ReviewFrontiers in Cancer Therapy

Drugging RAS: Know the enemy

See allHide authors and affiliations

Science  17 Mar 2017:
Vol. 355, Issue 6330, pp. 1158-1163
DOI: 10.1126/science.aam7622


The three RAS oncogenes make up the most frequently mutated gene family in human cancer. The well-validated role of mutationally activated RAS genes in driving cancer development and growth has stimulated comprehensive efforts to develop therapeutic strategies to block mutant RAS function for cancer treatment. Disappointingly, despite more than three decades of research effort, clinically effective anti-RAS therapies have remained elusive, prompting a perception that RAS may be undruggable. However, with a greater appreciation of the complexities of RAS that thwarted past efforts, and armed with new technologies and directions, the field is experiencing renewed excitement that mutant RAS may finally be conquered. Here we summarize where these efforts stand.

RAS genes have the distinct honor of being the first mutated genes identified in human cancer, ushering in the era of molecularly targeted anticancer drug discovery. Although our roster of cancer genes now exceeds 600 (COSMIC v80;, the three RAS genes constitute the most frequently mutated gene family in cancer, with RAS mutations found in ~25% of human tumors. Despite more than 30 years of intensive efforts to develop pharmacologic inhibitors of RAS, a clinically effective anti-RAS therapy remains elusive (13). The history of anti-RAS drug discovery is marked by mistakes, missteps, and misconceptions. As the military strategist General Sun Tzu wrote in the Art of War, “If you know neither the enemy nor yourself, you will succumb in every battle.” With the establishment of the RAS Initiative in 2013, the U.S. National Cancer Institute proclaimed a new war on RAS (4). Although our knowledge of RAS remains far from complete, there is a sense that the time is finally at hand to drug a protein once considered undruggable. In this Review, we provide an update and a perspective on current strategies to make anti-RAS therapies a reality (Fig. 1).

Fig. 1 Five general strategies for anti-RAS drug development.

(i) Molecules that directly bind RAS disrupt its interaction with guanine nucleotide exchange factors or with effectors such as the RAF serine-threonine kinases. Also shown are four indirect approaches that target (ii) proteins modulating RAS spatial organization and association with the plasma membrane (e.g., farnesyltransferase and PDEδ), (iii) RAS effector signaling (e.g., RAF and PI3K), (iv) synthetic lethal interactors of mutant RAS, and (v) RAS-regulated metabolic processes in cancer cells.


Aberrant RAS function in cancer

The three human RAS genes are not mutated at equivalent frequencies in cancer. KRAS is the most frequently mutated (85% of all RAS-driven cancers), followed by NRAS (12%) and HRAS (3%) (COSMIC v80). RAS mutations are most common in the top three cancers responsible for cancer deaths in the United States: pancreatic ductal adenocarcinomas (PDACs; 95%), colorectal adenocarcinomas (CRCs; 52%) and lung adenocarcinomas (LACs; 31%) (13). In contrast, RAS mutations are found rarely (<2%) in breast, ovarian, and brain cancers.

KRAS is the predominant isoform mutated in PDACs, CRCs, and LACs. NRAS is the predominant isoform mutated in cutaneous melanomas and acute myelogenous leukemia, whereas HRAS is the predominant isoform mutated in bladder and head and neck squamous cell carcinomas (1). Why KRAS is preferentially mutated overall, and why a particular RAS gene is preferentially mutated in specific cancers, are questions that remain to be fully answered.

Early failures in RAS drug discovery

Early approaches considered for developing RAS inhibitors were guided by understanding the biochemical defects of mutant RAS proteins (5). The three RAS genes encode four closely related small guanosine triphosphatases (GTPases, enzymes that hydrolyze guanosine triphosphate): HRAS, KRAS4A, KRAS4B, and NRAS. Driven by the convenience of available reagents (expression vectors and antibodies), much of the current knowledge of RAS is based largely on earlier studies centered on HRAS. The unsuccessful attempts to develop farnesyltransferase inhibitors (FTIs) as anti-RAS drugs (see below) resulted from the misconception that the four RAS proteins were identical in function. With an ever-growing appreciation that different RAS proteins will have distinct roles in cancer, the field has now shifted its focus to KRAS, the isoform most frequently mutated in cancer, and the predominant splice variant, KRAS4B (6).

