Many changes in environmental conditions and hormones are mediated by MAPK (mitogen-activated protein kinase) cascades in all eukaryotes, including plants. Studies of MAPK pathways in genetic model organisms are especially informative in revealing the molecular mechanisms by means of which MAPK cascades are controlled and modulate cellular processes. The present review highlights recent insights into MAPK-based signalling in Arabidopsis thaliana (thale cress), revealing the complexity and future challenges to understanding signal-transduction networks on a global scale.
- Arabidopsis thaliana (thale cress)
- biotic and abiotic stresses
- mitogen-activated protein kinase (MAPK)
- phosphorylation cascade
As plants are sessile organisms unable to escape changing environmental conditions, they developed mechanisms to sense and adapt their physiology, growth and development. These adaptations involve a rapid and dynamic regulation of enzymatic activities and the modification of gene-expression programmes.
Phosphorylation/dephosphorylation of proteins is among the most common post-translational modification in all organisms. Protein phosphorylation is carried out by protein kinases that are typically organized in signalling cascades. It is estimated that about 30% of all proteins are phosphorylated in a eukaryotic cell. Not surprisingly, about 5% of the genomes of green plants (Viridiplantae) code for protein kinases [1,2]. Roughly 10% of all plant kinases are involved in MAPK (mitogen-activated protein kinase) pathways. In the present review we discuss the function of plant MAPKs, with a particular focus on Arabidopsis thaliana (thale cress) studies, and Table 1 gives an overview of our current knowledge. Genetic studies coupled with biochemical approaches have allowed breakthroughs in this area and have shown that MAPKs are not only involved in biotic and abiotic signalling, but also in processes such as hormonal and developmental signalling. Astonishingly, the same kinases are often involved in distinct physiological processes. Moreover, MAPK modulators such as phosphatases were found to be involved in the control of signal strength and the duration of signalling processes.
MAPK cascades are conserved signalling modules found in all eukaryotic cells, including plants, fungi and animals. A MAPK cascade minimally consists of three kinases: a MAP3K (MAP2K kinase), a MAP2K (MAPK kinase) and a MAPK, which phosphorylate, and therefore activate, each other in a specific way. MAP3Ks are serine/threonine kinases phosphorylating two amino acids in the S/T-X3–5-S/T motif of the MAP2K activation loop. MAP2Ks are dual-specificity kinases that activate a MAPK through double phosphorylation of the T-X-Y motif in the activation loop. MAPKs are serine/threonine kinases able to phosphorylate a wide range of substrates, including other kinases and/or transcription factors. A fourth level of kinases, named MAP4Ks (MAP3K kinases), may act as adaptators linking upstream signalling steps to the core MAPK cascades. Interactions between kinases within a MAPK cascade occur through docking sites present in the kinases and/or with the help of external scaffolding proteins. MKPs (MAPK phosphatases) are involved in the time-dependent control or in the shut-down of the pathway after signalling.
Complete sequencing of the Arabidopsis genome has revealed a large number of genes coding for MAPK-related kinases: 20, 10 and 80 kinases share significant similarity with the MAPK, MAP2K and MAP3K families found in animals and fungi respectively . A similar repertoire of genes was observed in the sequenced genomes of other plants, such as rice (Oryza sativa), poplar (Populus sp.) or grapevine (Vitis vinifera). In general, MAPKs can be found in a large number of plant species, indicating that MAPK cascades are also very conserved signalling mechanisms among higher plants. MAPKs and MAP2Ks constitute homogeneous kinase families, mechanistically behaving like their homologues in animals and yeast and divided into four subfamilies (A–D) based on sequence similarities . The 80 MAP3Ks are a much more heterogeneous group of protein kinases  that can be further divided into three main subgroups: the MEKK (MAPK/ERK kinase kinase)-like MAP3Ks, for which we have functional evidence that they act as MAP3Ks in planta, and the Raf-like and ZIK-like subgroups, for which the only functional evidence comes from non-plant systems. In addition to their kinase activity, plant MAP3Ks often contain long N- or C-terminal regions that might function in regulation or scaffolding to recruit MAP2Ks and MAPKs, or in the integration of the input signals. Apart from identifying a MAP3K as a scaffold, no specific scaffolding protein was yet reported in plants. The Arabidopsis genome contains a large number of various phosphatases putatively involved in kinase dephosphorylation (for a review, see ). Among these, five genes of the dual-specificity MKP family exist which should be modulators of the MAPKs.
