Targeting the mitogen-activated protein kinase cascade to treat cancer

Key Points

• The mitogen-activated protein kinase (MAPK) pathway controls the growth and survival of a broad spectrum of human tumours.

• Activating mutations in RAS and RAF result in activation of the MAPK pathway and are present in a large percentage of solid tumours.

• The central role of RAF and MAPK kinase (MEK) in transmitting signals through the RAS–MAPK pathway make these kinases promising targets of anticancer drugs.

• MEK inhibitors are the first highly selective inhibitors of the MAPK pathway to enter the clinic.

• Emerging clinical data show promising hints that suppression of the MAPK pathway can be achieved without unacceptable toxicity levels.

Abstract

The RAS–mitogen activated protein kinase (MAPK) signalling pathway has long been viewed as an attractive pathway for anticancer therapies, based on its central role in regulating the growth and survival of cells from a broad spectrum of human tumours. Small-molecule inhibitors designed to target various steps of this pathway have entered clinical trials. What have we recently learned about their safety and effectiveness? Will the MAPK pathway prove amenable to therapeutic intervention?

Main

A diverse array of growth factors, cytokines and proto-oncogenes transduce their growth-promoting signals through the activation of the small G protein RAS, which leads to activation of the serine/threonine kinase RAF, and then to activation of the mitogen-activated protein kinase (MAPK) kinase (MEK). MEK then phosphorylates and activates MAPK, also known as extracellular signal-regulated kinase (ERK). Dating back to the early 1990s, elucidation of the various protein interactions and phosphorylation events that mediate this pathway has made it one of the best-characterized growth-factor-mediated signalling pathways.

Anchorage-independent cell-cycle progression is one hallmark of the neoplastic phenotype^1,2. The various growth factors and cell-adhesion receptors that are involved in regulating anchorage-dependent cell-cycle entry stimulate a myriad of signalling pathways, including the phosphatidylinositol 3-kinase (PI3K) and MEK–ERK cascades (Fig. 1). In addition, these pathways regulate other responses associated with cell adhesion such as anoikis (apoptosis that occurs when cells detach from their substrate) and cell motility (Fig. 2). These events are mediated by both transient reversible steps as well as by changes in gene expression. Comprehensive reviews of these regulatory processes have been published elsewhere^3,4. We will discuss what we have recently learned about the importance of the RAS–MEK–ERK pathway in tumour growth and progression from studies involving small-molecule inhibitors of one or more components of this cascade. Recent data from clinical trials with these inhibitors have provided important new information about the toxicity and efficacy of targeting the MAPK pathway.

Figure 1: The mitogen-activated protein kinase signalling pathway.

Following ligand binding, growth-factor receptor tyrosine kinases such as the platelet-derived growth factor receptor (PDGFR), the vascular endothelial growth factor receptor (VEGFR), the receptor ERBB, and the fibroblast growth factor receptor (FGFR) become activated. This induces the binding of adaptor proteins such as growth-factor-receptor-bound protein 2 (GRB2) that bind to specific phosphorylated residues on the cytoplasmic tails of activated receptors. In cooperation with GRB2, the guanine-nucleotide exchange factor SOS activates RAS, by catalysing the replacement of GDP with GTP. In its GTP-bound form, RAS activates the kinase activity of RAF and its downstream signalling cascade. RAF phosphorylates the mitogen-activated protein kinase (MAPK) kinase (MEK), which phosphorylates and activates MAPK/extracellular signal-regulated kinase (ERK). There exist several negative regulators of this pathway. These regulatory molecules include RAF kinase inhibitor protein (RKIP), which interferes with MEK phosphorylation⁸¹; RAS and RAB interactor 1 (RIN1), which competes with RAF for binding to activated RAS⁸²; IMP (impedes mitogenic signal propagation), which interacts with kinase suppressor of RAS (KSR) to prevent recruitment of MEK to activated RAF⁸³; and the MKP (MKP1,3) family of MAPK phosphatases, which reduce the activation state of the RAF–MEK–ERK module^84,85. Other negative regulators include both AKT and serum/glucocorticoid inducible kinase (SGK), which can phosphorylate BRAF^86,87. Positive regulators of RAF activity, such as protein kinase C (PKC), SRC, p21-activated kinase (PAK) and 14-3-3, activate RAF in a RAS-independent manner. This pathway can also become activated by adhesion of integrins to specific extracellular-matrix molecules. These interactions activate focal adhesion kinase (FAK) and phosphatidylinositol 3-kinase (PI3K). PI3K is also activated by RAS. Stimulation of this pathway leads to activation of the GTP-binding protein RAC, which interacts with the NCK–PAK complex, thereby leading to PAK, and subsequently RAF, activation.

