Tumor-derived exosomes drive immunosuppressive macrophages in a pre-metastatic niche through glycolytic dominant metabolic reprogramming

Abstract

One of the defining characteristics of a pre-metastatic niche, a fundamental requirement for primary tumor metastasis, is infiltration of immunosuppressive macrophages. How these macrophages acquire their phenotype remains largely unexplored. Here, we demonstrate that tumor-derived exosomes (TDEs) polarize macrophages toward an immunosuppressive phenotype characterized by increased PD-L1 expression through NF-kB-dependent, glycolytic-dominant metabolic reprogramming. TDE signaling through TLR2 and NF-κB leads to increased glucose uptake. TDEs also stimulate elevated NOS2, which inhibits mitochondrial oxidative phosphorylation resulting in increased conversion of pyruvate to lactate. Lactate feeds back on NF-κB, further increasing PD-L1. Analysis of metastasis-negative lymph nodes of non-small-cell lung cancer patients revealed that macrophage PD-L1 positively correlates with levels of GLUT-1 and vesicle release gene YKT6 from primary tumors. Collectively, our study provides a novel mechanism by which macrophages within a pre-metastatic niche acquire their immunosuppressive phenotype and identifies an important link among exosomes, metabolism, and metastasis.

Keywords: NF-kB; PD-L1; exosomes; glycolysis; immunosuppression; lactate; metastasis.

Figure 1.. The effect of TDE on macrophages in a pre-metastatic niche.

(A) Schematic for experimental design (B-E). (B) Micro-metastatic burden quantified by LLC-GFP% in the lungs of s.c. GFP-LLC tumor-bearing mice treated i.v. with MLE-12 or LLC exosomes. Representative dot plots of GFP expression in the lungs and summarized data of LLC-GFP% are shown. Each dot represents data from one mouse. *p<0.05, (one-way ANOVA with multiple comparisons). (C) Confocal microscopic analysis of lung GFP⁺ micro-metastases from primary s.c. LLC tumor-bearing mice. (D) Confocal microscopic analysis of PD-L1 expression in LLC Exo⁻ or LLC Exo⁺ treated lungs (left) and PD-L1 co-expression with CD11b and F4/80 in LLC Exo⁺ mice (right). Scale bar=100μm. (E) Micro-metastatic burden quantified by LLC-GFP% in the lungs of s.c. GFP-LLC tumor-bearing mice treated i.v. with LLC or Pan02 Exo.. Representative dot plots of GFP expression in the lungs (left) and summarized data of LLC-GFP% in the lungs (middle) and tissue specific LLC-GFP% of LLC Exo. treated mice (right) are shown. *p<0.05, **p<0.01, ***p<0.001 (one-way ANOVA with multiple comparisons). (F) Percentage of CD11b⁺F4/80⁺PD-L1⁺ IM in the lungs of control and exosome treated mice. (G) Expression of PD-L1 by flow cytometry in peritoneal macrophages stimulated with MLE-12 Exo. or LLC Exo.. Representative histograms and summarized MFI are shown (n=3). MFI= mean fluorescent intensity. ***p<0.001, ****p<0.001, one-way ANOVA with multiple comparisons. (H) Representative western blot showing increased PD-L1 expression in macrophages treated with LLC Exo. for 16 hours. (I) PD-L1 expression by flow cytometry on macrophages treated with exosomes from 4-T1, MC38, and B16-F10 cell lines. n=3. (J) The mRNA expression of PD-L1 in macrophages stimulated with LLC Exo. with pre-treatment of actinomycin D (ACT-D). n=3. Data was normalized to control expression level. The data are shown as mean ± SEM. **p<0.01, ***p<0.001, ****p<0.001 (unpaired student’s t test). See also Figure S1.

Figure 2.. TDE polarize macrophages towards an immunosuppressive phenotype.