MAPK CASCADES ARE INVOLVED IN PATHOGEN SIGNALLING
Plant–pathogen interactions involve very complex mechanisms ensuring survival in the competition arena. During evolution, higher plants developed an innate immune system to detect and rapidly respond to pathogen aggression . After pathogen infection and recognition of PAMPs (pathogen-associated molecular patterns) in the plant cell environment, multiple levels of defence are activated. PAMPs are small molecules usually derived from abundant pathogen structures such as structural proteins or cell-wall components that are shared by a large range of pathogen varieties. Within minutes of pathogen-derived PAMP recognition, the plant modifies enzymatic activities and gene-expression patterns to synthesize a large set of anti-microbial reagents, including AOS (active oxygen species) and phytoalexins. In order to suppress the innate immune response of plants, pathogens inject a number of virulence factors into the host cell. In specific cases, however, plants learnt to recognize virulence factors, resulting in the so-called ‘gene-for-gene resistance reactions’. Recognition of virulence factors occurs by plant NBS-LRR (nucleotide binding site–leucine-rich repeat) receptors to switch on defence signalling cascades. Whereas PAMP recognition is not specific to a given pathogen variety, gene-for-gene interactions are based on the recognition of a given pathovar. Among others factors, resistance against different pathogen types generally involves a regulation of the balance between SA (salicylic acid)- and ET/JA (ethylene/jasmonic acid)-dependent defence mechanisms and thereby long-term responses to pathogens . SA is linked to resistance to biotrophic pathogens and is important to trigger the HR (hypersensitive response), a PCD (programmed cell death) to locally counteract pathogen attack and progression. ET and JA play a role in the control of PCD spreading  and regulate resistance against necrotrophic pathogens. All three hormones regulate distinct sets of pathogen-related genes and are involved in inducing SAR (systemic acquired resistance), a long-range process of priming pathogen resistance in unaffected tissues.
In recent years, the number of publications revealing a function of MAPK cascades in pathogen signalling increased greatly. A set of three MAPKs – MPK3, MPK4 and MPK6 – was found to be activated on interaction with pathogens, explaining why they are by far the most well known plant MAPKs. Other MAPKs might also be involved in plant–pathogen signalling, but they have been only rarely observed so far, possibly solely because of a lack of appropriate tools such as specific antibodies to detect low protein amounts or kinase activities. Efforts to identify single and multiple knock-out lines for these kinases should also help to by-pass these limitations.
Given the heterogeneity in the protocols, the variety of biological systems and the favourite assays and topics of research of each laboratory, an integrative view of theoretical pathogen signalling pathways is still difficult. Nevertheless, MAPKs are unambiguously implicated in the signalling processes linked to innate immunity, whereas no function has so far been revealed for gene-for-gene defence mechanisms.
Flagellin, a model PAMP, helps in dissecting the MAPK cascades involved in innate immune responses
Flg22 is a 22-amino-acid-long peptide derived from flagellin, the main structural protein of the eubacterial flagellum . Flg22 acts as a PAMP on various plant species, inducing a wide range of pathogen-related responses. In Arabidopsis, the signal perception occurs at the plasma membrane through a receptor-like kinase complex composed of the proteins FLS2 (FLAGELLIN SENSING 2) and BAK1 (BRI1-ASSOCIATED KINASE 1) [10–12]. FLS2 alone carries the specificity toward flg22. Once bound to flg22, the FLS2–BAK1 complex initiates a signalling cascade, triggering a set of pathogen-related responses, including an early oxidative burst , the activation of various kinases, the consecutive modification of the phosphorylation state of many cellular proteins  and broad gene modulations .
In Arabidopsis, MPK3, MPK4 and MPK6 are activated by flg22 and other PAMPs . This activation occurs within 5 min of treatment, even in the presence of the translation inhibitor cycloheximide, indicating a direct link between receptors and the initiation of the MAPK signalling pathways . Flg22 perception by Arabidopsis became one of the main models to study the early signalling events upon pathogen recognition and is now broadly used in the scientific community.
Antagonistic control of flg22-induced defence responses by two MAPK cascades
Transient expression in protoplasts, together with biochemical and genetic approaches, identified the first MAPK cascade acting downstream of the FLS2–BAK1 receptor complex  (Figure 1a). This cascade consists of the MAP3K MEKK1 (named also MAP3K08), two MAP2Ks, namely MKK4 and MKK5, and two MAPKs, namely MPK3 and MPK6, and allows the early flg22-induced expression of WRKY29 (WRKY DNA-BINDING PROTEIN 29) and FRK1 (FLG22-INDUCED RECEPTOR KINASE 1). Expression of constitutively active forms of MEKK1 or MKK4 activate downstream events, mimicking flg22 treatment in terms of gene regulation. Interestingly, in the mekk1 background, flg22 is still able to activate MPK3 and MPK6, suggesting redundancy at the level of the MAP3K step in the MEKK1–MKK4/MKK5–MPK3/MPK6 signalling pathway [18,19]. It is presently unclear whether this kinase is YODA, which was recently shown as the MAP3K upstream of a developmental signalling pathway otherwise consisting of MKK4/MKK5 and MPK3/MPK6 .
Flg22-regulated MPK3/MPK6 targets start to be unveiled. For example, upon flg22 treatment, MPK6 specifically phosphorylates – and therefore stabilizes – ACS6 [ACC (1-amino-cyclopropane-1-carboxylic acid) SYNTHASE 6], an enzyme involved in biosynthesis of ET [21,22]. Interestingly, ET production is triggered by flg22. More recently, VIP1 (VirE1-INTERACTING PROTEIN 1), a bZIP (basic leucine zipper) transcription factor, was shown to be specifically phosphorylated by MPK3 upon flg22 treatment, inducing its re-localization from the cytoplasm to the nucleus to activate defence-related genes such as PR1 (PATHOGENESIS-RELATED 1) . The question which of the two signalling pathways regulates flg22-induced early genes WRKY29 and FRK1 remains to be addressed. However, MPK3 and MPK6 probably also target other proteins. Indeed, using a protein-chip approach, several putative targets were phosphorylated in vitro by these MAPKs, but this has not yet been confirmed in vivo .
