Direct Inhibition of RAS Reveals the Features of Oncogenic Signaling Driven by RAS G12 and Q61 Mutations

Abstract

RAS genes are frequently mutated in cancer, often at codons 12 and 61. With the recent introduction of RAS inhibitors, we can now directly investigate the effects of specific RAS mutations in cancer cells. In this study, we demonstrate that in tumors with RASG12X mutations, mutant RAS can be activated by receptor tyrosine kinases (RTK), and PI3K activation is dependent on mutant RAS. Conversely, RASQ61X mutations activate the MAPK cascade independently of RTKs, and inhibition of RASQ61X impairs MAPK pathway activation but leaves the PI3K pathway unaffected. Our characterization of these distinct features of G12X and Q61X mutations suggests that co-inhibition of RAS and RTKs selectively inhibits the growth of RASG12X-mutant tumors, both in vitro and in vivo, regardless of the RAS isoform and tumor type. Additionally, our findings offer a mechanistic explanation for the increased frequency of RASQ61X mutations as a secondary resistance mechanism against EGFR inhibition in colorectal cancer.

Significance: RAS inhibition in multiple tumor types reveals the difference between G12 mutants and Q61 mutants in their cooperation with upstream regulators and downstream effectors to promote oncogenic signaling. Our findings provide the rationale for combinatorial approaches and contribute to explaining the nonuniform distribution of RAS mutations, de novo and at resistance.

Figure 1.. RAS inhibition has a different impact on PI3K signaling in RAS^G12X- and RAS^Q61X-mutant cancer cells

(A) Comparison of the Areas Under the Curve (AUC) for a set of human cancer cell lines exposed to incremental doses of RMC-7977. Each point represents the area under an average dose-response curve of three biological replicates, each consisting of three technical replicates. (B) Western blot analysis of MEK and ERK phosphorylation dynamics following treatment with low-dose RMC-7977 in a subset of cell lines. Vinculin is used as the loading control. (C) Western blot analysis of early ERK, Akt (S473), and PRAS40 phosphorylation dynamics following treatment with RMC-7977 in a set of lines harboring KRAS or NRAS mutations at position G12. Vinculin is used as the loading control. (D) Western blot analysis of early ERK, Akt (S473), and PRAS40 phosphorylation dynamics following treatment with RMC-7977 in a set of lines harboring KRAS or NRAS mutations at position Q61. Vinculin is used as the loading control. (E) Western blot analysis of early Akt (T308) phosphorylation dynamics following treatment with RMC-7977 in a set of lines harboring KRAS or NRAS mutations at position G12 or Q61. The total Akt bands are the same as the corresponding total Akt bands in panels C and D. (F) Schematic of the procedure to derive liver cancer cell lines using the HTVI-GEMM system. (G) Western blot analysis of early signaling dynamics following treatment with RMC-7977 in two KRAS^G12D and two KRAS^Q61R lines. Each line originates from a different mouse. Vinculin is used as the loading control. (H) Schematic of RAS signaling based on the mutated codon. Created with BioRender.com

Figure 2.. The impact of RMC-7977 on PI3K correlates with sensitivity to RAS inhibition.

(A) Dose-response Western blot analysis of AsPC-1 and HPAF-II exposed to three different concentrations of RMC-7977 (1nM, 10nM, 100nM) for 1h, 8h, and 24h. Vinculin is used as a loading control. (B) Dose-response assay of AsPC-1 and AsPC-1 overexpressing myristoylated p110α (myr-p110α), treated with RMC-7977. The Western blot validates the expression of HA-tagged myr-p110α and shows its impact on PI3K signaling. The curves represent the average of three biological replicates, each consisting of two technical replicates. (C) Dose-response analysis of the HTVI-KRAS lines treated with RMC-7977. Statistical significance was calculated using the Wilcoxon matched-pairs signed rank test. Each curve represents the average of two different cell lines with the same genotype (HTVI-KRAS^G12D #1 and #2, HTVI-KRAS^Q61R #1 and #2), each obtained by averaging three technical replicates. (D) Time-course Western blot analysis of AsPC-1 treated with RMC-7977 or the combination of RMC-7977 and UCL-TRO-1938. (E) Time-course Western blot analysis of LIM1215, LIM1215 PIK3CA^E545K, and LIM1215 PIK3CA^H1047R treated with RMC-7977 (F) Dose-response pharmacological assays of LIM1215, LIM1215 PIK3CA^E545K, and LIM1215 PIK3CA^H1047R treated with RMC-7977 or trametinib. (G) Drug combination validation experiments using RMC-7977 and alpelisib/AZD-8186: heatmaps show viability through color-coding as percentage of cell viability normalized on untreated controls (left) and the Loewe score of the drug combination (right). Each point in the heatmap represents the average of three biological replicates, each consisting of two technical replicates. (H) Western blot analysis of the MAPK and PI3K pathway responses to RMC-7977 50nM (RAS inhibitor, R), trametinib 50nM (MEK inhibitor, M), alpelisib 1μM / AZD-8186 0.25μM (PI3K inhibitors, P), and the respective combinations RP (RMC-7977 + alpelisib/AZD-8186) and MP (trametinib + alpelisib/AZD-8186) in AsPC-1 and HS766.T. Vinculin is used as the loading control.

Figure 3.. In RAS^Q61X-mutant cancer cell lines, mutant RAS is decoupled from both PI3K and receptor tyrosine kinases.

