Mutation-Specific Approaches to KRAS Cancers: What We Can Learn from G12C-Directed Inhibitors
, by Sharon Campbell
Directly inhibiting oncogenic RAS has proven to be an arduous task. After more than thirty years of intensive investigation, no clinically relevant therapies exist1. Unlike other pharmacological targets that bind nucleotide ligands, such as protein kinases, RAS binds to GTP with high specificity and affinity, making the development of competitive inhibitors difficult. In addition, RAS proteins lack native pockets that can be targeted using small molecules2. However, by building on the foundation of knowledge acquired over the last three decades, the field has now significantly advanced. New promising efforts are in place, including the NCI RAS initiative.
RAS is a key anti-cancer target, as oncogenic mutations in RAS are found in ~ 30% of human cancers3. Despite findings that most hotspot oncogenic RAS mutations at positions 12, 13 and 61 are constitutively activated through a defect in the ability to hydrolyze GTP, it has become evident that not all oncogenic RAS mutations impart equal effects on signaling and tumor growth4. Hence, mutation-specific therapies are now being examined, with particular focus on KRAS, as it is the most prevalent mutated RAS isoform in cancer. Recently developed mutation-specific therapies have been directed to target KRAS G12C, a mutant that accounts for 12% of all KRAS G12 mutations and 40% of KRAS mutations in non-small-cell lung carcinomas5,6. The G12C cysteine residue provides a handle to covalently modify the KRAS nucleotide-binding pocket, and two distinct classes of compounds have been developed to tether thiol-reactive small molecules to the KRAS G12 cysteine for inhibition7-9.
Shokat and colleagues identified several novel small molecule fragments that covalently modify the mutant KRAS G12C thiol9. These compounds achieve RAS specificity by inducing the formation of an allosteric pocket proximal to Switch II. Switch II is one of two nucleotide-dependent and conformationally dynamic regions that are critical for RAS regulator and effector interactions. Modification of KRAS G12C with these cysteine-tethering agents appears to inactivate RAS by locking it in an inactive GDP-bound conformation, thus impairing effector binding and activation. Treatment of KRAS G12C-containing cell lines using one of the inhibitors, compound 12, demonstrated moderate efficacy, with a few cell lines showing decreased cellular viability and increased apoptosis.
Westover, Gray and colleagues also targeted the KRAS G12C thiol for covalent modification, but used a distinct approach. They developed thiol-reactive GDP analogs to inactivate RAS by directly competing with nucleotide binding7,8. The SML-8-73-1 nucleotide-analog covalently modifies the G12C thiol, forcing the protein into an inactive conformation mimicking the GDP-bound state. Using a cell-permeable analog of this compound, mild decreases in signaling and cell proliferation were seen in a KRAS G12C-containing lung cancer line; however, similar growth defects were also seen in a cell line lacking this KRAS mutation, suggesting limitations in specificity.
As these are first-generation compounds, limitations in compound efficacy have been observed. The class of compounds reported by Shokat and colleagues target the inactive GDP-bound state of KRAS G12C by design. As the G12C mutant is impaired in GTPase activating protein (GAP)-mediated GTP hydrolysis, the predominant species in vivo is likely GTP-bound RAS. Consequently, limited reactivity with hyperactivated RAS may limit the reaction of these compounds in vivo, and may explain the variable efficacy of compound 12 when used to treat mutant KRAS G12C cell lines9. Likely, efficient modification by these compounds will require KRAS G12C to undergo nucleotide exchange, though the degree of nucleotide cycling by oncogenic RAS mutants is not well characterized in vivo. Further, due to the nature of targeting a surface-available cysteine for chemical modification, off-target modifications are possible.
