ERK1 and ERK2
The transmission of extracellular signals to intracellular targets is mediated by a network of signaling pathways. The ERK signaling cascade is a central regulator to a large number of cellular processes such as proliferation, differentiation, and migration; it is also one of the most studied pathways. The kinases ERK1 and ERK2 are activated by MEK kinase in the signaling cascade Ras/Raf/MEK/ERK, and then they phosphorylate protein substrates on serine and threonine residues.
Several years prior to ERK1 and ERK2 cloning, respectively, in 1990 and 1991 (Boulton et al. 1991), the close correlation between mitogen activation and the increased double phosphorylation of two proteins of 41 and 43 kDa on a phospho-tyrosine residue and a phospho-threonine/phospho-serine was revealed by two-dimensional polyacrylamide-gel electrophoresis (reviewed in Chambard et al. 2007). Because of the sustained phosphorylation during the critical part of G0/G1 phase of the cell cycle, these two proteins known as p41 and p43 were suspected of playing a key role in cell cycle entry and they correspond, respectively, to ERK2 (42 kD) and ERK1 (44 kD). Their essential role in cell proliferation was then confirmed in fibroblasts by showing that ERK activation occurred in two phases, a rapid one in response to many stimuli and a late phase lasting several hours induced only by mitogenic inducers; moreover, decreasing ERK expression was shown to block cell cycle entry (Pages et al. 1993).
Mechanism of ERK Activation
ERK is a member of the CMGC Ser/Thr protein kinase family, which encompasses the CDKs, MAPK, GSK, and CDK-like (cyclin-dependent kinases, mitogen activated kinases, glycogen synthase kinase and cyclin-dependent-like kinases, respectively). ERK is a member of the MAPK cluster, and the closest kinase to ERK1/2 is ERK5 which lies on a distinct signaling cascade.
ERK can phosphorylate substrates on a serine or a threonine residue; the consensus phosphorylation site was established from studies with phosphorylated peptide banks, it is PXS/TP (X can be any amino acid and P-proline, S-serine, T-threonine). ERK is a proline-directed kinase since there is an absolute requirement for +1proline for the phosphorylation site of substrates (reviewed in Lee et al. 2004). This requirement was closely analyzed, which results from a close association of Ala-187 of ERK2 after remodeling of ERK2 to convert a small depression on the surface to a deep pocket where the proline of the substrate interacts. Interestingly ERKs are resistant to activating point mutations. Even the replacement of either or both phosphorylation site residues by negatively charged amino acids does not lead to enzymatic activation unlike many kinases. It is thought that the interaction of +1proline of substrates is crucial to prevent undesired activation of ERK, since this step requires the formation of an energetically unfavorable structure.
ERK is a compact protein encompassing two lobes. The active site of the kinase is located within the cleft between the two lobes. In the inactive state, the two lobes are rotated in comparison with the two lobes of PKA (constitutively active subunit), which is probably the reason for inactive ERK to display very low specific activity unless the kinase is doubly phosphorylated (reviewed in Lee et al. 2004). Activation of ERK is a combination of rotation of the two lobes upon double phosphorylation by MEK of the activation lip on threonine 185 and thyrosine 187 (T185EY187 sequence of human ERK2) and remodeling of ERK by binding of the +1proline of the substrate.
Many kinases are proline directed, such as the CDKs; hence, the sequence of the phosphorylation site is not sufficient to provide substrate specificity for ERK. Indeed, substrates must harbor at least one of two types of docking motifs to be specifically phosphorylated by ERK, the F-docking site, or the D-docking sites. The F-docking site of substrates (also called FXFP site or DEF domain, for Docking site for ERK, FXF) is located close to the small depression where the +1proline binds, which is constituted by an hydrophobic pocket (Lee et al. 2004). As a consequence of this proximity, the phosphorylation site is usually located close to the F-docking site. When ERK is inactive (not doubly phosphorylated), this hydrophobic pocket of the F-docking site is occluded; hence, substrates cannot bind to inactive ERK via this site. On the contrary, the second binding site to ERK, the D-docking site (also called DEJL or KIM for Kinase Interacting Motif), binds to ERK when it is active or inactive (cited in (Ebisuya et al. 2005; Murphy and Blenis 2006). Furthermore, the D-docking site is found in all types of ERK partners: substrates, activator (MEK), and inhibitors (DUSPs, for Dual Specificity Phosphatases). The D-docking site consists of a cluster of basic amino acids at the N-term of a hydrophobic motif.
