Physiological Reviews

Mitogen-Activated Protein Kinase: Conservation of a Three-Kinase Module From Yeast to Human



Widmann, Christian, Spencer Gibson, Matthew B. Jarpe, and Gary L. Johnson. Mitogen-Activated Protein Kinase: Conservation of a Three-Kinase Module From Yeast to Human. Physiol. Rev. 79: 143–180, 1999. — Mitogen-activated protein kinases (MAPK) are serine-threonine protein kinases that are activated by diverse stimuli ranging from cytokines, growth factors, neurotransmitters, hormones, cellular stress, and cell adherence. Mitogen-activated protein kinases are expressed in all eukaryotic cells. The basic assembly of MAPK pathways is a three-component module conserved from yeast to humans. The MAPK module includes three kinases that establish a sequential activation pathway comprising a MAPK kinase kinase (MKKK), MAPK kinase (MKK), and MAPK. Currently, there have been 14 MKKK, 7 MKK, and 12 MAPK identified in mammalian cells. The mammalian MAPK can be subdivided into five families: MAPKerk1/2, MAPKp38, MAPKjnk, MAPKerk3/4, and MAPKerk5. Each MAPK family has distinct biological functions. In Saccharomyces cerevisiae, there are five MAPK pathways involved in mating, cell wall remodelling, nutrient deprivation, and responses to stress stimuli such as osmolarity changes. Component members of the yeast pathways have conserved counterparts in mammalian cells. The number of different MKKK in MAPK modules allows for the diversity of inputs capable of activating MAPK pathways. In this review, we define all known MAPK module kinases from yeast to humans, what is known about their regulation, defined MAPK substrates, and the function of MAPK in cell physiology.


In the early 1980s, it was realized in several different cell types stimulated with growth factors including platelet-derived growth factor (PDGF) and epidermal growth factor (EGF) that a predominant protein phosphorylated on tyrosine had a size of 42 kDa (59 ,61). Proteins of the same size and behavior on two-dimensional gels were phosphorylated on tyrosine in response to phorbol esters (174), in virus-transformed cells (60), and in metaphase-arrested Xenopus laevis eggs (57). At the same time, serine-threonine protein kinases activated by the insulin receptor tyrosine kinase were being characterized (7 ,287 ,294). The hypothesis was considered that direct regulation by tyrosine phosphorylation catalyzed by the insulin receptor would regulate the activity of serine-threonine protein kinases (7 ,287). Ray and Sturgill (287) demonstrated that a 42-kDa serine-threonine protein kinase, referred to as mitogen-activated protein kinase (MAPK), isolated from insulin-stimulated 3T3-L1 cells was phosphorylated on both threonine and tyrosine. It was quickly realized that the 42-kDa MAPK characterized to be threonine and tyrosine phosphorylated in response to insulin was the same protein shown to be tyrosine phosphorylated in response to other growth factors, phorbol esters, viral transformation, and metaphase arrest in Xenopus eggs (294). It was also demonstrated that phosphorylation of both the threonine and tyrosine was required for MAPK activation (294).

The cDNA encoding MAPK was isolated by Boulton et al. (32) who renamed it ERK1 for extracellular signal-regulated kinase 1, because of the variety of extracellular signals that could stimulate its activity. The isolation of cDNA for ERK2 and ERK3 quickly followed (31). Alignment of the ERK sequences showed they were closely related to the Saccharomyces cerevisiae protein kinases Fus3 and Kss1, demonstrating the close homology between mammalian and yeast MAPK.

Biochemical characterization and molecular cloning identified the upstream kinases in the MAPKerk module in mammalian cells (67). The work defined a conserved three-kinase module in mammals. The MAPK regulatory system is a three-kinase module that establishes a sequential protein kinase activation pathway.

In parallel, geneticists were identifying the component genes in yeast mating. The sterile (ste) genes in S. cerevisiae involved in pheromone-induced mating were ordered and cloned (136). Quickly, the related genes in Schizosaccharomyces pombe were identified and their protein products characterized (260). Cumulatively, this work identified a conserved three-component protein kinase module that included the MAPK Fus3 and Kss1 in S. cerevisiae and Spk1 in S. pombe.

In this review, we detail the properties of the known MAPK modules. The expanding number of MAPK modules and their role in controlling complex cellular functions defines their importance in responsiveness of cells and organisms to their environment.


Pathways involving MAPK are activated in response to an extraordinarily diverse array of stimuli. These stimuli vary from growth factors and cytokines to irradiation, osmolarity, and the shear stress of fluid flowing over a cell. The basic assembly of MAPK pathways is a three-component module conserved from yeast to humans. The minimal MAPK module is composed of three kinases that establish a sequential activation pathway (Fig. 1). The first kinase of the three-component activation module is a MAPK kinase kinase (MKKK) (92). Specific MKKK have been shown to be activated either by phosphorylation by a MAPK kinase kinase kinase (MKKKK) or by interaction with a small GTP-binding protein of the Ras or Rho family. Other potential modes of activation include oligomerization and subcellular relocalization. The MKKK are serine/threonine kinases that when activated phosphorylate and activate the next kinase in the module, a MAPK kinase (MKK) (315). The MKK are kinases that recognize and phosphorylate a Thr-X-Tyr motif in the activation loop of MAPK (109), defining MKK as dual-specificity kinases. Mitogen-activated protein kinases are the final kinase in the three-kinase module and phosphorylate substrates on serine and threonine residues. The vast majority of defined substrates for MAPK are transcription factors. However, MAPK have the ability to phosphorylate many other substrates including other protein kinases, phospholipases, and cytoskeleton-associated proteins.

Fig. 1.

Core module of a mitogen-activated protein kinase (MAPK) pathway is composed of 3 kinases [MAPK kinase kinase (MKKK), MAPK kinase (MKK), and MAPK] that are sequentially activated by phosphorylation.

Why have MAPK activation modules evolved having three kinases? The reason probably lies in the unique activation properties of MAPK. Mitogen-activated protein kinase must be phosphorylated on both a threonine and tyrosine for their activation, a dual phosphorylation catalyzed by a specific MKK. The different MKK recognize a tertiary structure of specific MAPK and not simply a linear sequence surrounding the Thr-X-Tyr activation motif of MAPK. As described in this review, very specific MKK and MAPK combinations are found in a MAPK module. Specific MKK appear to recognize the tertiary structure of different MAPK, effectively restricting their regulation of different MAPK subtypes. In contrast, MKKK are able to mix and match with different MKK-MAPK combinations. In mammalian cells, there are more known MKKK than MAPK. The MKK represent the fewest members of the three-component MAPK modules. The completion of the human genome project will be required to define the exact number of kinases in each of these groups. The large number of MKKK allows for diversity of inputs from numerous stimuli to feed into specific MAPK pathways. Some kinases that appear to be MKKK may regulate pathways not involving MAPK [such as regulation of the NFκB pathway by MEKK1 (139 ,196 ,233 ,393)]. Thus the regulation of MAPK pathways at the level of MKKK may represent branch points in regulation of signal pathways in some cases.

Table 1 lists the members of the known MAPK three-component modules defined to date. Perusal of Table 1 boggles the mind and clearly demonstrates that the current nomenclature, or more appropriately the lack of nomenclature, in the MAPK field poses a major problem of understanding the MAPK literature, especially for readers not familiar with the field. In the present review, we decided to name a given component of a MAPK module according to its position in the pathway with the current name of the protein in superscript. For example, the yeast Ste11 MAPK kinase kinase will be described as MKKKste11. This notation should greatly facilitate the reading of this review, since it is not required to know the identity of each kinase to determine its position in a given MAPK pathway.

View this table:
Table 1.

Components of MAPK pathways

To date, 14 MKKK, 7 MKK, and 12 MAPK have been identified in mammalian cells (Table 1). Dendogram analysis indicates that these kinases belong to different subfamilies (Fig. 2). Four subfamilies among the MKKK can currently be defined. The Raf subfamily is the best characterized and comprises MKKKB-raf, MKKKA-raf, and MKKKraf1. The MEK kinase (MEKK) subfamilly is made of the four MEKK (MKKKmekk1–4). MKKKask1 and MKKKtpl2 appear to form a third MKKK subfamily. The fourth group in the dendogram is more diverse and comprises MKKKmst, MKKKsprk, MKKKmuk, MKKKtak1, and the most distantly related MKKK, MKKKmos. For the MKK, MKKmek1 and MKKmek2 are closely related as are MKKmkk3 and MKKmkk6. The MAPK can be categorized into five subfamilies: the MAPKerk1/2, the MAPKp38, the MAPKjnk, the MAPKerk5, and the MAPKerk3/4 subfamilies. These MAPK give the name to the MAPK pathways that employ them (i.e., the MAPK pathways using the MAPKjnk are called the JNK MAPK pathways). This review focuses on the organization, regulation, and function of the different MAPK modules in eukaryotic cells.

Fig. 2.

Phylogenetic trees of mammalian MAPK, MKK, and MKKK family members. Dendrograms were created with CLUSTAL X program (334) using human sequences when available or mouse sequences otherwise (see Table 1).


Mitogen-activated protein kinases are proline directed in that they only phosphorylate substrates that contain a proline in the P-1 site (219). A general consensus for MAPKerk1/2 is Pro-X-Ser/Thr-Pro (6). The activity of MAPK is controlled by dual phosphorylation in an amino acid sequence known as the activation loop. The sequence Thr-X-Tyr in the activation loop, where X can be different amino acids among the MAPK, is the site for dual phosphorylation catalyzed by specific MKK (4). For MAPKerk1/2, the phosphorylation sites correspond to Thr-183 and Tyr-185. Dual phosphorylation of these sites results in a >1,000-fold increase in specific activity of the MAPK.

The core three-dimensional structure of protein kinases, as resolved from the crystal structure of protein kinase A (PKA), is composed of two domains with the active site at the domain interface (177). Adenosine 5′-triphosphate binds in the active site cleft, and peptide substrate for some kinases has been shown to bind in a groove on the surface of the COOH-terminal domain of the active site (178). The phosphoreceptor amino acid (Ser, Thr, Tyr) of the substrate binds near the catalytic loop in this domain. A surface loop contiguous with this COOH-terminal domain is found in many protein kinases and encodes a phosphorylation site; this sequence is referred to as the activation loop or lip. Threonine-183 in MAPKerk1/2 is homologous to the phosphorylation site in the activation loop of other protein kinases (165). The tyrosine phosphorylation site in the activation loop of MAPK is unique.

The recent crystal structure of active, dual-phosphorylated MAPKerk2 has been recently solved (40). Phosphorylation of Thr-183 and Tyr-185 in the activation loop causes the loop to refold and interact with surface arginine-binding sites. The conformational changes in the activation loop and neighboring sequences result in activation of the kinase.

An interesting property of the requirement for dual phosphorylation of the Thr-X-Tyr sequence is that substitution of the Thr and Tyr by acidic amino acids like glutamate do not result in constitutive activation. The characteristics of the activation loop and the dual phosphorylation of the Thr and Tyr induce several conformational changes in the protein that cannot be mimicked by glutamates. For this reason, point mutations in MAPK do not activate their activity. The requirement for the P-Thr-X-P-Tyr regulation of MAPK activity has been proposed as why MAPK have never been identified as oncogenes (40), as could be expected from proteins regulating cell growth.


Yeast probably represents the experimental model where the organization and regulation of MAPK pathways are best understood. Presently, five MAPK pathways have been well characterized in S. cerevisiae: the haploid mating pathway, invasive growth, cell wall remodeling, and two pathways involved in stress responses such as hyperosmolarity. These five MAPK pathways are discussed in detail in section iv. An important concept on MAPK pathways that has emerged from yeast studies is that the kinases employed in MAPK pathways are organized into modules. As discussed in section iv A, this is achieved by tethering to scaffold proteins as well as by direct interaction between the different kinases of the module. Organization into modules ensures segregation of the pathways from other signaling events in the cells and also allows the use of a given component kinase in more than one MAPK module without affecting the specificity of the response mediated by the MAPK pathways.

A. Organization of MAPK Modules

Genetic and biochemical studies in yeast have demonstrated that specific MKKK-MKK-MAPK signaling modules are functionally segregated from one another (98). In the budding yeast, S. cerevisiae, the three component MAPK modules are assembled either by tethering to scaffolding proteins such as Ste5 or by binding domains encoded in the kinases themselves. Ste5 is a 110-kDa dimeric protein that is capable of binding the three-kinase components of the pheromone-regulated mating MAPK pathway, MKKKste11, MKKste7, and MAPKfus3. This ensures that this particular module is coordinately regulated and segregated from other signaling pathways. Tethering and segregation of MAPK modules may in fact be vital for MAPK pathways to function properly (217). Mitogen-activated protein kinase module kinases may have high affinity for each other in the absence of a tethering protein like Ste5. For example, MKKKfus3 and MKKste7 have an ∼5 nM dissociation constant binding affinity, indicating they could have significant interaction in the the absence of Ste5 (13). Similarly, MKKpbs2 has the ability to bind several different MKKK involved in osmoregulation (see sect. iv F). The complexity of regulation of MAPK pathways is further exemplified by the fact that some kinases participate in more than one MAPK module. For example, MKKKste11 functions in the MAPK pathways for mating, pseudohyphal development, and osmoregulation. However, because of differential upstream inputs from MKKKK, GTP-binding proteins, and the assembly of three kinases into modules, there is generally little if any cross talk between the different MAPK pathways in S. cerevisiae.

