The SOS1 Inhibitor MRTX0902 Blocks KRAS Activation and Demonstrates Antitumor Activity in Cancers Dependent on KRAS Nucleotide Loading

Abstract

KRAS is the most frequently mutated oncogene in human cancer and facilitates uncontrolled growth through hyperactivation of the receptor tyrosine kinase (RTK)/mitogen-activated protein kinase (MAPK) pathway. The Son of Sevenless homolog 1 (SOS1) protein functions as a guanine nucleotide exchange factor (GEF) for the RAS subfamily of small GTPases and represents a druggable target in the pathway. Using a structure-based drug discovery approach, MRTX0902 was identified as a selective and potent SOS1 inhibitor that disrupts the KRAS:SOS1 protein–protein interaction to prevent SOS1-mediated nucleotide exchange on KRAS and translates into an anti-proliferative effect in cancer cell lines with genetic alterations of the KRAS–MAPK pathway. MRTX0902 augmented the antitumor activity of the KRAS G12C inhibitor adagrasib when dosed in combination in eight out of 12 KRAS G12C–mutant human non–small cell lung cancer and colorectal cancer xenograft models. Pharmacogenomic profiling in preclinical models identified cell cycle genes and the SOS2 homolog as genetic co-dependencies and implicated tumor suppressor genes (NF1 and PTEN) in resistance following combination treatment. Lastly, combined vertical inhibition of RTK/MAPK pathway signaling by MRTX0902 with inhibitors of EGFR or RAF/MEK led to greater downregulation of pathway signaling and improved antitumor responses in KRAS–MAPK pathway–mutant models. These studies demonstrate the potential clinical application of dual inhibition of SOS1 and KRAS G12C and additional SOS1 combination strategies that will aide in the understanding of SOS1 and RTK/MAPK biology in targeted cancer therapy.

Introduction

Dysregulation of the RAS–RAF–MEK–ERK MAP kinase (RAS/MAPK) pathway is a critical factor in the pathogenesis of many human cancers. The rat sarcoma viral oncogene (RAS) superfamily of small guanosine triphosphatase proteins, which includes KRAS, HRAS, and NRAS, is mutated in nearly 19% of all human malignancies and is associated with approximately 3.4 million deaths per year worldwide (1). The integration of signal transduction from receptor tyrosine kinases (RTKs), such as the EGFR, to RAS is mediated by a number of key signaling molecules including SH2-containing protein tyrosine phosphatase-2 (SHP2), growth factor receptor-bound protein 2 (GRB2), and guanine nucleotide exchange factors (GEF) such as the Son of Sevenless (SOS) protein family. Following canonical transmembrane receptor activation, SHP2 recruits GRB2 to the membrane, thereby promoting its binding to both RTK and SOS proteins via its SH2 and SH3 domains, respectively (2, 3). Subsequently, SOS proteins directly engage with the switch one region of RAS family proteins which opens the nucleotide-binding site and facilitates the exchange of guanosine diphosphate (GDP) to guanosine triphosphate (GTP), ultimately rendering RAS into an active conformation (4). Although two homologs of SOS (SOS1 and SOS2) are ubiquitously expressed in mammalian cells, various studies have demonstrated a dominant role for SOS1 over SOS2 in the regulation of RAS-dependent embryonic development, cellular proliferation, and overall survival in adult mice (5, 6). The guanosine triphosphatase–activating protein (GAP) neurofibromin, encoded by the NF1 tumor suppressor gene, is a key GAP family member that opposes SOS protein function by promoting the conversion of GTP-RAS to GDP-RAS, thus negatively regulating RAS activation (7, 8). Genetic alterations in NF1 are prevalent in nearly 7% of all cancers with the greatest incidence in lung and colon adenocarcinoma, breast invasive ductal carcinoma, conventional glioblastoma multiforme, and cutaneous melanoma and at a lower frequency in the inherited neurofibromatosis type 1 disorder (7, 9). Activating SOS1 mutations and loss-of-function NF1 mutations are specifically found in approximately 1% and 10% of non–small cell lung cancer (NSCLC) cases, respectively (9, 10). A greater incidence of SOS1 mutations (up to 10%) is observed in RASopathies including Noonan’s syndrome and gingival fibromatosis (11–16).

SOS1 functions primarily as an activator of KRAS and is also a target of adaptive feedback signaling following KRAS-mediated activation of the MAPK pathway (17, 18). KRAS is the most frequently mutated oncogene amongst the RAS family of proteins and comprises approximately 90% of pancreatic ductal adenocarcinomas, 50% of colorectal carcinomas (CRC), and 30% of lung adenocarcinomas (LUAD; refs. 19–21). Proteogenomic and functional genomic studies have uncovered additional links between KRAS-mutant signaling and SOS1 activity: the former showed significant upregulation of SOS1 phosphorylation on serine residue 1161 in KRAS-mutant LUAD versus normal adjacent tissue, leading to membrane localization for pathway (RAS) activation (22), and the latter revealed that SOS1 depletion decreases the survival of KRAS–MAPK pathway-dependent cancer cells (13, 23). Consistent with in vitro observations, SOS1 inhibitor treatment of preclinical human xenograft models harboring mutations in KRAS resulted in MAPK pathway modulation and tumor growth inhibition (TGI; ref. 24). Thus, SOS1 represents a significant therapeutic node that maintains KRAS activation state, pathway equilibrium, and oncogenic signaling. Targeting the SOS1:KRAS protein–protein interaction (PPI) is anticipated to suppress KRAS-MAPK signaling.

Single-agent or combination approaches targeting the RAS/MAPK pathway have been shown to specifically decrease the survival of tumor cells harboring respective pathway mutations (24). Clinically, inhibitors that target KRAS G12C-mutant cancer have shown robust antitumor efficacy with tolerable safety profiles (25–28). KRAS G12C is the most frequent mutation variant with a prevalence of approximately 13% to 14% in LUAD (1, 29, 30). Notably, LUMAKRAS (sotorasib) and KRAZATI (adagrasib) have been granted accelerated approval by the FDA for the treatment of patients with KRAS G12C–mutant locally advanced or metastatic NSCLC who have received at least one prior systemic therapy. Although KRAS G12C inhibitors represent a meaningful new therapeutic option, additional therapies that enhance the depth and duration of response are needed. Clinical mechanisms of acquired resistance associated with KRAS G12C inhibitors include novel secondary KRAS mutations and amplifications that interfere with covalent drug binding within the switch II pocket, genomic amplification of KRAS, mutations in RTKs and RAS effector signaling pathway members, and oncogenic fusions (31–33). In human KRAS G12C–mutant cancer models, administration of the combination of a SOS1 inhibitor with a KRAS G12C inhibitor results in enhanced antitumor efficacy compared with the respective single agent (24). Of relevance, inhibition of the upstream target protein SHP2 (encoded by the PTPN11 gene) in combination with KRAS G12C inhibitors in KRAS-mutant cancers has demonstrated antitumor activity in preclinical and clinical studies (34–37). However, SHP2 has been shown to also function in both JAK/STAT pathway regulation and immune checkpoint signaling nodes and is critical for the viability of both normal and cancer cells (38). Although clinical activity has been reported for KRAS G12C and SHP2 inhibitors in combination, tolerability, and optimal dose schedule remain key challenges, and it is not yet clear whether the combinations are more effective than monotherapy (39, 40). The precise role of SOS1 in KRAS regulation may provide an alternative strategy to co-target the KRAS activation state with an anticipated wide safety margin and continuous administration schedule.

