MAP kinase signal transduction pathways in plants


  • Peter C. Morris

    Corresponding author
    1. Heriot-Watt University, Department of Biological Sciences, Riccarton, Edinburgh, EH14 4AS
      Author for correspondence: Tel: +44 131 451 3181 Fax: +44 131 451 3009
    Search for more papers by this author

Author for correspondence: Tel: +44 131 451 3181 Fax: +44 131 451 3009


The mitogen-activated protein kinase (MAP kinase) signal transduction cascades are routes through which eukaryotic cells deliver extracellular messages to the cytosol and nucleus. These signalling pathways direct cell division, cellular differentiation, metabolism, and both biotic and abiotic stress responses. In plants, MAP kinases and the upstream components of the cascades are represented by multigene families, organized into different pathways which are stimulated and interact in complex ways. Experimental strategies for the analysis of MAP kinase cascades include the yeast two-hybrid system; using this approach in vitro interactions between specific MAP kinase cascade components have been analysed and putative plant cascades postulated. Transient transformation of protoplasts with epitope-tagged kinases has allowed cascades to be tested in planta. There is clear evidence for the involvement of MAP kinases in plant cell division and in the regulation of auxin signalling. Biotic (pathogens and pathogen-derived elicitors from fungi, bacteria and viruses) and abiotic stresses including wounding, mechanical stimulation, cold, drought and ozone can elicit defence responses in plants through MAP kinase pathways. There are data suggesting that ABA signalling utilizes a MAP kinase pathway, and probably ethylene and perhaps cytokinins do so also. The objective of this paper is to review this rapidly advancing field.


Summary 67

I. Introduction 68

II. Background 68

III. MAP kinase targets and targeting specificity 69

IV. Assays and inhibitors 70

V. Two well characterized MAP kinase pathways, Hog1 and Sevenless 71

VI. MAP kinases in plants 73

VII. MAP kinases and cell division 76

VIII. MAP kinases and plant hormones 76

IX. MAP kinase and abiotic stress78

X. MAP kinase and biotic stress 80

XI.Future perspectives for MAP kinase research in plants 83

Acknowledgements 84

References 84

I. Introduction

In order for living cells to respond and adapt to external circumstances, changes in the extracellular environment must be communicated in a specific manner from outside of the cell to the inside, and ultimately to the nucleus where changes in gene expression may occur. During the course of evolution, cells have evolved many different mechanisms to accomplish this. Underlying many of these mechanisms are the processes of protein phosphorylation by specific protein kinases and de-phosphorylation by protein phosphatases (Hunter, 1995; Hardie, 1999). Activation and de-activation of enzymes through phosphorylation/de-phosphorylation by kinases and phosphorylases allows for fast and specific signal transduction, and amplification of external stimuli (Brown et al., 1997). One particular signal transduction mechanism, the MAP kinase cascade, plays an important role in many different eukaryotic organisms, from yeast, through Dictyostelium, Drosophila and Caenorhabditis to mammals, and also plants. An explosive increase in research and the literature on MAP kinase pathways has taken place in recent years, reflecting the importance and complexity of these signalling pathways. MAP kinase pathways are perhaps best understood and researched in the mammalian and yeast fields. However there has also been much interest and work on these signal transduction modules in the plant field, as reflected by the number of reviews devoted to MAP kinase pathways in plants (Jonak et al., 1994; Hirt, 1997; Mizoguchi et al., 1997; Hirt, 2000; Ichimura et al., 2000a; Meskiene & Hirt, 2000). The core of the MAP kinase pathway is formed by three protein kinases, MAP/ERK kinase ⇒ MEK ⇒ MAP kinase, which phosphorylate and activate each other in a linear pathway. The purpose of this review is to try and unravel some of the biochemical complexity that surrounds this simple cascade, and to illustrate the manifold processes that are regulated by MAP kinase pathways, and the progress that has been made to date in understanding these pathways in plants.

II. Background

MAP kinases first came to light when Sturgill & Ray (1986) identified a protein kinase from insulin-treated 3T3-L1 cell extracts that would phosphorylate microtubule associated protein-2 (MAP-2) on both serine and threonine. MAP-2 kinase was shown to be closely related to a set of previously identified proteins which are tyrosine phosphorylated in response to mitogens (agents which promote cell division) and was renamed p42 MAP kinase (Mitogen-Activated-Protein kinase) (Rossomando et al., 1988). It was later shown that the activity of the p42 MAP kinase protein was itself dependent on phosphorylation of both threonine and tyrosine (Anderson et al., 1990). Boulton et al. (1990) isolated and cloned an insulin activated protein kinase from chinese hamster ovary (CHO) cells, named Extracellular signal Regulated Kinase 1 (ERK1) or p44 MAP kinase, very similar in sequence to p42 MAP kinase. The protein sequence of ERK-1 is also 56% identical to the Saccharomyces cerevisiae proteins Fus3 and Kss1, protein kinases involved in the yeast cell cycle during mating. Further biochemical and molecular biological studies revealed yet more MAP kinase-like genes and proteins, and it soon became apparent that these enzymes are represented by multigene families in eukaryotic species. Despite the general name-mitogen activated protein kinase-, this family of kinases is not only activated by mitogens such as Epidermal Growth factor (EGF) or insulin, there are many other elicitors, such as hyper or hypoosmolarity, UV light, genotoxic agents, inflammatory mediators, thrombin, heat shock or mechanical stretching. MAP kinase pathways have now been implicated in a multiplicity of biological events including apoptosis (in which ERK and Jun N terminal kinase (JNK) or p38 MAP kinase pathways have opposing effects, Xia et al., 1995), transcription, translation, nucleotide synthesis (Graves et al., 2000), stress responses (Nebreda & Porras, 2000), cytoskeleton dynamics (Shiina et al., 1992), and the inflammatory response (Lin et al., 1993).

Active MAP kinase requires phosphorylation on tyrosine and threonine in the conserved threonine-x-tyrosine (TXY) sequence in kinase subdomain VIII (Payne et al., 1991) and can be de-activated by both tyrosine and serine/threonine specific phosphatases (Anderson et al., 1990). The yeast MAP kinase Kss1 has an inhibitory function in the unphosphorylated form (Cook et al., 1997). Since protein kinases are generally specific for either serine/threonine, or tyrosine, it was thought that two protein kinase pathways might operate to activate MAP kinase. Surprisingly, single protein dual-specificity activators were identified from EGF-treated Swiss 3T3 cells (Ahn et al., 1991). Subsequently it was shown that the MAP kinase activator is indeed a dual-specificity serine/threonine and tyrosine protein kinase with strong specificity for MAP kinase (Matsuda et al., 1992). This MAP kinase activator was found capable of activating p42 and p44 MAP kinase but not other related MAP kinases (Kyriakis et al., 1994), indicating considerable substrate specificity for this enzyme. The activating kinase was cloned and named MEK (MAP/ERK kinase) (Crews & Erikson, 1992). Specificity between MEK and MAP kinase may be aided by docking signals in the N terminus of some MEKs (Bardwell & Thorner, 1996), and scaffold proteins that interact with both MEK and MAP kinase are also known (for example MP1, Schaeffer et al., 1998). The amino acid sequence of MEK is related to that of the serine/threonine kinases; the basis of the dual specificity is unclear, however, dual function protein kinases other than MEK have been reported (for example Ali et al., 1994). As is the case for the MAP kinases, MEK (also known as MAP kinase kinase) is represented by multigene families in eukaryotic organisms. Active MEK is itself phosphorylated on serine and threonine residues (Ahn et al., 1993), and is de-activated by serine/threonine phosphatases (Matsuda et al., 1992), suggesting that serine/threonine kinase activity is responsible for activation. The unphosphorylated form of some MEKs can act as potent negative regulators of MAP kinase signalling (Sugiura et al., 1999).

Cells transformed with the v-raf oncogene (encoding a serine/threonine protein kinase) show constitutive MEK and ERK activity, and a constitutively active form of c-Raf-1 will phosphorylate and activate MEK in vitro (Kyriakis et al., 1992). Raf activates MEK by phosphorylation of conserved serine or threonine residues in the catalytic domain of the enzyme (Alessi et al., 1994). Raf kinase activity is associated with 14–3–3 proteins; the specific dissociation of 14–3–3 from Raf will inactivate Raf (Tzivion et al., 1998). A proline-rich sequence in mammalian MEK1 and MEK2 is required for Raf binding (Catling et al., 1995). MEK activators other than Raf were subsequently discovered, such as the Ste11 protein from Saccharomyces cerevisiae (Stevenson et al., 1992), and a mouse Ste11 homologue termed MEKK (MEK kinase) or MAP kinase kinase kinase (Lange-Carter et al., 1993). These kinases do not share much sequence similarity with Raf, suggesting multiple pathways for MEK activation. Upstream of Raf or MEKK are a variety of different effectors. In mammalian systems, MAP kinase pathways are found downstream from receptor tyrosine kinases (RTKs), and G-protein coupled receptors. RTKs when activated may stimulate the exchange of guanosine triphosphate (GTP) for guanosine diphosphate (GDP) on the G protein Ras. Activated Ras can interact with a number of potential partners such as PI-3 kinase or Cdc42, but also with Raf (Schlessinger, 2000). Yeast by contrast does not possess RTKs and here, receptors upstream of MAP kinase including a two-component histidine protein kinase and G-protein coupled seven-transmembrane receptors. Other upstream intermediates in yeast include the small G-protein Rho1 and protein kinase C.

Mammalian MAP kinases are represented by three families with multiple members in each (some derived from differential splicing), and with multiple upstream MEKs and MEK kinases: these families are; the ERK/MAP kinases, generally activated by the response of receptor tyrosine kinases to growth factors such as EGF; the JNK/SAPK (Jun N-terminal kinase/stress-activated protein kinase) family, activated by stress or inflammatory cytokines; and the p38/Hog family which is activated by cytokines, endotoxins and osmotic stress. The MAP kinases in the different families are characterized by the activating sequences: generally threonine-glutamic acid-tyrosine (TEY) for ERK/MAP, threonine-proline-tyrosine (TPY) for JNK/SAPK and threonine-aspartic acid-tyrosine (TDY) or threonine-glycine-tyrosine (TGY) for p38/Hog. The upstream enzymes for these families are also distinct and multiple. Extracellular stimuli may activate more than one pathway; for example EGF will activate both ERK and JNK/SAPK (Cano & Mahadevan, 1995).

