MAPKs (mitogen-activated protein kinases) are key components in cell signalling pathways. Under optimal growth conditions, their activity is kept off, but in response to stimulation it is dramatically evoked. Because of the high degree of evolutionary conservation at the levels of sequence and mode of activation, MAPKs are believed to share similar regulatory mechanisms in all eukaryotes and to be functionally substitutable between them. To assess the reliability of this notion, we systematically analysed the activity, regulation and phenotypic effects of mammalian MAPKs in yeast. Unexpectedly, all mammalian MAPKs tested were spontaneously phosphorylated in yeast. JNKs (c-Jun N-terminal kinases) lost their phosphorylation in pbs2Δ cells, but p38s and ERKs (extracellular-signal-regulated kinases) maintained their spontaneous phosphorylation even in pbs2Δste7Δmkk1Δmkk2Δ cells. Kinase-dead variants of ERKs and p38s were phosphorylated in strains lacking a single MEK (MAPK/ERK kinase), but not in pbs2Δste7Δmkk1Δmkk2Δ cells. Thus, in yeast, p38 and ERKs are phosphorylated via a combined mechanism of autophosphorylation and MEK-mediated phosphorylation (any MEK). We further addressed the mechanism allowing mammalian MAPKs to exploit yeast MEKs in the absence of any activating signal. We suggest that mammalian MAPKs lost during evolution a C-terminal region that exists in some yeast MAPKs. Indeed, removal of this region from Hog1 and Mpk1 rendered them spontaneously and highly phosphorylated. It implies that MAPKs possess an efficient inherent autoposphorylation capability that is suppressed in yeast MAPKs via a C-terminal domain and in mammalian MAPKs via as yet unknown means.
MAPKs (mitogen-activated protein kinases) are key components in cell signalling pathways. In mammals, MAPKs are classified into sub-families that include the ERKs (extracellular-signal-regulated kinases), the JNKs (c-Jun N-terminal kinases) and the p38s [1,2]. Other known mammalian MAPKs are ERK5 (Bmk1) and ERK7 . Members of all subfamilies are concomitantly activated (to different levels) in response to any of a variety of stimuli, including growth factors, cytokines, radiations, high osmolarity and shear stress . Some MAPKs are essential for embryonic development [5,6], as well as for proper differentiation and functionality of the brain , muscle  and the immune system . MAPKs are highly conserved throughout evolution in all eukaryotes at the levels of sequence and the principal mechanism of their activation (requirement of dual phosphorylation; see below). This conservation led to the view that MAPKs are similarly regulated and could be functionally substitutable between organisms. This notion was supported by several studies [10–13].
Catalytic activity of MAPKs is strictly regulated. In optimally growing cells, not evoked by stresses, growth factors or hormones, their catalytic activity is kept off. In response to stimulation they are activated by phosphorylation of a conserved tyrosine and threonine residue, separated by only one amino acid, located in a domain called the phosphorylation lip [1,14,15]. This phosphorylation is catalysed by a dual-specificity enzyme from a family known as MAP kinase kinases [MAPKKs (MAPK kinases) or MEKs (MAPK/ERK kinase)]. MEKs are highly specific to their particular MAPK targets, which are their only substrates [16,17]. A docking groove comprising the common docking (CD) domain and the ED (glutamate/aspartate) domain is responsible for the interactions of MAPKs with activators, phosphatases and substrates [18–20]. However, additional domains seem to be also responsible for specific MEK–MAPK recognition .
MEK-independent mechanisms of MAPK activation were also reported. For example, p38α could autophosphorylate its own Thr180 and Tyr182 residues following interaction with Tab1 [TAK1 (transforming growth factor β-activated protein kinase 1)-binding protein 1]  or following phosphorylation on Tyr323 . It is not clear if ERKs and JNKs are also activated via regulatable autophosphorylation, but recombinant ERKs and some isoforms of JNKs possess an autophosphorylation activity [15,23].
Although usually most tightly regulated, we show here that catalytic activity of mammalian MAPKs becomes unrestricted in yeast.