B. Role of MAPK Modules in Yeast

Yeast may exist either as haploid or diploid cells. The haploid cells have two sexual phenotypes characterized by the expression of a set of genes involved in mating that are not expressed in diploids. The mating response to generate diploids is controlled by the a and α-pheromones that bind to their respective receptors that are coupled to a heterotrimeric G protein. Pheromone binding to its receptor leads to G protein activation and the dissociation of the βγ-subunit complex from α-GTP. In budding yeast, the βγ-complex binds to Ste5 and by a process still poorly understood stimulates activation of the mating MAPK pathway. In addition to mating, yeast respond to their environment with metabolic changes that involve MAPK pathways. For example, in response to starvation, yeast undergo dramatic morphological changes involving pseudohyphal formation and invasiveness in an attempt to find nutrients that are controlled in part by a MAPK pathway. Yeast adapt to additional adverse environmental conditions such as high or low osmolarity by activation of specific MAPK pathways. Thus many aspects of yeast physiology controlling their response to environmental cues are, at least in part, regulated by MAPK pathways (136). Knowledge of the entire genome sequence for the budding yeast S. cerevisiae has defined the number of protein kinases in the organism (147). Relative to MAPK modules, it appears that the S. cerevisiae genome encodes four MKKK, four MKK, and six MAPK (Table 1). Among the six different MAPK present in S. cerevisiae (147), only two cannot be attributed to one of the five well-characterized MAPK pathways: MAPKsmk1, which is involved in spore wall assembly, and the putative MAPKYKL161C identified by genome sequencing. MAPKsmk1 and MAPKYKL161C may be involved in as yet undefined MAPK pathways. MAPKYKL161C has a K-X-Y motif in its activation loop distinguishing it from other MAPK.

C. Mating Pathways in Haploid Yeast

The best-defined yeast MAPK pathway in S. cerevisae is involved in the mating of haploid cells (21 ,58 ,89 ,305) (Fig. 3). The two haploid cell types (a and α) of S. cerevisiae, upon binding the sexual pheromone secreted by the opposite cell type (a-factor and α-factor), stop growing and differentiate into mating-competent cells by inducing transcription of mating genes. The S. cerevisiae genes whose disruption inhibited mating and caused sterility were designated as sterile (ste) genes. The seven transmembrane receptors for the α- and a-factors are designated as Ste2 and Ste3, respectively, and are coupled to a heterotrimeric G protein. Pheromone activation of the G protein induces the dissociation of the heterotrimeric G protein subunits designated Gpa1 (α-subunit), Ste4 (β-subunit), and Ste18 (γ-subunit). The released βγ-subunit complex (Ste4/Ste18) activates Ste20 and interacts with the scaffolding protein Ste5, resulting in the stimulation of the MAPK module MKKKste11/MKKste7/MAPKfus3. It should be noted that involvement of Ste20 in the activation of the mating MAPK pathway is still debated, since it has not yet been unequivocally proven that Ste20 phosphorylates and activates MKKKste11. Activated MAPKfus3 regulates the activity of transcription factors required for the expression of components of the mating pathway itself and genes necessary for cell cycle arrest and cell fusion (136).

Fig. 3.

Saccharomyces cerevisiae MAPK pathway used for mating. Note that Ste5 is a scaffolding protein that is able to bind to Ste4, MKKKste11, MKKste7, and MAPKfus3.

It was initially thought that MAPKfus3 and MAPKkss1 were redundant kinases in the mating response, because genetic experiments indicated that expression of one protein could complement the mating defect induced by the absence of the other. This notion was widely accepted in the MAPK field even though a number of observations indicated that MAPKfus3 and MAPKkss1 were differently regulated and were in fact components of different signaling pathways. First, MAPKfus3 and MAPKkss1 are not activated by the same stimuli. MAPKfus3 activity is very low, whereas MAPKkss1 exhibits high activity, in the absence of mating pheromones. Conversely, in presence of pheromones, MAPKfus3 is strongly activated but MAPKkss1 only weakly if at all (13 ,85). Second, MAPKfus3 is expressed only in haploid cells, whereas MAPKkss1 is expressed both in haploid and diploid cells, indicating that these two MAPK have different functions. Third, MAPKfus3 and MAPKkss1 have different substrate targets (259). With regard to substrate recognition, it has been shown that MAPKfus3, but not MAPKkss1, phosphorylates Far1 (85 ,275) that represses the transcription of the G1 cyclins Cln1 and Cln2 (83), leading to cell cycle arrest that is a feature associated with mating. Cumulatively, these studies suggested that MAPKfus3 is the MAPK involved in mating, whereas MAPKkss1 is not. This was confirmed recently with experiments showing that the expression of a kinase inactive mutant of MAPKfus3 in a MAPKfus3 null background hampered the ability of MAPKkss1 to fulfill mating MAPK functions (217). The findings indicated that MAPKkss1 cannot activate the mating MAPK pathway in wild-type cells because it is excluded from this pathway by MAPKfus3. Only in null mutants lacking MAPKfus3 can MAPKkss1 substitute for MAPKfus3 in the mating pathway, explaining the earlier results that suggested redundancy between the two MAPK. Thus MAPKfus3 has an additional function that is to segregate the mating MAPK pathway from imposter proteins, in this case MAPKkss1, that are normally involved in other signaling pathways. Segregation of the MAPK module involved in mating is further achieved by Ste5, the scaffold molecule which has no apparent enzymatic activity itself. Different domains of Ste5 bind, in the two-hybrid system, to MKKKste11 and MAPKfus3 (53 ,151 ,184 ,226 ,280) as well as to Ste4 (G protein β-subunit) (365) and is predicted to couple the βγ-complex (Ste4/Ste18) to activation of the MAPK mating pathway (Fig. 3).

The NH2-terminal portion of Ste5 (residues 177–229) contains a cysteine-rich region that is the prototype for the RING-H2 motif. Proteins of the zinc RING family possess two fingerlike domains connected by a linking region and require zinc for folding (18 ,29). Crystal structures indicate that RING domains are globular pseudosymmetric folds that coordinate two zinc atoms through a cross-bridging element (14 ,28). When this structure is mutated, the corresponding Ste5 mutants are unable to complement the mating defect of yeast strains lacking Ste5 (ste5Δ cells) (152). The RING-H2 mutants, while being able to bind MKKKste11, MKKste7, and MAPKfus3 as efficiently as the wild-type protein, were not capable of binding Ste4. Moreover, the RING-H2 mutants could not dimerize (152). When the Schistosoma japonicum glutathione S-transferase protein, known to form stable dimers (230 ,349), was fused to the COOH terminus of the RING-H2 Ste5 mutants, the resulting protein could complement the mating defect of ste5Δ cells but could not associate with Ste4. Oligomerization of Ste5 is, however, not sufficient to transduce the mating signal, since wild-type Ste5 fused to the S. japonicum glutathione S-transferase was able to complement ste5Δ cells but not ste4Δ ste5Δ cells (152). The conclusion of these studies is that the RING-H2 domain has inhibitory functions that can be alleviated by Ste4. Once the inhibitory role of the RING-H2 domain is suppressed, it oligomerizes, leading to the activation of the MAPK mating module, possibly by allowing Ste20 to phosphorylate MKKKste11. It is presently unclear if MKKKste11 phosphorylation is the primary regulatory event that controls its activity, even though it is a substrate for Ste20.

Stimulation of the MAPK mating pathway leads to the activation or repression of the activity of a variety of proteins. For example, Far1 phosphorylated by MAPKfus3 binds to and inactivates the cell division control (Cdc)28/Cln kinase complex and thus inhibits cell growth. Phosphorylation of Far1 is critical, since a mutant allele of Far1 that is not fully phosphorylated is unable to mediate pheromone-induced cell cycle arrest (275). Ste12, a transcription factor mediating the induction of pheromone-response genes, is also phosphorylated by MAPKfus3 (85). One of the functions of Ste12 is to induce Far1 transcription (266). Thus MAPKfus3 regulates Far1 in two complementary ways: a direct phosphorylation-dependent activation and an increase in transcription through activation of Ste12.

Yeast cells that are ready to undergo mating must make sure that they do not grow, and this is achieved in part by the MAPKfus3 phosphorylation and activation of Far1. Similarly, cells undergoing cell division should not be responsive to pheromones. Interestingly, the G1 cyclins that are inhibited by Far1 in mating competent cells are themselves inhibitors of the mating pathways in dividing cells. It has indeed been shown that Cln2 represses the mating MAPK pathway, and this inhibition takes place at the level of MKKKste11 (362). There are other examples demonstrating the reciprocal actions of the mating MAPK pathway and cyclin-dependent kinases. For example, the rat glucocorticoid receptor ectopically expressed in S. cerevisiae is phosphorylated by cyclin-dependent kinases and MAPK at distinct sites. Glucocorticoid receptor-dependent transcriptional enhancement is reduced in a yeast strain deficient in cyclin-dependent kinases but is increased in a strain devoid of MAPKfus3 (186).

In the fission yeast S. pombe, mating involves a MAPK module containing homologs of the S. cerevisiae MAPK mating pathway (MKKKbyr2 is homologous to MKKKste11, MKKbyr1 to MKKste7, and MAPKspk1 to MAPKfus3) (89) (Fig. 4). The S. pombe and S. cerevisiae mating pathways are controlled by different upstream regulators. First, in the budding yeast, the Gβγ complex transmits the signal, whereas in S. pombe, it is the Gα subunit that transmits the signal. Second, although haploid budding yeast express the genes required for the pheromone response in all nutritional conditions, differentiation of fission yeast into mating competent cells is strictly dependent on nutritional starvation. The S. pombe Ras homolog Ras1 plays a role in the starvation-dependent control of the mating pathway, possibly in a Gα-independent manner (384). In response to mating pheromones, mutants defective in Ras1 or in Ste6, the fission yeast homolog of the Ras GDP/GTP exchange factor, are unable to induce transcription of the mat1-Pm gene that controls entry into meiosis (262). Ras1 may be directly linked to the S. pombe MAPK module involved in mating as suggested by the observation that Ras1 can bind MKKKbyr2 in a two-hybrid assay (48 ,344). Interestingly, a Ste5 equivalent has not been defined in S. pombe, nor is there evidence that phosphorylation of MKKKbyr2 by a Ste20 homolog is required. Current evidence argues that GTP-binding proteins regulate MKKKbyr2 activity.

Fig. 4.

Schizosaccharomyces pombe MAPK pathway used for mating.

D. Invasive Pathways: Pseudohyphal Pathway of Diploids and Invasive Growth of Haploids

In response to nitrogen starvation, diploid S. cerevisiae activate an intracellular pathway that will eventually lead to profound morphological changes. When starved for nitrogen, the elliptical diploid yeast undergo an asymmetric cell division to produce a long thin daughter cell that will keep producing long daughter cells. Because the mother and daughter cells remain attached, reiteration of this unipolar division pattern produces filaments composed of a linear chain of elongated cells called a pseudohypha. The mother yeast produce colonies on the surface of an agar plate, whereas the pseudohypha invades the agar (112 ,113). Haploid cells can also invade the agar and grow beneath the surface. The response of haploids is not as dependent on nutritional starvation as the pseudohyphal development of diploid cells. In particular, nitrogen limitation seems to play no role in the invasive growth response of haploids (293). A MAPK module consisting of MKKKste11, MKKste7, and MAPKkss1 can transduce the signals leading to invasive growth (Fig. 5). Mitogen-activated protein kinasekss1 was initially thought to play no role in invasiveness because diploid mutants lacking MAPKkss1 had a normal pseudohyphal response (136). However, with the use of strains bearing hypermorphic alleles of MKKKste11 and MKKste7 to increase filamentous growth, it was shown that the absence of MAPKkss1 blocked hyperfilamentation, as well as the enhancement of the activity of promoters containing filamentation and invasion response elements. The kinase activity of MAPKkss1 is required for filamentous growth, both in haploids and diploids, because removal of MAPKKste7, the upstream regulator of MAPKkss1, or point mutations impairing the catalytic activity of MAPKkss1 result in severe filamentation defects (121 ,217). If MAPKkss1 is required for a normal invasive response, how is it that MAPKkss1 knockout mutants have no pseudohyphal defect? The postulated answer is that MAPKkss1 also has an inhibitory function in the invasive response and that there are MAPK-independent signaling pathways mediating invasiveness (121 ,217). In the absence of MAPKkss1, the MAPK-independent pathways can stimulate the invasive growth because they are not hampered by the inhibitory function of MAPKkss1. It has been possible to genetically separate the inhibitory and stimulatory functions of MAPKkss1 on pseudohyphal development. There are mutants of MAPKkss1 that have lost their inhibitory function that when expressed lead to hyperfilamentation in a MAPKkss1 null background (217). In contrast, in the situation where MAPKkss1 cannot be activated (i.e., when MAPKKste7 is absent), filamentation is drastically reduced because MAPKkss1, while unable to activate filamentation since this requires the stimulation of its catalytic activity, retains its negative function (121 ,217). This also shows that the inhibitory function of MAPKkss1 does not require its catalytic activity. When not activated by MKKste7, MAPKkss1 is thus an inhibitor or silencer of the invasive growth response. In contrast, when MAPKkss1 is activated by MKKste7, it stimulates invasive growth, and this response depends on an intact kinase activity. The negative function of MAPKkss1 is mediated by its interaction with its target Ste12. MAPKkss1 interacts with Ste12 predominantly in its inactive state. This interaction inhibits the activity of Ste12. Once MAPKkss1 is activated, it no longer binds and inhibits Ste12; rather, active MAPKkss1 phosphorylates and activates Ste12, leading to activation of genes under the control of promoters containing filamentation and invasion response elements (217).

Fig. 5.

Saccharomyces cerevisiae MAPK pathway used during invasive responses.

