EGFR-driven lung adenocarcinomas coopt alveolar macrophage metabolism and function to support EGFR signaling and growth

Abstract

The limited efficacy of currently approved immunotherapies in EGFR-driven lung adenocarcinoma (LUAD) underscores the need to better understand alternative mechanisms governing local immunosuppression to fuel novel therapies. Elevated surfactant and GM-CSF secretion from the transformed epithelium induces tumor-associated alveolar macrophage (TA-AM) proliferation which supports tumor growth by rewiring inflammatory functions and lipid metabolism. TA-AM properties are driven by increased GM-CSF-PPARγ signaling and inhibition of airway GM-CSF or PPARγ in TA-AMs suppresses cholesterol efflux to tumor cells, which impairs EGFR phosphorylation and restrains LUAD progression. In the absence of TA-AM metabolic support, LUAD cells compensate by increasing cholesterol synthesis, and blocking PPARγ in TA-AMs simultaneous with statin therapy further suppresses tumor progression and increases proinflammatory immune responses. These results reveal new therapeutic combinations for immunotherapy resistant EGFR-mutant LUADs and demonstrate how cancer cells can metabolically co-opt TA-AMs through GM-CSF-PPARγ signaling to provide nutrients that promote oncogenic signaling and growth.

Figure 1.. Alveolar Macrophages accumulate during lung tumorigenesis, become increasingly tolerogenic, and promote LUAD growth.

(A) Schematic of lung adenocarcinoma induction and immune profiling in a genetically inducible EGFR^L858R mouse LUAD model. Tumors were initiated via feeding LUAD mice (Ccsp-rtTA; TetO-EGFR^L858R) or littermate control ‘Healthy’ mice (TetO-EGFR^L858R or Ccsp-rtTA) doxycycline (DOX) in chow diet. Unless otherwise noted, mice were analyzed 6–8 weeks on DOX. (B) Serial MRI of a mouse after 4 and 6 or 7 weeks on DOX. (C) Immune infiltrates were quantified by flow cytometry during disease initiation (~2 weeks), at emergence of macroscopic disease (~4 weeks) and with fully established disease (~6 weeks). Alveolar macrophages (AMs) were defined as CD45⁺CD11b⁻SigF⁺CD11c⁺ and interstitial macrophages (IMs) were defined as CD45⁺AM⁻CD11b⁺Ly6G⁻. CD4⁺ and CD8⁺ T cells were gated on CD45⁺CD3⁺ cells. (D) as in C, but AMs were stimulated with LPS and T cells were stimulated with PMA and ionomycin for 4.5 hrs prior to intracellular TNF staining. (E-F) After 6–8 weeks on DOX, AMs were isolated from LUAD mice (red) or littermate controls (black) and amounts of the (E) inhibitory receptor CD200R or (F) checkpoint ligand PD-L1 were measured by flow cytometry based on mean fluorescent intensity (MFI). (G) Bulk RNA-sequencing of AMs from late stage EGFR^L858R lung tumors were analyzed for expression of several pro-inflammatory cytokines (n=3). Data from Ayeni et al. 2019 (40). (H-I) LUAD mice were administered clodronate liposomes 2x weekly retro-orbitally at weeks 4–6 of DOX and then sacrificed at week 6 of Dox. Vehicle treated (red) or clodronate treated (blue) lungs were further analyzed. (H) AM frequency was measured by flow cytometry and (I) tumor burden was assessed by the dry lung weight. (J) AM proliferation rates were assessed by Ki67-staining and flow cytometry. (K) GM-CSF was measured in healthy and LUAD lung lysates using ELISA. (L) AM surface expression of GM-CSFR was measured using flow cytometry. (M) Representative flow plots showing intracellular GM-CSF and Ki67 in healthy and tumor-associated AT2 cells (CD45⁻Epcam⁺CD31⁻MHCII^hiproSPC⁺, comprised of tumor cells (majority of cells) and non-transformed AT2 cells) incubated with BFA for 4.5 hours. (N-P) GM-CSF blocking (blue) or isotype control mAbs (red) were administered twice weekly (0.5mg/mouse) intra peritoneal to LUAD mice during weeks 4–7 on DOX. (N) We then measured the number of TA-AMs present in the BALF, (O) tumor burden using lung dry weight as a surrogate, and (P) AM TNF secretion following LPS stimulation. Data shown are mean ± SEM, and statistical analysis were performed by two-tailed unpaired Student’s test (E-P). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are representative from ≥3 experiments (C,D,M) or pooled from ≥3 experiments (E-F,H-L,N-P) with each group containing 3 (C,D), 8–7 (E), 10 (F), 6–5 (H), 6 (I-J,L), 7 (K), or 7–6 (N-P) mice.