III. MAP kinase targets and targeting specificity

After stimulation by an appropriate extra-cellular agent, the MAP kinase response can be transient or sustained (Marshall, 1995). Although MAP kinases are normally classified with the serine/threonine protein kinases, they also show some tyrosine kinase activity, and may autophosphorylate on both threonine and tyrosine (Wu et al., 1991). Autophosphorylation on tyrosine may also increase the affinity of MEK for MAP kinase (Haystead et al., 1992). The phosphorylation targets of activated MAP kinases include both nuclear and cytosolic proteins, including upstream elements of the MAP kinase cascade. Target specificity is determined at least in part through docking domains in the substrate protein (Sharrocks et al., 2000). Cytosolic targets include phospholipase A2 (Lin et al., 1993), cytoskeletal proteins (Sturgill & Ray, 1986; Shiina et al., 1992), and ribosomal protein S6 kinase (Sturgill et al., 1988). Upstream targets include the EGF receptor, Son of sevenless (Sos), Raf and MEK (Denhardt, 1996). Activated MAP kinase may be also translocated to the nucleus (Chen et al., 1992). Nuclear targets of MAP kinase include transcription factors such as Jun (Pulverer et al., 1991), Myc (Seth et al., 1992), Elk1 (Gille et al., 1992), and ATF2 (Abdel-Hafiz et al., 1992). Some MAP kinases, for example ERK3, are known to be located constitutively in the nucleus (Cheng et al., 1996). The control of intracellular targeting may operate in several ways; studies on Xenopus cells have shown that MEK is retained in the cytosol because it contains a nuclear export sequence towards the N-terminus. Since MEK binds MAP kinase through the N terminal docking domain, MAP kinase is also retained in the cytosol. Activation of MAP kinase through phosphorylation by MEK promotes dissociation from MEK, followed by dimerization of MAP kinase and active nuclear import. Mammalian ERK2 undergoes a conformational change after phosphorylation by MEK such that the activation loop bearing TEY can interact with a C terminal domain termed L16 on another ERK2 molecule, promoting dimerization. MAP kinase heterodimers are not thought to form because the geometry of the interaction is very specific. The exact mechanisms leading to active import are not understood; it is possible that dimerized MAP kinase interacts with other proteins that bear nuclear localization sequences. The conformational changes that lead to MAP kinase dimerization expose a domain termed the MAP kinase insertion, which may function as a bipartite nuclear localization sequence. Additionally, dimerization may also conceal a putative nuclear export sequence (Cobb & Goldsmith, 2000). MAP kinase may enter the nucleus by passive diffusion, which does not require dimerization (Fukuda et al., 1997; Adachi et al., 1999). MAP kinase may be retained in the nucleus through MAP kinase-induced nuclear anchoring proteins (Lenormand et al., 1998).

An important feature of MAP kinase pathways is signalling specificity; with a multiplicity of MAP kinase pathway components, how does a given signal become directed to the correct intracellular target? One way in which specificity is ensured, even when MAP kinase pathways share common elements, is through enzyme complexes referred to as ‘signalosomes’ (Chang & Karin, 2001; Whitmarsh & Davis, 1998). These complexes may form between different elements of a particular pathway (for example in the Hog1 pathway, Fig. 2) or scaffold proteins may hold signalling molecules together (for example in the Sevenless pathway, Fig. 3). In this way, unwanted ‘crosstalk’ between different pathways is avoided and signalling is more rapid.

Mammalian and yeast MAP kinases are de-activated by serine/threonine protein phosphatases and tyrosine protein phosphatases, or by dual function phosphatases, which may themselves be regulated by MAP kinases (Cobb & Goldsmith, 1995; Keyse, 1998). These phosphatases may be very specific in their action with high substrate affinity for MAP kinases. For example MKP-3 is selectively activated by ERK2 which binds in a kinase-independent manner to the phosphatase and will specifically de-activate ERK family MAP kinases but not other MAP kinases such as JNK (Camps et al., 1998). Other components of MAP kinase pathways are also de-activated by phosphatases, for example Raf is de-activated by membrane-associated protein phosphatases (Dent et al., 1995). A generalized scheme illustrating the different elements in MAP kinase signalling is shown in Fig. 1.

Figure 1.

Generalized mitogen-activated protein (MAP) kinase signal transduction cascade. Signals perceived by receptors are transduced through signalling intermediates to the first core kinase, MAP/ERK kinase. Activated MEK kinase phosphorylates and thus activates MEK on conserved serine (S) and threonine (T) residues. Active MEK phosphorylates and activates MAP kinase on conserved T and tyrosine (Y) residues. MEK kinase, MEK and MAP kinase may be held together in a signalosome complex by scaffold proteins. Activated kinases are de-activated by specific phosphatases. Active MAP kinase is released from the signalosome, dimerizes and activates by phosphorylation target proteins in the cytoplasm or nucleus. P, phosphorylated aminoacid.

IV. Assays and inhibitors

Regulation of MAP kinase activity in cells can be on the basis of transcription, translation, activation or de-activation of enzyme activity by phosphorylation and de-phosphorylation, respectively. The assay of specific MAP kinase protein phosphorylation activity is the most direct indicator of whether a particular physiological event is associated with a MAP kinase signal transduction pathway. MAP kinase enzyme activity is assayed by its ability to phosphorylate a specific substrate, usually myelin basic protein (MBP), which contains a consensus phosphorylation site for MAP kinase. MAP kinases are proline-directed serine-threonine protein kinases, the preferred sequence for Erk-1 and 2 being proline-X-serine/threonine-proline where X is neutral or basic, however, the −2 proline is not an absolute requirement (Clark-Lewis et al., 1991). γ32P ATP is used as a phosphate donor and the incorporated radioactivity in MBP determined by scintillation counting. It should be noted however, that MPB is a substrate for many other protein kinases such as protein kinase C, cAMP-dependent protein kinase, and calmodulin-dependent protein kinase II. The most convincing data for MAP kinase enzyme activity therefore comes from experiments that use immunoprecipitation with antisera specific to the protein, or to an introduced epitope tag before assay (Reuter et al., 1995). In the absence of such supporting evidence, the most that can be said is that MBP protein kinase activity can be detected. Many commercial sources of antisera are available, however, the specificity and noninhibitory properties of these antisera should be carefully checked, since an antibody raised against a mammalian MAP kinase epitope will not necessarily be specific for a given plant MAP kinase. Protein phosphatase inhibitors such as sodium orthovanadate and pyrophosphate are important components of the extraction and reaction buffers for MAP kinase assays since of course de-phosphorylation will de-activate the kinase activity of the protein. A frequently used assay is the ‘in gel’ kinase assay in which the MBP substrate is incorporated into a polyacrylamide gel. Cell protein extracts or immunoprecipitates are fractionated in the gel, renatured and incubated with radiolabled ATP. The advantage of this technique is that it gives an estimate of the molecular weight of the active kinase, although, again, other protein kinases might comigrate with the putative MAP kinase (Marshall & Leevers, 1995). MEK and MEK kinase are assayed indirectly in a similar manner to MAP kinase. MEK kinase is used to activate MEK, which is used to activate MAP kinase, which in turn is then used to phosphorylate MBP (Alessi et al., 1995a; Kuroda et al., 1995; Lange-Carter & Johnson, 1995). This all assumes that the correct active partners for the relevant MEK kinase and MEK are available either as recombinant proteins or from cell extracts.

Functional complementation of mutants, usually in yeast, is a powerful method of establishing functionality of clones encoding MAP kinase cascade components. However although a given protein may act as a MAP kinase or a MEK in a particular yeast MAP kinase pathway, this does not mean that the protein need play the same physiological role in the plant, it simply establishes a property of the enzyme. Several plant MAP kinase cascade members have been functionally analysed in this way (for example Mizoguchi et al., 1996; Popping et al., 1996; Covic et al., 1999). An experimental approach which also uses yeast is the two-hybrid method of identifying and analysing interacting partners between different protein partners (Fields & Song, 1989). This has been used to analyse interactions between cloned plant genes (Mizoguchi et al., 1998), and to isolate new MAP kinase pathway partners (Ichimura et al., 1998; Liu et al., 2000).

Potentially useful tools in the analysis of MAP kinase pathway are commercially available specific inhibitors of certain steps in the pathway. PD 098059 (also referred to as PD 98059) is an inhibitor of MEK activation by Raf or MEK kinase at 10 µM concentration (Alessi et al., 1995b; Dudley et al., 1995). PD 098059 is very specific for the mammalian MEK MAPKK1 (which activates the ERK/MAP kinases MAPK1 and MAPK2) with much less effect on the activation of other MEK enzymes. The compound U0126 will inhibit both mammalian MEK1 and MEK2 activity at 10 µM, being a noncompetitive inhibitor with respect to ATP and ERK (Favata et al., 1998). SB 203580 is a specific inhibitor of p38/Hog MAP kinases, but not other MAP kinases, at concentrations of 10 µM (Cuenda et al., 1995). PD 169316 and SB 202190 are similar, but more potent compounds than SB 203580. The specificity of SB 203580 lies however, not in the TGY sequence of the activating domain of p38/Hog, but in a sequence with a conserved threonine in the ATP-binding pocket. Substitution of this sequence into other MAP kinases will render them sensitive to the compound (Gum et al., 1998). As with any inhibitor studies, great care is needed in the design and interpretation of experiments using these compounds, to avoid any misinterpretation through nonspecificity.

V. Two well-characterized MAP kinase pathways, Hog1 and Sevenless

MAP kinase pathways were first described in organisms other than plants, and progress in the analysis of these systems is correspondingly more advanced. In order to illustrate the complexities of MAP kinase pathways, two well-characterized systems are examined in a bit more detail: the yeast Hog1 pathway (Fig. 2) and the Drosophila Sevenless pathway (Fig. 3).

Figure 2.

The Hog1 pathway in Saccharomyces cerevisiae. Hyperosmotic conditions are perceived through the osmosensor Sho1, which activates the MEK (MAP/ERK kinase) kinase Ste11, which then activates the MEK Pbs2; this in turn activates the mitogen-activated protein (MAP) kinase Hog1, initiating the high osmolarity glycerol response. Pbs2 also acts as a scaffold protein to ensure specific activation of Hog1. Additionally, the two-component histidine kinase osmosensor Sln1 autophosphorylates under normal conditions, and phosphorylates Ypd1, which phosphorylates and de-activates Ssk1. Under hyperosmotic conditions, Ssk1 is not phosphorylated, and can activate the two MEK kinases Ssk22 and Ssk2, which in turn activate Pbs2 and thus Hog1. Hog1 is de-activated by the protein tyrosine phosphatases Ptp2 and Ptp3.

