KRASG12D drives immunosuppression in lung adenocarcinoma through paracrine signaling

Abstract

Lung cancer is the leading cause of cancer deaths in the United States. New targeted therapies against the once-deemed undruggable oncogenic KRAS are changing current therapeutic paradigms. However, resistance to targeted KRAS inhibitors almost inevitably occurs; resistance can be driven by tumor cell-intrinsic changes or by changes in the microenvironment. Here, we utilized a genetically engineered mouse model of KRASG12D-driven lung cancer that allows for inducible and reversible expression of the oncogene: activation of oncogenic KRASG12D induces tumor growth; conversely, inactivation of KRASG12D causes tumor regression. We showed that in addition to regulating cancer cell growth and survival, oncogenic KRAS regulated the transcriptional status of cancer-associated fibroblasts and macrophages in this model. Utilizing ex vivo approaches, we showed that secreted factors from cancer cells induced the expression of multiple cytokines in lung fibroblasts, and in turn drove expression of immunosuppressive factors, such as arginase 1, in macrophages. In summary, fibroblasts emerged as a key source of immune regulatory signals, and a potential therapeutic target for improving the efficacy of KRAS inhibitors in lung cancer.

Keywords: Cellular immune response; Lung cancer; Oncogenes; Oncology.

Figure 1. Inducible and reversible L-iKRAS^G12D mouse model of lung adenocarcinoma.

(A) Schematic depicting L-iKRAS^G12D mouse model and its inducibility and reversibility of KRAS^G12D expression in club cells of the lung with dox and activation of mutant Trp53 expression by ad-Cre. (B) Kaplan-Meier survival analysis comparing control mice (n = 20, Ccsp-rtTa or TetO-Kras^G12D with or without Trp53^–/+ allele and with and without dox or ad-Cre), Kras^G12D (n = 11, Ccsp-rtTa; TetO-Kras^G12D; Trp53^–/+ on dox, but no ad-Cre or Ccsp-rtTa; TetO-Kras^G12D on dox, with or without ad-Cre), and Kras^G12D/Trp53^R172H/+ (n = 23, Ccsp-rtTa; TetO-Kras^G12D; Trp53^LSL-R172H/+ on dox plus ad-Cre). Log-rank (Mantel-Cox) test with statistically significant P value of 0.0033. Median survival for all groups is indicated in inset. One-way ANOVA with Tukey’s post hoc test showed the median survival of the Kras^G12D/Trp53^R172H/+ group was significantly lower than that of the control group (P value: 0.0014) and the Kras^G12D group (P value: 0.0148). (C) Timeline for KRAS^G12D induction (ON) and KRAS^G12D inhibition (OFF) in triple-transgenic and control mice (single transgenics). (D) Western blot using anti-KRAS^G12D antibody or anti–total RAS (T-RAS) with corresponding β-actin blots as loading control of L-iKRAS lung tissue from all groups (ON: 20 weeks, OFF: 20 weeks ON and 4 weeks OFF). (E) Representative images of H&E. Scale bars: 50 mM. ON: 17–20 weeks, OFF 1 and 4 weeks. (F) Quantification of percentage tumor area over total lung area from whole slide scanned images in control, ON (17–25 weeks) and OFF (1, 2, and 4 weeks combined). (G) Representative images of Ki67/ECAD/DAPI. Scale bars: 25 mM. (H) Quantification of percentage Ki67⁺ cells among total ECAD⁺ cells. (I) Representative images of p-ERK/ECAD/DAPI. Scale bars: 25 mM. (J) Quantification of percentage p-ERK⁺ cells among total ECAD⁺ cells. (K) Representative images of CC3/ECAD/DAPI. Scale bars: 25 mM. (L) Quantification of percentage CC3⁺ cells among total ECAD⁺ cells. Data in F, J, H, and L are presented as mean ± SEM.

Figure 2. Oncogenic KRAS changes epithelial gene expression in the tumor microenvironment.