The yeast Saccharomyces cerevisiae possesses five MAPK pathways. The mating pathway and its MAPK Fus3 controls cell cycle arrest and cell fusion [4,24]. The filamentation-invasion pathway, with Kss1 as the MAPK, transduces signals leading to filamentous and invasive growth . The cell integrity pathway with the Mpk1 MAPK is essential for synthesis and maintenance of the cell wall [26–29]. The high osmolarity growth pathway, with the Hog1 MAPK, is essential for growth under high osmotic pressure . The spore wall assembly pathway (Smk1) was not studied yet in depth . The phosphor-acceptor residues of Kss1, Fus3 and Mpk1 are part of a TEY motif, making them members of the ERK family. Hog1 has a TGY motif similar to p38 proteins. Smk1 has a TNY motif.
Several attempts to replace a yeast MAPK with its mammalian orthologue were already made. It was reported that JNK1, but not JNK2, rescued hog1Δ cells and allowed growth on medium supplemented with NaCl [12,32,33]. p38α was also shown to partially rescue hog1Δ cells . Another study showed that only mutated, less active, p38α could rescue hog1Δ cells . However, Alonso-Monge et al. [10,33] showed that p38α could not rescue hog1Δ cells. Atienza et al.  reported that expression of a MEKK (MEK kinase)–ERK1 fusion protein in yeast triggered phenotypes reminiscent of filamentous growth, suggesting that this chimera may mimic activated Kss1. Blumer et al.  showed that the mammalian MAPKKK (MAPKK kinase), MEKK1, suppressed the phenotype of bck1Δ cells (Bck1 is the MAPKKK for Mpk1). mpk1Δ cells were shown to be rescued by ERK5 . These studies relate the mammalian JNKs and p38s to the Hog1 pathway and the ERK1 to the filamentous/Kss1 pathway. Besides observing the phenotypic effects, most of these studies did not monitor catalytic activity or regulation of mammalian MAPKs in yeast. We decided, therefore, to systematically express mammalian MAPKs of all families in yeast and to analyse their activity, regulation and phenotypic effects.
We found, unexpectedly, that the mammalian MAPKs are spontaneously phosphorylated in yeast. We further revealed the mechanism underlying the spontaneous phosphorylation. We show that phosphorylation of JNK is mediated via Pbs2, whereas ERK1 and p38s are spontaneously phosphorylated in yeast by a combined mechanism of autophosphorylation and MEK-mediated phosphorylation. Strikingly, any yeast MEK seems capable of phosphorylating p38 and ERKs. Finally, we investigated the structural basis for the promiscuity toward yeast MEKs and the acquirement of autophosphorylation capability, and suggest that the mammalian MAPKs are not properly regulated in yeast because during the course of evolution they lost a region in the C-terminus that now exists only in yeast MAPKs and is responsible, in part, for their regulation.
MATERIALS AND METHODS
Yeast strains and media
Yeast strains used are featured in Supplementary Table S1 (http://www.BiochemJ.org/bj/417/bj4170331add.htm). Cultures were grown on YPD (1% yeast extract, 2% Bacto Peptone, 2% glucose) or on the synthetic medium YNB-URA [0.17% yeast nitrogen base without amino acids and (NH4)2SO4, 0.5% ammonium sulfate, 2% glucose and 40 mg/litre of the required nutrients].
Yeast expression vectors are described in Supplementary Table S2 (http://www.BiochemJ.org/bj/417/bj4170331add.htm).
Preparation of cell lysates and Western blot analysis
Cell cultures (12 ml) were grown to an A600 of 0.4–0.6. Cultures were pelleted and resuspended in 10 ml of 20% trichloroacetic acid. After the samples were repelleted, they were resuspended in 200 ml of 20% trichloroacetic acid at room temperature (25 °C), and 650 mg of glass beads was added. Each sample was vortex-mixed twice for 4 min each time. Supernatants were transferred to new Eppendorf tubes, and glass beads were rinsed twice with 200 ml of 5% trichloroacetic acid (the final concentration of trichloroacetic acid was 10%). Following centrifugation, pellets were resuspended in 200 ml of 2× Laemmli sample buffer followed by addition of 100 ml of 1 M Tris base. Samples were vortex-mixed for 30 s and boiled for 3 min prior to centrifugation. Supernatant was used.