Figure 3.

The Rolled pathway in Drosophila. Binding of Bride of Sevenless (Boss) in cell R8 to Sevenless (Sev) in cell R7 triggers a mitogen activated protein kinase pathway which results in the differentiation of cell R8 into a functional neurone. Sev binds the adapter Downstream of receptor kinase (Drk), which in turn binds Son of sevenless (Sos). Sos binds the GTPase Ras1, causing the exchange of guanosine diphosphate (GDP) for guanosine triphosphate (GTP). This activates Ras1 which binds and activates the MEK (MAP/ERK kinase) kinase D-Raf. D-Raf in turn activates the MEK Dsor1, which activates the mitogen-activated protein (MAP) kinase Rolled, resulting in cellular differentiation. The MEK kinase, MEK and MAP kinase proteins are held together by the Ksr1 scaffold protein. Ras1 is de-activated by the GTPase activating protein Gap1. Sev is regulated through Daughter of sevenless (Dos) both positively by Corkscrew (CSW) and negatively by Phospholipase Cγ (PLCγ).

Sequencing of the yeast genome and genetic analysis has allowed the assignment of specific extracellularly stimulated signalling pathways to five of the six known MAP kinases (Hunter & Plowman, 1997). The Hog1 pathway allows yeast to adapt to conditions of high osmolarity by the production of glycerol (Fig. 2). High osmolarity results in stimulation of two osmolarity-sensitive receptors, Sho1 and Sln1. Signal transmission from Sho1 to the MAP kinase Hog1 (high osmolarity glycerol response) is accomplished through a MAP kinase cascade comprising of Ste11 (a MEK kinase), Pbs2 (MEK) and Hog1 (MAP kinase) (Maeda et al., 1995). The Ste11 MEK kinase is also functional in the Fus3 mating response cascade which involves a different MEK, Ste7. To ensure that activated Ste11 is directed to the correct downstream MEK, the MAP kinase pathways are assembled and function within protein scaffold complexes. Thus in the Hog1 pathway, the MEK Pbs2 is thought to function as a scaffold to which Ste11 and Hog1 attach, and the whole assembly interacts with the Sho1 receptor (Posas & Saito, 1997). In the Fus3 pathway, Ste11, Ste7 and Fus3 are held together on a separate scaffold protein called Ste5.

Alternatively, Hog1 can be activated by a separate pathway initiated by the two-component histidine kinase osmosensor, Sln1 (Posas et al., 1996). Under normal conditions, Sln1 autophosphorylates and in turn phosphorylates the protein Ypd1, which passes the phosphate to the response regulator Ssk1. This de-activates Ssk1. Under hyperosmotic conditions, Ssk1 is not phosphorylated and is active. Active Ssk1 in turn activates the two redundant MEK kinases Ssk2 and Ssk22. These active MEK kinases then activate Pbs2 and thus Hog1 (Maeda et al., 1994). Hog1 can be de-phosphorylated and thus de-activated by the tyrosine phosphatases Ptp2 and Ptp3, which may themselves be activated by Hog1 (Wurgler-Murphy et al., 1997). The Hog1 pathway is complementary to the Mpk1 pathway, which is activated by low osmolarity, the two pathways share some crosstalk since Mpk1 is required for Ptp2 expression (Huang & Symington, 1995).

The role of MAP kinase pathways in the development of Drosophila eyes has been studied through the analysis of mutants in which the structure of the individual ommatidia of the compound eye are disturbed (Raabe, 2000). Each ommatidium contains eight photosensitive neurones called R-cells. A mutation in a RTK called Sevenless (Sev) specifically inhibits the development of the R7 cell. Normally Sev binds a protein called Bride of Sevenless (Boss) found on the surface of the R8 cell, signalling the precursor R7 cell to develop into a functional neurone (Krämer et al., 1991). Activated Sev dimerizes, autophosphorylates, and binds an adapter protein, Drk, which contains protein-binding SH2 and SH3 domains. Drk binds a guanine nucleotide exchange factor called Son of sevenless (Sos) (Simon et al., 1993). Positive (Corkscrew, CSW) and negative (Phospholipase C-γ, PLCγ) regulatory proteins interact with Sev through the Daughter of Sevenless (Dos) protein (Perkins et al., 1996; Raabe et al., 1996). Sos in turn binds Ras1 and catalyses the exchange of GDP for GTP, thus activating Ras1. Ras1 turnover is catalysed by the GTPase activating protein Gap1. Activated Ras1 in turn binds and activates the MEK kinase D-Raf, which initiates a MAP kinase pathway resulting in the activation of the Map kinase Rolled (Biggs et al., 1994) and the phosphorylation of transcription factors which initiate the development of the R7 cell. The route of this pathway, involving receptors, adapters, guanine nucleotide exchange factors and small GTP binding proteins is commonly found in the MAP/ERK family. As in the example of the Hog1 pathway, the Sevenless MAP kinase pathway is held together by a scaffold protein called Ksr-1 (Cacace et al., 1999) (Fig. 3). A gain of function Rolled MAP kinase will however, activate not only the Sevenless pathway, but other developmental pathways such as those regulated by the RTK torso (Brunner et al., 1994).

VI. MAP kinases in plants

Numerous genes encoding MAP kinases and other components of MAP kinase signal transduction pathways have been identified in plants. Changes in gene expression or enzyme activity in response to external stimuli or physiological processes have been associated with many of these, and in some cases, changes in phenotype are linked with mutations in these genes. Details of which stimuli are allied with which genes or proteins are summarized in Table 1. The year 1993 saw a number of reports detailing plant MAP kinase genes. Duerr et al. (1993) described a cDNA from Alfalfa (Medicago sativa), MSERK1, encoding a MAP kinase homologue; this cDNA was isolated by using a plant CDC2 gene as a probe to screen a root cDNA library (cdc2 and MAP kinases are members of the same protein kinase subfamily). Southern blot analysis indicated that this gene was representative of a small gene family. The recombinant protein, and fractionated plant protein extracts, cross-reacted with antisera raised against sea-star and rat MAP kinases, and the E. coli-expressed protein could autophosphorylate, and phosphorylate the substrate MBP. Kinase activity was dependent on tyrosine phosphorylation at position 215 and to a lesser extent, threonine phosphorylation at position 213. These positions correspond to the TXY motif found in other MAP kinases. Stafstrom et al. (1993) isolated a pea (Pisum sativum) MAP kinase cDNA, termed D5, using degenerate primers corresponding to amino acid sequences common to all known protein kinases. The deduced amino acid sequence showed some 50% identity to yeast and mammalian MAP kinases and the recombinant protein cross-reacted with antiserum raised against human ERK. The messenger RNA appeared to be present in all tissues at similar levels. A tobacco (Nicotiana tabacum) MAP kinase homologue was isolated from cell suspension cells by PCR amplification of conserved MAP kinase domains; the cDNA NTF3 was expressed in all tissues tested (Wilson et al., 1993). Two more tobacco MAP kinase cDNAs, NTF4 and NTF6 were later characterized (Wilson et al., 1995). Jonak et al. (1993) isolated a Medicago sativa MAP kinase cDNA (MSK7) which showed stem and root specific expression, and at higher levels during S and G2 phases of the cell cycle. Mizoguchi et al. (1993) described a gene family encoding MAP kinases in Arabidopsis thaliana (cDNAs ATMPK1-ATMPK7) and characterized two of these cDNAs (ATMPK1 and ATMPK2) in more detail (Mizoguchi et al., 1994). Transcripts for ATMPK1 and 2 could be found in all tissues but at higher levels in stems and lower levels in seeds. The recombinant proteins encoded by these cDNAs could be phosphorylated, and the protein kinase function activated by, Xenopus MEK. Extracts prepared from tobacco cells 10 min after treatment with 1 µM auxin contained enhanced MBP kinase activity and would also activate the recombinant proteins. These early reports therefore confirmed the presence of multiple MAP kinase genes and proteins in plants, and showed that they could have differential gene expression patterns, and similar enzymatic properties to the yeast and mammalian homologues. The number of MAP kinase genes reported from plants has been increasing steadily; a recent count gave a total of 23 published full-length cDNAs and 45 ESTs. 13 of the 24 Arabidopsis ESTs might represent new MAP kinase genes, but since these are only partial sequences that cannot be aligned, the number is likely to be less. Phylogenetic analysis of the plant MAP kinases allows them to be put into groups, the number of groups depends on which criteria are considered and can vary from 3 to 7 (Ligterink, 2000). These groupings may or may not have a relationship to function; however, functional redundancy between MAP kinases is a real possibility. The majority of plant MAP kinases described so far belong to the ERK/Map kinases grouping, based on the TEY sequence in the activation domain. However ArabidopsisATMPK8 and ATMPK9 (Mizoguchi et al., 1997), the predominantly root-expressed alfalfa TDY1 gene (Schoenbeck et al., 1999) and the wound-induced BWMK1 gene from rice (He et al., 1999) encode a TDY sequence at the activation domain, characteristic of the p38/Hog group. No plant MAP kinases are yet described with the TPY activation domain sequence.

Table 1.  Summary of reported associations of specific plant MAP kinase cascade components with different stimuli. Dark blue shading indicates a reported change in gene expression, enzyme activity or mutant phenotype associated with each kinase. Red shading indicates no change observed Thumbnail image of

The Arabidopsis genomic sequence is now complete (The Arabidopsis Genome Initiative, 2000). A search of the Arabidopsis genome and protein databases at MIPS and NCBI yielded 640 genes encoding protein kinases and of these, there were 21 different genes encoding potential MAP kinase-like enzymes. The genes encoding ATMPK1 through ATMPK7 were amongst these. Genes for ATMPK8 and 9 were not identified however, the encoded sequence T10F20–20 closely resembles ATMPK8, with the addition of a 47 amino acid insertion at the C terminus, and AT3g18040 closely resembles ATMPK9 with a 114 amino acid insertion at the N terminus. 12 of the 21 genes encoded MAP kinases with a TEY motif, and 9 with a TDY motif. No MAP kinases with a TPY or TGY motif were found.

