|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 23, No 27 (September 20), 2005: pp. 6771-6790 © 2005 American Society of Clinical Oncology. DOI: 10.1200/JCO.2005.08.036
Raf: A Strategic Target for Therapeutic Development Against CancerFrom The University of Texas Health Science Center at San Antonio and Institute for Drug Development, Cancer Therapy and Research Center, San Antonio, TX Address reprint requests to Eric K. Rowinsky, MD, 11710 Spotted Eagle, San Antonio, TX, 78248; e-mail: erowinsk{at}oncodrugs.com
The mitogen-activated protein kinase (MAPK) signaling pathway plays a critical role in transmitting proliferative signals generated by cell surface receptors and cytoplasmic signaling elements to the nucleus. Several important signaling elements of the MAPK pathway, particularly Ras and Raf, are encoded by oncogenes, and as such, their structures and functions can be modified, rendering them constitutively active. Because the MAPK pathway is dysregulated in a notable proportion of human malignancies, many of its aberrant and critical components represent strategic targets for therapeutic development against cancer. Raf, which is an essential serine/threonine kinase constituent of the MAPK pathway and a downstream effector of the central signal transduction mediator Ras, is activated in a wide range of human malignancies by aberrant signaling upstream of the protein (eg, growth factor receptors and mutant Ras) and activating mutations of the protein itself, both of which confer a proliferative advantage. Three isoforms of Raf have been identified, and therapeutics targeting Raf, including small-molecule inhibitors and antisense oligodeoxyribonucleotides (ASON), are undergoing clinical evaluation. The outcomes of these investigations may have far-reaching implications in the management of many types of human cancer. This review outlines the structure and diverse functions of Raf, the rationale for targeting Raf as a therapeutic strategy against cancer, and the present status of various therapeutic approaches including ASONs and small molecules, particularly sorafenib (BAY 43-9006).
The ras family of oncogenes and encoded proteins has been evaluated as a putative target for anticancer therapeutic development. These efforts have resulted in new insights into Ras-mediated cell signaling as it relates to human cancer. Ras plays a central role in an intricate array of signal transduction pathways, characterized by cross talk, feedback loops, and multicomponent signaling complexes.1-3 One strategy to overcome the challenges inherent in developing therapeutics against signaling elements situated in redundant pathways is to target elements downstream of convergence points of critical signaling modules. This reasoning has led, in part, to interest in Raf kinase, which is one of several downstream effectors of Ras, as a target for therapeutic development against cancer. The Raf serine/threonine kinases are the principal effectors of Ras in the mitogen-activated protein kinase (MAPK) pathway (Fig 1). Raf activation occurs immediately downstream of membrane and cytoplasmic receptors that relay mitogenic signals.4 Although principally activated by Ras, Raf may also be activated by Ras-independent elements and, in turn, propagates signals through diverse effectors that mediate proliferation, angiogenesis, metastases, and survival.5 Raf may be activated by signaling upstream or constitutively. Constitutive activation of Raf and Ras are indistinguishable in their potential to induce malignant transformation.6-8 Activating raf mutations have been identified in melanoma, thyroid, colon, and other cancers (Table 1). 9-43 Furthermore, the disappointing clinical results of farnesyltransferase (FTase) inhibitors (FTIs) that were developed based on a flawed premise that they would effectively target malignancies with a high incidence of Ras mutation has led to scrutiny of signaling elements downstream of Ras, such as Raf, as therapeutic targets.44 This review highlights relevant information about the biology of Raf and novel strategies directed at exploiting this knowledge to more effectively treat malignant diseases.
The molecular mechanisms and signaling pathways that regulate cell proliferation and survival are receiving considerable attention as potential targets for anticancer strategies.45,46 Recently, there has been a notable increase in efforts directed at targeting the MAPK pathway, which integrates a wide array of proliferative signals initiated by receptor tyrosine kinases (RTKs) and G protein–coupled receptors.47,48 The network of signals emanating from the MAPK pathway are transmitted by proteins that serve as chemical switches, cycling between phosphorylated (activated) and dephosphorylated (inactivated) states.49 These on and off switches are regulated by kinases and phosphatases, respectively. Activated signaling elements, in turn, phosphorylate amino acid residues on downstream signaling proteins in a cascade-like and expansive manner.50 Not only does the centrality of the MAPK pathway render its components important targets for therapeutic development, but many genes that encode for its critical signaling elements undergo mutations, constitutively activating downstream signaling elements and conferring the potential for transformation and autonomous growth.50-52 In addition to efforts directed at Ras, therapeutic strategies directed at the MAPK pathway are targeting the cascade of downstream effector proteins including Raf, MAPK kinase (MAPKK; also called MEK or extracellular signal–regulated kinase [ERK] kinase), and ERK. The Raf/MEK/ERK module (Fig 1) of the MAPK pathway, which is immediately downstream of Ras, may be less redundant and innately resistant to therapeutic manipulations compared with Ras (as discussed in the next section).