Antibodies used for identification of different proteins and their phosphorylated form are described in Table 1.
|anti-HOG1 (y215)||rabbit||1:15000||Santa Cruz SC9079|
|anti-JNK||rabbit||1:2500||Santa Cruz SC(N18)|
|anti-p38||goat||1:3500||Santa Cruz SC535-G|
|anti-p-ERK||mouse||1:2000||Santa Cruz SC7383|
|anti-p-JNK||mouse||1:2000||Santa Cruz SC6254|
|anti-p-p38||rabbit||1:3000||Santa Cruz SC23763R|
|anti-HOG1 (y215)||rabbit||1:15000||Santa Cruz SC9079|
|anti-JNK||rabbit||1:2500||Santa Cruz SC(N18)|
|anti-p38||goat||1:3500||Santa Cruz SC535-G|
|anti-p-ERK||mouse||1:2000||Santa Cruz SC7383|
|anti-p-JNK||mouse||1:2000||Santa Cruz SC6254|
|anti-p-p38||rabbit||1:3000||Santa Cruz SC23763R|
Mating and β-galactosidase assay
Transformation of mammalian MAPKs to fus3Δkss1ΔFUS1-lacZ cells was performed. Colonies obtained were grown to D600=0.5 and split. Half of the culture was treated with α-factor at 10 μg/ml for 2 h. β-Galactosidase reactions were performed as previously described . In each case, the reaction was performed in duplicate.
mpk1Δ cells exprssing the mammalian MAPKs were streaked on YPD plates without caffeine or on YPD plates supplemented with 7.5 mM, 10 mM or 12 mM caffeine. Yeast growth was monitored for 5 days.
hog1Δ cells expressing the mammalian MAPKs were streaked on YPD plates without NaCl or on YPD plates supplemented with 0.7 M, 0.9 M or 1.1 M NaCl. Yeast growth was monitored for 5 days.
Many of the mammalian MAPKs are spontaneously phosphorylated when expressed in yeast
To study the biochemistry and biology of mammalian MAPKs in yeast, we subcloned cDNAs encoding 8 mammalian MAPKs (p38α, p38β, p38γ, p38δ, JNK1, JNK2, ERK1 and ERK2) into yeast expression vectors. Each cDNA was inserted into both a 2 μ-based plasmid (allowing the presence of about 100 copies per cell) and into an integrative plasmid (integrated as one copy per cell). All plasmids and their parental empty vectors (as a control) were introduced into various yeast strains. In wild-type strains of three different genetic backgrounds (BY4741, SP1 and Σ1278b), most plasmids carrying mammalian MAPKs gave rise to several thousands colonies per plate, just as the empty vectors did. Two plasmids, pAES426-ERK1 and pES86-p38α, showed a different behaviour. Only a few colonies appeared following transfection with pAES426-ERK1, and these colonies were eventually found not to express the ERK1 protein. This result is rather surprising as ERK1 and ERK2 are almost identical proteins (∼85% identity) and yet, yeast cells cannot produce colonies when introduced with ERK1 on a 2 μ-based plasmid, but are most permissive to ERK2 overexpression. They are also permissive for ERK1 expression from an integrative plasmid. In addition to overexpression of ERK1, only expression of p38α led to a partial growth inhibition. The effect of p38α on yeast is in agreement with previously reported results [34,40].
We next wondered whether the mammalian MAPKs are being regulated normally in the yeast strains, namely phosphorylated and activated in response to appropriate signals. We tested the phosphorylation state of the MAPKs using specific antibodies, which recognize dually phosphorylated forms of each MAPK, and found that all mammalian MAPKs were constitutively phosphorylated in yeast (Figure 1). Using phospho-specific antibodies we verified that the MAPKs are phosphorylated on both threonine and tyrosine residues (results not shown). Thus, when expressed in yeast, all mammalian MAPKs lose proper regulation and are dually phosphorylated even in the absence of external signals.
Mammalian MAPKs are spontaneously phosphorylated when expressed in yeast
Phosphorylation of JNKs is Pbs2-dependent, but p38s and ERKs are phosphorylated in all MEK mutants
To test which of the yeast MAPKKs is responsible for phosphorylation of each mammalian MAPK, we expressed each of the mammalian MAPKs in five different yeast strains, each lacking a MAPKK, i.e. the pbs2Δ, ste7Δ, mkk1Δ, mkk2Δ and the mkk1/2Δ strains. ERK2 was dually phosphorylated in all strains (Figure 2A), although to a somewhat lower intensity in the mkk2Δ and mkk1/2Δ strains. Similarly, ERK1 and p38α were also phosphorylated in all strains (Figures 2B and 2E). In contrast, JNK1 and JNK2 were found to be non-phosphorylated in the pbs2Δ strain (Figures 2C and 2D). This result may suggest that JNKs are closer relatives of Hog1 than p38s. Support for this notion comes from the ability of JNK1 to rescue (partially) hog1Δ cells under osmotic stress and from the inability of p38α to do so (see below).
