Portal de Periódicos da CAPES

Alike but Different: RAF Paralogs and Their Signaling Outputs

2015; Cell Press; Volume: 161; Issue: 5 Linguagem: Inglês

10.1016/j.cell.2015.04.045

1097-4172

Enrico Desideri, Anna Lina Cavallo, Manuela Baccarini,

Cytokine Signaling Pathways and Interactions

RAF links RAS, one of the most potent human oncogenes, to its effector ERK and to proliferation. This role is evolutionarily conserved, but while simpler multicellular organisms express one RAF, mammals have three. This Minireview highlights common and divergent features of RAF paralogs, their signaling outputs, and roles in tumorigenesis. RAF links RAS, one of the most potent human oncogenes, to its effector ERK and to proliferation. This role is evolutionarily conserved, but while simpler multicellular organisms express one RAF, mammals have three. This Minireview highlights common and divergent features of RAF paralogs, their signaling outputs, and roles in tumorigenesis. The RAF family of enzymes comprises three evolutionarily conserved cytosolic serine/threonine kinases (ARAF, BRAF and RAF1, AKA CRAF). RAFs are best known for activating ERK signaling, which transduces extracellular stimuli (e.g., growth factors) into a phosphorylation cascade to induce biological outcomes, including proliferation, migration, differentiation, and survival. Activated downstream of the small G protein RAS by a much-studied yet incompletely understood mechanism, RAFs phosphorylate the dual-specificity kinase MEK, which in turn activates ERK. ERK phosphorylates an impressive roster of membrane, cytosolic, and nuclear targets; in contrast, RAFs and MEK mainly phosphorylate one substrate, MEK for RAF and ERK for MEK. Considering this stringent substrate specificity and the fact that simpler organisms possess only one Raf gene, the presence of three RAF isoforms in vertebrates is intriguing. All paralogs contain three conserved regions: the N-terminal CR1, comprising the Ras-binding domain (RBD) and cysteine-rich domain (CRD), which stabilize the inactive conformation; CR2, which contains residues important for RAF membrane recruitment during activation; and CR3, containing the kinase domain (Figure 1A). Phylogenetic comparisons reveal that BRAF is the isoform most similar to the single RAF kinase homolog LIN-45 in Caenorhabditis elegans and D-Raf/lethal-1-polehole in Drosophila, suggesting that BRAF is the prototypical RAF kinase, while ARAF and RAF1 may have evolved to carry out additional functions. Upon stimulation by growth factors, active RAS recruits RAFs to the plasma membrane and promotes the formation of functionally asymmetric RAF homo- and heterodimers in which one monomer, typically BRAF, allosterically stimulates the kinase activity of the other. Phosphorylation/dephosphorylation of several serine residues is also necessary for activation. In particular, the residue required for binding of 14-3-3 to the CR2 (Ser259 in RAF1) must be dephosphorylated for membrane recruitment and activation; in contrast, the C-terminal 14-3-3 binding site (Ser 621 in RAF1) must be phosphorylated to facilitate Raf dimerization. Finally, the N region upstream of CR3 must be negatively charged for activation. In ARAF and RAF1, this is achieved by phosphorylation of specific residues, while in BRAF, this region features a negatively charged Asp residue and a constitutively phosphorylated Ser residue. Thus, BRAF is poised for activation by RAS. The efficiency with which the three RAFs phosphorylate MEK also varies, with BRAF and ARAF being the most and least potent MEK activators, respectively. This may reflect not only BRAF's higher kinase activity, but also its ability to heterodimerize with MEK in the cytosol of quiescent cells. It is envisioned that, upon pathway activation, this