In 1983, the cloning of the acutely transforming replication-defective mouse type C virus 3611-MSV and characterization of its acquired oncogene was reported.1 Since 3611-MSV induces rapidly growing fibrosarcoma in mice, the transduced oncogene was called v-raf, whereas its cellular homologue was named c-raf.1 Shortly thereafter, Bister and coworkers described the cloning of the avian acute leukemia retrovirus Mil Hill No. 2 (MH2).2 MH2 was found to carry a second potential oncogene in addition to v-myc, which was termed v-mil, whereas its cellular counterpart was called c-mil.3 Already a hybridization analysis and the gross comparison of restriction maps of v-raf and v-mil pointed at sequence homologies between these 2 genes.4 By direct sequencing of both genes it was finally proven that 3611-MSV and MH2 have integrated orthologues of the same gene into their genomes.5 In the following we are going to use the nomenclature for Raf kinases as proposed in a recent review from Wellbrock et al.6
In contrast to all other oncogene kinases known at that time, no tyrosine kinase activity was detected in C-Raf.7 Indeed, it was found that C-Raf is a serine/threonine kinase.8, 9 This and the observation that Raf coexists with the Myc oncogene in retroviruses10 led to two novel concepts: (i) There is a tyrosine to serine phosphorylation switch as the growth factor signal enters the cell, and (ii) the mechanistic basis for cooperation between nuclear and cytoplasmic oncogenes as well as for signal transduction from cell surface receptors to the nucleus is the phosphorylation of transcription factor class oncogenes such as Myc.11, 12, 13 Growth factor abrogation and oncogene cooperation studies were performed to corroborate this signaling scheme.14, 15, 16, 17 Additional important early findings were (i) the cooperation between Raf and Myc in growth factor independent proliferation, immortalization and tumor induction,14, 18, 19 leading to the Raf-Myc balance model,20 and (ii) the cell lineage-switch activity of the Raf-Myc oncogene combination.21, 22, 23 The use of a kinase-dead Raf version and its presumed negative autoregulatory N-terminal half as a dominant-negative mutant24, 25, 26 enabled the identification of Raf as the first effector of Ras and positioned Ras at the top of the mitogenic cascade.27 Our finding that constitutive MAP kinase activity in v-raf transformed cells quickly led to the identification of the mitogen-activated kinase kinase MEK as first physiological Raf substrate, thereby completing the description of the classical mitogenic cascade.28, 29, 30
Subsequently, the paralogues A-Raf31, 32, 33 and B-Raf34, 35 were identified. Although all 3 Raf isoforms share considerable sequence similarity,36 they exhibit beneath common also individual functions, which are still far from being fully understood.37, 38 Additional key findings in the early days of Raf kinase research were the demonstration of C-Raf as a mutational target in a lung tumor model suggesting that Raf is a critical effector of Ras,39 the identification of Raf as an apoptosis suppressor cooperating with BCL2 at the outer mitochondrial membrane,40, 41, 42 the dependence of the cellular response to Raf signaling on signaling intensity and context,43, 44 and the ability of Raf to activate the NF-κB transcription factor, a major regulator of inflammatory responses and mesenchymal-epithelial transitions.45, 46
Raf kinase signaling
As already described above, the classical mitogenic cascade was originally described as unidirectional highway between extracellular growth factor signals and nuclear transcription events (Fig. 1a). RAS GTPases are activated by the majority of growth factor receptors and bind and recruit Raf to the cell membrane upon activation. The central signaling components Raf, MEK and ERK are then sequentially phosphorylated and activated by each other. More than 70 nuclear and nonnuclear effector molecules of the mitogenic cascade have been identified so far. In addition, Raf kinase signaling in a cascade-independent fashion has been described; for review see Ref.47. This includes the activation of the NF-κB transcription factor,45, 46 the prevention of apoptosis by antagonizing proapoptotic factors such as MST2, the mammalian sterile 20-like kinase,48 ASK1, the apoptosis signal-regulating kinase 1,49 and BAD, the BCL-2-antagonist of cell death,41 and finally the positive regulation of cell migration via the Rho effector kinase Rok-α.50 The regulation of Raf kinase activity is quite complex, far from being fully understood and beyond the scope of this review. For recent comprehensive reviews, please see Refs.6,38,51, 52, 53.