Figure 2.. Lung surfactants accumulate in the TME despite increased levels of AM lipid uptake

(A) Bar graphs show SP-D concentrations in BALF collected from patients undergoing diagnostic bronchial alveolar lung lavages (BAL) for lung cancer (n=5), COPD (n=3), or mycobacterial infection (n=2) as measured by ELISA. (B) Concentration of SP-D and SP-A from the BALF of littermate control (WT Heathy, black) mice or those with LUAD (WT Tumor, red) as measured by ELISA. (C) Bar graph shows the fold change in surfactant proteins mRNAs from EGFR^L858R lung epithelium (EPCAM⁺) in LUAD lungs. (D) SPA^−/− SPD^−/− double knockout mice were crossed with CCSP-rtTA;TetO-EGFR^L858R mice and placed on DOX for 6–8 weeks and dry lung weight was measured. (E-F) Similar to (A-B), heatmaps show (E) lipids in the BALF from patients with lung cancer (red), COPD (black), or mycobacterial infection (gray) (F) or from littermate control (WT Heathy, black) mice or those with LUAD (WT Tumor, red) as measured by LC/MS. Heatmaps depict the relative abundance of each class of lipid species normalized to volume (shown as a row Z score). Human data depict abundance averaged amongst the samples. (G-H) Import of free fatty acids, cholesterol and phospholipids (Bodipy C12, NBD cholesterol, DPPE respectively) were compared between AMs isolated littermate controls (H-AM, black) and LUAD mice (TA-AM, red) using flow cytometry and displayed as (G) representative histograms of lipid import and (H) cumulative bar graphs of MFI. (I-K) H-AMs (black) and TA-AMs (red) were isolated 6–8 weeks on DOX and rates of basal mitochondrial respiration was measured using (I) Seahorse Flux Analyzer or cells were labeled with (J) ¹³C-palmitate or (I) ¹³C-glucose and measured for rates of fatty acid oxidation and other metabolites, respectively. Data shown are mean ± SEM, and statistical analysis were performed with a two-tailed unpaired Students test (A-C, H-K) or a two way ANOVA (D). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are pooled from ≥3 experiments with each group containing 6 (B), 7–9 (D), 5–13 (H) animals or is representative from ≥2 experiments with each group containing 3 (G,I-K) animals. Human samples were collected from 3 (COPD), 2 (mycobacterial infection) or 5 (lung cancer) and data was pooled (A) or averaged (D).