RAS proteins function as molecular switches that regulate a diversity of cytoplasmic signal transduction networks (2). RAS can exist in two states: the guanosine diphosphate (GDP)–bound “off” state and the GTP-bound “on” state. Active RAS-GTP binds to a spectrum of catalytically diverse downstream effectors. The RAS GDP-GTP cycle is regulated by guanine nucleotide exchange factors (GEFs; e.g., SOS1) that promote nucleotide exchange and formation of RAS-GTP. GTPase-activating proteins (GAPs; e.g., neurofibromin) stimulate the hydrolysis of the bound GTP, forming inactive RAS-GDP (7).

Cancer-associated RAS genes harbor missense mutations that produce single amino acid substitutions primarily at codons 12, 13, or 61. These mutations impair intrinsic and GAP-stimulated GTP hydrolysis rates and/or increase intrinsic exchange rates, favoring stimulus-independent formation of active RAS-GTP. Thus, the earliest ideas centered on developing small-molecule antagonists of GTP binding. However, the picomolar affinity of RAS for GTP and the millimolar cellular concentrations of GTP rendered such efforts fruitless (1). The search for small-molecule mimetics of GAP active on mutant RAS proteins was likewise unsuccessful. These early experiences, coupled with the lack of well-defined hydrophobic pockets on the surface of RAS proteins, contributed to a perception that RAS may be not tractable (8). This perception led to the pursuit of indirect strategies to target proteins that promote RAS membrane interaction or effector signaling. Later, the concept of synthetic lethality was applied to identify genetic interactors with mutant RAS. With a rebirth in the interest in cancer cell metabolism, a search for RAS-dependent metabolism began (9, 10). Only recently has the field returned to the issue of direct RAS inhibitors, with unexpected findings that suggest that RAS may be a tractable drug target after all. Below, we highlight key recent findings in these areas.

Targeting RAS plasma membrane localization

With the recognition that RAS oncogenic activity is dependent on the protein’s association with the inner face of the plasma membrane, and the subsequent identification of the posttranslational modifications that modulated this association, came the next important direction for anti-RAS drug discovery (11). The RAS isoforms are synthesized initially as cytosolic, inactive proteins. The RAS C-terminal CAAX (C, cysteine; A, aliphatic amino acid; X, terminal amino acid) tetrapeptide motif gives signals for a series of posttranslational modifications (12). The first is farnesyltransferase-catalyzed covalent addition of a farnesyl moiety to the cysteine residue of the CAAX motif. The second, which occurs at the cytosolic surface of the endoplasmic reticulum, is the proteolytic removal of the last three amino acids by RAS converting enzyme 1 (RCE1). Last, isoprenylcysteine carboxyl methyltransferase (ICMT) facilitates methyl transfer to the C-terminal amino acid to negate the negative charge and prevent plasma membrane repulsion. The CAAX-signaled modifications, together with the addition of a palmitate fatty acid (HRAS, KRAS4A, and NRAS) and/or polylysine sequences (KRAS4A and KRAS4B), promote RAS association with the plasma membrane (Fig. 2).

Fig. 2 Regulation of RAS subcellular localization and membrane association.

The C-terminal CAAX motif is recognized by the cytosolic farnesyltransferase (FTASE), which facilitates a covalent addition of a farnesyl isoprenoid to the cysteine residue. In the next step, the endoplasmic reticulum (ER)–associated protease RAS converting enzyme 1 (RCE1) removes the AAX residues, which is followed by ER-associated isoprenylcysteine carboxyl methyltransferase (ICMT)–mediated methyl esterification of the now terminal farnesylated cysteine. Modification of the CAAX box is necessary but not sufficient to promote the plasma membrane association and subcellular localization that are essential for RAS protein function. A second membrane-targeting motif is needed: cysteine residues (in the case of HRAS, NRAS, and KRAS4A) that are covalently modified by addition of a palmitoyl fatty acid, which is executed by Golgi-associated palmitoyl acyltransferases (PATs), or a polylysine sequence (in the case of KRAS4A and KRAS4B). Enrichment of RAS molecules at the plasma membrane requires PDEδ. PDEδ recognizes the farnesyl group of RAS, solubilizes the protein from the cytosol, and returns it to the recycling endosome or the Golgi, from which directed vesicular transport shuttles RAS to the plasma membrane; PDEδ is blocked by small molecules such as deltarasin (i). Farnesyltransferase inhibitors (FTIs) block the first CAAX-signaled modification, preventing all subsequent modifications (ii). In the absence of farnesyltransferase activity, KRAS and NRAS are substrates for geranylgeranyl transferase I (GGTASE-I) and alternative prenylation by covalent addition of a geranylgeranyl isoprenoid, followed by the RCE1 and ICMT modifications. APT, acyl protein thioesterase; Me, methyl.


