Personalizing KRAS-Mutant Allele–Specific Therapies

Mutations that constitutively activate the small GTPase KRAS are the most common oncogenic events in human cancers. They occur in nearly 25% of all tumors worldwide (∼4.25 million cases/year), with a prevalence in adenocarcinomas, and are commonly associated with poor prognosis and therapeutic resistance (1). One of the most challenging yet unresolved questions in oncology is the tissue- and context-specific mutation pattern of KRAS (2): Mutation frequencies vary from 90% in pancreatic ductal adenocarcinoma (PDAC), to 40% in colorectal cancer and <5% in breast and kidney cancers. In addition, the position and type of KRAS mutations are heterogeneous and differ substantially among cancer entities (1). Most point mutations cluster in three hotspots resulting in alteration of amino acids G12, G13, or Q61, with the majority of mutations causing substitutions at codon G12. Thereby, KRAS remains in the GTP-bound active state, leading to the activation of multiple downstream effectors, such as the RAF kinases, the p110α subunit of PI3K, and RalGEFs, all of which are capable of transmitting oncogenic signals to the cell (1). Interestingly, the type of amino acid substitution at codon 12 again shows tissue specificity. In PDAC, for example, KRAS primarily presents a G12D mutation, whereas lung cancers harbor predominantly G12C substitutions. Furthermore, in the spectrum of possible G12 missense substitutions, KRAS^G12R mutations are rare in colorectal cancer and lung cancer (∼1%), but are the third most predominant alteration in PDAC (∼17%; ref. 3).

This context-specific pattern of KRAS mutations suggests fundamental biologically and therapeutically relevant differences that depend on the tissue of origin and mutation type. Indeed, G12C mutation–specific pharmacologic inhibitors have entered clinical trials, demonstrating the possibility to target oncogenic KRAS with mutant allele–specific precision (4). Therefore, it is of fundamental importance to elucidate the causes and consequences of distinct KRAS mutations in context. This might provide us with unique possibilities to develop novel mutation-specific therapies for KRAS-driven cancers.

In this issue of Cancer Discovery, Hobbs and colleagues (3) illustrate allele-specific differences in downstream effector signaling pathways of oncogenic KRAS^G12R (Fig. 1). They show that the G12R substitution is defective in interaction and activation of a key KRAS downstream target, the class IA PI3K p110α (PIK3CA). This interaction, as well as the subsequent KRAS-mediated induction of PI3K signaling, has been demonstrated to be important for oncogenic KRAS^G12D-induced lung and pancreatic carcinogenesis (2, 5). Disruption of the KRAS/p110α interaction reduced the number of KRAS^G12D-induced lung tumors substantially (5). Furthermore, conditional deletion of the catalytic domain of p110α or the PI3K downstream target Pdkp1 blocked KRAS^G12D-driven PDAC development completely (2).

How do pancreatic tumors harboring the KRAS^G12R mutation evolve and how do they compensate their inability to activate p110α/PI3K directly? This study sheds light on this question by demonstrating differences in the metabolic rewiring of PDAC cells harboring G12D, G12V, and G12R mutations. They show that KRAS^G12D/V, but not KRAS^G12R, drive macropinocytosis in a MYC-dependent manner. Macropinocytosis is an important nutrient-scavenging pathway that helps cancer cells to replenish their intracellular amino acid and lipid pools to meet their extremely high metabolic demands. Therefore, even under hypoxic and nutrient-deprived conditions, a hallmark of the desmoplastic and poorly perfused PDAC microenvironment, macropinocytosis sustains tumor-cell survival and proliferation. Macropinocytosis depends on the remodeling of the actin cytoskeleton and leads to the engulfment of extracellular macromolecules and fluids into intracellular vesicles, which fuse with lysosomes for the digestion of the cargo. This process is controlled by small GTPases, such as RAC1 and ARF6, PAK1, and phosphoinositide signaling, including PI3Ks. It has been shown that macropinocytosis leads to increased cell proliferation, and inhibition of this process in mice using a macropinosome formation inhibitor (EIPA) reduced tumor growth, demonstrating therapeutic potential (6).

Surprisingly, KRAS^G12R-mutated PDAC cells do not lack the ability to perform macropinocytosis. Hobbs and colleagues demonstrate that they compensate and bypass the lack of KRAS-mediated p110α stimulation by relying on p110γ (PIK3CG), a class IB PI3K (Fig. 1). PI3K are a family of lipid kinases involved in cell growth, division, and survival. PI3K comes in several isoforms. PIK3CA and PIK3CB have a broad tissue distribution, whereas expression of PIK3CG and p110δ (PIK3CD) is restricted and predominantly detected in leukocytes. However, the authors show that both isoforms are aberrantly expressed in KRAS-mutant PDAC cells, and that pharmacologic blockade or siRNA-based downregulation of p110γ reduces their ability to perform macropinocytosis. Unexpectedly, these effects were independent of the G12 mutation variant. p110γ inhibition and knockdown decreased macropinocytosis not only in G12R- but also in G12D/V-mutant PDAC cells. This suggests that p110γ plays an important role in both a KRAS-dependent and KRAS-independent manner (Fig. 1). The underlying molecular mechanisms are yet unknown, and additional work is clearly needed to decipher the function and therapeutic value of the various PI3K subunits. In Dictyostelium cells it has been shown that different PI3Ks are differentially activated by RAS isoforms and that this might affect the regulation of macropinocytosis (7).

