mutants

Redox homeostasis is achieved through the action of ROS scavenging enzymes. In Arabidopsis, ROS scavenging enzymes are encoded by 148 genes representing 15 gene families (Mittler et al., 2004). Not unexpectedly, a significant proportion of these genes, 23, representing nine of the 15 gene families, are differentially expressed in mekk1, mkk1/2, and/or mpk4 plants, indicating major imbalances in redox homeostasis (Table 4). The corresponding gene products are of diverse sub-cellular localization, indicating that redox imbalances due to loss of MEKK1, MKK1/MKK2, and/or MPK4 activity are not restricted to a particular cell compartment. For instance, all three mutant lines have reduced transcript levels of the gene encoding thylakoidal peroxidase (tAPX), which uses ascorbic acid for peroxide removal (2Asc + H₂O₂ → 2MDA + 2H₂O) (Table 4). The repression of tAPX in mekk1, mkk1/2, and mpk4 is likely to contribute to the enhanced ROS levels in these mutants. The importance of tAPX in the removal of H₂O₂ has been emphasized in previous studies showing that overexpression of tAPX renders plants more tolerant to treatment with the superoxide-generating herbicide paraquat (PQ) (Murgia et al., 2004), whereas antisense reduction of tAPX results in PQ hypersensitivity (Tarantino et al., 2005). In addition to tAPX, the expression of all three members of the CAT family, encoding the peroxisomal hydrogen peroxide-scavenging enzyme catalase (CAT) (2H₂O₂→H₂O + O₂) is affected in our analyzed mutants. Interestingly, there are differences between the mutants with respect to differential expression of individual catalase genes. The expression of CAT2 encoding the major H₂O₂ scavenging enzyme (Vandenabeele et al., 2004) is reduced in mekk1 and mpk4, but unaffected in mkk1/2. This repression is likely to contribute to the highly elevated levels of H₂O₂ in mekk1 and mpk4 plants as has been observed by staining of mutants with the H₂O₂-specific dye DAB (Nakagami et al., 2006). Surprisingly, the steady-state transcript levels of CAT1 and CAT3 are elevated in mekk1, and also in mkk1/2, but not in mpk4. The simultaneous repression of CAT2 and induction of CAT1 and CAT3 in mekk1 might point to a feedback regulatory mechanism—an ‘attempt’ of these mutants to compensate for insufficient CAT2 levels by enhancing CAT1 and CAT3 synthesis. That such feedback mechanisms might exist is supported by a study investigating the kinetics and localization of catalase gene expression in Arabidopsis during plant development (Zimmermann et al., 2006). The expression and activity of CAT2, which are specific for photosynthetically active tissue, decrease during senescence before a detectable loss of chlorophyll. CAT2 down-regulation is assumed to be the initial step in the H₂O₂ peak during bolting time, whereas the decrease in APX1 (ascorbate peroxidise) activity presumably is only a secondary and amplifying effect (Zimmermann et al., 2006). CAT3 displays vasculature-specific expression, is induced with age and corresponds to an accumulation of H₂O₂ in the vascular bundles. The H₂O₂ peak resulting from senescence-induced CAT2 down-regulation results in CAT3 enzyme activation, followed by a recovery of APX1 activity and subsequent decline of H₂O₂ (Zimmermann et al., 2006). The lack of CAT3 induction specifically in mpk4, but not in mekk1 and mkk1/2 mutants despite high H₂O₂ levels, suggests a role of MPK4 in the sensing of vascular H₂O₂ independently of MEKK1 and MKK1/MKK2. Likewise, the down-regulation of CAT2 in mekk1 and mpk4, but not mkk1/2, mutants indicates (1) that disturbances of additional redoxregulatory mechanisms other than repression of CAT2 account for the H₂O₂ accumulation in mkk1/2 mutant plants and (2) that CAT2 expression is controlled by a MEKK1–MPK4 pathway bypassing MKK1/MKK2.

Table 4.

Altered Expression of ROS Scavenging Enzyme-Encoding Genes in mekk1, mkk1/2, and/or mpk4 Mutants.