Given the essential role of the farnesyl lipid modification for all subsequent posttranslational modifications and for RAS oncogenic activity, an intense search for FTIs was initiated in 1990. Many were developed and demonstrated to potently block HRAS-driven growth of cancer cells in vitro and in various mouse models of cancer. Two inhibitors (lonafarnib and tipifarnib) advanced to phase III clinical trials and were tested in cancers with KRAS mutations (pancreatic, colorectal, and lung cancer) but disappointingly showed no efficacy in cancers with high-frequency KRAS mutations. In retrospect, this result was foreshadowed by earlier cell culture studies. In contrast to HRAS, KRAS4B and NRAS had been found to retain membrane association in the presence of FTIs because these isoforms were modified by the related geranylgeranyl isoprenoid (13).

The experience with FTIs diminished the pharmaceutical industry’s interest in anti-RAS drug discovery, particularly with respect to targeting RAS membrane association. Cautiously, in recent years, the field is again revisiting this issue, with a focus on new targets. Several studies have applied unbiased functional screens for previously unidentified components that facilitate KRAS4B plasma membrane association. One chemical library screen identified fendiline, an L-type calcium channel blocker, as a selective inhibitor of KRAS4B but not HRAS or NRAS membrane association (14). Fendiline works relatively nonspecifically by inhibiting acid sphingomyelinase, causing a reduction in plasma membrane phosphatidylserine levels (15).

One of the more intriguing recently identified targets is the prenyl-binding protein phosphodiesterase δ (PDEδ). PDEδ acts as a solubilizing factor that facilitates the transit of RAS proteins to either the Golgi or the recycling endosomes, and from there, RAS is shuttled via directed vesicular transport to the plasma membrane (16) (Fig. 2). Deltarasin and deltazinone are two small molecules with different chemical scaffolds that can occupy the farnesyl-binding pocket of PDEδ. When these molecules were used to inhibit PDEδ, RAS was found to shift from the plasma membrane to the endomembranes; this in turn reduced oncogenic signaling and impaired the tumorigenic growth of RAS-mutant cancer cells in a xenograft model (17, 18). In principle, targeting PDEδ would overcome the concerns encountered with FTIs; however, because PDEδ regulates the function of other farnesylated proteins, inhibitors of PDEδ may be compromised by off-target RAS-independent activities. Because virtually all proteins that influence RAS membrane association are likely to have other substrates, the relative nonspecificity of targeting RAS membrane association remains a concern.

Targeting RAS downstream effector signaling

Blocking effector signaling is one of the most attractive and intensely pursued anti-RAS strategies. However, with at least 11 catalytically diverse downstream effector families, the key questions are which effectors should be targeted and whether concurrent inhibition of multiple effectors is required. The two effector pathways that have attracted the greatest attention are the RAF-MEK-ERK mitogen-activated protein kinase (MAPK) cascade and the PI3K-AKT-mTOR pathway (19, 20). Mutations in genes encoding components of each pathway (BRAF and PIK3CA) are known to drive human cancer development, and their gene products are druggable protein kinases.

Numerous inhibitors against each component of both the RAF-MEK-ERK and PI3K-AKT-mTOR effector pathways have been developed and are under clinical evaluation. The evidence to date suggests that the RAF pathway is the more critical effector in RAS-dependent cancer growth (19). For example, in PDAC, a substantial fraction of the rare cancers harboring wild-type KRAS also harbor either BRAF missense mutations or deletion mutations (2123). In contrast, PI3K mutations can co-occur with RAS mutations, arguing that RAS may not potently activate PI3K signaling. When evaluated in mouse models, mutant BRAF, but not PI3K, phenocopied mutant KRAS and drove the initiation and progression of PDAC (24). In the developmental disorders characterized by germline RAS mutations (the so-called “RASopathies”), mutations in the RAF-MEK-ERK cascade are also found. Last, only components of the RAF-MEK-ERK pathway, and not other effectors, could counteract the loss of RAS function and restore the growth of mouse embryo fibroblasts deficient in all Ras alleles (24). Because this pathway is the key effector for RAS in these divergent settings, we focus on recent developments in the therapeutic targeting of the RAF-MEK-ERK MAPK cascade.

