Ras/Raf/ERK, the pathway

ERK2/ERK1 (also known as p42/p44MAPK, respectively, and officially named MAPK 1 and 3) are two isoforms of extracellular signal-regulated kinase (ERK) that belong to the family of mitogen-activated protein kinases (MAPKs), which include ERK5, the c-JunNH₂-terminal kinases (JNK1/2/3) and the p38 MAP kinases (p38 α,β,δ,γ). These enzymes are activated through a sequential phosphorylation cascade that amplifies and transduces signals from the cell membrane to the nucleus. Upon receptor activation, membrane-bound GTP-loaded Ras recruits one of the Raf kinases, A-Raf, B-Raf and C-Raf (or Raf1), into a complex where it becomes activated. Then, Raf phosphorylates two serine residues on the kinase mitogen protein kinase kinase 1 and 2 (MEK1/2; also known as MAP2K1 and MAP2K2, respectively), which in turn activate ERK1/2 by tandem phosphorylation of threonine and tyrosine residues on the dual-specificity motif (T-E-Y). Finally, active ERKs regulate by phosphorylation many cytoplasmic and nuclear targets that perform important biological functions [1].

Depending on the duration, the magnitude and its subcellular localization, ERK activation controls various cell responses, such as proliferation, migration, differentiation and death [2]. Protein phosphatases play an important role as negative regulators by controlling the Ras/Raf/ERK signaling pathway at different levels. Phosphotyrosine phosphatases target the tyrosine kinase receptors, whereas phosphoserine/phosphothreonine phosphatases target the adapter protein Shc and MEK1/2. Dual-specificity phosphatases [DUSP; also called MAPK phosphatases (MKP)], are able to dephosphorylate both of the threonine and tyrosine residues within the activation loop of MAPK. Specific DUSPs tightly regulate subcellular ERK activity. DUSP1/MKP-1, DUSP2/PAC-1, DUSP4/MKP-2 and DUSP5 are mainly nuclear, whereas DUSP6/MKP-3, DUSP7/MKP-X and DUSP9/MKP-4 are cytoplasmic. Moreover, the expression of DUSP1, -2, -4 and -6 is increased following ERK activation [3,4], taking part in a negative feedback loop aimed at terminating Ras/Raf/ERK signaling pathway stimulation.

The Ras/Raf/ERK pathway is frequently deregulated in tumors as a result of activating mutations in Ras or B-Raf, observed particularly in malignant melanoma, pancreas intestine and thyroid tumors (cosmic database: http://www.sanger.ac.uk/genetics/CGP/cosmic). Many studies associate its oncogenic potential to increased cell survival, mainly by promoting the activity of antiapoptotic proteins, such as Bcl-2, Bcl-XL, Mcl-1, IAP, and repressing proapoptotic proteins, such as Bad and Bim [5].

Paradoxically, a growing number of studies also suggest that in certain conditions, aberrant ERK activation can promote cell death. This review will summarize the different cellular models in which the Ras/Raf/ERK pathway plays an antiproliferative role. The specific pro-death function of ERK activity in neurons [6] and lymphocytes [7] and its role in cadmium toxicity [8] will also be discussed in this minireview series.

Ras/Raf/ERK pathway induces apoptosis

Programmed cell death by apoptosis is a cell-autonomous mechanism that relies on pathway-controlled activation of caspases and nucleases leading to the death of the injured cells without affecting neighboring cells. The intrinsic pathway of apoptosis regulates the activity of the Bcl-2 family proteins that control the integrity of the mitochondrial membrane. The release of proapoptotic factors from the mitochondria, such as cytochrome c, into the cytoplasm promotes the activation of initiator caspase-9, which in turn activates effector caspases such as caspase-3 or -7. The extrinsic pathway relies on the activation of death receptors from the tumor necrosis factor (TNF) receptor family that promote the recruitment and activation of initiator caspase-8 via adaptor proteins such as Fas-associated death domain (FADD) or TNFRSF1A-associated via death domain (TRADD). Strong caspase-8 activity may directly activate effector caspases; it may also require signal amplification through induction of the intrinsic pathway via cleavage of the Bcl-2 family protein Bid [9].