Figure 2: Downstream effects of extracellular signal-regulated kinase activation.

Once the extracellular signal-regulated kinases ERK1 and ERK2 are activated through the RAS–RAF–MEK–MAPK signalling pathway shown in Fig. 1, they mediate a range of cellular effects. They activate several subsequent kinases, including mitogen-activated protein kinase (MAPK)-interacting kinases (MNK1 and MNK2), mitogen- and stress-activated kinases (MSK1 and MSK2), and 90-kDa ribosomal S6 protein kinase (RSK). They also activate transcription factors such as the peroxisome-proliferator-activated receptor-γ (PPARγ), ELK1, ETS, signal transducer and activator of transcription 1 (STAT1) and STAT3. The resulting changes in gene regulation affect cell-cycle progression and/or cell motility in a cell-dependent manner. MEK, MAPK kinase.

Cell-cycle regulation

Cell proliferation can be activated by activation of growth-factor receptors and/or binding of integrins to specific extracellular-matrix molecules during cell adhesion. Activation of these receptors is associated with the induction of cyclin D1 and downregulation of endogenous cyclin-dependent kinase (CDK) inhibitors such as p21 (also known as WAF1 or CIP1) and p27 (also known as KIP1) (Refs 5,6). Early studies showing a link between the MEK–ERK pathway and cell-cycle machinery included the demonstration that blockade of thrombin-induced proliferation of Chinese hamster lung fibroblasts correlated with suppression of ERK activation⁷. Similarly, dominant-negative forms of ERK, as well as ERK antisense nucleotides, inhibited proliferation of NIH3T3 fibroblasts⁸. In addition, studies using pharmacological inhibitors of MEK^9,10 or activated MAPK phosphatase 1 (MKP1), a negative regulator of MEK¹¹, significantly altered the ability of these cells to proliferate in response to growth-factor stimulation.

The expression of cyclin D1 was ultimately shown, through the expression of dominant-negative or activated forms of MEK, to be controlled by MEK–ERK signalling^12,13. This induction is thought to be mediated, in part, by activation of the AP-1 transcription factor¹⁴. Although data indicate that ERK activation is also required for cyclin D1 expression, the intensity and duration of the signalling through this pathway ultimately determines whether a cell undergoes differentiation, proliferation or cell-cycle arrest¹⁵.

Different RAF isoforms, for example, fine tune the cell-cycle regulatory machinery¹⁶. In response to a strong RAF signal, a G1-specific cell-cycle arrest, through induction of p21, leads to inhibition of cyclin-D- and cyclin-E-dependent kinases¹⁷. This finding is consistent with a report that sustained activation of MEK correlates with cell-cycle arrest¹⁸. However, moderate RAF activity is sufficient to induce cyclin D expression and DNA synthesis¹⁷. Negative regulation of RAF is mediated by a range of proteins, such as AKT and serum/glucocorticoid inducible kinase^19,20. Furthermore, the phosphorylation status and expression levels of a range of early gene products of the MAPK signalling pathway, such as of the transcription factors FOS, MYC and JUN, have also been shown to dictate biological outcome²¹. FOS has been shown to function as a sensor for ERK signal duration. When ERK activation is transient, its activity declines before FOS protein accumulates. By contrast, sustained ERK signalling results in phosphorylation of FOS, leading to its stabilization, thereby positively impacting cell-cycle entry²².

How does one integrate the observation that high levels of RAF activation can result in cell-cycle arrest with the well-established role of MEK–ERK signalling in promoting G1 progression? An explanation is provided by the finding that CDK-inhibitor family members can positively regulate the activation of cyclin-D-dependent kinases²³. This observation is consistent with the induction of p21 in growth-factor-stimulated fibroblasts occurring in the early part of the cell cycle followed by its decline at mid G1 (Ref. 24). This pattern resembles the activation of ERKs during the G1 to G1/S transition²⁵.

Given the role of the MEK–ERK signalling pathway in mediating both cell-cycle progression and arrest, it is surprising that most tumours have increased, sustained activation of the RAF–MEK–ERK pathway. One possible explanation could be that this pathway also contributes to one of the mechanisms that tumour cells use to promote their own survival, such as by controlling induction of apoptosis.

Survival signalling

Studies in several experimental model systems have provided insight into the involvement of RAF–MEK–ERK signalling in control of cell survival. As reviewed elsewhere, members of the RAF family (BRAF and c-RAF/RAF1) have been described as important mediators of anti-apoptotic activity²⁶. Although the anti-apoptotic activity of the RAF family seems to be established, the molecular steps are just beginning to be defined. Both the kinase activity and kinase-independent mechanisms of RAF are believed to mediate its anti-apoptotic functions. Independent of its ability to stimulate MEK–ERK signalling, RAF is believed to directly antagonize the death-promoting activity of apoptosis signal-regulating kinase 1 through a physical interaction between the two proteins²⁷.

