Figure S17. Cell deconvolution and cell specificity of spatial transcriptomic expression.

Figure S11. Identified ligand-receptor interactions involving CT-26 KRASG12C cells following no treatment.

Figure S14. Analysis of T-cell function, expansion and TCR clonotype analysis after KRAS(G12C)i and MEKi.

Figure S5. Classification of immune cell states in treated CT-26 KRASG12C tumors.

Table S1. Mouse T cell phenotyping antibody panel; Table S2. Mouse macrophage phenotyping antibody panel; Table S3. Mouse dendritic cell phenotyping antibody panel; Table S4. Mouse T cell function antibody panel; Table S5. FACS panel for RNA-seq; Table S6. Antibodies used to BMDC co-culture.

Abstract Although mutant-specific KRAS inhibitors are approved to treat cancer, a deeper understanding of intratumoral changes driven specifically by KRAS inhibition is needed to maximize therapeutic responses. In this study, we used single-cell RNA sequencing, flow cytometry, and spatial transcriptomics to distinguish mechanisms of tumor control after KRASG12C inhibition [KRAS(G12C)i] or MEK inhibition (MEKi). Despite both inhibiting the MAPK pathway, KRAS(G12C)i and MEKi drive the adaptation of distinct neoplastic cell fates affecting metabolism and cell-cycle regulation, and additive tumor suppression is observed after co-administration. KRAS(G12C)i results in the emergence of a specific, cDC1-driven mature conventional dendritic cell (cDC) state. Coculture of treated neoplastic cells with cDC1s is sufficient to upregulate maturation markers such as CCR7, and intercellular communication analyses suggest that activation is augmented through nonimmune mediators. Both KRAS(G12C)i and MEKi increase infiltration of cytotoxic T cells, but MEKi, which also targets nonmalignant cells, is associated with a reduced capacity for T-cell proliferation and degranulation, consistent with distinct adaptive immune activation mechanisms. We observe that combination treatment of KRAS(G12C)i with anti–PD-1 immunotherapy further expands effector T-cell states, increases clonal persistence, and induces proinflammatory macrophages associated with higher overall survival that were largely absent after KRAS(G12C)i alone. Furthermore, combination treatment enhances intercellular communication networks among non–PD-1+–expressing cells that can perpetuate cDC activation. Our findings delineate distinct tumor and immune responses to KRAS and MEK inhibition and identify molecular features of the responding tumor microenvironment that may be leveraged to improve therapeutic efficacy.

Abstract KRAS is the most frequently mutated oncogene in solid tumors. Covalent KRAS G12C-selective inhibitors have been approved for G12C-mutant non-small cell lung cancer (NSCLC). Despite their clinical success, the durability of response and emergence of resistance have tempered the efficacy of KRAS-targeted therapies, leading to the exploration of combination strategies to enhance clinical outcomes. AMG 410, currently in Phase 1, is a reversible pan-KRAS inhibitor capable of targeting KRAS mutants (e.g., G12D, G12V, and G12C) and wild-type amplification. These altered KRAS alleles are prevalent in multiple solid tumor indications, particularly colorectal cancer (CRC), pancreatic ductal adenocarcinoma (PDAC), and NSCLC. Preclinically, AMG 410 demonstrated significant tumor growth inhibition as a single agent in multiple cell line-derived xenograft (CDX) and patient-derived xenograft (PDX) models. Guided by reverse translation findings from the G12C-selective inhibitor sotorasib, we evaluated AMG 410 combination approaches in preclinical models to proactively mitigate potential primary and acquired resistance mechanisms. A large panel of diverse KRAS-mutant cancer cell lines representing three indications (NSCLC, CRC, and PDAC) was screened in combination with inhibitors of the RAS signaling pathway, DNA damage repair pathways, cell cycle regulators, and select chemotherapy agents. Robust synergistic effects on cell viability were observed with pan-HER kinase and PI3K/mTOR inhibitors. To further explore resistance mechanisms, an AMG 410-anchored genome-wide CRISPR screen was performed. The Hippo pathway and YAP1 were identified as key modifiers of response to KRAS inhibition. Co-treatment with YAP/TEAD inhibitors demonstrated strong synergy in multiple KRAS-mutant cell lines. To assess whether these observations translated to improved efficacy in vivo, rational combinations were evaluated in tumor xenograft models. Consistent with the in vitro synergy observed with pan-HER inhibition, the combination of AMG 410 with panitumumab resulted in tumor regression in a CRC PDX model. Additionally, enhanced anti-tumor activity was observed when AMG 410 was combined with chemotherapy agents in CRC and PDAC xenografts. Building on the potential benefit of clinical KRAS G12C immune-oncology combinations, treatment with AMG 410 and PD-1 blockade in a KRAS G12D syngeneic CRC model led to tumor regression and significantly enhanced survival. Finally, co-treatment with a TEAD inhibitor enhanced durability of response to AMG 410 in vivo. Taken together, these findings support the clinical investigation of AMG 410 combination strategies to extend the therapeutic benefit of KRAS inhibition across diverse KRAS-mutant cancers. Citation Format: Ying-Chu Chen, Tao Osgood, Kevin Gaida, Gilbert Diaz, Chun Su, Elissa Swearingen, Daniel Lu, Deanna Mohn, Anne Y. Saiki, Monica Leavitt, Upendra P. Dahal, Ryan P. Wurz, Brian A. Lanman, Jason DeVoss, Karen Rex, Paul E. Hughes, Rati Verma. Preclinical combination approaches with the pan-KRAS inhibitor AMG 410 in KRAS-mutant cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5866.