Invasiveness is more pronounced in the absence of MAPKfus3 (either in mutant haploids cells or in diploids that do not express MAPKfus3) (121 ,293), indicating that MAPKfus3 functions as a repressor of the invasive growth response. Thus, when activated, MAPKfus3 and MAPKkss1 have opposite functions in invasive growth. Possibly, one role of the MAPK pathway employing MAPKfus3 is to suppress invasive growth in haploid cells to facilitate their entry into the mating program where the cells must stop growing. In such a scenario, in the presence of pheromones, the inhibitory activity of MAPKfus3 on invasive growth would be greater than the stimulatory activity of MAPKkss1. This is consistent with the fact that pheromones activate MAPKfus3 much more strongly than MAPKkss1 (13 ,85). Clearly, the balance of these kinase activities, both regulated by MKKste7, will play a role in determining whether a cell will exhibit an invasive response.

The exact role of the MKKKste11/MKKste7/MAPKkss1 module in the control of invasive growth is still incompletely understood. The need for a better understanding of invasive growth becomes obvious when it is considered that removal of all the components of the MKKKste11/MKKste7/MAPKkss1 module does not markedly affect the invasive growth response (121). Thus, even though genetic studies show invasive growth can be strongly affected by mutants in the MKKKste11/MKKste7/MAPKkss1 module, the pathway itself is not required. This shows that filamentation and invasion response elements are under the control of alternative regulatory pathways in the absence of a functional MKKKste11/MKKste7/MAPKkss1 pathway. Such pathways may involve Ras, since it has been shown that activated forms of Ras potentitate the invasive growth response (113 ,121 ,250).

E. Cell Wall Remodeling Pathway

In yeast, growth depends on efficient cell wall remodeling, and this response also uses a MAPK pathway (Fig. 6). In the budding yeast, the MAPK module is composed of MKKKbck1, MKKmkk1 or MKKmkk2, and MAPKmpk1. The biological relevance of having two seemingly redundant MKK in the cell wall remodeling pathway is unclear. The upstream regulator of the MAPK module of the cell wall remodeling pathway is the yeast homolog of mammalian protein kinase C (PKC), PKC1, which could function as a MKKKK for this pathway although it has not been proven biochemically that PKC1 can directly phosphorylate and activate MKKKbck1. Mutants lacking PKC1, MKKKbck1, or MAPKmpk1 lyse under isotonic conditions because of a deficiency in cell wall construction (154 ,199 ,206). However, the phenotypes of the mutants defective in PKC1 are more severe than in the mutants defective in the downstream components of the pathway; PKC1 mutants have a lysis defect at all temperatures (200). This suggests that PKC1 controls two pathways required for an optimal cell-wall remodeling response, one of which contains the MKKKbck1/MKKmkk1 or MKKmkk2/MAPKmpk1 module.

Fig. 6.

Saccharomyces cerevisiae MAPK pathway used for cell wall construction. PKC, protein kinase C.

In the fission yeast S. pombe, MAPKpmk1 may be part of a MAPK pathway involved in maintenance of cell wall integrity (337 ,395). However, MAPKpmk1 may not be a downstream target of PKC in S. pombe but may function in coordination with PKC to regulate cell integrity (337). The upstream regulators of MAPKpmk1 remain to be characterized in fission yeast.

F. Osmosensor and Stress Pathways

The budding yeast S. cerevisiae activates two pathways in response to hyperosmolarity (220) (Figs. 7 and 8). The osmosensors that stimulate these pathways function in a very different manner. One pathway is regulated by a set of three proteins functioning like the prokaryotic two-component transduction system, and the other osmosensor is an integral membrane protein. These two sensors regulate two different MAPK pathways utilizing common kinase elements that will lead to the transcription of genes necessary for survival in hyperosmotic conditions such as those required for the synthesis of glycerol to increase the internal osmolarity (136). Although external high osmolarity will activate both of these MAPK pathways, the corresponding osmosensors do not detect the same osmolarity changes. The integral membrane osmosensor is activated by high osmolarity and stimulates its corresponding MAPK pathway. In contrast, the two-component osmosensor is active in low-osmolarity conditions, and this results in the repression of the associated MAPK module. In high-osmolarity conditions, however, this osmosensor is inactivated and no longer inhibits the activation of the MAPK. Apparently, fission yeast also employ two related MAPK pathways to sense changes in osmolarity. Two different MAPK pathways are thus activated in response to hyperosmotic conditions in both S. cerevisiae and S. pombe. The reason for this is not yet understood, but it is likely that each of the osmosensor MAPK pathways do not fulfill exactly the same function in response to increased osmolarity and that they may be differentially involved in sensing and responding to other types of stress.

Fig. 7.

“Two-component” osmosensor MAPK pathway of Saccharomyces cerevisiae. Solid and dotted boxes correspond to active and inactive state of proteins, respectively.

Fig. 8.

Sho1-dependent osmosensor MAPK pathway in Saccharomyces cerevisiae. MKKpbs2 functions both as a MKK and as a scaffold protein that binds to Sho1, MKKste11, and MAPKhog1.

1. “Two-component” osmosensor pathway

Two-component transduction systems are commonly found in prokaryotes. A two-component system is composed of a sensor molecule and a response-regulator molecule. Typically, a sensor protein has an extracellular input domain and a cytoplasmic histidine kinase domain. A typical response-regulator is a cytosolic protein containing the receiver domain and a DNA binding domain. When the sensor protein is activated, it phosphorylates a histidine residue within its kinase domain and transfers this phosphate group to an aspartic acid in the receiver domain of the cognate response-regulator molecule, resulting in the switching of its output function that is generally transcriptional activation. It is because the signaling pathway is composed of only two proteins that it is called a two-component system. The two-component osmosensor in yeast is, however, composed of three proteins, Sln1, Ypd1, and Ssk1, that funtionally behave as two linked two-component systems (Fig. 7). Sln1 corresponds to the first two-component system, since it contains both a histidine kinase domain and a receiver domain. The Ypd1-Ssk1 pair functions as the second two-component system. Ypd1 is phosphorylated on a histidine residue as a result of a transfer of the phosphate on the aspartic acid of Sln1. This phosphate is then transferred to Ssk1 on an aspartic acid (278). Sln1, Ypd1, and Ssk1 thus function as a multistep phosphorelay. The advantage of such a system may be its ability to be regulated at different levels. Ssk1 has a kinase activity and is able to stimulate a MAPK module composed of MAPKhog1, MKKpbs2, and two apparently redundant MKKK, MKKKssk2 and MKKKssk22 (23 ,35 ,220) (Fig. 7). Ssk1 is inactive when phosphorylated (i.e., in low-osmolarity conditions) and activated by high osmolarity (when the sensor Sln1 is inactive), which eventually leads to stimulation of MAPKhog1 and transcription of genes necessary for survival in hyperosmotic conditions.

2. Sho1-dependent osmosensor pathway

In addition to the Sln1-Ypd1-Ssk1 osmosensor pathway, the budding yeast has an alternate way of sensing hyperosmolarity that relies on the Sho1 osmosensor (220) (Fig. 8). Sho1 contains four transmembrane domains and a COOH-terminal cytoplasmic region with a Src homology 3 (SH3) domain (220). In high-osmolarity conditions, MKKKste11 is activated in a Sho1-dependent manner. MKKKste11 then phosphorylates and activates MKKpbs2, which activates MAPKhog1 in the module (277). Activated MAPKhog1 regulates the transcription of genes required for survival in hyperosmotic conditions (Fig. 8). In addition to its roles in the mating and the pseudohyphal pathways, MKKKste11 is also involved in an osmoregulation pathway, indicating that MKKKste11 is a component of at least three different MAPK modules. Activation of MKKKste11 in one module does not necessarily imply that MKKKste11 present in the other MAPK modules will be activated. The reason for this is the segregation and assembly of different MAPK modules regulated by different upstream inputs. For example, pheromones induce the phosphorylation and activation of MAPKfus3 but not MAPKhog1, even though MKKKste11 is the upstream regulator of both MAPK, indicating that MKKKste11 is activated in the mating pathway but not in the osmosensor pathway. Conversely, hyperosmolarity will activate MKKKste11 in the Sho1-dependent osmosensor pathway as assessed by the fact that MAPKhog1 gets phosphorylated, but hyperosmolarity does not lead to the activation of MKKKste11 in the mating pathway demonstrated by the fact that MAPKfus3 is not phosphorylated (277).

The basis for the differential control is accounted for by the regulation of the mating MAPK module by Ste5 and the Gβγ subunits (Ste4/Ste18) while MKKpbs2 is able to bind to Sho1, MKKKste11, and MAPKhog1 [the interaction of Sho1 and MKKpbs2 being mediated by the poly-proline-rich region of MKKpbs2 and the SH3 domain of Sho1 (220)]. Thus MKKpbs2 also functions as a scaffold protein to segregate the MKKKste11/MKKpbs2/MAPKp38 module for regulation by Sho1 (220 ,277). The properties of Ste5 and MAPKKpbs2 in the mating and Sho1-dependent MAPK pathways, respectively, ensure that the two pathways are segregated from one another.

3. Osmosensing pathways in S. pombe

In fission yeast, MAPK pathways are also activated under stress conditions to mediate survival responses. Similar to budding yeast, it appears that two related MAPK pathways are activated in response to environmental stress. One is composed of the MKKKwik1/MKKwis1/MAPKspc1 module, and the other is most likely composed of the MKKKwin1/MKKwis1/MAPKspc1 module (299 ,312 ,313) (Fig. 9). The pathway using MKKKwin1 may be predominant in conditions of osmotic stress because MKKKwik1 is not essential for osmosignaling (299). Interestingly, activation of MAPKspc1 can be achieved in the absence of MKKwis1 phosphorylation in response to heat shock and oxidative stress. The inactivation of the Pyp1 protein tyrosine phosphatase is likely to be involved in this alternative MAPKspc1 activation mechanism (299). The upstream regulator of MKKKwik1, Mcs4, is the homolog of the S. cerevisiae Ssk1 protein (312). Thus, in lower eukaryotes, the stress-activated MAPK pathway appears to be controlled by a conserved two-component system. In addition to involvement in stress responses, MKKwis1, MAPKspc1, and Pyp1 have roles in cell cycle progression. MKKwis1 was originally isolated as a mitotic inducer (360), and Pyp1 has been shown to negatively regulate entry into mitosis (236).

Fig. 9.

Stress-induced MAPK pathways in Schizosaccharomyces pombe.

G. Sporulation Pathway

Yeast sporulation is the process, involving meiosis, that leads to the packaging of haploid nuclei into spores. This reponse occurs only in diploids and is elicited by nutritional starvation (241). Upon completion of meiosis, the four-haploid nuclei, which still remain within a single nuclear membrane, are enveloped by the double membraneous prospore wall. The spore wall is then deposited from the space between the layers of the prospore wall. The final differentiated spore wall consists of four layers. The two inner layers appear indistinguishable from the vegetative cell wall, whereas the third layer is a spore-specific structure composed primarily of chitosan and chitin. The outermost electron-dense layer is a dityrosine coat (36 ,37). One gene encoding a MAPK, MAPKsmk1, has been shown to be involved in the sporulation pathway (MAPKsmk1 mutants are defective in spore wall assembly) (185). A putative MAPK module, which employs MAPKsmk1 and an uncharacterized MKKK and MKK, may thus be used to regulate the sporulation response. A MKKKK homolog (Sps1) has also been shown to be involved in the sporulation pathway. Both Sps1 and MAPKsmk1 mutants display similar phenotypes: they both proceed normally through meiosis but then are defective in spore wall assembly (136 ,185). This suggests that Sps1 controls the putative MAPK module mediating the sporulation response.


One important point that emerges from the the studies of the yeast MAPK pathways is that MAPK pathways form modules that are held together through protein-protein interactions. Activation of one MAPK pathway does not normally lead to the activation of other MAPK pathways, even if a given component is found within multiple pathways. Mitogen-activated protein kinase pathways are thus spatially regulated in yeast. It is thus predicted that this also applies to the mammalian MAPK pathways. In mammalian cells, however, multiple MAPK pathways can be activated by a single receptor type. For example, the high-affinity FcεR1 receptor for IgE activates the MAPKerk, MAPKjnk, and MAPKp38 pathways in mast cells. Mammalian cells also do not lend themselves to genetic studies as performed in yeast. For this reason, many studies in mammalian cells involve overexpression of activated and inhibitory mutant components of MAPK pathways. A potential pitfall, however, is that this methodology may perturb the segregation mechanisms holding individual MAPK modules. The challenge in the forthcoming years will be to improve existing methods or develop new technologies in mammalian systems to overcome these limitations. Obviously, gene knock-out strategies may be particularly usefull in this context.

Fig. 10.

MKKK, MKK, and MAPK that can be components of MAPKerk pathway. GPCR, G protein-coupled receptor; RTK, related tyrosine kinase; ERK, extracellular-regulated kinase; MEK, MAPK or ERK kinase.

In yeast, it appears that transcription factors comprise the majority of the known substrates regulated by MAPK pathways. In mammalian cells, many MAPK substrates defined to date are also transcription factors. In addition, several cytoskeletal proteins, protein kinases, and phopholipases are also substrates for specific MAPK. As our understanding of the yeast and mammalian MAPK pathways increases, it is likely that additional classes of MAPK substrates will be defined.

Four types of MAPK pathways have been defined to date in mammalian cells. However, within a given pathway, several MKKK, MKK, and MAPK can generally be found to be interchangeable. For example, there are 3 isoforms and 10 different splice variants of MAPK in the c-Jun kinase (MAPKjnk) pathway (123). The relevance of this complexity is poorly understood and is difficult to assess, because it is not straightforward to dissect the role of individual components of the MAPK pathways in mammalian cells. Mammalian MAPK pathways are involved in a diverse set of responses affecting cell fate, including cell proliferation and differentiation, adaptation to environmental stress, and apoptosis.

