Increased fatty acid delivery by tumor endothelium promotes metastatic outgrowth

Abstract

Metastatic outgrowth in distant microscopic niches requires sufficient nutrients, including fatty acids (FAs), to support tumor growth and to generate an immunosuppressive tumor microenvironment (TME). However, despite the important role of FAs in metastasis, the regulation of FA supply in metastatic niches has not been defined. In this report, we show that tumor endothelium actively promotes outgrowth and restricts antitumor cytolysis by transferring FAs into developing metastatic tumors. We describe a process of transendothelial FA delivery via endosomes that requires mTORC1 activity. Thus, endothelial cell-specific targeted deletion of Raptor (RptorECKO), a unique component of the mTORC1 complex, significantly reduced metastatic tumor burden that was associated with improved markers of T cell cytotoxicity. Low-dose everolimus that selectively inhibited endothelial mTORC1 improves immune checkpoint responses in metastatic disease models. This work reveals the importance of transendothelial nutrient delivery to the TME, highlighting a future target for therapeutic development.

Keywords: Cancer immunotherapy; Endothelial cells; Immunology; Metabolism; Oncology.

Figure 1. Raptor/mTORC1 loss in endothelium reduces metastatic outgrowth in the lung.

(A) Schematic of experimental procedures of tumor cell inoculation, tamoxifen treatment, and harvest. (B–E) WT (n = 6) or Rptor^ECKO (n = 8) male mice were inoculated with LLC-GFP-luc cells and treated with tamoxifen, as described in A. (B) Representative bioluminescence image from day 14 is shown. Scale bar shows counts. Total radiance flux was normalized to WT controls and presented as fold change (FC). (C) Representative GFP (left) and gross (right) lungs after harvest on day 18. Scale bar: 5 mm. Visible tumor area is outlined by yellow line. (D) Lung weights were recorded in grams (g) at harvest and (E) GFP intensity was calculated as arbitrary units (a.u.). (F–H) WT (n = 8) or Tsc2^ECKO (n = 6) female mice were inoculated with E0771-luc tumor cells as described in A. (F) A representative bioluminescence image is shown from day 14. Scale bar shows counts. Total radiance flux was normalized to WT controls. (G) Representative lungs after harvest on day 20 are shown. Scale bar: 5 mm. Visible tumor area is outlined by yellow line. (H) Lung weights were recorded in grams (g) at harvest. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001 by unpaired, 2-tailed Student’s t test.

Figure 2. Long-chain fatty acids are reduced upon loss of Raptor/mTORC1 in endothelial cells.

Metabolomics was performed on primary microvascular endothelial cells isolated from Rptor^fl/fl mice transduced with control (WT) or Cre-recombinase (Rptor-KO) adenoviruses. Cells were collected 24 hours after infection (n = 6 per group). (A and B) Summary of significantly altered metabolites by (A) major class and (B) lipid classes in Rptor-KO versus WT endothelial cells. (C) Heatmap of lipid metabolites in WT and Rptor-KO endothelial cells. Columns represent individual samples and rows are metabolites. Long-chain fatty acids (LCFAs) and long-chain polyunsaturated fatty acids (PUFAs) are indicated. (D and E) Normalized intensities (log₂ + 1) of representative (D) LCFA and (E) PUFA metabolites. The median along with the 25th and 75th percentile hinges are indicated within the box. The whiskers indicate minimum and maximum values within each group. The mean is shown as a plus sign (“+”). *P < 0.05 by 2-tailed Welch’s t test.

Figure 3. Endothelial Raptor/mTORC1 supports transendothelial delivery of fatty acids.