(A) Representative cytokine array panels of supernatants taken from 16-hour culture of peritoneal macrophages stimulated with media alone (top), MLE-12 exosomes (middle), or LLC exosomes (bottom). Representative plots and summarized mean pixel density of each protein are shown (n=3). (B) IL-6 and IL-10 concentration detected by ELISA from supernatants of macrophages stimulated with MLE-12 and LLC exosomes (n=3). (C) The mRNA expression levels of Arg-1, VEGF, and iNOS in peritoneal macrophages treated alone or with MLE-12 or LLC exosomes (n=3). (D) LLC exosome polarized macrophages suppress OT-I T-cell proliferation (top) and IFN-γ production (bottom) in a PD-L1 dependent manner. Representative histograms and summarized data are shown (n=3). The data are shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001 (one-way ANOVA with multiple comparisons). See also Figure S2.

Figure 3.. TDE mediate PD-L1 expression through TLR2 and NF-κB.

(A) The expression of PD-L1 by flow cytometry on WT or MyD88^−/− peritoneal macrophages stimulated with LLC Exo.. (B) Expression of PD-L1 via flow cytometry in WT or TLR2^−/− macrophages following stimulation with LLC Exo. (left). Macrophages from TLR4^−/−, TLR6^−/−, TLR7^−/−, and TLR9^−/− mice (n=3) showed increased PD-L1 expression similar to WT following TDE stimulation (right). (C) Summary of GFP⁺ micrometastases in the lungs of TLR2^−/− mice that had previously been injected s.c. with GFP-LLC tumor cells and then treated i.v. with MLE-12 or LLC exosomes. Each dot represents data from one mouse. (D) Comparison of GFP⁺ micrometastases between WT and TLR2^−/− LLC Exo. treated mice. Each dot represents data from one mouse. (E) Activation of NF-κB signaling cascade in macrophages following stimulation with MLE-12 or LLC Exo. at the indicated time determined by WB analysis. (F) PD-L1 mRNA expression level in macrophages pre-treated with the NF-κB inhibitor, BAY-11-7082 prior to stimulation with LLC Exo. Data was normalized to the control (n=3). (G) PD-L1 expression in macrophages stimulated with rmHMGB-1. Representative histograms and summarized MFI data are shown (n=5). **p<0.01 (unpaired student’s t test). (H) Comparison of PD-L1 expression between WT and MyD88^−/− macrophages following stimulation with rmHMGB-1. Representative histograms and summarized MFI are shown (n=4). (I) PD-L1 expression in TLR2^−/− macrophages compared to controls following rmHMGB-1 stimulation. Summarized MFI data is shown (n=3). (J) HMGB-1 expression in control siRNA or HMGB-1 siRNA transfected LLC cells assessed by Western blot. (K) PD-L1 expression on macrophages stimulated with TDE from control or HMGB-1 siRNA transfected LLC cells. Representative histogram and summarized data are shown (n=3). (L) Expression of HMGB-1 in either shControl or shHMGB-1 primary 4T1 s.c. tumors (left), H&E staining of lung (middle), and summarized number of metastatic tumor nodules (right) in the lungs. Scale bar=100μm. **p<0.01 (unpaired student’s t test). The data are shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001. See also Figure S3.

Figure 4.. TDE inhibit oxidative phosphorylation via NO.