JNKs are not phosphorylated in a pbs2Δ yeast strain
The mechanism responsible for ERKs and p38 constitutive phosphorylation in yeast is obscure. One possible explanation is that the yeast MAPKKs are not specific towards ERKs and p38, and more than one MAPKK could phosphorylate them. Another explanation could be autophosphorylation of these MAPKs when expressed in yeast.
A kinase-dead mutation in ERKs and p38α did not abolish the constitutive phosphorylation
To test the possibility of autophosphorylation, we mutated the catalytic lysine of these kinases to arginine or alanine, creating kinase-dead mutants of ERKs, JNKs and p38α. We expressed the kinase-dead MAPKs in wild-type yeast strains in order to monitor their phosphorylation state. Notably, normal growth of yeast cells was observed when we over-expressed the kinase-dead mutants of ERK1 and p38α, verifying that the inhibitory effect of ERK1 and p38α on cell proliferation is mediated via their catalytic activity and is not a mere consequence of their overexpression. Monitoring the phosphorylation state revealed that the kinase-dead variant of p38α (K53A) was phosphorylated in yeast, although to a lesser extent in comparison with p38α wild-type (Figure 3A). Kinase-dead ERK proteins were phosphorylated in the wild-type yeast strain (Figure 3B), and even in mkk1Δmkk2Δ cells. As expected, kinase-dead JNK proteins were phosphorylated in wild-type yeast strain, to similar levels as wild-type JNKs (Figure 3D). The fact that all kinase-dead versions of mammalian MAPKs were phosphorylated strongly suggests that the spontaneous phosphorylation of mammalian MAPKs in yeast is MEK-mediated. Yet it does not rule out the possibility that autophosphorylation occurs too.
Kinase-dead mutants of MAP kinases are phosphorylated in yeast cells
Kinase-dead variants of p38α and ERK1 lose their phosphorylation in a yeast strain lacking all four MEKs
To test the possibility that Pbs2 and Ste7 are redundant for ERKs and p38s phosphorylation, we constructed a pbs2Δste7Δ strain, expressed the mammalian MAPKs in it and monitored their phosphorylation state. We found that ERKs and p38α, but not JNKs, were dually phosphorylated (Figure 4). As ERKs and p38s are phosphorylated in mkk1Δmkk2Δ cells (Figure 3) and in pbs2Δste7Δ cells (Figure 4), we generated further a yeast strain lacking the four MEKs (Pbs2, Ste7, Mkk1 and Mkk2), expressed the wild-type and kinase-dead variants of p38α, ERK1 and ERK2 in it and monitored their phosphorylation state. Astonishingly, all wild-type ERKs and p38α retained their phosphorylation in this strain, strongly suggesting that they autophosphorylate (Figure 5). Strongly supporting the idea of autophosphorylation is the observation that the kinase-dead variants of p38α (Figure 5A) and ERK1 (Figure 5B) were no longer phosphorylated in pbs2Δste7Δmkk1Δmkk2Δ cells. The phosphorylation level of kinase-dead ERK2 was also dramatically reduced in these cells, although not abolished completely (Figure 5B, furthest right lane). Thus, we conclude that phosphorylation of p38α and ERKs, in yeast, is a consequence of a combined mechanism; phosphorylation by any MEK and autophosphorylation. Both MEK-mediated phosphorylation and autophosphorylation are spontaneous, occurring in unprovoked cells. Pbs2-mediated phosphorylation of JNKs is also spontaneous. An unknown kinase is responsible for the residual phosphorylation of kinase-dead ERK2 in pbs2Δ-ste7Δmkk1Δmkk2Δ cells.
