Spatial analysis of mixed KPAR1.3 subcutaneous tumors

Supplementary Figure 2. KRAS-driven inhibition of tumor-intrinsic IFN signaling promotes immune evasion

G12Ci as monotherapy promotes the outgrowth of drug-resistant subpopulations. A, Viability assays of Luc-eGFP KRAS G12D or eBFP KRAS G12C KPAR1.3 subclones compared with the parental counterparts. Cells were treated with adagrasib for 72 hours. Mean ± SEM of three biological replicates (n) with three technical replicates. B, Experimental setup for C–E. Subcutaneous tumors were engrafted with different ratios of eBFP KRAS G12C and Luc-eGFP KRAS G12D cells, as indicated. Mice were treated daily with the G12Ci adagrasib (50 mg/kg; gray area). n = 6–8 mice per group. C, Tumor growth over time for indicated ratios of KRAS G12C:KRAS G12D cells. D, Bioluminescence scans over time, detecting signal emitted by luciferase-expressing KRAS G12D cells. E, Flow cytometric analysis of Luc-eGFP KRAS G12D cells of all live cancer cells. All graphs display mean ± SEM; pair-end Student t test with Welch correction. F, Schematic of subpopulation dynamics. ***, P < 0.001; ****, P < 0.0001.

Immune rejection of mechanistically distinct G12Ci-resistant tumor subpopulations. A, Viability assays of KPAR.E2 cells and parental KPAR1.3 KRAS G12C cells (parent G12C). Cells were treated for 72 hours with the G12Cis adagrasib and RMC-4998. Mean ± SEM; n = 3. B, Western blots of cells treated at different timepoints with 100 nmol/L adagrasib. C and D, RNA-seq analysis comparing gene expression at baseline (DMSO-treated samples) between the KPAR1.3 KRAS G12C cell line (parent) and KPAR.E2 cells. Heatmaps for leading-edge genes from GSEA (MSigDB Hallmarks; KRAS up (C) and EMT signatures (D). Z-scores normalized to parent group mean. E, Western blots comparing basal protein expression between parental KPAR1.3 and KPAR.E2 cells. F, Tumor growth of Luc-eGFP KPAR.E2 subcutaneous tumors treated for 2 weeks with RMC-4998 (G12Ci). Growth of vehicle-treated mice is shown in Supplementary Fig. S15A. G and H, Mixed subcutaneous tumors (BFP KRAS G12C cells plus 0.04% Luc-eGFP-KPAR.E2 cells) were treated for 2 weeks with RMC-4998 (G12Ci) ± RMC-4550 (SHP2i) or anti–PD-1 (aPD-1). Growth of individual tumors. Complete responders (CR) indicated (G). H, Survival probability across treatment groups; cut-off value: 500 mm3(H). I, Rechallenge of complete responders (CRs) from G with untraced KPAR.E2 cells compared with naïve control mice. Immune-rejected (IR) tumors are indicated as the fraction of all CRs. Graph titles indicate the treatment that the primary tumor received.

Supplementary Figure 6. Tumor-intrinsic COX-2 remodels the lung tumor microenvironment