The first kinases in the MAPK pathway that is activated by RAS-GTP are the RAF serine-threonine kinases (ARAF, BRAF, and CRAF/RAF1). The only well-validated RAF substrates are the highly related MEK1 and MEK2 dual-specificity kinases. Similarly, MEK1 and -2 substrates are limited to the related ERK1 and ERK2 serine-threonine kinases. ERK1 and -2 then phosphorylate over 200 cytoplasmic and nuclear substrates; one of them is the transcription factor and oncoprotein MYC, which is one of the key mediators of ERK-dependent cancer growth.

The RAF-MEK-ERK cascade was initially perceived as a simple linear, unidirectional signaling pathway, so molecules targeting RAF and MEK were pursued as inhibitors of ERK activation (25). The first RAF inhibitors (vemurafenib and dabrafenib) are now approved for the treatment of BRAF-mutant melanomas. However, when these inhibitors were evaluated in RAS-mutant cancer cells, a paradoxical activation, rather than inactivation, of ERK signaling was observed. This activity was determined to be driven by mutant RAS–dependent formation of BRAF-CRAF heterodimers (26, 27). Because the first-generation RAF inhibitors are only BRAF-selective, they paradoxically cause activation of CRAF and thereby activation of MEK-ERK signaling. This observation led to the development of second-generation pan-RAF inhibitors (LY3009120 and PLX8394), which do not activate the MAPK pathway in the presence of a RAS mutation (28, 29).

Potent and highly selective inhibitors of MEK1 and -2 have also been developed, two of which (trametinib and cobimetinib) are clinically approved for the treatment of BRAF-mutant melanoma. However, these inhibitors have shown limited efficacy in RAS-mutant cancers, in part because of loss of ERK-mediated feedback inhibitory mechanisms (30). Although increased ERK signaling drives cancer growth, aberrantly high ERK activation can be growth-inhibitory. Thus, feedback inhibition mechanisms are initiated by ERK phosphorylation of components upstream of MEK (e.g., CRAF), dampening flux through the cascade. Consequently, upon MEK inhibition of ERK—which terminates feedback inhibition—increased flux through RAF and MEK can overwhelm the inhibitor block, leading to ERK reactivation. Cancer-associated MEK mutations have also been described that render these cancers insensitive to MEK inhibitors. These unanticipated consequences of MEK inhibition have prompted the development of inhibitors of ERK, with several now under clinical evaluation (25).

The PI3K-AKT-mTOR pathway may play a lesser role in RAS-dependent cancer growth, but it nonetheless serves a highly complementary role for the RAF-MEK-ERK cascade. Resistance to ERK-MAPK pathway inhibitors can be mediated by PI3K-AKT-mTOR activation. Consistent with this cooperative interaction, concurrent inhibition of components of both pathways has shown enhanced antitumor activity in mouse models. Disappointingly, these combinations have not demonstrated the same activities in clinical trials, in part because of toxicities not seen in mouse studies that limit the doses that can be used in cancer patients. Therefore, ongoing studies are aimed at defining combinations that can offset both the toxicity to normal cells and the onset of treatment-induced acquired resistance associated with inhibitors of RAF and PI3K signaling.

Recently, a different approach for blocking RAS effector signaling has been described. Rigosertib was developed originally as a multikinase inhibitor that does not compete with adenosine triphosphate, and it is now under clinical evaluation for the treatment of myeloid dysplastic syndrome. Rigosertib has been reported to have an unanticipated activity—namely, acting as a RAS mimetic to occupy the RAS-binding and RAS association domains of RAS effectors (31). Thus, rigosertib may have the potential to concurrently block RAS activation of RAF and other effectors. However, another recent study reported that rigosertib affinity for RAS-binding and RAS association domains may not be sufficient to block RAS effector signaling. Instead, the observed reduction in ERK was due to an indirect mechanism: stress-induced JNK activation and phosphorylation and inhibition of SOS1 and RAF (32). Because other non–RAS-related activities have been ascribed to rigosertib, future studies are needed to determine the extent to which this agent’s antitumor activity is due to its effects on RAS.

