All human and murine cell lines were cultured in RPMI-1640 media (Corning, Tewksbury, MA) supplemented with 5% fetal bovine serum (FBS) and 1% penicillin–streptomycin (Sigma-Aldrich, St. Louis, MO) at 37°C in a humidified 5% CO₂ incubator. All human cell lines used in this study were submitted to fingerprint analysis to confirm authenticity within a year of performing the studies described herein. A reference STR genotype was acquired through IDEXX Bioresearch for all murine lines and was later used to confirm genotypic integrity. CMT167 cells (24) and Lewis Lung Carcinoma cells (LLC) were obtained from the lab of Dr. Raphael Nemenoff. All cell lines were periodically tested for mycoplasma infection. To avoid cross-contamination and phenotypic changes, cells were maintained as frozen stocks and cultured for only two to four weeks before use in experiments. Authentication of cell lines based on morphology and growth curve analysis was performed regularly. No phenotypic changes were observed through the duration of the study.

B6;129S4-Kras^em1Ldow /J mice (Lukas Dow, Jackson Labs stock #033068) were crossed with B6.129P2-Trp53^tm1Brn /J mice (Jackson Labs stock #008462). F1 hybrid mice were genotyped for KRAS and p53 status using PCR on genomic DNA and the following primer pairs: KRAS^G12C, LSL-Kras Common 5’-TCCAATTCAGTGACTACAGATG-3’, LSL-Kras Mutant 5’-CTAGCCACCATGGCTTGAGT-3’, LSL-Kras Wildtype 5’-ATGTCTTTCCCCAGCACAGT-3’); Trp53, FWD 5’-GGTTAAACCCAGCTTGACCA-3’, REV-5’-GGAGGCAGAGACAGTTGGAG-3’). All PCR reactions employed PCRBIO VeriFi™ mix (PCRBIO, Wayne, Pennsylvania) according to manufacturer’s protocol with 60°C annealing temperature. Mice heterozygous for KRAS^G12C and varying p53 backgrounds (p53^+/+, p53^+/flox, and p53^flox/flox) were inoculated with Ad-Cre intratracheally and submitted to microCT imaging once monthly until tumor burden was detected. Tumors were harvested, minced, plated on tissue culture plastic and cultured until the stable cell line, mKRC.1, was obtained.

RNA was submitted to the University of Colorado Sequencing Core where library preparations were generated, and RNA was sequenced on the NovaSeq 6000 to generate 2 x 151 reads. Fastq files were quality checked with FastQC, illumina adapters trimmed with bbduk, and mapped to the mouse mm10 genome with STAR aligner. Counts were generated by STAR’s internal counter and reads were normalized to counts per million (CPM) using the edgeR R package. Differential expression was calculated using the limma R package and the voom function. Heatmaps were generated in Prism 9 (GraphPad Prism Software, San Diego, CA). Integrated Genomics Viewer analysis was completed by stripping section of sequence reads from RNAseq dataset and viewing in IGV software (25).

Fastq files were quality checked with FastQC, illumina adapters trimmed with bbduk, and paired reads were mapped to the UCSC mm10 BWA genome with samtools bwa-mem. Output sam files were converted to bam format, sorted, and duplicates were marked with the picard tool, MarkDuplicates. Samtools “mpileup” and bcftools “call” multiallelic caller were used to call variants and generate variant call files (VCF). To identify variants unique to each cell line, bcftools “isec” was used on each cell lines VCF files filtered with bcftools “view” excluding “G/T = 0/0” and including variants with at least a read depth of 10 and a quality score of 20 compared to unfiltered parental VCF files. Variants private to each cell line (001.vcf) were then annotated with ANNOVAR mm10db including refGene, cytoBand, and genomicSuperDups. VCFs were filtered excluding identified “intronic”, and “intergenic” annotated variants under the “Func.refGene tab in addition to “synonymous”, “.”, and “other” under the “ExonicFunc.refGene” tab. Variants were confirmed by interrogating specific regions from the BAM files and uploading those reads to the integrated genomics viewer (IGV). A variant was considered detected reliably if the variant was present in > 10% of the total read depth at the variant position. All code and the parameters used are not included in this report but can be provided upon request.