preformed complex translocates to the membrane, forming an active tetramer with another RAF:MEK dimer. Intriguingly, the BRAF:MEK complex is dissociated by BRAF mutations promoting BRAF:RAF1 dimerization, suggesting that MEK may negatively regulate RAF dimerization and pathway activation by competing with RAF1 for binding (Figure 1B) (Haling et al., 2014Haling J.R. Sudhamsu J. Yen I. Sideris S. Sandoval W. Phung W. Bravo B.J. Giannetti A.M. Peck A. Masselot A. et al.Cancer Cell. 2014; 26: 402-413Abstract Full Text Full Text PDF PubMed Scopus (142) Google Scholar). Ultimately, the ERK pathway is switched off by dephosphorylation of its components. Even before this happens, however, activated ERK regulates its upstream components via negative feedback phosphorylation, dimming RAS activation by the guanine nucleotide exchange factor SOS (Son of Sevenless), reducing RAF dimerization and activity by phosphorylating both BRAF and RAF1, and decreasing the activity of MEK1/MEK2 dimers by phosphorylating MEK1. Thus, activated ERK exerts tight temporal control on the RAS signal transduction. Pathway activation was recently reviewed in Cseh et al., 2014Cseh B. Doma E. Baccarini M. FEBS Lett. 2014; 588: 2398-2406Crossref PubMed Scopus (75) Google Scholar. The module described above is the minimal version of the RAF/MEK/ERK pathway. However, the extent of its activation, the strength of the signal it transduces, and its biological outputs are also determined by the interaction of pathway components with scaffold proteins that insulate and localize the response, the best known of which is the RAF-related pseudokinase KSR (kinase suppressor of ras). Mutations in these scaffolds are rare in cancer (http://www.cbioportal.org/), and protein-protein interactions are notoriously difficult therapeutic targets. Recently, however, Jameson et al., 2013Jameson K.L. Mazur P.K. Zehnder A.M. Zhang J. Zarnegar B. Sage J. Khavari P.A. Nat. Med. 2013; 19: 626-630Crossref PubMed Scopus (149) Google Scholar have shown that disrupting the function of the RAF/MEK/ERK scaffold IQGAP1 (IQ motif-containing Ras GTPase-activating-like protein) inhibits RAS and RAF-driven tumorigenesis, circumventing the acquired resistance to RAF and MEK inhibitors. Thus, appreciating the full complexity of the pathway will be instrumental in defining new potential intervention points. Conditional knockout models have provided insight into the essential functions and complexity of Raf paralogs (Figure 1C; reviewed in Cseh et al., 2014Cseh B. Doma E. Baccarini M. FEBS Lett. 2014; 588: 2398-2406Crossref PubMed Scopus (75) Google Scholar). Braf ablation causes embryonic lethality through defective placenta development and, when restricted to neuronal precursors, in a progressive neurodegenerative disease caused by a failure in oligodendrocyte differentiation and myelination. These phenotypes correlate with reduced MEK/ERK activation in the relevant cells and organs and can be phenocopied by treating wild-type cells with MEK inhibitors. Thus, BRAF is the main ERK activator in vivo, whether through its own MEK kinase activity or through RAF1 activation. In contrast, none of the phenotypes observed in RAF1-deficient cells or tissues is accompanied or caused by a decrease in ERK signaling. If anything, ERK phosphorylation is higher in Raf1 knockout cells and tissues. RAF1 is an essential regulator of keratinocyte migration and of the stability of nascent cell-cell junctions in endothelial cells. In embryonic liver and fibroblasts, it regulates apoptosis via the internalization of the death receptor FAS. All of these essential functions of RAF1 can be attributed to its ability to bind the Rho-activated, cytoskeleton-based kinase ROKα (Rho-associated kinase α), bringing it to specific subcellular localizations and/or dimming its activity. In addition, RAF1 promotes cell survival by antagonizing the proapoptotic kinase ASK1. ASK1 hyperactivation correlates with the defects in postnatal development observed in RAF1-deficient cardiac muscle, and Ask1 ablation rescues this (Yamaguchi et al., 2004Yamaguchi O. Watanabe T. Nishida K. Kashiwase K. Higuchi Y. Takeda T. Hikoso S. Hirotani S. Asahi M. Taniike M. et al.J. Clin. Invest. 