The Unfilled Promise of Targeting Ras
After synthesis as inactive cytosolic propeptides, Ras undergoes a series of post-translational modifications at its carboxyl terminus that increase its hydrophobicity.56,57 These modifications render Ras functional and capable of localizing to the lipid-rich inner surface of the cell membrane. The first and most critical modification, farnesylation, which is principally catalyzed by protein FTase, adds a 15-carbon hydrobobic farnesyl isoprenyl tail to the carboxyl terminus of Ras. It is in the cell membrane where Ras cycles between inactive guanosine diphosphate–bound and active guanosine triphosphate (GTP) –bound states, thereby activating a series of effector kinases that phosphorylate a cascade of signaling proteins.58 Ras mutants exhibit slightly less intrinsic GTPase activity than wild-type Ras; however, the principal consequence of the mutated proteins is a marked decrease in interactions between Ras and its GTPase activator protein.59 Instead of reverting to its inactive guanosine diphosphate–bound state, the modified conformation of mutant Ras favors its active GTP-bound state, which has a higher propensity to activate downstream effectors even in the absence of growth factor stimulation, conferring a proliferative advantage to tumors. The considerable attention paid to targeting Ras as a therapeutic strategy is based on the high incidence of activating ras mutations in human malignancies, including approximately 22% of non–small-cell lung, 50% of colorectal, and 90% of pancreatic cancers.53,60,61 Of the strategies directed at Ras, targeting FTase has received the most attention, but the FTIs are not Ras specific, and a bonafide Ras-specific therapeutic agent has not yet been evaluated in clinical trials.62,63 Fortunately, because K-ras mutations constitute most ras mutations in the aforementioned malignancies, in which the therapeutic expectations of FTIs were among the highest, the failure of this strategy should not be surprising because geranylgeranyl transferase I can alternatively prenylate K-ras, rendering it functional even when FTase is completely inhibited.64,65 Although the FTIs have shown notable antitumor activity in patients with advanced breast cancer and some hematologic malignancies, the low ras mutation rates in these cancers suggest that farnesylation of other critical proteins is being inhibited.66
Localization of GTP-bound Ras to the inner surface of the cell membrane activates several downstream effectors, most notably the serine/threonine kinase Raf, which is the first signaling element in the MAPK pathway.2,67,68 As shown in Figure 1, other downstream effectors of Ras include the PI3K cell survival pathway, the small GTP-binding proteins Rac and Rho, and the stress-activated protein kinase pathway (also referred to as the c-jun N-terminal kinase [JNK] pathway).69-71 In addition, in response to cellular stress and cytokine stimulation mediated through Ras, the dual-specificity p38MAPK kinases (MKK3 and MKK6) and the JNK kinases (MKK4 and MKK7) phosphorylate p38MAPK and JNK, respectively.72-76 GTP-bound Ras interacts directly with Raf and mobilizes the inactive protein from the cytoplasm (Figs 1 and 2). Once the Ras-Raf complex is translocated to the cell membrane, Ras activates the serine/threonine kinase function of Raf through an association between its Ras-binding domain (RBD) in the amino-terminal regulatory region and Ras-GTP. This is followed by a series of Ras-dependent phosphorylation events and conformational changes, which will be described later in this review.77-84 The regulatory mechanisms of various Raf isoforms differ in that A-Raf and C-Raf require additional phosphorylation reactions for activity, whereas B-Raf has a much higher level of basal kinase activity.85 Raf is also activated by Ras-independent activators, including the soluble non-RTK Src and Janus kinase 1, which are involved in cytokine signaling.86 Other Ras-independent activators of Raf include interferon beta, protein kinase C (PKC) alpha, antiapoptotic proteins (eg, Bcl-2), scaffolding proteins (eg, ceramide-activated protein kinase), ultraviolet light, ionizing radiation, retinoids, erythropoietin, and dimerization between Raf isoforms86-94 (Fig 3). In addition, several Raf mutations confer constitutive activity to Raf irrespective of signaling activity upstream.9,11,12 The multifactoral mechanisms of Raf activation imply that therapeutic strategies that depend on the abrogation of any single element of these pathways may not result in sufficiently robust tumor growth–inhibitory activity. Furthermore, the kinase activity of Raf is inhibited by its interactions with cholesterol-rich lipid rafts in the cell membrane and phosphorylation by protein kinases A (PKA) and B (PKB/Akt), as shown in Figure 2.95-98 In essence, the activation status of Raf depends on the integration of both activating and inhibitory stimuli, the net result of which determines the downstream messages.
The raf family of genes was first identified as oncogenes in retroviruses that are the causative vectors of tumors in mice and chicken.99,100 The first raf gene to be identified, v-raf, the transforming gene of the mouse sarcoma virus 3611, induces fibrosarcomas and erythroleukemia in newborn mice, and C-raf (also called raf-1) is its proto-oncogene homolog.101,102 A-raf was found next by screening a mouse spleen cDNA library at low stringency with a v-raf probe. Next, v-Rmil was identified as the transforming gene in the avian retrovirus Mill Hill No. 2 (MH2) from a spontaneous ovarian tumor and found to be homologous to v-raf.99 C-Rmil corresponded to a third mammalian raf gene, B-raf, which was also shown to be an oncogene.103,104 However, initial attempts to identify activated versions of raf in human cancers failed to demonstrate unique DNA rearrangements in any specific tumor type, which, in retrospect, can be attributed to the lack of requisite sensitivity of early assays to detect single point mutations. Furthermore, most early efforts were directed at C-raf, rather than at the more oncogenic B-raf. The mammalian raf family consists of the following three genes: A-raf, B-raf, and C-raf, which are located on chromosomes Xp11, 7q32, and 3p25, respectively. The raf proto-oncogenes encode three 68- to 74-kd cytosolic proteins, termed A-Raf, B-Raf, and C-Raf (Raf-1), which share highly conserved amino-terminal regulatory regions and catalytic domains at the carboxyl terminus (Fig 4). 10 As serine/threonine kinases, Raf proteins phosphorylate serine and threonine residues on essential modulatory proteins downstream of Ras. Each Raf species has a distinct expression profile in tissues, which suggests that individual Raf isoforms perform clearly defined functions.4 C-Raf is ubiquitously expressed in most tissues. Both A- and B-Raf have more restricted expression profiles than C-Raf, with A-Raf overexpressed in urogenital tissues (eg, kidney, ovary, prostate, and epididymis) and B-Raf overexpressed in neural, testicular, splenic, and hematopoietic tissues.105 Unlike A-raf and C-raf, B-raf undergoes differential splicing in exons 8b and 10a, and its spliced variants are translated into 10 B-Raf isoforms.106,107 Both A-Raf and C-Raf undergo localization to the mitochondria, which supports the notion that Raf regulates apoptosis, but the specific proportions of Raf isoforms that are localized to the mitochondria are not known.108-112 This localization may be a result of isoform-specific lipid- or protein-binding partners, which recruit Raf to distinct membrane rafts.