Cells were plated at 100 cells per well in 100 μL in 96-well tissue culture plates and allowed to attach for 24 hrs. Inhibitors were added at various doses as 2X concentrates in 100 μL. After incubation for 7-10 days, cell number per well was assessed using a CyQUANT Direct Cell Proliferation Assay (Invitrogen) according to the manufacturer’s instructions. The data are means and SEM (n=3) and presented as percent of the control values measured in DMSO-treated wells for each cell line.

Conditioned media was collected from treated and untreated murine lung cancer cell lines. Chemokine levels were measured using the Invitrogen ELISA Kit (Quantikine mouse/human CXCL10/IP-10; Invitrogen, Carlsbad, CA) following manufacturer’s instructions. Absorbance was measured at 450 nm. The measured concentration in each sample was normalized to the total cellular protein per dish and the data are presented as pg/μg protein.

Eight-week-old female C57B/6J mice were purchased from Jackson Labs (Bar Harbor, ME) and eight-week-old nu/nu mice were purchased from Envigo (Indianapolis, IN). For tumor cell inoculation, parental LLC, LLC 46 NRAS KO, LLC 23 NRAS KO and mKRC.1 cells were grown to 75% confluence, harvested, and resuspended in sterilized phosphate buffered saline. Cells were counted and resuspended to a final concentration of 125,000 cells per 40uL injection. Cell suspensions were directly injected into the left lung of mice through the ribcage. Mice were randomized into groups to receive diluent control, RMC-4550 (30 mg/kg; MedChem Express), MRTX-849 (30 mg/kg; MedChem Express), or the combination by daily oral gavage until primary experimental endpoints were met. Treatment was initiated following confirmation that at least 75% of mice had calculatable primary left lung tumors via microCT imaging (CUD Small Animal Imaging Core). MicroCT imaging was conducted weekly and ITK-SNAP software (26) was used to calculate cross-sectional tumor volumes in cubic millimeters. Mice exhibiting signs of morbidity according to the guidelines set by the Institutional Animal Care and Use Committee (IACUC) were sacrificed immediately.

Statistical Analysis: Prism 9 (GraphPad Software, San Diego, CA) was used to perform specific statistical analyses noted in the figure legends. Data are presented as the mean and standard error of the mean (SEM) as indicated. To calculate IC₅₀ values, the dose-response data were submitted to the “log(inhibitor) vs. normalized response” function with the Prism 9 software. An unpaired Student’s t test (two-tail) was used to determine statistical significance, unless otherwise noted. The P values are denoted by *(P < 0.05), **(P < 0.01), ***(P < 0.001), and ****(P < 0.0001) and were corrected for multiple comparisons (Dunnett). Drug synergy with combinations of KRAS-G12C inhibitors and SHP2 inhibitors was determined with Combenefit, a free software tool for visualization, analysis and quantification of drug combination effects. Data from drug combination assays was processed using the HSA synergy model.

The literature supports a role for innate and adaptive immunity in the overall tumoral responses to oncogene-specific inhibitors (16, 22, 27, 28). Herein, we generated three novel murine KRAS^G12C-driven LUAD cell lines to permit investigation of KRAS^G12C inhibitor responses in immune-competent hosts.