Synthetic lethal interaction partners of mutant RAS

Another indirect approach for targeting mutant RAS is based on the concept of synthetic lethality (33). Synthetic lethal interaction partners for mutant RAS are genes whose functions are essential in RAS-mutant but not wild-type RAS cells. Such genes may or may not encode proteins that are linked directly with the RAS signaling network. In 2009, several groups reported the identification of synthetic lethal interactors of mutant RAS, including several protein kinases (e.g., STK33 and TBK1). However, subsequent studies failed to validate a strong functional linkage of these hits with mutant RAS. To date, the search for synthetic lethal interactions of mutant RAS has not lived up to the high expectations envisioned for this strategy (34).

In retrospect, the high expectations may have been unrealistic, given the limitations in the performance of these RNA interference (RNAi)–based screens. A common strategy applied in these screens was the use of so-called isogenic human tumor cell lines that differed only in their RAS mutation status. These paired cell lines were established by stable deletion of the mutant KRAS allele in colorectal carcinoma cell lines heterozygous for KRAS G13D mutations (G, glycine; D, aspartic acid). Acute suppression of mutant RAS function severely impairs cell proliferation; thus, cells that arose after loss of mutant RAS were likely to have undergone adaptive changes to escape their RAS addiction. If so, RAS mutation status would not be the only difference between the paired cell lines. Another issue with these screens is the underappreciation of the substantial off-target activities of RNAi libraries. That a number of studies used the same cell model for screening, yet identified completely different sets of synthetic lethal genes, was perhaps another indication of the technology’s limitations.

Despite the setbacks from earlier screens, there is still enthusiasm that the search for synthetic lethal interactors of mutant RAS may bear fruit. To overcome some of the limitations of previous screens, Sabatini and colleagues performed a CRISPR-Cas9–based screen for which off-target activities are expected to be less, using a large panel of RAS-mutant and wild-type acute myelogenous leukemia cell lines. They identified five synthetic lethal genes. Although RCE1 and ICMT are enzymes involved in promoting RAS membrane association, and RAF1, SHOC2, and the PREX1 RAC-selective GEF are known regulators of the ERK-MAPK pathway, no previously unknown RAS modulators were identified (35). The tight association of the hits with RAS function supports the accuracy of these screens in identifying bona fide regulators of mutant RAS function. Going forward, synthetic lethality screens might be further improved by the use of three-dimensional culture systems (e.g., organoid cultures) or in vivo models. However, there is a realistic expectation that synthetic lethal interactors will not be universally linked to all RAS-mutant cancers and that their lethality may be influenced by genetic alterations unrelated to RAS mutations.

Targeting RAS-regulated metabolic processes

One of the 10 hallmarks of cancer is the need for cancer cells to reprogram energy and nutrient metabolism to support their elevated proliferative state (36). These increased needs are met by reprogramming various metabolic processes to either recycle intracellular fuel sources or to scavenge extracellular components (Fig. 3). Recent studies have implicated mutant RAS in the regulation of these metabolic processes (9, 10, 37). Thus, delineating the mechanisms by which aberrant RAS function alters cell metabolism may define new targets to exploit the vulnerability of RAS-mutant cancers.

Fig. 3 RAS-regulated metabolic processes.

(A) RAS-mutant cancers are characterized by increased macropinocytosis and autophagy (i). These processes generate additional energy and macromolecules needed to accommodate the accelerated growth rate of cancer cells. (B) Oncogenic KRAS directs glucose metabolism into hexosamine biosynthetic pathways by up-regulating many key enzymes involved in glycolysis metabolism (ii). Additionally, oncogenic KRAS induces the nonoxidative pentose phosphate pathway (iii) to support increased nucleic acid biosynthesis. PDACs also use a noncanonical pathway to process glutamine (iv), which is used to maintain redox balance and support growth. Red text indicates RAS-dependent gene and/or protein expression, with arrows indicating increased or decreased expression. GLUT1, glucose transporter 1; HK1/2, hexokinase 1 and 2; P, phosphate; BP, biphosphate; PFK1, phosphofructokinase 1; ENO1, enolase 1; PKM, pyruvate kinase; LDHA, lactate dehydrogenase A; GFPT1, glucosamine-fructose-6-phosphate aminotransferase 1; GlcN, glucosamine; GlcNAc, N-acetylglucosamine; UDP, uridine diphosphate glucose; RPE, ribulose-5-phosphate-3 epimerase; RPIA, ribulose-5-phosphate isomerase; GLUD1, glutamate dehydrogenase 1; GOT, aspartate transaminase; GLS, glutaminase; GSH, glutathione; GSSG, glutathione disulfide; MDH1, malate dehydrogenase; ME1, malic enzyme; NADPH, nicotinamide adenine dinucleotide phosphate (reduced); NADP+, nicotinamide adenine dinucleotide phosphate (oxidized); ROS, reactive oxygen species; TCA, tricarboxylic acid; CoA, coenzyme A.