The second MAPK cascade recruited by flg22 consists of the MAP3K MEKK1 and MPK4 (Figure 1a). MPK4 activity increases within a few minutes upon flg22 treatment , and this activation is abolished in mekk1 mutants [18,19]. mpk4 and mekk1 mutants have related phenotypes. Both mutants exhibit dwarfism, a phenotype which is likely linked to the overproduction of SA and H2O2 [18,19,25]. This dwarf phenotype is partially reversed by overexpression of the bacterial SA hydrolase nahG or by mutations in genes involved in the SA signalling pathways or biosynthesis [19,25]. Moreover, mekk1 and mpk4 mutants show spontaneous cell death on leaves and constitutive expression of pathogenesis-related genes such as PR1 and PDF1.2 (PLANT DEFENSIN 1.2). Consequently they are both more resistant to infection from pathogens such as Pseudomonas syringe or fungi [19,25]. Interestingly, yeast two-hybrid experiments showed that the MAP3K MEKK1 is able to interact with MPK4 through a short motif in the MEKK1 N-terminal tail . Taken together, these results show that MEKK1 and MPK4 are important elements of a flg22-induced signalling cascade involved in negative regulation of pathogen- or PAMP-induced cell death. Despite extensive investigations, no MAP2K mutants have yet been identified showing a mpk4- or mekk1-like phenotype, suggesting functional redundancy at this level. However, MKK1 and MKK2 were both identified to be specific upstream regulators of MPK4 , and Meszaros et al.  showed that flg22 fails to activate MPK4 in mkk1 mutants. Using a hyperactive form of MKK2, Brader et al.  showed enhanced MPK4 activation and altered pathogen responses, suggesting that MKK1 and MKK2 could perform redundant MAP2K functions on the flg22-induced MEKK1–MPK4 cascade.
The currently known targets of MPK4 are MKS1 (MPK4 substrate 1), as well as WRK25 and WRK33, two WRKY type transcription factors . Even if not tested upon flg22 treatment, these substrates could be targets of the flg22-activated MPK4 cascade. This is supported by the fact that MKS1-overexpressing plants show a weak mpk4 phenotype in terms of PR1 gene expression, SA accumulation, dwarfism and resistance to pathogens .
Are MPK3/6 and MPK4 cascades common signalling modules for various PAMPs?
So far, different studies have revealed that two MAPK signalling pathways are activated by the PAMP flg22. MKK4/MKK5–MPK3/MPK6 activates pathogen responses, whereas MEKK1–MPK4 acts negatively on the same responses. Some results suggest that these two distinct cascades are more interconnected than expected. For example, in mkk1 mutants , flg22-dependent activation of MPK3–MPK6 is impaired, which could indicate that MKK1 acts redundantly in the MPK3/6 branch of the signalling cascade. An alternative explanation could be that in the mkk1 mutant background, players in the MPK3/6 cascade are not expressed/regulated in the same way. Moreover, it was recently shown that the protein phosphatase AP2C1 is able to regulate both MPK6 and MPK4, confirming that neither of the MAPK cascades is regulated autonomously .
EF-Tu (elongation factor thermo-unstable), a peptide from the Escherichia coli elongation factor, was recently identified to trigger various pathogen-related events from activation of MAPKs to gene regulation . The authors identified EFR (EF-Tu RECEPTOR) as the EF-Tu sensor, belonging to the superfamily of RLKs (RECEPTOR-LIKE KINASES). Strikingly, despite being sensed by independent receptors, flg22 and EF-Tu regulate a very similar set of genes, indicating that they use the same signalling pathways . This hypothesis is supported by the fact that these two elicitors do not have additive effects, indicating the saturation of the signalling pathways. Very recently, using a reverse genetic approach on candidate RLKs, the receptor CERK1 (CHITIN ELICITOR RECEPTOR KINASE 1), sensing the fungal elicitor chitin, was identified . Chitin also activates MPK3 and MPK6, and this activation is totally impaired in the cerk1 mutant. This result indicates that flg22-activated MAPK pathways could be also recruited by fungus-derived PAMPs and, more importantly, that such a signalling network could be downstream of many PAMP-sensing receptors. Further characterization of other kinases upon chitin treatment will also help us to understand to what extent the signalling networks overlap. As a small step, Miya et al.  have shown that chitin in wild-type plants regulates a set of pathogen-related genes whose overlap with the ones regulated by flg22 and EF-Tu could help to test this hypothesis.