The sensitivity of the murine KRAS^G12C lung cancer cell lines to the KRAS^G12C inhibitors MRTX-1257 and AMG-510 as well as the MEK1/2 inhibitor, trametinib was assessed with in vitro growth assays (see Materials and Methods). Knockout of NRAS increased the sensitivity of LLC NRAS KO cell lines to both KRAS^G12C inhibitors (AMG-510, 14-34 fold; MRTX-1257, 82-148 fold) compared to parental LLC cells ( Figure 1A ), further supporting that mutated NRAS-Q61H significantly contributes as an oncogenic driver in these cells. This finding is consistent with that reported previously in the 3LL derivative of LLC cells (34). LLC NRAS KO cell lines exhibited similar sensitivity to trametinib as parental LLCs ( Figure 1A ), indicating that NRAS-Q61L serves as an additional proximal driver of MAPK signaling in LLC cells. Compared to parental CMT cells which are completely insensitive to MRTX-1257 and AMG-510 at concentrations of 3 nM and 300 nM, respectively, the CMT KRAS^G12C-engineered cell lines were highly sensitive to these agents with IC₅₀ values in the low nanomolar range ( Figure 1B ). Interestingly, parental CMT cells bearing KRAS^G12V were less sensitive to trametinib than the CMT KRAS^G12C clones ( Figure 1B ). This finding may also reflect the disruption of the non-edited KRAS-G12V allele as a result of the CRISPR editing resulting in a single functional KRAS-G12C gene in the CMT-54 and CMT-55 clones. The reduced KRAS copy number may also contribute to their decreased proliferation rates observed in vitro ( Suppl. Figure S3C ). Finally, the mKRC.1 cell line exhibits sensitivity to MRTX-1257 and AMG-510 ( Figure 1C ). RNAseq data from the parental and engineered cell lines was interrogated for KRAS mRNA levels. LLC 46 NRAS KO cells exhibited 1.4-fold higher KRAS mRNA levels relative to parental LLC cells (p = 0.04, one-way ANOVA), but no other significant differences in KRAS mRNA expression were noted ( Figure 1E ). In addition to cell growth, mRNA expression levels of the MAP kinase pathway target genes (38, 39) ETV4, DUSP6 and FOSL1 measured by RNAseq were markedly reduced in LLC 46 NRAS KO, CMT-KRAS-G12C.54.10 and mKRC.1 cells after 1 to 2 days of MRTX-1257 treatment ( Figure 1F ). Evidence for pathway re-activation was evident by 3 days of MRT-1257 treatment, especially in LLC 46 NRAS KO cells.

Analysis of MRTX-1257 and AMG-510-sensitivity in a panel of 13 human KRAS^G12C-mutant lung cancer cell lines revealed IC₅₀ values for these two drugs that ranged from 0.1 to 356 nM for MRTX-1257 and 0.3 to 2534 nM for AMG-510 ( Supplementary Figures S4A, B ). The inhibitor sensitivity exhibited by the murine KRAS^G12C cell lines ( Figure 1 ) overlapped with the most sensitive human lung cancer cell lines bearing KRAS^G12C, indicating that co-occurring mutations present in many of the human lines markedly reduce their sensitivity to MRTX-1257 and AMG-510 similar to the activity of GTPase-deficient NRAS in parental LLC cells. Overall, the murine LUAD cell lines exhibit sensitivity to KRAS^G12C inhibitors that are consistent with human KRAS^G12C mutant cell lines.

In a pilot study, LLC 46 NRAS KO cells were propagated as orthotopic tumors in the left lung of C57BL/6 mice and treated with diluent or the MRTX-1257 clinical-grade analogue, MRTX-849 (30 mg/kg daily). The results revealed a modest reduction in the rate of growth by single agent KRAS^G12C inhibitor (data not shown) and indicate that combination therapy with MRTX-849 will be required to achieve clinically significant anti-tumor responses. In this regard, the protein tyrosine phosphatase, PTPN11/SHP2, has emerged as a promising upstream target as it functions to dephosphorylate multiple proteins leading to sustained KRAS GTP cycling and activation (11–13, 40, 41), and highly specific inhibitors are available. SHP2 also plays a role in negatively regulating JAK-STAT signaling through dephosphorylation of activated STAT proteins (42) and T cell receptor (TCR) signaling within lymphocytes resulting in decreased T cell effector function (14, 15). Both JAK-STAT signaling, and adaptive immunity and T cell function have been shown to be necessary for oncogene-targeted therapy response (14–17).

