Covalent KRAS G12C inhibitors have changed the treatment landscape for NSCLC and CRC patients, but numerous RAS-mutant cancers lack approved targeted therapies. Here we describe RMC-8839, an oral RAS(ON) G13C-selective, covalent, tri-complex inhibitor that induced tumor regressions in selected KRAS G13C-mutant xenograft models. However, one-third of KRAS G13C-mutant human cancer cell lines in vitro showed incomplete RAS pathway suppression despite near-complete KRAS G13C engagement, suggesting a role for wild-type RAS(ON). We find that codon 13-mutant RAS differs from other KRAS mutations, exhibiting decreased stability, increased nucleotide exchange, and substantial intrinsic and GAP-stimulated GTP hydrolysis, which decreases oncogenicity. Furthermore, co-occurring RAS pathway mutations leading to increased wild-type RAS activation are enriched in codon 13 mutant tumors. Consistent with a role for wild-type RAS(ON) signaling, combination of RMC-8839 with a RAS(ON) multi-selective inhibitor resulted in deeper inhibition of KRAS G13C-mutant xenograft tumor growth than either inhibitor alone.

Oncogenic RAS mutations are among the most common in human cancers. To target the active, GTP-bound state of RAS(ON) directly, we employed an innovative tri-complex inhibitor (TCI) modality. Formation of a complex with an intracellular chaperone protein CypA, an inhibitor, and a target protein RAS blocks effector binding, inhibiting downstream RAS signaling and tumor cell proliferation. Herein, we describe the structure-guided SAR journey that led to the discovery of daraxonrasib (RMC-6236), a noncovalent, potent tri-complex inhibitor of multiple RAS mutant and wild-type (WT) variants. This orally bioavailable bRo5 macrocyclic molecule occupies a unique composite binding pocket comprising CypA and SWI/SWII regions of RAS(ON). To achieve broad-spectrum RAS isoform activity, we deployed an SAR campaign that focused on interactions with residues conserved between mutants and WT RAS isoforms. Concurrent optimization of potency and drug-like properties led to the discovery of daraxonrasib (RMC-6236), currently in clinical evaluation in RAS mutant advanced solid tumors (NCT05379985; NCT06040541; NCT06162221; NCT06445062; NCT06128551).

Supplementary Table S9 shows datasets used for PK/Efficacy and PK/PD modeling.

The discovery of elironrasib (RMC-6291) represents a significant breakthrough in targeting the previously deemed undruggable GTP-bound, active KRASG12C. To target the active state of RAS (RAS(ON)) directly, we have employed an innovative tri-complex inhibitor (TCI) modality involving formation of a complex with an inhibitor, the intracellular chaperone protein CypA, and the target protein KRASG12C in its GTP-bound form. The resulting tri-complex inhibits oncogenic signaling, inducing tumor regressions across various preclinical models of KRASG12C mutant human cancers. Here we report structure-guided medicinal chemistry efforts that led to the discovery of elironrasib, a potent, orally bioavailable, RAS(ON) G12C-selective, covalent, tri-complex inhibitor. The investigational agent elironrasib is currently undergoing phase 1 clinical trials (NCT05462717, NCT06128551, NCT06162221), with preliminary data indicating clinical activity in patients who had progressed on first-generation inactive state-selective KRASG12C inhibitors.