Regulation of Photosynthesis by MEKK1–MPK4

A regulated appropriate balance of photosynthetic input of H₂O and CO₂ and removal of O₂ is necessary for optimal photosynthesis. Under various suboptimal conditions, excess excitation energy initiates dangerous ROS production. A connection of MAPK-regulated ROS control of photosynthesis is provided through a recent report showing that activation of three tobacco MAPKs, SIPK/Ntf4/WIPK by DEX-induced expression of a constitutively active MAPKK (MEK2DD) rapidly inhibits photosynthesis concurrently with strong accumulation of H₂O₂ (Liu et al., 2007). More detailed studies revealed that MEK2DD-induced shutdown of carbon fixation triggers chloroplastic superoxide production, which is rapidly converted to H₂O₂ by superoxide dismutase. Prolonged activation of the MAPK cascade (10 h) leads to chloroplast damage in a light-dependent manner, resulting in an HR-like cell death. Also ROS generation upon challenge of plants with tobacco mosaic virus (TMV), which activates SIPK and WIPK (Zhang and Klessig, 1998), is greatly enhanced upon light. This observation is similar to the finding that incompatible pathogen-induced HR cell death is light-dependent (Bechtold et al., 2005). Examination of the kinetics of carbon fixation inhibition and onset of cell death upon MEK2DD expression indicates that the MAPK cascade actively blocks carbon fixation, leading to an excess of excitation energy in illuminated plants and—consequently—accumulation of ROS. The sustained generation of ROS cannot be compensated for by the action of antioxidant enzymes, and, after depletion of the antioxidant pool, leads to cell death (Liu et al., 2007).

Interestingly, among the genes down-regulated specifically in mekk1 and mpk4, but not in mkk1/2, mutants, those encoding plastidic or chloroplastic proteins are significantly overrepresented (Table 5), suggesting that photosynthetic suppression of genes through MEKK1 and MPK4 bypasses MKK1/MKK2. These results would suggest that MEKK1-dependent activation of MPK4 might also occur independently or through other MAPKKs. In this context, it should be noted that in yeast two-hybrid assays, MEKK1 was found to directly interact with MPK4 (Ichimura et al., 1998). Biochemical analysis later confirmed the specificity of the MEKK1–MPK4 interaction (Nakagami et al., 2006). It is presently unclear whether MEKK1 can activate MPK4 directly or whether MAPKKs other than MKK1 and MKK2 can function in MPK4 activation.

Table 5.

Abundance of Selected Annotations of Genes Differentially Regulated in mekk1, mkk1/2, and/or mpk4 Mutants Compared to the Overall Abundance in the Arabidopsis Genome.

P-values of Chi-square test are shown. Statistically significant values are shaded. Gene annotations were retrieved from the TAIR webtool.

MEKK1-Specific Genes

Although a considerable number of differentially expressed genes is shared between mekk1, mkk1/2, and mpk4 mutants (284 of a total of 1635, 1142, and 776, respectively), the transcriptome analysis of these mutants also revealed that expression of a large set of genes (694 of 1635) is only affected in mekk1 mutants (Figure 1). Within this pool of mekk1-specific genes, several SA- and ROS-responsive genes are found. Their proportion is not significantly different from that in the datasets of genes showing altered transcript levels also in mkk1/2 and/or mpk4.

These data suggest that:

• (1) MEKK1 is the upstream activator of MKK1/MKK2 and MPK4;

• (2) MEKK1 activates an independent pathway, not involving MKK1/MKK2 or MPK4;

• (3) MKK1/2 is not only regulating MPK4, but most likely also other MAPKs (MPK6?);

• (4) MEKK1 activates MPK4 to some extent in an MKK1/MKK2-independent manner (Figure 2);

• (5) some SA signalling might be regulated by MEKK1 via a MKK1/MKK2- and MPK4-independent mechanism.

We cannot exclude that some of the alterations in ROS gene expression are due to defective developmental programs in the mutants. Also, other non-overlapping roles of MEKK1, MKK1/2, and MPK4 might exist that would explain why some ROS-responsive genes are unaffected in one mutant line but differentially expressed in the other mutant lines.

