Raf Kinases: Function, Regulation and Role in Human Cancer

2.1. Insights from knockout and transgenic mice models

Knockout studies of individual Raf isoforms in mice point to their essential and non-overlapping roles during embryo- and organogenesis, inasmuch as knockout of each of the Raf isoforms results in a lethal phenotype in most of the examined mice strains [16–21]. The lack of functional overlap is also apparent from phenotypic differences among the various Raf knockout mice during embryogenesis [13, 15]. In general, Raf knockout mice display severe growth retardation of up to 50% in size and weight in addition to abnormal development of the liver, vascular and neuronal tissues.

The first C-Raf knockout description, which showed a lethal phenotype at embryonic day E10.5 to E12.5, expressed low amounts of an N-terminal truncated C-Raf protein that maintained some kinase activity, thus complicating the interpretation of the data [18]. In follow up knockout studies where the C-Raf protein expression was completely abolished, the mice showed embryonic lethality and poor development of the placenta, liver, and haematopoietic organs. These studies also suggested that C-Raf is required to restrain apoptosis during embryogenesis [20, 22]. Interestingly, activation of the MAPK pathway in whole embryos and cell lines derived form these knockout embryos appeared normal, though activation in specific cell types has not been examined. This finding prompted the authors to suggest that B-Raf could be the one mediating MAPK activation in this setting and that other C-Raf targets besides the MAPK pathway may be responsible for the observed lethal phenotype. Supporting this notion, a “knock in” mutation in the C-Raf gene that results in decreased kinase activity did not have a significant phenotype in the mice, suggesting that C-Raf kinase activity is not crucial for its function in the murine model [22]. This claim needs further confirmation, however, since the mutation used in this study, YY340/341FF, has only a moderate effect on C-Raf kinase activity [23] and the residual activity may suffice for its function. A more definitive examination will require a “knock in” of a mutation that completely abolishes the kinase activity, for example, a mutation in the ATP binding pocket such as K375M. In general, since residual expression of incomplete C-Raf transcripts coding for various C-Raf protein fragments in the knockout models could significantly affect data interpretation, it would be highly significant to extensively verify C-Raf expression in these mice at both the mRNA and protein levels for absence of any such variants.

As for other Raf isoforms, B-Raf knockouts die in utero by embryonic day E12.5 displaying vascular and neuronal defects. Massive hemorrhaging spills into major body cavities due to apoptotic death of differentiated endothelial cells results in the death of the mice during mid-gestation [17, 19]. In addition, these mice also show severe neuronal abnormalities [21].

A-Raf knockout mice are born alive but show severe intestinal and neurological abnormalities [16]. In the C57 Bl/6 background, the mice die within 7–21 days after birth. However, this lethal phenotype is reversed in mice outbred with the 129/OLA genetic background, which survive to adulthood but display several neurological pathologies. The phenotypic variance seen with the different mice strains has not been established yet and is also observed to a lesser extent with the C-Raf knockouts.

Collectively, these studies suggest that all Raf isoforms are required for maintaining normal embryogenesis and sustaining life. However, the question regarding the exact role of each isoform in embryogenesis remains open. The current data suggests that the Raf isoforms share both redundant and non-redundant functions. For example, single Raf knockouts result in relatively moderate effects on embryonic development until embryonic day E10.5, suggesting that Raf isoforms can compensate to some extent for each other until this developmental stage. Supporting this notion, double B-Raf/C-Raf knockouts display more severe phenotypes than the single knockouts, resulting in underdeveloped embryos, which grow into small clumps of cells representing two distinct lineages resembling the inner cell mass and the trophoblast cells of the mouse blastocyst [19]. This study also determined the expression patterns of the three Raf isoforms, showing that all three isoforms are equally expressed throughout the embryonic development and in the adult mouse, disavowing previous observations indicating tissue- and development-specific isoform expression patterns. Similar to the above, double knockouts of A-Raf and C-Raf hold a more severe phenotype than either gene alone, with the double knockouts dying by E10.5, looking small and abnormal [24].