Broad-spectrum antitumor activity of RMC-6236 in preclinical models of RAS-addicted cancers. A–C, Tumor response waterfall plots of KRASG12X NSCLC (A), PDAC (B), and colorectal cancer (C) xenograft models upon RMC-6236 daily treatment at 25 mg/kg. 29 NSCLC, 22 PDAC, and 23 colorectal cancer xenograft models were included (n = 1–10 per model). Average % mean tumor volume change ± SEM from baseline at response calling date are shown. mRECIST criteria were used to call tumor response as indicated on the right-hand side of each waterfall plot. Oncoplots illustrating gene alterations and expression levels in critical genes linked to the clinicopathologic characteristics of the indicated models are shown below each waterfall. Color coding represents dark green for mutations and light green for the absence of mutations. The ◈ symbol denotes that mRNA expression of corresponding genes not expressed, defined as having a gene-expression value of ≤0.5 CPM. The top row specifically highlights the mutation codon at KRASG12. D–F, Kaplan–Meier analyses of time to tumor doubling on treatment in individual tumor-bearing animals from KRASG12X NSCLC (D), PDAC (E), and colorectal cancer (F) xenograft models upon daily treatment of vehicle control or RMC-6236 at 25 mg/kg for up to 90 days. 29 NSCLC models (n = 135 animals each in control and RMC-6236 treatment groups), 22 PDAC models (n = 95 animals in control, n = 83 in RMC-6236 treatment group), and 23 colorectal cancer models (n = 95 animals in control, n = 93 in RMC-6236 treatment group) were included. Time to event was determined by the time on treatment until tumor volume doubling from baseline on survival plots by Kaplan–Meier analysis. Log-rank test was used to compare vehicle control with treatment groups, Cox proportional hazards models were used to estimate hazard ratios: KRASG12X NSCLC (HR 0.035, 95% interval 0.020–0.061, P < 2 × 10−16), KRASG12X PDAC (HR 0.008, 95% interval 0.002–0.026, P < 2 × 10−16) and KRASG12X colorectal cancer (HR 0.072, 95% interval 0.043–0.120, P < 2 × 10−16). G, Tumor response waterfall plot and Kaplan–Meier analysis of KRASG12X GAC and OVCA xenograft models upon daily treatment of vehicle control or RMC-6236 at 25 mg/kg for up to 90 days. Four models of GAC and 4 models of OVCA tumors were included. Average % mean tumor volume change ± SEM from baseline at the response calling date were plotted. mRECIST criteria were used to call tumor response as indicated on right-hand side of the waterfall plot. Time to event was determined above. H, Bar plots of mean tumor volume % change ± SEM from baseline for xenograft models of NSCLC with KRASG12X and KRASOther mutations. Data for both vehicle control and RMC-6236 treatment groups of 35 KRASMUT NSCLC models (29 KRASG12X and 6 KRASOther models) are shown with each model represented by one symbol. The genotype of each model was represented by color and shapes: KRASG12X (green dot), KRASOther (purple; KRASG13X, square; KRASQ61H, triangle; KRASK117N, star). Mean tumor volume % change from baseline of the vehicle control groups and RMC-6236 treatment groups for KRASG12X models are 708.1% and −13.7% respectively; for KRASOther models are 1,070% and 257.5%, respectively. Vehicle control and RMC-6236 treatment groups were compared by paired t test, with P < 0.001 (***) for KRASG12X models and P < 0.01 (**) for KRASOther models. The dotted line represents mean baseline tumor volume.

Translating RMC-6236 activity in NSCLC. A, Efficacy of RMC-6236 on KrasG12C, KrasG12D, KrasG12V, KrasG12A, KrasG13D, or KrasQ61H-driven autochthonous lung tumors in immunocompetent mice. A pool of lentiviral cDNA vectors encoding each oncogenic Kras variant was delivered intratracheally to the lungs of each mouse, and 13 weeks after tumor growth, mice were treated with RMC-6236 at 20 mg/kg po qd for 3 weeks prior to analysis. 95% confidence intervals are shown. B, Efficacy of RMC-6236 and adagrasib in the LUN055 NSCLC PDX model with KRASG12C allele copy-number gain. Immunoblot Western analyses (left) of RAS and KRAS protein levels in NCI-H358 (KRASG12C/WT, NSCLC), LU99 (KRASG12C/WT, NSCLC), NCI-H2122 (KRASG12C/G12C, NSCLC), and LUN055 (KRASG12C/WT, NSCLC) xenograft tumors. Relative copy-number (middle) of KRASWT or KRASG12C in LUN055 xenograft tumors (n = 2) were determined by ddPCR and normalized to ACTB. LUN055 xenograft tumor-bearing mice were treated with vehicle or RMC-6236 at 25 mg/kg po qd or adagrasib at 100 mg/kg po qd for 24 to 28 days (n = 3 per group, right). Mean tumor volumes of each group were plotted over the course of treatment. Dotted line indicates the initial average tumor volume. Error bars, SEM. C, Efficacy of RMC-6236 in the intracranially implanted LU99-Luc (KRASG12C/WT, NSCLC) xenograft model (n = 8 per group). RMC-6236 was dosed at 25 mg/kg daily for 21 days. Images of bioluminescence in individual mice were shown. Bioluminescence of ROI in vehicle control and RMC-6236 groups were compared by two-way repeated-measures ANOVA at day 21 (**, P < 0.01). Results were shown as mean ± SEM. D, Antitumor activity of RMC-6236 and the combination with anti–PD-1 (clone RMP1-14, rat IgG2a) following repeated administration in BALB/c mice bearing the murine colon carcinoma eCT26 (KrasG12C/G12C) shown as individual tumor growth curves (n = 10 per group). Graphs indicate the number of complete regressions per injected mice. RMC-6236 and anti–PD-1 treatment started on day 17 after implantation. RMC-6236 treatment was stopped at day 31 after implantation and anti–PD-1 at day 35 after implantation. E, Antitumor activity of RMC-6236 following repeated administration in NSG mice bearing the murine colon carcinoma eCT26 (KrasG12C/G12C) shown as individual tumor growth curves (n = 10 per group). Graphs indicate the number of complete regressions per injected mice. RMC-6236 treatment started on day 16 after implantation. F, Immune cell composition (CD8+ and CD4+ T cells, Ly6C+ and Ly6G+ myeloid-derived suppressor cells and M2 macrophages) in murine colon carcinoma eCT26 syngeneic tumors (KrasG12C/G12C) represented as percentage of CD45+ cells and expression of cell-surface markers on viable, CD45− large cells (assessed as tumor cells) 24 hours post 4 days of treatment with vehicle or RMC-6236 at 25 mg/kg po qd n = 3 biological replicates/group represented as mean; *, P < 0.05; **, P < 0.01; ns, nonsignificant by two-sided Student t test.