To test the effect of the SHP2 inhibitor RMC-4550 in combination with KRAS^G12C inhibition, we plated LLC 46 NRAS KO, CMT-KRAS-G12C.54 and mKRC.1 cell lines in 96-well plates and assessed cell growth in response to a range of concentrations of MRTX-1257 and RMC-4550 alone and in combination. Analysis of the resulting data (see Suppl. Figure S5 for primary data) with Combenefit (43) revealed strong in vitro synergy with combinations of MRTX-1257 and RMC-4550 in these cell lines ( Figure 2A ). Two human KRAS-G12C-positive lung cancer cell lines, Calu1 and H2030, were similarly submitted to growth assays with combinations of MRTX-1257 and RMC-4550. Analysis of the data with Combenefit indicated synergistic drug interactions in these human cell line models as well ( Suppl. Figure S6 ). Also, combination of these inhibitors yielded greater inhibition of pERK in mKRC.1 and LLC 46 NRAS KO over time compared to either single agent alone ( Figure 2C ), suggesting this combination induces synergistic growth inhibition, in part, through prevention of ERK reactivation. To assess if the MRTX-1257/RMC-4550 combination exerts cytostatic or cytotoxic effects, LLC 46 NRAS KO, CMT-KRAS-G12C.54 and mKRC.1 cells were seeded in 24-well plates and treated with DMSO diluent, 10 nM MRTX-1257, 300 nM RMC-4550 or the combination. Cell numbers were directly counted and the results ( Figure 2B ) suggest that the drug combination induces a cytostatic, but not a cytotoxic effect in the three cell lines. Also, immunoblot analysis of full-length PARP and cleaved PARP, a biochemical measure of apoptosis, revealed modest induction of PARP cleavage in LLC46 cells with single-agent MRTX-1257, but not further increased by combining with RMC-4550 ( Figure 2C ). No PARP cleavage was noted in mKRC.1 cells treated with either drug alone or in combination.

Based on the synergistic activity of the combination of MRTX-1257 and RMC-4550 in vitro ( Figure 2A ), the drug combination was tested in LLC 46 NRAS KO and mKRC.1 cells using an orthotopic lung tumor model (30). Following inoculation of LLC 46 NRAS KO cells into the left lung of C57BL/6 mice, tumors were assessed for growth over time using μCT imaging. Representative μCT images from this experiment are presented in Supplemental Figure S7 . Treatment with single agent RMC-4550 or MRTX-849 resulted in marginally reduced rate of tumor growth compared to diluent control ( Figure 3A ). However, the combination of the two agents yielded initial tumor regression and suppressed tumor growth until tumors ultimately progressed despite continued treatment ( Figure 3A ). Assessment of the individual tumor responses at day 21 (11 days of treatment) among the experimental groups revealed no significant change in tumor volume with single agents compared to diluent control mice, although modest improvement in overall survival was observed ( Figures 3B, C ). By contrast, four of the eight mice treated with the combination of MRTX-849 and RMC-4550 experienced tumor shrinkage and significantly improved survival compared to diluent control and each single agent alone ( Figures 3B, C ). In contrast to the LLC 46 NRAS KO model, orthotopic lung tumors established with the mKRC.1 cell line exhibited a significant single-agent anti-tumor response to MRTX-849 ( Figure 4A and representative μCT images in Suppl. Figure S8 ), although progression on continued drug treatment was observed within 18 days. RMC-4550 was ineffective as a single agent, but combination of MRTX-849 and RMC-4550 yielded durable tumor shrinkage to less than 10% of initial tumor volume. Assessment of individual tumor responses to 14 days of treatment with MRTX-849 alone or in combination with RMC-4550 is shown in Figure 4B and demonstrates shrinkage of all five of the combination-treated tumors relative to one of four in the MRTX-849 monotherapy group. When the daily MRTX-849 plus RMC-4550 treatment was terminated after 25 days, the residual tumor cells rapidly progressed over the subsequent 2 weeks, indicating that the combination treatment did not eliminate residual cancer cells in the mKRC.1 tumors.