However, a few general principles are noteworthy. The key feature involves assembly of the cascade at the membrane from pre-existing modules (Ras module, Raf module, KSR module). This process is paralleled by an intricate pattern of phosphorylation and dephosphorylation events leading to conformational changes of signaling molecules. The kinetics of this process depends on the presence of individual Raf isoenzymes and on the engagement of various positive and negative feedback loops. Primarily the phosphorylation status and the localization of Raf kinases determine the association with interacting partners, such as chaperones, other kinases, prolyl isomerases, phosphatases, scaffolding proteins and also lipids and vice versa (Fig. 2). Within this signaling zoo along the mitogenic cascade, there is still more room for novel players. They are definitely more than just additional signaling proteins and contribute significantly to our understanding how Raf kinase signaling really works. Good examples for this conclusion are Prohibitin, which is required for Ras-induced Raf-MEK-ERK activation and epithelial cell migration,54 Shoc2/Sur8, which functions in complex with the catalytic subunit of protein phosphatase 1 as M-Ras effector modulating Raf activity55 and Hyphen, which in Drosophila is part of a novel Raf controlling complex consisting of the scaffold KSR and the Raf binding protein CNK.56
In these multiprotein complexes Raf proteins may act as true kinases, but kinase-independent, scaffolding-like functions of Raf kinases have also been described. Work from several groups established that homo- and heterodimerization of Raf kinases clearly exist,57, 58, 59 and that heterodimerization can be Ras induced.59 In addition, it was shown that Raf heterodimerization is regulated by 14-3-3 proteins, mitogens and the Mixed-lineage kinase 3 and is also stabilized by MEK inhibition.59, 60, 61 Marais and coworkers described that heterodimerization is involved in the activation of C-RAF by B-RAF, but that wild-type and mutant B-RAF use different activation mechanisms.62, 63 Whereas wild-type B-RAF activates C-RAF via RAS-induced heterodimerization, mutant B-RAF heterodimerizes with and activates C-RAF in a RAS-independent manner, thereby generating a novel B-RAF > C-RAF > MEK > ERK pathway that is active in normal as well as in transformed cells (Fig. 1b).63 Whether signaling via this pathway just reflects some sort of crossactivation of Raf isoforms or whether it leads to a different set of effector functions in comparison to the classical RAF > MEK > ERK cascade remains to be determined in future.
Raf kinases and mouse cancer models
Animal models of human cancer have contributed significant insights into the mechanisms of tumor formation and progression64, 65 and are indispensable tools for drug discovery.66 During the early days of Raf kinase research, infection or transduction of newborn mice with retroviruses or retroviral vectors harbouring either wild-type or mutant Raf kinases were instrumental to prove their transforming potential.18, 67 In another line of experiments mutagen-induced tumor formation in mice was analyzed to detect whether Raf kinases are mutational targets.39, 68 Ethylnitrosourea treatment of mice led to the induction of adenocarcinoma and T-cell lymphoma carrying point mutations in craf in the absence of ras mutations.39 Mutations of b-raf were also detected in ∼20% of chemically induced liver tumors from C3H/He mice.68 In addition, mice have also been used as genetic tools for understanding solid tumor formation.69 When the Sleeping Beauty transposon was used for somatic insertion mutagenesis, accelerated tumor formation in mice predisposed to cancer was noted. Unexpectedly, the cloning of insertion sites revealed that the gene most frequently disrupted by the transposon was b-raf. The resulting truncated B-Raf was lacking N-terminal autoregulatory elements and had transforming properties in focus formation assays.69
The first transgenic mouse tumor model expressing Raf kinases was established in our lab by overexpression of wildtype or constitutive active C-Raf (C-Raf-BXB) under the control of the human surfactant protein-C promoter in type II alveolar pneumocytes.70 All mice expressing C-Raf-BXB developed benign lung adenomas within 4 months of life.71 Raf-induced tumors expanded continuously without any signs of apoptosis.72, 73 Lung tumor formation in vivo was substantially retarded by systemic administration of the MEK inhibitor CI-1040.74 Initially, progression of C-Raf-BXB driven tumors to metastasis was never observed,70 and the stability of this phenotype allowed us to determine the influence of other genes on tumor progression. For example, neither loss of the tumor suppressor p53 nor the cyclin-dependent kinase inhibitor p21 provoked metastasis, although tumor latency was dramatically reduced.23 The loss of the Bcl-2 proto-oncogene greatly retarded lung tumor formation in C-Raf-BXB mice suggesting that bcl-2 is a major susceptibility gene for development of lung cancer in mice and perhaps also in humans.72 The influence of the co-chaperone BAG-1, which is interacting with B-Raf and C-Raf as well as with Bcl-2, was also analyzed on lung tumor formation. In C-Raf-BXB mice, which are heterozygous for bag-1,75 lung adenoma growth was significantly reduced due to increased apoptosis implicating BAG-1 as a critical player in Raf-driven oncogenic transformation.73 Surprisingly, when cell–cell contacts were impaired by conditional expression of dominant-negative E-cadherin, tumor progression from lung adenoma to adenocarcinoma and lymph node metastasis was observed. Moreover, not only loss of cell–cell contacts but also the induction of angiogenesis within the tumor was dramatically enhanced (Ceteci and Rapp, unpublished observations). This provides further experimental support for the angiogenic switch paradigm76 and links for the first time the formation of the adherence complex and the induction of angiogenesis.