In a more recent study, Galabova-Kovacs et al [25] demonstrated using conditional B-Raf and C-Raf knockouts that B-Raf is required mainly for placenta development, possibly through controlling VEGF expression, and that it serves as the main MAPK activator in this organ. C-Raf on the other hand, appears important for ERK activation in the embryo proper and also for erythrocyte differentiation. Interestingly, this work also did not detect a significant reduction in phospho-ERK levels in embryos lacking B-Raf expression or significant effects on cell proliferation rates. These findings indicate that data obtained using mouse embryonic fibroblast regarding the role of the different Raf isoforms in MAPK activation is not recapitulated in the whole organism, reputing previous claims that B-Raf is the sole physiological MAPK activator. Thus, more detailed examination of the individual roles of the three Raf isoforms in specific tissues will be required to gain a better idea regarding the complex functions of each Raf isoform.

In parallel to the knockout studies, genetically modified transgenic mice were used to examine the role of the Raf-MAPK pathway activation in embryogenesis and development. Targeted overexpression in the lung of wild-type or a constitutively active C-Raf form containing a deletion of the regulatory N-terminal domain, called BXB-Raf resulted in normal mouse embryogenesis and development. However, both transgenics developed lung adenocarcinomas. The BXB-Raf transgenics displayed accelerated tumor development in comparison to wild-type C-Raf, pointing to a higher oncogenic potential of the active C-Raf protein [26, 27].

Transgenic mice expressing a conditional knock-in of a constitutively active B-Raf allele (^V600EB-Raf, an activating mutation occurring naturally in human cancers) during embryogenesis under the control of a Cre/Lox system, did not survive embryo development, with embryos dying at E7.5, indicating that expression of a constitutively active B-Raf allele during embryogenesis is lethal [28]. When the Cre/Lox ^V600EB-Raf mice where bred to induce B-Raf expression under the control of an IFN-inducible promoter, the mice survived embryogenesis but died before reaching 4 weeks of age. Thus, even at low conditional ^V600EB-Raf knock-in levels the effect was lethal. The transgenic mice showed expression of the transgene in various tissues including spleen, liver, lung, heart and kidney, which correlated with increased B-Raf kinase activity in these tissues. Though the cause of death was not provided, the mice displayed several hyperproliferative and developmental abnormalities. For example, the spleen was greatly enlarged, the liver showed both hyperproliferation and increased apoptotic regions and the bone marrow had diminished expression of erythrocytes. There were no apparent developmental defects in the lung, heart or kidney. Interestingly, though ^V600EB-Raf expression and increased B-Raf activity were detected in the lung and thymus, there was no increase in ERK phosphorylation. In contrast, both spleen and liver displayed a significant increase in ERK phosphorylation. Thus, there appeared to be a correlation between the pathological effects and the level of ERK phosphorylation induction. MEF cells derived from the ^V600EB-Raf Cre/Lox mice induced to express the mutant B-Raf allele by transfection of Cre displayed increased proliferation and some characteristics of cell transformation, such as the loss of contact inhibition and the ability to form colonies in soft agar. Thus, expression of an active B-Raf allele from the intrinsic locus can cause hyper-proliferation and induce transformed characteristics in certain cell lineages. It remains to be determined what controls these cell specific effects, a key question in understanding the oncogenic potential of B-Raf. This study did not examine the expression levels of the mutant B-Raf allele protein in the various tissues; therefore, it is not known whether some of the differences observed in these tissues reflected possible differences in protein expression levels.

2.2. Insights from tissue culture models

In the past 20 years there have been several thousand publications relating to Raf function using various culture models, thus we provide here only a sampling of these studies. Overall, key roles of Raf have been demonstrated in cell proliferation, differentiation, cell cycle progression, cell survival and several other physiological processes (Fig. 2). For example, using pharmacological inhibitors for Raf and MEK, several studies demonstrated that blocking the MAPK pathway attenuates cell cycle progression mainly by arresting cells at the Go/G1 boundary ([29–31] and our unpublished observation). However, as noted above, the role of individual Raf isoforms in this process is less clear and may depend on cell types or developmental stages. As for cell cycle targets, many have been proposed including cyclin D1 [24], Rb [32] and CDC25 [33] among others [34].