2004; 114: 937-943Crossref PubMed Scopus (166) Google Scholar). Finally, RAF1 competes with another RAS effector, RASSF1 (Ras association domain-containing protein 1) for the binding to MST2 (mammalian STE20-like kinase 2), preventing MST2 activation and impacting the Hippo pathway. Interaction with MST2 was also shown for ARAF and BRAF. Interestingly, the strength of interaction and inhibition inversely correlates with RAF kinase activity; it is strongest in the case of ARAF, while BRAF interaction with MST2 is extremely weak and possibly depends on RAF1 (reviewed in Nguyen et al., 2015Nguyen L.K. Matallanas D.G. Romano D. Kholodenko B.N. Kolch W. Cell Cycle. 2015; 14: 189-199Crossref PubMed Scopus (21) Google Scholar). The roles of ARAF and RAF1 in MST2 regulation, while intriguing, have not been demonstrated conclusively in vivo. It would thus appear that BRAF is the only RAF kinase lacking a role in protein-protein interaction-based cross-talk with other pathways; it is possible, however, that through its binding to RAF1 or ARAF, BRAF may compete with other interactors, ultimately impacting the cross-talk described above. In support of this hypothesis, in keratinocytes, Braf ablation results in increased complex formation between RAF1 and ROKα (Doma et al., 2013Doma E. Rupp C. Varga A. Kern F. Riegler B. Baccarini M. Cancer Res. 2013; 73: 6926-6937Crossref PubMed Scopus (20) Google Scholar). RAF was discovered in 1983 as an oncogene causing fibrosarcoma in mice. Nearly 30 years later, hundreds of reports elucidated roles of RAF paralogs in cancer, either as drivers or as critical downstream effectors of activated RAS (reviewed in Maurer et al., 2011Maurer G. Tarkowski B. Baccarini M. Oncogene. 2011; 30: 3477-3488Crossref PubMed Scopus (208) Google Scholar, Holderfield et al., 2014Holderfield M. Deuker M.M. McCormick F. McMahon M. Nat. Rev. Cancer. 2014; 14: 455-467Crossref PubMed Scopus (558) Google Scholar), providing an excellent rationale to design inhibitors of this pathway. These reports can be broadly divided in three categories, depending on the approaches taken: (1) large-scale sequencing studies, such as the one that first identified BRAF mutations in melanoma, which establish the identity and frequency of mutations in a specific tumor type or across a panel of cancer cell lines; (2) studies assessing the ability of RAF mutants, expressed in specific cells or tissues, to drive tumorigenesis; (3) studies assessing the requirement for a specific isoform in specific cancers. Somatic RAF mutations in human cancers are unequally distributed among isoforms. While mutations activating BRAF are frequent in hairy cell leukemia (100%), melanoma (50%–60%), and thyroid cancer (40%–60%), RAF1 and ARAF mutations are very rare and are associated with some lung adenocarcinomas and, in the case of ARAF, intrahepatic cholangiocarcinoma (Sia et al., 2015Sia D. Losic B. Moeini A. Cabellos L. Hao K. Revill K. Bonal D. Miltiadous O. Zhang Z. Hoshida Y. et al.Nat. Commun. 