(A) UMAP visualization of scRNA-seq data showing unsupervised clustering of cells from L-iKRAS lung samples (KRAS ON = 21 weeks ON dox; KRAS OFF = 20 weeks ON dox + 1 week OFF). Each color represents a distinct cell cluster. (B) UMAP visualization of scRNA-seq data showing overlap of KRAS ON and KRAS OFF groups. (C) Bar graph comparing cell cluster breakdown per sample (red blood cells were removed). (D) UMAP visualization of defined lung epithelial clusters from KRAS ON and KRAS OFF samples. (E) Violin plots of Cxcl2, Wnt4, and Ccl4 comparing expression levels between total epithelial cells from KRAS ON and KRAS OFF samples. Adjusted P value given above each violin plot. Median expression is indicated with a horizontal line. (F) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in epithelial cells curated from the differential gene expression list. Genes from E are marked with an asterisk. (G) GSEA plot of KRAS ON versus KRAS OFF lung epithelia showing the running enrichment score for the “HALLMARK_KRAS_SIGNALING_DN” gene set. Normalized enrichment score (NES) = –1.430058344. Adjusted P value = 0.043658317.

Figure 3. Oncogenic KRAS changes fibroblast gene expression in the tumor microenvironment.

(A) Representative images of αSMA/PDGFR/DAPI. Scale bars: 25 mM. (B) Quantification of percentage αSMA⁺ area of total PDGFR⁺ area. (C) Percentage PDGFR⁺ area of total area. (D) UMAP visualization of scRNA-seq data showing unsupervised clustering of subsetted fibroblasts from KRAS ON and KRAS OFF samples. (E) Violin plots of Ccl4, Gas6, Cxcl2, and Timp3 comparing expression levels between total epithelial cells from KRAS ON and KRAS OFF samples. Adjusted P value given above each violin plot. Median expression is indicated with a horizontal line. (F) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in fibroblasts (pericytes removed) curated from the differential gene expression list. Genes from E are marked with an asterisk.

Figure 4. Oncogenic KRAS regulates myeloid compartment of the lung adenocarcinoma microenvironment.

(A) Representative images of MPO/F4/80/ECAD/DAPI. Scale bars: 25 mM. (B) Quantification of percentage F4/80⁺ area of total area. (C) UMAP visualization of scRNA-seq data showing unsupervised clustering of subsetted myeloid cells from KRAS ON and KRAS OFF samples. (D) Violin plots of Apoe, Mrc1, Csf1r, and C1qa comparing expression levels between macrophages from KRAS ON and KRAS OFF samples. Median expression is indicated with a horizontal line. (E) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in macrophages (combined macrophages, interstitial macrophages, and alveolar macrophages) curated from the differential gene expression list. Genes from D are marked with an asterisk. (F) Violin plots of Ccl3, Cd274, and Icam1 comparing expression levels between neutrophils from KRAS ON and KRAS OFF samples. Median expression is indicated with a horizontal line. (G) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in neutrophils from differential gene expression list. Genes from F are marked with an asterisk. (H) Violin plots showing expression of Egfr, Tgfbr1, and Cxcr2 across all identified cell populations in KRAS ON and KRAS OFF samples combined.

Figure 5. Identification of KRAS^G12D-dependent immunosuppressive secretome.

(A) Schematic depicting generation of LC3-547, an L-iKRAS cancer cell line from a murine lung tumor. (B) Representative Western blot depicting RAS^G12D and p53 protein expression in LC3-547 cells, which were cultured in dox-containing media for 24 hours prior to withdrawal of dox, and in A549 cells as controls. (C) Representative Western blot of RAS^G12D expression, p-ERK1/2, and vinculin in LC3-547 cells treated with 1 μM MRTX or 1 μM sotorasib (Soto) for 6 hours. (D) Depiction of the experimental outline for the collection of RNA and tumor-conditioned media (TCM) from LC3-547 cells treated with DMSO, MRTX, or Soto for subsequent analyses. All experiments were repeated at least 3 times, each time with 3 technical replicas. (E) qRT-PCR for Tgfa and Cxcl5 of LC3-547 cells treated with DMSO or 500 nM MRTX for 48 hours. Data are represented as mean ± SEM, and statistical significance was determined with a 2-tailed Student’s t test for unpaired samples. (F) Heatmap with z score of Luminex data of TCM from LC3-547 cells treated with 500 nM MRTX, 500 nM Soto, or equimolar concentration DMSO for 48 hours depicting cytokine and growth factor secretion. (G) Quantification (pg/mL) of indicated cytokines in TCM from treated LC3-547 cells. Data are represented as mean ± SEM. Statistical significance was determined using 1-way ANOVA with Tukey’s post hoc test. (H) Representative images of p-EGFR/ECAD/DAPI. Scale bars: 25 mM. (I) Quantification of percentage p-EGFR⁺ area of total ECAD⁺ area. (J) Representative Western blot of KRAS^G12D expression and p-ERK1/2 in human KRAS^G12D lung adenocarcinoma cell line, A427, upon 3 hours treatment with 1 μM MRTX or 1 μM Soto or equimolar DMSO. (K) Experimental design: RNA and TCM were harvested from human KRAS^G12D cancer cells cultured with DMSO, 100 nM MRTX, or 500 nM Soto. All experiments were repeated at least 3 times, each time with 3 technical replicas. (L) Quantification of CXCL1, CCL5, and TGF-α cytokines in A427 TCM from cells treated with DMSO, MRTX, or Soto. Data are represented as mean ± SEM and differences were evaluated by 1-way ANOVA with post hoc Tukey’s HSD test.