Regarding cell proliferation, expression of active Raf forms, either B-Raf or C-Raf, have pro-proliferative effects, inducing uncontrolled proliferation and transformation, and as noted above, even at low conditional gene knock-in levels of the ^V600EB-Raf mutant allele, transgenic mice display hyper-proliferative disorders [28, 35–37]. As to the role of specific Raf isoforms in cell proliferation, the data is less straightforward. For example, in Ras-transformed melanocytes, B-Raf depletion by siRNA does not block the MEK-ERK pathway or cell cycle progression [38].

The role of Raf in cell survival and apoptosis is a topic of a separate review in this issue (Galmiche et al). In general, Raf proteins are regarded as pro-survival antiapoptotic factors. These functions can be mediated by both MEK-ERK dependent pathways, involving Raf kinase activity, and at least for C-Raf, some of the data points to kinase independent functions involving regulation of proteins such as ASK-1 [39], MST2 [40, 41] and BAD/Bcl family members [42–44]. In addition, a recent study demonstrated a role of C-Raf in preventing apoptosis in cardiac myocytes during induced cardiac ischemia [14, 45].

The role of the Raf-MAPK pathway in differentiation was originally examined in PC12 cells and showed a role both in cell differentiation and cell proliferation; a distinction that depended on MAPK activation levels versus its duration [46]. However, in other cell lines, for example erythroblasts, C-Raf delays differentiation by inhibiting caspase activation [47]. In epidermal development, C-Raf activation can induce terminal differentiation of keratinocytes [48]. However, the specific role of C-Raf in this process is not clear since epidermis-specific C-Raf knockout does not affect epidermis architecture or follicle development [49]. This topic requires more work since the knockout models suggest a complex role of the three Raf isoforms in various stages of development and different cell lineages.

Other reported cellular Raf functions include roles in cell motility, migration and wound healing [49, 50], T cell development and anergy [51, 52], and B cell development [53].

2.3. Insights from invertebrate studies

Raf homologues were discovered in various organisms such as invertebrates, birds, bony fish, and amphibians and many of the initial Raf studies were performed in invertebrates. Whereas mammals express three isoforms of Raf, invertebrates have only one Raf isoform resembling the mammalian B-Raf isoform; D-Raf in Drosophila [54] and Lin-45 in Caenorhabditis elegans [55] (Fig. 1) making for simpler genetic and overall analysis. This lack of redundancy is also evident along the signaling module as a whole, where only one MEK and ERK are present in Drosophila. D-Raf was found to have an important role during Drosophila embryogenesis and several phenotypes appear in D-Raf knockout flies. Most significant, the fly embryos fail to differentiate into structured embryos and, like in mice, display a lethal phenotype. These studies also showed a role of Raf in determining the dorsal/ventral polarity of the fly eggs [56–59]. Genetic screens for activating and inactivating allelic mutations also contributed significantly to the Raf study field. For example, mutations in the CR1 domain (R217L), now known for the RasGTP binding domain, showed a decrease in D-Raf activity and severe morphological disorganization of the fly eye [60]. As for gaining pathway understanding, studies in Drosophila identified D-Raf as a downstream target of the tyrosine kinase receptors Torso and Sevenless, which play an important role in fly development. These and other Drosophila EGF receptor (DER) signaling pathways helped in defining the mammalian pathway known as the EGF receptor-Grb2-SOS-Ras-Raf pathway [61–63]. In C.elegans, lin-45 was found downstream of the mammalian homologue of Ras, let60 [55]. These initial genetic studies helped define the role of Raf and the MAPK pathway.

More recent work helped in identifying other components that modulate this pathway. Therrien et al. discovered a potential Raf scaffold protein, kinase suppressor of Ras, KSR [64] and a connector enhancer of KSR, CNK [65], both evolutionary conserved and are discussed below in Raf Regulation.

Besides the role of Ras in Raf activation, other mechanistic aspects of Raf regulation are also conserved across species. For example, C-Raf phosphorylation at a S338 site by PAK is conserved in Drosophila, though Lin-45 holds a histidine at this site (Fig. 1). Also, two potential C-Raf activating phosphorylation sites, T491 and S494 have homologues in lin-45 and their substitution with aspartic acids results in a multi-vulval phenotype, indicating activation of the MAPK pathway [66].