A. MAPKerk Pathway

The best described MAPK signaling pathway in mammalian cells is the extracellular signal-regulated kinase (MAPKerk) pathway. This pathway can apparently include a number of different MKKK and MKK (Fig. 10). There are five MAPK defined as ERK (MAPKerk1–5) (52 ,274 ,308). However, amino acid sequence comparisons indicate that these proteins belong to different subfamilies of MAPK (see Fig. 2). There is actually more divergence between the MAPKerk1/MAPKerk2 subfamily and the MAPKerk3/MAPKerk4 subfamily than between the MAPKerk1/MAPKerk2 subfamily and any other MAPK. Of the collective group of MAPK referred to as ERK, MAPKerk1 and MAPKerk2 are the most extensively studied. MAPKerk1 and MAPKerk2 are 44- and 42-kDa isoforms, respectively. MAPKerk1/2 is activated by many different inputs to the cell that lead to the activation of several transcription factors and other serine threonine kinases contributing to cellular proliferation, differentiation, cell cycle regulation, and cell survival. The other ERK are less well characterized. MAPKerk3 is localized in the nucleus and is activated by PKC isoforms (52 ,302). MAPKerk4 has been shown to be activated in response to nerve growth factor (NGF) and EGF via a Ras-dependent pathway (274). The MAPK pathway employing MAPKerk3 and MAPKerk4 remains to be characterized. MAPKerk5 is discussed in section v D.

1. MAPKerk1/2

For further discussion in this section, we focus on the regulation of MAPKerk1 and MAPKerk2, which will be collectively referred to as MAPKerk1/2. Upon activation, MAPKerk1/2 phosphorylate substrate proteins on serine or threonine residues within a proline-directed motif. Pro-Leu-Ser/Thr-Pro is the most stringent consensus sequence for substrate recognition by MAPKerk1/2 (42). Several cytoplasmic proteins have been shown to be substrates for MAPKerk1/2. Proteins known to be phosphorylated by MAPKerk1/2 include the S6 kinase p90rsk, cytosolic phospholipase A2 , and the juxtamembrane region of the EGF receptor (210 ,308 ,366). Several microtubule-associated proteins (MAP) are also substrates for MAPKerk1/2, including MAP-1, MAP-2, MAP-4, and Tau (308).

MAPKerk1/2 phosphorylation of p90rsk at threonine-79 and threonine-396 contributes to activation of the kinase (261 ,308). One consequence of p90rsk activation is its translocation to the nucleus, where it phosphorylates c-Fos (308). p90rsk-induced phosphorylation of c-Fos occurs at serine-362 in the COOH-terminal transrepression domain. p90rsk also phosphorylates glycogen synthase kinase 3 (GSK3) at serine-9 near the NH2 terminus, resulting in the inhibition of GSK3 kinase activity. Glycogen synthase kinase 3 has been shown to negatively regulate c-Jun. Thus stimulation of p90rsk by MAPKerk1/2 results in the regulation of both c-Fos and c-Jun. The transient inactivation of GSK3 has been proposed to enable the rapid activation of c-Jun and possibly AP-1 activity (82 ,308). The regulation of additional kinases by MAPKerk1/2 can, therefore, result in amplification of the signaling downstream of MAPKerk1/2.

The regulation of cytosolic phospholipase A2 also generates active signaling molecules for the regulation of cellular physiology. Cytosolic phospholipase A2 is phosphorylated at serine-505 by MAPKerk1/2 (119). This phosphorylation event activates the enzyme that catalyzes the release of arachidonic acid. This is the rate-limiting step in the biosynthesis of eicosanoids (i.e., prostaglandins, leukotrienes). These compounds are important regulators of many physiological responses in cells (210 ,366).

MAPKerk1/2 is also able to phopshorylate the EGF receptor, the Ras exchange factor Sos, MKKKraf1, and MKKmek1. The phosphorylation of each of these proteins by MAPKerk1/2 is believed to reduce their catalytic activity. Each of these proteins is involved in a signal pathway involving MAPKerk1/2, and their phosphorylation would provide a feedback mechanism for controlling the activity of upstream regulators of the MAPKerk1/2 pathway (366).

In addition to phosphorylating cytoplasmic proteins, activated MAPKerk1/2 is translocated to the nucleus and phosphorylates several different transcription factors. Transcription factors phosphorylated and activated by MAPKerk1/2 include Elk1, Ets1, Sap1a, c-Myc, Tal, and signal transducer and activator of transcription (STAT) proteins; Myb activity is believed to be inhibited by MAPKerk1/2 (159 ,203 ,308 ,366 ,375).

2. MKKK in the MAPKerk1/2 pathway

Ligation of many receptors leads to the activation of MAPKerk1/2 through the activation of Ras. This was inferred from the observation that Ras activates MKKKraf1 and that activated MKKKraf1 is sufficient to stimulate the MAPKerk1/2 signaling pathway (225 ,319). Ras in the GTP-bound form binds to the NH2-terminal regulatory domain of MKKKraf1 (179). The NH2-terminal regulatory domain, by itself, of MKKKraf1 has high affinity for Ras⋅GTP and functions as an effective dominant negative mutant. Although MKKKraf1 interacts with GTP-bound Ras, it is unclear whether the Ras⋅GTP interaction with MKKKraf1 is sufficient to activate MKKKraf1 (400). One function of Ras⋅GTP is to localize MKKKraf1 to the plasma membrane where it becomes activated (80 ,225). Support for this hypothesis was demonstrated by the addition of the Ras prenylation sequence (CAAX box) to the COOH terminus of MKKKraf1 (131). The prenylated plasma membrane targeted MKKKraf1-CAAX chimera has seven times greater kinase activity at similar levels of expression relative to wild-type MKKKraf1 (201). MKKKraf1-CAAX is constitutively activated and is oncogenic in fibroblast transformation assays. Once MKKKraf1 is at the plasma membrane, it becomes tyrosine phosphorylated on Tyr-340 and Tyr-341 by membrane-bound tyrosine kinases including c-Src (223 ,225). Treatment of M K K K r a f 1  w i t h  t y r o s i n e  p h o s p h a t a s e s  i n a c t i v a t e s MKKKraf1, further supporting the importance of tyrosine phosphorylation for MKKKraf1 activation (72). However, the role that tyrosine phosphorylation plays in MKKKraf1 activation remains controversial, since mutation of the tyrosines on MKKKraf1 reduces but does not eliminate its kinase activity upon activation of Ras (223 ,225). A closely related homolog of MKKKraf1, MKKKB-raf does not have the tyrosines at positions 340 and 341 but is activated by Ras, suggesting that tyrosine phosphorylation is not obligatory for activation of all MKKKraf family members (225). In addition to tyrosine phosphorylation, MKKKraf1 is phosphorylated on serine residues 43, 259, 499, and 621 and is phosphorylated on threonine residue 268 (180 ,225). Because treatment of cultured cells with phorbol esters leads to MKKKraf1 activation and a strong MAPKerk1/2 activation (144), it is reasonable to suspect that PKC isoforms that are activated directly by phorbol esters phosphorylate MKKKraf1 leading to its activation. Indeed, coexpression of PKC-β with MKKKraf1 stimulates MKKKraf1 activation in a phorbol ester-dependent manner (227). Furthermore, PKC can also phosphorylate Ser-259 and Ser-499 of MKKKraf1 in vitro (180), suggesting that tyrosine and serine phosphorylation of MKKKraf1 controls its kinase activity when it is at the plasma membrane.

Recently, it was shown that MKKKraf1 must be associated with Ras⋅GTP for its activation by PKC (222). Phorbol esters stimulate GTP binding to Ras, but the activation of MKKKraf1 by PKC was not blocked by dominant negative Ras (N17Ras). N17Ras, which interacts with the exchange factor Sos, will block the activation of MKKKraf1 by receptor tyrosine kinases, suggesting that PKC activates Ras by a mechanism involving a Ras exchange factor different from the one used by tyrosine kinases.

Proteins in addition to Ras bind and regulate MKKKraf1 activity. The 14–3-3 proteins were initially discovered to interact with MKKKraf1 in two-hybrid screening (101 ,103). The 14–3-3 proteins have a variety of biological properties and recently have been shown to be involved in cell cycle control in both yeast and mammalian cells (5 ,94 ,135 ,225 ,291). The 14–3-3 proteins are dimers that may function as scaffolds or anchors to localize signaling proteins including protein kinases such as MKKKraf1. The 14–3-3 proteins also bind to motifs in proteins that frequently include phosphoserine and phosphothreonine residues (255 ,386). Binding of 14–3-3 protein to phosphoserine/threonine motifs in proteins has been demonstrated to protect these sites from dephosphorylation by phosphatases (56). Association of 14–3-3 proteins with MKKKraf1 may prevent MKKKraf1 inactivation by dephosphorylation and prolong its activation (225). Both 14–3-3β and 14–3-3ε associate with the NH2-terminal regulatory domain of MKKKraf1 (99). Mutation of Cys-162, Cys-168 in the cysteine-rich domain or Ser-259 of MKKKraf1 prevents the association of 14–3-3 isoforms with MKKKraf1. Mutation of Cys-162 or Cys-168 to serines has also been reported to block the activation of MKKKraf1 by oncogenic Ras (225), but this finding has not been repeated by other investigators (95 ,234). Despite the clear interaction of 14–3-3 proteins with MKKraf1, it is still not apparent what the exact function of this interaction is with regard to regulation. Different investigators have found that 14–3-3 binding to MKKKraf1 enhances, suppresses, or has no effect on MKKKraf1 kinase activity (95 ,101 ,153 ,234). It seems most likely that 14–3-3 proteins will be found to function primarily in organizing signal transduction systems such as upstream regulators of MAPK modules by controlling the proximity of these proteins with different MKKK (94).

MKKKraf1 can also be phosphorylated by MAPKerk1/2 and PKA (cAMP-dependent protein kinase) (225 ,240 ,308 ,348). These phosphorylations inhibit MKKKraf1 activity. The MAPKerk1/2-mediated inhibition of MKKKraf1 may be a mechanism to limit the extent of the activation of the pathway via a classical feedback inhibition mechanism. The inhibition of MKKKraf1 signaling by PKA is a mechanism whereby receptor systems that regulate cAMP synthesis can negatively regulate the MAPKerk1/2 pathway, allowing integration and control of these pathways by multiple inputs, both positive and negative.

Recently, MKKmek1, the downstream effector of MKKKraf1 in the MAPKerk1/2 pathway, has been shown to increase the activity of MKKKraf1 in a Ras- and Src-independent manner (408). The signal transmitted through the MAPKerk1/2 pathway may thus be amplified when it reaches the MKKmek1 level, but it is negatively regulated when MAPKerk1/2 is activated. This type of dual regulation may be important to determine the duration and strength of the MAPKerk1/2 response.

There are two additional MKKKraf members, MKKKA-raf and MKKKB-raf, whose pattern of expression is more restricted than that of MKKKraf1 (324). The large 96-kDa form of MKKKB-raf is expressed in many neuronal and neuroendocrine cell types and appears to be the major MKKmek activator in brain (43 ,388). A smaller 68-kDa splice variant is expressed in fibroblasts and many other cell types. MKKKB-raf is also activated in NGF- and EGF-stimulated PC-12 cells (158). MKKKA-raf activates MKKmek in cardiac myocytes after endothelin-1 stimulation (22). However, although MKKKB-raf is strongly activated by oncogenic Ras alone, maximal activation of both MKKKraf-1 and MKKKA-raf requires additional inputs (224). Such inputs could involve upstream kinases such as Src, PKC, other MKKKK, or additional protein interactions such as with 14–3-3 proteins. A glimpse of such differential regulation has recently been discovered with MKKKB-raf. The 96-kDa MKKKB-raf protein can be activated by a second GTP-binding protein, Rap1. Rap1 is phosphorylated near its COOH terminus by PKA (204). Treament of cells with cAMP leads to phosphorylation of Rap1, which enhances exchange of GDP for GTP. Rap1⋅GTP binds and activates MKKKB-raf in a manner similar to Ras activation of MKKKraf1 (348). Rap1 does not activate MKKKraf1. The 68-kDa form of MKKKB-raf neither interacts with nor is regulated by Rap1. Rap1 appears to be expressed in most if not all cells. Thus, in cells expressing the 96-kDa form of MKKKB-raf, cAMP activation of PKA will stimulate the MKKKB-raf/MKKmek/MAPKerk1/2 module while inhibiting MKKKraf-1.

3. MKK in the MAPKerk1/2 pathway

MKKmek1 and MKKmek2 are the MKK in the three-kinase component MAPKerk1/2 activation modules (225 ,366). MKKmek1 and MKKmek2 are highly homologous in primary amino acid sequence and function as dual threonine/tyrosine kinases. MAPKerk1/2 have a Thr-Glu-Tyr sequence in the activation loop of the kinase catalytic domain that is phosphorylated by activated MKKmek1/2 (366). MKKmek1 contains a nuclear export signal that excludes it from the nucleus (104). It has been proposed that MKKmek1 in an inactive state sequesters MAPKerk1/2 in the cytoplasm. Activation of the MAPKerk1/2 module results in MKKmek1 phosphorylation and activation of MAPKerk1/2. The activation of MAPKerk1/2 results in its dissociation from MKKmek1. Activated MAPKerk1/2 is rapidly translocated to the nucleus where it is functionally sequestered and can regulate the activity of nuclear proteins including transcription factors. The inactivation of MAPKerk1/2 requires its dephosphorylation by specific phosphatases (see below). Nuclear MAPKerk1/2 when dephosporylated returns to the cytoplasm and reassociates with MKKmek1.