The classical pathway by which RAF regulates apoptosis involves its kinase activity, which regulates both the transcriptional activity of the cell as well as the activity of pro-apoptotic proteins. Functionally, the signalling pathways can be divided between MEK-dependent and -independent events. In a MEK-dependent manner, activation of ERK and subsequently of the 90-kDa ribosomal S6 protein kinase (RSK, Fig. 2) leads to the phosphorylation and inactivation of the pro-apoptotic protein BAD²⁸. RSK activation also leads to phosphorylation of the transcription factor CREB, thereby promoting survival²⁹. However, studies showing that RAF-mediated activation of the pro-survival transcription factor NF-κB occurs in the absence of ERK activation indicates the existence of a MEK-independent mechanism in RAF's repertoire of anti-apoptotic functions³⁰. Data obtained from RAF1-null mice, as well as those obtained from studies using specific MEK inhibitors or activated and dominant-negative forms of MEK, support the occurrence of both MEK-independent and MEK-dependent anti-apoptotic mechanisms^31,32.

Targeting the MAPK signalling pathway

As the MAPK signalling pathway not only promotes cell proliferation, but also mediates cell survival and is upregulated in cancer cells, it seems to be a good therapeutic target. Several inhibitors of MAPK signalling have therefore been developed. What are the effects of inhibiting this pathway in animal models? Erk1^−/− mice are fertile and have no apparent abnormalities. However, they do have a twofold reduction in the number of mature thymocytes. Furthermore, T-cell proliferation is reduced in response to activating factors³³. The involvement of the MAPK pathway in immune signalling has therefore raised valid concerns about the potential toxicity of inhibitors directed against this pathway. In addition, Mek1^−/− mice undergo embryonic lethality, resulting from impaired placental development. Subsequent demonstration that Mek1^−/− fibroblasts were defective in fibronectin-mediated cell migration, despite the fact that MEK2 and ERK activation were normal, indicated that MEK1 might have additional roles other than activation of ERK³⁴. If so, blockade of this pathway could be toxic.

There has also been widespread concern that effective blockade of the MAPK pathway would prove too toxic to normal proliferating cells, such as those in the gastrointestinal tract and skin. Might it be possible to develop small-molecule inhibitors that have sufficient potency and selectivity to inhibit this signalling pathway specifically in cancer cells?

Approaches to therapeutic intervention

The regulation of a large number of cellular processes depends on the activation state of ERK, so blocking the RAS–MAPK pathway conceivably has use against a multitude of human diseases and conditions including cancer^35,36,37, inflammatory diseases^38,39, pain⁴⁰ and cardiac hypertrophy^41,42. Because of widespread concerns regarding the potential toxicity of MAPK-pathway inhibitors, efforts in this area have largely been expended in the oncology arena, because of the life-threatening nature of the disease. Clearly, there exists a strong rationale for exploring the potential of MAPK-pathway inhibitors as anticancer agents, in part based on the broad spectrum of human tumours known to show activated signalling through this pathway (Table 1). Three proteins have received the most attention as targets for pharmacological intervention of the MAPK pathway: RAS, RAF and MEK. Each of these proteins has unique features that make it a good therapeutic target, from a scientific rationale. What are their relative merits and what has been learned from the study of various small-molecule inhibitors, in both the preclinical and clinical arenas?

Table 1 Human tumours exhibiting activated RAS–MAPK signalling

Farnesyltransferase inhibitors. Extensive effort has been expended from numerous academic and pharmaceutical laboratories to investigate the therapeutic potential of inhibiting RAS activity with selective farnesyltransferase inhibitors (FTIs). RAS, which does not have a transmembrane domain, must have an ISOPRENOID group attached to it, through a process known as prenylation, to facilitate its localization to the plasma membrane. By inhibiting the post-translational addition of this farnesyl group to RAS by farnesyltransferase, it was thought that FTIs would be able to target a broad range of human tumours in which RAS was constitutively activated. The rationale for this approach has been discussed in greater detail elsewhere^43,44.

Preclinical evaluation of numerous small-molecule FTIs, covering chemically diverse structural classes, yielded promising reports of activity. Anticancer effects included antiproliferative, pro-apoptotic and anti-angiogenic activities⁴⁵. However, concerns about this approach began surfacing when investigators learned that higher concentrations of FTIs were required to inhibit oncogenic KRAS compared with wild-type RAS or oncogenic HRAS⁴⁶. It was subsequently shown in preclinical models that tumours that express KRAS — the most commonly detected form of activated RAS — escape FTI inhibition.