Previous reports highlighted that aberrant expression of PIK3CB, PIK3CD, and PIK3CG is sufficient to transform cultured cells (8). In contrast to the alpha and delta isoforms, signaling to p110γ depends on G protein–coupled receptors (GPCR; Fig. 1). Clearly, additional work is needed to uncover these signals and the GPCRs involved, as well as PIK3CG downstream signaling nodes that regulate macropinocytosis and other PI3K-dependent effector pathways. This might open novel possibilities of targeting KRAS^G12R mutants and their metabolic dependencies. Furthermore, it allows us to uncover additional mechanisms and vulnerabilities that result from the loss of the KRAS^G12R/p110α interaction and to understand whether p110γ is able to compensate and keep all the p110α effectors active. Finally, it might enable us to elucidate which intracellular signaling cues or paracrine signals from the tumor microenvironment permit selected G12 mutants to survive and form tumors in the pancreas.

Several studies connected p110α and p110γ to RAC GTPases. Once activated, class I PI3Ks produce PtdIns(3,4,5)P3, which functions as second messenger that recruits and activates phosphoinositide-binding proteins, such as RAC1. In its active state, RAC1 signals to several proteins to remodel the actin cytoskeleton and induce macropinocytosis. In this study, Hobbs and colleagues observed that siRNA suppression of RAC1 reduced macropinocytosis, irrespective of the KRAS mutation status, validating an essential role of RAC1 for macropinocytosis. The impact of PIK3CA and PIK3CG on RAC1 activation levels and RAC1-mediated macropinocytosis has yet to be characterized.

To uncover the therapeutic dependencies of KRAS^G12R-mutant PDAC, it will be fundamental to investigate the genetic makeup and the transcriptional and metabolic subtypes of these tumors. In KRAS^G12D-driven PDAC it has been shown that KRAS amplification or the amplification of alternative oncogenes, such as MYC, YAP, or NFKB2, plays an important role; this has implications for the phenotype and clinical behavior of the resulting tumors, such as grading, epithelial-to-mesenchymal transition, aggressiveness, metastasis, and survival (9). In addition, it is of particular interest to investigate whether components of the PI3K pathway are amplified or altered during KRAS^G12R-driven tumor evolution, thereby increasing PI3K signaling output and enabling bypass of p110α activation.

To explore whether KRAS^G12R mutations confer specific pharmacologic vulnerabilities, Hobbs and colleagues performed large-scale drug screens. They observed that PDAC cells harboring KRAS^G12R mutations were more sensitive to MEK inhibition than those with other KRAS mutations. The KRAS^G12R-mutant lines were also more sensitive to PI3K inhibition, likely due to lack of KRAS signaling through this pathway. Moreover, KRAS^G12R PDAC showed increased single-agent sensitivity to both chloroquine and mepacrine, two agents targeting autophagic flux. However, the decrease in viability upon dual treatment with ERK inhibitor plus chloroquine was only additive, with markedly less synergy than in non-G12R KRAS-mutant PDAC cells.

Although this study is an important step forward toward therapeutic targeting of allele-specific KRAS mutations, it raises also several questions. Single-agent PDAC therapies failed in the clinic, and the combination of ERK and autophagy inhibition of KRAS^G12R mutants lacks a high synergy score. Therefore, it remains unclear whether the G12R mutation confers specific therapeutic vulnerabilities that can be translated into clinical application; further studies including synthetic lethality and combinatorial drug screens (e.g., with MEK and PI3K inhibitors as backbones) are needed to uncover KRAS mutation–specific fitness factors. Whether direct targeting of macropinocytosis, for example by EIPA in combination with other drugs, is an option for G12R mutants remains to be investigated. Furthermore, comparative studies including standard-of-care treatments, such as FOLFIRINOX and gemcitabine/abraxane, are needed in the future to test side-by-side the efficacy and superiority of novel therapies.

Classifying solid cancers according to oncogenic alterations is shifting clinical decision making. Examples like the VIKTORY trial demonstrate the feasibility and the success of such concepts (10). However, despite being stratified by an oncogenic alteration, only a fraction of patients respond to the targeted therapy. Therefore, a more detailed, sophisticated, and mechanism-based stratification is needed to improve marker-selected therapies.

Individualizing therapies according to specific KRAS mutations is the ultimate goal. With the availability of G12C-specific KRAS inhibitors, these approaches entered clinical testing. Beyond direct inhibition, a comprehensive knowledge of KRAS mutation–specific signaling outputs, phenotypes, and vulnerabilities in tissue context is a complementary translational approach. Evaluating tumor heterogeneity induced, for example, by co-occurring alterations or by the tumor microenvironment that might affect the overall signaling level is of fundamental importance for patient stratification.

No potential conflicts of interest were disclosed.