We have recently reported the design and discovery of MRTX0902, a selective, orally bioavailable inhibitor of the SOS1:KRAS PPI (41, 42). In the current studies, the activity, mechanism of action, pharmacodynamics, and antitumor activity of MRTX0902 were evaluated in vitro and in vivo. These data demonstrate that MRTX0902 is effective as a monotherapy in select RTK/MAPK pathway-mutant models and that combination treatment with KRAS G12C, EGFR, and RAF/MEK inhibitors results in deeper and more durable antitumor responses.

Materials and Methods

Chemical reagents and cell lines

MRTX0902 [optically pure enantiomer (42, 43)] was synthesized at Wuxi AppTec and adagrasib at Array Biopharma, Inc., or WuXi AppTec. Osimertinib was procured from Selleck Chemicals (S7297, Lot # 07). Avutometinib (VS-6766, Lot 002) was provided by Verastem Oncology. Drugs (MRTX0902, osimertinib, and avutometinib) in powder form were stored at room temperature and protected from light. For in vitro studies, drugs were formulated in 100% DMSO to make 10 mmol/L stock solutions and aliquoted for long-term storage at −20 °C.

Minimum Essential Medium Alpha media (Cat. #32561), DMEM (#10566-016), RPMI-1640 media (Cat. #11875-093), Dulbecco’s PBS (#14190-136), Trypsin-EDTA (Cat. #25200-056), sodium pyruvate (Cat. #11360-070), and Trypan blue solution, 0.4% (Cat. #15250061) were obtained from Gibco/Thermo Fisher Scientific. FBS was obtained from Corning (Cat. #35-011-CV). Antibiotic–antimycotic solution was obtained from Caisson Labs (Cat. #ABL02).

The following cell lines were obtained from the ATCC between April 2014 and May 2021 and grown in recommended media: MIA PaCa-2 (Cat. #CRL-1420), NCI-H358 (Cat. #CRL-5807), SW837 (Cat. #CCL-235), NCI-H2122 (Cat. #CRL-5985), SW1573 (Cat. #CRL-2170), A549 (Cat. #CRM-CCL-185), LN229 (Cat. #CRL-2611), NCI-H1435 (Cat. #CRL-5870), NCI-H211 (Cat. #CRL-5824), NCI-H1975 (Cat. #CRL-5908), DLD1 (Cat. #CCL-221), NCI-H1838 (Cat. #CRL-5899), NCI-H508 (Cat. #CCL-253), NCI-H1666 (Cat. #CRL-5885), LS123 (Cat. #CCL-255), and AsPC-1 (Cat. #CRL-1682). Additional cell lines obtained between April 2014 and May 2021 and grown in standard growth conditions and evaluated in cellular assays include the following: OCI-AML5 (Deutsche Sammlung von Mikroorganismen und Zellkulturen, #ACC247), HT115 (Cat. #85061104; European Collection of Authenticated Cell Cultures), MKN1 (Cat. #0252; Japanese Collection of Research Bioresources Cell Bank), LU99 (Cat. #RCB1900; Riken), KYSE-410 (Cat. #94072023; Millipore Sigma), PC9 (Cat. #RCB4455; Riken), HCC1438 (Cat. #71438; Korean Cell Line Bank), HOP92 [DCTD Tumor Repository (National Institutes of Health)], SNU-C4 (Cat. #0000C4; Korean Cell Line Bank), RERF-LC-Ad1 (Cat. #JCRB1020; Japanese Collection of Research Bioresources Cell Bank), and MKN-74 (Cat. #ABC-TC0689; AcceGen Biotechnology). Human cancer cell lines were maintained at 37°C in a humidified incubator at 5% CO₂ and were periodically checked for Mycoplasma. Cell lines used for in vivo studies were confirmed pathogen and mycoplasma-free by IMPACT 1 assessment (IDEXX BioAnalytics) prior to implant. Cell lines were carried for no more than 12 to 15 cell passages in this work.

In vivo studies

All mouse studies have been approved by the Institutional Animal Care and Use Committee from the National Institutes of Health and were conducted in compliance with all applicable regulations and guidelines. Mice were maintained under pathogen-free conditions, and food and water were provided ad libitum. Six- to eight-week-old female Hsd:Athymic Nude-Foxn1nu mice (Envigo) were injected subcutaneously with tumor cells in 100 μL of PBS and Matrigel matrix (Corning #356237; Discovery Labware) in the right hind flank of each mouse with 5.0 × 10⁶ cells in 50:50 cells:Matrigel. Both human tumor cell line–derived xenograft (CDX) and patient-derived xenograft (PDX) models were utilized in these studies. Mouse health was monitored daily, and caliper measurements began when tumors were palpable. Tumor volume measurements were determined utilizing the formula 0.5 × L × W² in which L refers to the length and W refers to the width of each tumor. When tumors reached the desired average study start tumor volume (100–333 mm³ depending on model), mice were randomized into treatment groups and treated by oral gavage with either vehicle [0.5% methylcellulose (4,000 cps) + 0.2% Tween 80 in water], MRTX0902 in vehicle, adagrasib (10% Captisol in 50 mmol/L citrate buffer pH 5.0), osimertinib (5% DMSO: 45% PEG400: 50% H₂0), or avutometinib (5% DMSO, 10% hydroxypropyl-β-cyclodextrin in H₂0) at indicated doses and schedules. Formulations were completed weekly and dosing solutions were stored protected from light at 4 °C (Supplementary Materials and Methods). For efficacy studies, animals were orally administered drug or vehicle and monitored daily, tumors were measured two times per week, and body weights were measured two times per week (n = 5 per group). Study day on efficacy plots indicates the day after which drug treatment was initiated. For pharmacodynamic studies, mice were implanted with cells as previously described and randomized into treatment groups when tumors reached an average tumor volume of approximately 250 to 350 mm³ depending on the model. Mice were treated by oral gavage for 6 to 7 days with either vehicle or drug at indicated doses and tumors were collected for analysis. For studies conducted at Crown Biosciences China, 4- to 5-week-old female BALB/c nude mice were implanted with tumor fragments 2 to 3 mm in diameter into the right flank via trocar implant. Eight mice were dosed with either vehicle alone [0.5% methylcellulose (4,000 cps)/0.2% Tween80 in water], 50 mg/kg BID MRTX0902 (in vehicle formulation), or 100 mg/kg adagrasib (10% Captisol in 50 mmol/L citrate buffer pH 5.0) daily by oral gavage for 27 to 42 days (n = 8 per group). GraphPad Prism 8 was used to graph the data and complete a two-tailed Student t test statistical analysis between the vehicle-treated and drug-treated cohorts.

Data availability

The data generated in this study are available upon request from the corresponding author.

Results

MRTX0902 is a potent and selective inhibitor of SOS1 and RAS-dependent signal transduction and cell viability in vitro

A structure-based drug discovery approach was utilized to design SOS1 binders that disrupt the SOS1:KRAS PPI, which led to the discovery of the orally bioavailable small molecule development candidate MRTX0902 (Fig. 1A; ref. 42). MRTX0902 bound to SOS1 with an inhibitory constant (K_i) value of 2.1 nmol/L and disrupted the SOS1:KRAS complex with IC₅₀ values of 13.8 nmol/L (wild-type KRAS), 16.6 nmol/L (KRAS G12D), 24.1 nmol/L (KRAS G12V), and 30.7 nmol/L (KRAS G12C; Fig. 1B). Furthermore, MRTX0902 inhibited SOS1-mediated GTP exchange of KRAS in a Homogeneous Time Resolved Fluorescence (HTRF)–based biochemical assay with an IC₅₀ value of 15 nmol/L. No appreciable effect on SOS2-mediated GTP exchange of KRAS (IC₅₀ >10 µmol/L) was observed. These data demonstrate that MRTX0902 exhibits selective binding and functional inhibition of SOS1 (Fig. 1B).