2015; 6: 6087Crossref PubMed Scopus (215) Google Scholar). The higher frequency of BRAF mutations compared to ARAF and RAF1 mutations correlates well with BRAF's higher basal activity toward MEK and its simpler mechanism of activation. While one single mutation is sufficient to activate BRAF, other RAF isoforms may require multiple mutations or additional upstream events to reach maximal activation. The most frequently observed mutation, a valine 600 → glutamic acid substitution (BRAFV600E), disrupts the inactive conformation of the kinase and abolishes the requirement for dimerization, inducing persistent activation of BRAFV600E monomers and, consequently, of the MEK/ERK pathway. BRAFV600E is oncogenic in cultured cells and in many tissues, as demonstrated by Cre-inducible expression of the mutant Braf allele. Depending on the tissue, BRAFV600E promotes cancer development (in thyrocytes or in the gastrointestinal tract) or causes benign tumors, which become malignant upon loss of tumor suppressors (lung and melanocytes). Tumor formation is always accompanied by ERK activation and is sensitive to MEK inhibitors. Less common BRAF mutants with impaired kinase activity have also been identified. These mutants can constitutively activate MEK/ERK pathway, in particular in the context of oncogenic RAS, thanks to their intrinsic ability to bind and allosterically activate RAF1. Accordingly, in the only in vivo study of this sort, melanocyte-restricted expression of the kinase-impaired, dimerization-competent BRAFD594A mutant is tumorigenic only when combined with an activating KRAS mutation (Heidorn et al., 2010Heidorn S.J. Milagre C. Whittaker S. Nourry A. Niculescu-Duvas I. Dhomen N. Hussain J. Reis-Filho J.S. Springer C.J. Pritchard C. Marais R. Cell. 2010; 140: 209-221Abstract Full Text Full Text PDF PubMed Scopus (1191) Google Scholar). Unlike BRAF, ARAF and RAF1 mutations are rare in human cancers. A S214A-activating ARAF mutation has been identified in lung adenocarcinoma patients, and S214A ARAF has been shown to transform immortalized airway cells (Imielinski et al., 2014Imielinski M. Greulich H. Kaplan B. Araujo L. Amann J. Horn L. Schiller J. Villalona-Calero M.A. Meyerson M. Carbone D.P. J. Clin. Invest. 2014; 124: 1582-1586Crossref PubMed Scopus (85) Google Scholar). Two further ARAF mutations, G322S in the kinase domain and N217I in the CR2, have been identified in intrahepatic cholangiocarcinoma. Both constitutively activate MEK/ERK but fail to transform NIH 3T3 cells (Sia et al., 2015Sia D. Losic B. Moeini A. Cabellos L. Hao K. Revill K. Bonal D. Miltiadous O. Zhang Z. Hoshida Y. et al.Nat. Commun. 2015; 6: 6087Crossref PubMed Scopus (215) Google Scholar). Thus, their oncogenic potential is unclear. Beyond activating mutations, ARAF has been proposed to act as a scaffold protein by binding to BRAF and stabilizing BRAF:RAF1 heterodimers, particularly in the presence of RAF inhibitors (Rebocho and Marais, 2013Rebocho A.P. Marais R. Oncogene. 2013; 32: 3207-3212Crossref PubMed Scopus (51) Google Scholar). The overall frequency of RAF1 mutations in cancer is below 1% (http://www.cbioportal.org). The most frequent, S257 and S259 mutations, were reported in several tumor types, including lung adenocarcinoma and colorectal cancer (Imielinski et al., 2014Imielinski M. Greulich H. Kaplan B. Araujo L. Amann J. Horn L. Schiller J. Villalona-Calero M.A. Meyerson M. Carbone D.P. J. Clin. Invest. 2014; 124: 1582-1586Crossref PubMed Scopus (85) Google Scholar). Intriguingly, S259 mutations prevent phosphorylation of the residue that restrains RAF1 membrane recruitment by RAS, making RAF1 more susceptible to activation. While mutations are rare, overexpression of RAF1 occurs in bladder cancer, hepatocellular carcinoma, squamous cell carcinoma, and lung adenocarcinoma (Maurer et al., 2011Maurer G. Tarkowski B. Baccarini M. Oncogene. 