This D-docking site of substrates binds to the CDS (Common Docking Site), a docking groove located on the back of ERK (the front of the kinase being the catalytic cleft between the two lobes (reviewed in Ebisuya et al. 2005; Murphy and Blenis 2006). The fact that the CDS is located on the back of the kinase allows greater flexibility for the localization of the D-docking site on substrates (by convention the front of the kinase is the phosphorylation site). The F-docking site of substrates binds to the F-recruitment site (FRS) of ERK. Some substrates possess a D-docking site, some possess an F-docking site, and others possess an array of these docking sites.
Subcellular Localization of ERK
ERK translocate to the nucleus after acute stimulation of the cell and it accumulates in the nucleus during prolonged stimulation (Lenormand et al. 1993). Nuclear translocation of ERK is required for cell cycle entry. This was demonstrated by retaining ERK in the cytoplasm upon expression of inactive MKP3/DUSP6 that binds tightly to ERK and remains strictly located in the cytoplasm. Retention of ERK in the cytoplasm has been shown to alter neither ERK kinase activity nor phosphorylation of cytoplasmic substrates, while ERK-dependent transcription and cell proliferation is blocked (Brunet et al. 1999).
MEK remains in the cytoplasm as a consequence of active export out of the nucleus mediated by its NES sequence (Nuclear Export Sequence). MEK binds to inactive ERK in resting cells; therefore, MEK sequesters inactive ERK in the cytoplasm. Upon activation of ERK by MEK, the conformational changes liberate ERK that can diffuse freely and interact with substrates and new partners in all cell compartments. The fact that MEK behaves as the cytoplasmic anchor for ERK was demonstrated by showing that only MEK co-overexpression could maintain the cytoplasmic localization of overexpressed ERK, whereas expression of saturating levels of the ERK-binding site of MEK abrogated ERK export from the nucleus (reviewed in Ebisuya et al. 2005).
Regulation of ERK nuclear translocation is an essential feature of the Ras/Raf/MEK/ERK signaling cascade. It has been shown that the stimulation-induced nuclear accumulation of ERK requires the synthesis of short-lived nuclear anchors (reviewed in Chambard et al. 2007). If protein synthesis is blocked during stimulation, ERK enters the nucleus but cannot accumulate there. In vivo regulation of ERK translocation has been observed upon gradual expression of cytoplasmic anchors such as PEA15 and Sef or nuclear anchors such as DUSP5 and Vanishing (reviewed in Ebisuya et al. 2005). As an example, cytoplasmic retention of ERK by PEA-15 has been linked to Ras-induced senescence in fibroblasts (reviewed in Ebisuya et al. 2005).
ERK lacks a nuclear localization sequence (NLS), suggesting that ERK may enter by a piggy-back mechanism via binding to NLS-containing proteins. NLS-dependent mechanisms require cytosolic factors and energy for Ran-dependent cycling of importins. However, reconstituted import assays have shown that ERK can bind directly to FXFG repeated sequences of nucleoporin in the lumen of the NPC (Nuclear Pore Complex), indicating that it may enter the nucleus in the absence of energy sources or cytosolic factors (reviewed in Ebisuya et al. 2005; Lidke et al. 2010). Point mutations of ERK revealed that inactive and active ERK interact with nucleoporins via different domains; thus, both active and inactive ERK can be transported across the nuclear pore in an energy-independent fashion (Yazicioglu et al. 2007). Nonetheless, in a reconstituted import assay, nuclear import of ERK2 fused to beta-galactosidase required the presence of energy, revealing an active nuclear import. Overall, these observations suggest a role for both an energy-dependent and an energy-independent mechanism (via direct binding to nucleoporins) in ERK cytoplasmic-nuclear translocation. Since ERK was shown to form dimers when crystallized and when purified in vitro, the possibility that ERK dimerization could play a role in nuclear entry was scrutinized. Several independent studies have failed to detect existence of ERK dimers in vivo, either by lack of FRET between GFP-ERKs of different colors, by advanced microscopic techniques such as emFRET and fluorescence correlation spectroscopy or by an array of biochemical methods (reviewed in (Roskoski 2012). An ERK “dimerization-mutant” that failed to dimerize in vitro was shown to be more slowly activated than wild-type ERK and to enter more slowly in the nucleus, but this mutant accumulated normally in the nucleus. These experiments confirm that the activation step is central for ERK nuclear entry, not dimerization. Half-maximal nuclear accumulation of ERK is reached within 3 min (movie in supplemental data of Lidke et al. 2010).