Figure 1.

MRTX0902 is a potent and selective inhibitor of the SOS1:KRAS protein–protein interaction in vitro. A, Structure of MRTX0902. B, (left) HTRF binding assay with MRTX0902 utilizing a recombinant human SOS1 polypeptide and Cy5-labeled Tracer compound; (center) HTRF biochemical protein–protein interaction assay between recombinant SOS1 and recombinant wild-type (WT) KRAS and mutant KRAS (G12C, G12D, G12V), with MRTX0902; (right) HTRF functional assay that measures the SOS1- or SOS2-mediated exchange of GDP for GTP on KRAS, in the presence of MRTX0902. Activity of MRTX0902 was normalized to vehicle (0.2% DMSO) control. C, In-cell Western blot assay to evaluate the modulation of ERK1/2 phosphorylation (pERK) in KRAS–MAPK pathway-mutant cells treated with MRTX0902. D, CellTiter-Glo® (CTG) assay to evaluate cell viability performed on a panel of KRAS–MAPK pathway-mutant or WT KRAS cell lines grown in 3D tissue culture conditions in a 7-day or 14-day assay. A–D were previously published in Ketcham and colleagues (42). E, Western blot analyses of KRAS–MAPK pathway target pERK in MIA PaCa-2 (KRAS G12C–mutant) cells treated with MRTX0902 and/or adagrasib. Immunoblots represent multiple independent blots as detailed in the Supplementary Data section.

We next determined the cellular activity of MRTX0902 by measuring inhibition of ERK1/2 phosphorylation (pERK), a target engagement marker for KRAS–MAPK pathway activation. Additionally, the effect on viability in 3D growth conditions was characterized in a panel of cancer cell lines harboring mutations in the KRAS–MAPK pathway (Figs. 1C and D; Supplementary Table S1). Significant modulation of pERK and cell viability were observed following MRTX0902 treatment in EGFR mutant (NCI-H1975 and PC9), PTPN11 mutant (LN229), SOS1 mutant (SNU-C4 and HOP92), NF1 mutant (NCI-H1838 and HCC1438), KRAS G12C (NCI-H358), KRAS G13D (DLD1), KRAS G12S (A549), and class III BRAF (NCI-H508 and NCI-H1666) mutant cell lines with IC₅₀ values <250 nmol/L. In the wild-type amplified KRAS cell line, MKN1, pERK was modulated with MRTX0902 treatment with an IC₅₀ value of 39.6 nmol/L with no effect on viability observed. In summary, MRTX0902 inhibited pERK (IC₅₀ value <100 nmol/L) in 16 of the KRAS-MAPK pathway–mutated cell lines evaluated and exhibited antiproliferative activity (IC₅₀ value <250 nmol/L) in 13 of the 20 cell lines tested. This data suggests that MRTX0902 inhibits KRAS-mediated regulation of viability in cell lines harboring mutations within the MAPK pathway.

MRTX0902 treatment in combination with adagrasib leads to increased KRAS G12C modification and KRAS pathway inhibition in vitro

We next determined the cellular activity of MRTX0902 in combination with adagrasib utilizing the KRAS G12C–mutant MIA PaCa-2 pancreatic cancer cell line. As adagrasib is an irreversible inhibitor that specifically binds to the GDP (inactive) form of KRAS G12C, the addition of MRTX0902 is anticipated to increase the pool of inactive GDP-loaded KRAS G12C, thereby increasing its susceptibility to adagrasib binding. Adagrasib covalently modified KRAS G12C when incubated at 30 and 100 nmol/L for 1 and 3 hours, as previously published (34). Following incubation of MIA PaCa-2 cells with 1 µmol/L MRTX0902 and either 30 or 100 nmol/L adagrasib, we observed a time-dependent increase in modified KRAS G12C protein versus single agent adagrasib treatment as measured by % KRAS mobility shift (Fig. 1E; Supplementary Fig. S1A). Additionally, a time-dependent increase in inhibition of active RAS was observed compared with adagrasib monotherapy as determined by a RAF RAS-binding domain capture ELISA assay (Supplementary Fig. S1A). Lastly, combination treatment demonstrated greater inhibition of pERK compared with single agent adagrasib at both timepoints evaluated (Fig. 1E; Supplementary Fig. S1A). Thus, MRTX0902 treatment facilitates an increase in the covalent binding of adagrasib to the GDP-bound mutant KRAS G12C and prevents conversion to the active GTP-bound state.

CRISPR/Cas9 screen identifies vulnerabilities and modifiers of response to the combination of MRTX0902 with adagrasib in KRAS G12C–mutant cancer cell lines in vitro and in vivo

To gain additional insight into the mechanisms of therapeutic response or resistance following combination treatment, we executed a CRISPR/Cas9 knockout screen targeting approximately 1,000 genes implicated in oncogenic signaling. This was conducted in KRAS G12C–mutant MIA PaCa-2 and LU99 cells in vitro and xenografts in vivo in the presence and absence of adagrasib plus MRTX0902 (Supplementary Fig. S1B and S1C). In the adagrasib/MRTX0902-anchored screen in vitro, single guide RNAs (sgRNA) that target RTK signaling (EGFR, FGFR1, and CRKL), KRAS signaling (SOS2, SHOC2, and MYC), mTOR pathway (MTOR and PIK3CA), or cell cycle genes (CCNA2 and CDCA5, CDK1, and CCND1) reduced cell viability in at least one cell line. sgRNAs that target KEAP1, TSC1/2, PTEN, and NF1 were enriched in the in vitro anchored screen, demonstrating enhanced cellular fitness through loss of tumor suppressor genes. sgRNAs that target SOS1 and KRAS demonstrated less pronounced dropout in the adagrasib/MRTX0902-treated cells compared with vehicle control–treated cells as anticipated. In the adagrasib/MRTX0902-treated animals, sgRNAs targeting RTK/RAS signaling, cell cycle, MYC, and mTOR pathway genes remained among the top depleted sgRNAs and demonstrated enhanced TGI. Consistent with the in vitro screen, sgRNAs targeting tumor suppressor genes including KEAP1, PTEN, TSC1/2, and NF1 were enriched in adagrasib/MRTX0902-treated animals further implicating these genes in acquired resistance. Together, these data reveal additional targets that may mediate the bypass of therapeutic response to adagrasib/MRTX0902 or resistance associated with combination treatment with SOS1 and KRAS G12C inhibitors (Supplementary Fig. S1B and S1C).

As the SOS1 homolog SOS2 was uncovered as a top dropout gene following combination versus vehicle treatment in both KRAS G12C–mutant LU99 and MIA PaCa-2 cell lines, we hypothesized that the additional loss of SOS2 function would converge on increased RAS inhibition and deeper pERK modulation in the KRAS G12C–mutant background. We next generated both SOS1 and SOS2 knockout (KO) clones utilizing the KRAS G12C–mutant MIA PaCa-2 parental strain and profiled MRTX0902 cellular activity in MIA PaCa-2 parental (wild-type SOS1/2 proteins), SOS1 KO (B2), and SOS2 KO (F9) clones (Supplementary Fig. S1D and S1E). The KRAS G12C–mutant MIA PaCa-2 SOS2 knockout clone (F9) exhibited greater sensitivity to SOS1 inhibition compared with the parental KRAS G12C–mutant MIA PaCa-2 clone, with Hill’s slope values closer to 1 and comparable pERK and 3D viability IC₅₀ values. As expected, MRTX0902 treatment in the KRAS G12C–mutant MIA PaCa-2 SOS1 knockout clone (B2) demonstrated little to no effect in both assays. These findings suggest SOS2 plays a compensatory role in the absence of SOS1, and that dual inhibition could lead to greater antiproliferative activity in KRAS-MAPK pathway-mutant cell lines.