2.4. Physiological Raf targets and effectors

Both biochemical and genetic studies established a linear phosphorylation cascade downstream of Raf where Raf phosphorylates and activates MEK-1/2 that in turn phosphorylates and activates ERK-1/2 [6]. This initial view of Raf function has been repeatedly contested, however, with the identification of so called MEK-independent Raf functions in the past 6–8 years [41, 67]. Interestingly, some of the attributed Raf functions are proposed to be independent of kinase activity, proposing scaffold and/or structural modulator functions for Raf [15, 39, 40] (for a recent review, see Galmiche et al in this issue). Many of the MEK-independent Raf functions were determined in experiments using MEK inhibitors or Raf mutants that abrogate kinase activity. Only a very few studies identified Raf targets other than MEK. Among the emerging new Raf targets are proteins involved in programmed cell death control such as ASK-1 [39] and MST-2 [40], proteins involved in cell cycle regulation such as RB [32] and Cdc25 [33], cell survival, e.g., NFKB [68] and BAD/Bcl family members [42–44], and proteins involved in cytoskeletal organization such as vimentin [69], Rokα [49] and keratin 8 [70]. Though some of these studies provide solid evidence in cell culture models, their relevance in more physiological contexts is yet unknown. In addition, the exact role of Raf in regulating these targets and the associated mechanisms remain to be determined. Thus, awaiting further confirmation of MEK independent Raf functions, MEK remains the only well established physiological Raf target.

3.1. Role of C-Raf as a prime oncogene

The only naturally occurring transforming form of C-Raf described so far is a viral-transmitted form that has a deletion at the N-terminal regulatory region, resulting in a constitutively active kinase [1, 3, 4]. This form can transform NIH 3T3 and other cell lines in culture, resulting in various cancerous phenotypes and the ability to form tumors in nude mice. It also induces tumor formation when expressed in transgenic animals [10, 26, 27, 75, 76]. Other mutations, including point mutations that result in a constitutive kinase also display transforming activity in cell culture models, however, to date, these mutations have not been identified in primary cancers [10, 77–79]. Thus, C-Raf has not been considered as a de facto cellular oncogene. This notion has been somewhat questioned recently with the identification of several naturally occurring C-Raf mutations showing elevated kinase and transforming activity (e.g., E478K, S427G and I448V) [78, 80]. However, further examination as to the prevalence of these mutations in human cancers and determining their role in the tumorigenic process is needed before establishing the significance of these mutations and excluding the possibility of allelic alternations or random mutations.

Besides activating mutations, attention has been given also to the expression status of the wildtype C-Raf form. The initial studies were unable to demonstrate C-Raf activation or transforming activity as a result of its overexpression in a variety of cell culture models [36]. However, more recent studies targeting C-Raf overexpression to the lung demonstrated its transforming potential, suggesting that under certain circumstances C-Raf overexpression can display a tumor promoting or initiating potential [26, 27, 75]. Providing physiological relevance to these studies, C-Raf has been found overexpressed in a variety of primary human cancers, such as lung, liver, prostate, primitive neuroectodermal tumors, myeloid leukemia and head and neck squamous cell carcinoma [75, 81–85]. Also here, a more thorough examination of C-Raf expression status in human cancers and its role in the oncogenesis process is needed to determine the true significance of these initial observations.

3.2 Role of C-Raf in mediating Ras, growth factor receptor and other oncogenes effects

In contrast to the uncertainty regarding the role of C-Raf as a prime oncogene, its significance in mediating oncogenic signals in a large number of human cancers is well established [6, 86–95]. Being the primary target downstream of Ras, which is found mutated in up to 30% of all human cancers, the key role of C-Raf in human cancer is readily evident. This role has been established in studies using a variety of agents that block C-Raf function or expression (e.g., anti-sense RNA, small interfering RNA and small molecule kinase inhibitors), demonstrating reversal of the transformed phenotype induced by oncogenic Ras [6, 86, 91, 92, 96–101]. In addition, the role of Raf in mediating Ras-induced transformation was demonstrated in studies using Ras mutants that bind exclusively to Raf proteins [91].

