MAP kinase and its relatives 12/2/2002

MAP kinase was first identified in 1987 by Ray and Sturgill using microtubule associated protein 2 (MAP2) as substrate, and so was named microtubule associated protein kinase. It was a major ser/thr kinase of the microtubule preparation. Eventually a 44kDa protein was cloned and sequenced with the typical sequence characteristics of a protein kinase. This protein kinase is quite abundant and was isolated independently in several other ways. For example cells such as lymphocytes can be treated with mitogens, which are usually lectin molecules which bidn to cell surface proteins such as receptors and adhesion molecules and induce cell division. In these cells it is fairly easy to look for kinases which are activated by mitogens and workers described a mitogen activated protein kinase, (so coincidentally also MAP kinase), which turned out to be the same molecule. Other groups were isolating protein kinases which became rapidly activated after growth factor stimulation, and found early response kinases (ERKs), which also were subsequently found to be the same molecules. Most of the early studies were aided by the fact that MAP kinase is a major phosphoprotein, so it is also a substrate for a protein kinase, allowing it to be radiolabelled with ³²P if you add ATP-g³²P0₄. It is also a major cellular protein. So nowadays the molecule is called MAP kinase or ERK, (this can all get extremely confusing…). Mammals have two very closely related and apparently functionally interchangable MAP kinases called MAP kinase 1 and 2 or Erk 1 and Erk 2. They are slightly different in molecular weight on gels, 42kDa and 44kDa, so are also known as p42^MAPK/ERK2 and p44^MAPK/ERK1 (as if they didn't have enough names already).

Other relatives of the MAP kinases were quickly identified in various different ways. One of these was the family of Jun N-terminal kinases (JNK), found by searching for kinases which would phosphorylate the transcriptional regulator c-Jun. Jnk kinase was also independently isolated by looking for kinases activated by stress (i.e. UV radiation, changes in medium concentration, serum removal in tissue culture etc.) and so is also called stress activated protein kinase (SAPK). There are 10 different mammalian Jnk proteins produced by three genes. The genes are Jnk1, Jnk2 and Jnk3. Each gene produces 2 proteins of size 46kDa and 55-56kDa by alternate transcription, producing one form with a short C-terminal extension versus another with a longer extension. It’s not completely known what the significance of the different extensions is, though the longer forms appear to interact with proteins that the shorter forms do not, suggesting that the extensions contain specific binding domains. Jnk1 and 2 also have two alternate transcripts which produce different protein sequences in the catalytic domain, and produce the a and b subtypes, which are the same molecular size. These two forms also have different protein binding partners. Jnk1 and Jnk 2 are ubiquitous while Jnk 3 is more enriched in nervous tissues.

A third major group of MAP kinase relatives are the p38 kinases, of which there are four, p38a (a.k.a. cytokine suppressive anti-inflammatory drug binding protein 1 or CSBP1), p38b (a.k.a. CSBP2), p38g (a.k.a. SAPK3 or ERK6) and p38d (a.k.a. SAPK4). All p38s are mammalian homologues of yeast Hog1, which in yeast is involved in the response to changes in osmolarity (Hog stands for high osmolarity and glycerol, and is activated by both these treatments). Another MAP kinase family member is Bmk (big MAP kinase), a little less studied to date. There are various other MAP kinase family members in human such as ERK3 and ERK4 which are at present even less well understood.

MAP kinase is activated by a wide variety of signals acting through receptor tyrosine kinases, G-protein coupled receptors, PKC etc. In the activation loop is the sequence TEY (amino acids 183-185). Both the tyr and thr must be phosphorylated in order to get MAP kinase fully activated. Interestingly MAP kinase is phosphorylated by MAP/Erk kinase, or MEK, a.k.a MAP kinase kinase which puts phosphates on both of these sites, and appears to have only MAP kinase as a substrate. This is called a dual specificity kinase and is unusual in that it can phosphorylate on both ser/thr and tyr residues. The JNK, p38 and other MAP kinase family members all are regulated by their own MAP kinase kinase. The Jnks contain the sequences TPY, p38s contain the sequence TGY in their activation loops and ERK5/bmk contains the peptide TEY, like Erk1. Interestingly the only known function of these MAP kinase kinases is to phosphorylate their respective MAP kinase. These phosphates can be removed by regular ser/thr and tyrosine phosphatases, but there is also a class of dual specificity phosphatases the MAP kinase phosphatases, whose sole function appears to be do this.