We and others have shown that RAS is susceptible to cysteine oxidation and that RAS activity can be regulated by oxidative modification at native thiol residues10-12. Preliminary work from our lab has shown that the KRAS G12C thiol has a greatly diminished pKa compared to that of a free cysteine (pH 7.4 vs. pH 8.5)13. This altered pKa populates the KRAS G12C thiol in its thiolate state (-S• vs. –SH), making it more reactive to both oxidation and electrophilic modification. Although a distinct highly solvent accessible cysteine exists in the core RAS nucleotide-binding domain, this cysteine (C118) does not have an altered pKa and is less reactive to thiol reactive compounds relative to G12C. Consistent with these observations, we find that KRAS G12C is readily oxidized at physiological pH by cellular oxidants. These findings suggest that KRAS G12C will undergo modification by cellular oxidants in vivo, which could potentially compete with KRAS G12C inhibitors. The type of oxidative modification will vary depending on the specific redox microenvironment. Of note, the KRAS G12C mutation appears most frequently in lung adenocarcinomas, an organ-specific cancer which has a unique redox state distinct from other tissues. Further complicating the diverse array of redox reactions that can take place in a tumor, the redox microenvironment is dependent on spatial and temporal expression of cellular oxidants, antioxidants and redox proteins, which can be heterogeneous.
Thus, cellular redox reactions and the reactivity of KRAS G12C further complicate specific targeting of this oncogenic mutant in vivo. Despite the impressive efficacy of modification and ability to lock RAS into an inactive conformation in vitro, the compounds reported so far by Shokat and Westover and Gray show limited KRAS G12C-specific efficacy in cells. Another challenge moving forward lies in the therapeutic window or the fraction of oncogenic RAS that needs to be modified to reduce KRAS G12C tumorigenicity. As mentioned above, the redox-sensitivity of KRAS G12C may limit reaction of the KRAS G12C thiol to covalent modification by inhibitors due to direct competition with cellular oxidants. In-cell studies may be insufficient to test competition of cellular oxidants with cysteine-direct inhibitors, as the redox environment in cells and intact tissue is quite distinct. There is also a potential for off-target modification inherent in this approach, as a number of GTPases contain cysteines in or near the nucleotide-binding pocket. The lack of GTPase specificity observed for the cysteine-directed nucleotide analogs is a particular concern, given the large number of GTPases and biomolecules that bind guanine nucleotides. Further optimization may be needed to increase potency and specificity for effective modification in vivo. Regardless of these limitations, these cysteine-directed approaches represent the first small molecules to directly target an oncogenic RAS mutant, and these early attempts have significant room for improvement.
Importantly, these new strategies for directly and specifically inhibiting an oncogenic RAS mutant may pave the way for targeting other RAS mutants. For example, another prominent RAS mutant is KRAS G12D, which accounts for ~34% of all KRAS mutations. The G12D mutant, like G12C, lies within the phosphoryl-binding loop. Chemical approaches that facilitate tethering the reactive aspartate side chain within the nucleotide-binding site could be employed for inhibitor development. Further, the KRAS G12D mutation is also prevalent in pancreatic (50%) and colorectal cancers (34%), which are among the most deadly cancers; hence, this is a key oncogenic RAS mutant to target. However, RAS-driven cancers may possess distinct properties that depend on both the specific mutation as well as the tissue type, requiring the development of tailored inhibitors for each respective case. Further complicating this issue is that most tumors contain other mutations that contribute to tissue-specific tumorigenic properties and consequently affect the response to inhibitor therapies. Given the past problems inhibiting KRAS with farnesyl transferase inhibitors14,15, careful consideration of how RAS directed inhibitors work in vivo will be required in the development of new therapeutic approaches.
Regardless of the past difficulties in directly targeting oncogenic RAS, these cysteine-directed targeting approaches show promise and highlight the use of new and innovative approaches to specifically target oncogenic RAS. Generation of highly selective irreversible inhibitors may mitigate concerns with nucleotide cycling, compound specificity and competition with cellular oxidants. Finally, the mutant-directed approaches outlined herein reveal that, despite previous reservations, RAS may indeed be druggable.
About the Author
Sharon Campbell received her Ph.D. in Chemistry from the Yale University and conducted post-doctoral work with Alfred Redfield at Brandeis University. Following her postdoctoral work, she became a research scientist in central research and development (CR&D) division at DuPont in Wilmington, Delaware. She began her independent career at the University of North Carolina. Dr. Campbell is currently a Professor in the Department of Biochemistry and Biophysics at the University of North Carolina. A graduate student in Dr. Campbell's lab, Minh Hyunh, made significant contributions to this work.