Upstream components of MAP kinase pathways have also been isolated from plants; Shibata et al. (1995) reported a tobacco cDNA encoding a protein kinase, NPK2, similar to mammalian MEK. The ArabidopsisMEK1 gene shows transcriptional regulation during development, transcript levels are higher in de-etiolating seedlings and also after wounding of leaves (Morris et al., 1997). Three further Arabidopsis MEK cDNAs (ATMKK2, ATMKK3 and ATMKK4) have also been identified (Ichimura et al., 1998). NPK1 is a tobacco gene encoding a MEK kinase homologue, similar to yeast BCK1, a MEK kinase in the cell wall integrity pathway (Levin et al., 1994); a truncated form of NPK1 has constitutive kinase activity and can complement yeast bck1 mutant (Banno et al., 1993). Four Arabidopsis homologues of NPK1 have been isolated (ANP1, 2, 3 and 4 ). The gene encoding ANP1 undergoes differential splicing to produce variant cDNAs (Nishihama et al., 1997; Ichimura et al., 1998). An analysis of plant MEK kinases shows 16 distinct and diverse Arabidopsis sequences (including cDNAs, ESTs and genomic sequences). These sequences fall into two broad groups: the Raf family or the MEK kinase family (Jouannic et al., 1999).

Potential upstream MAP kinase pathway receptor-effectors have also been found in plants. In Arabidopsis, there are at least 10 genes encoding proteins with similarity to histidine kinase two-component systems, and at least 14 genes for independent response regulator proteins (D’Agostino & Kieber, 1999). RTKs are not found in plants, but the receptor-like kinases (RLKs) constitute a large family of transmembrane serine/threonine protein kinases with divergent extracellular domains (Stone & Walker, 1995). RLKs are associated with self-incompatibility, brassinosteroid signal transduction, meristem development and defence responses (Torii, 2000). Recently, an Arabidopsis seven-transmembrane receptor that may be involved in cytokinin signalling was identified (Plakidou-Dymock et al., 1998).

The interactions between plant MAP kinase cascade components have been analysed using protein expression in yeast as a molecular tool. The yeast 2-hybrid system showed a positive interaction specifically between Arabidopsis MEK1 and ATMPK4, and also between MEK1 and the MEK kinase ATMEKK1. The yeast mpk1 mutant could be complemented by coexpression of MEK1 and ATMPK4, showing that these proteins can also functionally interact. Similarly, coexpression of MEK1 and ATMEKK1 would complement the yeast pbs2 mutant (Mizoguchi et al., 1998). An Arabidopsis cDNA library was screened by the yeast two-hybrid method using ATMEKK1 as bait and yielded ATMKK2, which is similar to MEK1 in sequence and also ATMPK4, indicating direct contact between the MEK kinase and the MAP kinase levels. In a pairwise yeast two–hybrid interaction analysis, ATMEKK1 interacted (through the catalytic domain) with ATMKK2 as well as MEK1, and ATMPK4 interacted with both MEK1 and ATMKK2. ATMPK4 also interacted directly with the regulatory domain of ATMEKK1. The plant MEKs appear to have a slightly different activation domain amino acid sequence to animal MEKs, and when either of two conserved threonines in this domain were converted to alanine, ATMKK2 was no longer able to complement the yeast pbs2 mutant (Ichimura et al., 1998). Recombinant MEK1 will specifically phosphorylate and activate recombinant ATMPK4 in vitro, but this phosphorylation was found primarily on threonine. Tyrosine phosphorylation is needed for ATMPK4 activation since de-phosphorylation with a protein tyrosine phosphatase will de-activated ATMPK4. Recombinant ATMPK4 was found to autophosphorylate predominantly on tyrosine, although it is not yet known if this tyrosine is that in the TEY activation domain (Huang et al., 2000). This in vitro study suggests that in contrast to animal systems, Arabidopsis MEK1 may not be a dual function protein kinase, and activation of ATMPK4 takes place by MEK1-catalysed threonine phosphorylation and either tyrosine autophosphorylation, or alternatively tyrosine phosphorylation by a second MEK such as ATMKK2 (Ichimura et al., 1998), however, it should be noted that this was an in vitro study using fusion proteins. Together these data suggest that ATMEKK1, ATMKK2/MEK1 and ATMPK4 may constitute a plant MAP kinase cascade, and may even form a signalosome complex.

Not only have MAP kinases been described in plants, but also some of the phosphatases that may de-activate MAP kinases. A tyrosine-specific protein phosphatase encoding cDNA (VH-PTP13) was isolated from the green alga Chlamydamonas; the recombinant phosphatase was capable of de-phosphorylating alfalfa MMK2 in vitro (Haring et al., 1995). However Chlamydamonas MAP kinases have not yet been characterized nor has the substrate specificity of VH-PTP13 for MAP kinases been established. Xu et al. (1998) identified an Arabidopsis gene encoding a tyrosine-specific phosphatase, ATPTP1. Transcript levels for ATPTP1 were enhanced by salt and down-regulated by cold treatment, and recombinant ATPTP1 will de-phosphorylate and de-activate an Arabidopsis MAP kinase in vitro (Luan, 1998). Meskiene et al. (1998) employed a clever screen for MAP kinase pathway de-activating factors by introducing a cDNA library from Medicago into a yeast strain that is pheromone oversensitive, and will not grow in the presence of pheromone. One of the effects of yeast pheromone is to activate the Fus3 MAP kinase pathway. A yeast transformant was screened out that grew in the presence of pheromone; it contained a plant cDNA, MP2C, encoding a putative phosphatase 2C (a serine/threonine specific phosphatase). MP2C was found to act in yeast on the enzyme Ste11 kinase, one step before the MEK kinase, Ste11. An Arabidopsis dual-specificity protein phosphatase (ATDSPTP1) has also been identified. The recombinant protein will de-phosphorylate both MBP and also recombinant ATMPK4 (Gupta et al., 1998).

VII. MAP kinases and cell division

The name MAP kinase implies a relationship with cell division; indeed it has been demonstrated that some plant MAP kinases are transiently expressed and activated during mitosis (Jonak et al., 1993). In a study on tobacco cell cultures, these were starved of phosphate to arrest them in G1, and caused to recommence division by addition of phosphate. After approx. 45 min of phosphate addition, a 45-kDa MBP kinase activity was induced; this activity was immunoprecipitable with antisera raised against the alfalfa MAP kinase MSK7. From the known cross-reactivity of this antiserum with tobacco MAP kinases, the enzyme is thought to be NTF4 (Wilson et al., 1998). Tobacco NTF6 was also shown, by the use of specific antisera, to be activated specifically during late anaphase of mitosis, with transient localization at the phragmoplast (Calderini et al., 1998). Similarly, using cell cultures synchronized with aphidicolin, the alfalfa MAP kinase MMK3 was shown, by the use of MMK3-specific antisera, to be present in dividing cells only throughout the cell cycle, but to be transiently activated only after metaphase. MMK3 was found to be located at the plane of division of the cell (Bögre et al., 1999). An investigation into the expression pattern of the NPK1 gene showed that NPK1 message is found in tissues containing dividing cells, such as root primordia and apical meristems (Nakashima et al., 1998). The NPK1 protein appears to interact with microtubule-based motor proteins (NACK1 and NACK2), transcripts of which accumulate during M phase. Both NPK1 and NACK1 have been localized at the centre of the mitotic spindle and phragmoplast of dividing tobacco cells (Nishihama & Machida, 2000).

VIII. MAP kinases and plant hormones

The observations that extracts from auxin treated tobacco cells would activate MAP kinases (Mizoguchi et al., 1994) were followed up by an elegant investigation into the role of the tobacco MEK kinase, NPK1, in auxin signal transduction (Kovtun et al., 1998). A transient expression system using maize leaf protoplasts was used to monitor the expression of an auxin-inducible promoter reporter gene construct (GH3-sGFP) when cotransformed with the NPK1 gene under the regulation of a constitutive promoter. It was found that NPK1 expression specifically blocked auxin-induced gene expression. A C-terminal deletion of the NPK1 protein was found to be more effective in inhibiting auxin activation than the full protein, confirming a regulatory effect of the C-terminus (Banno et al., 1993). NPK1 transfection into maize protoplasts enhanced the endogenous MAP kinase activity of the cells. De-activation of MAP kinase activity by cotransfection with a MAP kinase-specific phosphatase (MPK1) blocked repression of GH3, indicating that active MAP kinase enzyme is not required for auxin-induced gene expression. Transgenic tobacco plants over-expressing NPK1 were found to have defects in seed development. Surviving plants were however, wildtype in appearance and although high levels of NPK1 transcript were found, the protein could not be detected, suggesting tight regulation of the protein level. These results indicate therefore that auxin signalling is negatively regulated through a MAP kinase cascade (Kovtun et al., 1998). It will be of interest to see if the expression of all auxin-induced genes is negatively regulated by NPK1, and if cytokinin-induced gene expression is similarly affected. Treatment of mature leaf tissue with both auxin and cytokinin together (which stimulates mitotic divisions), results in the accumulation of NPK1 transcripts; however, neither auxin or cytokinin on their own would do so. Aphidicolin, an inhibitor of DNA replication, did not prevent NPK1 accumulation in meristematic tissue (Nakashima et al., 1998). These data suggest that NPK1 plays a role in regulating auxin and cytokinin-driven cell division, perhaps by inhibiting the expression of a subset of genes normally induced by auxin.

How then does this fit with the observations by Mizoguchi et al. (1994)? Tena & Renaudin (1998) have shown that low doses of auxin would not induce MBP kinase activation in the same tobacco cell lines, but weak acids such as butyric acid would do so. High auxin doses were shown to induce MBP kinase activity, probably as a result of cytosolic acidification. These authors suggest that the result of Mizoguchi et al. (1994) may be an artefact of the suspension culture system employed, since MBP kinase activity could also be induced by subculturing the cells, irrespective of the addition of auxin. However Mockaitis & Howell (2000), working with whole plants grown in a hydroponic system, have provided good evidence that low levels of auxin (2 µM) will induce a rapid, transient increase in MBP kinase activity in Arabidopsis roots. Inactive auxin analogues would not do so, nor would acetic acid. The auxin-induced kinase activity showed MBP substrate specificity (as opposed to casein or BSA), and could be immunoprecipitated with antisera directed against mammalian ERK, or antiphosphotyrosine. The auxin resistant mutant axr4 showed a much-reduced auxin-inducible MBP kinase response, although salt-induction of MBP kinase activity remained the same in the mutant. Auxin-induced gene expression was investigated using a transgenic reporter line expressing GUS under the control of the auxin activated IAA4/3 ci promoter. 20 µM IAA resulted in GUS expression in the roots of these plants, however, preincubation in the MEK activation inhibitors PD 098059 (20 µM) or U0126 (10 µM) inhibited auxin induction of GUS activity. Surprisingly, both PD 098059 and U0126 enhanced auxin-induced MBP kinase activity. These results suggest that the auxin-induced MAP kinase detected by immunoprecipitation and in gel kinase activity is not responsible for auxin-inducible gene expression, and lies in a MAP kinase pathway whose MEK is not blocked by the inhibitors. However this first pathway may be negatively regulated by a second MAP kinase pathway, which is blocked by the inhibitors. This second pathway may also more directly regulate auxin-inducible gene activation.