(A) Schematic of transendothelial transport assay. (B) Representative images of BODIPY-C16 (green) in LLC tumor cells after 1 hour. VEGFR1-Fc was used in control samples to bind tumor cell–derived VEGF-B. Endothelial cells act as a physical barrier for BODIPY-C16 access, demonstrated in a representative image from an endothelial cell–free control (“No ECs”). A control performed without tumor cells (“No TCs”) is shown to confirm removal of endothelial cells prior to fluorescence imaging of basolateral tumor cells. Scale bar: 100 μm. (C) BODIPY-C16 fluorescence in LLC tumor cells, normalized to WT plus VEGFR1-Fc control (n = 3 per group). (D) Representative confocal images of WT or Rptor-KO endothelial cells treated with BODIPY-C16 for 5 minutes. Nuclear (Hoechst, blue) and actin (phalloidin, red) staining was used to detect perinuclear and cell boundaries, respectively. Total (all z-planes) and basolateral (bottom 10% of z-planes) BODIPY-C16 (green) staining are also shown, with the basolateral BODIPY presented at higher exposure (HE). Scale bar: 20 μm. BODIPY-C16 intensities at the basolateral surface were normalized to total signal in each cell. WT (n = 16) and Rptor-KO (n = 10) cells were analyzed from 2 independent experiments. (E and F) WT (n = 4) or Rptor^ECKO (n = 5) mice were inoculated with LLC tumor cells as described in Figure 1A. One hour prior to tumor harvest, animals were injected with BODIPY-C16 (50 μg). Median fluorescence intensity (MFI) was determined by flow cytometry in (E) CD45^–CD31⁺ endothelial cells and (F) CD45^–EpCAM⁺FSC^hi tumor cell–enriched populations and normalized to littermate WT controls. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001 by 2-way ANOVA with Tukey’s post hoc test (B) or unpaired, 2-tailed Student’s t test (D–F).

Figure 4. Raptor/mTORC1 loss reduces vesicle trafficking of fatty acids in endothelial cells.

(A and B) Gene set enrichment analysis of WT or Rptor-KO endothelial cells (n = 4 per group). (A) Top enriched pathways are shown. Normalized enrichment score (NES) and false discovery rate (FDR) q values are indicated. (B) Enrichment plots for RAB trafficking pathways are shown. (C) Immunofluorescence of RAB5 (left) or RAB7 (right) (both red) was performed on WT or Rptor-KO endothelial cells (n = 3 per group). Representative images are shown. Nuclei were stained with DAPI (blue). Scale bars: 100 μm. (D) Gene expression volcano plot from data in A and B. Differentially upregulated genes in Rptor-KO cells are displayed in red, while downregulated genes are in blue. (E) Transendothelial transport of BODIPY-C16 in WT or Rptor-KO endothelial cells transfected with siRNA against Clstn1 (n = 3 per group). BODIPY-C16 intensity in LLC tumor cells was normalized to the non-targeting control (NTC) and presented as fold change (FC). (F) LLC tumor cells were cultured in fractionated conditioned media from WT or Rptor-KO endothelial cells (n = 4 per group). *P < 0.05, **P < 0.01, ****P < 0.001 by unpaired, 2-tailed Student’s t test (C), 2-way ANOVA with Tukey’s post hoc test (E), or 2-way ANOVA with Šidák’s multiple-comparison post hoc test (F).

Figure 5. Targeting endothelial mTORC1 reduces fatty acid uptake and improves antitumor immunity to reduce lung metastatic outgrowth.

(A) BODIPY-C16 MFI of CD8⁺ T cells from LLC metastatic tumors in WT (n = 4) or Rptor^ECKO (n = 5) male mice. (B) GZMB⁺CD8⁺ T cells in E0771-luc lung metastatic tumors from WT (n = 5) or Rptor^ECKO (n = 5) female mice. (C) Correlation of the mTORC1^ECKO and cytotoxic T lymphocyte (CTL) gene signatures in TCGA-BRCA. (D) Breast cancer recurrence-free survival (RFS) in the TCGA BRCA (n = 1218) dataset, stratified by mTORC1^ECKO gene signature (ECKO GS). Number of at-risk patients in each group is shown. HR is 0.6679 (95% CI = 0.5732–0.7782). (E–G) Pharmacological mTORC1 inhibition combined with αPD-1 immunotherapy. Control (n = 5), αPD-1 (n = 6), RAD001 (n = 6), RAD001/αPD-1 (n = 5). (E) Schematic of experimental procedures. (F) Bioluminescence of lung tumors on day 17. (G) GZMB⁺CD8⁺ T cells were determined by flow cytometry and normalized to Control. (H) Proposed model of long-chain fatty acid (LCFA) transport across endothelial cells into tumor tissue. (a) VEGF-B activates mTORC1, leading to (b) trafficking of LCFAs via RABs. (c) CLSTN1 expression (d) promotes anterograde transport of LCFA-filled cargos. (e) Released LCFAs are taken up by cancer cells and CD8⁺ T cells, (f) suppressing cytotoxicity. *P < 0.05, ***P < 0.005 by unpaired, 2-tailed Student’s t test (A and B) or 1-way ANOVA with Dunnett’s post hoc test (E–G).