(A) Uptake of 2-NBDG in LLC or MLE-12 Exo. stimulated macrophages. Representative histograms and summarized data are shown (n=3). (B) PD-L1 expression on macrophages stimulated with LLC Exo. in the presence or absence of 1mM 2-DG. Representative histogram and summarized data are shown (n=3). (C) The mRNA expression levels of GLUT-1, HIF-1α, LDHA, PDK1, HK-1, and IDH1 by RT-PCR in macrophages. Data normalized to control expression level (n=3). (D) Representative western blots showing expression of GLUT-1, LDHA, HK-1, HIF-1α and PDK1 in macrophages treated with MLE-12 Exo. or LLC Exo.. (E) Seahorse glycolysis stress test with sequential addition of glucose, oligomycin, and 2-DG in macrophages pre-treated with MLE-12 or LLC Exo. for 16 hours. Data are representative for n=4 independent experiments. (F) OCR measurement in TDE stimulated macrophages. Data are representative of three independent experiments with similar results. (G) OCR Bioenergetic profiling showing relative values of parameters for representative Seahorse assay in Figure 4F. (H) Intracellular NOS2 expression via flow cytometry in macrophages stimulated with MLE-12 or LLC Exo.. Representative histogram and summarized data are shown (n=3). (I) OCR in macrophages treated with LLC Exo. in the presence of 2mM or 4mM SEITU (left). ECAR in the presence of SEITU (2mM) treated macrophages as measured by glycolysis stress test (right). Data are representative of two independent Seahorse Mito stress tests and two glycolytic stress test experiments each with similar results. (J) Expression of PD-L1 on macrophages treated with LLC Exo. in the presence or absence of SEITU (4mM). Representative histogram and summarized MFI data are shown (n=3). ***p<0.001 by unpaired student’s t-test. (K) Intracellular NOS2 expression via flow cytometry in WT or MyD88^−/− macrophages stimulated with LLC Exo.. n=4. (L) Intracellular NOS2 expression in HIF-1α^f/f and LysM-cre;HIF-1α^f/f macrophages upon LLC Exo. stimulation. (M) Micro-metastatic burden quantified by LLC-GFP% in the lungs of s.c. GFP-LLC tumor-bearing WT or NOS2^−/− mice treated with LLC Exo. (n=5 mice per group). Representative dot plots and summarized data are shown. (N) PD-L1 expression on lung IMs of WT and NOS2^−/− LLC Exo. treated mice (n=5 mice per group). The data are shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.00, ****p<0.0001 (one-way ANOVA with multiple comparisons). See also Figure S4.

Figure 5.. Glycolytic metabolism fuels lactate production and PD-L1 expression.

(A) Levels of L-lactate in the supernatants of macrophages treated with MLE-12 or LLC Exo. (n=3). (B) Level of L-lactate in the supernatant of WT or MyD88^−/− macrophages stimulated with LLC Exo. (n=3). (C) L-lactate levels in the supernatant of macrophages pre-treated with BAY-11-7082 prior to stimulation with or without LLC Exo. (n=3). (D) Supernatants from HIF-1α^f/f and LysM-cre;HIF-1α^f/f peritoneal macrophages treated with LLC Exo. were collected and L-lactate concentration was determined (n=6). (E) The mRNA expression levels of MCT-4 on macrophages treated with MLE-12 or LLC Exo. (n=3). Expression levels were normalized to the control. (F) Summarized PD-L1 expression via flow cytometry on macrophages stimulated with 20mM L-lactate (n=3). *p<0.05 (unpaired student’s t test). (G) PD-L1 expression on macrophages treated with lactate in the presence of the MCT-1 inhibitor AZD3965 (250nM) (n=3). (H) Intracellular phosphorylated NF-κBp65 (pNF-κBp65) expression following stimulation with L-Lactate for the indicated time. Representative histograms and summarized MFI data are shown (n=3). (I) Confocal translocation assay demonstrating increased co-localization of NF-κBp65 to the nucleus following 30 minutes stimulation with L-lactate. Scale bar=10μm. (J) PD-L1 expression on macrophages pretreated with BAY prior to stimulation with or without L-lactate. Representative histogram and summarized MFI data are shown (n=3). (K) mRNA levels of PD-L1 in macrophages pretreated with BAY prior to stimulation with or without L-lactate (n=4). (L) Schema depicting TDE-induced PD-L1 expression via direct NF-κB transcription and NF-κB mediated glycolytic metabolism. The data are shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (one-way ANOVA with multiple comparisons). See also Figure S5.

Figure 6.. In vivo assessment of TDE induced glycolytic and PD-L1 phenotype.