The ability of several FTIs to treat a broad range of human cancers has nonetheless been tested in clinical trials. These agents, including BMS-214662, L-778123, SCH-66336 (Sarasar), R115777 (Zarnestra) and AZD3409, represent structurally distinct small molecules. For a comprehensive review of the clinical testing status of these agents, see Ref. 45. The most extensive clinical evaluation has been carried out with Zarnestra, which has been studied in patients diagnosed with a range of cancers, encompassing acute and chronic leukaemias, multiple myeloma, non-small-cell lung cancer, and breast, prostate, colorectal and pancreatic cancers. Clinical results with R115777 have largely been disappointing (Table 2). The most notable effects were observed when this drug was administered as a single agent to patients with haematopoietic malignancies. The lack of objective responses in a Phase III clinical trial of patients with advanced pancreatic cancer⁴⁷, where the incidence of KRAS mutations is extremely high, provides further evidence that FTIs are not successful in treating tumours that are associated with RAS activation.

Table 2 Clinical trials of RAS–MAPK pathway inhibitors

There is little doubt, however, that these drugs actually inhibit farnesyltransferase activity in patients. Translational studies with the FTIs R115777 and SCH66336 have shown significant inhibition of prenylated proteins, namely the chaperone protein HDJ2 and the intranuclear intermediate-filament protein prelamin A. FTI research is now focused on the search for farnesylated proteins that are required for disease pathogenesis and that are strongly inhibited by these drugs. Candidate proteins include centromere-associated proteins, the PRL family of phosphatases, the RAS-related protein RHEB, and the Rho family member RND⁴⁴. Alternatively, the real target(s) of FTI-mediated efficacy might not be farnesylated. It is known for instance that RHOB is not only farnesylated, but is also a substrate for geranylgeranyltransferase — another prenylation enzyme, which adds a 20-carbon, rather than a 15-carbon, lipid moiety to signalling proteins. A gain-of-function role for geranylgeranylated RHOB has been described⁴⁸. Although there could be a specific therapeutic niche in which FTIs are most effective, clinical data obtained so far leads us to conclude that these agents do not represent a viable approach to blocking signal transduction through the RAS–MAPK pathway.

RAF inhibitors. The RAF family — ARAF, BRAF and RAF1 — are at the centre of a highly regulated control module involving both negative and positive regulatory interactions with a large array of signalling molecules. RAF phosphorylates the downstream kinase MEK on two serine residues to activate it. Initially, interest in inhibitors of the kinase activity of RAF came about as a consequence of the finding that RAF was an important downstream effector of RAS⁴⁹. It was therefore reasonable to speculate that inhibition of this kinase might afford a broader spectrum of antitumour activity than agents directed against a single upstream growth-factor receptor, such as members of the ERBB or vascular endothelial growth factor (VEGF) receptor families. The discovery of oncogenic BRAF mutations in human tumours has fuelled further interest in exploring the anticancer potential of inhibitors that block the kinase activity of RAF⁵⁰.

BRAF mutations are most prevalent in melanomas, occurring in nearly 70% of tumour samples examined^50,51, but are also found at a significant frequency in other cancers, including cancers of colorectal^50,52,53, ovarian⁵⁴ and thyroid origin^55,56. Interestingly, roughly 90% of the activating mutations involve the replacement of Val599 with a glutamate residue. This site lies within the kinase activation loop, resulting in constitutive activation of BRAF. BRAF activity is enhanced by this mutational change presumably because its negative charge effectively mimics phosphorylation of the activation loop.

The functional consequences of knocking down BRAF activity in melanoma with small interfering RNA has confirmed that its oncogenic activity mediates ERK signalling, induction of proliferation and protection from apoptosis^57,58. These studies have provided further impetus in exploiting BRAF for pharmacological intervention in patients with cancer. As reviewed by Bollag et al., several small-molecule RAF inhibitors have now been reported⁵⁹. According to published reports, only one of these, BAY 43-9006, has reached the clinical testing stage. BAY 43-9006 was discovered through a combinatorial chemistry approach aimed at studying structural modifications to a 3-thienyl urea chemical series⁶⁰. Subsequently, this bis-aryl, which inhibits recombinant RAF with an IC₅₀ of 12 nM, was shown to have in vivo anti-tumour activity against colon, pancreatic and ovarian xenograft models⁶⁰.