Oncogenic RAS is known to drive macropinocytosis, an endocytic process in which cells use their plasma membrane to internalize nutrient-containing fluids. KRAS-mutant PDAC cell lines and a KRAS-driven mouse model of PDAC were found to have increased macropinocytotic activity (38). There are no pharmacologic inhibitors that selectively block macropinocytosis, but EIPA [5-(N-ethyl-N-isopropyl)amiloride], an inhibitor of Na+/H+ exchange, was reported to inhibit this process and thereby impair PDAC tumor formation in vivo. Although the elevated macropinocytotic activity in PDAC is RAS-dependent, the effector signaling mechanisms driving this activity remain poorly understood. One of the key components ingested by macropinocytosis is albumin, which is then used as an additional amino acid source. The process of macropinocytosis is speculated to be hijacked by nanoparticles, which are labeled with the albumin-bound chemotherapy agent paclitaxel (nab-paclitaxel), a standard of care for PDAC.

Another route used by oncogenic RAS-driven cancer cells to generate the macromolecules required for cell proliferation is macroautophagy (hereafter referred to as autophagy). During autophagy, the cell degrades unneeded or dysfunctional intracellular components, including cellular organelles, which provides a supply of amino acids, lipids, and nucleic acids. Autophagy was observed to be elevated in KRAS-driven PDAC cell lines, and chloroquine inhibition of lysosomal acidification and preventing autolysosome formation impaired PDAC tumorigenic growth in mice (39). Chloroquine is a relatively nonspecific inhibitor of autophagy and can also inhibit macropinocytosis. The related approved drug, hydroxychloroquine, is now under clinical evaluation for PDAC ( identifier, NCT01506973). Autophagy is a mechanistically complex process and offers potentially more selective targets for therapeutic intervention (e.g., the ULK serine-threonine kinases) (40).

Mutant RAS has been linked to increased glucose metabolism and the diversion of glucose metabolites into nucleotide and lipid biosynthetic pathways (41). RAS can drive increased glucose uptake by up-regulating expression of the glucose transporter GLUT1. KRAS also up-regulates glycolytic enzymes to enhance conversion of pyruvate to lactate. In addition, RAS regulates genes encoding enzymes that route glucose intermediates into the nonoxidative pentose phosphate and hexosamine biosynthetic pathways. KRAS regulates these enzymes through the RAF-MEK-ERK pathway and regulation of MYC expression. Thus, inhibitors of the ERK-MAPK pathway may block KRAS-driven glycolytic perturbations.

Last, RAS-mutant PDAC is characterized by increased dependency on glutamine for NADPH (nicotinamide adenine dinucleotide phosphate, reduced form) production to maintain redox balance (42). Mutant KRAS regulates flux through a previously undescribed pathway in which the mitochondrial aspartate aminotransferase (GOT2) generates glutamine-derived aspartate; this aspartate is then converted to oxaloacetate by the cytosolic aspartate aminotransferase (GOT1). Malate dehydrogenase (MDH1) converts the oxaloacetate to malate. Malic enzyme (ME1) then uses malate to produce NADPH to maintain redox balance. CB-839, an inhibitor of glutaminase (GLS1) that regulates the first step in glutamine metabolism, is under clinical evaluation for leukemia and solid cancers ( Inhibitors of other components of this metabolic pathway are being pursued in preclinical studies. In summary, although promising directions have been established, it is still early days in mining RAS-dependent metabolic functions for cancer treatment.