(A) Schematic for experimental design (B-D). (B) Radioactivity measured by Biodex Atomlab^™ 500 following intraperitoneal injection of ¹⁸FDG in the lungs of C57BL/6 mice treated with control MLE-12 Exo. or LLC Exo. Each dot represents data from one mouse. (C) ¹⁸FDG uptake in the macrophage enriched population from the lungs of MLE-12 Exo. or LLC Exo. treated mice. **p<0.01 via Mann-Whitney U test. (D) Flow cytometry expression of PD-L1 and mRNA expression of HIF-1α and MCT-4 in the macrophage enriched population. (E) Primary tumor volume 28 days post injection with SCRAM or Rab27a^−/− 4T-1 tumor cells. (F-G) CCR2 and PD-L1 expression on CD11b⁺ IM in the lungs of SCRAM versus RAB27^−/− tumor bearing mice. (H) Percentage of CD8⁺ T cells in the lung. (I-J) Percentage and PD-1 expression of CD4⁺ T cells in the lung. Representative dot plots and summarized data are shown. The data are shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001 (unpaired student’s t test). See also Figure S6.

Figure 7.. The effect of human TDE on the formation of a pre-metastatic niche.

(A) PD-L1 expression via flow cytometry on healthy donor sorted CD14⁺ monocytes (n=3 donors) stimulated with control HBEC or A549 Exo.. Representative histogram and summarized data are shown. *p<0.05, **p<0.01, (one-way ANOVA with multiple comparisons). (B) PD-L1 expression on CD14⁺ cells stimulated with exosomes from MCF-7 (left) or S2-013 (right). Representative histograms and summarized data are shown. n=3. ***p<0.001, ****p<0.0001 by unpaired student’s t-test. (C) The expression of CD206 and PD-L1 on CD68⁺ macrophages in the LNs of non-cancerous healthy donor (HD) lung transplants versus T1–3N0 NSCLC patients. (D) PD-L1 expression in CD123⁺ pDCs and cDCs compared to CD206⁺ macrophages in dLNs of NSCLC patients (n=6). Representative contour plots and summarized data are shown. **p<0.01, ***p<0.001 (one-way ANOVA with multiple comparisons). (E) Pearson correlation between CD206/PD-L1⁺ macrophages and CD8⁺ PD-1⁺ T cells within the dLN. (F) PCA plot of the first two principal components (“Dim1” and “Dim2”), depicting the relationships of NSCLC (red triangles) and HD (blue circles) subjects. The percent variance explained by each principal component is indicated in parentheses. The 95% confidence ellipses and centroids (enlarged symbols) for each group of samples (NSCLC and HD) are also shown (left). Boxplot depicting the values for the first principal component, partitioned by disease state (NSCLC vs. HD), as determined by PCA (right). ****p<0.0001 generated by linear regression. (G) GLUT-1 expression in CD206^hiPD-L1^hi versus CD206^loPD-L1^lo CD68⁺ macrophages obtained from dLN samples. Representative histograms and summarized data are shown. (H) Pearson correlation between CD206/PD-L1 and GLUT-1 expression in CD68⁺ macrophages in NSCLC dLN. (I) Boxplot showing YKT6 expression in patients with negative nodal staging (N0) compared to positive nodal staging (N1-N3) in LUAD patients from the TCGA database. Plots were made with Deseq2’s function plotCounts with count normalized to library size. Positive (N1-N3) n=85, Negative (NX/N0) n=156 (J) Kaplan-Meier survival curve generated by Oncolnc for TCGA dataset LUAD patients based on expression profile (High= top 25%) and (low=bottom 25%) of YKT6. n=123 per group. **p<0.01 (log rank test). (K) YKT6 mRNA expression levels in CD45⁻ and CD45⁺ cells sorted from primary NSCLC tumor samples. n=8. (L) YKT6 mRNA expression levels in NSCLC patient primary tumors grouped by CD206/PD-L1 expression level (high >10% CD68⁺CD206⁺PD-L⁺ and low <10% CD68⁺CD206⁺PD-L⁺ expression) in CD68⁺ macrophages within the dLN. n=4 patients per group. The data are shown as mean ± SEM. **p<0.01, ***p<0.001 (unpaired student’s t test). See also Figure S7.