Besides its role in mediating oncogenic Ras effects, many studies established a role of C-Raf also in mediating the transforming effects of a variety of active growth factor receptors, oncogenes and mitogens [86, 94, 102, 103]. For example, C-Raf was found to be the predominant Raf isoform mediating growth factor-stimulated proliferation in a large panel of ovarian cancer cell lines [94] and a dominant negative C-Raf mutant was shown to prevent v-Src-induced transformation [104]. C-Raf was also shown to cooperate with other oncogenes such as v-myc and v-Ras in cell transformation, as well as in cases involving deregulated tumor suppressor gene expression, such as p53 and RB [32, 96, 105, 106]. In addition, several recent studies suggest that some of the effects of activating/oncogenic B-Raf mutations could be mediated through C-Raf [107, 108].

C-Raf has been implied also in cell invasion. A recent study demonstrated that metastatic prostate cancer cell lines have decreased expression of the Raf inhibitor protein RKIP, a protein that attenuates C-Raf/MEK interaction. These cells displayed elevated activation of ERK, which correlated with increased tumor invasiveness [109, 110]. This phenomenon has also been recently observed in hepatomas [111].

3.3. Role of B-Raf as a prime oncogene

As mentioned above, B-Raf has been the second Raf isoform to be identified and since in contrast to the ubiquitous C-Raf expression was considered initially to be mainly a brain specific isoform, much of the early research focus has been on C-Raf. This focus shifted however to some degree with the identification of B-Raf in 2002 as a mutational target in various human cancers [9, 37, 71, 112]. As of now, B-Raf mutations have been detected in 70% of melanomas, 30% of thyroid cancers, 15% of colon cancers and at lower frequencies in several other cancer types [9, 28, 78]. Most of the mutations were identified within the kinase domain, with a single amino acid substitution, V600E, accounting for up to 90% of the mutations. This mutation results in a constitutive activation of the kinase domain and correlates in vivo with constitutive activation of its downstream targets MEK and ERK (both in experimental models and pathological cancer samples) [38, 71, 113]. Interestingly, it appears that one hit to the MAPK pathway is enough for oncogenesis since the current data shows very little overlap of B-Raf and Ras activating mutations in cancer samples. For example, a recent study in colorectal cancers found 36% B-Raf mutations and 18% Ras mutations, yet each mutation occurred exclusively [114]. Similarly, in 21 gallbladder carcinoma specimens examined by PCR the B-Raf activating mutation occurred without the Ras mutation and vise versa [115].

Interestingly, a corresponding mutation to B-Raf V600E was not detected in C-Raf or A-Raf in human cancers. This is most likely due to the fact that C-Raf and A-Raf hold a tighter regulation of their kinase domain than B-Raf, thus, a single mutation is insufficient for a substantial kinase activation [78, 79, 116, 117]. As elaborated below, C-Raf activation requires phosphorylation at serine 338/339 and tyrosine 340/341 at the aminoterminal region of the kinase domain, whereas in B-Raf the comparable serine residue at 445 holds constitutive phosphorylation and residues 448/449 hold phosphomimetic aspartic acids, priming B-Raf for activation. A-Raf appears to be regulated in a similar fashion as C-Raf.

Regarding the transforming effects of B-Raf mutations, the opinions vary to some extent. Some of the data indicates a strong oncogenic activity[28, 37, 38, 118], however, the identification of B-Raf mutations also in pre-malignant lesions, such as benign nevi [113, 118, 119], points to the possibility that B-Raf mutations may be insufficient by themselves to drive transformation and that other cooperating alternations have to occur for a fully transformed phenotype.

The involvement of B-Raf in mediating oncogenic signals downstream of active Ras or growth factor receptors has not been sufficiently examined yet. However, several initial studies point to a more dominant role of C-Raf than B-Raf in these systems [94, 95, 101]. More detailed examination of various cancer types and models will be required to get a better understanding of this key question which may have significant implications in cancer treatment and future drug development.