BAY 43-9006 is now undergoing clinical evaluation, with several Phase I and Phase II trials now completed (Table 2). It has been reported that this agent is generally well tolerated, with the most common toxicities involving the gastrointestinal tract (diarrhoea) and the skin (rash). These side effects seem consistent with those expected from a MAPK-pathway inhibitor, which would be predicted to affect tissues known to undergo rapid turnover, like the gut and skin. Although clinical trials have revealed substantial interpatient variability in the pharmacokinetics of BAY 43-9006, dose-dependent increases in exposure were consistently observed⁶¹. A total of three Phase I trials were carried out, in which the incidence of stable disease ranged from 33–43%. Furthermore, partial responses were observed for two patients, one with renal-cell cancer and one with hepatacellular carcinoma⁶². Whereas Phase II trials were carried out in patients with several different tumour types, including melanoma, as well as renal-cell, colorectal and hepatocellular carcinomas, Phase III trials seem to be focusing on patients with renal-cell carcinoma. Will RAF prove to be a valid anticancer drug target? There are several studies indicating that this question can not be addressed by BAY 43-9006, because of its lack of specificity. It was earlier reported that BAY 43-9006 inhibits RAF1 without affecting other protein kinases⁶⁰. Recent studies, however, have reported that this agent inhibits a range of other kinases, including many that are known to regulate proliferation and survival of tumour cells, such as platelet-derived growth factor receptor (PDGFR) and kinase insert domain receptor (KDR, also known as VEGFR2) tyrosine kinases¹⁰⁹. So, although BAY 43-9006 has promise as an anticancer agent, based on reports of its clinical activity, it is difficult to assess the impact of RAF inhibition on observed clinical outcomes.

Clinical activity of BAY 43-9006 against renal-cell carcinoma could very likely be a consequence of its effective inhibition of the VEGF receptor KDR. This supposition is based on a report that bevacizumab and SU11248, which are distinctly different types of VEGFR inhibitors, both have antitumour activity against human renal cancers. The known susceptibility of renal tumours to VEGFR inhibitors, along with recent reports on VEGFR inhibition by BAY 43-9006, make it tempting to speculate that VEGFR is this drugs true antitumour target, rather than RAF⁶³. Clinical samples from patients treated with BAY 43-9006 could also be examined to determine if MAPK signalling is reduced, and to determine the relative contribution of RAF inhibition in patient response. There have been encouraging reports that this drug inhibits phorbol-ester-stimulated ERK levels in patients' peripheral-blood lymphocytes, so this could be a useful biomarker⁶⁴. Further studies should include a similar biomarker analysis of human tumour biopsies.

Ongoing efforts in this area are no doubt being directed towards the design and development of BRAF-specific kinase inhibitors, as in human cancers, oncogenic mutations have only been identified in the BRAF isoform. RNA interference studies have successfully demonstrated that depletion of BRAF, but not ARAF or RAF1, in melanoma cell lines blocks ERK activity⁵⁷. It is not clear, however, whether BAY 43-9006 has effects on these other isoforms, as the RAF homologue kinase domains are relatively highly conserved — there is 78% and 81% identity between BRAF compared with ARAF and RAF1, respectively. The observation that activating mutations are always found in BRAF might be due to differences in the number of RAS-GTP-dependent phosphorylation sites required for maximal activation — two versus four sites in the case of BRAF and RAF1, respectively^65,66.

A key breakthrough in our understanding of how BRAF mutations lead to transformation was provided by the recent elucidation of the crystal structure of this enzyme⁶⁷. Richard Marais and his colleagues successfully solved the structure of isolated wild-type and mutant BRAF kinase domains (Box 1) bound to the inhibitor BAY 43-9006. The binding pattern of this inhibitor shows that the hinge region and the conserved Asp–Phe–Gly motif present in the activation loop (kinase subdomain VIII) are important for the interaction between this inhibitor with the enzyme. Information provided by these crystal-structure analyses also explained why the ATP-binding site and the activation loop are preferential targets for mutational events. Interestingly, oncogenic mutations affecting the equivalent position to Val599 in BRAF are known to occur in various tyrosine kinases, including PDGFR kinase family members such as KIT⁶⁸. Structural studies are proving to be invaluable for addressing the reasons that such different kinases have oncogenic hot spots in common. In the meantime, this information can also be exploited in the design of cancer drugs tailored to these molecular targets.

MEK inhibitors. The rationale for targeting MEK is similar to that of targeting RAF — because it has a central role in the MAPK pathway and in transducing proliferative stimuli in a broad spectrum of human cancer cells. In vitro expression of constitutively active forms of MEK has been shown to result in transformation, giving rise to highly tumorigenic cell lines^69,70. Similarly, MEK has been implicated in the development of a broad range of human tumours⁷¹. Studies carried out in a large number of tumour types have indicated the importance of MEK, which activates ERK, in driving their proliferation and, often, progression (Table 1).