Early reports of a proapoptotic function of the Ras/Raf/ERK pathway appeared in 1996. Depletion of Raf by the benzoquinone ansamycin geldanamycin was shown to protect MCF-7 cells from apoptosis induced by the antitumor compound taxol [10], whereas MEK antisense cDNA expression prevented bufalin-induced apoptosis in U937 leukemic cells [11]. A growing number of studies using MEK inhibitors (PD98059, U0126) and expression of dominant negative or constitutively active forms of Ras, Raf, MEK or ERK have confirmed the implication of the Ras/Raf/ERK pathway in the induction of apoptosis (see Table 1 for details).

The proapoptotic function of the Ras/Raf/ERK pathway is well documented for apoptosis induced by DNA-damaging agents, such as etoposide [12–15], doxorubicin [13,16–20], UV [13] and gamma irradiation [21]. ERK activity has been particularly implicated in cisplatin-mediated apoptosis in renal cells [15,22–32] (Table 1).

ERK activity has also been involved in cell death induced by various antitumor compounds, such as resveratrol [33], quercetin [34], phenethyl isothiocyanate [35], betulinic acid [36], apigenin [37], oridonin [38], miltefosine [39], shikonin [40] or taxol [10,41] (Table 1).

Most of these drugs induce the intrinsic apoptotic pathway. However, ERK activity has also been involved in activation of the extrinsic pathway by death receptor ligands such as TNF-related apoptosis-inducing ligand (TRAIL) [34,42–46], TNFα [47–49], Fas [50,51] or CD40 ligand [52,53]. Cell death induced by other death pathways that occur in response to zinc [54,55], oxidation [56–59], especially in response to ONOO⁻ [60], H₂O₂ [61–64] or NO treatment [65–67], toxic compounds such as cadmium, [17,68,69], benzo[a]pyrene [70], asbestos [71] or arsenic [17,72], also require ERK activity. Many death stimuli, such as estradiol [73] or its antagonist tamoxifen [74], interferon-α [75], cephalosporin [76], the calcium mobilizer calcimycin [77], epidermal growth factor (EGF) deprivation [78], leptin [79], bufalin [11], bacterial infection [80,81], chelerythrine [82] and the dominant negative form of Rac and Cdc42 [83,84], are sensitive to inhibition of the Ras/Raf/ERK pathway (Table 1).

Conversely, constitutive activation of ERK by dominant active Raf1 combined with c-Myc expression [85] or p53 induction [86], or constitutively active MEK combined with death-associated protein kinase (DAPK) expression [87], could induce apoptosis without any other stimulus. Moreover, in rare cases, activation of the Raf/ERK pathway alone mediated by Raf1 [88,89], RasV12S35 [89], insulin-like growth factor type I (IGF-I) receptor [90] expression or by ΔRaf1:ER induction [91] was sufficient to promote cell death.

Mechanisms of Ras/Raf/ERK-mediated apoptosis

ERK activity has been associated with classical markers of apoptosis execution, such as effector caspase-3 activation, poly(ADP-ribose) polymerase (PARP) cleavage, annexin-V staining, DNA condensation or DNA fragmentation (see Table 1 for details). Depending on the cell type and the nature of the injury, activation of the Ras/Raf/ERK pathway is associated with the intrinsic apoptotic pathway characterized by the release of cytochrome c from mitochondria [22,38,57,77,79] and activation of initiator caspase-9 [20,27,29,57,59,82] or with the extrinsic apoptotic pathway, which relies on the activation of an initiator caspase-8 [27,28,40,42,68,82].

Implication of Ras/Raf/ERK pathway during apoptosis in vivo

Cytoplasmic vacuolization: lysosomal cell death and autophagy

Autophagy is a genetically regulated program, initially identified as a cell survival mechanism to protect from nutrient deprivation. However, in certain conditions, autophagy results in a form of cell death now described as type II programmed cell death [112].

