Lymphatic-derived oxysterols promote anti-tumor immunity and response to immunotherapy in melanoma

Abstract

In melanoma, lymphangiogenesis correlates with metastasis and poor prognosis and promotes immunosuppression. However, it also potentiates immunotherapy by supporting immune cell trafficking. We show in a lymphangiogenic murine melanoma that lymphatic endothelial cells (LECs) upregulate the enzyme Ch25h, which catalyzes the formation of 25-hydroxycholesterol (25-HC) from cholesterol and plays important roles in lipid metabolism, gene regulation, and immune activation. We identify a role for LECs as a source of extracellular 25-HC in tumors inhibiting PPAR-γ in intra-tumoral macrophages and monocytes, preventing their immunosuppressive function and instead promoting their conversion into proinflammatory myeloid cells that support effector T cell functions. In human melanoma, LECs also upregulate Ch25h, and its expression correlates with the lymphatic vessel signature, infiltration of pro-inflammatory macrophages, better patient survival, and better response to immunotherapy. We identify here in mechanistic detail an important LEC function that supports anti-tumor immunity, which can be therapeutically exploited in combination with immunotherapy.

Fig. 1. LECs express Ch25h and control extracellular 25-hydroxycholesterol (25-HC) levels in lymphangiogenic B16F10-OVA tumors.

A B16F10-OVA VEGF-C tumors were inoculated in C57BL/6 mice. LECs were sorted by flow cytometry (CD45^negCD31⁺GP38⁺) from tumors, tumor-draining LNs (TdLN) and non-draining LNs (NdLN) after 11 days. Ch25h mRNA levels (RPKM) provided by RNA sequencing. n = 5 mice/group. Data were presented as mean values ± SD. B Correlation between Ch25h expression and lymphatic vessel (LV) signature (LV signature 1: pdpn, vegfc, lyve1, and LV signature 2: prox1, flt4, lyve1, pdpn, vegfc) in SKCM (Skin cutaneous melanoma) patients (TPM transcript per million). SKCM patient overall survival (OS) of high and low Ch25h expression. Data extracted from TCGA. C Predictive activity (OS) of Ch25h high and Ch25h low-expressors SKCM patients treated with anti-PD-1 and anti-CTLA-4, from an integrated dataset of multiple clinical trial studies. Significance was determined by log-rank analysis. D–F B16F10-OVA VEGF-C tumor cells were injected into Ch25h-GFP and WT mice. D, E Ch25h expression by tumor infiltrated cells on day 14. (DC dendritic cells, CAF cancer-associated fibroblasts, BEC blood endothelial cells). Results are representative of two independent experiments, with n = 4 mice/ group. F Ch25h expression by LECs in indicated organs at different time points. Results are pooled from two independent experiments, with n = 2–3 mice/group each. Two-way ANOVA, ^****P < 0.0001. G, H B16F10-OVA VEGF-C tumor cells were injected in LEC^ΔCh25h and LEC^WT mice. G Ch25h expression by LECs and BECs in tumors on day 11. Results are representative of two independent experiments, with n = 5 mice/group. D–G Ch25h expression is represented as MFI Ch25h-GFP – MFI-WT mice. H 25-HC levels in tumor interstitial fluid measured by liquid chromatography–mass spectrometry. Results are pooled from two experiments, with n = 4–13 mice /group. G, H Two-tailed unpaired t-test. ^*P < 0.05; ^****P < 0.0001. I VEGF-C expression in human melanoma cell lines by Q-PCR. Histograms depict technical triplicates from one experiment. E–I Data were presented as mean values ± SD. J, K Human melanoma VEGF-C^high T362C and VEGF-C^low T618A cells were injected in NSG (WT), NSG-LEC^ΔCh25h, and NSG-LEC^WT mice. Ch25h expression (MFI Ch25h-GFP-MFI-WT mice) by LECs from tumors and skin was assessed by flow cytometry. Data were presented as mean values ± SD. Results are representative of two independent experiments. L Ch25h mRNA and PDPN staining on human melanoma sections (n = 3 patients). Representative images are shown for patients 1 (Pt1) and 2 (Pt2). Scale bar, 20 and 5 µm (zoomed).

Fig. 2. Loss of Ch25h expression by tumor LECs enhances tumor growth by dampening anti-tumor immunity.