p38α and ERKs are phosphorylated in a pbs2Δste7Δ yeast strain
Kinase-dead mutants of p38α and ERK1 are not phosphorylated in yeast cells lacking four MAP kinases
Removal of the C-terminal tail of Hog1 and Mpk1 renders them constitutively and highly phosphorylated
To reveal the mechanism underlying the promiscuous behaviour towards MEKs and the acquirement of autophosphorylation capability of mammalian MAPKs in yeast, we compared the sequences of the relevant mammalian and yeast proteins. This comparison (Figures 6A and 6D) showed that, although the yeast MAPKs Hog1 and Mpk1 are highly homologous to their mammalian orthologues, they contain an extra C-terminal tail of approx. 100 amino acids. It could be, therefore, that this tail is important for maintaining the low activity of the yeast MAPKs. Perhaps, because of losing this tail through the course of evolution, mammalian MAPKs are not properly regulated when expressed in yeast. To check if indeed the yeast-specific C-terminal tails are responsible for the proper regulation of the yeast MAPKs, we created 3 truncated Hog1 molecules and 3 truncated Mpk1 molecules, missing parts of the C-terminal tail. Truncations were planned based on the Hog1–p38α and Mpk1–ERK2 sequence alignment, so that the truncated molecules are now more similar to the mammalian MAPKs (see arrows in Figures 6A and 6D). We first tested whether the truncated molecules are still biologically active. We found that Mpk1D394 and Mpk1Q364 can still rescue mpk1Δ cells in the presence of 15mM caffeine, although not as efficiently as wild-type Mpk1. Mpk1E354 cannot rescue mpk1Δ cells (Figure 6B). Similarly, Hog1A366, Hog1S356 and Hog1F343 allowed hog1Δ cells to grow under osmotic stress (1 M NaCl), just as wild-type Hog1 (Figure 6E). Thus, the long C-terminal tail is not essential for the principal biological activities of Mpk1 or Hog1. Next, we monitored the phosphorylation state of the truncated Mpk1 and Hog1 variants. When expressed in either wild-type or mpk1Δ cells (of the BY4741 genetic background), grown under optimal growth conditions, the Mpk1D394 truncated protein was significantly more phosphorylated compared with wild-type Mpk1. Mpk1Q364 was as phosphorylated as wild-type Mpk1. Mpk1E354, which was not able to rescue cells (Figure 6B), was not over-phosphorylated, and even lost the basal phosphorylation of wild-type Mpk1 (Figure 6C). Thus, the C-terminal domain of Mpk1 is an inhibitory domain, roughly mapped to the region between residues 484–394. When expressed in hog1Δ cells, grown under optimal growth conditions, all three truncated Hog1 variants were spontaneously phosphorylated, in contrast with wild-type Hog1, which was phosphorylated only after exposure of cells to NaCl (Figure 6F). Notably, the shortest Hog1 molecule (Hog1F343) was phosphorylated to a higher level than Hog1A366. The spontaneous phosphorylation of the Hog1S356 and Hog1F343 truncated variants was found to be Pbs2-independent. Namely, these proteins were phosphorylated when expressed in hog1Δpbs2Δ cells, with and without exposure to osmotic stress (results not shown). The results with both Mpk1 and Hog1 indicate that removal of the C-terminal tail of Mpk1 and Hog1 renders them spontaneously phosphorylated; i.e. similar to ERK2 and p38α respectively, with respect to their biochemical behaviour in yeast.
Truncation of the C-terminal tail of Hog1 and Mpk1 rendered them spontaneously and highly phosphorylated compared with the wild-type protein
ERK2 partially rescues mpk1Δ cells and JNK1 partially rescues hog1Δ cells
As the mammalian MAPKs are catalytically active in yeast, we wondered if they are also biologically active and may functionally replace any of their yeast orthologues. To test this question we expressed each of the mammalian MAPKs in the mpk1Δ, hog1Δ and fus3Δ strains. First, we tested their phosphorylation status in these mutants and found that they are spontaneously phosphorylated, just as they are in wild-type yeast cells (Supplementary Figure S1, at http://www.BiochemJ.org/bj/417/bj4170331add.htm). Next, to test if any mammalian MAPK can replace Mpk1, transformants of the mpk1Δ cells expressing mammalian MAPKs were streaked on YPD plates, supplemented with caffeine (caffeine concentrations of 7.5 mM, 10 mM and 12 mM were tested; Figure 7). As positive controls, we streaked a wild-type strain and also a mpk1Δ strain into which we introduced a native mpk1 gene. Cells harbouring an empty vector served as a negative control. Besides the positive controls, only cells harbouring pAES426-ERK2 grew well on the 7.5 mM caffeine plate. On the 10 mM caffeine plate, ERK2 could just partially rescue mpk1Δ cells (Figure 7). These results indicate that among mammalian MAPKs tested, ERK2 is the functional homologue of Mpk1, although it could not support growth on high concentrations of caffeine. Importantly, not only mpk1Δ cells, but also mkk1Δmkk2Δ cells containing the pAES426-ERK2 plasmid, were able to grow on plates containing 7.5 mM caffeine (results not shown), showing that mammalian ERK2 is not only catalytically but also physiologically independent of Mkk1/2.