Figure 3:. AM PPARγ is required for the accumulation and phenotype in the TME

(A) Violin plots depict PPARG mRNA expression in macrophage subsets from scSeq from patients with NSCLC (red) or matched healthy adjacent tissue (black). Patients are segregated by WT EGFR and EGFR-mutant tumors (included EGFR^Del19 and EGFR^L858R mutations). Data and cell types from GSE131907. (B) MFI of PPARγ in various myeloid subsets in paired samples from NSCLC tumors or healthy adjacent tissue from the same patient measured by CYTOF analysis (data from Lavin et al. 2017 (50)). The following gating strategy was used: AMs: CD11b⁺CD64⁺CD163⁺CD206^hi, IMs: CD11b⁺CD64⁺CD163⁻, CD14⁺Monocytes: CD11b⁺CD64⁻CD14⁺, and CD16⁺ Monocytes: CD11b⁺CD64⁻CD16⁺. (C-G) LUAD mice with deletion of PPARγ in macrophages (Pparγ^Fl/Fl; Csf1r-CRE, Ccsp-rtTA; TetO-EGFR^L858R)(referred to as PPARγ^−/− Tumors, blue) and littermate controls (Pparγ^Fl/F;Ccsp-rtTA; TetO-EGFR^L858R) (referred to as WT Tumors, red) were placed on DOX for 7–9 weeks at which point tumor burden and immune infiltrates were examined by scRNA-seq, microscopy and flow cytometry. Healthy lungs (black) were from control TetO-EGFR^L858R mice on DOX. (C) Bar graph of dry lung weights. (D) Immunofluorescence microscopy was performed on lung sections from WT Tumor (red) and PPARγ^−/−Tumor (blue) measuring density of tumor cells (EGFR^L858R), AMs (F4/80) and nuclei (dapi). (E) Heatmap shows Z-score by row of PPAR-target gene expression (from scRNA-seq) in the AM cluster from healthy (H, black) lungs or those with LUAD from WT (T, red) and PPARγ^−/− (KO, blue) Tumors. (F-H) Percentage of (F) total AMs, (G) frequency of immature AMs (CD45⁺CD11c⁺SigF⁺CD11b⁺) within the AM compartment (H) and (I) MFI of PD-L1 on AMs as assessed by flow cytometry from the three groups of mice. (J-K) H-AMs (black) or TA-AMs from WT (red) and PPARγ^−/− (blue) Tumors were isolated and stimulated with LPS to measure TNF production by flow cytometry. (L) Spheroids from human EGFR^del19 cell line HCC827 were cultured on low attachment plates for a week and then AMs isolated from LUAD from WT (red) or PPARγ^−/−(blue) Tumors were added to cultures along with GM-CSF (20 pg/mL) for three days and proliferation was measured by ki67 staining. (M-P) LUAD mice and littermate controls were placed on DOX for four weeks and then treated intra tracheal 5x/week with the PPARγ antagonist GW9662 (1mg/kg) (blue) in corn oil or vehicle alone for three more weeks (red). (M) Representative MRI images and (N) quantification of tumor burden in untreated and antagonist treated lungs. (O) Percentage of AMs and those (P) producing TNF after LPS stimulation were assessed by flow cytometry. Data shown are mean ± SEM, and statistical analysis were performed with a two-tailed unpaired Students test (D,L,N-P) or a two way ANOVA (C,F,G,I,K). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are pooled from ≥3 experiments with each group containing 13 (B) patients and 4 (D), 17–21 (F), 8–9 (G), 12 (I), 8–10 (K) 10–12 (L), 10–15 (N,O) and 10–14 (P) mice.

Figure 4:. PPARγ rewires Alveolar macrophage metabolism within the TME.

LUAD mice lacking PPARγ in macrophages (PPARγ^−/− Tumors or KO Tumors (blue)) and littermate ‘WT Tumor’ controls (PPARγ^Fl/F; CCSP-rtTA; TetO-EGFR^L858R(red)) along with ‘WT Healthy’ lung controls (TetO-EGFR^L858R or Ccsp-rtTA (black)) were placed on DOX for 7–9 weeks. (A) Heatmap depicts the relative abundance (averaged across 3 samples/group) of the indicated lipid species in AMs normalized to the total cell number as measured by LC/MS. (B) WT and PPARγ^−/− KO TA-AMs were compared by scRNA-seq to identify the top 10 differentially upregulated pathways using gene set enrichment analysis. (C) Heatmap shows expression of fatty acid metabolism genes by row Z-score (from scRNA-seq) in WT AMs from Healthy lungs (black) vs. WT (red) and PPARg^−/− (KO, blue) TA-AMs from LUAD lungs. (D-H) AMs isolated from WT Healthy lungs (black) and TA-AMs isolated WT (red) or PPARγ^−/− (blue) Tumors were incubated with (D) [¹³C]-palmitate, (E) [¹⁴C]-acetate, (F) [¹³C]-palmitate, or (I) [¹³C]-glucose and assayed for (D) free fatty acid (FFA) import, (E) FA synthesis, (F) FA b-oxidation (G) or relative flux into several metabolites. (H) Basal mitochondrial respiration in AMs was performed using seahorse extracellular flux assay. (I) Heatmap shows expression of cholesterol synthesis and metabolism genes by row Z-score (from scRNA-seq) in WT AMs from Healthy lungs (black) vs. WT (red) and PPARg^−/− (KO, blue) TA-AMs from LUAD lungs. (J) AMs isolated from WT Healthy lungs (black) and TA-AMs isolated WT (red) or PPARγ^−/− (blue) Tumors were incubated with [¹⁴C]-acetate and sterol synthesis was assessed by incorporation of radio label into the sterol fraction. (K) TA-AMs were incubated with NBD cholesterol for 30 mins to look at rates of cholesterol import by flow cytometry. (L) To assess cholesterol efflux, the labeled TA-AMs and H-AMs were then equilibrated overnight in serum free media followed by incubation with FBS as a cholesterol acceptor for 4 hours and effluxed NBD-Cholesterol in the supernatant was measured by fluorescence. (M) ABCG1 expression on AMs was assessed by flow cytometry. (N) TA-AMs were incubated with NBD cholesterol for one hour and examined by fluorescent microscopy to look at lipid droplets. Data shown are mean ± SEM, and statistical analysis were performed with a two way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are representative from 2 experiments with an N=3 (A). Data are pooled from ≥2 experiments with each group containing 6 (D), 5 (E, F), 3 (G-H),5 (J), 11–12 (K), 5–11 (L) animals.