We and others have shown that constitutive activation of ERK by active Raf [88,91], cadmium [68] or IGF-I receptor [90] induced a form of cell death that correlated with extensive cell rounding and the formation of cytoplasmic macrovacuoles, which pushed the nucleus and the cytoplasm to the side of the dying cell. Although cell death was associated with caspase-8 activation [68,91], this massive vacuolization is unrelated to the classical features of apoptosis. This morphology could be a sign of autophagic programmed cell death, but also of paraptosis, a form of caspase-independent cell death associated with cytoplasmic vacuolization [90]. Interestingly, other studies using cadmium [69,113] or TNFα treatment [49] have clearly associated ERK activation with autophagic programmed cell death rather than with apoptosis. This is supported by several studies that have associated ERK activity with neuron autophagic cell death in the course of a neurodegenerative disease [6,104,105]. In addition, ERK activity has also been associated with autophagy and autophagic cell death in many non-neuronal cellular models (see Table 1) in response to different stresses, such as amino acid depletion [114] and aurintricarboxylic acid (ATA) [115] in human colorectal cancer cell line HT29, soyasaponins [116] in human colon adenocarcinoma HCT-15, lindane [117] in the mouse Sertoli cell line, dihydrocapsaicin [118] in WI38 lung fibroblasts, cadmium in mesengial MES-13 cells [69,113] and TNFα treatment in MCF-7 [49] and L929 cells [48]. Interestingly, in human ovarian cancer cells, cytoplasmic sequestration of ERK by PEA-15 has been shown to promote autophagy [119]. Moreover, direct ERK activation by overexpression of constitutively active MEK can promote autophagy without any other stimulus [117].

ERK-dependent autophagic activity is associated with classical markers of autophagy, such as induction of LC3 and conversion of LC3-I to LC3-II [48,49,118], induction of beclin 1 [48], induction of BNIP-3 [66] and activation of G-interacting protein (GAIP) by phosphorylation on serine 151 [114]. p53 is also associated with the autophagic process, as ERK-mediated phosphorylation of p53 on serine 392 [48] was involved in TNFα-induced autophagy.

The lysosomal compartment plays an important role in autophagy by fusing with autolysosome vacuoles. In NIH3T3 and in human colon carcinoma HCT-116 cells, oncogenic forms of Ras, respectively, v-H-Ras and K-Ras, lead to increased sensitivity to the lysosomal cell death pathway induced by cisplatin or etoposide in an MEK-dependent manner. In these models, constitutive ERK activation leads to a decrease in the levels of lysosome-associated membrane protein-1 and -2 due to the induction and activation of cysteine–cathepsin B [15].

In humans, ERK activity is potentially associated with limb sporadic inclusion body myositis, a disease characterized by cytoplasm vacuolization of muscle fibers. Interestingly, immunostaining of muscle samples from patients revealed a strong ERK accumulation in cytoplasmic vacuoles [120].

Together, these results suggest that the Ras/Raf/ERK pathway can mediate autophagic type II programmed cell death.

ERK-induced cytoplasm vacuolization associated with autophagy has some similarity with senescence-associated vacuoles. Interestingly, autophagy has recently been reported to be required for the efficient establishment of senescence induced by a constitutively active form of Ras or MEK [121].

The Ras/Raf/ERK pathway and senescence

Cellular senescence is an irreversible form of cell cycle arrest that prevents proliferation of damaged cells or cells that have surpassed their capacity to proliferate. In response to oncogenic hyperproliferative signals, primary cells undergo cell cycle arrest leading to premature oncogene-induced senescence [122]. Although aberrant activation of the Ras/Raf/ERK pathway promotes oncogenic transformation of immortalized cells, it is also tightly associated with senescence of primary cells. In human and rodent primary fibroblastic and melanocytic cells, senescence is triggered by constitutively active forms of Ras [123–126], PAK4 [125,127], Raf [97,126,128–133] or MEK [123,124,132,134]. This process is often prevented by use of MEK inhibitors (See Table 2).

The hallmarks of ERK-mediated cell death: sustained and sequestered activity

The importance of subcellular localization of ERK activity

In most cases, prolonged ERK activation alone, such as in models expressing constitutively active forms of upstream kinases, is not sufficient to promote cell death [15,63,64,66,82,85–87]. In normal cells, subcellular localization of ERK is tightly regulated by scaffold proteins and docking phosphatases that allow nuclear accumulation of dephosphorylated ERK to terminate signaling [2]. Thus, in addition to a sustained ERK activity, the outcome of ERK-mediated cell death might also rely on an aberrant subcellular localization. Indeed, apoptosis induced by estradiol [73], tamoxifen [74], zinc [54,55] cephaloridine [76], doxorubicin [20], revestratol [33] or dominant negative mutant of Rac or Cdc42 [84] correlated with sustained nuclear ERK activity. As mentioned previously, nuclear activation of ERK is also associated with apoptosis during Ciona intestinalis development [106] and in mouse testis deficient for PEA-15 [106]. Interestingly, some of the compounds that induce nuclear ERK activity are associated with the production of ROS [54,55,76], which could promote nuclear accumulation of active ERK due to inhibition of DUSP. In MDA-MB-231 human breast cancer cells, taxol-induced apoptosis was abrogated by induction of nuclear DUSP1. In this study, DUSP1 induction clearly inhibited both ERK and JNK activity [150]. Because nuclear DUSPs (especially DUSP1 and -4) also control JNK and p38 phosphorylation [3,4], any modification of DUSP activity or expression could also increase cell death by activation of the stress pathways.