In contrast with flg22, EF-Tu and chitin, harpin from Ps. syringue and NLPs (Nep1-like proteins) from various oomycetes trigger HR-related PCD in Arabidopsis and other species. Desikan et al. [35,36] showed that harpin activates both MPK4 and MPK6 in Arabidopsis cell cultures, and NLPs induce activation of MAPKs as well . Even if these kinases are not always identified at the molecular level, these results suggest that the same MAPKs might be targeted by harpin, NLPs, flg22 and EF-Tu. Moreover, about 50% of the flg22-regulated genes are also regulated on treatment with NLP . A question that needs to be addressed is whether the relative strength of the signal spreading through either the MPK6 or the MPK4 branches might determine different responses, such as the occurrence of PCD. Another possibility is that HR might not be determined by the MPK4 and MPK6 pathways alone, but also by the activity of other, yet unknown, cascades or components. In any case, it is clear that the flg22 system has helped us to map some signalling pathways, but our knowledge is clearly not sufficient to understand the whole range of responses observed after a plant–microbial interaction.
Are AOS acting as early second messengers of PAMP responses?
AOS production is one of the earliest known signalling step in response to environmental stimuli, such as pathogen interactions – or PAMP treatment – and abiotic stresses, but also in response to various hormones and in developmental processes. The plant AOS content controls the expression of many genes and the activity of cellular enzymes. In addition, upon pathogen attack, an oxidative burst occurs, which is supposed to have an antimicrobial effect, be involved in cell-wall strengthening to counteract pathogen progression and to promote HR-related cell death. Exogenous H2O2 was shown to activate various Arabidopsis MAPK cascades, including MPK3/MPK6 (Figure 1b). Using a protoplast system, Kovtun et al.  showed H2O2-dependent activation of ANP1 [Arabidopsis NPK1 (Nicotiana protein kinase 1)-related kinase 1; also termed MAP3K01] and the downstream activation of MPK3 and MPK6. The MAP2Ks involved in this signalling pathway could, once again, be MKK4/MKK5. Indeed, Ren et al.  showed that the expression of hyperactive forms of MKK4 and MKK5 in Arabidopsis activates MPK3/6 and thereby triggers H2O2 production and cell death. MKK4 and MPK6 seem to be part of a feed-forward loop, since Doczi et al.  recently showed that MKK4 is involved in H2O2-dependent activation of MPK6. Several other investigations completed the picture by showing that H2O2-dependent MPK3/MPK6 activation is also modulated by NDP kinase 2 , which, unexpectedly, directly interacts with the two MAPKs. OXI1 (OXIDATIVE SIGNAL-INDUCED 1) protein kinase is also an upstream actor of this MPK3/MPK6-containing cascade: oxi1 mutants are impaired in H2O2-dependent activation of MPK3 and MPK6 and compromised in AOS-dependent developmental processes such as root-hair elongation, but also in basal resistance to the fungal pathogen Peronospora parasitica (downy mildew) . How OXI1 interacts with MPK3/MPK6 remains to be addressed. Another H2O2-induced MAPK cascade was recently defined as containing MEKK1 and MPK4 . Upon H2O2 treatment, MEKK1 is activated and, in mekk1 mutants, H2O2-dependent activation of MPK4, but not of MPK3 and MPK6, is impaired, confirming that MEKK1 is necessary only for the MPK4 branch. Adding further complexity, Doczi et al.  recently dissected the MKK3–MPK1/2/7/14 signalling pathway. This cascade is involved in defence against Ps. syringae, but, surprisingly, is not induced by flg22. On the other hand, H2O2 triggers the MKK3-dependent activation of MPK7 through the inhibition of the proteasome-dependent degradation of MKK3. In addition, Ortiz-Masia et al.  recently showed that MPK1 and MPK2 are also activated upon H2O2 treatment. Taken together, these results suggest that AOS act upstream of several MAPK cascades. A striking feature about H2O2-dependent activation of MAPK cascades is the stabilization of kinases that has been observed for MEKK1 , MKK3 and MKK4 . The question of whether H2O2 acts as a general activator or potentiator of the MAPK-related kinases in plant cascades remains to be addressed.
The flg22-induced oxidative burst in Arabidopsis leaves is caused by activation of the NADPH oxidase RbohD (Respiratory-burst oxidase protein D) . This change in activity occurs within 3 min of treatment and is triggered by phosphorylation of two S-Q sites in the C-terminal part of the RbohD protein. The rapid change in activity, together with the fact that these S-Q sites are probably not targeted by MAPKs which recognize S/T-P sites, suggest that MAPK cascades are downstream of this RbohD-dependent oxidative-burst step, nicely fitting the model of flg22-induced H2O2-dependent MAPK activation. However, this might not be true for all PAMPs, as shown for pep 13, a 13-amino-acid-long peptide derived from the oomycete Phytophthora sojae (soybean stem and root rot) growing on parsley (Petroselinum crispum) suspension cells. In this system, DPI (diphenylene iodonium), an inhibitor of NADPH oxidase, was shown to be unable to block pep 13-induced MAPK activation  and MAPK activation was not induced by external H2O2 . This result suggests that, opposite to flg22, pep13-induced MAPK activation does not require NADPH-dependent AOS production.
mpk4 and mekk1 mutants show a strong accumulation of AOS as detected by DAB (3,3′-diaminobenzidine) staining [25,43]. Moreover, a number of genes encoding proteins involved in the production and detoxification of AOS are misregulated. There is also evidence for a feedback mechanism, since the MEKK1/MPK4 cascade is itself regulated by AOS . Taken together, these results suggest an important function for the MEKK1–MPK4 cascade in the regulation of AOS homoeostasis.