To explore the role of adaptive immunity in contribution to the tumor response to KRAS^G12C inhibition alone or in combination with SHP2 inhibition, mKRC.1 cells were propagated as orthotopic tumors in the left lungs of athymic nu/nu mice. The findings in Figure 5A show that the MRTX-849 response (fold of initial volume after a 14-day treatment) in C57BL/6 and nu/nu hosts was 1.2 ± 0.3 and 7.6 ± 2.9, respectively. Similarly, LLC 46 NRAS KO cells were inoculated into the left lung of athymic nu/nu mice and tumors were allowed to establish. Subsequently, tumor-bearing nu/nu mice were treated with diluent control, RMC-4550, MRTX-849, or the combination. While either drug alone reduced tumor growth similar to the result in C57BL/6 mice ( Figure 3A ), the combination failed to yield any additional benefit ( Figure 5B ). This was further supported by assessing percent tumor volume change in the individual mice, as the combination treatment group was not different than the individual tumors within the single agent groups ( Figure 5C ). These results suggest a role for adaptive immune cells as participants in the MRTX-849 response in mKRC.1 tumors and the enhanced response to the MRTX-849 and RMC-4550 combination in LLC 46 NRAS KO tumors.

Previous studies have demonstrated that KRAS-SHP2-MAP kinase pathway-targeted agents induce expression of multiple chemokines in KRAS-mutant lung cancer cell lines (21, 44, 45) and may provide a mechanism by which oncogene-targeted drugs induce cross-talk between cancer cells and the tumor microenvironment. Interrogation of RNAseq data generated from LLC 46 NRAS KO, CMT-KRAS-G12C.54 and mKRC.1 cells treated in vitro for 6 hr to 3 days with 30 nM MRTX-1257 reveals distinct expression patterns of chemokines with defined anti- and pro-tumorigenic functions (46) ( Figure 6A ). LLC 46 NRAS KO cells constitutively expressed Ccl2 and Csf1 and MRTX-1257 modestly induced Ccl28 mRNA. By contrast, CMT-KRAS-G12C.54 cells exhibited low levels of the chemokines surveyed, but Cxcl2, Cxcl5 and Ccl28 mRNAs were induced upon MRTX-1257 treatment. Inspection of RNAseq data from mKRC.1 cells revealed constitutive expression of Cxcl1 as well as baseline expression of Csf2 that was reduced upon MRTX-1257 treatment. In addition, Cxcl2, Cxcl5, Cxcl10, and Csf1 were induced upon MRTX-1257 treatment. Among these secreted chemokines, Cxcl10 is defined as anti-tumorigenic through recruitment of T cells (47) and was strongly induced in mKRC.1 cells relative to LLC 46 NRAS KO and CMT-KRAS-G12C.54 cells. To confirm this enhanced expression in mKRC.1 cells, secreted levels of CXCL10 were measured by ELISA in conditioned medium obtained from LLC 46 NRAS KO and mKRC.1 cells treated for 2 days with DMSO, MRTX-1257, RMC-4550 and the combination of the two drugs. The combination of MRTX-1257 and RMC-4550 stimulated significant, but modest induction of CXCL10 protein compared to single agent MRTX-1257 or RMC-4550 in LLC 46 NRAS KO cells ( Figure 6B ). Moreover, mKRC.1 cells exhibited increased CXCL10 secretion to single agent MRTX-1257 or RMC-4550 and when combined, further increased levels of CXCL10 protein relative to the single agents ( Figure 6B ).