Heterogeneity in RAS mutations: One size does not fit all

Abstract

Although oncogenic driver mutations in RAS occur in 20% of cancers, heterogeneity in the biologic outputs of different RAS mutants has hampered efforts to develop effective treatments for RAS-mutated cancers. In this issue of Science Signaling, Huynh et al. show that even among KRAS^Q61 mutants, the specific amino acid that is substituted substantially affects mutant KRAS biologic activity and oncogenicity.

The genes KRAS, NRAS, and HRAS encode four highly homologous protein isoforms: KRAS4B, KRAS4A, NRAS, and HRAS. Whereas 85% of RAS mutations occur in KRAS, mutations in NRAS and HRAS also drive oncogenesis. RAS-mutated cancers respond poorly to standard chemotherapy, and for over 40 years, scientists have searched for adequate approaches to treating RAS-mutated cancers. At each stage, we have learned that not all RAS mutants are equal—that RAS is not necessarily RAS. This lesson was first learned when developing and testing farnesyl transferase inhibitors (FTIs) as anti-RAS therapeutics. RAS proteins are lipid-modified by farnesylation in their C-terminal CAAX motif before membrane localization. However, although FTIs showed preclinical promise in HRAS-mutated cancer, they had no benefit in clinical trials for KRAS-mutated pancreatic cancer. Subsequent analysis revealed that FTIs uniquely inhibit HRAS; KRAS and NRAS can be alternatively geranylgeranylated and thus are resistant to FTIs. This first lesson, that all RAS proteins do not have equal biochemical regulation, has affected subsequent efforts to understand the biology of mutated RAS. RAS isoforms show unique biological activities toward their main effector kinases, PI3K and RAF. Compared to HRAS, mutant KRAS is a weak activator of PI3K signaling—it requires wild-type RAS activation, induced by upstream signaling through the guanine nucleotide exchange factor SOS2, to fully activate PI3K (1)—but strongly activates RAF (BRAF, CRAF, and ARAF) through high-affinity interactions (2). HRAS strongly interacts only with CRAF, making HRAS-mutated cancers uniquely sensitive to CRAF deletion (Fig. 1A) (2). NRAS shows an intermediate phenotype; mutant NRAS-PI3K signaling is augmented by SOS2 (1) and has moderate affinity for ARAF and BRAF (2). Thus, KRAS is not NRAS is not HRAS.

Fig. 1. Heterogeneity in the biologic activity of mutant RAS exists at multiple levels.

(A) Mutant RAS family members show differential sensitivities to FTIs and differential transduction to RAF-MEK-ERK (1, 2) versus PI3K-AKT (1) pathways. (B) In KRAS, hotspot mutants use different mechanisms to alter GTP/GDP cycling (3) and increase RAS^GTP levels. These differences affect both sensitivity to proximal RTK pathway inhibitors (4) and the extent of ERK activation (5, 7, 10). Increasing ERK signaling enhances oncogenicity in G13 and G12 mutants but may induce oncogenic stress in Q61 mutants to limit oncogenicity (5–7, 10). (C) Focusing further still on Q61 specifically, the various KRAS^Q61 mutant alleles show relative abundancies that can be classified as common (Q61H, Q61L, and Q61R), less common (Q61K), or rare (Q61E and Q61P) (7). Expression of the common Q61 mutants induces transformation by increasing KRAS-dependent cell metabolism and destabilizing actin stress fibers (7), whereas rare Q61P/Q61E mutants are nontransforming (7). Although Q61K is also transforming, its expression is inhibited by a rare cryptic splice donor site and requires a silent (G60G) co-mutation to be expressed (8). These defects in either expression (Q61K) or biologic activity (Q61P/Q61E) help explain why these Q61 mutant alleles occur in cancers less frequently than would be predicted on the basis of mutational spectra of each cancer type.