Substrates and Cellular Functions of ERK
ERK can phosphorylate many substrates (von Kriegsheim et al. 2009); it has been demonstrated that ERK activation is required for cell proliferation, cell growth, cell differentiation, development, memory formation, senescence, and apoptosis among others. Antiproliferative mechanisms triggered by ERK activation are reviewed in Cagnol and Chambard (2010) and proliferative roles of ERK are reviewed in Torii et al. (2006) and Chambard et al. (2007). Biological responses to ERK activation depend on the cellular context and on the strength and duration of ERK activation. For example, in PC12 cells a transient ERK stimulation leads to cell division, whereas long lasting ERK activation is required for differentiation, contrarily to fibroblasts where long lasting ERK activation is required for proliferation (reviewed in Ebisuya et al. 2005; Murphy and Blenis 2006; Chambard et al. 2007). The focus of this chapter will be to understand how to convert a gradual ERK signal to an on/off switch, such as the decision for an individual cell to proliferate or not.
In fibroblasts , ERK activation occurs in two phases, an initial rapid phase lasting about 30 minutes in response to many stimuli and a late phase lasting several hours that is induced only by the persistent presence of mitogenic agonists. Whereas mitogenic and nonmitogenic agonists can activate as efficiently ERK during the rapid phase, only long-lasting stimulations of ERK induce the robust accumulation of immediate-early gene products (IEG products), such as c-Fos protein that ultimately leads to cyclin D1 expression and cell cycle entry (reviewed in Murphy and Blenis 2006). Typically, in quiescent cells, activation of surface receptors leads to ERK activation and immediate entry of ERK in the nucleus to phosphorylate preexisting transcription factors, such as Elk-1, which in turn induce transcription of IEGs, such as c-fos. When ERK stimulation is transient, the c-Fos protein is not accumulated enough prior to the decline of ERK activation and it is degraded. On the contrary, sustained ERK activity can phosphorylate C-terminal phosphorylation sites on c-Fos, which in turn unmasks an F-docking site which increases c-Fos binding to ERK to ensure phosphorylation of other sites for full stabilization of the c-Fos protein. Consequently, persistent activation of ERK leads to persistent expression of c-Fos in the nucleus versus almost no c-Fos protein expression during transient activation of ERK, converting the duration of ERK activation to an on/off switch.
In order to increase markedly transcription of IEGs, ERK also induces chromatin remodeling and increases protein translation (directly or indirectly by activating downstream kinases such as RSK, MSK, and MNK (substrates reviewed in Yoon and Seger 2006; Chambard et al. 2007).
Another way for ERK to “sense” the duration of activation occurs via the downregulation of antiproliferative genes (Yamamoto et al. 2006). As soon as ERK activation decreases, these genes are re-expressed and block cell cycle progression; only persistent ERK activation and persistent downregulation of these genes allows proliferation. When studying ERK activation in individual cells, it was shown that ERK was activated in discreet pulses, and cell proliferation is triggered when pulses occur at high frequency with elevated amplitude (Albeck et al. 2013).