MRTX0902 treatment leads to antitumor activity and KRAS–MAPK pathway inhibition in vivo

A series of in vivo mouse pharmacology studies were used to characterize the pharmacokinetic, pharmacodynamic, and dose-dependent TGI effects of oral administration of MRTX0902 as a single agent in KRAS–MAPK pathway-mutant human tumor CDX models. Administration of 25 or 50 mg/kg twice daily (BID) MRTX0902 to athymic mice bearing NCI-H1435 (NF1 K615N-mutant) NSCLC xenografts resulted in concentration-dependent TGI of 50% and 73%, respectively (Fig. 2A). Furthermore, administration of 50 mg/kg BID for 6 days resulted in 72% inhibition of pERK protein levels at 1-hour post-dosing and 54% inhibition at the 3-hour timepoint (Fig. 2B). Similar pERK inhibition by MRTX0902 was observed in the PTPN11 A72S–mutant LN229 glioblastoma model with 34% TGI observed at the 50 mg/kg BID dose (Supplementary Fig. S2). In both models, loss of pERK inhibition was observed 6 hours post-administration, with the extent of pERK inhibition at each timepoint correlating with plasma concentrations of MRTX0902 (Fig. 2B; Supplementary Fig. S2B). Thus, MRTX0902 demonstrated dose-dependent pERK inhibition and TGI when administered orally twice daily to immunocompromised mice bearing human xenograft tumor models containing genetic alterations in the KRAS–MAPK pathway.

Figure 2.

MRTX0902 inhibits KRAS–MAPK pathway signaling and tumor growth in vivo. A, MRTX0902 was administered via oral gavage to mice bearing established NCI-H1435 tumor xenografts. Data for n = 5 animals/group is shown as mean tumor volume ± SEM. Tumor growth inhibition was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05 compared with the vehicle-treated group. B and C, MRTX0902 was administered to NCI-H1435 tumor-bearing animals for 6 days and pERK levels in tumor lysates were analyzed via immunoblot and quantified by densitometric analysis. Blood (plasma) was collected for MRTX0902 pharmacokinetic analysis. The data shown represent the individual data for three tumors per treatment group ± SD. Reduction of pERK relative intensity was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05; or ****, P < 0.0001 compared with vehicle-treated group. Immunoblot image is a composite of multiple independent blots ran as detailed in the Supplementary Data section.

Combination treatment with MRTX0902 and adagrasib leads to broad anti-tumor activity in KRAS G12C-mutant human tumor xenograft models

The efficacy of MRTX0902 as a single agent and in combination with adagrasib was initially characterized in the adagrasib-sensitive KRAS G12C–mutant pancreatic tumor cell line-derived MIA PaCa-2 model (Fig. 3; Supplementary Figs. S3A and S8; ref. 42). Oral administration of MRTX0902 at 25 or 50 mg/kg BID resulted in 41% and 53% TGI, respectively, whereas administration of the sub-maximally efficacious dose of 10 mg/kg QD adagrasib as a single agent resulted in 94% TGI (Fig. 3A). Increased antitumor activity was observed following co-administration of MRTX0902 at 25 or 50 mg/kg BID with 10 mg/kg QD adagrasib, which resulted in −54% and −92% tumor regression (including a tumor-free animal in the latter group), respectively (Fig. 3A). We further evaluated the effect of MRTX0902 monotherapy versus combination treatment on KRAS–MAPK pathway signaling by measuring inhibition of pERK in tumors following 6 days of treatment. Following dosing of MRTX0902 at 25 and 50 mg/kg BID in combination with 10 mg/kg QD of adagrasib, pERK levels were reduced by 66% and 81%, respectively, compared with 21% inhibition with single agent adagrasib at 10 mg/kg QD (Fig. 3B; Supplementary Fig. S3A).

Figure 3.

Antitumor activity of the MRTX0902/adagrasib combination in KRAS G12C–mutant human tumor xenograft models. A, MRTX0902 and adagrasib were administered orally to mice bearing established MIA PaCa-2 xenograft tumors. Data for n = 5 animals/group is shown as mean tumor volume ± SEM. Tumor growth inhibition or regression was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; or ****, P < 0.0001. B, MRTX0902 and adagrasib were administered as described in A for 6 days. Tumors were collected 4 hours after the last dose, and pERK levels were analyzed by immunoblot and quantified by densitometric analysis (Supplementary Fig. S3A). The data shown represent the individual data for three tumors per treatment group ± SD. Reduction in pERK relative fluorescence intensity was determined to be statistically significant using the two-tailed Student t test with annotations as described in A. Data from A and B are reproduced with permission (42). C, MRTX0902 and adagrasib were administered via oral gavage for 22–43 days to immunocompromised mice bearing human KRAS G12C–mutant tumor xenografts. Data is shown as a percent change in baseline tumor volume on ∼Day 28 of the study (Supplementary Table S2). Status of mutations and alterations in key genes for each model is tabulated in Supplementary Table S3. D, Tumor growth inhibition plots from representative human KRAS G12C–mutant CDX and PDX models (Supplementary Table S2). Tumor growth inhibition or regression was determined to be statistically significant using the two-tailed Student t test with annotations as described in A.

Next, the breadth of antitumor efficacy of MRTX0902 as a single agent and in combination with adagrasib was evaluated in a panel of genetically and histologically heterogeneous human KRAS G12C–mutant CDX and PDX models grown in immunocompromised mice (Fig. 3C). The response to MRTX0902 monotherapy across the panel of models varied from no effect to 43% TGI (Fig. 3C; Supplementary Table S2). As these models were selected based on the intermediate/moderate effectiveness of adagrasib monotherapy in prior studies, adagrasib administered at its maximally efficacious dose (100 mg/kg QD) demonstrated 27% to 99% TGI in the KRAS G12C–mutant CR2528 (CRC), SW1573 (NSCLC), NCI-H2122 (NSCLC), PA1383 (PANC), LU11692 (NSCLC), and CR6256 (CRC; Fig. 3C and D; Supplementary Table S2). Combination treatment with adagrasib plus MRTX0902 led to increased antitumor activity compared with either monotherapy alone across the majority of models and led to strong (>50%) tumor regression in four models (Fig. 3C and D; Supplementary Table S2). Notably, MRTX0902 and adagrasib as a monotherapy or in combination were well-tolerated and did not significantly affect body weight or demonstrate clinical signs in all studies at the doses reported. In summary, MRTX0902 treatment inhibits KRAS–MAPK pathway signaling and improves the antitumor activity of adagrasib when dosed as a combination in eight of 12 KRAS-mutated models. These data demonstrate that MRTX0902 treatment in combination with adagrasib can lead to deeper and more durable antitumor activity versus single agent adagrasib administration in human KRAS G12C–mutant tumor models in vivo. The implications of these findings are significant for patients with KRAS G12C–mutant cancers as this approach may provide a strategy to impart increased clinical benefit.