Direct inhibition of RAS

When the crystal structure of HRAS was reported in 1989 (43), the surface topology did not reveal “pockets” that could serve as selective interaction sites for small molecules. This deterred the search for direct RAS-binding molecules for several years, but more recently, considerable progress has been made in this arena. In 2012, using chemical library screens, several groups identified small molecules that directly interact with and perturb the function of recombinant KRAS4B. Several compounds interacted with KRAS4B at a site important for SOS1 interaction, preventing effective stimulation of exchange and formation of active KRAS-GTP (38, 44). One set of compounds bound to the KRAS effector interaction domain, prevented its interaction with CRAF, and additionally impaired activation of other effectors (45). These molecules exhibited low (micromolar) affinity in vitro. Although they inhibited cancer cell growth in cell-based assays, whether this effect was due to antagonism of KRAS function or to off-target activities is unclear. Nevertheless, these early successes spurred other investigators to search for direct inhibitors of RAS. For this strategy to be successful, the RAS-selective small molecules must bind with high affinity, target a face of KRAS4B that is essential for oncogenic function, and have the pharmacologic properties needed for clinical development.

Among the molecules that directly bind RAS, the most provocative class comprises those designed to recognize the specific RAS mutation G12C. Although G12C accounts for only 12% of KRAS mutations in all human cancers, it is the major KRAS mutation in lung cancer. Shokat and colleagues developed a GDP analog that binds selectively and covalently to oncogenic RAS G12C, locking it in the inactive state (46). Westover and colleagues independently developed another G12C-selective inhibitor (47). These initial G12C inhibitors were important proof-of-concept compounds but had limited cellular activity. Subsequent refinement led to the development of a compound called ARS-853 with the same mechanism of action and with improved biochemical and cellular activities; this compound strongly inhibited proliferation of cancer cells harboring RAS G12C mutations (48, 49).

An important revelation from the evaluation of ARS-853 was that different mutations can have different consequences on RAS function. Before this work, it was generally assumed that mutant RAS proteins were persistently GTP-bound, independently of signaling activities that stimulated GEF-dependent activity. Instead, the full activation state of KRAS G12C was found to depend on epidermal growth factor receptor (EGFR)–mediated and GEF-dependent activation. Thus, concurrent EGFR inhibition together with ARS-853 treatment more effectively blocked KRAS G12C function. Therefore, inhibitors of GEF stimulation of RAS may be useful for at least some RAS mutants. This finding should stimulate the search for mutation-selective defects for the development of mutation-specific anti-RAS therapies.

Concluding thoughts and future perspectives

Despite the tortuous history of anti-RAS drug discovery, there is renewed optimism that the goal will be achieved in the not-too-distant future. Tempering that optimism, it is also clear that RAS-mutant cancer cells still hold many secrets, and these may lead to further defeats on the RAS battlefield. One emerging reality is that when we are ultimately successful, it will not be a simple one-size-fits-all therapy that will be effective in all RAS-mutant cancers. The function of mutant RAS is greatly affected by cell and genetic context. Given the genetic heterogeneity of RAS-mutant cancers, multiple anti-RAS strategies will need to be developed and matched to subsets of RAS-mutant cancers. New discoveries are emerging at an accelerated pace. Investigators new to the RAS field have recently joined the effort, bringing fresh ideas and approaches to target RAS. Which approach will lead to the first clinically approved RAS inhibitor? Which approach will yield the most effective RAS inhibitors? With inhibitors already under clinical evaluation, we predict that effector inhibitor–based therapies will be the first effective anti-RAS therapies to reach the clinic. However, direct inhibitors of RAS may prove to be the most effective anti-RAS therapy in the long term. Will other approaches prove more successful? With the rebirth of cancer immunotherapy, perhaps linkages between RAS signaling and tumor immunity will be identified and exploited. Suppression of mutant RAS expression by RNAi or CRISPR-Cas9 could also potentially be effective, if delivery and potency issues are addressed. In summary, although the hurdles to success remain considerable, there is confidence that researchers are ready to finally defeat RAS.

References and Notes

Acknowledgments: C.J.D. is on the Scientific Advisory Boards of Warp Drive Bio, a company developing therapeutics for proteins that cannot be targeted by conventional drug discovery approaches, and Kyras Therapeutics, a company developing RAS-targeted drugs. C.J.D. is also a paid consultant for Astex Pharmaceutics, Novartis, LifeSci Advisors, and Cullinan Pharmaceuticals.

Stay Connected to Science

Navigate This Article