Transfection analysis and biochemical reconstitution in vitro indicates that MKKKraf1 can activate both MKKmek1 and MKKmek2. Stimulation of cells through receptor tyrosine kinases, nonreceptor tyrosine kinases, and G protein-coupled receptors also leads to activation of both MKKmek1 and MKKmek2 (225 ,366). There are several conditions where MKKmek1 and MKKmek2 appear to be differentially regulated. When recombinant Ras⋅GTP is used to bind effectors from cell lysates, a complex of MKKKraf1 and MKKmek1 or MKKKB-raf and MKKmek1 is found. MKKmek2 has not been isolated in these complexes. Also, cells transformed by oncogenic forms of Ras display increased MKKmek1 activity compared with MKKmek2, suggesting that Ras and MKKKraf1 preferentially signal to MAPKerk1/2 via MKKmek1 (161 ,225). In response to phorbol esters or endothelin-1 treatment of cardiomyocytes, MKKKA-raf activated MKKmek1 (22). MKKKB-raf activated MKKmek1 in PC-12 cells in response to NGF, EGF, and PDGF. Serum stimulation of NIH 3T3 cells stimulates MKKmek1 (190 ,366), and MKKKA-raf selectively activates MKKmek1 compared with MKKmek2 in response to EGF treatment of HeLa cells (225 ,366). Thus MKKmek1 and MKKmek2 appear to be differentially activated by MKKKraf1, MKKKA-raf, and MKKB-raf in different cell types and in response to different extracellular stimuli. Stimuli that preferentially activate MKKmek2 are still to be identified.

There are several reports in the literature that demonstrate MAPKerk1/2 activation when MKKKraf activation could not be demonstrated. For example, EGF treatment of Swiss 3T3 cells, insulin stimulation of adipocytes, and tumor necrosis factor-α (TNF-α) challenge of macrophages activated MKKmek1 and MAPKerk1/2 but not MKKKraf1 (133 ,225 ,374 ,404). Because in mouse macrophages neither MKKKB-raf nor MKKKA-raf was expressed, these results suggest that a MKKK other than MKKKraf was involved in activating the MAPKerk1/2 pathway. Reported activators of MKKmek1/2 other than MKKKraf subfamily members include M K K K m o s,  M K K K t p l 2,  M K K K m e k k 1,  M K K K m e k k 2,  a n d MKKKmekk3 (92) (Fig. 10). A COOH-terminal truncation of MKKKtpl2 referred to as MKKKcot, MKKKmekk1, MKKKmekk2, and MKKKmekk3 is also capable of activating the MAPKjnk signaling pathway (92). This indicates that some MKKmek1 activators can feed into the MAPKjnk pathway by activation of MKKsek1. This finding is reminiscent of budding yeast, where specific MKKK like MKKKste11 can function in more than one MAPK module. Both MKKKmos and MKKKtpl2 have restricted patterns of expression; MKKKmos is expressed in germ line cells where it is a key regulator of meiosis and has been found expressed in some cervical cancer cell lines (92 ,366), and MKKKtpl2 is mainly expressed in lymphoid and hemopoietic cells (92). Thus MKKKmos and MKKKtpl2 may serve as tissue-specific activators of MKKmek1 or MKKmek2. MKKKmekk1, 2 and 3 are expressed in a wide variety of tissues and have been shown to activate the MAPKerk1/2 pathway in transfection assays. Only kinase-inactive MKKKmekk1 and not MKKKmekk2/3 has been shown to inhibit receptor stimulation of MAPKerk1/2 (93). MKKKmekk1 has also been shown to bind to Ras⋅GTP (296), suggesting that it has the necessary regulatory functions to be an MKKK in the MAPKerk1/2 module. In addition to Ras association, MKKKmekk1 also binds Rac and Cdc42, which may direct its activity toward the MAPKjnk signaling pathway (93). The implication of these findings is that several different MKKK may function in mammaliam MAPKerk1/2 signaling modules. The prediction is that upstream inputs involving MKKKK, GTP-binding proteins, or other signals that are differentially regulated by extracellular stimuli could regulate different MAPKerk1/2 modules, allowing discrimination and integration of signals. MKKKmekk1 encodes a predicted pleckstrin homology domain that may bind phosphatidylinositol 3,4,5-trisphosphate or phosphatidylinositol 4,5-bisphosphate, reaction products of the phosphatidylinositol 3-kinase (PI3K)catalyzed reaction. A difference, therefore, between MKKKmekk1 and MKKKraf1 could be the activation of MKKKmekk1 by PI3K reaction products.

Reversal of MAPKerk activation involves dephosphorylation of the Thr(P)-Glu-Tyr(P) activation motif by protein kinase phosphatases. Protein kinase phosphatases that can inactivate MAPK include MKP-1 (also known as CL100, 3CH134, Erp, and hVH-1), MKP-2 (hVH-2, Typ-1), MKP-3 (rVH6, Pyst1), MKP-4, PAC1, VHR, B23 (hVH-3), and M3/6 (hVH-5) (55 ,251253 ,366). Among these phosphatases, MKP-1, MKP-3, and MKP-4 are more selective in dephosphorylating and inactivating MAPKerk1/2 compared with other MAPK (100 ,122 ,252 ,253). Except M3/6 that is very specific for MAPKp38 and MAPKjnk, the other phosphatases do not seem to discriminate between different MAPK. MKP-3 and MKP-4 are primarily in the cytoplasm, whereas the other phosphatases are nuclear. This finding indicates a specific control of the MAPK by MKP-3 and MKP-4 compared with the other phosphatases based on their cellular location.

MAPKerk1/2 has many substrates with diverse functions in cells, and the question should be asked what the biological outcome is of MAPKerk1/2 activation? The unsatisfying answer is that the role of MAPKerk1/2 is probably as variable as there are differentiated cell types. In cell culture systems, there is a good correlation between MAPKerk1/2 activation and proliferation of cells. Growth factors such as EGF and PDGF that stimulate proliferation of cells also frequently give a strong and persistent stimulation of MAPKerk1/2 activity. Interfering with components of the MAPKerk1/2 signaling pathway with dominant negative mutants or antisense constructs for MKKKraf1 or MAPKerk1 shows significant inhibition of cell proliferation (238 ,272 ,308). Conversely, constitutively activated MKKmek1 persistently stimulates MAPKerk1 activity, resulting in enhanced cell proliferation (308).

MAPKerk2 activation may not always be required for proliferation. Interleukin-4 stimulation of B cells does not significantly activate MAPKerk signaling but still induces proliferation. Conversely, okadaic acid (a phosphatase inhibitor) treatment of B lymphocytes activates MAPKerk2 but inhibits rather than augments cellular proliferation (42 ,353), indicating that regulation of the phosphorylation status of other proteins by okadaic acid can have growth inhibitory effects even though MAPKerk2 is activated. In smooth muscle cells, activation of MAPKerk1/2 leads to PGE2 secretion that inhibits cellular proliferation. However, in smooth muscle cells lacking the inducible form of cyclooxygenase, activation of MAPKerk1/2 does not lead to secretion of PGE2 , and the cells proliferate (30). Therefore, the MAPKerk1/2 signaling pathway can mediate either proliferation or growth inhibition depending on the repertoire of genes that are regulated in a specific cell type. Consistent with this theme is the ability of constitutively active MKKmek1 or MAPKerk2 to induce differentiation in some cell types including PC-12 pheochromocytoma (282) and K562 erythroleukemia cells (283). Treatment of PC-12 cells with either EGF or NGF activates the MAPKerk1/2 signaling pathway, yet EGF stimulates proliferation while NGF induces growth arrest and differentiation (282). This apparent difference in the outcome between EGF and NGF treatment has been postulated to be related to differences in the duration and magnitude of the MAPKerk1/2 activation. Epidermal growth factor-stimulated MAPKerk1/2 activity in PC-12 cells is transient, returning to basal or near-basal levels within 1–2 h. In contrast, the NGF-stimulated activity is more sustained. Overexpression of the EGF receptor in PC-12 cells changes the response to EGF from proliferation to differentiation, a response that is correlated with prolonged MAPKerk1/2 activation (76). Taken together, these results indicate that the strength and duration of the signals transmitted through MAPKerk1/2 will contribute to determine whether a cell proliferates or differentiates in response to a specific stimulus.

MAPKerk1/2 may also contribute to cell cycle regulation in some cell types. In Chinese hamster ovary cells, MAPKerk1/2 is activated in the G1 and M phase of the cell cycle (332). MAPKerk1/2 is associated with the microtubule-organizing center (MTOC) during M phase of the cell cycle, suggesting an involvement of MAPKerk1/2 in controlling MTOC function (347). Furthermore, MAPKerk1/2 function in the spindle assembly checkpoint in Xenopus egg extracts (239) (see sect. vi D). The substrates for MAPKerk1/2 in cell cycle control are not defined. Persistent activation of MAPKerk1/2 has also been shown to be required to pass the G1 restriction point in fibroblasts (194). It was shown that MAPKerk1/2 activation increases cyclin D1 promoter activity and cyclin D1 protein expression (193). These results demonstrate that MAPKerk1/2 can modulate cyclin D1 expression and its associated cyclin-dependent kinase activities for the control of G1 progression.

Activation of MAPKerk1/2 may also provide protection against apoptosis in some cell types. In PC-12 cells, the withdrawal of NGF led to the inhibition of MAPKerk1/2 activity and cell death (383). Constitutive activation of the MAPKerk1/2 pathway in these cells inhibited apotosis. In L929 cells, blockage of MAPKerk1/2 activation prevents the protection against TNF-α-induced apoptosis that is mediated by fibroblast growth factor (FGF)-2 (108). In Jurkat cells, the ERK pathway is activated when Fas is stimulated (117 ,368); however, the ERK activation is transient, possibly as a consequence of MKKKraf1 degradation in apoptotic cells (368). It is believed that the function of the inactivation of the MAPKerk1/2 response pathway in Fas-stimulated cells is to prevent MAPKerk1/2-mediated protection against the apoptotic response. Support for the hypothesis that the MAPKerk1/2 pathway components are involved in survival signaling comes from MKKKB-raf knock-out mice. These mice die of vascular defects during mid-gestation and show increased numbers of endothelial precursors cells, enlarged blood vessels and apoptotis of differentiated endothelial cells (376). Thus MKKKB-raf appears critical as a signaling element in the development of the vascular system. This appears to be due, at least in part, to its ability to protect maturing endothelial cells from apoptosis. In some cases, however, the MAPKerk1/2 pathway may be positively involved in cell death as indicated by the observation that inhibition of MKKmek1 activity inhibits crocidolite asbestos (a carcinogen)-induced apoptosis (164) and Fas-induced apoptosis (117). How inhibition of MKKmek1 activity would suppress apoptosis is unclear from these studies.

4. Receptor activation of the MAPKerk1/2 pathway

Many different receptor types are able to activate the MAPKerk1/2 pathway (Fig. 11). Upon stimulation of receptor tyrosine kinases such as the EGF receptor, PDGF receptor, or insulin receptor, their intrinsic tyrosine kinase domains are activated, leading to tyrosine phosphorylation of specific substrates including themselves (225 ,308 ,366). Tyrosine phosphorylation of the receptor allows the binding of adapter proteins to the receptor. Adapter proteins within their amino acid sequence have specific motifs that are involved in protein-protein interactions. The adapter protein Shc consists of a phosphotyrosine binding (PTB) domain, Src homology 2 (SH2) domain [which, as PTB, also binds phosphotyrosine residues], and a Src homology 3 (SH3) domain. After activation of the EGF receptor, Shc binds to a specific phosphotyrosine of the receptor via its PTB domain (80 ,225). The association of Shc with the EGF receptor permits the tyrosine phosphorylation of Shc by the receptor itself or intracellular tyrosine kinases such as Src (15). This phosphorylation allows the binding of another adapter protein, Grb2, consisting of a SH2 domain and two SH3 domains (80 ,301). The association between Shc and Grb2 is mediated by the SH2 domain of Grb2. Grb2 can also bind directly to the EGF receptor through its SH2 domain (15). These associations amplify the amount of Grb2 associated with the receptor. Because the guanine nucleotide exchange factor Sos (son of sevenless) binds constitutively to the Grb2 SH3 domain, the binding of Grb2 with the EGF receptor also recruits Sos (80 ,301). The localization of Sos to the EGFR leads to the exchange of Ras⋅GDP for GTP at the plasma membrane (80). Ras⋅GTP interacts with MKKKraf1 or other MKKK fitting into the MAPKerk1/2 module and leads to the activation of the MAPKerk1/2 pathway. The activation of the MAPKerk1/2 signaling pathway by other receptor tyrosine kinases involves similar recruitment of adapter proteins leading to Ras activation.

Fig. 11.

Receptor–mediated activation of MAPKerk pathway. See text for details. JAK, Janus kinase; STAT, signal transducer and activator of transcription; PLCβ, phospholipase C-β; PI-3K, phosphatidylinositol 3-kinase.

A second group of receptors that activate MAPKerk1/2 includes specific cytokine receptors, the T-cell receptor, CD28, and the B-cell receptor. Upon ligation of these receptors, there is a rapid and transient increase in tyrosine phosphorylated proteins. Because these receptors lack an intrinsic tyrosine kinase domain, tyrosine phosphorylation of target proteins is accomplished by the activation of Src family tyrosine kinases including Lck, Lyn, Fyn, and others (46 ,333). Similar to receptor tyrosine kinases, the receptor itself becomes tyrosine phosphorylated, enabling the recruitment of the adapter proteins Shc and Grb2 to the cytoplasmic surface of the receptor effectively recruiting Sos to the receptor. These events lead to GTP loading of Ras and MAPKerk1/2 activation (46).