2011; 30: 3477-3488Crossref PubMed Scopus (208) Google Scholar). Based on the results of these studies, the roles of BRAF and RAF1 in cancer are just as distinct as in development, with BRAF driving MEK/ERK activation and RAF1 performing other essential functions. This hypothesis has been tested in RAS-driven models of epithelial cancers combined with conditional ablation of either BRaf or Raf1. Both paralogs are essential for development and maintenance of RAS-induced epidermal tumors. BRAF promotes ERK activation and cell proliferation, and its ablation halts tumor growth (Kern et al., 2013Kern F. Doma E. Rupp C. Niault T. Baccarini M. Oncogene. 2013; 32: 2483-2492Crossref PubMed Scopus (24) Google Scholar). In contrast, the role of RAF1 is ERK independent and relies on its ability to bind and inhibit ROKα, preventing keratinocytes dedifferentiation. Raf1 ablation rapidly induces keratinocyte differentiation, causing complete tumor regression in this system (Ehrenreiter et al., 2009Ehrenreiter K. Kern F. Velamoor V. Meissl K. Galabova-Kovacs G. Sibilia M. Baccarini M. Cancer Cell. 2009; 16: 149-160Abstract Full Text Full Text PDF PubMed Scopus (82) Google Scholar). The discrepancy between the two Raf isoforms is even more evident in a KRAS-driven model of NSCLC, in which BRAF is completely dispensable while RAF1 plays an essential role that, despite being unknown, was shown to be MEK independent (Blasco et al., 2011Blasco R.B. Francoz S. Santamaría D. Cañamero M. Dubus P. Charron J. Baccarini M. Barbacid M. Cancer Cell. 2011; 19: 652-663Abstract Full Text Full Text PDF PubMed Scopus (224) Google Scholar, Karreth et al., 2011Karreth F.A. Frese K.K. DeNicola G.M. Baccarini M. Tuveson D.A. Cancer Discov. 2011; 1: 128-136Crossref PubMed Scopus (106) Google Scholar). Thus, RAF1 has essential but MEK-independent roles in these two models. RAF1, however, is dispensable for the development of KRAS-driven pancreatic cancer in which the PI3K/PDK1 axis is essential. The role of BRAF has not been tested in this system, but it has been previously shown that BRAFV600E can drive tumorigenesis in the pancreas (Eser et al., 2014Eser S. Schnieke A. Schneider G. Saur D. Br. J. Cancer. 2014; 111: 817-822Crossref PubMed Scopus (374) Google Scholar). Together, these results are consistent with the idea that BRAF is the entry point in the MEK/ERK pathway and promotes proliferation either downstream of RAS or, when mutated, as an oncogenic driver. In contrast, RAF1 mutations are rare, but at least in skin and lung epithelia, the impact of their ERK independent functions on RAS-driven tumorigenesis is profound. The deregulation of the RAF/MEK/ERK signaling observed in human cancer has boosted the hunt for drugs specific to this pathway. RAF inhibitors have been developed and tested; some have gained FDA approval for treatment of specific cancers, possibly in the context of personalized medicine. Advantages and drawbacks of RAF inhibitors have been recently reviewed (Lito et al., 2013Lito P. Rosen N. Solit D.B. Nat. Med. 2013; 19: 1401-1409Crossref PubMed Scopus (419) Google Scholar, Holderfield et al., 2014Holderfield M. Deuker M.M. McCormick F. McMahon M. Nat. Rev. Cancer. 2014; 14: 455-467Crossref PubMed Scopus (558) Google Scholar, Cseh et al., 2014Cseh B. Doma E. Baccarini M. FEBS Lett. 2014; 588: 2398-2406Crossref PubMed Scopus (75) Google Scholar). Vemurafenib and dabrafenib, which specifically target BRAFV600E monomers, have brought unprecedented clinical benefits to patients with BRAFV600E-expressing melanoma. They are also used in hairy cell leukemias refractory to conventional treatment with purine analogs and are currently being tested in BRAFV600E-positive NSCLC patients. Unfortunately, these inhibitors are ineffective in other BRAFV600E-expressing malignancies such as metastatic colorectal carcinoma and thyroid cancer. In the latter, a negative feedback loop attenuating growth factor receptor (e.g., EGFR) signaling is lost upon inhibitor treatment; as a result, the attainable level of ERK inhibition never reaches the 80% required for therapeutic