(A) Profile of Akt (T308) phosphorylation as a function of time after stimulation with EGF (20ng/ml), in cells previously serum-starved overnight. RMC-7977 100nM (red curve) or DMSO (black curve) were added to the cells 2h before the EGF stimulation. The curves were derived from densitometric quantification of the three Western blot replicates in Fig. S5A. The intensities of phospho-Akt (T308) bands were normalized to the loading control and expressed as a ratio to the intensity in the non-stimulated, DMSO-treated, state. (B) Western blot of two different HTVI-KRAS^G12D and two different HTVI-KRAS^Q61R lines, serum-starved overnight, then pretreated with DMSO or RMC-7977 (100nM) for 2h and finally stimulated with murine EGF (20ng/ml). Vinculin is used as the loading control. (C) Kinetic profile of MEK, ERK, Akt (T308), and Akt (S473) phosphorylation after treatment with RMC-7977 (50nM) in AsPC-1 or HS766.T grown in regular 10% FBS medium or 10%FBS supplemented with EGF (100ng/ml). The curves are derived from densitometric quantification of the three Western blot replicates in Fig. S6B. (D) Western blot analysis of AsPC-1, COLO678, Sk-Mel-176, and HS766.T pretreated with DMSO or RMC-7977 (50nM) for 24h and then stimulated with PBS or different growth factors (at a concentration of 100ng/ml) for 10 minutes. β-actin is used as the loading control.

Figure 4.. Inhibitors of GTP exchange are synergistic with RMC-7977 in RAS^G12X-mutant cancer cell lines.

(A) (Left) one representative pulldown of GTP-bound RAS in AsPC-1 and HS766.T, serum-starved overnight, then pretreated with DMSO or RMC-7977 (100nM), and finally stimulated with EGF (20ng/ml). The antibodies against RAS^G12D and RAS^Q61X were used to detect GTP-bound RAS and KRAS was used as a loading control. (Middle) The graph shows the ratio of RAS-GTP over RAS normalized to the respective non-stimulated controls, determined from densitometric quantification of the three active RAS pulldown replicates in Fig. S7. (Right) Profile of MEK phosphorylation in AsPC-1 and HS766.T as a function of time after EGF (20ng/ml) in the presence of DMSO or RMC-7977 (100nM), derived from densitometric quantification of the three Western blot replicates in Fig. S5A. The intensities of phospho-MEK bands were normalized to the loading control and expressed as a ratio to the intensity in the non-stimulated, DMSO-treated, state. (B) Drug combination validation experiments using RMC-7977 and cetuximab: heatmaps show viability through color-coding as percentage of cell viability normalized on untreated controls (left) and the Loewe score of the drug combination (right). Each point in the heatmap represents the average of three biological replicates, each consisting of two technical replicates. (C) Values of the differences in area under the curve (delta AUC) for the dose-response curves of RMC-7977 with and without cetuximab (25μg/ml), shown in Fig. S9. (D) Dose-response curves of RMC-7977 with or without BI-3406 (2μM). (E) Pulldown of GTP-bound RAS in four different lines treated for 24h with DMSO, RMC-7977 (10nM), cetuximab (25μg/ml) or RMC-7977+cetuximab. Detection of GTP-bound RAS was achieved with mutant-specific antibodies (except for Sk-Mel-173, as Q61K mutations are not detected by the RAS^Q61X-specific antibody), while KRAS or NRAS were used as loading controls. The Western blots also include important signaling markers and vinculin as an additional loading control.

Figure 5.. Efficacy of RMC-7977 + cetuximab in vivo

(A) Waterfall plot of percent tumor growth relative to baseline for a patient-derived xenograft (PDX) of PDAC harboring a KRAS^G12D mutation treated with RMC-7977 alone and in combination with cetuximab. Each bar represents an individual mouse. (B) Waterfall plot of percent tumor growth relative to baseline for a patient-derived xenograft (PDX) of MEL harboring an NRAS^Q61K mutation treated with RMC-7977 alone and in combination with cetuximab. Each bar represents an individual mouse. (C) Addition of cetuximab to a KRAS^G12D PDAC PDX that showed evidence of progression on RMC-7977 was able to induce tumor regression. (D) Drug combination validation experiments using RMC-7977 and cetuximab in four patient-derived organoids (PDOs): the surface plots show the Loewe score of the drug combination. Each point represents the average of two biological replicates, each consisting of three technical replicates.

Figure 6.. Emergence of RAS^Q61X mutations in colorectal cancer treated with anti-EGFR therapies.

(A) Distribution of RAS^G12X, RAS^Q61X, RAS^G13X, and RAS^A146X mutations in patients with CRC, either de novo or acquired as a mechanism of resistance to anti-EGFR treatment. Patient data from MD Anderson Cancer Center institutional datasets. (B) Distribution of CRC patients with RAS^G12X, RAS^Q61X, or co-existence of RAS^G12X and RAS^Q61X mutations as a mechanism of resistance to anti-EGFR treatment. The graph also includes patients who developed mutations in other genes (other mts.). Data from the MSKCC cohort. (C) Drug combination validation experiments using RMC-7977 and cetuximab: heatmaps show viability through color-coding as percentage of cell viability normalized on untreated controls (left) and the Loewe score of the drug combination (right). Each point in the heatmap represents the average of three biological replicates, each consisting of two technical replicates. (D) Pulldown of GTP-bound RAS in the three isogenic OXCO2 lines treated for 24h with DMSO, RMC-7977 (10nM), cetuximab (25μg/ml), or RMC-7977+cetuximab. Detection of GTP-bound RAS was achieved with mutant-specific antibodies whenever needed. KRAS, NRAS, RAS^G12D, or RAS^Q61R were used as loading controls. (E) Schematic representation of how the acquisition of RAS^G12X or RAS^Q61X mutations as a mechanism of secondary resistance to cetuximab (orange antibody) affects signaling in CRC. Created with BioRender.com. (F) Proposed model to explain the bias towards RAS^G12X mutations in adenocarcinomas and RAS^Q61X mutations in melanoma. See Discussion for details. The pie charts are from (11). Created with BioRender.com.