So far, only 2 transgenic mouse models with either constitutive77 or conditional78 expression of mutant B-Raf have been published. Constitutive expression of the mutant human V600EB-RAF under the control of the bovine thyroglobulin promoter induced goiter and invasive papillary thyroid cancers (PTC) with tall-cell features, which later transitioned to poorly differentiated carcinomas.77 This resembles closely the phenotype of human PTC and together with the striking histopathologic similarity between mice and human harbouring V600EB-Raf suggests that this mouse cancer model might be useful for the analysis of molecular events leading to dedifferentiation of PTC. In addition, we have generated SPC V600EB-Raf mice and observed cystic lung hyperplasia, signs of inflammation and progression to metastasis, which are further analyzed in ongoing experiments (Goetz and Rapp, unpublished observation). Pritchard and coworkers applied a Cre/lox strategy to generate a conditional knock-in allele of V600EB-Raf.78 Ubiquitous expression of V600EB-Raf resulted in embryonic death before E7.5, whereas expression in adult somatic tissues induced hyperproliferation and bone marrow failure.78
Up to now, 2 Raf-dependent mouse tumor models have been generated that allow a preclinical evaluation of novel therapeutic modalities and potential anticancer drugs. As already mentioned above, the C-Raf BXB mice with constitutive expression of either C-Raf or C-Raf BXB develop lung adenoma70 and have been used to study the in vivo influence of the mitogenic cascade blockers BAY 43-9006 and CI-1040.74 Although both inhibitors reached comparable serum levels in mice, only CI-1040 was able to reduce lung tumor formation, which might be explained by inefficient cellular uptake of BAY 43-9006 or an increased dependency of the Raf-induced lung tumor on MEK activity.74 The lab of Andreeff established a conditional mouse model to study the effects of small molecule inhibitors in vivo using Raf-expressing FDC-P1 hematopoietic cells.79 All 3 Raf isoforms were individually expressed in an N-terminal truncated form and fused to the hormone-binding domain of the estrogen receptor, thereby enabling induction of Raf activity upon β-estradiol addition, as well as to the green fluorescent protein (GFP) allowing immunofluorescence detection. Tumor formation of FDC-P1 cells expressing individual Raf fusion proteins was analyzed in SCID mice. Estradiol pellet implantation into these mice significantly accelerated tumor onset and enhanced tumor growth and disseminated leukemia was observed using bioluminescence imaging.79 Whereas in both animal cancer models the MEK inhibitor CI-1040 reduced proliferation as well as tumor formation in vivo, no obvious effect on apoptosis was observed in our lung model in comparison to the leukemia model, which might be explained by cell-type specific responses.
Raf kinases and human cancer
For a long time, it was thought that constitutive activation of the mitogenic cascade,80e.g., by either mutational activation of RAS proteins81 or by overexpression of C-RAF,82 is the main mechanism contributing to tumor formation in humans. A couple of recent publications addressing the mutational status of A-RAF and C-RAF genes in more than 600 cancer cell lines and 500 primary human samples obtained from colorectal and gastric carcinoma, acute leukemias, gliomas, as well as lung, ovarian and testis tumors suggest that mutations in both RAF genes are very rare or nonexistent events.83, 84, 85 The most detailed analysis of 545 cancer cell lines and about 80 primary tumor samples by Marais and coworkers detected no A-RAF mutations and only 4 rare polymorphisms in C-RAF corresponding to a mutational rate of 0.7%.85 The identified 4 C-RAF mutants (P207S, V226I, Q335H, E478K) displayed differences with respect to basal and Ras induced kinase activities but were similar active in standard 3T3 colony formation assays.85
Surprising—but not unexpected—was the discovery of B-RAF mutations in ∼8% of human cancers.86 Initially, as an early outcome of the Cancer Genome Project at the Wellcome Trust Sanger Institute B-RAF mutations were detected in 66% of malignant melanomas and at a lower frequency in a wide range of other human tumors.87 All mutations described there were within or close to the kinase domain of B-RAF and the majority of mutations involved a single base substitution in the kinase domain (T1799A) leading to a V600E amino acid exchange.87 Moreover, mutant V600EB-RAF kinase displayed elevated in vitro kinase activity, constitutive MEK-ERK signaling, as well as enhanced transformation activity in fibroblasts and melanocytes.62, 87, 88, 89, 90 This pivotal discovery started a hunt for B-RAF mutations. In May 2006 the annotated database COSMIC (Catalogue of Somatic Mutations in Cancer) was listing 2,989 B-RAF mutations in 16,120 tumor samples incorporating curated mutation data from 180 publications.91 The tissues represented with the highest mutation frequencies being skin (43% of samples mutated), thyroid (27%), large intestine (15%), ovary (15%) and biliary tract (15%). No B-RAF mutations are so far listed for the urinary tract, testis, prostate, pleura, kidney, genital tract, cervix and bone, but up to now only a smaller number of samples have been analyzed from these tissues.