Cytoplasmic sequestration of ERK has also been associated with different forms of cell death. Cytoplasmic sequestration of ERK by binding to PEA-15 promotes autophagy [119], whereas sustained cytoplasmic ERK activity induces senescence in human primary fibroblasts [135–137]. Together, these data suggest that sustained activation of ERK in different subcellular compartments is not tolerated and results in different forms of cell death (see Fig. 1).

The limits of ERK1/2-mediated cell death studies, the specificity of MEK inhibitors

In many studies, implication of the Ras/Raf/ERK pathway in the induction of cell death is based uniquely on the sole use of MEK1/2 inhibitors PD05059 or U0126 (see Table 1). The weakness of these inhibitors is that they inhibit both MEK1/2 and MEK5 [151]. Interestingly, it has recently been shown that constitutive activation of the MEK5/ERK5 pathway could promote apoptosis of meduloblastoma cells [152] or thymocytes [153,154], through Nur77-dependent mechanisms [153,154]. As a consequence, in those types of cell, some of the effect attributed to ERK1/2 might also be caused by the MEK5/ERK5 pathway. The use of PD184352, a more recent MEK1/2 inhibitor that does not target MEK5 [155], could help to distinguish between the effects of MEK1/2 and MEK5 in those cells.

Conclusions

Together, these data clearly demonstrate that the Ras/Raf/ERK pathway plays a critical role in promoting several forms of cell death in response to numerous stress stimuli both in vitro, with various cellular models, and in vivo. A common hallmark of this response is the sustained activation of ERK, which contrasts with the transient nature of ERK stimulation found in situations where ERK regulates other cell fates. As depicted in Fig. 1, ERK activates its own phosphatases, inducing a feedback loop that, within hours, restores a basal level of ERK activity. At least in the cellular models depicted in Table 1, ERK stimulation induces the expression of gene products with death-promoting activity. We can speculate that the feedback loop decreasing subcellular ERK activity over time prevents these death-promoting factors reaching a threshold concentration that triggers cell death. Consequently, any agent affecting the kinetics of ERK activity in a given cellular compartment (such as the ROS that inhibit DUSP in Fig. 1) has a potential to induce cell death. Given the importance of the spatiotemporal regulation of ERK activity for the control of cell division [2], the induction of cell death could be seen as negative feedback mechanism preventing uncontrolled cell proliferation.

The Ras/Raf/ERK pathway is among the most commonly deregulated pathways identified in tumors, as indicated by frequently observed activating mutations in Ras or B-Raf oncogenes. Thus, this pathway is currently the target of new antitumor strategies, based on the inhibition of upstream ERK regulators. However, because ERK activation is implicated in DNA-damaging agent-induced cell death (see Table 1), inhibiting ERK activity in combination therapy with classical antitumor compounds, such as cisplatin or doxorubicin, might affect the efficiency of such compounds.

Because prolonged ERK activation has been shown to promote the death of human cancer cell lines from different origins (see Table 1), this property of the Ras/Raf/ERK pathway to induce cell death could be used to target cancer cells. However, although tumor cells escape Ras/Raf/ERK pathway-induced senescence by inactivating effectors of senescence, such as p53 or p16/INK4A, mechanisms involved in ERK-induced cell death might also be silenced in tumor cells. Tumor cells with high ERK activity might also have re-modeled the ERK signaling to escape ERK-mediated cell death. Thus, the crucial biochemical events underlying sensitivity or resistance to ERK-mediated cell death remain to be fully understood. We propose the hypothesis that in tumor cells harboring strong ERK activity, the alteration of compensating pathways (PI3K/AKT, Wnt, etc…) would unleash the cell killing ability of ERK. Alternatively, if reagents were able to sequester ERK in a given subcellular compartment, the changes in the spatiotemporal regulation of ERK might be lethal. Moreover, because sustained ERK activity is required to promote cell death, such strategies would only target cancer cells with deregulated ERK activity and not normal cells in which ERK activation is transient.