Figure 6. CAFs provide KRAS^G12D-dependent immunosuppressive paracrine signals.

(A) Schematic depicting derivation of normal lung fibroblasts, NLF-2522, from syngeneic L-iKRAS control mouse. (B) Experimental outline: lung fibroblasts were treated with TCM from ON, MRTX-treated LC3-547 cells, or DMEM alone as no-cell control. All experiments were repeated at least 3 times, each time with 3 technical replicas. (C) RNA was isolated from NLF-2522 fibroblasts incubated with TCM from ON or MRTX groups and expression of indicated genes was compared to DMEM-treated cells by qRT-PCR. Data are represented as mean ± SEM and statistical analysis was performed by 1-way ANOVA with post hoc Tukey’s HSD test. (D) TCM or DMEM was heated at 95°C–100°C for 10 minutes and subsequently added to NLF-2522 fibroblasts for 24 hours. Expression of Cxcl5 was assessed by qRT-PCR and compared to cells incubated with non-boiled TCM or DMEM. (E) Representative images of p-STAT3/αSMA/PDGFR/DAPI. Scale bars: 25 mM. (F) Quantification of p-STAT3⁺ cells as percentage of total PDGFR⁺ cells. (G) Experimental design of BMDMs cultured in conditioned medium (CM) from ON, OFF, MRTX-, and DMEM-treated fibroblasts. All experiments were repeated at least 3 times, each time with 3 technical replicas. (H) qRT-PCR for expression of M2 markers Retnla, Chil3, Cd274, and Arg1 in BMDMs treated with media from ON, OFF, and MRTX-treated L-iKRAS cells (TCM) or media from fibroblasts treated with ON, OFF, and MRTX-treated L-iKRAS cells (FB CM). Data are represented as mean ± SEM and statistical analysis was performed by 1-way ANOVA with post hoc Tukey’s HSD test.

Figure 7. scRNA-seq analysis of human lung adenocarcinoma recapitulates gene expression of distinct immunosuppressive markers from L-iKRAS epithelial and fibroblast secretomes.

(A) UMAP visualization of scRNA-seq data showing unsupervised clustering of cells from lung cancer samples. KRAS status: KRAS^G12D or wild-type KRAS (24). Each color represents a distinct cell cluster. (B) Interactome showing potential ligand/receptor pair interactions that are significantly (adjusted P value < 0.05) upregulated in KRAS^G12D-driven lung cancer samples. (C) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in epithelial cells curated from the differential gene expression list. Genes from D are marked with an asterisk. (D) Violin plots of CXCL1, CXCL2, and HDGF comparing expression levels between total epithelial cells from KRAS^G12D and KRAS–wild-type samples. Adjusted P value given above each violin plot. Median expression is indicated with a horizontal line. (E) Heatmap showing averaged scRNA-seq expression data (relative to the highest expressor) for genes in human macrophages (24) curated from the differential gene expression list. Genes from F are marked with an asterisk. (F) Violin plots of C1QB and C1QC comparing expression levels between total epithelial cells from KRAS^G12D versus KRAS–wild-type samples. Adjusted P value given above each violin plot. Median expression is indicated with a horizontal line.