From a functional standpoint, although all Raf proteins are serine/threonine kinases and capable of activating the MAPK cascade, they have distinct downstream phosphorylation targets and play unique roles in signaling.113 Their distinct roles are supported by Raf knockout studies, in which mice lacking each of the three Raf proteins have disparate phenotypes.113,114 B-raf knockouts die in utero by day 12, usually as a result of massive internal hemorrhage, whereas A-raf and C-raf knockouts die postpartum with extensive intestinal distension (A-raf knockout) or failure of lung maturation (C-raf knockout). Additional support for the diverse functionality of Raf family members is provided by the disparate responses of B-Raf and C-Raf to identical stimuli, as well as the distinct messages that each isoform relays downstream to Rap1, which is a small GTPase that functions as both an activator and repressor of Raf.115 For example, Rap1-mediated stimulation of B-Raf by cyclic adenosine monophosphate (cAMP) phosphorylates ERK, whereas stimulation of C-Raf inhibits ERK phosphorylation.115
The Structure of Raf
Regulation of Raf Kinase Activity C-Raf exists in the cytoplasm as a 300- to 500-kd protein complex. The complex consists of C-Raf, heat shock protein 90 (Hsp90), and the dimeric protein cofactor 14-3-3. 14-3-3 binds to two specific phosphoserine residues of C-Raf, which masks its kinase domain and inactivates the protein. The binding of Ras to C-Raf displaces the 14-3-3 dimer, rendering C-Raf accessible to dephosphorylation by protein phosphatase 2A.126 This action enables subsequent activation of C-Raf by mitogenic stimuli. The stability and function of C-Raf are also regulated by the phosphorylation status of C-Raf itself, the binding of C-Raf to Ras, and interactions between C-Raf and cellular lipids. Activation by Ras and other small GTPases. The initial event in the activation of Raf is its recruitment to the inner surface of the cell membrane by the small GTPase Ras. The effector domain of GTP-bound Ras binds to C-Raf through the RBD and CRD in the CR1. Although binding to both sites is required for Raf activation, the most critical interaction is between Ras-GTP and the RBD.127 All Ras isoforms are capable of interacting with Raf, but K-Ras is the most potent activator, whereas N-Ras is much more efficient than H-Ras.128 The interaction between Ras and C-Raf alone is insufficient to activate C-Raf, but it serves to translocate C-Raf to the cell membrane where it can be activated. The activation of B-Raf by Ras has been less well studied; however, the interacting amino acids in the Ras-Raf interface are identical for B-Raf and C-Raf.129,130 The association of Ras with B-Raf also translocates B-Raf to the cell membrane where it is activated.124 Interestingly, a membrane-free complex of B-Raf and 14-3-3 can be activated in vitro by recombinant Ras. This is in stark contrast to A-Raf and C-Raf, which must undergo a series of phosphorylation reactions on serine and tyrosine residues in the cell membrane and cannot be activated by Ras alone.124,130 Of the Raf isoforms, B-Raf is activated first, and on stimulation by Ras, it heterodimerizes with C-Raf, the significance of which is not known.94 Both B-Raf and C-Raf can bind to other small GTPases, most notably Rap1.115,131,132 The effector domains of Rap1 and Ras are nearly identical, but activation of these proteins produces vastly different downstream effects. Furthermore, Rap1 mediates distinct effects after binding to various Raf isoforms. The B-Raf–Rap1 complex activates B-Raf, whereas the C-Raf–Rap1 interaction does not activate C-Raf and, in fact, may be inhibitory.115,132,133 This occurs because Rap1 binds to the CRD of C-Raf with higher affinity than Ras and excludes Ras from binding. The Rho family of small GTPases, consisting of Rho, Rac, and cyclin-dependent kinase (Cdc) 42, regulate cytoskeletal organization during the cell cycle and also mediate Ras-induced activation of Raf, especially C-Raf.134-136 These GTPases do not directly bind to Raf but, instead, signal by activating downstream kinases. Rho signals by activating the serine/threonine protein kinases N1 and N2 and Rho-associated kinase 1, whereas Rac and Cdc42 signal through p21-activated kinase (PAK).134-136 Phosphorylation. Raf is principally activated by phosphorylation of specific amino acid residues as shown for each isoform in Figure 4. From an evolutionary standpoint, the Raf activation sites are highly conserved from yeast to humans. Several amino acids in Raf, particularly serine (S) 259 and S621, which bind 14-3-3 and maintain C-Raf in a closed auto-inhibited conformation, are phosphorylated in the basal state.137 On stimulation, Ras-GTP displaces 14-3-3 from S259, and C-Raf is translocated to the cell membrane, where it can be dephosphorylated at S259 by protein phosphatase 2A or other phosphatases.126 S259 also represents the site of inhibitory phosphorylation by PKB/Akt, PKA, and serum glucocorticoid-inducible kinase.121,138,139 Phosphorylation at S621 seems to have greater significance because mutations at this site inactivate Raf's kinase activity. Hence, a balance of phosphorylation and dephosphorylation is required to prime Raf in the basal state before stimulation by Ras or mitogens.137 Raf is also phosphorylated at other serine and threonine residues, the most important of which are S338 and tyrosine (Y) 341, which are situated adjacent to the C-Raf kinase domain.140 Phosphorylation of these residues relieves the inhibitory effects of the regulatory domain on the kinase domain.141 S338, which is the evolutionarily conserved PAK phosphorylation site that resides on the amino-terminal side of the kinase domain, is critical for Raf activation.134-136,140,142 This site is also phosphorylated in response to stimulation by growth factors, integrins, Ras, PAK1, and PAK3.78,136,143 The homologous site on B-Raf, S445, is constitutively phosphorylated, accounting for the higher basal