effects. This feedback reactivation is not observed in melanomas, which express low levels of EGFR, unless they upregulate this receptor. Thus, the presence of the BRAFV600E mutation does not always guarantee responsiveness to BRAF inhibitors, and the cellular context in which they are used must be considered as a key factor. In addition, most tumors develop resistance to BRAF inhibitors within months. Multiple mechanisms of acquired resistance exist. Cells escape RAF inhibition by increasing expression of receptor tyrosine kinases, which can reactivate ERK or signal through parallel survival pathways such as PI3K-AKT, diminishing the dependence on ERK pathway. ERK reactivation mechanisms by pathway components include BRAFV600E amplification or the expression of inhibitor-insensitive splice variants, as well as RAF1 amplification or mutational activation of NRAS, which preferentially signals through RAF1. Alternatively, BRAF can be bypassed by activating MEK mutations or by overexpressing COT (cancer Osaka thyroid oncogene) kinase, which activates MEK in a RAF-independent manner. Another undesirable consequence of RAF inhibition is paradoxical ERK activation. This is observed in BRAF wild-type cells in the presence of activated RAS and is due to the fact that the inhibitors, unless present in saturating concentrations, promote RAF dimerization and therefore activation. This mechanism is responsible for the occurrence of skin tumors in patients treated with vemurafenib (20%–26%) and dabrafenib (6%–11%). To prevent MEK/ERK activation, a combination of dabrafenib and the MEK inhibitor trametinib has been approved for the treatment of BRAFV600E/K melanoma. Unfortunately, melanomas can circumvent RAF/MEK inhibition by combining mechanisms of resistance to BRAF inhibitors. Intriguingly, double-resistant cells are drug addicted, which may be exploited for therapy schedules involving intermittent cycles of inhibitor (Moriceau et al., 2015Moriceau G. Hugo W. Hong A. Shi H. Kong X. Yu C.C. Koya R.C. Samatar A.A. Khanlou N. Braun J. et al.Cancer Cell. 2015; 27: 240-256Abstract Full Text Full Text PDF PubMed Scopus (247) Google Scholar). New efforts are required to develop improved strategies that combine the excellent clinical response of RAF inhibition with reduced side effects and prolonged efficacy. A new generation of inhibitors termed "paradox breakers" have been designed and tested in preclinical settings. Two compounds (CCT196969 and CCT241161), which inhibit RAF1 and BRAFV600E (but not wild-type BRAF), as well as SRC-family kinases, which are downstream of receptor tyrosine kinases and contribute to ERK reactivation in resistant cells, showed improved efficacy in BRAF and NRAS mutant melanoma. The same drugs also inhibit the growth of patient-derived, BRAF mutant melanoma xenografts resistant to the vemurafenib analog PLX4720, implying that they may be beneficial as second-line treatment (Girotti et al., 2015Girotti M.R. Lopes F. Preece N. Niculescu-Duvaz D. Zambon A. Davies L. Whittaker S. Saturno G. Viros A. Pedersen M. et al.Cancer Cell. 2015; 27: 85-96Abstract Full Text Full Text PDF PubMed Scopus (150) Google Scholar). It is clear that the RAF paralogs have very distinct essential functions in spite of their synergy in ERK signaling. BRAF, the main ERK activator, is the target of somatic mutations that increase its enzymatic activity and can drive tumorigenesis in a number of systems. RAF1, in contrast, is not essential for ERK activation. Its function in development and tumorigenesis is mainly the protein-protein interaction-based cross-talk with other signaling pathways, which generates non-oncogene addiction in RAS-driven models. Consistently, RAF1 is not the target of activating mutations, but it is often found overexpressed in tumors.