Although more than 30 B-RAF missense mutations have been published, the mutation most often represented in the database in ∼85% of all entries is the point mutation V600E. The high mutation rate at this position may be biased due to limited hot-spot mutation detection and the question about the predominance of this V600EB-RAF mutation needs to be clarified in future. The crystal structures of the kinase domains of wildtype and V600EB-RAF in complex with the small molecule inhibitor BAY 43-9006 were solved at 2.9 and 3.4 Å resolution, respectively, and provided some insights into the activation and signaling properties of B-RAF wildtype and mutants.62 It was proposed that B-RAF mutants such as the V600EB-RAF mutant, which was classified into a group called “activated mutants” displaying elevated in vitro kinase activities, have no impairment with respect to their overall enzymatic activity and are destabilized in their inactive conformation.86, 92 Another remarkable feature of the V600EB-RAF mutation was unraveled, when human cancer cell lines from various tissues were treated with the MEK inhibitor CI-1040.93 It was shown that the V600EB-RAF mutation—in contrast to the independent Q61RNRAS mutation—is associated with enhanced sensitivity to this drug. This sensitivity was correlated with downregulation of cyclin D, p27 induction and G1 arrest in all and a reduction of RB phosphorylation and increased apoptosis in some of the investigated cell lines with the V600EB-RAF mutation.93 Whether this finding has implications for the treatment of human cancer patients using MEK inhibitors, has to be addressed in future clinical studies with patients with established B-RAF mutation status.52
In radiation-induced thyroid papillary carcinomas another type of B-RAF mutation leading to a novel mode of B-RAF activation was found.94 The paracentric inversion of the long arm of chromosome 7 results in an in-frame fusion of the N-terminus of the A-kinase anchor protein 9 (exons 1–8) and the C-terminal catalytic domain (exons 9–18) of B-RAF. The resulting fusion protein has increased B-RAF kinase activity and is considerably active in NIH 3T3 focus formation assay and tumor formation in nude mice.94 Whether this type of mutational activation is found exclusively in thyroid carcinomas with recent radiation exposure needs to be determined in future. The group of Nikiforov was also successful in describing an additional mechanism of B-RAF activation.95 Using fluorescence in situ hybridization with B-RAF specific and centromeric chromosome 7 probes they were able to detect additional copies of B-RAF or of chromosome 7 in a high percentage follicular thyroid carcinomas. These changes were absent in follicular tumors with RAS mutations and as determined by Western blotting resulted in increased B-RAF protein levels.95 In summary, more and more mechanisms are elucidated contributing to hyperactivation or deregulation of RAF-mediated signaling pathways in human cancer. These are (i) Activating mutations of upstream RAF regulators, (ii) Overexpression of nonmutated RAF proteins due to transcriptional upregulation or due to mutations leading to a net increase of RAF copy numbers and (iii) Mutational activation by either point mutations or chromosomal rearrangements.
Before the year 2006 no germline mutations in RAF genes have been described. This was not unforeseen, since our early attempts to generate raf transgenic mice with promoters that are active early during development invariably failed and we had to choose a tissue-specific promoter active at later developmental stages (see above). Unexpectedly, 2 germline mutations, S427GC-RAF and I448VC-RAF, have been identified in 82 samples of patients with therapy-related acute myeloid leukemia.96 Both mutations were located in the kinase domain and able to sustain growth in soft agar assays, but did not induce morphological transformation of NIH 3T3 cells. The authors proposed that these C-RAF mutations constitute a novel tumor-predisposing factor.96 By mutational analysis of individuals with the cardio-facio-cutaneous syndrome, a developmental disorder not associated so far with an increased risk of cancer, 2 groups independently reported 15 different B-RAF missense mutations.97, 98 The majority of these 15 germline mutations had not been detected previously in human tumors and their kinase activities ranged from low (kinase-impaired) to highly elevated (high kinase). Whether these B-RAF mutants also display enhanced sensitivity to MEK inhibitors remains to be established.