Supplementary Table S6 shows PK/PD data used to derive PK/PD relationship curves in Fig. 5 and Fig. S5.

Molecular glues are small molecules that nucleate novel or stabilize natural, protein-protein interactions resulting in a ternary complex. Their success in targeting difficult to drug proteins of interest has led to ever-increasing interest in their use as therapeutics and research tools. While molecular glues and their targets vary in structure, inspection of diverse ternary complexes reveals commonalities. Whether of high or low molecular weight, molecular glues are often rigid and form direct hydrophobic interactions with their target protein. There is growing evidence that these hotspots can accommodate multiple ternary complex binding modes and enable targeting of traditionally undruggable targets. Advances in screening from the molecular glue degrader literature and insights in structure-based drug design, especially from the non-degrading tri-complex work, are likely intersectional.

Supplementary Table S5 shows time to tumor doubling of individual animals in RMC-6236 mouse clinical trial.

Supplementary Figure 1 shows RMC-6236 crystal structure in tri-complex, as well as biophysical and cellular potencies of RMC-6236 by genotype. Supplementary Figure 2 shows RMC-6236 demonstrates dose-dependent anti-tumor activities at tolerable doses; and pharmacodynamic effects on RAS signaling in NCI-H441 xenograft tumors as assessed by IHC, and in relatively refractory KP-4 and NCI-H2122 xenograft tumors as assessed by human DUSP6 mRNA expression in vivo. Supplementary Figure 3 shows genotype dependent response of RMC-6236 across NSCLC, PDAC and CRC; and potential modifiers to the durability of response of KRASG12C NSCLC models upon RMC-6236 treatment assessed by Kaplan-Meier analyses. Supplementary Figure 4 shows Efficacy of RMC-6236 on KrasG12C–driven autochthonous lung tumors harboring cis second-site mutations within KrasG12C (KrasG12C,H95D or KrasG12C,Y96C) and eCT26 (KrasG12D/G12D) syngeneic model in immunocompetent mice; anti-tumor immunity of RMC-6236; and in intracranially implanted NCI-H1373-Luc xenograft model on nude mice. Supplementary Figure 5 shows effects of RMC-6236 mediated pharmacological modulation in KP-4 xenograft tumors and normal colon tissues isolated from xenograft tumor bearing mice. Supplementary Figure 6 shows a graphical representation of the combined mouse PK-Efficacy and PK/PD model.