activity of B-Raf. Ras presumably phosphorylates this site by activating PI3K. Activated mutants of Rac and Cdc42 are also capable of inducing phosphorylation of S338 by activating PAK. Y341 is phosphorylated by the Src family of non-RTKs, Janus kinase, and erythropoietin.85,92 The substitution of this tyrosine residue by aspartate in B-Raf may explain why B-Raf is fully inducible by Ras, whereas A-Raf and C-Raf require both Ras and Src for full activation.124 However, Ras-mediated recruitment of C-Raf to the cell membrane and Src activation are not the only steps involved in the activation of C-Raf. A-Raf, which is structurally similar to C-Raf, is activated in a similar manner; however, the pertinent structural and activational aspects of B-Raf differ from those of A-Raf and C-Raf. Although the structural domains and phosphorylation sites of Raf proteins differ, the greater degree of phosphorylated amino acids in B-Raf confers a 15- to 20-fold higher level of kinase activity in the basal state than either A-Raf or C-Raf, and B-Raf is therefore a much more robust activator of ERK phosphorylation.85,144,145 The differential splicing of B-raf may also account for the distinct kinase activity of the protein. In addition, the structures of several B-Raf mutants mimic the conformational changes unique to phosphorylated wild-type B-Raf, which may explain the ability of B-Raf mutants to activate ERK in the absence of stimulation. Other interactions. In addition to phosphorylation events, the activation status of C-Raf is regulated by protein-protein and protein-lipid interactions. As shown in Figure 2, C-Raf interacts with a diverse array of scaffolding proteins (kinase suppressor of Ras and MEK partner-1), adaptor proteins (Bcl-2–associated athanogene-1), chaperone proteins (Hsp90 and Hsp70), substrates (retinoblastoma protein [Rb]), lipids (phosphatidic acid, cholesterol-rich caveolae, and cytosolic lipid rafts), and cellular constituents, many of which, in turn, modulate its kinase activity.95
Activation of Downstream Effectors by Raf The respective downstream substrates of MEK1 and MEK2 are ERK1 (p44MAPK) and ERK2 (p42MAPK), which are translocated to the nucleus where they ultimately induce an array of cytoplasmic and nuclear regulatory proteins.50,159-162 Effectors include the nuclear transcription factors Elk-1, Fos, Jun, AP-1, and Myc, which regulate genes encoding proteins that play key roles in proliferation, angiogenesis, metastasis, and resistance to anticancer therapeutics.51 As a result, cell cycle regulators, such as cyclins D1 and E and Cdc activator 25 phosphatase, are positively regulated,163,164 whereas p27kip-1 and other inhibitors of cyclin-dependent kinases (cdk) are negatively regulated.51 These actions favor progression through cell cycle checkpoints, aberrant growth, dedifferentiation, and cell survival.
C-Raf activates other cellular effectors, but the extent of the interdependence of these actions on MEK1 and MEK2 is not clear. For example, C-Raf activation regulates cytoskeleton formation by modulating the polymerization status of vimentin.165 Cell survival signaling is also regulated by C-Raf, which induces phosphorylation of I
General Constitutively active mutant Raf proteins are predominately a result of point (missense) mutations, deletions, amplification, and rearrangements of raf.173-176 Such genetic alterations have been identified in malignant melanoma, hematopoietic cancers, and cancers of the thyroid, breast, kidney, liver, larynx, biliary tract, and other organs, as shown in Table 1.173-176 Although initial efforts at identifying raf mutations in human cancer focused on C-raf, the advent of high-throughput gene sequencing led to the identification of activating B-raf mutations as the predominant genetic aberrations.11,122,145,177
B-raf Mutations Sequence analysis of B-raf in human cancer has identified more than 30 single-site missense mutations, principally encoding amino acids in the kinase domain of B-Raf, whereas the constitutive activity and transforming potential of C-Raf result from loss of the auto-inhibitory amino-terminal region, as well as gene rearrangements.6,175,176 Most B-raf mutations are caused by thymidine-to-adenine transversions at nucleotide position 1796 in exon 11 or 15, which encode a valine-to-glutamic acid substitution at amino acid 599 (V599EB-Raf) in the activation segment (kinase domain) of the protein. Interestingly, structural changes in the activation segment as a result of the insertion of an acidic residue close to a site of regulated phosphorylation mimic phosphorylated B-Raf.9 V599EB-Raf possesses the hallmarks of a conventional oncogene because the kinase activity of its encoded protein is greatly elevated; it constitutively stimulates ERK in vivo in the absence of Ras activation; and it transforms NIH3T3 cells.144 Furthermore, the basal kinase activity of V599EB-Raf is 12.5-fold higher than that of wild-type B-Raf, and its responsiveness to stimulation by oncogenic H-Ras is diminished. Furthermore, the transforming capacity of V599EB-Raf in NIH3T3 cells is 667-fold more efficient than that of wild-type B-Raf, whereas the equivalent mutation introduced into C-Raf (V492E) confers 10-fold lower kinase activity and transforming capacity.85 The discovery of B-raf mutations in 60% to 70% of malignant melanomas is surprising because early studies attributed the hyperactivation of the Raf/MEK/ERK module of the MAPK pathway in melanoma to an abundance of autocrine and paracrine growth factors. Interestingly, B-raf mutations are not found in uveal melanoma, which differs from cutaneous melanoma in that abnormalities of chromosome 6 are found only in uveal melanoma, suggesting that there are distinct pathways for melanoma formation.178,179 Further studies evaluating the function of V599EB-Raf in benign and dysplastic nevi may yield important information about the type and timing of events required for tumorigenesis. Interestingly, the V599EB-Raf