So far, only a few studies link the mutational status of B-RAF with disease progression, clinical outcome and patient survival. For example, it was shown that the constitutive activation of the Ras-Raf signaling pathway due to either B-RAF or NRAS mutation is associated with poorer prognosis, i.e., shortened survival, in metastatic melanoma99 and that oncogenic B-RAF mutations rather correlate with progression than initiation of human melanoma.100B-RAF mutations were also correlated with poor survival in microsatellite-stable colon cancers101 and with a poorer clinicopathological outcome for patients with PTC.102 The great majority of publications however do not link the clinical course with the mutational status of B-RAF and due to great variations in sample size and methods used for tumor isolation, mutation detection or statistical analysis many conflicting reports on the impact of B-RAF mutations on tumorigenesis are currently published. Therefore, we are not going to comment these partially conflicting results and refer the reader to some recent excellent reviews on B-RAF mutations in general6, 86, 103 and on overviews focusing on B-RAF and melanoma,104, 105 thyroid106, 107 or ovarian cancer.108
Another rising field of research is to investigate how the immune system deals with B-RAF mutations, or in other words, whether mutated B-RAF is targeted by immune surveillance.109 The immunogenicity of constitutively active V600EB-RAF was successfully demonstrated by the detection of CD8+ responses, which were either HLA-B*2705-restricted110 or HLA-A*0201-restricted111 and of a V600EB-RAF specific CD4+ T-cell response.112 In addition, the loss of the V600EB-RAF genotype in a patient during progression from primary to metastatic melanoma suggested an active immune selection of nonmutated melanoma clones.99 Moreover, B-RAF and V600EB-RAF specific antibodies emerging at late stages of melanoma progression have been discovered.113 Together, these findings make it likely that in future at least advanced stage melanoma patients may benefit from targeting of the V600EB-RAF mutation.
Raf kinases and cancer drug discovery
Intense efforts have been taken in the past to develop novel anticancer strategies targeting Raf dependent signaling pathways; for reviews see Refs.114, 115, 116. Considerable clinical progress has been made with the small-molecule inhibitor Nexavar® (Sorafenib, BAY 43-9006), antisense and heat shock protein 90 (HSP90) inhibitors. An overview on the clinical status of different Raf targeting strategies in oncology is given in Figure 3.
In 2004, the National Cancer Institute director Andrew von Eschenbach set a ambitious goal for the NCI: the elimination of suffering and death due to cancer by the year 2015.117 A major step towards this direction was the positive evaluation of Nexavar (Sorafenib, BAY 43-9006), a potent small-molecule inhibitor of Raf kinases, in ongoing clinical trials. Sorafenib belongs to the bi-aryl urea class of protein kinase inhibitors118 and its development by Bayer HealthCare and Onyx Pharmaceuticals, Inc. as a C-Raf inhibitor119, 120 was based on the large body of evidence for a role of C-Raf as a major transformation effector of Ras.36 In early biochemical and cellular-based assays, this inhibitor was identified as bonafide C-Raf inhibitor and was demonstrated to reduce tumor cell proliferation in vitro as well as tumor growth in human tumor xenograft models in vivo.121, 122 Upon further characterization BAY 43-9006 was shown not only to inhibit B-Raf and V600EB-Raf in vitro, but also to target efficiently other receptor tyrosine kinases involved in neovascularization and tumor progression, including VEGFR-2, VEGFR-3, Flt-3, c-KIT, PDGFR-α, PDGFR-β and RET.123, 124, 125, 126 Daily oral dosing of BAY 43-9006 demonstrated broad-spectrum antitumor activity in various cancer xenograft models.123 The basis of this antitumor activity was the inhibition of the mitogenic cascade as well as the reduction of neovessel formation demonstrating that BAY 43-9006 is a novel dual action inhibitor targeting tumor cell proliferation as well as tumor angiogenesis.123 Moreover, it was also shown recently that downregulation of the antiapoptotic Bcl-2 family member Mcl-1 by BAY 43-9006 in a MEK/ERK signaling-independent manner contributes to the proapoptotic effects of this inhibitor.127 Additional multiple proapoptotic effects of BAY 43-9006 have been described in human melanoma cells.128 Among them were dephosphorylation of BAD on Ser75 and Ser99 activation of Bak and Bax, downmodulation of Bcl-2 and Bcl-XL levels, downregulation of the mitochondrial transmembrane potential, caspase activation as monitored by PARP cleavage and release of cytochrome c and SMAC from mitochondria. Surprisingly, in some but not all cell lines it was demonstrated by the use of a pan-caspase inhibitor that apoptosis induction by BAY 43-9006 is largely caspase independent and depends primarily on the nuclear translocation of apoptosis inducing factor AIF.128