An investigation into abscisic acid (ABA) signalling pathways in barley aleurone layers has shown that MAP kinase pathways are probably involved here also (Knetsch et al., 1996). Western blotting of aleurone protoplast extracts using antisera specific for rabbit ERK1 gave three cross-reacting bands in the molecular weight range of potential MAP kinase enzymes. The amount of cross reacting protein did not change after incubation of the protoplasts with 10 µM ABA, but immunoprecipitable MBP kinase activity transiently increased fivefold after 3 min, before returning to basal levels after 5 min. Anti-phosphotyrosine antisera precipitated MBP kinase activity from ABA treated cells with the same kinetics as anti-ERK1 antisera. The tyrosine phosphatase inhibitor PAO inhibited ABA-stimulated MBP kinase activity, and also expression of the ABA-inducible gene, Rab16. A MAP kinase cDNA, Asmap1 that is transcriptionally negatively regulated by gibberellic acid (an antagonist of ABA) has been isolated from oat (Avena sativa) aleurone cells (Huttley & Phillips, 1995), but it is not known what affect ABA has on Asmap1 expression or activity, or if this might represent the oat homologue of the MBP kinase activity found in barley.

An important target of ABA action is the regulation of stomatal activity. An analysis of 515 expressed sequence tags (EST) sequences derived from a Brassica campestris guard cell library yielded two different MAP kinases, with amino acid sequence similarities to ATMPK3 and ATMPK4 (Kwak et al., 1997) and ABA-inducible MBP kinase activity has been found in guard cell protoplasts from Vicia faba (Mori & Muto, 1997). Burnett et al. (2000), working with epidermal peels from pea have shown that treatment with 100 µM MEK inhibitor 098059 inhibited ABA-inhibition of closed stomata and ABA-stimulation of open stomata. High levels of PD 098059 also inhibited the expression of the ABA-induced gene encoding dehydrin. 10–100 µM ABA treatment results in the rapid and transient activation of MBP kinase activity in epidermal peels. Treatment with protein tyrosine phosphatase gave a slight decrease in activity, as did treatment with 100 µM MEK inhibitor PD 098059. Using the same source of antisera as Knetsch et al. (1996), western blotting revealed three cross-reacting proteins of approx. 45, 49 and 51 kDa in epidermal and mesophyll cells, and proteins of 51, 45 and 43 kDa in guard cells. It is not clear which of these might be responsible for the ABA-inducible MBP kinase activity. The results presented by both Knetsch et al. (1996) and Burnett et al. (2000) throw an interesting light onto the involvement of MAP kinase pathways in ABA signal transduction, however, since these findings depend on the specificity of antisera raised against mammalian proteins and on relatively high levels of MEK inhibitor, further investigations are required to confirm the issue.

A key component of the ABA signal transduction pathway in Arabidopsis, the ABI1 gene, encodes a protein phosphatase 2C (Leung et al., 1994; Meyer et al., 1994) and this has led to speculation that ABI1 might de-phosphorylate a MAP kinase pathway component in the ABA signal transduction pathway (Heimovaara-Dijkstra et al., 2000). Provocatively, the ABI1 gene lies directly next to the MEK1 gene (Morris et al., 1997). The alfalfa PP2C MPC2 gene when expressed in yeast will inactivate the Ste11 MEK kinase (Meskiene et al., 1998). Further, mutations in genes encoding PP2Cs in fission yeast results in osmotically sensitive cells. A genetic suppresser of this condition has been identified as the MEK encoding gene Wis1 (the Schizosaccharomyces pombe homologue of Pbs2), suggesting that PP2Cs might be negative regulators of Wis1 and/or other components of the pathway (Shiozaki & Russell, 1995). The argument is slightly diluted by the later finding that PP2C actually acts downstream of the MAP kinase pathway in yeast (Gaits et al., 1997), but remains an attractive hypothesis.

Perhaps the most progress has been made in elucidating the signal transduction pathway for ethylene; there have been many excellent recent reviews on the subject (Kieber, 1997; Stepanova & Ecker, 2000). The ETR1 ethylene receptor family in Arabidopsis is known to comprise of 5 membrane proteins. These proteins are similar to two-component histidine protein kinase regulators, such as that commencing the Hog1 osmosensing pathway in yeast (Chang et al., 1993; Hua & Meyerowitz, 1998). The ethylene receptor is thought to associate with CTR1 (Clark et al., 1998). CTR1 is similar to the MEK kinase Raf, and genetic lesions in CTR1 result in constitutive ethylene signalling (Kieber et al., 1993). Since at least two components of the ethylene signalling pathway belong to component of MAP kinase cascades in other organisms, it is very possible that a down-stream MEK and MAP kinase will be involved in negatively regulating ethylene signal transduction, but currently hard biochemical evidence for this is lacking. A recent report details an Arabidopsis MBP kinase which is enhanced by ethylene and immunoprecipitable by antisera directed against ERK1 or phosphotyrosine. Ethylene-induced MBP kinase activity was found to be reduced in the etr1 ethylene insensitive mutant and enhanced in the ctr1 constitutive ethylene response mutant. This MBP activity is however, unlikely to be due to a downstream MAP kinase in a direct ethylene signal transduction pathway leading from ETR1 through CTR1, because in the absence of CTR1, a decrease in downstream MAP kinase activity would be expected (Novikova et al., 2000).

Overexpression of the ArabidopsisCKI1 gene leads to cytokinin independence in tissue culture cells. CKI1 encodes a histidine kinase sensor-like protein (Kakimoto, 1996). The ArabidopsisCRE1 gene also encodes a histidine kinase; when mutated this gene leads to reduced cytokinin responses. Expression of CRE1 in yeast (lacking the endogenous histidine kinase SLN1) gave a cytokinin-dependent growth phenotype, indicating that CRE1 is a cytokinin receptor (Inoue et al., 2001). Genes encoding response regulator domains were also found to be rapidly up-regulated by cytokinin (Brandstatter & Kieber, 1998; D’Agostino & Kieber, 1999). These finding raise the possibility that cytokinin signalling, like ethylene signalling, might proceed through a MAP kinase pathway. It should of course be borne in mind that histidine protein kinases might initiate signal transduction pathways in plants other than MAP kinase pathways.

IX. MAP kinase and abiotic stress

Abiotic stresses such as wind, cold, UV and wounding elicit many changes in gene expression in plants, and MAP kinase pathways have been implicated in the signal transduction of these environmental insults. Arabidopsis plants treated with touch, low temperature or salinity stress showed transient increases in transcript levels within a few minutes of stimulation for the genes encoding the MEK kinase ATMEKK1, the MAP kinase ATMPK3 and also ATPK1, a p90 ribosomal S6 kinase (Mizoguchi et al., 1996). S6 kinases are known to be activated by MAP kinases in mammalian systems (Sturgill et al., 1988). By contrast, salt rather than cold or touch treatment increased ATMPK1 mRNA levels. Specific antisera were raised against Arabidopsis ATMPK4 and ATMPK6 peptides and used to investigate the activation of these enzymes in response to stress. Low temperature, low humidity, salt stress, sorbitol, touch and wounding all induced a rapid, transient activation of both enzymes, with slightly different kinetics. Gene expression and protein levels for both proteins did not significantly change after abiotic stress. In contrast, no activation of ATMPK3 could be detected after stress treatment (Ichimura et al., 2000b).

Working with tobacco, Seo et al. (1995) cloned a cDNA encoding a 43-kDa wound inducible MAP kinase, termed WIPK (wound-induced protein kinase). Cutting the stem resulted in a transient (one hour) systemic increase in WIPK transcript as early as one minute after wounding. A 46-kDa MBP kinase activity from leaf extracts was also shown to increase transiently after wounding. Transgenic tobacco plants transformed with WIPK under the control of the CaMV 35S promoter showed constitutive expression of WIPK and constitutive low levels of MBP kinase activity in leaf extracts. However there was also the loss of wound induction of the endogenous message, probably through cosuppression. These plants also lost wound induction of typically wound-induced messages such as PI-II. Curiously, genes normally associated with pathogen attack, PR-1 and PR-2 were induced instead. Plants normally respond to wounding by enhancing levels of jasmonic acid which in turn is involved in the induction of wound induced genes such as PI-II (Farmer & Ryan, 1992); the cosuppressed transgenics produced much lower levels of jasmonic acid after wounding than did the controls. However, salicylic acid, normally produced in response to pathogen attack and thought to be responsible for the induction of pathogenesis related gene expression such as PR-1, actually increased in the wounded leaves, in contrast to the controls. The authors speculate that wound activated WIPK may regulate jasmonic acid synthesis by the phosphorylation of cytoplasmic phospholipase A2 (as in animal cells, Lin et al., 1993), and that jasmonic acid suppresses salicylic acid synthesis. Usami et al. (1995) presented similar data; they described a 46-kDa MBP kinase transiently activated by cutting leaves from several plant species. Kinase activation was independent of protein synthesis, and both serine/threonine and tyrosine phosphatases inhibited activation of the MBP kinase activity, consistent with the activity being attributable to a MAP kinase.

Jasmonic acid is synthesized from linoleic acid; the enzyme ω-3 fatty acid desaturase, encoded by the FAD7 gene, converts linoleic acid to the intermediary compound linolenic acid and FAD7 gene expression is enhanced by wounding. Transgenic tobacco plants bearing an ArabidopsisFAD7 promoter-Gus fusion construct were analysed for GUS expression after wounding; FAD7 gene expression was inhibited in stems and roots (but not leaves) by the protein kinase inhibitor staurosporin and also the phosphatase inhibitor calyculin A. FAD7-Gus expression was monitored in a background of transgenic tobacco over-expressing or cosuppressing WIPK, however, no difference was seen in FAD7 gene expression after wounding in leaves and roots, and only minor differences in stems; it appears that wound-induced FAD7 gene expression is not governed by WIPK (Kodama et al., 2000).