MAPK cascades are targets of pathogen virulence factors
Successful pathogens have developed mechanisms to suppress plant defences. An important strategy consists in injecting a set of virulence factors into the host plant cell to suppress immunity . Some of these virulent factors were shown to target MAPK cascades activated upon pathogen infection. In a recent study, He et al.  showed that Ps. syringae AvrPto and AvrPtoB effectors were potent suppressors of MAPK signalling by targeting an upstream activation step. More directly, the Pseudomonas HopAI1 effector inhibits Arabidopsis MAPK by dephosphorylating the phosphothreonine (pT) in the pTEpY motif of the MAPK activation loop .
Another striking mechanism of the war between plants and pathogens is demonstrated by the hijacking of the pathogen-activated MPK3 defence pathway by agrobacteria . Agrobacteria and flg22 trigger the MPK3-dependent phosphorylation of the bZIP transcripion factor VIP1, which consequently re-localizes from the cytoplasm to the nucleus to activate pathogen-response genes. Agrobacteria make use of this nucleocytoplasmic shuttle system to transfer their T-DNA (Transfer DNA) into the plant nucleus before chromosomal integration.
MAPK cascades modulate and are modulated by stress-related phytohormones
The endoplasmic-reticulum-localized ET receptor ETR1 (ET RESPONSE 1) targets a MAP3K of the Raf subfamily termed CTR1 (constitutive triple reponse 1) [50,51]. Forward genetic approaches failed to identify the downstream kinases of the cascade. Nevertheless, ET was shown to activate a 47 kDa MAPK-like kinase  that was identified as MPK6 . ET-dependent MPK6 activation required early players in the signalling pathway, such as ETR1, but not the late ones such as EIN2 (ETINSENSITIVE 2), indicating a role of MPK6 during early signalling steps or in a parallel independent signalling pathway . The picture of ET signalling was very recently elegantly completed with the identification of MKK9 as the upstream activator of both MPK3 and MPK6  (Figure 2a). Upon ACC treatment, MKK9 re-localizes from the cytoplasm to the nucleus to activate MPK3 and MPK6. Coherently in mkk9 mutants, ACC-induced MPK3 and MPK6 activation fails and these mutants exhibit some ET-insensitive phenotype traits such as a slight ACC insensitivity of hypocotyl elongation and ET-responsive gene misregulation. ACC-activated MPK3 and MPK6 are able to phosphorylate the ET-related transcription factor EIN3 [55,56], which is thereby stabilized and able to regulate downstream gene expression. The authors also showed that an MKK9–MPK3/6-independent EIN2-dependent pathway is activated upon ACC treatment to conjointly regulate gene expression . The MAP3K upstream of MKK9 is still a matter of debate: CTR1 belongs to the Raf-kinase family, which acts in other systems as an MAP2K activator, but genetic data indicate that CTR1 is a negative regulator of ET signalling. Direct physical interaction between CTR1 and MKK9 could help to clarify this.
Many abiotic and biotic factors activate MPK6 and might thereby regulate stress-induced ET production  (Figure 2a). This hypothesis is supported by the identification of ACS6 as a substrate of MPK6 [21,22]. The upstream activators of this ET-producing signalling pathway are probably MKK4 or MKK5, since expression of hyperactive MKK4 or MKK5 triggers MPK3 and MPK6 activation and sustained ET production. Interestingly, inducible expression of hyperactive MKK5 also triggers ET-dependent PCD, probably corresponding to ET overaccumulation above a threshold .
The question how these two ET-related cascades, despite sharing kinases, do not interfere, remains to be investigated. One explanation could be that the proteins involved in each of the cascades are not located in the same compartment. This idea is supported by the fact that activated MKK9 relocalizes from the cytosol to the nucleus, whereas active MKK4 is located in the cytosol . Each of these MAP2Ks could possibly activate a subpopulation of MPK6, which is located in both compartments and could therefore phosphorylate distinct targets, namely ACS6 in the cytosol and EIN3 in the nucleus.
MKK3-dependent MPK6 activation was recently identified as part of a negative regulator of JA signalling  (Figure 2b). MPK1 and MPK2 are also activated by MeJA (methyl jasmonate) . Together with MPK7 and MPK14, these two MAPKs belong to the C subgroup of Arabidopsis MAPKs. Since MPK7, but not MPK6, was shown in planta to physically interact with MKK3 , MeJA activation of MPK6 could be a secondary consequence of MeJA-dependent MKK3–MPK1/2/7/14 activation.