MEK proteins are dual-specificity kinases that contain two consensus kinase motifs, one involved in phosphorylation of serine/threonine residues and another in phosphorylation of tyrosine residues. Two MEK homologues, MEK1 and MEK2 are ubiquitously expressed in mammals. Both of these MEKs sequentially phosphorylate ERK1 and ERK2 at two sites — Tyr185 followed by Thr183 (Ref. 72). For a discussion of the biological attributes that delineate key differences between MEK1 and MEK2, see Ref. 73. Their high degree of sequence homology makes it likely that a small-molecule MEK inhibitor would target both homologues.

The development of pharmacological inhibitors of MEK was launched with the discovery of PD098059 (Ref. 9). Because of its pharmaceutical limitations, this compound was made available to the academic community and subsequently became an invaluable tool for exploring the role of the MAPK pathway in an array of physiological processes. U0126, a MEK inhibitor shown to be more potent than PD98059, has also become a widely used reagent for investigating the role of this pathway in cellular events⁷⁴. The first MEK inhibitor reported to inhibit tumour growth in vivo was PD184352 (subsequently named CI-1040)⁷⁵. This compound, which was shown to be orally active in early studies, was shown to significantly inhibit the growth of colon carcinomas in animal models. Importantly, efficacy was achieved at well-tolerated doses and correlated with a reduction in the levels of activated ERK in tumour samples. Based on a promising preclinical profile that encompassed effects on proliferation, survival, invasion and secretion of pro-angiogenic growth factors, this agent was advanced into clinical oncology trials^75,76.

CI-1040 has been evaluated in Phase I and Phase II trials^76,77. Because of the poor metabolic stability and bioavailability of this agent observed in Phase I trials, high doses were administered to patients in Phase II trials (Table 2). Whereas there was encouraging evidence of antitumour activity in patients in the Phase I trial, similar results were not observed in Phase II trials, so development of this agent was terminated. Nonetheless, the finding that levels of phosphorylated ERK were reduced by roughly 50% or greater in all of the evaluable tumour biopsies taken during the Phase I trial indicated that this drug did affect its target in humans, and that it might be possible to improve the pharmaceutical properties of CI-1040. As CI-1040 was clinically well tolerated at doses that clearly resulted in significant MEK inhibition, would it be possible to develop more potent MEK-targeted agents with greater systemic exposure?

PD0325901, which is structurally highly similar to CI-1040, was subsequently developed as a significantly more potent MEK inhibitor. It has an IC₅₀ of 1 nM against activated MEK1 and MEK2 (J.S.S.-L., unpublished observations). This compound, which has subnanomolar IC₅₀ values in cells, is also significantly more potent than CI-1040 in vivo. Anticancer activity of PD 0325901 has been demonstrated for a broad spectrum of human tumour xenografts, significantly inhibiting the growth of six out of seven human tumour models tested. The improved anticancer activity of PD0325901 over CI-1040 is probably due to several contributing pharmacological factors, including longer duration and greater potency of MEK inhibition, as well as greater solubility leading to improved bioavailability, and increased metabolic stability. Phase I trials with this agent are now underway and are focusing on patients with solid tumours who are expected to show activation of the MAPK pathway.

The benzimidazole ARRY-142886 has also been reported to be a highly potent MEK inhibitor, with an IC₅₀ of 12 nM against purified MEK¹¹⁰. This compound was reported to inhibit basal phosphorylation of ERK in a range of human tumour cell lines, with an IC₅₀ in the 10 nM range. ARRY-142886 has also been reported to be efficacious against a panel of tumour xenografts, encompassing tumours of pancreatic, colon, breast, lung and skin origin. Based on its collective preclinical profile, ARRY-142886 was advanced into full development and recently entered clinical trials.

All three of these recently developed MEK inhibitors, CI-1040, PD0325901 and ARRY-142886, are highly selective for MEK, as indicated by their inability to inhibit a multitude of serine/threonine and tyrosine kinases. None of these inhibitors compete with ATP binding. Why are these inhibitors so selective for MEK? The answer to this question was provided through structural analysis of human MEK1 or MEK2, each determined as a ternary complex of enzyme, PD0325901-like inhibitor and MgATP⁷⁸ (Box 2). These studies showed that NON-COMPETITIVE INHIBITORS such as PD0325901 can bind to MEK without perturbing the ATP site. The concurrent binding of MgATP and these inhibitors does, however, cause several conformational changes in MEK1 and MEK2. These changes are thought to contribute to their unique non-competitive mechanism of inhibition and help explain the remarkable selectivity of these drugs, as this binding pocket is in a region without sequence homology to other kinases.

Validation of inhibitors of the MAPK pathway as anticancer therapeutics has been long awaited. This quest has been further hampered by the lack of suitable drug candidates to provide clear answers. The high degree of specificity of PD0325901 and ARRY-142886, combined with their favourable pharmaceutical profiles, make these compounds ideal clinical candidates for fulfilling this need.