Predicted MHC-I binding affinities for eBFP peptides

Spectral flow cytometry gating strategy for Emv2 tetramer

Triple combination increases the generation of complete responders

RNA-seq analysis of inflammation-associated tumor cell states

Combination therapies promote elimination of resistant subclones. A and B, Luc-eGFP KRAS G12D KPAR1.3 subcutaneous tumors treated with vehicle (n = 7) or RMC-4550 (SHP2i; 30 mg/kg; n = 8)(A); isotype control (10 mg/kg, IgG, n = 6) or anti–PD-1 (aPD-1; 10 mg/kg; n = 7) (B). Doses indicated with gray area (SHP2i) or vertical dotted lines (anti–PD-1). Mean ± SEM. C–F, Mixed subcutaneous tumors (BFP KRAS G12C cells plus 0.04% Luc-eGFP KRAS G12D cells) were treated for 2 weeks with RAS G12C (on) inhibitor RMC-4998 (G12Ci; 100 mg/kg) ± SHP2i RMC-4550 (30 mg/kg) or anti–PD-1 (10 mg/kg). Tumor volume over time of individual tumors. Complete responder (CR) mice are indicated in respective plots (C). Survival probability across different treatment groups. Cutoff value: 500 mm3. C, G12Ci; P, aPD-1; S, SHP2i (D). Bioluminescence scans, indicating the relative abundance of Luc-eGFP KRAS G12D cells, measured in total flux, photons per second (p/second) (E). Representative tumors were isolated from vehicle and single-therapy treatments, and the fraction of Luc-eGFP KRAS G12D cells within all cancer cells was determined via flow cytometry. Each dot represents one tumor; mean ± SEM; one-way ANOVA test comparing each of the treatment groups with the vehicle (F). G, Rechallenge experiment: CR mice from C were injected in the opposite flank with Luc-eGFP KRAS G12D cells. H, Tumor growth over time of rechallenged tumors in naïve mice or CRs. Mice that immune-rejected (IR) the re-injected cancer cells are indicated. Graph titles indicate the treatment that the primary tumor received. ****, P < 0.0001.

Supplementary Figure 5. Flow cytometry gating strategies

Supplementary Figure 9. Oncogenic KRAS drives tumor-intrinsic expression of COX-2 in LUAD

TME phenotyping via spectral flow cytometry

Supplementary Figure 7. Genetic loss of COX-2 signaling synergizes with ICB

Spectral flow cytometry staining panel

Combination strategies to enhance treatment responses

Supplementary Figure 4. COX-2 deficient tumors are sensitized to anti-tumor immunity

Supplementary Table S2. Antibodies used for FACS analysis of mouse tumors

Outline of image analysis pipeline

G12Ci-resistant subpopulation undergoes transcriptional changes in response to G12Ci and combinations. A, Schematic of the RNA-seq experiment. Mixed subcutaneous tumors (BFP KRAS G12C cells plus 7.5% Luc-eGFP KRAS G12D cells) were treated for 5 days with RMC-4998 (G12Ci) ± RMC-4550 (SHP2i). After treatment, cancer cells were sorted and bulk RNA-seq (bRNA-seq) was performed. B, Principal component analysis (PCA) across all samples and treatment groups; genotype of respective samples is indicated. PC, principal component. C, GSEA of top 10 differentially regulated Hallmark gene sets; vehicle set as baseline. NES, normalized enrichment score. D, Heatmaps for top 20 leading edge genes of the E2F target collection from the comparison between vehicle and G12Ci for KRAS G12C (left) and KRAS G12D cells (right). Z-scores normalized to the average of five vehicle samples (set to zero). E, GSEA of secondary treatment effects for KRAS G12C and KRAS G12D cancer cells. Second treatment on x-axis refers to baseline. Scoring as indicated in C. F, Expression of individual genes related to inflammatory response. Box plots indicate median and interquartile range. Statistical significance between vehicle- and inhibitor-treated samples was assessed using the pairwise Wilcoxon rank-sum tests (two-sided). G, Mixed subcutaneous tumors (BFP KRAS G12C cells plus 0.5% Luc-eGFP KRAS G12D cells) were treated for 6 days with RMC-4998 (G12Ci) ± RMC-4550 (SHP2i) or anti–PD-1 (aPD-1). KRAS G12D cells were sorted, and bulk RNA-seq was performed. GSEA of top 12 differentially regulated gene sets (MSigDB Hallmarks) for the comparison indicated in x-axis; the latter condition is used as the reference. Scoring as indicated in C. H, Mixed subcutaneous tumors with BFP KRAS G12C cells plus 0.04% of Luc-eGFP KRAS G12D cells expressing Ifngr wild-type (WT) or KO were treated for 2 weeks with RMC-4998 plus RMC-4550 (G12Ci + SHP2i) or anti–PD-1 (G12Ci + aPD-1). Bioluminescence imaging indicating relative abundance of KRAS G12D cells 18 days after treatment withdrawn (day 41 from treatment start). One-way ANOVA comparing WT vs. KO. *, P < 0.05; **, P < 0.01; ns, not significant.