MPK4 is involved in the modulation of the balance between SA and JA/ET-related defences . An mpk4 mutant accumulates SA, constitutively expresses SA-related PR genes and is impaired in ET- and MeJA-dependent gene regulation. Interestingly, eds1 (enhanced disease susceptibility 1) and pad4 (phytoalexin deficient 4) mutants and transgenic plants with decreased SA contents partially revert the mpk4 phenotype: double mutants are less dwarfed than mpk4, contain much less SA and show partial restoration of MeJA-dependent PDF1.2 expression. These results indicate that MPK4 acts on the hormonal balance through the EDS1/PAD4 module . Given the mutant phenotype similarities and the fact that MEKK1 physically interacts with MPK4 in yeast two-hybrid assays , MEKK1 could act upstream of MPK4 to modulate the defence-related hormonal balance. Interestingly, expression of a hyperactive form of MKK7 also triggers a higher SA content, constitutive PR1 expression and pathogen resistance . This suggests that an unknown MAPK cascade could antagonistically modulate this balance. Given the antagonistic roles of the MEKK1–MPK4 and MKK4/MKK5–MPK3/MPK6 cascades during the immune response and MPK6 regulation of ACS6 activity and ET synthesis, the MPK3/MPK6 branch is of course a very good candidate for such a role.
MAPKS INVOLVED IN ABIOTIC STRESSES
In several species, including Arabidopsis, MAPK cascades have also been shown to be involved in signalling pathways activated by abiotic stresses such as cold, salt, touch/wind, wounding, heat, UV or osmotic shock. Once again, the main activated kinases are MPK3, MPK4 and MPK6, or their orthologues in other species.
MAPK cascades in salt and cold stresses
The most complete MAPK cascade functioning in abiotic stresses consists of the MAP3K MEKK1 activating MKK2 and MPK4/MPK6  (Figure 3a). In the mkk2 background, cold and salt activation of MPK4 and MPK6 are impaired and mkk2 mutant plants are hypersensitive to cold and salt stresses. Using the protoplast expression system, the authors showed that whereas MKK2 activates MPK4 in response to cold and salt, activation of MPK4 by H2O2, flg22 and β-glucan is triggered by MKK1 (Figure 3b). This result is apparently in contradiction with genetic data suggesting that MEK1 (MKK1) also functions in salt tolerance .
Water stress, ABA (abscisic acid) and osmotic shock
ABA is a phytohormone involved in various physiological processes, including the adaptation to water stress and the control of seed dormancy. The fact that a MAP2K inhibitor, PD98059, was able to decrease ABA-induced stomatal closure indicates that a MAPK cascade could function in ABA signalling  (Figure 3c). More direct evidence came from guard-cell-specific MPK3 silencing in Arabidopsis . In these RNAi (RNA interference) lines, ABA partially failed to maintain stomatal closure upon opening conditions, indicating a function of MPK3 in ABA-dependent inhibition of the stomatal opening machinery, but not of ABA-induced closure. Additionally, H2O2, known to be an important second messenger of ABA in the closure of stomata, gave very similar results, indicating MAPK activation downstream of ABA-induced AOS production. Stomata of knock-out plants of the MAP2K gene MEK1 (also named MKK1) were also found to be unable to respond to ABA and to be impaired in ABA-induced expression of the leaf catalase gene CAT1 . To our knowledge, MKK1 has not been shown to regulate MPK3 activity in response to ABA. More indirect genetic data of a function for a MAPK cascade in ABA signalling comes from the study of two mutants impaired in putative MKPs: ibr5 (indole-3-butyric acid response 5) showed partial resistance to ABA in a root elongation assay but was unfortunately not tested for its stomatal response to ABA . By contrast, phs1-3 (propyzamide-hypersensitive 1-3) exhibits ABA hypersensitivity in seed germination and stomatal opening [64a]. ABI1 (ABA-INSENSITIVE 1), a member of the PP2C (protein phosphatase 2C) family known to function in ABA signalling, was also shown to interact with MPK6 in vitro, in yeast two-hybrid assays and in planta . Additionally, expression of dominant-negative forms of MPK6 rendered plants hypersensitive to ABA . Direct proof of MAPK activation by ABA was obtained recently by Ortiz-Masia et al.  for MPK1 and MPK2, who used an Arabidopsis protoplast expression system and revealed these two MAPKs as putative candidates mediating ABA signalling in plant cells.
An important component of water stress is the osmotic shock created by washing or concentration of the extracellular medium. In suspension-cultured Arabidopsis cells, hyperosmotic and hypo-osmotic shock were shown to activate MPK3, MPK4 and MPK6 [16,66] (Figure 3d). Interestingly, even if MPK4 is not activated by hyperosmotic shock in plantlets, mpk4 mutants exhibit resistance to osmotic stress in germination and root-growth assays, indicating a negative function in osmotic tolerance . Interestingly, in Arabidopsis and other species, osmotic shock is known to trigger a strong AOS production, which could be the input signal activating the MAPK network.
O3 (ozone)-induced MAPK cascade
The major pollutant O3 was shown to activate a MAPK signalling pathway through AOS production and to modify ET, SA and JA contents, leading to a PCD similar to HR. These similarities between responses to pathogens, H2O2 and O3 allowed researchers to use O3 to manipulate the H2O2 content of plants. In Arabidopsis, Ahlfors et al.  observed MPK3 and MPK6 activation within 30 min of exposure to O3 (Figure 3e). MAPK activation was normal in ET, SA and JA signalling mutants, indicating (i) the existence of a hormone-independent pathway for MPK3 and MPK6 activation and (ii) that O3-induced hormone changes could be a long-term process under the control of the MAPK cascade. Plants with diminished MPK6 or lacking MPK3 exhibit O3 hypersensitivity, as shown by O3-induced leaf damage . Interestingly, in each of these mutant backgrounds, the other MAPK remained activated longer . Recently, MKP2 was identified as an important regulator for controlling both O3-induced MPK3, and MPK6 and MKP2 RNAi plants were shown to exhibit hypersensitivity to O3 .