3.4. Therapeutics targeting Raf inhibition

This topic has been covered recently in several excellent reviews [11, 72, 73, 98]. Shortly, several approaches have been undertaken through the years for achieving Raf inhibition. These included targeting the Raf kinase activity using small molecule inhibitors, targeting Raf expression using RNAi and antisense RNA molecules, and targeting the Ras-Raf interaction. Of the small molecule inhibitors, BAY 43-9006 (Sorafenib) is now under advanced clinical trials [93, 120]. Another small molecule Raf inhibitor, SB-590885 has been recently reported from GlaxoSmithKline, however, its potency in a whole organism remains to be evaluated [31]. As for BAY 43-9006, crystallography data indicates its binding to the B-Raf kinase domain and stabilization of the inactive conformation [107]. As regards to the biological functions of BAY 43-9006, not all of its in vivo effects are attributed to Raf inhibition, since it also inhibits the VEGF and PDGF receptor families. Thus, some of the anti-tumor effects may be mediated through inhibition of angiogenesis and tumor vascularization. Phase I/II clinical trials established the maximum tolerated dose at 400 mg/kg/d of continuous dosing and demonstrated a potential for disease stabilization and tumor shrinkage in patients with advanced kidney cancer [120–122]. Currently, combination trails are underway using BAY 43-9006 with carboplatin, paclitaxel and other chemotherapeutic agents, including a Phase III trial for advanced melanoma.

Regarding the targeting of Raf expression, various antisense and RNAi compounds have been examined. For example, ISIS 5132 which is a 20-base phosphorothioate oligonucleotide antisense RNA for human C-Raf showed efficient tumor inhibition in animal models and in several past trials also some disease stabilization in ovarian carcinoma [123]. More recent Phase II trials, however, showed no benefit when given as a single agent. These trials included epithelial ovarian cancers [124], colorectal carcinoma [125], small-cell lung cancer (SCLC) or non-small cell lung cancer (NSCLC) [126], and hormone-refractory prostate cancer [127]. It is likely that future cancer therapeutic strategies will combine one or more molecular targeting agents such as ISIS 5132 with conventional chemotherapies to improve patient outcome. In general, a significant shortfall in the above studies was that many of them did not examine patient tumors for MAPK pathway activation, before and after treatment. Therefore, there is no knowledge of whether the administered drug actually targeted Raf or the MAPK pathway.

The success in developing drugs targeting Ras-Raf interaction has been very limited and no clinical data is currently available.

4.1. Raf regulation by phosphorylation

As noted above, most of the studies regarding Raf regulation have been done with C-Raf, thus we use here the numbering of C-Raf in referencing the known phosphorylation sites. In resting cells, C-Raf is inactive and is found phosphorylated at several sites, considered to be the basal C-Raf phosphorylation sites. These sites include S43, S259, S621 and several other sites that remain to be identified [142, 151, 157–159]. S259 and S621 phosphorylations serve for generating binding sites for 14-3-3 [138, 150] and S43 is thought to be a negative regulatory site targeted by PKA [160–162]. Following cell stimulation with growth factors, C-Raf is activated, concomitant with its increased phosphorylation. These phosphorylations are believed to maintain the active C-Raf conformation, since treatment of active C-Raf with phosphatases results in a complete inactivation of its kinase activity [151, 163]. The identities of some of the growth factor-induced sites have been established, yet the question remains whether other unidentified sites can contribute to C-Raf activation.

To date, the major identified catalytically significant, growth factor-induced C-Raf phosphorylation sites are located within a region outside the actual catalytic domain and involve S338/9 and Y340/1 [164–167]. Phosphorylations at these sites can be mediated by the p21 Rac-activated kinase family (PAK) and the Src family of tyrosine kinases respectively. These phosphorylations are thought to release the inhibitory effect of the N-terminal part on the catalytic domain, stabilizing the active form of C-Raf [140, 142, 168]. However, this model is somewhat questioned by the fact that these residues are also required for the activation of the truncated C-Raf form that lacks the N-terminal regulatory region [150]. Interestingly, the S338/339 sites are conserved among most of the known Raf proteins (Fig. 1), with the exception of A-Raf that has a glycine at the corresponding S339 site and the C. elegans Raf, Lin-45 that has histidines at its corresponding sites. The YY340/341 motif has more variations, with B-Raf, D-Raf and Lin-45 having phosphomimetic, charged residues (Glu or Asp) at their corresponding sites. These differences are believed to prime these kinases for activation, thus resulting in kinases that require less activation-related modifications for full activation [9, 78, 117]. This fact, however, combined with the report that B-Raf S446 (corresponding to C-Raf S338) is found constitutively phosphorylated in resting cells, where the kinase is inactive, suggest that phosphorylation at these sites, though may be an essential step in the activation process, does not activate Raf per se. Thus, the activating Raf phosphorylation site(s), if exist, remains elusive.