The spontaneous HR-like cell death in the mutants can be the trigger of differential gene expression of some ROS genes, while it might be the consequence of differential gene expression of other ROS genes. It has been reported that mekk1 and mpk4 mutants, when grown at higher temperatures, develop normally; and this coincides with a decrease of the SA accumulation in these mutants.

A comparison of the transcriptome of the mutants grown at elevated temperature, and a subsequent shift to normal (i.e. HR-supporting) temperature would help to distinguish between direct and secondary effects of the MAPK components on ROS gene expression.

Feed-Forward and Feed-Back Loops in MAPK Signalling

With an increasing number of microarray data becoming publicly available, a more and more complex pattern of MAPK regulation arises. Not only are MAPK cascade components post-translationally activated in the process of signal transduction from receptor to MAPK-targeted effector, but genes encoding several MAPK pathway components are subject to transcriptional regulation themselves. For instance, in a search for genes whose expression is rapidly induced upon wounding (5 min), MPKK9 and MPK3 as well as AP2C1, a PP2C-type phosphatase with a MAPK interaction motif, have been identified (Walley et al., 2007), indicating that transcriptional control of both phosphorylation and dephosphorylation of MAPK signalling components is involved in transduction of initial stress signalling events.

These data suggest that our ‘classical’ signalling concepts are oversimplified and multiple feed-forward and feed-back mechanisms are involved so that gene expression might be part of the signalling process itself. A good example might be provided by MPK3. The fast kinetics of MPK3 activation upon a series of challenging conditions (Walley et al., 2007; Pitzschke and Hirt, 2006; Nakagami et al., 2005) suggests that, initially, already existing MPK3 protein is being used for signal transduction. The stress-induced accumulation of MPK3 transcripts over an extended period, as has been observed in several gene expression studies (Mizoguchi et al., 1996; Wan et al., 2004; Walley et al., 2007), might be indicative of the need for a continuous supply of MPK3 enzyme, which would feed into the cascade and thereby contribute to amplifying the stress signal. Interestingly, in contrast to MPK3, its closest homolog, MPK6, underlies neither transcriptional nor translational control (Ulm et al., 2002). Why enhanced levels of some MAPK components persist long after MAPK activation has declined is a puzzling question.

An even more complex scenario of MAPK-regulated gene expression became apparent from the isolation of the Arabidopsis MAPKKK MEKK1 from a screen for proteins binding to the promoter of the WRKY53 gene. WRKY53 is a member of the plant-specific transcription factor family of WRKYs, which, in A. thaliana, comprises 74 members, many of which are transcriptionally inducible upon pathogen infection and other defence-related stimuli (Dong et al., 2003; Kalde et al., 2003). MEKK1 does not only interact with the WRKY53 promoter, but also binds to and phosphorylates the WRKY53 gene product (Miao et al., 2007). Previous studies had shown WRKY53 to induce the expression of stress and defence-related as well as senescence-associated genes (Miao et al., 2004). Characterization of the WRKY53 promoter region bound by MEKK1 revealed that it is important for the switch of WRKY53 expression from a leaf age-dependent to a systemic plant age-dependent expression during bolting. As in-vitro analysis and studies with a WRKY53 promoter-driven reporter gene have shown, MEKK1 binding to and phosphorylation of WRKY53 enhances the DNA-binding capacity of WRKY53 (Miao et al., 2007). WRKY53 expression is induced by H₂O₂ (Miao et al., 2004), and results from transient expression studies with WRKY53 promoter deletion constructs suggest MEKK1 to be involved in the hydrogen peroxide response of WRKY53 (Miao et al., 2007).

The apparent short-cut in mitogen-activated protein kinase (MAPK) signalling by direct phosphorylation of a transcription factor through a MAPKKK is so far unique to MEKK1. WRKY53 expression is also induced upon treatment with the fungal cell wall-derived elicitor chitin (review Montesano et al., 2003) known to activate MPK3 and MPK6 (Nühse et al., 2000; Gust et al., 2007), as well as by ectopic expression of the tobacco MAPKK NtMEK2 active mutant NtMEK2^DD (Wan et al., 2004). We also found WRKY53 within the dataset of genes commonly up-regulated in mekk1, mkk1/2, and mpk4. Together, these data suggest that WRKY53 expression is not only controlled by the MEKK1 short-cut but most likely also through classical MAPK signalling pathways.