STRESS-UNRELATED FUNCTIONS FOR MAPK CASCADES
A number of studies have also revealed the role of MAPK pathways in other processes, including developmental and auxin signalling.
YODA–MKK4/MKK5–MPK3/MPK6 regulates stomatal patterning in epidermis
The main finding concerning MAPK function in development came from studies on the MAP3K YODA (also named MAP3K04) and its downstream MAPK components MKK4/5 and MPK3/6 (Figure 3f). yoda was initially identified in a genetic screen for plants impaired with respect to normal cell division during early embryo development . Another study revealed that yoda mutants also exhibit typical clustering of stomata in the leaf epidermis corresponding to a transgression of the ‘one-cell spacing’ rule . Using a reverse-genetic approach, MKK4–MKK5 and MPK3–MPK6 were then shown to be redundant downstream steps of this MAPK signalling cascade: both mkk4mkk5 and mpk3mpk6 double mutants fail to develop further than the cotyledon stage and show a typical yoda-like stomatal patterning at the epidermis . Using a dominant-negative form of MPK6 to by-pass functional redundancy, Bush confirmed MPK6 function in stomatal patterning and also published work suggesting a convincing role for it during embryo and inflorescence development . The embryo-development phenotype in mkk4mkk5 and mpk3mpk6 double mutants still remains to be studied. Despite the upstream signal of YODA–MKK4/5–MPK3/6 remains still being unknown, it defines unambiguously a function for MAPKs in the intercellular communication during development.
MAPK cascades and auxin signalling
Interesting results on the function of MAPKs in auxin signalling came from the phenotypic study of the semi-dominant bushy and dwarf1 (bud1) Arabidopsis mutant, which turned out to be an overexpresser of MKK7 . Auxin was known to trigger MAPK-like activities in tobacco (Nicotiana tabacum)  and Arabidopsis . Using tobacco and maize (Zea mays) protoplasts, however, Kovtun et al. previously showed a negative function for a MAP3K NPK1-dependent MAPK cascade in auxin signalling . Genetic evidence for a role of the Arabidopsis NPK1 homologues ANP1, ANP2 and ANP3 as postive regulators in cytokinesis was provided in a reverse-genetic approach. Whereas single-mutant plants displayed no obvious abnormal phenotypes, double-mutant combinations displayed defects in cell division and growth [76a]. Whether ANPs have a role in auxin signalling is currently unclear and remains to be determined. A role for MKK7 in auxin transport was recently unravelled by analysing the phenotypes of the bud1 mutants which exhibit several auxin-related phenotypes, such as less ramifications in root architecture, a simpler venation of leaves, an enhanced curvature of the hypocotyls in response to gravity and a modified polar auxin transport. MKK7 could thus act as a negative regulator of auxin transport . Later the same authors reported a MKK7 function in defence regulation . Studies of loss-of-function mkk7 alleles should help to clarify the auxin-related MKK7 function in planta. Support for the notion that auxin signalling could involve a MAPK cascade also comes from studies of the MKP mutant ibr5, which was isolated on the basis of its auxin resistance and which shows, in common with some other auxin-related mutants, aberrant vascular patterning, longer roots and shorter hypocotyls . The IBR5-inactivated MAPKs remain to be identified.
In Arabidopsis mekk1 and mpk4 mutants, Nakagami et al.  showed an impaired auxin-induced activation of the auxin-regulated genes IAA3 (indole-3-acetic acid 3), IAA5 and IAA14, a phenotype that was suggested to arise as a consequence of H2O2 down-regulation of auxin signalling. Despite extensive forward-genetic approaches, no mutants of MAPK components were isolated with an auxin phenotype, probably attributable to redundancy in the extensive MAPK gene family.
MAP3Kϵ1 and MAP3Kϵ2 function during pollen maturation
A recent study  has shown that MAP3Kϵ1 and MAP3Kϵ2 (also known as MAP3K07 and MAP3K06) function redundantly during male gametophyte maturation. Whereas single mutants do not show any obvious phenotype, homozygous double mutants never recover, because of a lack of viable pollen grains. Additionally, the authors showed that double-mutant pollen grains often fail to go beyond the second mitosis, showing, as they do, additional plasma-membrane structures and pollen-wall modifications. Interestingly, MAP3Kϵ1 and MAP3Kϵ2 were previously reported to be expressed acutely during the cell cycle [77a]. Chaiwongsar et al.  did not exclude the possibility that MAP3Kϵ1 and MAP3Kϵ2 could also act redundantly in mitosis during the sporophytic phase. This question remains difficult to tackle because of the lack of viable double-mutant plants.