allele is found in as many as 80% of benign nevi, suggesting a role for oncogenic B-raf in nevus formation and melanoma initiation.180 However, there is no direct evidence that benign nevi harboring V599EB-Raf progress to malignancy, and most cases may actually represent terminally differentiated lesions analogous to nondysplastic colorectal aberrant crypt foci that harbor K-ras mutations in the absence of adenomatous polyposis coli (APC) mutations. APC mutations are generally considered to be of low malignant potential, whereas K-ras mutations that arise after APC mutations promote colorectal tumor progression.181,182 Further studies are also needed to determine whether the prevalence of B-raf mutations in melanoma relates to the site of the primary tumor, sun exposure, and radiation damage. Similar findings have been noted in the setting of papillary thyroid carcinoma, in which up to 69% of tumors harbor V599EB-Raf, whereas benign thyroid tumors and both follicular and medullary thyroid carcinomas do not.18,19 It is notable that B-raf mutations are common in melanoma and thyroid cancers and that both melanocyte and thymocyte growth is positively regulated by cAMP. Interestingly, B-Raf is thought to be the key Raf isoform that transduces cAMP-dependent growth signals in both cell types, which may account for their vulnerability to transformation by activating mutations of this kinase.183,184 Analysis of other much less common oncogenic B-Raf mutants, most of which cluster adjacent to valine 599 or in the G loop ATP-binding region, suggest that the mutated proteins stimulate kinase activity in a manner similar to V599EB-Raf.9 Nevertheless, it is intriguing that several of these mutations involve highly conserved or invariant residues in the catalytic domain, which are required by other kinases for optimal activity. This raises the question of how these mutants promote tumorigenesis.8,9,185 It should also be noted that B-raf mutations outside the kinase domain have been identified, and other mutations will likely be identified as the gene is sequenced in other types of malignancies.11 Mutations of B-raf and ras are essentially mutually exclusive, implying that these genes belong to the oncogenic signaling pathway. Fewer than 1% of cancers with B-raf mutations have simultaneous ras mutations, and of the 1% that have mutations of both genes, the B-raf mutations are almost never V599EB-Raf.8,9,14 In colorectal carcinoma, both genes are mutated at high frequencies in the same types of premalignant lesions and at the same stages in the transition from adenoma to carcinoma.8,14 A strong association exists between mismatch repair deficiency and the presence of the mutant V599EB-Raf protein in colorectal carcinoma, which may be a result of the underlying DNA repair defect.14 Further reflecting the redundancy of the MAPK pathway, a high fraction of papillary thyroid cancers harbor either V599EB-Raf, mutant K-ras, or mutant RET.9,20 Harboring more than one mutation is quite rare, although a moderate fraction of low-grade ovarian tumors harbor either V599EB-Raf or mutant K-ras.9,20 This finding may represent a unique paradigm of human tumorigenesis through mutations of these signaling proteins that lie in tandem.8,9 However, concomitant ras mutations have been identified in cancers that harbor uncommon B-raf mutations in the G loop region, suggesting that there may be differences in molecular pathways used by distinct mutant B-Raf proteins.9
C-raf Mutations
Given the high proportion of cancers with constitutively activated Raf, Ras mutations, or growth factor hyperactivity, which result in increased signaling through Raf, Raf is an ideal target for therapeutic development. Although there have been many attempts to develop therapeutics against Raf, most efforts have been directed at C-Raf rather than B-Raf. To decrease Raf production and inhibit its activation, antisense oligonucleotides (ASONs), small-molecule kinase inhibitors, and dominant interfering DNA constructs are being developed. In addition, other therapeutics that indirectly target Raf include inhibitors of chaperone proteins (eg, geldanamycin analogs), which destabilize Raf, and histone deacetylase inhibitors, which reduces raf expression.188
The specificity of nucleotide base pairing provides the rationale for using ASONs as therapeutics against Raf.189,190 This approach relies on the intracellular uptake of short synthetic ASONs that are complementary to Raf mRNA by mechanisms that have not been clearly elucidated. The ASON then hybridizes with its cognate mRNA, leading to RNAase H–mediated degradation of the complex. Alternatively, the ASON can sterically inhibit translation, which reduces synthesis of the encoded protein. ISIS 5132 (CGP 69846A; ISIS Pharmaceuticals Inc, Carlsbad, CA) is a 20-base phosphorothioate ASON designed to hybridize to the 3' untranslated sequence of C-raf.191 Binding induces degradation of the C-Raf mRNA, which, in turn, decreases synthesis of C-Raf in a concentration-dependent manner.192 The 50% inhibitory concentration (IC50) value for both tumor proliferation and C-Raf expression is approximately 100 nmol/L.192 Furthermore, treatment of mice bearing human lung and breast cancer xenografts produces impressive decrements in C-Raf, as well as antitumor activity.191 In other models, ISIS 5132 decreases C-Raf expression and enhances sensitivity to both cytotoxics and radiation.193 The phosphorothioate backbone of ISIS 5132 was engineered to confer resistance to digestive nucleases, which is manifested by plasma half-life values ranging from 30 to 85 minutes and extensive tissue distribution in mice.194-196 The feasibility of administering ISIS 5132 was explored in patients with advanced solid neoplasms on the following schedules: (1) 21-day continuous intravenous (IV) infusion (CIVI) every 28 days; (2) 2-hour IV infusion thrice weekly for 3 weeks every 28 days; and (3) 24-hour IV infusion weekly for 3 weeks every 28 days.197-199 The