Several cytokine receptors activate the MAPKerk1/2 pathway through the activation of JAK (Janus kinases including JAK1, -2, -3, and Tyk2). In addition to its ability to phosphorylate and activate STAT, JAK1 can phosphorylate Shc on tyrosines leading to activation of the MAPKerk1/2 pathway (375). In turn, MAPKerks may phosphorylate and potentiate the activity of the STAT (375). In human multiple myeloma cells, interleukin (IL)-6 triggers cell growth through the MAPKerk1/2 pathway in a JAK- and Ras-dependent manner (267). However, JAK-mediated MAPKerk1/2 activation can occur independently of Ras activation. For example, interferon-β induces MKKKraf1 activation in HeLa cells in a Ras-independent manner, whereas oncostatin M stimulation of MKKKraf1 correlated with GTP loading of Ras (321).

G-protein coupled receptors (GPCR) can also engage the MAPKerk1/2 pathway. With many Gi-coupled receptors, treatment of cells with pertussis toxin inhibits MAPKerk activation. When Gi-coupled receptors are involved, MAPKerk1/2 activation is believed to be primarily mediated by the βγ-subunit complex. One mechanism for βγ-stimulated MAPKerk activity appears to be PI3Kγ dependent (213). This pathway was shown to be specific for PI3Kγ, since PI3Kα could not be used as a substitute for PI3Kγ (213). Signaling from PI3Kγ to the MAPKerk pathway required a tyrosine kinase, Shc, Grb2, Sos, Ras, and MKKKraf1 (213), indicating that the GPCR that activate the MAPKerk pathway in a PI3Kγ-dependent manner can transactivate receptor tyrosine kinases (Fig. 11). Some GPCR activate the MAPKerk pathway in a pertussis toxin-insensitive manner. This activation mechanism involves the stimulation of Gαq/11 proteins and the activation of phospholipase C-β. The stimulation of intracellular calcium activates the proline-rich tyrosine kinase (Pyk2) (205). Pyk2 activates Ras through the tyrosine phosphorylation of Shc and/or by recruitment of the Grb2/Sos complex similar to receptor tyrosine kinases (327). In response to endothelin-1, lysophosphatidic acid, or thrombin, the EGF receptor is rapidly tyrosine-phosphorylated, leading to the recruitment of adaptor proteins such as Grb2 (69 ,327). This recruitment eventually leads to MAPKerk1/2 activation as described above. Inhibition of the EGF receptor function suppressed MAPKerk1/2 activation by the GPCR (69). It has also been suggested that the GPCR-induced tyrosine phosphorylation of EGF receptor is mediated by a Src tyrosine kinase (215).

In lymphoid cells, targeted deletion of Lyn or Csk tyrosine kinases blocks the stimulation of MAPKerk1/2 by Gq- but not Gi-coupled receptors (350 ,351). In cells deficient for Btk tyrosine kinase, Gi-coupled receptors failed to activate ERK, whereas Gq-coupled receptor-mediated stimulation was unaffected (350). Cells lacking Syk were deficient in both Gq- and Gi-mediated MAPKerk stimulation (350 ,351). Syk appears to integrate the signals stimulated by different GPCR, leading to the activation of the MAPKerk1/2 pathway (Fig. 11). In addition, there is evidence that some GPCR activate MAPKerk1/2 in a PKC-dependent, Ras-independent manner (366).

Finally, the MAPKerk pathway can be activated by engagement and clustering of integrins. Integrins are a family of transmembrane receptors that bind to proteins of the extracellular matrix, such as fibronectin, collagen, and vitronectin. Binding and clustering of integrins leads to the formation of focal adhesion structures, in which integrins connect to actin stress fibers. Integrin ligation induces the activation of a variety of signaling events, including the activation of the MAPKerk1/2 pathway (50 ,242 ,249 ,407). The activation of the MAPKerk1/2 after integrin ligation is mediated in part by Rho (290). Cell adhesion is also important for survival of adherent cells. The regulation of the MAPKerk1/2 pathway by integrins may contribute to the survival response observed with integrin-mediated adherence.

B. MAPKjnk Pathway

A new MAPK was identified biochemically in 1991 (281) that was distinguished from the MAPKerks by two characteristics: 1) it was activated by cell stress such as ultraviolet irradiation, and 2) it phosphorylated c-Jun at the NH2-terminal activating sites rather than the COOH-terminal inhibitory sites phosphorylated by the MAPKerk2. When the gene encoding this kinase was cloned by two groups, the human homolog was named c-Jun NH2-terminal kinase (MAPKjnk) (73 ,188), and the rat homolog was named stress-activated protein kinase (SAPK) (188). Two more genes encoding MAPKjnk family members have also been cloned (123 ,169 ,188). These members of the MAPKjnk group of MAPK, along with their MKKK and MKK, form the MAPKjnk module (Fig. 12).

Fig. 12.

MKKK, MKK, and MAPK that can be components of MAPKjnk pathway.

Differential splicing and exon usage yield a total of 10 different MAPKjnk isoforms from the 3 genes (123). Each kinase is expressed as a short form and a long form, and MAPKjnk1 and MAPKjnk2 have an alternative sequence that appears in the kinase domain in some transcripts. A definitive analysis of the expression patterns of these 10 different isoforms in vivo has not been done. MAPKjnk1 and MAPKjnk2 are expressed ubiquitously, whereas the expression of MAPKjnk3 appears to be limited to the brain (390). The alternative forms of the MAPKjnks differ in their ability to bind, and presumably activate, different transcription factors (123). There is also evidence that the different forms of the enzyme can be differently regulated. In mouse macrophages stimulated with TNF-α, a 46-kDa MAPKjnk is activated, whereas a 54-kDa isoform is not (47). Because of extensive sequence homology between these paralogues, biochemical identification using antibodies is very difficult, if at all possible. Whether these measurable differences reflect a real effect on the determination of cell fate remains to be determined.

MAPKjnk was first described as a SAPK, and it is the response to stress that has been most widely studied. In spite of the volume of research in this area, little is known about the signal transduction pathways leading from cell stress, such as ultraviolet irradiation, to the activation of the MAPKjnk cascade. Cell stress can come in several forms, including heat shock, direct damage of DNA (γ-irradiation, cytosine arabinoside), generation of reactive oxygen species (H2O2), and conditions of hyperosmolarity. Until the exact sensors of cell stress are identified and characterized, the pathways that lead from stress to activation of the cascade will remain elusive. Cellular redox state, tyrosine kinases, and phosphatases are potentially involved in some stress responses, but the mechanisms for regulation have not yet been defined.

The MAPKjnk response to extracellular ligands is far better characterized. MAPKjnks can be activated through different receptor types, and the signal transduction pathways that can converge on the MAPKjnk cascade vary widely. MAPKjnks have been shown to be activated through cell surface receptors from a variety of families, including the TNF receptor family, GPCR, tyrosine kinase receptors, and cytokine receptors. The growing list of signaling proteins that are capable of activating the MAPKjnk pathway has been recently reviewed (92).

MAPKjnks are activated by phosphorylation on threonine and tyrosine of the Thr-X-Tyr activation motif by one of two cloned dual specificity kinases, MKKmkk4 (300) and MKKmkk7 (340). These kinases are in turn activated by an MKKK, of which several examples have been identified (Fig. 12); MKKKmekk1–4 (20 ,111 ,191), MKKKask1 (357), MKKKtak1 (387), MKKKmst (138), MKKKsprk (286), MKKKmuk (137), and MKKKtpl2 (298) have been shown to phosphorylate MKKmkk4 and activate its kinase activity and thus are designated as MKKK for the MAPKjnk pathway.

Other kinases are capable of activating MAPKjnks when they are overexpressed in cells but have not been demonstrated to act as MKKK. In yeast, Ste20 is often placed directly upstream of the MKKKmekk1 homolog MKKKste11 (note again that this has yet to be proven genetically and biochemically). Therefore, Ste20 homologs, such as p21-activated kinase, germinal center kinase (GCK), Nck-interacting kinase (NIK), hematopoietic progenitor kinase 1 (HPK1), and GCK-like kinase (GLK) may regulate MAPKjnk by phosphorylation of specific MKKK. However, despite a great deal of interest in these kinases and the MAPKjnk pathway, there has never been a direct demonstration of the activation of a MKKK by a Ste20 homolog. Although these kinases may activate MAPKjnk when overexpressed in cells, their regulation of different MKKK has not been demonstrated biochemically or genetically. This is the case for NIK and HPK1 that, although capable of binding MKKKmekk1 and MKKKsprk, respectively (175 ,326), have not been shown to directly activate MKKKmekk1 and MKKKsprk.

Other signaling proteins that act as upstream activators of the MAPKjnk module include the low-molecular-weight GTP-binding proteins of the Rho family. These proteins, in particular Rac and Cdc42, have been demonstrated to activate MAPKjnk when expressed as constitutively active forms. The competitively inhibitory mutant forms of Rac and Cdc42 can block MAPKjnk activation induced by EGF or TNF-α (62). They may do so by binding to a MKKK (93 ,111) or to one of the Ste20 homologs discussed above (11).

MAPKjnk activities can be downregulated by dual-specificity protein phosphatases, including M3/6 (also known as hVH-5) (253) and MKP-1 (also known as CL100, 3CH134, Erp, and hVH-1) (100 ,140). These phosphatases display selectivity toward MAPKjnk family members. MKP-1 has been shown to be induced by MAPK pathways. However, there are conflicting results as to which MAPK pathway is responsible for this induction. In U 937 cells, MAPKerk2, but not MAPKp38α or MAPKjnk1, mediates the induction of MKP-1 expression (100). In contrast, in NIH 3T3 fibroblasts, the MAPKjnk pathway, but not the MAPKerk pathway, induces MKP-1 induction (25). It is possible that this difference is because of the different cell lines used, but further work is required to clarify this issue. The upstream regulation of M3/6 has not been demonstrated, so it is not known whether the phosphatase is a selective point of MAPKjnk regulation or a constitutively active switch that turns off MAPKjnk activity when the upstream activators are silenced.

The expression of another protein, MAPKjnk interacting protein-1 (JIP-1) causes a decrease in the MAPKjnk activation of transcription factors (75), suggesting that JIP-1 acts as an inhibitor of MAPKjnk. Several other biochemical and biological functions of MAPKjnk are inhibited by a fragment of JIP-1 (the MAPKjnk binding domain, or JBD), including activation of c-Jun and Elk-1, transformation of cells by break point cluster region-Abelson proto-oncogene (Bcr-Abl), and apoptosis induced by NGF withdrawal. However, it is not obvious that the inhibitory functions of the fragment are physiological functions of the full-length protein or merely a consequence of a loss of a functional domain of the protein. The full-length JIP-1 acts as a MAPKjnk substrate and may have downstream activities of its own, which are blocked by overexpression of the MAPKjnk binding domain. Recently, the rat homolog of JIP-1 (IB-1) has been cloned (27). Compared with the mouse JIP-1 sequence, IB-1 encodes a 47-amino acid insertion in a phosphotyrosine interaction domain. Polymerase chain reaction analysis of mouse and rat tissues has only revealed the sequences corresponding to IB-1 (that is the sequence with 47-amino acid insertion) (G. Waeber, personal communication), raising the possibility that JIP-1 is a truncated version of the natural protein.

The MAPKjnk substrates are, to date, exclusively transcription factors, in contrast to the MAPKerk family that appears to have substrates outside the nucleus. The substrates that have been identified for MAPKjnks include c-Jun, ATF-2 (124), Elk-1 (409), p53 (145 ,237), DPC4 (3), and NFAT4 (54). Phosphorylation of c-Jun at serine-63 and serine-73 by MAPKjnk results in an increase in the formation of Jun/Jun homodimers and Jun/ATF2 heterodimers. c-Jun phosphorylated by the MAPKjnks is also more resistant to ubiquitin-dependent degradation (256). Thus the MAPKjnks have the abilility to activate transcription factors as well as stabilizing them, resulting in efficient activation of the target genes controlled by the MAPKjnk-regulated transcription factors. Phosporylation of ATF-2 by MAPKjnk also leads to an increase in transcriptional activity. The transcription factor Elk-1 represents a point of convergence for the MAPKerk and MAPKjnk pathways (409). Phosphorylation of NFAT4 by MAPKjnk inhibits its function by preventing its translocation to the nucleus (54). Mutation of the serines that MAPKjnk phosphorylates leads to a NFAT4 mutant that is constitutively active and located in the nucleus (54). The inhibitory function of MAPKjnk on NFAT4 can be opposed by the phosphatase calcineurin (54). DPC4 is a human mothers against decapentaplegic (Mad)-related transcriptional factor regulated by transforming growth factor-β (TGF-β) in a MAPKjnk pathway-dependent manner (3). Overexpression of DPC4 leads to apoptosis, and this response can be potentiated by mad3, another transcription factor that associates with and is phosphorylated by the TGF-β receptor (401). This suggests that mad3 links the TGF-β receptor to DPC4 activation and possibly apoptosis. Finally, the transcription factor p53 is phosphorylated on serine-34 by all three isoforms of MAPKjnk (145). However, the functional significance of this phosphorylation is unclear.

The mechanism by which MAPKjnk recognizes these substrates involves a bipartite sequence. As with the other MAPK, MAPKjnk phosphorylates substrates at a Ser/Thr-X-Pro motif. However, this sequence is not sufficient for MAPKjnk to phosphorylate a protein, because JunB contains such a site and is not efficiently phosphorylated by MAPKjnk (123). An additional docking site that is present on c-Jun is necessary for phosphorylation of that protein. The recruitment of MAPKjnk to this docking site effectively increases the local concentration of the kinase and directs activity to the proper NH2-terminal phosphorylation motif on c-Jun. Because c-Jun has the ability to heterodimerize with other AP-1 components, the binding of MAPKjnk to this docking site allows the kinase to come into contact with transcription factors that lack a docking site of their own (169). It has been suggested that the differences between the activities of MAPKjnk1 and MAPKjnk2 are explained by their differing affinities for c-Jun. MAPKjnk2 shows a much higher affinity to c-Jun because of a specificity-determining region that MAPKjnk1 lacks (169). MAPKjnk also binds to ATF-2 at a site just NH2 terminal to the phosphorylation motif, and thus it interacts with this transcription factor in a bipartite manner as well (124).