When scrutinizing the role of ERK in differentiation, up to 284 proteins were identified to bind to ERK by quantitative proteomics (von Kriegsheim et al. 2009). Furthermore, 60 of these proteins were shown to change their binding to ERK during differentiation. Considering that several downstream kinases increase markedly the repertoire of ERK targets, this work illustrate why ERK is central in the cascade of activation to drive cell-surface receptor activation (upstream elements in the cascade have only few substrates; for example, MEK has only one proven substrate, ERK).
It is important to mention that ERK2 was shown to bind directly to DNA on the sequence G/CAAAG/C in a manner independent of kinase activity. ERK2 can behave as a transcriptional repressor of interferon signaling, which increases markedly the multiplicity of ERK functions (Hu et al. 2009).
ERK in Human Pathologies
The signaling cascade leading to ERK activation is deregulated several human diseases. For example, about 30% of all human cancers display activating mutations of the ERK pathway, either at the level of cell surface receptors (such as the EGFR2 receptor), at the level of Ras (K12V mutation for example), at the level of Raf (B-Raf V600E mutation for example), or at the level of MEK (MEK1 K57N for example reviewed by Roskoski (2012) and Busca et al. (2016)). There are also indices that modulators of ERK cascade such as Sprouty and Sef may play a role in tumorigenesis (Torii et al. 2006). Similarly, invalidation of the phosphatase DUSP5 in mice increases papilloma formation in a model of skin carcinogenesis (Kidger and Keyse 2016). DUSP5 is a negative regulator of ERK that plays a role in regulating the strength of the signal by inactivating ERK in the nucleus, therefore removing DUSP5 increases/prolongs ERK activation.
Activating germline mutations of components of ERK pathway also leads to a range of pathological disorders such as type 1 neurofibromatosis and Noonan syndrome (pathologies clustered under the name rasopathies (Rauen 2013)). On the opposite, diminution of ERK activation during development has also negative consequences on human health. For example, children having lost one allele of erk2 display reduced ERK2 protein levels and exhibit microcephaly, impaired cognition, and developmental delay (Samuels et al. 2008). These neuronal phenotypes are likely consequences of the multiple roles played by ERK on synaptic plasticity and memory (Kelleher et al. 2004), but also result from the important role of ERK on development. Blocking ERK activity plays also a role in the pathogenicity of Bacillus anthracis infection. This bacterium secretes lethal factor (a component of anthrax lethal toxin) that inactivates MEK1/2 through proteolysis of their amino termini (Duesbery et al. 1998). Proteolysis of MEK1/2 reduces interaction of MEK with ERK and also the phosphorylation of ERK by MEK, which leads to total loss of ERK activity in the cell during infection.
Mutations of ERK pathway components can be bona fide driver mutations in cancer. This was demonstrated upon mutation of endogenous B-Raf allele by CRE-mediated recombination in mice (Dankort et al. 2009; Dhomen et al. 2009). This B-Raf mutation of endogenous gene (without overexpression) was sufficient to trigger adenomas and further removal of PTEN or Tp53 protein (Shai et al. 2015) promoted progression towards lung adenocarcinoma. Normally the cells are protected from abnormal hyper activation of the ERK pathway since extremely elevated and long-lasting ERK activation blocks cell proliferation by increasing the levels of the cell cycle inhibitor p21 (reviewed in Murphy and Blenis 2006; Torii et al. 2006; Chambard et al. 2007). As a consequence the cell must evolve and acquire an additional mutation to overcome this cell cycle block to become oncogenic.
Most melanomas harbor activating mutation of B-Raf (66%); however, nearly all remaining melanomas harbor KRas or MEK activator mutations. Although all these mutations increase ERK activity, the biological consequences are not identical as evaluated by the genes transcribed, which is probably due to differences in retro-inhibitory mechanisms that are specific for each level of the signaling cascade (Pratilas et al. 2009).