Figure S1. Quality filtering of scRNA-seq of all cells from CT-26 KRASG12C syngeneic tumors and classification of cell types.

Figure S18. Spatial transcriptomic analysis of immune signatures and enhanced intercellular communication following KRAS(G12C)i with anti-PD-1 blockade.

Abstract Bispecific T-cell engagers (TCEs) have demonstrated transformative clinical efficacy by redirecting T cells to eliminate tumor cells. However, clinical outcomes highlight opportunities to further enhance therapeutic benefits, particularly by improving overall response rates and durability of responses. Sustained CD3 engagement by TCEs can lead to T cell exhaustion, resulting in diminished effector function and reduced persistence. To address these challenges, we developed AMG 728, a PD-L1-targeted 4-1BB bispecific molecule designed to enhance T cell anti-tumor activity and persistence. AMG 728 simultaneously blocks PD-1/PD-L1 inhibitory signaling and activates 4-1BB costimulatory signaling in a PD-L1-dependent manner. This dual mechanism is intended to enhance T cell activation and persistence while minimizing the risk of immune activation and off-tumor toxicity associated with systemic 4-1BB agonism. By combining the CD3-mediated signaling from TCEs with 4-1BB costimulation provided by AMG 728, we aimed to fully activate T cells, thereby promoting cytolytic effector differentiation, improving survival, and expanding memory T cell populations. A mouse surrogate PD-L1-4-1BB bispecific molecule demonstrated dose-dependent anti-tumor efficacy as a monotherapy in a human PD-L1/4-1BB double knock-in mouse model bearing syngeneic, human PD-L1-expressing tumors. Mechanistic characterization of this surrogate molecule further revealed several potential pharmacodynamic biomarkers in vivo, including increased CD8+ T cell proliferation, expansion of central memory T cells in draining lymph nodes, enhanced cytolytic effector T cell differentiation within tumors, and elevated levels of soluble 4-1BB. Furthermore, AMG 728 was evaluated for its potential to improve the efficacy of tarlatamab, a DLL3-targeted bispecific TCE in vitro and in a T cell humanized small cell lung cancer (SCLC) preclinical mouse model. Selective 4-1BB agonism on PD-L1-positive tumors enhanced the cytotoxic activity of tarlatamab in vitro. In the human PD-L1-overexpressing SHP-77 SCLC xenograft implanted in NSG mice, combination treatment with AMG 728 and tarlatamab at suboptimal doses resulted in significantly greater tumor growth inhibition, extended survival, and a higher frequency of complete responses, compared with either monotherapy. Notably, AMG 728 treatment led to a significant increase in central memory T cells in the tumors from tarlatamab-treated mice. Collectively, these findings suggest that the addition of AMG 728 to tarlatamab represents a rational combination strategy that enhances T cell activation and longevity, resulting in improved antitumor activity and survival. Citation Format: Felipe Vences Catalan, Willy Tsai, Wendy Chen, Anja Henn, Kevin Cook, Maryam Yousefi, Tanya Vagner, Andrew Jimena, Khushboo Sharma, Deepali Sawant, Jason DeVoss, Matthias Friedrich, Mithun Khattar, Julie Bailis, Andrew Rankin, Sungeun Kim. PD-L1-dependent 4-1BB costimulation enhances anti-tumor efficacy and T cell persistence as monotherapy or in combination with tarlatamab [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 7762.

Figure S4. Classification of immune cell types in treated CT-26 KRASG12C syngeneic tumors.

Figure S6. Characterization of classical dendritic cell states (cDC) in treated tumors.

Figure S7. Flow cytometry analyses of dendritic cell infiltration and maturation.

Figure S8. In-vitro bone marrow-derived dendritic cell (BMDC) co-culture with treated KRASG12C cell lines.

Figure S10. Identified ligand-receptor interactions involving neoplastic cells following KRAS(G12C)i and MEKi treatment.

Figure S2. Characterization of CT-26 KRASG12C neoplastic cell states in vivo.

Figure S3. Quality assessment of single-cell RNA-seq in CD45+ tumor-infiltrating immune cells.

Figure S15. Cell state dynamics of intratumoral monocytes and macrophages after KRAS(G12C)i and MEKi.

Figure S9. Summary of intratumoral inter-cellular signaling interactions in CT-26 KRASG12C.

Figure S16. CD8 T cell intratumoral dynamics following KRAS(G12C)i and anti-PD1 combination therapy.