Clinical challenges

In many ways, the challenges that we face in the design of clinical trials of inhibitors of the RAS–MAPK pathway do not differ substantially from those faced with other anticancer agents. Whether testing a signal-transduction inhibitor or even a cytotoxic agent, one does not know at the outset which tumour types will be most sensitive. However, unlike the case for cytotoxic agents, patients who benefit from treatment with a signal-transduction inhibitor are likely to share common molecular alterations of the target or pathway, such as RAS or RAF mutations driving activation of the MAPK pathway. This means that a subset of patients with certain tumour genotypes will respond to a certain drug in a more consistent manner than will an entire patient population of the same histological tumour type. Designing clinical trials that accrue patients based on the molecular profile of the tumour is therefore a special challenge. However, advances in our understanding of the molecular events that trigger oncogenic transformation make certain tumour subpopulations particularly attractive targets for a given agent. Knowing for instance that melanomas and pancreatic cancers have a high incidence of BRAF- and KRAS-activating mutations, respectively, make patients with these tumour types strong candidates for inclusion in clinical trials with MAPK-pathway inhibitors.

Are trial sponsors ready to take this leap of faith based on strong preclinical science? Only time will tell — clinical trials have often been directed towards the largest patient populations, including patients with lung, breast and colon cancers. The time has clearly arrived to believe in our targeted agents and to test in populations in which we feel that they have a good chance of being most effective, including the smaller patient populations that are enriched for a specific molecular defect. Otherwise, we face a real risk of shelving viable anticancer drugs for the wrong reasons. Along these same lines, it will be important to retrospectively profile tumour samples taken from patients, subsequently correlating the presence of various molecular markers with response outcome. By amassing information on genetic and epigenetic markers, such as RAS and RAF mutational status, commonalities in response determinants will hopefully be deduced. Although analysis of tumour samples for specific biomarkers is important for determining the in vivo effects of signal-transduction inhibitors, we will continue to face the technical challenges of securing sufficient biopsy material to perform these types of studies. Archived tumour material obtained at the time of surgery will probably serve as an invaluable resource when fresh biopsy material is not available.

Will inhibitors of the MAPK pathway be effective as single agents? As for all potential anticancer drugs, this is a formidable challenge, as a high percentage of human tumours possess more than one genetic defect. So it is likely that the ultimate niche for a MAPK-pathway inhibitor will be in combination with another signal-transduction inhibitor. For example, tumours that have upregulated signalling through the RAS–MAPK pathway also frequently have high levels of activated AKT, which is activated by PI3K. The overall result alters the balance between survival versus apoptosis signaling, resulting in cell survival⁵⁴. For maximal therapeutic benefit, it might be necessary to combine agents directed at blocking both pathways. As AKT and PI3K inhibitors, or other inhibitors of this pathway become available for clinical testing, their combination with MAPK-pathway inhibitors should be a high priority. In the meantime, a multitude of other targeted agents have become available that should also be tested in combination with MAPK-pathway inhibitors.

Will toxicity of RAF and MEK inhibitors preclude their ability to sufficiently shut down signalling through the MAPK pathway in tumour cells? Preclinical studies have shown that complete and continuous suppression of MAPK activation is required for maximal therapeutic effects (J.S.S.-L., unpublished observations). Ongoing studies with PD0325901 and ARRY-142866 will determine exactly how much target suppression can be tolerated in patients before unacceptable levels of damage to the gastrointestinal tract and skin occurs.

It is too early to tell whether clinical resistance to MAPK-pathway inhibitors, as has been the case with imatinib, will be encountered. A molecular explanation for acquired resistance to imatinib has been provided by detailed studies characterizing its target, the BCR–ABL chromosome-translocation product. This fusion protein undergoes mutations in its kinase domain that change Thr315 to an isoleucine residue⁷⁹. This hot spot in the ATP-binding site has been also identified in other kinases, such as epidermal growth factor receptor and PDGFR, and might therefore undergo mutations that confer resistance to other drugs that target tyrosine kinases. Indeed, a patient with a gastrointestinal stromal tumour that developed resistance to imatinib was found to contain an analogous mutation in KIT⁸⁰. As we know that the BRAF kinase domain also contains hot spots for oncogenic mutations, RAF inhibitors could well be subject to the same mechanisms of resistance.It has, however, been reported that BAY 43-9006 interacts with an inactive conformation of BRAF, so oncogenic mutants of BRAF might be less sensitive to a RAF-kinase inhibitor, when compared with wild-type BRAF⁶⁷. It is tempting to speculate that the non-ATP-competitive inhibitors of MEK that are now in clinical trials will not be subject to this type of resistance. The very absence of activating mutations, which rendered MEK an undesirable drug target to many researchers years ago, could ultimately allow this enzyme to be an effective therapeutic target.