In an interesting twist, Zhang & Klessig, (1998) presented evidence that wounding would activate the 48 kDa tobacco SIPK MAP kinase, also associated with biotic stress and salicylic acid (see later) but not WIPK. Since at that time activity of the WIPK protein had not been analysed with specific antisera, it had not been rigorously proven that the MBP kinase activity that accompanies WIPK message production upon wounding is indeed due to WIPK protein. Using antisera directed against the specific N-termini of the WIPK and SIPK proteins, only active SIPK could be immunoprecipitated from wounded leaves, but not WIPK, although WIPK, but not SIPK, message levels increased after wounding, as reported by Seo et al. (1995). Seo et al. (1999) prepared WIPK-specific antisera using the same N-terminal peptide as Zhang & Klessig (1998). In contrast to Zhang & Klessig, they found MBP kinase activity in immunoprecipitates from wounded tobacco leaves when using this antiserum. Seo et al. (1999) suggest in their paper that the MBP in-gel kinase assay protocol of Zhang and Klessig may not have been optimal, but the 46 kDa MBP kinase activity reported previously (Seo et al., 1995) may be due to a combination of both WIPK and SIPK. They also identified transgenic tobacco plants that over-expressed active WIPK, by looking for those plants that had constitutive PI-II expression. These plants also had constitutive MBP kinase activity and elevated jasmonic acid levels. These lines however, lost these phenotypic attributes when mature (5-month-old). More recent data from these workers has shown that WIPK is activated in both the cytosol and the nuclei of cells from wounded leaves, but the nuclear levels decreased transiently whereas the cytosolic levels remained constant. They suggest that wound-induced transcription of WIPK is required to maintain the steady state levels of WIPK in cytosol and nucleus after nuclear WIPK degradation and recruitment of cytosolic WIPK (Seo & Ohashi, 2000).

A classic system for the study of wound stresses in plants is the tomato, in which wounding induces the systemic synthesis of defence-related genes. An 18-amino acid polypeptide called systemin appears to be a primary wound signal and will induce jasmonic acid production (Pearce et al., 1991; McGurl et al., 1992). Stratmann & Ryan (1997) have shown that wounding in tomato systemically induces a 48-kDa tyrosine-phosphorylated MBP kinase activity, as would application of 25 nM systemin to the plants. However neither jasmonic acid, salicylic acid nor ABA would induce above-background levels of MBP kinase activity. Tomato plants mutated in a gene of the jasmonic acid biosynthetic pathway, and thus unable to accumulate jasmonic acid after wounding, would still respond to wounding or systemin application by the activation of MBP kinase. These results suggest that MBP kinase activation lies biochemically between systemin (and other primary wound signals such as polygalacturonic acid) and jasmonic acid biosynthesis. Although the MBP kinase activity reported here has not been formally proven to be a MAP kinase, given the similarities between tobacco and tomato, it would be surprising if this were not the case.

Jonak et al. (1996) investigated MAP kinase activity in stressed alfalfa by means of MBP kinase assays of immunoprecipitations using antisera raised against kinase-specific peptides. MKK4 kinase activity is activated rapidly and transiently by cold stress and drought, but not by heat or salt; MKK4 gene expression is also activated by cold and drought, however, MKK4 protein levels remained constant. MKK2 and MKK 3 were not activated by these conditions. In a thorough investigation into mechanical stimulation and wound-induced Map kinase activity by Bögre et al. (1996, 1997), it was shown that for MMK4-specific immunoprecipitations from wounded alfalfa leaves analysed for MBP kinase activity, that mechanical stimulation or wounding rapidly induced a transient MBP kinase activity. Although MMK4 protein levels remained constant after wounding, transcript levels increased after 30 min. Incubation of leaves with α-amanitin or cycloheximide did not inhibit wound induced MKK4 activation, however, de-activation of MKK4 was prevented by both inhibitors, suggesting that activation is independent of transcription or translation but de-activation depends on newly transcribed and translated factors, probably a protein phosphatase. Factors thought to be involved in transducing the wound response in plants, such as ABA and jasmonic acid, were not found to activate MKK4. Wound induction of MKK4 was found to be refractory to re-stimulation for some 35 min after wounding (i.e. when the initial induction of kinase activity had decayed). By contrast to the results found with MKK4 specific antisera, antisera specific for the alfalfa MAP kinases MKK2 and MKK3 did not precipitate wound-or mechanically activated MAP kinase. Thus in alfalfa, MKK4 activation appears to be associated with many different forms of stress, and Bögre et al. (1997) suggest renaming MKK4 to SAM kinase (stress-activated MAP kinase). However Zhang & Klessig (1998) raised the possibility that antisera directed against the C-terminus of MKK4 may also recognize the SIPK-like MKK1 MAP kinase, therefore MKK1 and/or MKK4 might be involved in wound-responsive MAP kinase signalling in alfalfa. Gene expression for the protein phosphatase 2C MP2C (Meskiene et al., 1998) was found to be transiently induced after wounding. Extract from wounded leaves were incubated with active recombinant MP2C, the phosphatase inactivated SAM kinase activity in vitro, raising the possibility that MP2C is also a negative regulator of SAM kinase in vivo. Interestingly, the in vitro phosphatase activity of recombinant MP2C is inhibited by 10–500 µM linolenic acid (the jasmonic acid precursor) but not by saturated fatty acids. This suggests that the wound-induced activation of MAP kinases could be reinforced by inhibition of the corresponding de-activating phosphatase (Baudouin et al., 1999).

Another environmental insult that appears to activate MAP kinases in plants is ozone (Samuel et al., 2000). Within 5 min of exposure to ozone, a MBP kinase activity of about 46 kDa molecular weight is induced in tobacco leaves and cells. Anti-SIPK, but not anti-WIPK antisera immunoprecipitated a 46-kDa MBP kinase after ozone treatment. It is probable that reactive oxygen species generated by ozone are responsible for stimulating the activation of SIPK, since hydrogen peroxide would elicit the same response as ozone but pretreatment with the free radical scavenger mercaptopropionyl glycine would inhibit the ozone response. La3+ ions inhibited activation of SIPK, suggesting the involvement of calcium in the ozone response.

The water status of the cell and its environment is another cause of stress to plants, and here MAP kinases are also implicated in signal transduction. Hypo-osmotic shock results in the activation of 50, 75 and 80 kDa MBP kinase activities from tobacco cells. The 50 kDa activity is immunoprecipitable with antiphosphotyrosine monoclonal antibody, and de-activation of this kinase was inhibited by the protein phosphatase inhibitor calyculin A, consistent with the 50 kDa activity being a MAP kinase. Hypo-osmotic shock results in calcium influx from the medium across the plasma membrane. Chelation of extracellular calcium prevented the increase in intracellular calcium after hypo-osmotic shock, and also prevented the activation of MBP kinase activity. Addition of calcium to the shocked cells stimulated MBP kinase activity (Takahashi et al., 1997). The tobacco MAP kinase NTF4 is transcriptionally upregulated during pollen maturation (during desiccation) and active NTF4 enzyme can be specifically immunoprecipitated out of maturing pollen. Conversely, when dry, mature pollen rehydrates, there is a fast (maximum activity at five minutes) and transient increase in immunoprecipitable NTF4 MBP kinase activity (Wilson et al., 1997). MBP kinase activity is also induced by hypo-osmotic stress of tobacco cell suspensions; some of this activity is immunoprecipitable by antisera directed against NTF4 (Cazaléet al., 1999). Salt stress induced MBP kinase activity in alfalfa; sodium chloride at between 125 and 750 mM induced a rapid, transient, and specifically immunoprecipitable activity of MSK7 (renamed SIMK, salt stress-inducible MAP kinase), and salt at between 750 and 1000 mM induced an unknown MBP kinase of 38 kDa molecular mass. Sorbitol at 1000 mM would induce SIMK, suggesting that sodium ion toxicity might be responsible for preventing SIMK activation at above 750 mM sodium chloride. Immunofluorescence microscopy revealed that intracellular translocation of SIMK did not occur after salt treatment, SIMK is constitutively localized in the nucleus (Munnik et al., 1999). The yeast two-hybrid system was used to isolate a MEK, SIMKK, that interacts with SIMK. SIMKK will specifically phosphorylate and activate SIMK in vivo, as shown by transient transformation of parsley protoplasts with vectors encoding SIMKK and epitope-tagged SIMK, immunoprecipitation and an in vitro MBP kinase assay. Salt stressing the protoplasts led to a considerable enhancement of MBP kinase activity (Kiegerl et al., 2000). Sodium chloride (250 mM) and sorbitol (900 mM) will also activate a 48-kDa MBP kinase from tobacco; this kinase is inactivated by tyrosine and serine/threonine protein phosphatases, and is immunoprecipitable with antisera raised against SIPK. A second salt-induced MBP kinase activity was not inhibited by protein tyrosine phosphatase; the kinase was purified, sequenced and shown to be similar to the Arabidopsis serine/threonine kinase ASK1, illustrating the importance of not categorizing all MBP kinase activity as MAP kinase activity (Mikolajczyk et al., 2000). Since the yeast Hog1 osmoregulatory MAP kinase pathway has as one of the osmosensing components a two-component histidine protein kinase, this has prompted the search for similar osmosensors in plants. Urao et al. (1999) report an Arabidopsis cDNA, ATHK1, isolated from droughted plants, that encodes such a histidine kinase. This cDNA is able to functionally complement a sln1Δ, sho1Δ yeast mutant, allowing growth on 0.9 M sodium chloride, and transducing the sodium chloride concentration-dependent tyrosine phosphorylation of Hog1. The Arabidopsis gene is predominantly expressed in roots, but is upregulated in leaves by 250 mM salt, or cold treatment. Distilled water also induces ATHK1 gene expression; the authors reason that ATHK1 is transcriptionally regulated by changes in external osmolarity, and encodes a potential plant osmosensor.

X. MAP kinase and biotic stress

Plants respond to potential pathogens by recognizing their presence and activating defence mechanisms, including transcriptional activation of defence-related genes (Hammond-Kosack & Jones, 1996). Plants may sense a biotic threat through the recognition of elicitors derived from the pathogen, or its action on plant cell walls. Suzuki & Shinsi (1995) studied the effect of an elicitor (PiE) from cell walls of Phytophthora infestans. They found that PiE would rapidly and transiently activate a tyrosine phosphorylated, cycloheximide-independent, 47 kDa MBP serine/threonine kinase (consistent with the kinase activity being due to a MAP kinase) in tobacco suspension cells. Kinase activity and tyrosine phosphorylation was inhibited by simultaneous addition of 1 µM staurosporine (a general protein kinase inhibitor) and also by addition of 0.5 mM Gd3+, a plasma membrane calcium channel blocker. Hypersensitive cell death in tobacco cells in response to elicitors has also been investigated; cell death can be induced by the protein elicitor TvX (xylanase from Trichoderma viride), but not by PiE. TvX induced a p47 MBP kinase activity in the cells, as did PiE, but compared to the effects of PiE, the MBP kinase activity induced by TvX was delayed and prolonged over about 6 h. Antisera directed against phosphotyrosine or mammalian ERK would immunoprecipitate TvX-induced p47 MBP kinase activity. Staurosporine and also Gd3+ prevented both the activation of p47 and hypersensitive cell death by TvX (Suzuki et al., 1999).