ERK2 can partially rescue mpk1Δ cells
To test if any of the mammalian MAPKs can replace Hog1, hog1Δ cells expressing mammalian MAPKs were streaked on plates supplemented with NaCl (0.7 M, 0.9 M or 1.1 M). Of the mammalian MAPKs tested, only JNK1 supported a very weak growth in the presence of 0.7 M NaCl (results not shown).
In order to test which of the mammalian MAPKs may functionally replace Fus3 in the mating pathway, we used the JPHY10 strain (fus3Δkss1Δ) that harbours the FUS1-lacZ reporter. We expressed all mammalian MAPKs in this strain and measured its β-galactosidase activity when cells were exposed to α-factor. As a positive control, we expressed the yeast Fus3 MAPK. High β-galactosidase activity was obtained only with the expression of Fus3 in the presence of α-factor (results not shown), suggesting that none of the mammalian kinases can replace the yeast MAPK in the mating pathway.
Because of their major effect on the fate of cells, MAPKs are very tightly regulated. They are rendered active only when dually phosphorylated. This phosphorylation occurs when the relevant upstream pathway is induced by a stimulus. As this regulation is conserved from yeast to mammals, it was proposed that mammalian MAPKs could be incorporated into, and properly regulated by, the orthologous yeast pathways. We showed here, however, that when expressed in yeast, mammalian MAPKs evade the need for stimulation and are spontaneously phosphorylated at high levels. Thus, the mammalian MAPKs are not properly incorporated into the yeast pathways and rather enforce these pathways to activate them.
The mechanisms underlying spontaneous phosphorylation of the mammalian MAPKs in yeast seems to be specific for each one. JNK1 and JNK2 are phosphorylated by Pbs2. This phosphorylation occurs spontaneously, namely, in cells not exposed to stress and, therefore, their Pbs2 should not be active. How do JNKs enforce Pbs2 to activate them? The possibility that the JNKs activate a downstream autocrine positive loop that ultimately activates Pbs2 is ruled out by the fact that kinase-dead versions of JNKs are also spontaneously phosphorylated in yeast. It seems, therefore, that a particular structural domain of JNKs is responsible for binding and activating Pbs2. This property of JNKs is reminiscent of that of intrinsically active Hog1 mutants that are also capable of enforcing Pbs2 to phosphorylate them [38,41,42]. The mechanism responsible for ERKs and p38 phosphorylation in yeast is more complex, as these kinases are spontaneously phosphorylated in cells lacking the PBS2, STE7, MKK1 and MKK2 genes. Our results suggest that any of the yeast MEKs is capable of phosphorylating p38s and ERKs and, in addition, these kinases undergo autophosphorylation. The observation that kinase-dead mutants of ERKs and p38α are phosphorylated in wild-type yeast strongly suggests MEK-mediated phosphorylation. But the fact that the kinase-dead molecules are not phosphorylated in pbs2Δste7Δmkk1Δmkk2Δ cells (whereas wild-type ERKs and p38s are) strongly implies that autophosphorylation is also part of the phosphorylation mechanism. The fact that any yeast MEK could phosphorylate p38s and ERKs is unexpected, because MEKs are highly specific enzymes. It was shown before, however, that under certain conditions some MEKs would phosphorylate MAPKs that are not their preferred substrates . It was also shown that minor changes (point mutations) are sufficient to manipulate the specificity of MEKs .
Regardless of the exact mechanism underlying the unregulated activity of mammalian MAPKs in yeast, the question remains, what is the structural basis of this mechanism? The answer could be the absence in p38s and ERKs of the C-terminal tail present in Hog1 and Mpk1. Indeed, truncating the C-terminal tail of Mpk1 and Hog1 led to spontaneous and high phosphorylation of the truncated variants, mimicking the behaviour of ERK2 and p38α in yeast. Thus, these C-terminal tails have a regulatory function in yeast, preventing autophosphorylation. Therefore the mammalian MAPKs, lacking this C-terminal tail, evade proper regulation when expressed in yeast. This suggests that MAPKs possess an efficient capability of autophosphorylation. In yeast MAPKs, this machinery is suppressed by the C-terminal tails, whereas in mammalian MAPKs it is suppressed by other means.