BAY 43-9006 rapidly advanced in clinical trials and was awarded fast track status by the US Food and Drug Administration (FDA) in 2004. In the middle of 2005, 59 clinical trials were sponsored by either the National Cancer Institute or Bayer/Onyx for the use of Sorafenib as single or as combination agent as treatment for a wide variety of solid and lymphoid tumors, including Phase III clinical trials for advanced renal carcinoma, advanced malignant melanoma and primary hepatic cancer.129, 130 In February 2006 the initiation of a randomized, double-blind, placebo-controlled Phase III clinical trial studying Sorafenib administered in combination with the chemotherapeutic agents carboplatin and paclitaxel in patients with non-small cell lung cancer (NSCLC) was announced. At the 2005 Annual ASCO Meeting Sorafenib was reported to double progression-free survival in advanced renal cell carcinoma by the BAY 43-9006 TARGETs Clinical Trial Group.131 In December 2005 Sorafenib was finally approved by the FDA for the treatment of patients with advanced renal cell carcinoma and is now marketed as Nexavar in the US. In the meantime, Nexavar was also approved in Suisse and Mexico; EU wide approval is expected in the second half of 2006.
Early experiments using vector-driven antisense expression of full-length C-RAF already demonstrated that antisense inhibition is a valuable approach to reduce significantly the proliferation and tumorigenesis of transformed cell lines.132 The antitumor activity of ISIS 5132, a 20-nucleotide phosphorothioate 2′ deoxynucleotide targeting the 3′ untranslated region of C-RAF mRNA, was very promising at the early preclinical stage.133, 134 However, no major clinical benefits were observed in several Phase I135, 136, 137 or Phase II clinical trials.138, 139, 140, 141 In addition, the use of ISIS 13650, a second generation antisense oligonucleotide with further modifications in the sugar moiety targeting the same C-RAF sequence as ISIS 5132, has not been superior in certain preclinical settings.142 LErafAON, a C-RAF antisense oligodeoxyribonucleotide applied as a liposome-encapsulated formulation, which is supposed to increase the stability and the cellular uptake of the antisense inhibitor,143 is currently evaluated in several Phase I clinical trials as monotherapy as well as in combination with either radiation or chemotherapy.144, 145 Moreover, additional antisense approaches are currently tested for antitumor activity in preclinical models. So far, depletion of either B-RAF or mutant V600EB-RAF by small-interfering RNAs (siRNAs) reduced proliferation and invasiveness of melanoma cell lines,146, 147, 148 and also the growth and vascular development of malignant melanoma tumors.149 Reduction of in vivo tumor growth by application of C-RAF siRNA was also reported in xenograft models of human prostate150 and also breast cancer.151
The benzoquinone ansamycin Geldanamycin (GA) was originally isolated as compound with antifungal activities and was later found to reduce oncogene-dependent proliferation of tumor cells; for review see Ref.152. This is primarily due to binding and inhibition of a major cellular chaperone, HSP90, which prevents the conformational maturation of HSP90 client proteins and promotes their proteasomal degradation.153 More than 50 client proteins of HSP90, including Raf kinases, Akt/PKB, ErbB2 and Met, have been identified. The therapeutic selectivity of HSP90 inhibitors in tumor versus normal cells was explained by increased HSP90 expression and the concomitant formation of a supersensitive multichaperone complex in cancer cells.154 Due to severe hepatotoxicity in animal studies GA was not further clinically evaluated,155 whereas 17-AAG (17-allylamino-17-demethoxygeldanamycin), a GA derivate with a significantly reduced toxicity, entered several Phase I and now also Phase II clinical trials with encouraging results.156 Interestingly, it was shown recently that wild-type and mutant B-RAF proteins are also HSP90 client proteins and are targeted to ubiquitin-dependent proteolysis by 17-AAG.157 The V600EB-RAF protein and most of the other mutant B-RAF kinases were 17-AAG hypersensitive in comparison to nonmutated B-RAF.157, 158 17-DMAG, a second HSP90 inhibitor with higher solubility and antitumor efficacy,159, 160 is currently tested on patients with solid tumors and lymphomas in 3 Phase I clinical trials sponsored by the NCI. Although HSP90 inhibitors are promising, the usefulness of HSP90 inhibitors for cancer therapy is discussed controversially. It was shown recently that inhibition of HSP90 by GA or 17-AAG concomitantly induced massive expression of HSP40, HSP70 and HSP90 proteins.161 Overexpression of HSP family members, such as the well-recognized antiapoptotic factor HSP70, in turn contributes to drug resistance and a poor response to combination-chemotherapy regimens thereby helping tumor cells to survive.162 The group of Bhalla used pretreatment with K562, a benzylidine lactam HSP70 inhibitor, in tissue culture to block HSP70 induction which was accompanied by an increase of the antitumor activity of 17-AAG.163 On the other hand, increased HSP70 levels may promote the formation of stable complexes with tumor antigens, which upon release from the cell may lead to the activation of antigen-presenting cells thereby breaking tolerance to tumor antigens and eliciting CD8+ tumor-specific responses.164 More basic and clinical research is clearly needed to find out for which type of tumors mono- or combination therapies with HSP90 inhibitors are beneficial.