Temporal effects of adagrasib versus combination treatment on oncogenic signaling and relationship to antitumor activity following repeat dosing in human patient-derived xenograft models

Previous studies have shown that adagrasib broadly affects multiple KRAS-regulated pathways, including MYC, mTOR, cell cycle and apoptosis/BCL2 pathway gene sets (34). To evaluate the temporal changes in global gene expression following adagrasib versus combination treatment, we performed RNA sequencing analysis on tumors from animals treated with MRTX0902 (50 mg/kg BID), adagrasib (100 mg/kg QD), or adagrasib plus MRTX0902, in the KRAS G12C–mutant LU11692 (NSCLC) and CR6256 (CRC) PDX models. In the LU11692 model, a greater change in the modulation of gene expression for the MYC, E2F, and mTOR gene set enrichment analysis hallmark pathways and KRAS pathway targets was observed in the combination versus adagrasib treatment groups (Supplementary Fig. S3B and S3C). Further analysis of cell cycle and apoptosis pathway genes uncovered downregulation of multiple pro-survival genes following combination treatment. Expression of the BIRC5 and AURKB genes was significantly decreased whereas the proapoptotic BBC3 gene was increased in the combination versus adagrasib treatment groups. In the CR6256 model, significant alterations in KRAS signaling and in the gene set enrichment analysis hallmark and cell cycle gene sets included downregulation of MYC targets in all three treatment groups and more significant downregulation of MYC, G2M, and E2F targets and CDC25B gene expression following combination versus adagrasib treatment (Supplementary Fig. S3D and S3E). In summary, the RNA sequencing data uncovered molecular signatures associated with both SOS1 and KRAS inhibition and highlighted significantly greater alterations in cell cycle regulation that resulted in superior antitumor activity following combination treatment across KRAS G12C–mutant sensitive tumor models.

MRT0902 modulates KRAS pathway signaling in KRAS G12C–mutant human xenograft tumor models and whole blood

To further characterize the pharmacodynamic effects of MRTX0902, we monitored KRAS–MAPK pathway modulation in tumor samples (pERK) and whole blood (DUSP6 mRNA) from KRAS G12C–mutant tumor xenograft-bearing mice. DUSP6 is a transcriptionally regulated phosphatase downstream of ERK and a validated biomarker of KRAS–MAPK pathway activity (44, 45). Following 7 days of treatment with 50 mg/kg BID MRTX0902, 10 mg/kg QD adagrasib, adagrasib with MRTX0902, or 0.3 mg/kg BID Q2D avutometinib (RAF/MEK clamp) in the MIA PACa-2 tumor xenograft model, tumor pERK levels were significantly reduced by 23%, 54%, 84%, and 85%, respectively, compared with vehicle treatment 4 hours posttreatment (Fig. 4A; Supplementary Fig. S4). Levels of the downstream biomarker of KRAS-PI3K pathway signaling, p4EBP1, were not significantly modulated in any of the four treatment groups, however an approximately 30% reduction was observed following combination treatment (Supplementary Fig. S4). In whole blood, MRTX0902 monotherapy treatment led to a 63% reduction in DUSP6 transcript levels 4 hours posttreatment (Fig. 4B). As adagrasib is a KRAS G12C–mutant tumor-specific inhibitor, expectedly, negligible inhibition of KRAS–MAPK pathway activity in whole blood was observed. In a similar study utilizing naïve animals, following 7 days of treatment, MRTX0902 and avutometinib administration led to significant reductions in blood DUSP6 transcript levels (58% and 89%, respectively) 4 hours posttreatment, which recovered by the 6-hour timepoint (Fig. 4C). Together, these data demonstrate on-target inhibition of KRAS–MAPK pathway signaling in both blood and tumor xenografts by MRTX0902.

Figure 4.

MRTX0902 modulates KRAS–MAPK pathway signaling in human xenograft tumors and mouse whole blood. A, MIA PaCa-2 tumor-bearing animals were treated with vehicle or drug as indicated for 7 days. Tumors were collected 4 hours after the final dose, and pERK levels were analyzed by immunoblot and quantified by densitometric analysis. The data shown represent the individual data for three tumors per treatment group ± SD. Reduction of pERK relative fluorescence intensity was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; or ****, P < 0.0001. B, Whole blood was collected 4 hours after the last dose for analysis of DUSP6 transcript levels by RT-qPCR. The data shown represent the individual data for three blood samples per treatment group ± SD. Reduction of DUSP6 transcript levels for the treatment groups was determined to be statistically significant vs. vehicle control using the two-tailed Student t test with annotations as described in A. C, Naïve non-tumor-bearing animals (n = 3) were treated as described in A. Whole blood was collected prior to treatment (pre-treatment) and after the last dose for analysis of DUSP6 transcript levels. Reduction of DUSP6 transcript levels for each treatment group was determined to be statistically significant vs. vehicle control using the two-tailed Student t test with annotations as described in A.

Clinically tractable strategy to address acquired resistance associated with KRAS amplification

Resistance to targeted therapy is a common challenge in the treatment of cancer. Following treatment with adagrasib, acquired resistance associated with amplification of the KRAS G12C–mutant allele specifically on extrachromosomal DNA [KRAS (+) ecDNA] was previously observed in the murine KRAS G12C–mutant-engineered CT26 model of CRC (referred to hereafter as CT26; ref. 46). With SOS1 being both a proximal and common intermediate in RTK/MAPK signaling, inhibition is anticipated to address mechanisms of acquired resistance. To test this hypothesis, we next evaluated if the combination could overcome acquired resistance associated with adagrasib treatment in CT26 tumor-bearing immunocompromised mice. Treatment of MRTX0902 with adagrasib led to greater antitumor activity (81% and 99% TGI) versus 30 and 100 mg/kg QD adagrasib monotherapy treatment (53% and 88% TGI; Fig. 5A). Notably, in the combination-treated tumors that were actively growing at a similar growth rate compared with vehicle-treated tumors at the time of harvest, we observed an additional 1.5- to 2-fold increase in Kras G12C gene and KRAS protein levels and significantly greater pS6 activity compared with tumors treated with adagrasib only (Fig. 5B–E; Supplementary Fig. S5). These data suggest the Kras G12C DNA was amplified in order for unmodified KRAS G12C protein levels to increase in treated tumors to reach a comparable level of MAPK signaling as observed in untreated tumors. In summary, the combination of adagrasib with MRTX0902 led to more durable antitumor activity compared with adagrasib monotherapy treatment, and the combination has the potential to delay KRAS (+) ecDNA amplification-associated acquired resistance.

Figure 5.

Combination treatment with MRTX0902 and adagrasib has the potential to overcome acquired resistance associated with KRAS amplification. A, MRTX0902 and adagrasib were administered via oral gavage as indicated for 13 days to immunocompromised mice bearing murine CT26 tumors. The data for n = 5 animals/group are shown as mean tumor volume ± SEM. Tumor growth inhibition between two experimental groups was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; or ****, P < 0.0001. Tumors were collected 3 hours after the last dose, and Kras G12C gene levels were quantified via whole genome sequencing (B) and KRAS, pERK, and pS6 proteins were analyzed by immunoblot and quantified by densitometric analysis (C–E). The data shown represent the individual value of 2 to 5 tumors per treatment group ± SD. Alterations in KRAS, pERK, and pS6 relative fluorescence intensity were determined to be statistically significant using the two-tailed Student t test with annotations as described in (A).