MEK and related kinases are in turn phosphorylated and activated by further ser/thr kinases, the MEK kinases or the MAP kinase kinase kinases. One of the MEK kinases or MAP kinase kinase kinases is c-Raf, one of the effectors of Ras, but there are numerous others. In fact each MAP kinase kinase can be activated by several other kinases, so information can be fed in from several different systems. Phosphorylation of MEK by Raf-1 is on two sites in the activation loop, though these are two ser or thr residues, and do not involve tyrosine phosphorylation. This presence of two ser or thr residues in the activation loop appears to be generally a feature of the activation of MAP kinase kinases and presumably affects the kinetics of activation.

In summary the activation of a MAP kinase is dependent on a specific MAP kinase kinase, though the molecule that activates this, the MAP kinase kinase kinase, is much more variable and may have other substrates. Often a small G protein is involved, such as ras for the MAP kinases, rac/rho for Jnk etc. The net effect is to funnel a lot of different inputs through the various MAP kinase kinase kinases onto a specific MAP kinase kinase and its associated MAP kinase. The activated MAP kinase in turn phosphorylates and regulates many other molecules, including many other protein kinases, signaling enzymes and transcription factors.

Raf-1 is activated by GTP bound p21 Ras. This activation was originally thought to simply involve the localization of raf-1 to the membrane by means of binding to activated p21 Ras, which is isoprenylated and hence membrane localized. Part of the evidence for this was that if you add a CAAX box to raf-1 it binds to membranes constitutively and is constitutively activated. However its now obvious that activation it is much more complex than that and activation requires binding of the cysteine rich C2 domain to membrane phospholipids and also the phosphorylation of raf-1 by several other protein kinases which appear to activate it. Most likely the major significance of the binding of raf-1 to p21 ras is allowing it to become a substrate for these kinases, which then activate the raf-1 molecule.

The cascades are interesting to study in the computer and much work has been done on this. To get a feel for the properties of these cascades, a typical mammalian cell contains 20,000 Ras molecules, 10,000 Raf-1 molecules, 36-80,000 MEK1 molecules, and 1,000,000 MAP kinase 1 and 2 proteins. There may be only a few hundred growth factor receptors per cell. The stimulation of only 10-50% of Ras to the GTP bound form is enough to activate almost all of the MAP kinase 1 and 2 molecules, so 2,000-10,000 Ras molecules can activate 1,000,000 MAP kinase 1 and 2 molecules, a 100-500 fold amplification. This is basically how very small amounts of specific growth factor can exert such powerful effects on cells.

We can get a clue as to how the mammalian cascades are likely to function by looking at yeast MAP kinases, which should be easier to understand. There are six homologous MAP kinases in yeast cells, and five of these are known to function in five different 3 protein kinase cascades, each homologous to the human MAP kinase cascade. Each of the five yeast cascades is activated by different phenomena and the activation of each results in different responses. So one cascade responds to mating pheremone and its activation results in mating. Another is activated by starvation and results in filamentation, which is something that yeast cells do in response to starvation. So each cascade appears to have a defined function. Presumably this will also be true for the cascades in mammalian cells. Interestingly some of the same proteins are components of each the yeast cascades, so that for example the MAP kinase kinase kinase ste11 and the MAP kinase kinase ste7 kinases are in both mating and filamentation pathways. Possibly the solution to this problem is that each cascade is a single protein complex with specific interactions conferred by scaffolding proteins. A whole series of scaffolding proteins have been characterized in yeast. Specific scaffolding proteins are now known in mammalian cells also; sometimes these are proteins dedicated to this function and other times they correspond to inserts in one or other of the kinase subunits. For instance MAP kinase 1 and MEK1 associate by means of an interaction mediated by the scaffolding protein MP1. JNK enzymes are known to associate with Jip-1 (Jnk interacting protein – 1). It may be generally true that kinases associate with scaffolding proteins which funnel their activity to specific substrates, and that these are particularly well characterized in the case of the MAP kinases since these are so well studied.