RMC-6236 inhibits RAS signaling and tumor growth and drives tumor regressions in vivo.A, Blood and tumor PK profiles of RMC-6236 in Capan-2 (KRASG12V/WT, PDAC) xenograft tumor-bearing BALB/c nude mice. Tumor-bearing mice were treated with a single dose of vehicle or RMC-6236 at 3, 10, or 25 mg/kg. Blood and tumors were harvested at indicated time points (n = 3/time point/dose). PK profiles are shown as RMC-6236 concentration in tumors (green lines) and blood (red lines) over time. Shades of green or red represent PK profiles at three tested doses. The dashed lines represent EC50 and EC90 potency of RMC-6236 in inhibiting DUSP6 mRNA expression in Capan-2 tumors derived from the PK/PD relationship curve in Fig. 5A. Values are plotted as mean ± SEM. B, PD of RMC-6236 in Capan-2 (KRASG12V/WT, PDAC) xenograft tumors, shown as the relative change in DUSP6 mRNA expression. Tumor-bearing mice were treated with a single dose (solid lines) of vehicle, RMC-6236 at 3, 10, or 25 mg/kg, or 7 consecutive daily doses of RMC-6236 at 25 mg/kg (dashed lines). Shades of green represent three tested doses. Solid lines represent a single dose while the dashed line represents repeat dosing. Values are plotted as mean ± SEM. C, Histopathology analysis of Capan-2 xenograft tumors treated with a single dose of vehicle control, or RMC-6236 at 3, 10, or 25 mg/kg or 7 consecutive daily doses of RMC-6236 at 25 mg/kg and collected at indicated time points (n = 2–3/time point/dose). pERK staining in tumor areas was quantified and compared with vehicle using one-way ANOVA followed by Dunnett multiple comparison test (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Representative images are shown at 200× magnification from samples closest to the mean of the group. Scale bar, 50 µm. D–G, Dose-dependent antitumor activity of RMC-6236 in subcutaneous xenograft models of (D) Capan-2 (KRASG12V/WT, PDAC; n = 8 per group), po qd, per os quaqua (once a day) (E) NCI-H441 (KRASG12V/WT, NSCLC; n = 10 per group), (F) HPAC (KRASG12D/WT, PDAC; n = 10 per group), and (G) NCI-H358 (KRASG12C/WT, NSCLC; n = 8–10 per group). Tumor-bearing mice were treated with vehicle or RMC-6236 at indicated doses for 27–28 days, and mean tumor volumes of each group were plotted over the course of treatment. Vehicle control and RMC-6236 groups were compared by two-way repeated-measures ANOVA on the last measurement day of the vehicle group (***, P < 0.001). The dotted line indicates the initial average tumor volume. Error bars, SEM. # indicates 1 animal terminated upon reaching a tumor burden endpoint.

Supplementary Table S3 shows PK and/or PD data of RMC-6236 in subcutaneous xenograft tumors, intracranially implanted xenograft tumors, normal brain from naüve mice, and ear skin and colon from xenograft tumor bearing mice.

Supplementary Table S4 shows % mean tumor volume change, % mean body weight change of RMC-6236 treatment group on mRECIST response calling date; and PPIA mRNA expression levels of each xenograft model in RMC-6236 mouse clinical trial.

Abstract RAS-driven cancers comprise up to 30% of human cancers. RMC-6236 is a RAS(ON) multi-selective noncovalent inhibitor of the active, GTP-bound state of both mutant and wild-type variants of canonical RAS isoforms with broad therapeutic potential for the aforementioned unmet medical need. RMC-6236 exhibited potent anticancer activity across RAS-addicted cell lines, particularly those harboring mutations at codon 12 of KRAS. Notably, oral administration of RMC-6236 was tolerated in vivo and drove profound tumor regressions across multiple tumor types in a mouse clinical trial with KRASG12X xenograft models. Translational PK/efficacy and PK/PD modeling predicted that daily doses of 100 mg and 300 mg would achieve tumor control and objective responses, respectively, in patients with RAS-driven tumors. Consistent with this, we describe here objective responses in two patients (at 300 mg daily) with advanced KRASG12X lung and pancreatic adenocarcinoma, respectively, demonstrating the initial activity of RMC-6236 in an ongoing phase I/Ib clinical trial (NCT05379985). Significance: The discovery of RMC-6236 enables the first-ever therapeutic evaluation of targeted and concurrent inhibition of canonical mutant and wild-type RAS-GTP in RAS-driven cancers. We demonstrate that broad-spectrum RAS-GTP inhibition is tolerable at exposures that induce profound tumor regressions in preclinical models of, and in patients with, such tumors. This article is featured in Selected Articles from This Issue, p. 897