Salicylic acid (SA) is an endogenous signal for the propagation of plant defence systems against pathogens (Malamy et al., 1990). Zhang & Klessig, (1997, 1998) in a biochemical tour de force described how SA induced a rapid and transient 48 kDa MBP kinase activity in tobacco, which they called SIPK (SA-induced protein kinase). Western blotting of protein extracts from SA-treated cells with phosphotyrosine monoclonal antibody showed an increase in a 48-kDa phosphotyrosine-containing polypeptide with the same kinetics as the induction of MBP kinase activity. Immunoprecipitation of SA treated cell extracts with the antiphosphotyrosine antibody and subsequent in-gel MBP kinase assay also showed the same activation kinetics. The MBP kinase was purified to homogeneity, amino acid sequencing carried out on internal tryptic peptides, a cDNA fragment obtained by RT-PCR and the full cDNA (which encoded a MAP kinase) by screening of a tobacco cDNA library. Further work showed that SIPK could be activated not just by SA, but also by various fungal elicitors of different natures; a crude cell wall carbohydrate elicitor, and two different purified protein elicitors; parasiticein and cryptogein (Zhang et al., 1998a). Additionally, 44 and 40 kDa MBP kinases were also activated, a few hours later than SIPK, by the protein elicitors. The 44 kDa MBP kinase was later identified as WIPK (Zhang et al., 2000). MBP kinase assays were carried out on immunoprecipitations from treated tobacco cells using SIPK-specific antisera. Induction of SIPK activity was found to parallel or slightly precede other events associated with cellular defence against pathogens, such as PAL gene expression, suggesting a role in their induction. SIPK protein levels and SIPK transcript levels remained unchanged during induction by elicitor or SA, however, activation was associated with SIPK tyrosine phosphorylation. Protein kinase inhibitors prevented the activation of SIPK and also the induction of PAL gene expression by fungal elicitors (Zhang et al., 1998a). Lebrun-Garcia et al. (1998) also showed that cryptogein would induce in tobacco cells the rapid and transient activation of two MBP kinase activities of 50 and 46 kDa in size. Both activities could be immunoprecipitated with antiphosphotyrosine antisera; they probably correspond to SIPK and WIPK. Interestingly, both La3+ ions and EGTA, which will inhibit Ca2+ influx, inhibited cryptogein induction of MBP kinase activity.

Activation of SIPK by cryptogein and parasiticein was found to be post-translational; by contrast WIPK was also regulated by transcription and translation, since only WIPK activation was inhibited by actinomycin D or cycloheximide. As well as activating SIPK and WIPK in tobacco cells, cryptogein and parasiticein also induced 100% cell death within 24 h, a hypersensitive response-like riposte. Cell death could be inhibited by the broad-spectrum protein kinase inhibitors K-252a and staurosporine, the kinetics of inhibition of cell death correlated with that of WIPK gene expression. A carbohydrate elicitor from Phytophthora parasitica in contrast to the protein inhibitors, will not induce cell death. This elicitor induced SIPK enzyme activity, WIPK message accumulation and WIPK protein accumulation, but active WIPK activity was not immunoprecipitable, strongly suggesting that elicitor induced cell death is associated with WIPK enzyme activity (Zhang et al., 2000).

A 42-kDa glycoprotein elicitor from Phytophthora sojae has been characterized in some detail; a 13 amino acid peptide (Pep13) derived from this protein is sufficient to elicit a defence response (including an oxidative burst) in parsley cells by binding to a plasmamembrane receptor with subsequent rapid ion fluxes across the cell membrane (Nürnberger et al., 1994). The Pep13 elicitor will rapidly and transiently activate MBP kinase activity in parsley cells. This activity is specifically immunoprecipitable with antisera raised against the alfalfa MMK4 MAP kinase (SAM kinase). A cDNA (ERM kinase– elicitor-responsive MAP kinase) corresponding to the parsley homologue of MMK4 was isolated, transcript levels of this gene increased after Pep13 treatment of cells. Immunolocalization of ERM kinase with anti-MMK4 antisera showed rapid translocation of ERM kinase into the nucleus after elicitor treatment of cells. The ion channel blocker A9C prevents Pep13-stimulated ion fluxes, and associated defence responses; AC9 also prevented ERM kinase activation. By contrast, amphotericin B mimics elicitor-induced ion fluxes, and will induce defence responses, and also ERM kinase activation. diphenyleneiodonium (DPI), which inhibits NADPH oxidase, an enzyme thought to be involved in the synthesis of reactive oxygen species (ROS), did not inhibit ERM kinase activation, suggesting that MAP kinase activation is either independent, or upstream of the oxidative burst (Ligterink et al., 1997).

MAP kinases are not only implicated in defence mechanisms resulting from fungal infection; both SIPK and WIPK are rapidly activated by infection with tobacco mosaic virus (TMV) in TMV-resistant tobacco plants, cultivar Xanthi (NN) (Zhang et al., 1998b). Resistance in this cultivar is determined by the N locus and inhibited by high temperature; shifting infected plants to a lower temperature will swiftly induce plant defence mechanisms in a synchronized manner. In a similar manner as for wound-stressed plants (Seo et al., 1995), WIPK messenger RNA accumulated in infected plants after the move to lower temperatures. The increase in WIPK kinase activity and WIPK message levels was transient, returning to background levels after 24 h. But in contrast to wounded plants, there was a sustained increased in WIPK protein levels over at least 72 h. SIPK protein and messenger levels remains unchanged by infection. TMV infection resulted in a systemic induction of WIPK mRNA. Treatment of leaf extracts with serine/threonine or tyrosine specific protein phosphatases abolished the TMV infection-induced kinase activity of WIPK and SIPK. Tobacco plants, cultivar Xanthi (nn), which do not possess resistance to TMV did not show an increase in WIPK kinase or WIPK message after TMV infection, thus WIPK activation is dependent on the N resistance gene. WIPK activation was also shown not to be dependent on salicylic acid, since transgenic tobacco expressing NahG, which encodes a salicylic acid metabolizing enzyme, would respond to infection by WIPK activation and message accumulation. SIPKK, an upstream MEK, which interacts with SIPK (but not WIPK), has been identified using the yeast two-hybrid system. SIPKK shows transcriptional activation after TMV infection or wounding (Liu et al., 2000). The Arabidopsis homologues of WIPK and SIPK (based on sequence comparisons) appear to be ATMPK3 and ATMPK6, respectively. Bacterial, fungal and plant-derived defence response elicitors induced 45 and 49 kDa MBP kinase activity in Arabidopsis suspension culture cells and leaf tissues. The 49 kDa activity is immunoprecipitable with antisera cross-reacting to ATMPK6, however, no activity was found when ATMPK3 specific antisera was used (Nühse et al., 2000).

SA is not the only player in mediating resistance to pathogens in plants, jasmonic acid, as well as its role in promoting wound-induced gene expression, will mediate resistance to some pathogens, as will ethylene. Indeed both of these signalling molecules may act in the same signalling pathway (Penninckx et al., 1998). ACC is a precursor to ethylene, and will be converted to ethylene by ACC oxidase. Using specific antisera, Kumar & Klessig (2000) showed that neither SIPK nor WIPK are activated by jasmonic acid or ethylene, however, both compounds induced a 48-kDa MBP kinase activity. Nitric oxide (NO), a compound normally associated with mammalian signalling, has recently also found to be involved in plant defence responses, and will increase SA levels (Delledonne et al., 1998). Infiltration of tobacco leaves with nitric oxide synthase and NO-producing substrates resulted in the specific, rapid and transient activation of SIPK. WIPK however, was not activated by NO. SIPK activation by NO could be suppressed in a NahG transgenic background, indicating that NO has its effect on SIPK activation through SA (Kumar & Klessig, 2000). Clarke et al. (2000) also found NO to induce a MBP kinase in Arabidopsis, in connection with a study on pathogen stimulated NO-induced cell death.

The recessive Arabidopsisedr1 mutant confers resistance to Pseudomonas syringae and to Erysiphe cichoracearum (powdery mildew). Resistance is not through constitutive defence gene expression, the mutant has a wild-type appearance, however, induced defence responses occur faster than in the wildtype. The mutant phenotype is suppressed in backgrounds with reduced SA production or perception, but not in ethylene or jasmonate mutant backgrounds, although edr1 appears to be more sensitive to ethylene during senescence, where SA is also thought to play a role. The EDR1 gene encodes a CTR1-like MEK kinase, whose function seems to be the negative regulation of SA-inducible defence responses (Frye et al., 2001). By contrast to EDR1, inactivation of ATMPK4 results in constitutive systemic acquired resistance (Petersen et al., 2000). A transposon insertion in the ATMPK4 gene generated the loss-of function mpk4 mutant which is dwarfed, has increased resistance to pathogens and shows constitutive expression of many genes associated with plant defence. SA levels are much higher in mpk4 than the wildtype, and the phenotype (including defence gene expression) can be suppressed by NahG, indicating that ATMPK4 is a negative regulator of SA accumulation. Jasmonate-induced genes are not expressed in mpk4 after jasmonate treatment even in a NahG background, showing a positive role for ATMPK4 in regulating the plant response to jasmonic acid.

Bacterial plant pathogens also produce elicitors that can induce defence responses in plants; the protein harpin from Pseudomonas syringae is such an elicitor (He et al., 1993). Infiltration of harpin at 40 µg per ml into tobacco leaves was found to activate a 49-kDa MBP kinase activity (Ádám et al., 1997). Similarly, harpin treatment at 1 µg per ml will induce rapid, transient MBP kinase activity of 39 and 44 kDa in Arabidopsis suspension culture cells (Desikan et al., 1999a). This activity was abolished by treatment of cell extracts with protein tyrosine phosphatase; enhanced MBP kinase activity after harpin treatment could be immunoprecipitated with antityrosine kinase antisera. Harpin activation of the defence related gene PAL could be abolished, and a minor decrease in MBP kinase activity was seen after pretreatment of cells with high levels (100 µM) of the MEK inhibitor PD 098059; however, 10 µM levels of PD 098059 had no effect.