A significant breakthrough in understanding Raf kinase regulation came with the identification of cancer related mutations in B-Raf [71]. Though many mutations have been reported thus far, some with holding elevated kinase activity and others not, the large majority of the mutations were confined to a single amino acid, V599 (due to an initial sequencing error, V600 was assumed as V599; more current literature cites the site as V600, thus we use here the new nomenclature). The mutations identified in V600 were substitutions to charged residues, i.e., Glu, Asp or Lys, resulting in a constitutively active kinase displaying an oncogenic potential [38, 107, 169, 170]. As noted above, however, these mutations do not necessarily lead to a full malignant transformation in humans [113, 118, 171]. The ability of the V600E substitution to activate the B-Raf kinase, combined with data from the kinase domain crystal structure showing that this mutation forces a conformational change that allows a more open configuration and potentially a better access for the substrate, suggest that phosphorylations at proximal sites to V600 may be responsible for Raf activation in response to growth factor stimulation under normal conditions [9, 107]. The generality of this proposed activation mechanism when considering the other Raf isoforms remains to be determined, especially in light of the fact that activation of the other isoforms involves a more complex mechanism. Nevertheless, several phosphorylation sites in close proximity to B-Raf V600 (V492 in C-Raf) have been reported in both in B-Raf and in C-Raf (Fig. 1) [66, 151, 172, 173]. These sites include C-Raf S497/9 [172], two phosphorylation sites that were identified initially in B-Raf, Thr599 and S602 [173] and confirmed in C-Raf (T491/S494) [66] and S471, identified initially in C-Raf [151].

S497 and S499 were reported as activation sites targeted by PKC [172], however, this notion has been later challenged [174] and the role of these sites in C-Raf activation by growth factors, Ras or even TPA appears to be limited [23].

The B-Raf T599/S602 sites were identified by Zhang et al [173] and confirmed using phosphospecific antibodies developed against these sites. These phosphorylations are not present on B-Raf isolated from resting cells and are induced by RasV12. Substitution of these sites with alanine results in an inactive kinase and their substitution with phosphomimetic charged residues results in a constitutive kinase, implying that these sites are potential de facto B-Raf activation sites. These sites are also conserved in C-Raf; however, the data confirming their phosphorylation or demonstrating a significant role of these sites in C-Raf regulation is limited [66] and subject to debate [23]. The physiological regulation and significance of these sites in vivo, both for B-Raf and C-Raf, remains unknown. C-Raf S471 was identified by mass spectrometry as a growth factor-induced C-Raf phosphorylation site and mutational analysis demonstrated that its substitution with several other residues; Ala, Cys, Glu or Asp results in an inactive kinase, but substitution with Thr preserved some of the kinase activity [151]. Similarly to B-Raf T599/S602 sites, C-Raf S471 is conserved among all known Rafs except Lin-45 that has a Thr at its corresponding site (T605, fig. 1). Mutations of the corresponding sites in B-Raf or Bxb-Raf also resulted in inactive kinases that were unresponsive to growth factor activation. Importantly, this inactivation was restored in B-Raf by the V600E substitution, suggesting that phosphorylation at this site is important for growth factor induced Raf activation but is dispensable when V600 contains a charged residue.

Thus, both of the above studies imply that phosphorylation within this CR3 region is important for Raf activation. However, more evidence supporting the identity of these phosphorylation sites and the identification of a kinase(s) responsible for their phosphorylation, capable of Raf activation is needed to fully establish their role in Raf regulation. Furthermore, the physiological roles of these phosphorylations under normal and transformed conditions remain to be examined.