Previous studies of Hog1 and Mpk1 did not note any role for these tails that disappeared in evolution. To check if the tails of either Hog1 or Mpk1 contain any known domains or functional elements, or whether these inhibitory domains were conserved in evolution in other proteins, we analysed the sequence of the tails via a battery of bioinformatic tools. None of the bioinformatic algorithms could find sequences similar to the tails of Hog1 or Mpk1 in any other proteins in the database, nor in any known domains or regulatory sites (results not shown). The only observation worth mentioning was that both tails are highly rich in glutamines (Figure 6), but the meaning of this is elusive to us. Strikingly, JNK1 and JNK2 appear in different splice variants, some with a long tail and some with a short one [44,45].
At the biological level, we showed that ERK2 can partially substitute for Mpk1 and allows mpk1Δ cells to grow on medium supplemented with caffeine. Since ERK1 and ERK2 are ∼85% identical, it is surprising that only ERK2 can replace Mpk1. Recently, Truman et al.  showed that ERK5 complements the loss of Mpk1 in the cell integrity pathway in the presence of a high concentration of caffeine (12 mM). ERK2 did not allow growth on 12 mM caffeine. In our system, too, ERK2 did not allow growth on 12 mM caffeine, but did allow growth on 7.5 and 10 mM (Figure 7). It seems that both ERK2 and ERK5 could compensate for Mpk1. As ERK1 and ERK2 are so similar, it is also unexplained why yeast cells cannot give rise to transformants overexpressing ERK1, but are permissive to overexpression of ERK2. It may be that ERK1 activates the pseudohyphal growth machinery , explaining our inability to obtain colonies overexpressing ERK1. However, colonies were also not obtained in strains with a defective pseudohyphal system (results not shown). Interestingly, we were able to obtain normal colonies when we introduced ERK1 on plasmids with inducible promoters (GAL1 or MET3). When ERK1 expression was induced in these colonies, high levels of the protein were obtained, but no growth inhibition was measured (results not shown). Altogether, the results with ERK1 suggest that this protein interferes with the cells' recovery from the transformation process. Notably, differences in functionality between ERK1 and ERK2 were reported in other systems as well [5,46].
JNK1 barely rescues hog1Δ cells in the presence of NaCl (0.7 M and 0.9 M). At 1.1 M NaCl, JNK1 could not rescue hog1Δ cells at all. This result contradicts the report of Galcheva-Gargova et al. , who showed that JNK1 fully restores growth of hog1Δ cells on media supplemented with 0.9 M NaCl. Similarly, in our experiments, none of the mammalian p38 isoforms could rescue hog1Δ cells, although Han et al.  and Kumar et al.  showed that p38α partially rescues hog1Δ cells. The results shown by Alonso-Monge et al.  that p38α was not able to rescue hog1Δ cells, while JNK1 did rescue, partially, hog1Δ cells, are in agreement with our findings.
None of the mammalian MAPKs tested could functionally replace Fus3 in the mating pathway. This result may be explained by the fact that the Fus3 pathway is not a genuine orthologue of the ERK cascade because it evolved in yeast after the divergence from animals . In addition, Fus3 and ERKs have a relatively low sequence identity (49%). Finally, a bootstrap analysis does not support the grouping of these pathways at the MEK and MEKK levels (42 and 34% respectively) and at the MAPK level (58%) .
The results presented here suggest that, although MAPK pathways were highly conserved during evolution, a conceptual change in their regulation occurred. Particularly, our findings suggest that MAPKs possess an efficient autophosphorylation capability that must be suppressed. In yeast, the autophosphorylation of MAPKs is perhaps prevented by the C-terminal tail. There also seems to be a mechanism preventing activation of MEKs by MAPKs. Thus, MAPKs are regulated in a number of mechanisms [2,14,21,22], with MEK-mediated phosphorylation being just one of them. Some of the mechanisms are conceptually different between yeast and mammals.
We thank Dr Edward Winter (Department of Biochemistry and Molecular Biology, Thomas Jefferson University, Philadelphia, PA, U.S.A.) for strains. We thank Jonah Beenstock for a critical review of the manuscript.
This study was supported by grant from The Israel Science Foundation [grant number 656/06] and from the U.S.-Israel Binational Science Foundation [grant number 2005002].