Clinically tractable strategy to address RTK-mediated feedback associated with EGFR inhibition

Tagrisso® (osimertinib) is approved as first-line therapy for adjuvant post-tumor resection or metastatic NSCLC in patients harboring EGFR mutations including exon 19 deletions, exon 21 L858R, and/or exon 20 T790M (47). Following treatment with osimertinib, resistance to therapy is characterized by both EGFR-dependent or EGFR-independent mechanisms including amplification of neighboring receptors MET and HER2, and subsequent magnification of oncogenic signaling and tumor growth. Preclinically, it has been shown that EGFR-mutant NSCLC cell lines are sensitive to inhibition of SOS1 activity (48). To test if proximal RTK/MAPK pathway inhibition via MRTX0902 may circumvent or delay therapeutic resistance to osimertinib, we evaluated the effects of the combination of MRTX0902 with osimertinib versus osimertinib monotherapy on cell signaling and viability in two EGFR mutant cell lines in vitro (Supplementary Fig. S6). Combination treatment led to the 27% to 72% reduction in active RAS levels following 4 to 12 hours of treatment, greater and more durable suppression (76%–96%) of pERK and pAKT signaling over 48 hours, and additive cell death in PC9 (EGFR ex19del) cells (Supplementary Fig. S6A). In the NCI-H1975 (EGFR L858R/T790M-mutant) cell line, which contains an additional gatekeeper mutation, combination treatment also led to approximately 60% reduction in active RAS, most notably beyond 12 hours, both deeper and sustained suppression (77%–98%) of pERK and pAKT signaling over 48 hours, and additive cell death (Supplementary Fig. S6B). Overall, the combination led to significant therapeutic benefits compared with monotherapy in vitro.

Next, a series of in vivo mouse pharmacology studies were used to characterize the pharmacodynamic effects of MRTX0902 as a single agent or in combination with osimertinib in EGFR mutant human tumor CDX models. Following 6 days of administration of 50 mg/kg BID MRTX0902, 5 mg/kg QD osimertinib, or the combination, tumors were collected at 0.25 hour [∼T_max for MRTX0902 (42)] and 4 hours postdose, and pERK and pAKT levels were analyzed. Combination treatment led to a significant reduction in pERK levels at the early timepoint compared with osimertinib monotherapy treatment in both models (Fig. 6A and B; Supplementary Fig. S6C). Similar, yet statistically insignificant, trends in pERK modulation were observed at the 4-hour timepoint. Finally, pAKT levels were reduced to similar extents following both osimertinib and combination treatment at both timepoints (Fig. 6A and B; Supplementary Fig. S6C). In summary, MRTX0902 with osimertinib leads to greater suppression of KRAS–MAPK versus monotherapy treatment in vivo.

Figure 6.

Rational MAPK combinations of MRTX0902 with osimertinib (EGFRi) or avutometinib (RAF/MEK clamp) demonstrate improved efficacy in KRAS-dependent cancer. A and B, MRTX0902 and osimertinib were administered via oral gavage as indicated for 6 days to immunocompromised mice bearing established human EGFR mutant xenografts models (PC-9 and NCI-H1975). Tumors were collected at the indicated timepoints (0.25 hour—solid circle; 4 hours—open circle), and KRAS pathway target proteins were analyzed by immunoblot and quantified by densitometric analysis as previously described. The data shown represent the individual value of 2 to 3 tumors per treatment group ± SD. Reduction of pERK and pAKT relative fluorescence intensity was determined to be statistically significant using the two-tailed Student t test. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; or ****, P < 0.0001. C, MRTX0902 and osimertinib were administered as described in A and B for approximately 80 days. The data for n = 5 animals/group for both PC-9 and NCI-H1975 CDX models are shown as mean tumor volume ± SEM. Tumor growth inhibition was determined to be statistically significant using the two-tailed Student t test with annotations as described in A and B (Supplementary Table S4). D and E, MRTX0902 and avutometinib were administered as indicated for 7 days to immunocompromised mice bearing established NCI-H1435 or LN229 xenograft tumors. Tumors were collected at the indicated timepoints, and KRAS pathway target proteins were analyzed by immunoblot and quantified by densitometric analysis as previously described. Data (pERK) for vehicle and 50 mg/kg BID MRTX0902 treatment groups are regraphed from Fig. 2B and Supplementary Fig. S2B. The data shown represent the individual value of three tumors per treatment group plus SD. Reduction of pERK and p4EBP1 relative fluorescence intensity was determined to be statistically significant using the two-tailed Student t test with annotations as described in A and B. F, MRTX0902 and avutometinib were administered via oral gavage as described in A and B for 27 to 32 days. The data for vehicle and 50 mg/kg BID MRTX0902 treatment groups are regraphed from Fig. 2A and Supplementary Fig. S2A. The data for n = 5 animals/group for both models are shown as mean tumor volume ± SEM. Statistical significance was determined for each model and is shown in Supplementary Table S5. Tumor growth inhibition or regression between two experimental groups was determined to be statistically significant using the two-tailed Student t test with annotations as described in A and B (Supplementary Table S5).

To investigate whether the combination of osimertinib with MRTX0902 translated to greater antitumor activity, we next evaluated the efficacy of a sub-maximally efficacious dose of osimertinib (5-mg/kg QD) as a single agent versus combination with 50 mg/kg BID MRTX0902 in both models in vivo (Fig. 6C; Supplementary Table S4). Following approximately 80 days of treatment in the PC9 model, combination treatment of MRTX0902 with osimertinib resulted in significantly enhanced antitumor activity, with −35% tumor regression observed versus 98% TGI for osimertinib alone. A similar favorable trend was observed in the NCI-H1975 model, with 72% TGI observed for 5 mg/kg QD osimertinib treatment and −43% tumor regression for the combination. Treatment with MRTX0902 or osimertinib as a monotherapy or in combination did not result in any adverse effects on body weight or remarkable clinical signs in both xenograft models tested.

Clinically tractable strategy to address RTK-mediated feedback associated with RAF/MEK inhibition

Another rational combination strategy of interest includes inhibition via MRTX0902 coupled with downstream inhibition of the KRAS–MAPK pathway. Several nodes within the MAPK pathway have been explored for their function and druggability and inhibition of RAF family members as well as MEK1/2 have long been pursued for the treatment of KRAS–MAPK pathway-mutant and hyperactive cancers. Both MEK1/2 and RAF inhibitors induce reactivation of RAS pathway signaling and lead to modest efficacy in RAS-mutant cancer. Moreover, MEK1/2 inhibitors suffer from a low therapeutic index in humans when administered as a monotherapy in KRAS-mutant cancers, resulting in poor clinical outcomes (20). The RAF/MEK clamp, avutometinib, blocks both RAF and MEK kinase activities, thereby preventing the reactivation of MEK by RAF and leading to antitumor activity in RAS/MAPK-dependent cancers (49, 50). Moreover, inhibition of MEK leads to a release of a negative feedback loop within the MAPK pathway, resulting in increased SOS1 activity (17, 18); thus, inhibition of RAF, MEK, and SOS1 could lead to greater and more durable antiproliferative activity in tumors with MAPK activation. To test this hypothesis, we evaluated the effects of the combination versus avutometinib monotherapy on cell signaling and viability in two mutant cell lines that exhibited moderate TGI following MRTX0902 treatment, the PTPN11 A72S-mutant LN229 and NF1 K615N-mutant NCI-H1435 cell lines (Supplementary Fig. S7). MRTX0902 treatment demonstrated a 43% reduction in RAS-GTP (at 4 hours), whereas combination treatment led to similar decreases in pERK levels, greater and more durable suppression of p4EBP1 signaling (at 48 hours), and increased cell death versus avutometinib monotherapy in the LN229 cell line (Supplementary Fig. S7A). In the NCI-H1435 cell line, MRTX0902 monotherapy demonstrated reduction (∼75%) in GTP-bound RAS following 48 hours of treatment, whereas combination treatment resulted in both deeper and sustained suppression (∼75%–90%) of pERK and p4EBP1 signaling for 48 to 72 hours and increased cell death compared with avutometinib treatment (Supplementary Fig. S7B). Overall, the combination led to significantly greater inhibition of tumor cell viability compared with monotherapy in vitro.