Future directions

There continues to be a high level of interest in targeting the RAS–MAPK pathway for the development of improved cancer therapies. So, how far have we come? MEK inhibitors represent the only 'clean', or highly selective, MAPK-pathway inhibitors now available. With clinical trial data pending, the MEK inhibitors PD0325901 and ARRY-142886 are positioned to be the first selective inhibitors of the RAS–MAPK pathway to to be validated as effective cancer therapies. However, this does not mean that RAF is no longer an important target. It has simply not been validated as such, because of the non-selectivity issue surrounding BAY 43-9006. Interest in targeting the RAS–MAPK pathway has not waned, in part due to the discovery of BRAF-activating mutations in human tumours. Many research programmes are now focused on the search for selective BRAF inhibitors. As these agents emerge, it will be of great interest to examine their pharmacological properties in the context of those observed for MEK inhibitors. As RAF is known to be involved in MEK-independent survival signalling, it is conceivable that selective BRAF inhibitors could have broader-spectrum anticancer activity than MEK inhibitors. By the same token, they could concurrently possess a higher toxicity potential.

Clearly, RAF and MEK are the only components of the RAS–MAPK pathway that are currently receiving attention as therapeutic targets. However, a multitude of other potential targets in this pathway are available for therapeutic exploration (Fig. 1). For example, a search for inhibitors of downstream targets from ERK, such as RSK, as well as positive allosteric agonists of phosphatases, such as MKP1, represent a largely unexplored area. Although the patent literature cites many examples of ERK inhibitors, none have been evaluated in clinical trials. Of all of the members of the MAPK cascade, it is tempting to speculate that SOS represents one of the most attractive antitumour targets from a functional standpoint. By disrupting the RAS–SOS interaction, one might expect that signalling through both the MAPK and PI3K pathways would be significantly impaired, thereby inhibiting both proliferation and survival of tumour cells. Unfortunately, antagonists of protein–protein interactions face high pharmaceutical hurdles, as they are difficult to develop. Overcoming these hurdles could have huge therapeutic impact.

As we look forward, we need smart drugs and smart clinical trials that take into account known genetic defects of tumours in a given patient population. In this way, we will maximize our chances of delivering improved therapies to patients in desperate need of our success.

Box 1 | BRAF inhibition

Significant advances in our understanding of BRAF function, and the way in which mutations lead to transformation were made by determining the crystal structure of this enzyme⁶⁷. The figure shows a ribbon diagram of the structure of the BRAF wild-type kinase domain looking into the kinase active site cleft occupied by the inhibitor BAY 43-9006. The overall structures of the wild-type BRAF and V599E mutant (not shown) of BRAF are comparable to other inactivated protein kinases like ABL and p38. The BRAF inhibitor binds in the ATP pocket and interacts with residues in the P-loop and with the kinase activation loop. Inhibition is most likely achieved by preventing the activation loop and the catalytic residues from adopting a conformation that is competent to bind and phosphorylate substrate. Interestingly, most oncogenic mutations of BRAF occur in either the P-loop or in the activation loop region. It is believed that the mutations act to destabilize the inactive conformation of the protein kinase, favouring an activated state of the enzyme.

Box 2 | Basis for the specificity of non-ATP-competitive binding of CI-1040-like inhibitors

Important information about mitogen-activated protein kinase kinase (MEK) inhibition was also obtained from structural analysis. The figures shows a ribbon diagram of the structure of MEK1 looking into the ATP pocket. The overall structure of MEK1 appears to be similar to other protein kinases, with an amino-terminal lobe composed of β-strands (green) and a carboxy-terminal lobe that is primarily α-helical (blue). The ATP-binding site is situated in the cleft between these two lobes. The uniqueness of this structure is revealed by the existence of an inhibitor-binding site located in an interior hydrophobic pocket adjacent to, but distinct from, the MgATP-binding site. When CI-1040-like inhibitors such as PD318088 or PD325901 bind to MEK1, it has been proposed that they induce a conformational change in the protein that traps the kinase activation loop into a two-turn C-helix that displaces the C-helix towards the amino terminus. As a result of inhibitor binding, the MEK1 catalytic site is partially occluded such that the extracellular signal-regulated kinase (ERK)-activation loop can no longer access the catalytic pocket and becomes phosphorylated. CI-1040-like inhibitors can thereby prevent the autophosphorylation of MEK1 and MEK2 as well as the phosphorylation of the ERK substrate by trapping MEK in a closed and catalytically inactive conformation.