Harpin will induce an oxidative burst in plant cells, resulting in the production of H2O2 and reactive oxygen species (Desikan et al., 1996); it is possible that the MBP kinase-inducing effects of harpin are attributable to this; Desikan et al. (1999b) showed that H2O2 would induce a rapid transient 44 kDa MBP kinase activity in Arabidopsis suspension culture cells. An Arabidopsis line harbouring a gst1 promoter-luciferase marker was used to investigate the plant response to avirulent Pseudomonas syringae strains, or infiltration with glucose and glucose oxidase, which results in sublethal accumulation of H2O2 (Grant et al., 2000). Both treatments resulted in luciferase expression, independently of ethylene, SA or jasmonic acid. However coinoculation of 100 µM PD 098059 inhibited luciferase activity by 46%, and the phosphatase inhibitor cantharidin at 5 µM induced luciferase activity. Infiltration of leaves with glucose/glucose oxidase rapidly and transiently induced 46 and 48 kDa MBP kinase activity.

The effect of H2O2 on MAP kinase activation was also investigated by Kovtun et al. (2000). Using Arabidopsis protoplasts, these workers found that H2O2 activated two MBP kinases at around 44 and 42 kDa molecular mass. H2O2 also activated stress responsive promoters such as the oxidative stress responsive GST6 and the heat shock responsive HSP18.2, but not the ABA-responsive promoter RD29A. Since H2O2 can regulate cell cycle progression in nonstressed cells (Fuchs et al., 1998), and the tobacco MAP kinase NPK1 is thought to be involved in cell cycle regulation (Nakashima et al., 1998), these workers reasoned that the NPK1 class of MEK kinases might mediate oxidative responses in plant cells. NPK1 is closely related to three Arabidopsis MEK kinases, ANP1, ANP2 and ANP3 (Nishihama et al., 1997). Truncated versions of NPK1/ANP are constitutively active since the C terminal contains a regulatory domain. Transient expression in Arabidopsis protoplasts of ANP driven by the CaMV 35S promoter and with a double HA epitope tag was used as a tool to investigate the function of ANP. All three constitutively active ANP MEK kinases would activate two MBP kinase activities of about the same molecular mass as those activated by H2O2. Constitutively active ANP1 would also activate GST6 and HSP18.2 promoters in the absence of H2O2; this depends on active ANP1 function since a kinase-dead version of ANP1 would not activate these promoters. Co-transfection of constitutively active ANP with individual epitope-tagged Arabidopsis MAP kinases followed by immunoprecipitation and in-gel MBP kinase assay allowed the identification of ATMPK3 and ATMPK6 as substrates for ANP1, ANP2 and ANP3, but not ATMPK2, ATMPK4, ATMPK5 or ATMPK7. Active CTR1 would not activate ATMPK3 or ATMPK6. Epitope-tagged ATMPK3 was introduced into Arabidopsis protoplasts under different conditions and the HA immunoprecipitates assayed for MBP kinase activity. ATMPK3 was activated by H2O2 but not by auxin. The effects of ANP on auxin induced genes was also studied. The GH3 promoter is strongly induced in Arabidopsis protoplasts by auxin; constitutively active ANP1, ANP2 and ANP3 suppressed GH3 activation by auxin, in a similar manner to the suppression of auxin-regulated promoters in tobacco by NPK1 (Kovtun et al., 1998). Similarly, it was found that H2O2 would also repress GH3 promoter activity. Transgenic tobacco plants expressing constitutively active NPK1 were generated; these plants were phenotypically normal but showed enhanced cold, heat and salt tolerance. Further analysis of these plants is required to see which stress-tolerance genes are activated and if all normal auxin-induced responses are down-regulated.

Disease resistance in plants may be nonspecific, as illustrated by the induction of defence mechanisms by fungal and bacterial elicitors, or may be race-specific. Here, pathogens that express a given avirulence gene (Avr) can be recognized by plants expressing the corresponding resistance gene (R). R gene products are thought to act as receptors for the Avr gene products. A well characterized Avr/R gene pair are Avr9 from Cladosporiumfulvum, which encodes a secreted 28 amino acid peptide (Van den Ackerveken et al., 1992), and the tomato Cf-9 which encodes a membrane-anchored extracytoplasmic glycoprotein containing 27 leucine-rich repeats (Jones et al., 1994). The Avr9/Cf-9 interaction is characterized by an oxidative burst leading to synthesis of reactive oxygen species. Transgenic tobacco plants or suspension cultures expressing Cf-9 were treated with Avr9 peptide; this resulted in a rapid and transient six-fold increase of tyrosine phosphorylated MBP kinase activity of molecular mass 46 and 48 kDa (Romeis et al., 1999). MBP kinase activation and ROS synthesis was however, inhibited when Avr9 treatment was combined with the general protein kinase inhibitor staurosporine, and also with La3+, or W7, an antagonist of calmodulin. DPI, which prevents the oxidative burst, did not inhibit MBP kinase activation, but did inhibit ROS production. PD 098059 (albeit at the very high concentration of 250 µM) inhibited MBP kinase activation but not ROS production. Independent activators of ROS (cantharidin and amphotericin B) did not induce MBP kinase activity; together, these data suggests that MBP kinase activation and ROS production may be independent of each other. 46 and 48 kDa Avr9-induced MBP kinase activity was immunoprecipitable with antisera specific for WIPK and SIPK, respectively. WIPK mRNA levels were also enhanced by Avr9 stimulation, in a similar manner to the accumulation of transcript in response to nonspecific elicitors or wounding. Thus race-specific fungal elicitors as well as non-specific elicitors also activate the WIPK and SIPK MAP kinases.

XI. Future perspectives for MAP kinase research in plants

Plant biologists will certainly have much employment in uncovering and understanding all the complex signal transduction pathways that utilize MAP kinase cascades. Arabidopsis contains at least 21 possible MAP kinase genes, and even allowing for redundancy of function amongst this family of enzymes, we have only just begun to understand the form and purpose of some of these pathways. It appears that plant MAP kinase pathways are not neatly delineated into separate parallel cascades. Some MAP kinases are known to be activated by common upstream elements, for example ATMPK3 and ATMPK6 by ANP1 through oxidative stress (Kovtun et al., 2000). However, although ATMPK4 and ATMPK6 share many similarities in their activation by stresses such as wounding (Ichimura et al., 2000b), ANP1 will not activate ATMPK4, which is thought to be activated by ATMEKK1 (Ichimura et al., 1998). It seems then that different stimuli activate different, but overlapping patterns of MAP kinases. How then is signalling specificity established at the MAP kinase level? Cellular and physiological competence may be of importance; different cells under different circumstances may present different MAP kinase target proteins. Scaffold proteins such as the yeast Spa2p protein, may also play a role in localizing MAP kinase activation to specific intracellular sites (Whitmarsh & Davis, 1998).

The literature reviewed here also underlines the need for solid evidence in order to unequivocally establish the involvement of a MAP kinase pathway in a physiological or biochemical process. There are many false leads that could be followed in the mistaken believe that a particular MAP kinase action is being monitored. Enhanced gene expression for a given MAP kinase may not correlate with enhanced kinase activity, as seen for WIPK (Zhang & Klessig, 1998). MBP kinase activity is not on its own diagnostic of MAP kinase activity, see for example Mikolajczyk et al. (2000). Antisera may not be specific for a given MAP kinase, and even if specific antisera are available, there is still room for misunderstanding, as seen in the case of WIPK and SIPK (Zhang & Klessig, 1998; Seo et al., 1999).

Clearly MAP kinase signal transduction is complex. However it is certain that MAP kinases are involved in mitotic processes and in the signal transduction of abiotic and biotic stresses in plants. There is convincing evidence that MAP kinases negatively regulate auxin responses. We have also been left with some puzzles. Why will H2O2 and oxidative stress sometimes induce MAP kinase activity and sometimes not (Ligterink et al., 1997; Desikan et al., 1999b; Romeis et al., 1999; Kovtun et al., 2000)? What, if any, is the MAP kinase downstream of CTR1? What is the role of calcium in MAP kinase signal transduction (Suzuki & Shinsi, 1995; Takahashi et al., 1997; Lebrun-Garcia et al., 1998; Romeis et al., 1999; Samuel et al., 2000)? Do plants have scaffold proteins to aid in the formation of signalosomes? Why are there so many MAP kinases in the Arabidopsis genome; how much redundancy has been built in?

Answers to these questions, and many others, will be given by the genetic, phenotypic and biochemical analysis of specific knockouts in individual and multiple MAP kinases. Already some significant insights into plant disease and stress signalling have been made through the analysis of the first two MAP kinase pathway knockouts in plants (Petersen et al., 2000; Frye et al., 2001). Screening the Arabidopsis T-DNA and transposon insertion mutant resources will permit the identification of knockouts in eventually all of these genes, however, this alone will not bring all of the answers. Some MAP kinase elements will inevitably be redundant; this will have to be addressed by crosses to make multiple mutants. Some knockouts may have a lethal phenotype, or have complex, pleiotropic effects. These genes will have to be investigated by other means; ectopic expression, or partial downregulation by antisense using tissue specific promoters, or the identification of mutant alleles with non-lethal point mutations. Another powerful approach will be to generate transgenic plants bearing kinases with constitutive activity (Miyata et al., 1999). Biochemical methods will also allow the identification of signalling complexes; recombinant MAP kinases can be used as immobilized ligands for affinity chromatography, or used to screen expression libraries. This would also help to identify the downstream targets of MAP kinases in plants, for which there is very little data as yet. Also of great importance will be the production of specific antisera, or the generation of transgenic plants bearing individual epitope tagged MAP kinases. This would help remove many of the current uncertainties as to which particular MAP kinase is represented by a MBP kinase activity. Immunoprecipitation coupled with Matrix Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF) would also permit the identification of scaffold proteins and kinases in signalosome complexes. These experimental strategies will go a long way toward unravelling stress and hormone signalling. We may find for example that MAP kinase pathways tend to act in tandem, with both stimulatory and inhibitory pathways operating together for a finer level of control. We may well uncover totally new areas where MAP kinases might operate in plant biology, such as in differentiation and morphogenesis. The tools are available, we need the time and energy to apply them.


Many thanks to friends and colleagues for reprints and preprints.