MAPKjnk activity has been implicated in the response to cell stress, specifically apoptosis. Although it has never been demonstrated that a MAPKjnk is sufficient for apoptosis in any system, it seems necessary for this process to occur in at least some cases (77). Inhibition of MAPKjnk signaling by introduction of dominant inhibitory mutants of MAPKjnk1, its main downstream target c-Jun, or its major upstream activator MKKmkk4, show that MAPKjnk is necessary for apoptosis in the response to growth factor withdrawal (125 ,383), stress (51), DNA damage (170), and ligation of Fas on the cell surface (117). Depending on the cell type, Fas-induced apoptosis may or may not necessitate the activation of the MAPKjnk pathway (391). Whether the MAPKjnk pathway is required for Fas-induced apoptosis could depend on the cell's sensitivity to the death response and/or the presence of Daax, a Fas-binding protein that activates MAPKjnk (391).

When the MAPKjnk pathway is required for apoptosis, it is expected that c-Jun-dependent transcription leads to the de novo synthesis of pro-apoptotic proteins. Characterization of the pro-apoptotic proteins that are induced by MAPK pathways is clearly an important step in elucidating the role of these signaling events in apoptosis. Expression of Fas ligand (FasL) is one pathway of transcription-dependent apoptosis. In response to DNA damage, FasL is expressed at the surface of T cells where it activates Fas and consequently induces apoptosis. Deoxyribonucleic acid damage-induced FasL expression necessitates the activation of the AP1 transcription factors in a MAPKjnk-dependent manner (170). Expression of FasL establishes thus a direct link between activation of the MAPKjnk pathway and apoptosis. MKKKmekk1 appears to be the MKKK in the MAPKjnk module that is required for FasL expression (97 ,170). Interestingly, MKKKmekk1 has also the potential to activate the NFκB transcription factor (139 ,196). Because NFκB is also required for DNA damage-induced FasL expression (the FasL promoter contains AP-1 and NFκB binding sites) (170), it is possible that MKKKmekk1 integrates the stress signals that lead to activation of both AP1 and NFκB.

Yang et al. (390) demonstrated a specific role for MAPKjnk3 in excitotoxicity. In MAPKjnk3-deficient mice, administration of an epileptogenic dose of kainic acid did not cause seizures that were as severe as those in similarly treated wild-type mice. The onset of seizures coincided with cell death in an area of the hippocampus. No cell death occurred in this area in the MAPKjnk3 knockout mice. This is probably the best evidence for the role of a specific MAPKjnk in an apoptotic response. The function of MAPKjnk3 in this excitotoxicity is undefined. Despite these observations, other investigators have suggested that MAPKjnk are not involved in apoptotic responses (160 ,211). The discordance is mostly related to Fas- and TNF-α receptor-mediated apoptosis, where the receptors may regulate caspases by a pathway different from that of other cellular stresses.

MKKKmekk1, an upstream regulator of the MAPKjnk pathway, is able to regulate the apoptotic response possibly in a MAPKjnk-independent manner (192 ,367). In its full-length 196-kDa form, MKKKmekk1 does not seem to promote apoptosis (41 ,367). In fact, activation of full-length MKKKmekk1 may induce survival responses, including the activation of the MAPKerk and NFκB pathways (139 ,191 ,196). Both MAPKerk and NFκB pathways have been shown to have survival responses in different cell types (see sect. v A3 and Refs. 229 ,345 ,352). MKKKmekk1 is a caspase-3 substrate, and when cleaved into a 91-kDa kinase fragment during apoptosis, it becomes a strong apoptotic inducer (41 ,192 ,367 ,369). The cleavage of MKKKmekk1 could thus switch the function of MKKKmekk1 from a protective response to a cell death-promoting response.

Although MAPKjnk in several cell lines appears to mediate a stress response that is often associated with subsequent cell death, in some cases it may act to promote survival or growth. BAF3 pre-B cells undergo apoptosis when deprived of IL-3. Readdition of IL-3 stimulates a MAPKjnk response in these cells, and inhibition of MAPKjnk activity by expression of a MAPKjnk specific phosphatase inhibits IL-3 induced proliferation, while having no effect on apoptosis induced by IL-3 withdrawal (318). In T98G glioblastoma cells, the MAPKjnk pathway seems to regulate DNA repair, and its inhibition sensitizes the tumor cells to cisplatin-induced death (279).

In both T and B cells, presentation of an antigen to the antigen receptors [the T-cell receptor (TCR) and the B-cell receptor, or surface IgM] alone causes growth arrest and/or apoptosis. However, presentation of antigen by an antigen-presenting cell allows the coactivation of the antigen receptor and a second receptor, CD28 in the T cell and CD40 in the B cell, that rescues the cell and allows it to proliferate. MAPKjnk has been demonstrated to play a role in this signal integration. Ligation of the TCR or CD28 alone is unable to activate MAPKjnk or induce the synthesis of IL-2. Ligation of both receptors simultaneously activates MAPKjnk and IL-2 production (325). Blocking the MAPKjnk pathway with dominant negative MKKKmekk1 blocks IL-2 production in Jurkat cells in response to phorbol ester and calcium ionophore, which can substitute for CD28 ligation in these cells (96). Finally, in lymphocytes that are lacking MKKmkk4, CD28 and CD3 ligation (a similar treatment to TCR ligation) are not as capable of inducing IL-2 production (265). However, although these cells are completely lacking a MAPKjnk response to phorbol ester and calcium ionophore, they are still capable of proliferating and making IL-2, albeit at a lower rate. This suggests that some MAPKjnk-independent signal is operating in the proliferative response to these signals. What must be remembered is that none of the MAPK pathways including MAPKjnk is acting alone in these responses; rather, the MAPK pathways are integrated with many other metabolic changes in the cell. The sum of these responses and their integration ultimately determines cell fate.

C. MAPKp38 Pathway

The mammalian MAPKp38 family consists of at least four different homologous proteins, MAPKp38α, MAPKp38β, MAPKp38γ, and MAPKp38δ (163 ,323 ,366). These MAPKp38 have the greatest homology to the yeast MAPKhog1, which is activated by hyperosmotic shock (366). Similarly to MAPKhog1, different MAPKp38 are activated by cellular stress (ultraviolet irradiation, osmotic shock, heat shock, lipopolysaccharide, protein synthesis inhibitors), certain cytokines (IL-1, TNF-α), and GPCR (366). The activation of MAPKp38 through either cellular stress or ligation of cell surface receptors involves the activation of specific protein kinases in an ordered activation module (MKKK/MKK/MAPK) (Fig. 13).

Fig. 13.

MKKK, MKK, and MAPK that can be components of MAPKp38 pathway.

Similar to MAPKerk and other MAPK families, MAPKp38 is activated by dual phosphorylation of Thr and Tyr in the Thr-Gly-Tyr activation motif (284 ,366). Upon stimulation of MAPKp38 activity, several specific substrates are phosphorylated. MAPKp38 can phosphorylate and activate the MAPK-activated protein (MAPKAP) kinase 2 and 3 which phosphorylate small heat-shock proteins such as 27-kDa heat-shock protein (295). Other substrates for MAPKp38 have been identified as transcription factors. ATF2 is phosphorylated by MAPKp38 at Thr-69 and Thr-71 within its NH2-terminal activation domain, resulting in increased transcriptional activity (42 ,284). Elk1, an effective MAPKerk substrate, also can be phosphorylated by MAPKp38 on several sites in its COOH-terminal activation domain (285). Chop (also known as GADD153) is a member of the C/EBP family of transcription factors that is phosphorylated by MAPKp38 on serine residues 78 and 81 (355). Phosphorylation of these residues enhances the ability of Chop to function as a transcriptional activator. Finally, the transcription factor Max binds to a COOH-terminal truncated isoform of MAPKp38α, leading to phosphorylation of Max (397). Max heterodimerizes with c-Myc, a MAPKerk substrate, raising the possibility that c-Myc/Max heterodimers represent a point of integration between the MAPKerk and MAPKp38 signaling pathways.

Because MAPKp38 can phosphorylate many different substrates, it is reasonable to suggest that MAPKp38 affects many different biological functions. To investigate biological functions of MAPKp38, a kinase inhibitor of MAPKp38 [pyridinyl-imidazole compound SmithKline Beecham (SB)-203580] is often used. It was found that administration of SB-203580 blocks production of cytokines (IL-1 and TNF-α) in stimulated monocytes (197). Furthermore, blockage of MAPKp38 activity inhibits IL-2 production in T cells (361). This suggests that MAPKp38 is important for the production of cytokines in hematopoietic cells. In addition, inhibition of MAPKp38 by SB-203580 prevents IL-2- and IL-7-driven proliferation, indicating a role for MAPKp38 in cytokine-stimulated cellular proliferation (64). MAPKp38 is also implicated in apoptosis. Several cellular stresses such as osmotic shock and ultraviolet irradiation that cause apoptosis activate MAPKp38 kinase activity (366). In addition, death receptors (Fas and TNF receptors) upon ligation cause activation of MAPKp38 (34). Blockage of MAPKp38 activation by pyridinyl-imidazole compounds impairs Fas-induced apoptosis in T cells. Overexpression of MKKKask1, which activates MAPKp38 and MAPKjnk activity, induces apoptosis, and TNF-induced apoptosis is blocked by kinase dead MKKKask1 (149). In addition, treatment of cells with sodium salicylate induces apoptosis and activates MAPKp38 activity. This induction of apoptosis was blocked by SB-203580 (306). Just as the case with MAPKjnk, there are cells where MAPKp38 is activated without apoptosis (329). Thus the involvement of MAPKp38 in apoptosis may be specific to the type of induction of apoptosis and the cell type used. It is likely that the involvement of MAPKp38 in survival versus death responses is integrated into the sum of metabolic changes that determine if the activation of caspases will irreversibly commit a cell to apoptosis.

The function of MAPKp38 in differentiated cells is poorly defined. MAPKp38 is activated by thrombin in platelets (183), suggesting that MAPKp38 plays a role in platelet activation, but no definitive studies have been reported. The MAPKp38 pathway may be involved in the cardiac hypertrophic growth program, since expression of an active form of MKKmkk6, which activates MAPKp38, leads to augmentation of cell size, induction of the genes encoding A- and B-natriuretic peptides, and increase of α-skeletal actin expression. The active form of MKKmkk6 also elicited sarcomeric organization (396).

Regulation of MAPK activation can occur through the activation of small G proteins of the Rho family. In the MAPKjnk pathway, Cdc42 and Rac1 mediate MAPKjnk activation (366). Coexpression of constitutively active forms of Cdc42 and Rac with MAPKp38 leads to increased activity of MAPKp38. Expression of dominant negative Cdc42 and Rac inhibits the ability of IL-1 to increase MAPKp38 activity. These findings suggest that Cdc42 and Rac mediated MAPKp38 activation (11 ,366 ,398). Because Cdc42 and Rac do not directly activate MAPKp38 activity, they must activate other signaling molecules leading to MAPKp38 activation. These small G proteins have been shown to activate a family of serine and threonine kinases called p21-activated kinase (Pak) (176 ,221). Expression of constitutively active Pak stimulates MAPKp38 activity, and a dominant negative Pak suppresses IL-1-induced MAPKp38 activity (11 ,398). Coexpression of Cdc42 or Rac with dominant negative Pak suppresses Cdc42/Rac-mediated MAPKp38 activation (11 ,398). Thus Cdc42 and Rac regulate the MAPKp38 signaling pathway in response to at least some cellular stimuli such as IL-1, and this apparently occurs through activation of Pak. Pak also associates with the adapter protein Nck through its SH3 domain, which localizes Pak to membranes causing its activation (214). Nck could thus function to link activated surface receptors to the Cdc42/Pak signaling pathway and consequently MAPKp38 activation. However, there is no evidence indicating that membrane localization of Pak leads to MAPKp38 activation. Moreover, no proteins downstream of Pak that could link Pak to the activation of MAPKp38 have been characterized. There are, however, additional MKKK candidates for MAPKp38 activation. MKKKtak1 is activated by Tab1 (Tak1-binding protein) mediating TGF-β signal transduction, and the expression of MKKKtak1 in cells leads to MAPKp38 activation (354). Another candidate is MKKKask1, which upon expression in cells activates MAPKp38 activity. MKKKask1 is activated in response to TNF-α treatment of cells that also leads to MAPKp38 and MAPKjnk activation (149). Finally, MKKKsprk may also activate MAPKp38 activity upon expression in cells (335) (Fig. 13). The connection between these MKKK to Cdc42/Rac-mediated activation of MAPKp38 is unknown, but it appears there will be multiple protein kinases leading to MAPKp38 activation similar to the MAPKerk and MAPKjnk signaling pathways.

The protein kinases responsible for phosphorylation and activation of MAPKp38 are MAPKKmkk3 and MAPKKmkk6 (247 ,366) (Fig. 13). Both of these MKK are activated by MKKKtak1, MKKKask1, and MKKKsprk. MKKKsprk associates with MKKmkk6 and phosphorylates amino acid residues on MKKmkk3 required for its activation (335).