Many kinases with unknown functions
On the 120 kinases putatively involved in MAPK cascades, only a handful have been characterized so far. By far the most studied MAPKs are MPK3, MPK4 and MPK6, which were identified in many distinct processes ranging from stress responses to developmental processes. Some activity data are now available for some of the other 17 MAPKs, but, to our knowledge, not a single loss-of-function mutant has been reported yet, even though insertion lines for almost all genes are available in various insertion mutant databases. An open question remains as to why it is so easy to monitor the activity of MPK3, MPK4 and MPK6 from plant extracts by in gel kinase assays, whereas the other MAPK activities are so rarely observed. Could this be for technical reasons? Are those MAPKs activated in such discrete conditions that they escaped detection? The situation is better for the ten MAP2Ks: the smaller number of genes, together with the possibility to generate constitutively active forms, enabled one to attribute functions to the majority of the MAP2Ks. Only the functions of MKK8 and the atypical MKK10 remain unknown. Nevertheless, gain-of-function approaches, such as strong overexpression of hyperactive kinases, indicate roles in pathogen responses, but they could also hide more subtle functions in planta. With the notable exception of some well studied kinases such as CTR1, MEKK1 and YODA, very little is known about the function or mechanism of action of the plant MAP3Ks on their downstream MAP2Ks. Moreover, no information is yet available to explain how the MAP3Ks are connected to their respective sensors and receptors.
A few kinases do all?
Plants use the same signalling pathways in very different signal-transduction processes. For example, MPK6 is involved in O3, PAMP, H2O2, ET, ABA and JA signalling pathways, but also in important developmental processes such as epidermal patterning and anther and embryo development. At this time it is difficult to conclude whether the involvement in many signalling pathways is a common feature of the 20 MAPKs or is specific for the three well-known MAPKs, namely MPK3, MPK4 and MPK6.
Nevertheless, one of the explanations for MPK6's various roles could be that all input signals in fact recruit a common second messenger, which activates a conserved signalling pathway. In this scenario, AOS production could be a good candidate, since an oxidative burst is known to be generated in response to many stresses, to several phytohormones and also during developmental processes such as root-hair growth. To our knowledge, H2O2 has not yet been shown to be involved in positioning-dependent differentiation during epidermal patterning, but such processes should be discrete and therefore difficult to detect. Even though the notion cannot be excluded, it is difficult to imagine that the set of genes and responses regulated by MPK6 during a developmental process or during a stress response are the same. Specificity could be under control of parallel signalling pathways that remain to be identified.
A second explanation why MPK6 is a common mediator of several distinct cascades could be that the other members and targets of the pathway are only expressed in particular cell types, in particular subcellular compartments, at particular developmental stages or under particular environmental conditions. This notion is supported, for example, by the fact that functional interaction of MPK6 was demonstrated for a wide set of MAP2Ks such as MKK2 , MKK3 , MKK4, MKK5  and MKK9 . The fact that MPK4 activity increases in response to an osmotic shock in Arabidopsis suspension cells but not in plantlets supports the idea of cell-specificity for MAPK cascades . ET-related functions of MPK6 could provide an example of subcellular-dependent function: MKK4–MPK6–ACS6 is used to control ET production, whereas MKK9–MPK6–EIN3 is involved in ET signalling. Yoo et al.  showed that active MKK4 is mainly a cytoplasmic protein, whereas active MKK9 re-localizes in the nucleus. Both MAP2Ks could target subpopulations of MPK6, either in the nucleus or in the cytoplasm, which thereby phosphorylate distinct targets and initiate different functional responses. Integration of expression data of MAPK-related kinases will help to simplify the system and highlight kinases susceptible to meet and build conditional cascades. In this scenario, scaffolding proteins have particularly important functions, as they help kinases to build functional cascades.
Abbreviations: ABA, abscisic acid; ACC, 1-aminocyclopropane-1-carboxylic acid; ACS6, ACC synthase 6; ANP1, Arabidopsis NPK1 (Nicotiana protein kinase 1)-related kinase 1; AOS, active oxygen species; BAK1, BRI1-associated kinase 1; bZIP, basic leucine zipper; CERK1, chitin elicitor receptor kinase 1; CTR1, constitutive triple reponse 1; EDS1, enhanced disease susceptibility 1; EF-Tu, elongation factor thermo-unstable); EIN2, ethylene-insentive 2; ET, ethylene; Flg22, a 22-amino-acid-long peptide derived from flagellin; ETR1, ET response 1; FLS2, FLAGELLIN SENSING 2; HR, hypersensitive response; JA, jasmonic acid; MAPK/MPK, mitogen-activated protein kinase; MAP2K, MAPK kinase; MAP3K, MAP2K kinase; MeJA, methyl jasmonate; MEKK, MAPK/ERK kinase kinase; MKP, MAPK phosphatase; MKS1, MPK4 substrate 1; NLP, Nep1-like protein; OXI1, oxidative signal-induced 1; PAD4, phytoalexin-deficient 4; PAMP, pathogen-associated molecular pattern; PCD, programmed cell death; PR1, pathogenesis-related 1 gene; RbohD, respiratory-burst oxidase protein D; RLK, receptor-like kinase; RNAi, RNA interference; SA, salicylic acid; VIP1, VirE1-interacting protein 1
- © The Authors Journal compilation © 2008 Biochemical Society