Specific Roles of ERK1 and ERK2
ERK activation is carried by ERK1 and ERK2 that are 84% identical at the amino acid level in humans. No agonist is known to more specifically activate ERK1 over ERK2, and both ERK1 and ERK2 were shown to translocate to the nucleus upon stimulation. MEK1/2 can indiscriminately phosphorylate ERK1 and ERK2 on their identical TEY sequence. Both ERK1 and ERK2 phosphorylate substrates on the consensus PXS/TP sites with similar specific activities in vitro, measured with bacterially expressed ERKs and with immuno-precipitated ERKs (reviewed in Busca et al. 2016; Saba-El-Leil et al. 2016). The D-docking sites diverge only for a leucine instead of an isoleucine between ERK1 and ERK2, and the F-docking sites are fully conserved between ERK1 and ERK2. More recently 284 partners of ERKs that were identified by quantitative proteomics following immunoprecipitation were shown to associate well with ERK1 and ERK2 (von Kriegsheim et al. 2009).
All these observations point towards a redundant role for ERK1 and ERK2; however, invalidation of ERKs in mice gave strikingly opposite results; lack of ERK2 led to early embryonic cell death, whereas mice lacking ERK1 live and reproduce normally (reviewed in Roskoski (2012).
ERK1 and ERK2 isoforms were demonstrated to have appeared about 400 million years ago during the course of the whole genome duplication (WGD) of early vertebrates. All mammals tested so far express both ERK isoforms; however, several tetrapods have lost the erk1 gene such as birds and frogs. Furthermore, it was shown that lizards, snakes, and geckos express only the ERK1 protein despite possessing both genes in their genomes (opposite situation to that observed with crocodiles). Therefore, it was suggested that ERK1 and ERK2 are functionally redundant in vertebrates since they can live by expression either one or both ERKs (Busca et al. 2015). This conclusion is confirmed by the observation that transgenic expression of ERK1 in mice was sufficient to compensate for the loss of ERK2 (Frémin et al. 2015). Therefore, mice have been shown to live and reproduce normally expressing only ERK2 or only ERK1 (pending increasing expression of ERK1). Upon scrutinizing carefully the literature on publications that have studied the consequences of invalidating ERK1 and/or ERK2, it was concluded that nearly all studies point towards a direct correlation between the expression level of an ERK isoform and its contribution to signaling (Busca et al. 2016). In most mouse tissues, ERK1 is expressed at lower levels than ERK2 which could explain why it is possible to invalidate erk1 gene in mice without the need to increase ERK2 expression. Therefore, ERK1 and ERK2 are functionally redundant; however, the regulation of their expression is different. The strength of erk2 proximal promoter is stronger than that of erk1 and the size of erk2 gene is much larger than that of erk1 gene, especially in vertebrates. More work is needed to understand the complex regulation of ERK protein expression and to define the signaling properties conferred by the elevated expression of ERK1/2 proteins. This is becoming more important in light of the observed erk2 gene amplification in primary lung cancers (Campbell et al. 2016), and the amplification of erk2 gene as a resistance mechanism of lung cancers to anti-EGFR treatment (Ercan et al. 2012).
ERK is a central player in the Ras/Raf/MEK/ERK cascade from its unique ability to phosphorylate a number of substrates in all cell compartments. The efforts to study ERK stemmed initially from the recognition that it is a key downstream target of tyrosine kinase receptors activated during G1 phase of cell cycle; interest for ERK remained elevated when this signaling cascade was found to be abnormally activated in about 30% of all cancers. It was even demonstrated that activating mutations of the ERK-pathway are initiator mutations in cancer.
ERK1 and ERK2 are functionally redundant kinases since wild tetrapod vertebrates express either one or both isoforms. In laboratory mice, loss of erk1 gene is not detrimental for normal development nor for reproduction; moreover, the lethality of erk2 gene loss can be compensated by increasing expression of ERK1 protein. We and others propose that ERK isoforms convey signaling according to their expression levels. Nonetheless, removal of a single ERK isoform can induce phenotypical consequences, likely this is due to the lowering of the global ERK quantity under a threshold. It is fascinating that mice can live normally upon expressing only transgenic ERK1 protein (its expression being driven solely by the chicken beta-actin promoter while no endogenous erk alleles are present in the whole animal (Frémin et al. 2015)).