Another set of inhibitors, which is currently under active clinical development as anticancer drug, is the heterogeneous group of histone deacetylase (HDAC) inhibitors.165, 166 Although histones are the primary targets of HDACs and transcriptional reactivation of “dormant” tumor-suppressor genes is clearly one major mechanism of action, other cellular targets and pleiotropic effects have been described.167 In context with Raf kinases, HDAC inhibitors seem to have 2 ways of action: (i) in human multiple myeloma C-RAF mRNA expression was substantially reduced in response to treatment with the HDAC inhibitor SAHA168 and (ii) in human leukemia cell lines it was shown that HDAC inhibitors such as LAQ824 induce acetylation and inactivation of HSP90 which in turn promotes degradation of HSP90 client proteins including C-RAF.169, 170, 171 SAHA is currently in Phase II/III trials for the treatment of advanced cutaneous T-cell lymphoma and relapsed diffuse large B-cell lymphoma, whereas LAQ824 is in Phase I.
Understanding the human immune response to cancer is a prerequisite for the development of effective cancer immunotherapies.172, 173 As already mentioned above, a couple of early preclinical studies have evaluated the usefulness of B-RAF as target for immunotherapy. In melanoma patients harboring the V600EB-RAF mutation, a 29-mer B-RAF peptide incorporating the V600E mutation was used for in vitro stimulation of lymphocytes, generating MHC class II-restricted CD4+ T cells specific for this peptide as well as for melanoma cells expressing V600EB-RAF.112 A spontaneous HLA-B*2705-restricted cytotoxic T-cell response against an epitope derived from V600EB-RAF was demonstrated for the first time in melanoma patients.110 Moreover, in this publication also the loss of the V600EB-RAF genotype during progression from primary to metastatic melanoma in patients with V600EB-RAF specific T-cell responses was shown. This suggests an active immune selection of nonmutated melanoma clones by the tumor-bearing host.110 Finally, the screening of sera of 148 patients with advanced melanoma revealed that in ∼9% of sera B-RAF/V600EB-RAF specific antibodies are present demonstrating a humoral response of the immune system.113 In summary, further studies investigating cell-mediated and humoral responses are clearly needed to validate B-RAF/V600EB-RAF as a specific target for immunotherapy.
The new facets of Raf kinases that have been discovered over the past 5 years are really fascinating. However, a number of pressing questions need to be addressed in near future in more detail.
1What is still missing is an in vitro reconstituted system with highly purified components that allows us to study the signaling properties of wildtype and mutant Raf kinases. This might not only clarify the impact of individual phosphorylation events or binding partners, but also help us together with the determination of Raf kinase structure at higher resolution to design better Raf kinase inhibitors. In addition, an in vitro reconstituted system would be instrumental in verifying and fine-tuning the pathway predictions recently obtained by computational and mathematical modeling of the mitogenic cascade174; for a recent comprehensive review see Ref.175.