Effects of RMC-6236 mediated pharmacologic modulation of RAS pathway signaling in tumor-bearing mice. A, PK/PD relationship between RMC-6236 concentration and inhibition of DUSP6 expression in Capan-2 (EC50 = 90 nmol/L and EC90 = 809 nmol/L), NCI-H441 (EC50 = 117 nmol/L and EC90 = 1,121 nmol/L), and HPAC (EC50 = 135 nmol/L and EC90 = 925 nmol/L) xenograft tumors. Subcutaneous xenograft tumors were treated with vehicle or RMC-6236 ranging from 0.3 to 100 mg/kg (Capan-2 and H441) or to 50 mg/kg (HPAC). B, PK/PD relationship between RMC-6236 concentration and inhibition of Dusp6 expression in ear skin (EC50 = 1,164 nmol/L and EC90 = 10,279 nmol/L) isolated from tumor-bearing BALB/c nude mice treated with vehicle or RMC-6236 ranging from 3 mg/kg to 100 mg/kg. A and B, Tumors and ear skin from tumor-bearing BALB/c nude mice were harvested at indicated time points (n = 3/timepoint/dose). A 3-parameter sigmoidal exposure–response model was fitted to the data to derive EC50 and EC90 values. Time points are represented by colors and doses are represented by symbol shapes. C–G, Histopathology of tumors and ear skin from the Capan-2 xenograft model collected at indicated time points post a single dose of vehicle control, RMC-6236 at 25 mg/kg or 7 consecutive daily doses of RMC-6236 at 25 mg/kg (n = 3–6/time point/dose). Staining of indicated markers in the tumor area or ear skin was quantified and compared with vehicle using one-way ANOVA followed by the Dunnett multiple comparison test (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Representative images are shown at 200× magnification from samples closest to the mean of the respective groups. Scale bar, 50 µm.

PK/PD/Efficacy modeling to predict clinically active dose range. A, Comparison of observed and predicted tumor growth data at multiple dose levels of RMC-6236 in the NCI-H441 xenograft tumor model. Tumor growth was predicted using the Simeoni tumor growth model. The dotted line indicates the initial average tumor volume. B, Comparison of observed and predicted tumor growth data at multiple dose levels of RMC-6236 in the Capan-2 xenograft tumor model. Tumor growth was predicted using the Simeoni tumor growth model. The dotted line indicates the initial average tumor volume. C, Comparison of observed vs. simulated PK and PD data at multiple dose levels of RMC-6236 in mice bearing NCI-H441 xenograft tumors. Single-dose data from all dose levels are presented from 0 to 24 hours, whereas repeat-dose data from 25 and 40 mg/kg repeat daily dosing is presented from 216 to 240 hours. Simulated blood and tumor PK data are indicated by the solid and dashed green and blue lines, respectively. Simulated PD data are indicated by the solid purple lines. Observed data are indicated by dots (blood PK and PD) or squares (tumor PK). The dotted line indicates the 10% expression level of DUSP6 mRNA as normalized to the vehicle control group. D, Predicted profiles of human whole blood and tumor PK as well as tumor PD at clinical dose levels at steady state. Blood PK and tumor PK are indicated by the solid and dashed green lines whereas tumor PD is indicated by the solid purple lines. Repeat dose data are presented from 336 to 360 hours after two weeks of simulated daily dosing. The dotted line indicates a 10% expression level of DUSP6 mRNA as normalized to the vehicle control group.

Supplementary Table S7 showes final processing and refinement statistics used for RMC-6236 Crystallography data collection and refinement.

Mutant RAS proteins are among the most prevalent drivers of human cancer, and the glycine to aspartic acid mutation at codon 12 (G12D) is the most common variant. Mutation-selective covalent inhibitors spare RAS in healthy tissue and enable extended pharmacodynamic effect, but covalent targeting of RAS G12D is hindered by low nucleophilicity and high proteomic abundance of carboxylic acids. We overcame these challenges with compounds that bind cyclophilin A (CYPA) to create a neomorphic protein-protein interface between CYPA and active RAS that enables selective, enzyme-like rate enhancement of the covalent reaction between D12 and electrophilic warheads with exceptionally low intrinsic reactivity. This approach yielded orally bioavailable compounds with marked antitumor activity in multiple preclinical models of KRAS G12D cancers, including the investigational agent zoldonrasib (RMC-9805) currently undergoing clinical evaluation (NCT06040541).

Supplementary Table S2 shows cell panel screening data with a panel of 78 cancer cell lines harboring mutant and wild-type RAS selected for screening at Crown Bioscience.

Supplementary Table S8 shows all parameter estimates for PK/Efficacy and PK/PD modeling.