Figure 5:. GM-CSF PPARγ signaling drives cholesterol efflux from TA-AMs to tumor cells to promote tumor growth

(A-D) WT and PPARγ^−/− TA-AMs were stimulated overnight in the presence or absence of GM-CSF (20 pg/mL) and assessed for (A) PD-L1 expression by flow cytometry (B), rates of fatty acid synthesis (C), cholesterol efflux, or (D) cholesterol esterification. Data were normalized to untreated samples with a dashed line at 1. In (C), cells were also treated with a cholesterol esterification inhibitor SOAT1 to increase efflux. (E) Diagram depicting the experimental schema of TA-AM cholesterol transfer experiments wherein TA-AMs were loaded with fluorescent NBD-cholesterol, washed and then co-cultured with dissociated tumor-associated AT2 cells overnight. (F) Transfer of fluorescently labelled cholesterol from TA-AMs to tumor-associated AT2 cells was assessed by flow cytometry and percentage of AT2 cells with imported NBD-cholesterol is shown in bar graph. (G) Heatmap shows expression of cholesterol synthesis and metabolism genes by row Z-score (from scRNA-seq) in AT2 cells isolated from healthy lungs (black) or LUAD lungs that contain WT (red) or PPARγ^−/− TA-AMs (blue). (H) Rates of sterol synthesis in sorted tumor-associated AT2 cells isolated from LUAD lungs that contain WT (red) or PPARγ^−/− TA-AMs (blue). Data shown are mean ± SEM, and statistical analysis were performed with a two-tailed unpaired Students test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are pooled from ≥2 experiments with each group containing 4 (A), 3–5 (B), 6 (C), 2 (D) or 6–7 (F) 8 (G,H) mice.

Fig 6:. AM PPARγ rewires AT2 metabolism to support oncogenic signaling

(A) Heatmap shows expression of EGFR signaling genes by row Z-score (from scRNA-seq) in AT2 cells isolated from healthy lungs (black) or LUAD lungs that contain WT (red) or PPARγ^−/− TA-AMs (blue). (B-C) Lungs from same samples as in (A) were stained with mAbs to phospho-EGFR (pEGFR) and the AT2 cell marker pro-SPC and analyzed by (B) confocal microscopy. Data are representative of sections taken from four separate mice. (C) The intensity of pEGFR staining in AT2 was quantified using IMARIS. (D-E) The EGFR^Del19 mutant human cell line HCC827 was cultured for 1–3 days in the absence or presence of 10 μM atorvastatin and (D) each day the viable cell number was enumerated and (E) the amounts of pEGFR relative to total EGFR were assessed by Western blotting. (F-K) LUAD lungs containing WT (red) or PPARγ^−/− (blue) were treated by oral gavauge with pravastatin (0.5 mg, 5X/week) from weeks 3–7 on DOX. (F-H) Tumor burden was assessed using: (F) Representative histological sections staining for H&E are depicted for each tumor group; (G-H) IHC for the EGFR^L858R oncogene is depicted along with quantification of the frequency of oncogene positive cells and their density amongst lung sections; (I) and by dry lung weight. (J-L) Intracellular cytokine staining and flow cytometry was used to assess the frequency of TNF-producing AMs (I), CD4⁺ and CD8⁺ T cells (J, K) after stimulation with LPS (I) or PMA and ionomycin (J, K). Data shown are mean ± SEM, and statistical analysis were performed with a two-tailed unpaired Students test (D,G,H-L) or a two-way ANOVA (C). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are representative of ≥3 mice (B,C) and are pooled from ≥3 experiments with each group containing 50–149 cells (C) and 5–14 (F,G), 10–13 (H), 9–12 (J), 5–7 (I), 9–10 (K-L) mice.