2Although Nexavar®/Sorafenib has certainly blockbuster potential, lessons learned on resistance development induced by several inhibitors in clinical use suggest that there is better no halt in the development of novel Raf kinase inhibitors. Several companies are actively pursuing this task (see overview Table I) and try to generate novel RAF mutation- or isoform-specific inhibitors in the light of the importance of the V600EB-RAF mutation. At the 97th AACR Annual Meeting in April 2006 the Chiron Corporation introduced the novel RAF inhibitor CHIR-265 and announced also the initiation of a Phase I clinical trial for melanoma patients.177, 178, 179 At the same meeting Plexxikon Inc. introduced the B-RAF specific inhibitor PLX4032.180 Although a couple of other companies, such as Sunesis, Exelixis and ArQule, are currently broadcasting the successful development of B-RAF selective inhibitors, there have been no data published on these so far.In the past, small-molecule Raf kinase inhibitors were primarily targeting the catalytic active site of these enzymes, whereas there have been no reports on allosteric inhibitors. This class of inhibitors offers an alternative approach to the inhibition of protein activities in particular for proteins undergoing conformational changes during their activation cycle, as previously shown for the N-WASP inhibitor Wiskostatin.189 Another way to go is the design of therapeutic approaches that specifically disrupt specific protein–protein interactions,190 in which Raf kinases are participating. If indeed isoform-exclusive interactions between individual Raf kinases and interacting proteins exist (Fig. 2), this might certainly help to bypass the missing isoform-specific selectivity of current active-site Raf inhibitors. A particularly interesting interaction to target in this respect is the interaction between C-Raf and Prohibitin, a highly conserved protein that was recently shown to be indispensable for RAS-induced activation of the mitogenic cascade and epithelial cell migration.54
3Whereas the detection of Raf-specific mutations is a top priority for oncologists nowadays, a systematic mutational analysis of genes encoding proteins negatively regulating the mitogenic cascade, such as RKIP, Spred and Sprouty, is still missing. So far, only reduced expression levels at certain stages of tumor development imply these proteins as potential tumor or metastasis suppressor genes. Another interesting point of future investigations is the functional analysis of microRNAs targeting Raf genes during normal development and tumor formation. As predicted from miRNA target databases as miRBase,191 all 3 isoforms are targeted by specific miRNAs (A-RAF:miR-412, B-RAF:miR-29a/b/c, C-RAF:miR-19a/b/c/d).
4The functional consequences of RAF mutations in human tumors have to be addressed using state-of-the art genomic and proteomic approaches. Two recent reports described the detection of V600EB-RAF specific gene expression patterns by microarray analyses in PTC192 and colorectal cancer193 and discussed the relevance of specific oncogene pathway signatures for diagnosis, clinical disease management and target discovery for directed therapies. Statements on the impact of B-RAF mutations on tumor establishment and progression absolutely require large clinical studies with patients being stratified for their B-RAF mutational status and with correctly staged tumor samples of a statistically significant amount of patients.
5Another neglected branch of investigation is the functional analysis of the role Raf kinases may have in stem cell biology and epigenetic regulation. Although by genetic analyses of transgenic and knock-out mice we have a gross understanding of the roles Raf kinases play during organismal development and during differentiation of multiple cell lineages, there are currently no published data associating Raf kinases directly to stem cell renewal or early stem cell differentiation. An indirect hint that Raf kinases might play a role in these processes is derived from the observation that hypermitogenic signaling via Raf kinases targets Bmi-1, a Polycomb family transcriptional repressor, which is required for the maintenance of stem cells in multiple tissues including the nervous and hematopoietic system.194 We have previously described a proliferative switch that links Raf signals to the phosphorylation and release of Bmi-1 from repressive chromatin complexes.195 Whereas low intensity Raf signals are required for normal proliferative responses, high intensity Raf signals induce an antiproliferative response that may include delayed cell cycle progression and subsequent differentiation, apoptosis or senescence.195 Since Bmi-1 is an epigenetic chromatin modifier that is also linked to cancer development it might be of interest to investigate whether the Raf-Bmi-1 connection has an impact on tumorigenesis. This might offer the possibility to manipulate cancer or cancer stem cells in a way that may lead either to apoptosis or prevents uncontrolled proliferation and differentiation.
Overview on selected small molecule Raf inhibitors. Compound activity in general is measured in vitro by their ability to inhibit RAF-mediated phosphorylation of kinase dead MAP kinase kinase MEK. IC50 values are presented to give an estimate of the in vitro activity of these inhibitors on C-RAF, B-RAF and V600EB-RAF.
Only a range of IC50 values for all three Raf kinases was given.
For SB-590885 data from a fluorescent ligand displacement assay have been incorporated and expressed as Kd (nM). Poor solubility in water may prevent the further clinical development of certain drugs. Abbreviations used: n.d., not determined; HTS, high-throughput screen.
Approved for patients with advanced renal cell carcinoma