principal toxicities were fever and malaise. Thrombocytopenia and anemia, which were typically moderate in severity, brief, and not cumulative, were also noted. Transient prolongation of the activated partial thromboplastin time and activation of the alternate complement pathway, which have been attributed to the phosphorothioate backbone of ISIS 5132, occurred in a dose-dependent manner. Dose-dependent elevations of the compliment component C3a, but not Bb or C5a, were noted. Although maximum tolerated doses were not clearly defined in the first two studies, plasma concentrations of intact ISIS 5132 achieved at the highest doses (6 and 4 mg/kg/d) exceeded IC50 values derived in vitro and were known to activate the alternate complement pathway in monkeys.198 In the third study, an unacceptably high incidence of intolerable toxicities, particularly Coombs hemolytic anemia and acute renal insufficiency, was noted in patients treated at doses greater than 24 mg/kg/wk. The toxicities of ISIS 5132 were similar to those of other ASONs and, therefore, should not be interpreted as being related to target inhibition. Although several patients experienced protracted periods of stable disease, major tumor regression did not occur. C-raf mRNA levels in peripheral-blood mononuclear cells were consistently suppressed in patients receiving ISIS 5132 as a 2-hour IV infusion thrice weekly for 3 weeks, but suppression of C-raf mRNA was not detected on the schedule of 24-hour CIVI weekly for 3 weeks every 28 days and not evaluated in the study of ISIS 5132 as a 21-day CIVI. The antitumor activity of ISIS 5132 was evaluated in phase II studies in patients with advanced colorectal (15 patients, no prior treatment for metastatic disease), hormone-refractory prostate (16 patients, no prior chemotherapy), ovarian (22 patients, one to two prior systemic therapies), small-cell lung (four patients, one prior therapy), and non–small-cell lung (18 patients, no prior therapies) carcinomas.200-203 Stable disease lasting 2.5 to 5.5 months was the best response in a sizeable proportion of patients, but there were no major tumor regressions. Nonetheless, these disappointing results should not diminish the potential importance of Raf as a therapeutic target because several alternative hypotheses, including the lack of validation of ASON technology as a platform that can confer robust anticancer activity and lack of documentation of raf mutational status in these clinical studies, may explain these results.
Small-Molecule Inhibitors of Raf Kinase
Sorafenib (BAY 43-9006): Mechanism of Action and Preclinical Results The bi-aryl urea sorafenib (4-{4-[3-(4-chloro-3-trifluoromethyl-phenyl)-ureido]-phenoxyl}-pyridine-2-carboxylic acidmethylamide-4-methylbenzene-sulfonate; BAY 43-9006) is the first molecule of its class to undergo clinical development. Originally identified by high-throughput screening of small molecules against C-Raf kinase, sorafenib was found to be a potent competitive inhibitor of ATP binding in the catalytic domains of C-Raf, wild-type B-Raf, and V599EB-Raf mutant. As shown in Table 2, the IC50 values of sorafenib against C-Raf, wild-type B-Raf, and the V599EB-Raf mutant in a biochemical assay are in the low nanomolar range, whereas sorafenib does not inhibit MEK1, ERK1, erbB1, or erbB2.206 Sorafenib potently inhibited activation of the MAPK pathway and ERK phosphorylation in human cancer cell lines, irrespective of whether they harbored K-ras mutations, V599EB-Raf, or both.207 Further characterization of sorafenib in biochemical assays showed potent inhibition of pertinent RTKs involved in tumor progression and angiogenesis, including human and murine vascular endothelial growth factor receptor (VEGFR) -2, VEGFR-3, platelet-derived growth factor receptor-beta (PDGFR-ß), Flt-3, c-Kit, p38  , and fibroblast growth factor receptor-1 (Table 2). In contrast, erbB1, insulin-like growth factor-1, c-met, and erbB2 RTKs were not inhibited. The kinase activities of PKA, PKB, PKC, PKC , cdk1/cyclin B, and pim-1 were also insensitive.208 Furthermore, sorafenib inhibited various nonkinase targets, including adenosine A3, dopamine D1, and muscarine M3, albeit at much higher (micromolar) concentrations than kinase targets.209 In cellular assays, sorafenib reduced basal phosphorylation of the MAPK pathway in a panel of human breast, melanoma, pancreatic, and colon cancer cell lines expressing either mutant K-Ras, mutant B-Raf, or wild-type Ras or Raf. Interestingly, several non–small-cell lung cancer cell lines expressing mutant K-ras were resistant, presumably because Raf-independent activation of MEK is operative in these cells.208,210 In other cell-based assays, sorafenib inhibited phosphorylation of several pro-angiogenic RTKs, including both human VEGFR-2 and murine VEGFR-2, murine VEGFR-3, PDGFR-ß, and Flt-3.208,210 In nude mice bearing human xenografts derived from melanoma and colon, pancreas, breast, and lung carcinomas, sorafenib treatment resulted in a high level of tumor growth inhibition without appreciable toxicity.206 Most early evaluations used the HCT116 human colon xenograft because its tumorigenicity depends on K-Ras activation. Extending the duration of sorafenib treatment attained protracted antitumor efficacy, even when treatment was initiated in settings of high tumor burden. Significant growth inhibition was noted after treatment of well-established human xenografts with B-raf mutations (HT-29, Colo205, and DLD-1 colon), K-ras mutations (NIH-H460 and A459 lung; MiaPaCa pancreas), and both K-ras and B-raf mutations (MDA-MB-231 breast) with sorafenib at doses of 7.5 to 60 mg/kg daily for 9 days. B-raf–mutated MDA-MB-231 breast cancer xenografts, which were reduced in size by 42%, on average, after only 9 days of treatment with 30 mg/kg of sorafenib, were the most sensitive.208,210 Relevant activity against the human SKOV-3 ovarian xenograft that harbors wild-type Ras but overexpresses both erbB1 and erbB2 was also noted.