DISCUSSION

Over the last couple of years, considerable efforts have been undertaken to disentangle the networks regulating basal ROS levels and targeted ROS synthesis in plants. Several MAPK cascade components were found to be involved, often in more than one aspect/process of ROS signalling. Clearly, MAPK pathways play a key role in controlling normal development and dynamic processes, such as flower development, stomatal patterning, and stomatal aperture, and are thus indispensable for ROS homeostasis. MAPK(KK) activities are regulated by a sophisticated network involving transcriptional, translational and posttranslational control as well as a diverse pattern of feedback loops. In a number of very recent studies, additional fascinating features of stress-response or developmental regulation of and through MAPK(KK)s have been identified. Our present analysis contributes to the understanding of redoxregulation through MAPKs. The strong overlap of the transcriptomes of mekk1, mkk1/2, and mpk4 mutants and the reported stress-responsiveness of the corresponding genes plausibly shows the existence of a full MAPK cascade (MEKK1–MKK1/MKK2–MPK4) as a key regulator of ROS- and SA-initiated stress signalling. Surely, many more astonishing scenarios of MAPK-controlled ROS production and ROS-controlled MAPK activity will be disclosed in the future.

METHODS

Plant Material

Surface-sterilized seeds were sown on agar plates containing half-strength MS medium, incubated at 4°C for 2 d and subsequently transferred to a controlled-environment room (22°C, 8 h photoperiod).

Ecotype Columbia mekk1, mpk4 (Nakagami et al., 2006), mkk1 (Mészáros et al., 2006) and mkk2 mutant lines (Teige et al., 2004) have been described previously.

Homozygous mkk1 and mkk2 mutant plants were crossed. Two plants carrying both T-DNA insertions were identified by genotyping. Genotyping of the progeny (30 individuals) of these plants showed that dwarfed individuals were homozygous for both insertions.

Fourteen-day-old seedlings for microarray analysis were selected based on their dwarfed phenotype. An aliquot of the RNA extracted for microarray analysis was used to confirm via RT–PCR that the harvested material was indeed homozygous for the respective insertions.

Transcriptome Studies

The microarray analysis was performed using the CATMA array containing 24 576 gene-specific tags (GSTs) corresponding to 22 089 genes, including 21 612 AGI-predicted and 477 Eugene-predicted genes (Crowe et al., 2003; Hilson et al., 2004). For each comparison (mutant line versus Col-O wild-type), we performed a biological repeat using an independent second set of samples. In addition, to avoid dye bias and gene-specific dye bias, a dye-swap experiment was carried out. Therefore, four arrays were used for each comparison assay. For each sample, total RNA was extracted from pools of 50 plantlets using the RNeasy Plant Mini Kit (Qiagen) following the manufacturer's protocol. RNA integrity, cDNA synthesis, hybridization, and array scanning were performed as described by Lurin et al. (2004).

Statistical Treatment of Microarray Data

Normalization and statistical analysis were based on two dye swaps, i.e. four arrays, as described in Gagnot et al. (2008). To determine differentially expressed genes, we performed a paired t-test on the log ratios, assuming that the variance of the log ratios was the same for all genes. Spots displaying extreme variance (too small or too large) were excluded. The raw P-values were adjusted by the Bonferroni method, which controls the Family Wise Error Rate. We use the Bonferroni method (with a type I error equal to 5%) in order to keep a strong control of the false positives in a multiple-comparison context (Ge et al., 2003). We considered as being differentially expressed the genes with a Bonferroni P-value ≤ 0.05, as described in Gagnot et al. (2008).

Microarray data from this article were deposited at Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/; accession no. GSE10763/GSE10764) and at CATdb (http://urgv.evry.inra.fr/CATdb/; Project: ADT06-03_MKK2-cold) according to the Minimum Information About a Microarray Experiment standards.

FUNDING

No conflict of interest declared.