It remains to be discovered why ERK is required to be expressed at high levels (in mouse fibroblasts ERK2 was measured to be the most expressed signaling kinase). The first exon of erk2 gene that is so large that a specific splicing mechanism is required, does this provide additional regulatory properties? It will be crucial to understand the role of scaffolding proteins for conducting signaling; similarly it is not understood yet why the N-terminal ends of ERK1 and ERK2 are constituted of stretches of alanine and glycine residues: MAAAAAQGGGGG- for human ERK1 and MAAAAAAGAG- for human ERK2. Finally how to reconcile that long-term activation of ERK triggers proliferation in some cells while it stops proliferation of other cells while triggering differentiation?
In order to improve mathematical modelling of ERK signaling, it will be necessary to find new tools to understand the exact role of individual phosphatases that inactivate ERK upon removal of activating phosphate groups on the TEY sequence. ERK can be inactivated by many phosphatases: threonine phosphatases such as PP2A, tyrosine phosphatases such as STEP, and DUSPs/MKPs (MAPK phosphatases) that dephosphorylate both the tyrosine and threonine residues. At least 11 DUSPs can bind specifically to MAPKs and several of them can inactivate ERK in vitro (Kidger and Keyse 2016). When these phosphatases were invalidated individually in the whole animal, unexpected results/specificities were observed such as during DUSP2/PAC1 invalidation in mice.
As a first targeted therapy against the ERK-pathway, the B-Raf inhibitor Vemurafenib was approved in 2011 for treating metastatic melanoma harboring B-Raf V600E mutations. Unfortunately, after stunning regression of the tumors in many patients, resistances were acquired and the disease progressed again. Then patients were treated by MEK inhibitors such as trametinib, also retarding disease progression but not providing long-term cure. Currently, for melanoma patients, harboring B-Raf V600E, combinatorial therapies aim at blocking B-Raf and MEK simultaneously to avoid acquisition of resistance to treatment. However, B-Raf inhibitors trigger the formation of B-Raf/C-Raf hetero dimers, where the drug-bound partner drives activation of the drug-free partner through scaffolding or conformational modifications, leading to paradoxical activation by cRAF inhibitors (reviewed in Uehling and Harris 2015). Tumor relapse has been shown to occur following gene amplification or appearance of activating mutations at other levels of the cascade. These observations prompted the targeting of ERK, the downstream effector kinase of the cascade, hoping to bypass all upstream down-regulatory mechanisms. On one side, pharmacological agents are being developed to target ERK translocation into the nucleus (only nuclear substrates are not activated, which was demonstrated long ago to be sufficient to block cell cycle entry, reviewed in Busca et al. (2016). More classically, chemical inhibitors targeting specifically ERK kinase activity have also being developed, despite the difficulty to target specifically ERK due to the great similarity between the kinase pockets of ERK and other kinases such as CDKs. As of 2016, five clinical trials are ongoing with four different ERK inhibitors, hoping to elude the disease relapse by targeting directly the effector kinase of the pathway. Unfortunately during long-term treatment of cancer cells in culture, it has already been shown that the ERK kinase inhibitor lost effectiveness via mutation of ERK1 protein (Jha et al. 2016). Therefore, combining ERK inhibition with that of Raf or MEK may remain necessary. It is worth mentioning that full inhibition of ERK activity is not possible since ERK activity is required for mammalian life; indeed, removal of erk alleles in mice was shown to trigger cell death within 2 weeks by multiple organ failures (reviewed in Busca et al. 2016). Finally, one puzzling challenge is to understand why cancers driven by the same mutation can be treated effectively in one organ and not in another one.
Due to extreme space limitations, we were unable to cite directly many excellent contributions to this field. We would like to apologize to all authors whose work was only cited indirectly from reviews or more recent publications.
- Frémin C, Saba-El-Leil MK, Lévesque K, Ang S-L, Meloche S. Functional redundancy of ERK1 and ERK2 MAP kinases during development. Cell Rep. 2015;12(6):913–21.Google Scholar