Concurrent with the demonstration that sorafenib is efficacious in a molecularly diverse range of human tumor xenografts, translational studies have demonstrated inhibition of the MAPK pathway after 5 days of sorafenib treatment of HT-29, DLD-1, HCT-116, and MDA-MB-231, but not Colo-205 xenografts. In the Colo-205 tumors, in which concurrent assessments of vascular effects were performed, tumor neovascularization was reduced dramatically. The cumulative results of these studies suggest that sorafenib inhibits tumor progression by blocking cellular proliferation that is dependent on activation of the MAPK pathway and/or inhibiting tumor angiogenesis through VEGFR-2, VEGFR-3, and/or PDGFR-ß. Recent studies suggest C-Raf inhibition may also promote the death of endothelial cells as a result of their specific requirements for stimulation by VEGFR-2.211 The results indicate that sorafenib is efficacious not only against human tumors with ras and/or raf mutations, but also against tumors that overexpress growth factor receptors that signal through Ras and the Raf/MEK/ERK module. However, it is important to note that the relative potency of sorafenib against various kinases, particularly VEGF, must be considered in assessing the value of sorafenib as a Raf kinase inhibitor, as well as in drawing conclusions about the value of Raf kinase as a molecular target against cancer. Favorable cytotoxic effects were noted after treatment of a broad spectrum of human cancer cell lines and xenografts harboring both wild-type and mutated forms of ras or raf with sorafenib and either fluorouracil, paclitaxel, gemcitabine, gefitinib, vinorelbine, doxorubicin, irinotecan, or its active SN-38 metabolite.212 Treatment of human tumor xenografts with sorafenib plus paclitaxel, irinotecan, gemcitabine, or cisplatin did not enhance the toxicity or diminish the activities of the therapeutics. Pharmacokinetic studies in rodents and dogs have demonstrated that sorafenib clearance is much lower than normal liver plasma flow. Its low steady-state volume of distribution (approximately 0.7 to 0.93 L/kg) suggests that tissue affinity is low and plasma protein binding is high (mean free fraction, 1.2% [human] to 2.5% [mouse]). The pharmacokinetics in mice are dose proportional over a biologically relevant dosing range, and tissue concentrations are several fold higher than IC50 values in vitro.206,208,213 At higher doses, drug exposure increases disproportionately, possibly because of saturation of gastrointestinal absorption. Autoradiographic studies have revealed homogeneous drug distribution to peripheral tissues and modest penetration across the blood-brain barrier. The mean terminal half-life ranges from 6 to 7 hours. In rodents, oral bioavailability is high (approximately 79%). Drug disposition is principally by CYP3A4 metabolism, followed by biliary and fecal excretion (approximately 90%). CYP1A, CYP2C9, CYP2C19, and CYP3A are not induced after incubating drug with microsomal extracts from human hepatocytes. However, in vitro metabolism studies in human systems indicate extensive metabolism by CYP3A, and early clinical data indicate that disposition is principally by hepatic metabolism and fecal excretion. Sorafenib is a modest inhibitor of CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, and CYP3A4, and the propensity for interactions between sorafenib and drugs that inhibit and induce P-450 systems exists. In rodents and dogs, sorafenib is well tolerated. Principal toxicities include emesis, diarrhea, and transaminase elevations. Histopathologic studies have revealed dose-related degenerative changes in the liver, stomach, duodenum, pancreas, kidneys, heart, testes, and ovaries and regenerative changes in the liver, pancreas, duodenum, and kidneys. Hypocellularity and necrosis of hematopoietic and lymphoid tissues and unusual findings involving the teeth and growth plate of the femur have been noted.