We further evaluated MAPK pathway inhibition by MRTX0902 as a single agent and combination partner of avutometinib in the established LN229 and NCI-H1435 tumors implanted in immunocompromised mice. Following 7 days of administration of 50 mg/kg BID MRTX0902, 0.3 mg/kg BID Q2D avutometinib, or the MRTX0902/avutometinib combination, tumors were collected at 1-, 3-, and 6-hour after the last dose, and pERK (MAPK pathway) and p4EBP1 (PI3K pathway) levels were analyzed. Combination treatment led to significant reduction of pERK at 3- and 6-hour posttreatment in the LN229 model and 1- and 3-hour posttreatment in the NCI-H1435 model (Fig. 6D and E; Supplementary Fig. S7C). In summary, MRTX0902 with avutometinib leads to significantly greater suppression of KRAS–MAPK pathway signaling in vivo compared with single-agent treatment.

To evaluate the antitumor activity of the combination, LN229 and NCI-H1435 CDX tumor-bearing animals were treated with MRTX0902, avutometinib, or MRTX0902 plus avutometinib for approximately 4 weeks. In both models, combination treatment resulted in significantly enhanced antitumor activity, with 99% TGI compared with 80% TGI following avutometinib treatment (Fig. 6F; Supplementary Table S5). Of note, administration of MRTX0902 or avutometinib as a monotherapy or in combination did not result in any adverse effects on body weight or remarkable clinical signs in both studies. Thus, combination treatment led to greater TGI compared with avutometinib monotherapy treatment, supporting the rationale of SOS1 and RAF/MEK inhibition for enhanced antitumor activity in KRAS pathway-dependent cancers characterized by mutations in proteins that are either directly upstream of (PTPN11) or are the negative regulator of (NF1) SOS1-mediated RAS activation.

Discussion

The identification of MRTX0902 as a selective SOS1 inhibitor that enhances the efficacy of KRAS G12C inhibition (adagrasib) provides insight into a clinically-relevant strategy that is currently being evaluated in patients with solid tumors harboring mutations in the KRAS–MAPK pathway (ClinicalTrials.gov, Identifier: NCT05578092). In this study, MRTX0902 treatment led to significant antiproliferative activity in KRAS–MAPK pathway mutant cell lines harboring mutations in EGFR, PTPN11, SOS1, NF1, KRAS G12C/G12S/G13D, and (class III) BRAF, and notable in vivo activity as a monotherapy in an NF1 mutant model, similar to previous findings with the SOS1 inhibitor BI-3406 (24) and the SHP2 inhibitor RMC-4550 (36). The activity of RMC-4550 is attributed to its ability to prevent SOS1-mediated activation of RAS, thereby indirectly inhibiting RAS and subsequent oncogenic RAS-RAF-MEK-ERK signaling. Given the pleiotropic functions of SHP2, including in normal tissues, SHP2 inhibitors may lack sufficient KRAS-pathway specificity to achieve a tolerated continuous dose regimen and consistent target coverage, thereby potentially limiting their therapeutic index in the clinic (38, 39, 51). In contrast, SOS1 inhibition by MRTX0902 is specific to the KRAS:SOS1 PPI and is anticipated to be better tolerated when administered continuously as both a monotherapy and combination partner of KRAS G12C inhibitors. Moreover, given the high incidence of KRAS–MAPK pathway germline mutations in RASopathies, there is additional therapeutic potential for MRTX0902 in SOS1-dependent genetic syndromes and related cancers. Interestingly, we show on-target activity of MRTX0902 can be monitored non-invasively as a surrogate for tumor tissue using whole blood. Additional insights will be gained as patients harboring various KRAS–MAPK pathway mutations are evaluated following MRTX0902 treatment.

To identify additional vulnerabilities and resistance mechanisms associated with MRTX0902 monotherapy and MRTX0902/adagrasib treatment, we conducted drug-anchored CRISPR screens in two KRAS G12C–mutant xenograft models and found that the SOS1 homolog, SOS2, emerges as a co-dependency under selective pressure of combination treatment and likely compensates for the inhibition of SOS1. Thus, SOS2 protein levels in human tumor cell lines may predict sensitivity or resistance to MRTX0902. Additional in vitro characterization revealed greater antiproliferative effects of MRTX0902 in the absence of SOS2, indicating that dual inhibition could lead to greater efficacy in models of KRAS–MAPK pathway–dependent cancer. While there may be increased efficacy by co-targeting SOS1 and SOS2, there also may be limitations as a previous study revealed that dual inactivation resulted in death within 2 weeks of treatment in a tamoxifen-inducible null mutant model (5). It is also noteworthy that there is an appreciable divergence between SOS1 and SOS2 protein in the binding pocket for MRTX0902 such that this molecule demonstrates strong SOS1 selectivity. Whether the development of a SOS2-selective inhibitor or a SOS1/SOS2 small molecule inhibitor to more fully block KRAS activation is feasible and whether co-targeting both proteins would be generally tolerated represent key questions in the field.

Although exploring additional clinically relevant strategies to address therapeutic resistance, we observed that combined vertical inhibition of RTK/MAPK pathway signaling by osimertinib and MRTX0902 improved the antitumor activity of osimertinib in both PC9 and NCI-H1975 EGFR mutant models, the latter of which contains the EGFR T790M gatekeeper mutation that can arise following treatment with first-generation EGFR inhibitors (47). This finding is particularly relevant for patients with EGFR mutations who have disease progression during or after first-line therapy with osimertinib, as the recommended subsequent therapy options include continuing osimertinib, surgery, or first-line systemic therapy regimen for metastatic NSCLC (such as carboplatin/paclitaxel; NCCN guidelines April 2022). Thus, this combination can directly target RTK/MAPK-mediated bypass of EGFR dependence to circumvent or delay therapeutic resistance associated with third-generation EGFR tyrosine kinase inhibitors. Finally, administration of MRTX0902 plus avutometinib resulting in dual inhibition of SOS1 and RAF/MEK, led to greater antitumor activity in PTPN11- and NF1-mutant models, similar to effects observed in KRAS G12C–mutant cancer models (24). MEK inhibitors as a single agent in KRAS-mutant cancers suffer from a low therapeutic index in humans (52–56); thus, the paradigm has shifted to look at combinations with other targeted inhibitors to more effectively and safely suppress MAPK pathway signaling (20). Co-administration of MRTX0902 with avutometinib demonstrated increased efficacy and was well-tolerated with no adverse side effects in mice.

Collectively, the present studies support the potential broad utility of a SOS1 inhibitor for the treatment of cancers with mutations in the KRAS–MAPK pathway and inform potential clinical strategies for patients likely to benefit from either monotherapy or rational combination strategies.

Supplementary Material

Supplementary Materials and Methods

Supplementary Materials and Methods section includes detailed protocols for the following: Experimental Preparation of MRTX0902 and Avutometinib, SOS1 Biochemical Binding Assay, SOS1 and SOS2 Functional Assays, KRAS-SOS1 Protein-Protiein Interaction (PPI) HTRF Assay, In-Cell Western Assay, 3D Ultra-Low Attachment (ULA) Viability Assay, Immunoblotting and Densitometry Analysis, DUSP6 Quantification from Naive and Tumor-Bearing Mouse Blood, Bioanalysis and Pharmacokinetic Analysis, Synergy Analysis, CRISPR/Cas9 Screening and Data Analysis Methodology, RNAseq Pre-Processing, RNAseq Data Analysis, and Whole Genome Sequencing.

Supplementary Figure S1

Supplementary Figure S1 details the genetic vulnerabilities and modifiers of response associated with combination treatment of MRTX0902 with adagrasib in the KRAS G12C-mutant MIA PaCa-2 (S1A, S1B, S1D, and S1E) and LU99 (S1C) cell lines in vitro and in vivo.