Oncogenic KRAS mutations (98%) occur at one of three mutational hotspots—Gly¹², Gly¹³, or Gln⁶¹ (commonly referred to as G12, G13, or Q61)—that dysregulate RAS GTP/GDP cycling to promote RAS^GTP loading. The mechanism leading to increased RAS^GTP is unique to each hotspot (3), and these differences have therapeutic implications for targeting RAS-mutated cancers. Whereas mutations at all three hotspots decrease GTP hydrolysis, Q61 mutants show markedly reduced intrinsic hydrolysis compared to G12 and G13 mutants and are thus relatively insensitive to upstream inhibition of SOS1 or SHP2 (4). In contrast, although G13 mutants show increased intrinsic nucleotide exchange, they are weaker oncogenes than G12 or Q61 mutants due to an open, disordered, and unstable active site (5), an effect that may account for the less-than-expected frequency of KRAS^G13-mutated cancers (Fig. 1B) (6). Thus, among KRAS mutants, G12 is not G13 is not Q61.

Huynh et al. (7) add to the growing body of evidence that a third level of specificity, allele and/or tissue specificity, regulates the biological activity of KRAS-mutated cancers. By comparing predicted versus observed frequencies of Q61 mutant alleles [Glu (E), His (H), Lys (K), Leu (L), Pro (P), and Arg (R)] within the pool of KRAS^Q61-mutated tumors, the authors found substantial deviation between the expected and actual mutation frequencies for several Q61 mutants indicative of a more dynamic process of tumor initiation than could be explained by random mutagenesis. In general, Q61H, Q61L, and Q61R mutations occurred as, or more frequently than, expected, whereas Q61K, Q61P, and Q61E mutations were rare and less frequent. Although not explored by Huynh et al., another study, revealing the induction of a splice donor site and requirement of a silent co-mutation (8), helps explain the low frequency of Q61K-mutated tumors despite its high biological activity.

The authors further explored five of these mutant alleles and found that, when expressed in immortalized fibroblasts or pancreatic cells, Q61H, Q61L, and Q61R mutants were transforming, whereas Q61P and Q61E mutants were not, which may account for their low frequency in tumors. Intriguingly, the authors found that these differences were not due to changes in nucleotide exchange, effector binding, or RAF-MEK-ERK pathway activation. Instead, they found that cells expressing transforming mutants had greater metabolic function and fewer F-actin stress fibers than those expressing nontransforming mutants (Fig. 1C).

The variance in biological outcomes for allelic Q61 mutants is similar to an observation previously reported by the Der laboratory where, in pancreatic cancer, G12R-mutant KRAS showed unique biology compared to G12D and G12V mutants (9); G12R showed defective PI3Kα signaling and was thus unable to induce KRAS-dependent macropinocytosis or suppress autophagy compared to G12D and G12V mutants. Further, whereas KRAS^G12R cells showed enhanced sensitivity to inhibitors of MEK-ERK signaling or of autophagy as single agents, combination therapy was additive, in contrast to the strong synergy seen in KRAS^G12D/V cells. Thus, even among Q61 or among G12 alleles, Q61H is not Q61E, G12D is not G12R, etc.

KRAS^Q61-mutated cancers occur much less frequently than would be expected on the basis of mutational signatures (6). Huynh et al. provide insight into the relative paucity of KRAS^Q61-mutated pancreatic cancers. Compared to G12D, Q61-mutant cells exhibited increased RAF-MEK-ERK pathway activation (7), likely due to both a decrease in KRAS GTP hydrolysis and an increase in nucleotide exchange (3). This increased oncogenic stress may enhance both senescence and apoptotic signaling, thereby reducing the oncogenicity of Q61 mutants compared to G12 mutants (10). Alternatively, Q61 mutants stimulated macropinocytosis to a lesser extent than did G12D; thus, reduced nutrient availability may contribute to the relative lack of KRAS^Q61-mutated pancreatic cancers (7). Despite these signaling changes, the authors found that, similar to G12D/V-mutated cells, combined inhibition of both ERK and autophagy showed modest synergy and effectively killed KRAS^Q61-mutated pancreatic cancer cells, giving further support for exploring this treatment regimen in PDAC.

Studying the unique biological effects of RAS mutations among RAS family members, RAS hotspot mutations, and individual mutant RAS alleles is essential to understanding therapeutic vulnerabilities in RAS-mutated cancers. RAS mutants with decreased oncogenicity may have greater dependency on secondary mutations (7, 10) or require tissue-specific signaling environments to promote cancer growth (9). These secondary changes can create a heterogeneous signaling environment that may subsequently alter therapeutic efficacy. Careful characterization of RAS mutants with respect to allele and tissue specificity is imperative to the successful pursuit of targeting RAS in cancer.