Clinical Evaluations Disease-directed studies. The principal paradigm adopted for disease-directed evaluations of sorafenib represents a radical departure from traditional phase II approaches. Although phase II studies are being performed in malignancies of high interest, the principal disease-directed evaluation strategy was a randomized discontinuation trial. This unorthodox approach was undertaken because the predominant clinical benefit of the agent, particularly in patients whose tumors were not screened for molecular aberrations known to increase the probability of responding, was projected to be increased progression-free survival (PFS), which was also the principal beneficial effect in preclinical studies. In addition, because sorafenib inhibits multiple kinases, the use of any empiric screening and/or enrichment strategy, as well as any particular malignancy, could produce false-negative results. In contrast to randomized phase II studies, which lack sufficient statistical power to discern small to moderate, albeit relevant, differences between treatments, the randomized discontinuation study is designed so that there is an initial process of natural enrichment of the study population with patients who may have experienced benefit to treatment before patients are randomly assigned to either continue or discontinue drug treatment.217 The randomized discontinuation study is felt to be ideal for sorafenib and agents whose main benefit is expected to be tumor growth delay, which is not readily detected in nonrandomized studies. At the end of an initial lead-in phase, in which all patients receive the study drug, patients who experience a relevant degree of tumor growth are removed from the study. This weeding out process enriches the study population with patients who will most likely benefit from further treatment, thereby increasing the probability that the randomization step will be more efficient at detecting tumor growth inhibition related to drug. In essence, the lead-in period may furnish data about the inherent potential of the agent to induce tumor regression and can suffice as multiple phase II studies, each of which can be sized in real time to provide a requisite level of statistical power. At the end of the lead in period, patients whose tumors have not progressed are randomly assigned to either continue or discontinue treatment, ideally in a double-blinded, placebo-controlled fashion. The natural selection or enrichment of the population before random assignment increases the efficiency of the trial, with as few as 20% of the standard number of randomly assigned patients. Nonetheless, a shortcoming of this approach relates to its inability to precisely quantify the magnitude of antitumor activity. However, if there is a clear difference in PFS between the randomly assigned arms, conclusions can still be generated about the general activity of the agent. Nevertheless, if the results meet a sufficient level of interest, resource-intensive phase III studies may ensue. The randomized discontinuation study, as depicted in Figure 6, was designed to discern differences in PFS between patients treated with either sorafenib or placebo in the randomization period. The randomization stage was sized to discern PFS in patients with colorectal carcinoma, which frequently harbors ras mutations, although patients with many tumor types were enrolled. At the end of the 12-week period, in which all patients received sorafenib 400 mg twice daily, patients whose target lesions had increased in excess of 25% were taken off study. Because of concerns about randomly assigning patients who had potentially benefited from treatment, patients whose target lesions had regressed by greater than 25% were not randomly assigned and, instead, continued treatment until disease progression. Patients who experienced neither objective benefit of this magnitude nor disease progression were randomly assigned to either continue treatment with sorafenib or placebo. Because PFS was the primary end point in the randomization phase, placebo-treated patients who experienced progressive disease could be re-treated with sorafenib.
Patient accrual ended in January 2004, with 484 patients accrued at a rate of 36 patients per month at only five institutions in the United States and Europe. Of these, 408 patients were the focus of a recent report.218 Tumor regression was noted in previously treated patients with advanced melanoma, sarcoma, RCC, and colorectal, thyroid, and pancreatic cancers. Most colorectal carcinoma patients treated with sorafenib developed disease progression before random assignment. RCC emerged as a central focus of the study, and the accrual targets for RCC were reset higher as the activity of sorafenib in RCC became increasingly evident. In the run-in phase, 202 patients with RCC were treated with sorafenib and were the focus of a recent report in May 2005.219 At 12 weeks, 144 patients (71%) experienced tumor shrinkage or disease stabilization. Independently confirmed tumor shrinkage of  25% (partial response) was noted in eight (4%) of these patients. A total of 65 patients were entered onto the randomization phase, of whom 32 were treated with sorafenib and the rest received placebo. Sixteen patients (50%) in the sorafenib arm were progression free at 12 weeks postrandomization, compared with six patients (18%) in the blacebo arm (P = .0077). The median progression-free survival from randomization in the two arms was 24 weeks and 6 weeks, respectively (P = .0087). The unique design of this study served the broad mechanism of action of this promiscuous tyrosine kinase inhibitor, providing a sufficient level of flexibility to evaluate a wide range of end points and tumor types. The intriguing results in patients with RCC provided the basis for a phase III study, in which 800 patients with unresectable and/or measurable RCC who have received at least one prior systemic therapy are being randomly assigned to treatment with either sorafenib or placebo. The primary and secondary end points were overall survival and progression free survival, respectively. A total of 905 patients were randomly assigned, of whom 769 patients were the focus of a recent report.220 Three hundred eighty-four patients were randomly assigned to the sorafenib arm, and the rest to the placebo arm. An independent response assessment was performed in 574 patients. Seven partial responses (2%) were noted in the sorafenib arm compared with none in the placebo arm. Disease stabilization was seen in 261 (78%) versus 186 (55%) patients in the sorafenib and placebo arms, respectively, and disease progression was noted in 29 (9%) versus 102 (30%) patients, respectively. The median progression-free survival for patients in the sorafenib arm was 24 weeks, compared with 12 weeks in the placebo arm (hazard ratio = 0.44; P = .000001). The substantial benefit due to sorafenib was apparent across all patient subgroups. Because of the magnitude of the benefit noted upon analysis, treatment unblinding was performed, and patients randomly assigned to placebo were allowed to cross over to sorafenib treatment. There was considerable interest in the melanoma patients who participated in the randomized discontinuation study based on the high incidence of B-raf mutations in melanoma.218 In the June 2004 report that focused on the first 20 patients enrolled, five patients developed cutaneous toxicity of grade 3 severity, and two patients developed hypertension that required intervention.221 Of 19 patients whose disease had been evaluated, 15 patients developed progressive disease before or at the planned 12-week assessment, whereas one patient had a partial res |
TOP
ABSTRACT
INTRODUCTION
B, nuclear factor-kappa B; p90RSK, 90-kd ribosomal S6 kinase; PPAR