Supplementary Figure S2

Supplementary Figure S2 shows the antitumor activity of MRTX0902 in the LN229 (PTPN11 A72S-mutant) model, with tumor growth inhibition data displayed in Figure S2A and ERK phosphorylation graphed in Figure S2B.

Supplementary Figure S3

Supplementary Figure S3 details the antitumor effects and gene set enrichment analysis data associated with coadministration of MRTX0902 and adagrasib in KRAS G12C-mutant human tumor xenograft models.

Supplementary Figure S4

Supplementary Figure S4 shows levels of KRAS-MAPK pathway modulation associated with coadministration of MRTX0902 and adagrasib in the MIA PaCa-2 (KRAS G12C-mutant) model.

Supplementary Figure S5

Supplementary Figure S5 shows levels of KRAS-MAPK pathway modulation associated with coadministration of MRTX0902 and adagrasib in the murine CT26 (KRAS G12C-mutant) model.

Supplementary Figure S6

Supplementary Figure S6 details the improved antiproliferative activity observed with the MRTX0902/osimertinib combination in EGFR mutant models, PC9 (EGFR ex19del) and NCI-H1975 (EGFR L858R/T790M).

Supplementary Figure S7

Supplementary Figure S7 shows the improved antiproliferative activity observed with the MRTX0902/avutometinib combination in KRAS-MAPK pathway mutant models, LN229 (PTPN11 A72S) and NCI-H1435 (NF1 K615N).

Supplementary Figure S8

Supplementary Figure S8 details the antitumor effects associated with coadministration of MRTX0902 and adagrasib in the MIA PaCa-2 (KRAS G12C-mutant) model.

Supplementary Table S1

Supplementary Table S1 details the efficacy of MRTX0902 in a panel of cancer cell lines.

Supplementary Table S2

Supplementary Table S2 shows the antitumor activity of MRTX0902 and adagrasib in a panel of human KRAS G12C-mutant tumor cell line-derived and patient-derived xenograft models.

Supplementary Table S3

Supplementary Table S3 displays the various genetic co-alterations present in human KRAS G12C-mutant tumor cell line-derived and patient-derived xenograft models evaluated.

Supplementary Table S4

Supplementary Table S4 shows the antitumor activity of MRTX0902 and osimertinib in human EGFR-mutant tumor cell line-derived xenograft models.

Supplementary Table S5

Supplementary Table S5 details the antitumor activity of MRTX0902 and avutometinib in human KRAS-MAPK pathway mutant tumor cell line-derived xenograft models.

Authors’ Disclosures

N. Sudhakar reports other support from Mirati Therapeutics outside the submitted work. L. Yan reports personal fees from Mirati Therapeutic during the conduct of the study. L.D. Engstrom reports other support from Mirati Therapeutics outside the submitted work. A. Calinisan reports personal fees from Mirati Therapeutics during the conduct of the study, as well as personal fees from Mirati Therapeutics outside the submitted work. A. Hebbert reports personal fees from Mirati Therapeutics outside the submitted work. L. Waters reports other support from Mirati Therapeutics outside the submitted work. V. Bowcut reports other support from Mirati therapeutics outside the submitted work. L. Vegar reports personal fees from Mirati Therapeutics outside the submitted work. J.M. Ketcham reports other support from Mirati Therapeutics outside the submitted work; additionally, J.M. Ketcham has a patent for US11702418 issued, a patent for US2022/0395507 pending, and a patent for US2022/034314 pending. A. Ivetac reports other support from Mirati Therapeutics outside the submitted work; in addition, A. Ivetac has a patent for WO/2021/127429 pending, WO/2022/026465 pending, WO/2022/271679 pending, and WO/2023/250165 pending. J.D. Lawson reports a patent for US11702418 issued, WO/2021/173524 pending, WO/2022/271679 pending, and WO/2023/250165 pending. L. Rahbaek reports a patent for Wo2022/251193 pending, Wo2023/196086 pending, and Wo2023/196218 pending. B. Saechao reports other support from Mirati Therapeutics outside the submitted work. C. Parker reports other support from Mirati Therapeutics outside the submitted work. A.J. Elliott reports other support from Mitati Therapeutics outside the submitted work. S. Coma reports other support from Verastem Oncology during the conduct of the study; and other support from Verastem Oncology outside the submitted work; in addition, S. Coma has a patent for Several patents in combination with avutometinib pending and issued. J.A. Pachter reports other support from Verastem outside the submitted work; in addition, J.A. Pachter has a patent for Several patents related to avutometinib combinations pending and issued. J. Hallin reports personal fees from Mirati Therapeutics outside the submitted work. M.A. Marx reports other support from Mirati Therapeutics outside the submitted work; moreover, M.A. Marx has a patent for WO2022/251193 pending, WO2023/196086 pending, and WO2023/196218 pending. D.M. Briere reports other support from Mirati Therapeutics outside the submitted work. J.G. Christensen reports personal fees from Mirati Therapeutics during the conduct of the study and personal fees from Mirati Therapeutics outside the submitted work; in addition, J.G. Christensen has a patent of 10,633,381 pending and a patent of 20,220,331,324 issued and is an employee of Bristol Myers Squibb (via acquisition of Mirati Therapeutics). P. Olson reports other support from Mirati Therapeutics during the conduct of the study, as well as other support from Pfizer and Tango Therapeutics outside the submitted work. J. Haling reports other support from Mirati Therapeutics outside the submitted work; additionally, J. Haling has a patent for WO2022/251193 pending, WO2023/196086 pending, and WO2023/196218 pending. S. Khare reports other support from Mirati Therapeutics outside the submitted work; in addition, S. Khare has a patent for WO2022/251193 pending, WO2023/196086 pending, and WO2023/196218 pending. No disclosures were reported by the other authors.

Authors’ Contributions

N. Sudhakar: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology. L. Yan: Data curation, validation, methodology. F. Qiryaqos: Data curation, validation, methodology. L.D. Engstrom: Conceptualization, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology. J. Laguer: Data curation, methodology. A. Calinisan: Data curation, methodology. A. Hebbert: Data curation, methodology. L. Waters: Data curation, validation, methodology. K. Moya: Data curation, methodology. V. Bowcut: Data curation, methodology. L. Vegar: Data curation, methodology. J.M. Ketcham: Conceptualization, data curation. A. Ivetac: Conceptualization, data curation. C.R. Smith: Conceptualization, resources. J.D. Lawson: Conceptualization, data curation. L. Rahbaek: Data curation, supervision. J. Clarine: Data curation, supervision, validation. N. Nguyen: Data curation, methodology. B. Saechao: Data curation, validation, methodology. C. Parker: Data curation, methodology. A.J. Elliott: Data curation, methodology. D. Vanderpool: Data curation, supervision, methodology. L. He: Data curation, software, formal analysis, validation, visualization, methodology. L.D. Hover: Data curation, software, formal analysis, validation, visualization, methodology. J. Fernandez-Banet: Resources, software, formal analysis, supervision, methodology. S. Coma: Resources, methodology. J.A. Pachter: Resources, methodology. J. Hallin: Resources, investigation, methodology. M.A. Marx: Conceptualization, resources, supervision, project administration. D.M. Briere: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology. J.G. Christensen: Conceptualization, resources, formal analysis, supervision, investigation, visualization, methodology, writing—original draft, project administration, writing—review and editing. P. Olson: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing—original draft, project administration, writing—review and editing. J. Haling: Conceptualization, data curation, supervision, investigation, methodology, personal email contact: jrhaling@gmail.com. S. Khare: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing—original draft, writing—review and editing.