MAP kinase effectors

All MAP kinases, (i.e. ERK1, ERK2, ERK3, ERK4, ERK5, the JNKs and p38s) are proline directed ser/thr kinases. MAP kinase substrates contain the consensus PX,S/T,P; such as MAP2, tau and other cytoskeletal proteins. Also many other protein kinases such as p90 RSK (ribosomal S6 kinase), MAPKAPK-2 (mitogen activated protein kinase activated protein kinase), Raf, Mek, b-adrenergic receptor kinase 1/GRK-2, MAP kinase phosphatase and the EGF receptor. The effect of phosphorylation on the upstream regulators suggests complex feedback mechanisms, and some work has been performed on the MAP kinase phosphorylation of MAP kinase phosphatase, which stabilizes this protein, thus generating a negative feedback loop. Activated MAP kinase dimerizes, and the dimerization favors the entry of a lot of activated MAP kinase into the nucleus. Apparently dimerization allows nuclear entry by covering up a nuclear exclusion sequence. In the nucleus activated MAP kinase phosphorylates a variety of transcriptional regulators, such as c-Myc, c-Jun, c-Fos, NF-IL6, and the ETS family. This is a large family including the TCF (ternary complex factor) subfamily.

The ETS superfamily of proteins bind to specific DNA sequences in promoters for genes and influence the expression or non-expression of the corresponding genes. The ETS superfamily consists of ETS, Yan, ELG, PEA3, ERF and ternary complex factor (TCF), many of which are substrates for MAP kinase and are generaly as a result activated. For example Ets-1 binds to sequences often neighboring AP-1 binding sites, which together form the ras response element (RRE). RRE are found in many genes and consist of an ETS binding sequence immediately 5’ to an AP-1 site (AP1 = activator protein 1). AP1 sites are where jun-jun dimers or jun-fos dimers bind, a well studied family of widespread and powerful transcriptional regulators. Phosphorylation of ETS-1, c-fos and/or c-jun by MAP kinase strongly favors the expression of genes the promoters of which contain ras responsive elements. Genes such as those for matrix metalloproteinases, heparin binding EGF, keratin 18, granulocyte-macrophage colony stimulating factor (GM-CSF) and macrophage scavenger receptor all work in this way.

As another example neighboring binding sites for TCF and for the SRF dimer (SRF= serum response factor, very conserved DNA binding protein, found in yeast and other species) can favor expression of the appropriate genes following MAP kinase phosphorylation of TCF. There are many TCF factor proteins. Genes regulated in this way include c-fos, Jun-B, b-actin, vinculin and many others.

Finally neighboring binding sites for Ets-1 and pit-1 provide an example of how proteins can be expressed only in certain tissues in a regulated, growth factor dependent manner. Pit-1 is a transcriptional regulator expressed only in pituitory cells, so Ets-1/Pit-1 complexes can only form on specific DNA promoter sequences in these cells. Phosphorylation of Ets-1 can there favor specifically the expression of certain genes only in these pituitary cells. Prolactin growth hormone and thyrotropin b genes are regulated in this way.

The Jnk and p38 cascades also have specific substrates, mostly different from those acted on by MAP kinase, and so Jnk and p38 phosphorylation events should have different consequences. For example, Jnk phosphorylates proteins with the motif S*/T*-P-X_n-R/K, which is different from the MAP kinase motif (P-X-S*/T*-P), though some sites might meet both sequence criteria. In the nucleus Jnk acts preferentially on c-jun (obviously) and Elk-1 and Atf2 (activating transcription factor 2, also known as cyclic AMP response element binding protein 2). In the case of c-Jun, Jnk phosphorylation inhibits ubiquitin mediated degradation of c-Jun, which effectively increases the amount of c-Jun around and favors transcription of genes with AP-1 sites. The p38 kinase has been less well studied in the nucleus but also seems to act preferentially on Atf2.

Presumably the different mammalian cascades, as in yeast, each have specific functions. In general MAP kinases 1 and 2 seem to mediate growth factor responses resulting in growth and differentiation. Jnks are activated by stress, growth factors and during development and regulate growth, differentiation and apoptosis. p38 family proteins seem to respond primarily to stress and their activation results in cytokine production, inflammatory responses and apoptosis. Equally presumably we will have to wait a few more years before we understand all this.