Hints in the literature suggest that some ligands can activate one MAPK pathway and at the same time inhibit another MAPK pathway. For example, glia maturation factor activates the MAPKp38 pathway and at the same time inhibits the MAPKerk pathway. Glia maturation factor is a 17-kDa brain protein that, when phosphorylated by the cAMP-regulated PKA, specifically enhances the activity of MAPKp38 (209). On the other hand, PKA-phosphorylated glia maturation factor is a strong inhibitor of MAPKerk1 and MAPKerk2 (394). This indicates that a PKA-dependent pathway is able to differentially regulate the activity of MAPK pathways.

Just as for the MAPKerk pathway, GPCR in addition to the thrombin receptor can induce the activity of MAPKp38. In HEK 293 cells, MAPKp38 can be activated through the Gq/G11-coupled m1 muscarinic receptor, through the Gi-coupled m2 muscarinic receptor, and through the Gs-coupled β-adrenergic receptor. Overexpression of Gβγ or G11α, but not Gsα or Giα, can stimulate MAPKp38 (389). Thus, depending on the GPCR, MAPKp38 activation can be mediated by the βγ-subunit complex or the α-subunit of heterotrimeric G proteins.

MAPKp38 family members are negatively regulated by protein kinase phosphatases. These phosphatases dephosphorylate the Thr-Gly-Tyr motif in MAPKp38, thereby inhibiting its kinase activity. There are some protein kinase phosphatases that inhibit the activity of all MAPK, whereas others are specific (366). The protein kinase phosphatase M3/6 inhibits the activity of p38 and MAPKjnk family members but is a poor inhibitor of the MAPKerk family members (253) (see sect. v A3).

D. MKK5/MAPKerk Pathway

The least known mammalian MAPK pathway employs MKKmkk5 and MAPKerk5. The MKKK in this MAPK module has not been yet identified (87). MAPKerk5 can be activated by oxidative stress and hyperosmolarity (1), but also by nonstress stimuli such as serum (172). MAPKerk5 can bind MKKmkk5 in vitro (405) and, in contrast to MAPKerk2, MAPKp38α, and MAPKjnk1, is activated in vivo by expression of an activated mutant of MKKmkk5 (172). This indicates that MKKmkk5 is the upstream regulatory kinase of MAPKerk5. The activation of MAPKerk5 induced its translocation from the cytosol to the nucleus. In response to serum, the activation of MKKmkk5 and MAPKerk5 leads to the activation of the transcription factor MEF2C, which induces the expression of c-Jun (172). MEF2C can also be regulated by the MAPKp38 pathway in macrophages induced with gram-negative bacterial lipopolysaccharide (127). This shows again that different mammalian MAPK pathways can have similar downstream targets, as seen in yeast. However, the p38 and the MKKmkk5/MAPKerk5 pathway activates MEF2C through the use of different phosphorylation sites (172). This suggests that MEF2C could integrate, in a noncompetitive fashion, signals originating from different MAPK pathways. Additional work is required to define the other components and signaling proteins that regulate the MAPKerk5 module.


Components of MAPK pathways are rapidly being characterized in a number of nonmammalian experimental systems, including X. laevis, Drosophila melanogaster, Caenorhabditis elegans, and plant species. The role played by MAPK pathways in these organisms is still fragmentory, but because they represent powerful experimental systems, they will undoubtely provide many interesting insights on the function and role of MAPK pathways in important physiological processes (i.e., development). To illustrate this point, we very briefly discuss MAPK modules and their regulation in nonmammalian and nonyeast organisms.

A. Dictyostelium discoideum

In Dictyostelium, three independent MAPK pathways regulate growth and multicellular development. The Dictyostelium MEK (MKKDdmek1) is required for chemotaxis toward cAMP during aggregation (216). MKKDdmek1 is required at the time of cAMP stimulation for the activation of guanylyl cyclase and the production of cGMP, the second messenger for chemotaxis in these cells. Dictyostelium MAPKerk1 is required for vegetative growth and probably plays a role during multicellular development (110). The Dictyostelium MAPKerk2 is activated by extracellular cAMP and is required for receptor activation of adenylyl cyclase during aggregation, prespore-specific gene expression, and morphogenesis in a partially GPCR-dependent manner (218 ,307). The chemoattractant folic acid, which allows the amoebae form of Dictyostelium to find bacteria in the wild, activates MAPKerk2. This response requires the Dictyostelium Gα protein (Gα4) (219). In Dictyostelium, there are thus two independent pathways to MAPKerk2, one that is mediated by folic acid and is totally dependent on G protein-dependent pathways and one that is mediated by cAMP and is partially G protein independent.

B. Caenorhabditis elegans

The ras homolog of C. elegans (Let-60) is a key player in the signal transduction pathway that controls the choice between vulval and epidermal differentiation in response to extracellular signals. The pathway downstream of Let-60 is composed of a MAPK module employing the MKKKraf1 homolog MKKKlin45 (129), the MKKmek homolog cMKKmek2 (181 ,381), and the MAPKerk homolog MAPKsur1 (380). MAPKsur1 acts upstream of the transcription factor Lin-1 (17 ,380). Worms deficient in MAPKsur1 can be rescued by rat MAPKerk2, demonstrating the conservation of a Ras-mediated pathway between worms and mammals. Another MKKKraf1 homolog, MKKKksr1, has been isolated by screening suppressors of the multi-vulva phenotype caused by an activated Let-60 Ras allele (328). It is, however, unclear whether MKKKksr1 lies downstream of Let-60 Ras or if it is part of a MAPK pathway that functions in parallel to the MKKlin45/MAPKsur1 pathway.

C. Drosophila melanogaster

The gene product encoded by the rolled locus is a homolog of the mammalian MAPKerk proteins. This kinase, dMAPKrolled, is required for the sevenless receptor tyrosine kinase expressed in the R7 precursor cells to transduce the signal that leads to the development of the R7 photoreceptor cell (19). The Drosophila homolog of MKKKraf1, dMKKKraf1, is required for early larval development (263), indicating that the Drosphila MAPKerk pathway is involved in this process. The Drosophila MAPKerk pathway seems to be involved in several developmental pathways in the fruit fly.

Dorsal closure, a morphogenetic process occurring during Drosophila embryogenesis, is regulated by the gene products of hemipterus (dMKKhep) and basket (dMAPKbsk), the homologs of MKKmkk7 and MAPKjnk1, respectively. Embryos lacking these kinases exhibit a dorsal closure phenotype (292). Similarly, embryos lacking the Drosophila Jun homolog (Djun) also have a defect in dorsal closure (143). This suggests that Jun is a target of dMKKhep/dMAPKbsk signaling. However, the role of dMAPKbsk in Djun regulation has been disputed (143). dMAPKbsk may also be implicated in an immune response toward bacterial infection in Drosophila because it is activated by lipopolysaccharide, a component of the bacterial cell wall (317).

MAPKp38 Drosophila homologs (dMAPKp38a and dMAPKp38b) have recently been cloned (126 ,130). Expectedly, the dMAPKp38 are activated by stress (osmotic shock, heat shock, H2O2) in Drosophila cells (126 ,130). Surprisingly, however, the dMAPKp38 do not positively regulate immunity signaling in contrast to what appears to be the case in mammalian cells; rather, the dMAPKp38 seem to act as attenuators of immunity gene expression (i.e., repression of transcription of the antimicrobial peptides attacin and acropin) (130). This attenuation of immunity gene expression only occurs at late stages of infection and could represent a control mechanism that prevents overstimulation of the immune system and its associated detrimental effects.

In conclusion, the three types of MAPK pathways found in mammalian cells (MAPKerk, MAPKjnk, and MAPKp38 pathways) appear to be present in insect cells. The Drosophila system will undoubtly be useful to assess the roles of MAPK pathways, since genetic studies are relatively easily performed in the fruit fly.

D. Xenopus laevis

G2 arrest of Xenopus oocytes requires that preformed cyclin B-cdc2 complexes [prematuration-promoting factor (MPF)] be kept in an inactive form that is largely due to inhibitory phosphorylation of MPF. xMAPKerk downregulates the mechanism that inactivates cyclin B-cdc2 kinase and may thus be involved in the exit of the G2 cell cycle block (2). Xenopus MKKKmos (xMAPKKKmos) is a germ-cell-specific protein that is absent from immature oocytes and is synthesized from stored maternal mRNA in response to progesterone. Translation of xMKKKmos is necessary for progesterone- and insulin-induced maturation of oocytes. xMKKKmos is the only protein that must be synthesized to initiate maturation. xMKKKmos induces rapid activation of MKK and MAPK in oocytes as well as during xMKKKmos-induced mitotic arrest in early embryos (276). xMKKKmos action on oocyte maturation may require the Xenopus MKKKraf1, but probably not in a direct manner, since MKKKraf1 activation occurs only several hours after xMKKKmos expression (254).

The spindle assembly checkpoint prevents cells whose spindles are defective or chromosomes are misaligned from initiating anaphase and leaving mitosis. Spindle assembly checkpoint requires the activity of the Ras-dependent MAPK pathway employing the Xenopus MAPKerk1 and MAPKerk2 proteins (331 ,356). In contrast, this MAPK pathway appears dispensable for the normal M phase entry and exit (331 ,356). In the developing Xenopus embryo, the Ras-dependent MAPK pathway is also involved in mesoderm induction (342).

The Xenopus system provides a good example of the involvement of MAPK kinase pathways in the regulation of the cell cycle and may prove to be particularly useful in studying the role of MAPK pathways in development.

E. MAPK Pathways in Plants

Stress such as drought, cold, wounding, or pathogen attacks activate MAPK in several plant species and mediates the appropriate defense or survival mechanisms (166 ,208 ,310 ,399). Mechanical stress in the form of touch, rain, and wind also activate MAPK (33).

Auxins are phytohormones that control plant growth and development. Kinases able to phosphorylate and activate ntMAPKmpk2 are stimulated in tobacco cells incubated with auxins (244). This suggests that growth and/or development is also controlled, at least in part, by MAPK pathways.

The increasing number of plant MAPK pathways members that are being characterized (Table 1) indicates that plant MAPK pathways mediate many different aspects of plant physiology. It is also anticipated that there is a great degree of conservation between the plant and the mammalian MAPK pathways, as indicated, for example, by the fact that MAPK pathways transduce stress signals in both types of organisms.


Saccharomyces cerevisiae, a single-cell eukaryote, has six MAPK, four MKK, and five MKKK encoded in its genome. The corresponding numbers in higher eukaryotes are larger, obviously reflecting the increased complexity of these organisms. For example, the number of characterized MAPK is 12 in mammals, and this will almost certainly increase when the human genome is completely sequenced. The number of MKKK appears to be equal to or larger than that of the MAPK. Why have so many MKKK? The most likely explanation is that different MKKK are integrated into different MAPK modules for responsiveness to different upstream signals such as those originating from cytokines, hormones, cell-cell interactions, and different stress responses (heat, changes in osmolarity, etc.). It is also likely that not only will specific MKKK function in more than one MAPK module, but they will probably have substrates other than MKK for regulation of cell physiology. MEKK1 is a good example, being capable of regulating the MAPKjnk pathway and the NFκB pathway.

What lies ahead in the discovery of MAPK pathways? Obviously, the genome projects for humans, C. elegans, and D. melanogaster will be enormously insightful. The number of MAPK, MKK, and MKKK in an organism will be predicted from this information. Selective expression of specific MAPK module kinases in given cell types will also be informative. The difficult task will be to characterize functionally and biochemically all the different MAPK modules operating in a cell. The biochemical characterization should determine which MAPK, MKK, and MKKK is used for a given MAPK module, and the functional characterization should define the upstream inputs and downstream substrates for each MAPK module. In addition to biochemistry and experiments employing transfection protocols, genetic approaches, such as gene-targeted disruption (knockouts), will have to be taken to unequivocally define the physiological functions of MAPK pathways. Apparent redundancy in function, particularly at the level of MKKK, makes this task difficult but also extremely important. Finally, the subcellular location of specific MAPK modules is predicted to be a major regulatory property for their regulation. The different structural motifs found in the MKKK suggest different protein-protein interactions for their regulation. These motifs include leucine zippers, proline-rich sequences, 14–3-3 interaction sites, GTP-binding protein interactions, serine/threonine and tyrosine phosphorylation sites, and more. All of these regulatory sites can influence the subcellular location, in addition to the activity of the kinases. Our prediction is that specific MAPK modules will be shown to function as sensors localized in specific regions of the cell to respond to defined extracellular and intracellular imputs for the control of gene expression, metabolism, and the cytoskeleton.


  • Present addresses: C. Widmann, Institut de Biologie Cellulaire et de Morphologie, University of Lausanne, Ch. du Bugnon 9, 1005 Lausanne, Switzerland; M. B. Jarpe, Biogen Corporation, Boston, MA.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
  187. 187.
  188. 188.
  189. 189.
  190. 190.
  191. 191.
  192. 192.
  193. 193.
  194. 194.
  195. 195.
  196. 196.
  197. 197.
  198. 198.
  199. 199.
  200. 200.
  201. 201.
  202. 202.
  203. 203.
  204. 204.
  205. 205.
  206. 206.
  207. 207.
  208. 208.
  209. 209.
  210. 210.
  211. 211.
  212. 212.
  213. 213.
  214. 214.
  215. 215.
  216. 216.
  217. 217.
  218. 218.
  219. 219.
  220. 220.
  221. 221.
  222. 222.
  223. 223.
  224. 224.
  225. 225.
  226. 226.
  227. 227.
  228. 228.
  229. 229.
  230. 230.
  231. 231.
  232. 232.
  233. 233.
  234. 234.
  235. 235.
  236. 236.
  237. 237.
  238. 238.
  239. 239.
  240. 240.
  241. 241.
  242. 242.
  243. 243.
  244. 244.
  245. 245.
  246. 246.
  247. 247.
  248. 248.
  249. 249.
  250. 250.
  251. 251.