In addition to the search for activating Raf phosphorylation sites, several reports recently described regulatory Raf phosphorylation sites that, while not directly affecting the catalytic activity of Raf, participate in Raf regulation by affecting its inactivation rate [158, 159, 175]. Albeit these sites were described and confirmed independently by several groups, the functional role of these phosphorylations remains controversial. Combined, these studies reported a total of 5 novel C-Raf phosphorylation sites: S29 at the N-terminal regulatory region, a stretch of three phosphorylation sites close to the CR2 region at S296, S289 and S301, and one at the carboxy-terminal end at S642 (see Fig. 5 for the location of these sites on a 2D phosphopeptide map of C-Raf). The common feature of these sites is having a proline-directed phosphorylation motif. Indeed, phosphorylation at these sites was shown to be largely mediated by ERK, which belongs to the proline directed kinase family. As regards to the functional debate, Dougherty at al [158] and Hekman et al [175] ascribe a negative feedback role to these phosphorylations while Balan et al [159] claim a positive, kinase stabilizing role. Establishing the true physiological significance of these phosphorylations will require a more detailed examination and the resolution of these conflicting observations. Importantly, the physiological role of these phosphorylations appears significant since their phosphorylation on endogenous C-Raf was induced by growth factors and during cell cycle progression ([159] and our unpublished observation). In addition, besides establishing a new mode of feedback C-Raf regulation, these studies reveal a potential mechanism for isoform specific Raf regulation, since these sites are not well conserved among the three-mammalian Raf isoforms (Fig. 1 and 3). Thus, C-Raf may be uniquely regulated by ERK-mediated feedback phosphorylation, affecting its activation duration and possibly other functions.

When considering the available literature on the critical role of phosphorylation in C-Raf activation and in maintaining its active conformation on one hand and the identified Raf phosphorylation sites on the other, it is almost inevitable to assume that there are other as-of-yet unidentified C-Raf phosphorylation sites that play a role in C-Raf regulation. Thus, a more comprehensive approach will be required for their full identification. In addition, it is possible that C-Raf activation could be achieved via separate modes, involving different phosphorylation sites. This possibility has been demonstrated for Raf activation by VEGF and PDGF [176, 177] and for v-Ras and v-Src involving the S338/9 and Y340/1 sites [23, 167]. The large plethora of factors that activate Raf, including various cytokines, stress factors, ionizing irradiation, cell cycle state etc., imply that the modes of C-Raf activation may be even more diverse than we currently contemplate.

Beside the above-mentioned sites, also S233, S268 and T269 were reported to be C-Raf phosphorylation sites. S233 is thought to be a negative regulatory site targeted by PKA [95], while S268 and T269 are presumed to be an autophosphorylation site and a target site for a ceramide activated kinase or KSR, respectively [157, 178, 179]. Also here, the validity of the sites and their significance in C-Raf regulation, as well as the kinase/s responsible for their phosphorylation is a subject to debate and needs further examination.

4.2. Raf regulation by other modes

Besides Raf regulation through phosphorylation, various protein-protein interactions have been shown to play key roles in Raf regulation [128, 141, 180]. These include binding of Raf to Ras [6] and 14-3-3 proteins [138], Raf homo- [181] and heterodimerization [108], and association with chaperones such as HSP-90/p50(cdc37) [182, 183] and HSP/HSC-70 [151]. Raf is also regulated through association with various scaffold proteins such as KSR [184, 185], CNK [65, 186], RKIP [187, 188] and others, and possibly through its association with its target proteins, predominantly MEK (unpublished observation). These aspects of Raf regulation are covered in several excellent recent reviews [141, 180] and by Kyriakis JM in this issue.

Surprisingly, after more than 20 years of study, our knowledge on the regulation of Raf transcription and translation is very limited. Similarly, we know very little on Raf protein recycling and inactivation. For example, Raf degradation has been proposed to involve a proteasome-mediated pathways [189–191]. A recent report implicated PP5 in Raf inactivation by dephosphorylating S338 [192], providing an initial insight into this process. This aspect of Raf regulation might turnout to be as complex as the Raf activation process and will require more intensive examination, both under normal and disease conditions. Also the question relating to the regulation of Raf expression is of special importance since, as mentioned above, C-Raf overexpression in cancer appears significantly prevalent.