A–D B16F10-OVA VEGF-C cells were injected in LEC^ΔCh25h and LEC^WT mice (A, C, D), in NSG-LEC^ΔCh25h and NSG-LEC^WT mice, or in Rag2^KO-LEC^ΔCh25h and Rag2^KO-LEC^WT mice (B). A, B Tumor growth was followed and normalized to the size of tumors in LEC^WT, NSG-LEC^WT, and Rag2^KO-LEC^WT mice, respectively, at day 6 for tumor growth, two-way ANOVA, ^*P < 0.05. Data were presented as mean values ± SEM. Tumor cell proliferation (Ki67⁺) was evaluated by flow cytometry on day 11. Data were presented as mean values ± SD. Results are pooled from two independent experiments (n = 4–8 mice per group each) and normalized to WT (A), and Ch25h expression (MFI Ch25h-GFP – MFI-WT mice) was measured by flow cytometry at day 12 (B). C, D Tumors were harvested on day 11. C CD4⁺ and CD8⁺ T cells were analyzed for their expression of indicated markers. Representative flow cytometry dot plots and histograms provide the frequency of positive cells among T cell subsets. Data are presented as mean values ± SD. Results are pooled from two independent experiments (n = 4–8 mice/ group each) and normalized to WT. D Myeloid cells (CD11b⁺CD68⁺) were separated into monocytes (Ly6C⁺F4/80^-) and macrophages (Ly6C^-F4/80⁺) and were analyzed for their expression of indicated markers. Representative flow cytometry dot plots provide the frequency of positive cells among indicated cells. Data were presented as mean values ± SD. Results are representative of two independent experiments, n = 5–7 mice/group each. A–D For FACS analysis, two-tailed unpaired t-test. ^*P < 0.05; ^**P < 0.01; ^***P < 0.0001.

Fig. 3. Loss of Ch25h in LECs dampens immunotherapy efficacy in lymphangiogenic melanoma.

A–E LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C (B16-OVA-VC) cells and vaccinated with OVA protein and CpG-B at day 5 (A) or at day 4 and day 8 (B–E). A Ch25h expression (GFP MFI) was measured by flow cytometry on day 11. n = 5 mice/ group. B Tumor growth was followed and compared to unvaccinated B16F10-OVA VEGF-C tumor-bearing WT mice. Representative of four independent experiments, n = 5–6 mice/group each. C–E Tumor were harvested on day 24 and analyzed by flow cytometry. C CD45⁺/tumor cell ratio and tumor cell proliferation (Ki67⁺) and CD45⁺ cell frequency in living cells. D Myeloid cells (CD11b⁺CD68⁺) were separated into monocytes (Ly6C⁺F4/80^-) and macrophages (Ly6C^-F4/80⁺) and were analyzed for their expression of indicated markers. E CD4⁺ and CD8⁺ T cells were analyzed for their expression of indicated markers. C–E Representative of two independent experiments, n = 5–8 mice/group each. Data were presented as mean values ± SD (F, G) LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells and adoptively transferred 8 days later with OT-1 effector T cells. Tumor growth was followed and normalized to the size of WT at day 6 (F), and OT-1 cells were analyzed in tumors 2 days after transfer for indicated markers (G). Data were pooled from two independent experiments, n = 2–5 mice/group each, and are presented as mean values ± SEM (F) or as mean values ± SD (G). B, F Two-way ANOVA, ^*P < 0. 05. (A, C, D, E, G) Two-tailed unpaired t-test. ^*P < 0.05; ^**P < 0.01; ^***P < 0.0001.

Fig. 4. 25-HC treatment reverses tumor-associated macrophage immunosuppressive phenotype and functions in vitro.

A LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells and tumors were harvested at day 11. Montages of maximum projected 3D confocal images of representative sections (n = 3 mice/group) immunostained for lymphatic vessels (Lyve-1, green), T cells (CD8, red), myeloid cells (CD68, white), and nuclei (blue). Images were obtained using a 10x objective, including a 3x relative magnification. Selected regions of interest are indicated by dashed squares and denoted magnified areas are shown in images beside. Scale bars, 1 mm, 100 μm (zoomed). B BMDMs were incubated with B16F10-OVA VEGF-C tumor-conditioned medium (TCM) and treated or not with 25-HC for 48 h. BMDM phenotype was analyzed by flow cytometry. Data were representative of three independent experiments with n = 3–6 replicates each. Data were presented as mean values ± SD. C BMDM were treated or not with IFN-γ or IL-4 and further treated or not with 25-HC for 48 h. BMDM phenotype was analyzed by flow cytometry. Data were representative of two independent experiments with n = 4–8 replicates each. Data were presented as mean values ± SD. D TCM-exposed BMDM were treated or not with 25-HC for 48 h, washed, and incubated with OT-1 effector cells (5:1 ratio) for 48 h. T-cell phenotype was analyzed by flow cytometry. Data were representative of two independent experiments with n = 3–5 replicates each. Data were presented as mean values ± SD. B Two-tailed unpaired t-test. C Two-way ANOVA, D One-way ANOVA. ^*P < 0.05; ^**P < 0.01; ^***P < 0.001; ^****P < 0.0001.

Fig. 5. LEC-derived 25-HC modulates tumor growth by impacting tumor-associated macrophages and monocytes.

A LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells and vaccinated with OVA protein and CpG-B at day 4 and day 8. Mice were additionally injected with anti-CSF1R (1 mg / mouse at day 10, followed by 500 µg/mouse every 3 days or PBS), as represented in the scheme. Tumor growth was followed. Two-way ANOVA, **P < 0.01. Data were representative of two independent experiments with n = 5–7 mice/group each. B NSG-LEC^ΔCh25h and NSG-LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells mixed or not with BMDM. Tumor-bearing mice were further adoptively transferred with OT-1 effectors on day 8, as represented in the scheme. Tumor growth was followed and normalized to the size individually from day 6. Two-way ANOVA, **P < 0.01. Data were pooled from two independent experiments with n = 5–8 mice/group each. A, B Data were presented as mean values ± SEM.

Fig. 6. Altered monocyte and macrophage transcriptomes in the absence of Ch25h expression by LECs in lymphangiogenic melanoma.

A–C LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells and vaccinated with OVA protein and CpG-B on day 4 and day 8. Tumors were harvested on day 24, and monocytes and macrophages were sorted by flow cytometry (see Fig. 2D for gating strategy). RNA sequencing was performed (Illumina). A Top 10 Hallmark pathways expressed in monocytes and macrophages in tumors from LEC^ΔCh25h and LEC^WT mice. B Heatmap showing expression levels of genes implicated in “anti-tumor” or “pro-tumor” immune signatures in monocytes and macrophages and in tumors from LEC^ΔCh25h and LEC^WT mice. C GSEA enrichment of M1 and M2 signatures in monocytes and macrophages from tumor in LEC^ΔCh25h and LEC^WT mice using gene set from ref. . D B16F10-OVA VEGF-C tumor-bearing C57BL/6 mice were treated with 25-HC every 2 days, starting at day 4. Tumor growth was followed. Two-way ANOVA, *P < 0.05. Data were a pool of three independent experiments with n = 4–6 mice/group each. On day 11, tumors were harvested, and the CD68⁺CD11b⁺ myeloid cell (macrophage, F4/80⁺Ly6C^- and monocytes, F4/80^-Ly6C⁺) phenotype was assessed by flow cytometry. Ratio iNOS⁺/CD206⁺ cells are provided. Two-tailed unpaired t-test, ***p < 0.001. E Correlation between Ch25h expression and M1/M2 signatures in normal skin and in SKCM patients. Data were extracted from the TCGA database.

Fig. 7. LEC-derived 25-HC dampens the immunosuppressive functions of tumor-associated macrophages by inhibiting PPAR-γ expression.

A BMDM were incubated with B16F10-OVA VEGF-C tumor-conditioned medium (TCM), treated or not with 25-HC for 48 h, and analyzed by Q-PCR for PPAR-γ mRNA levels. Data were representative of two experiments with n = 2–3 replicates each. Data were presented as mean values ± SD. Two-tailed unpaired t-test. *P < 0.05. B BMDMs were incubated with tumor-conditioned medium (TCM), treated or not with 25-HC, Troglitazone, or the combination of 25-HC and Troglitazone for 48 h. BMDM phenotype was analyzed by flow cytometry. Data were representative of two independent experiments with n = 3 replicates each. Data were presented as mean values ± SD. One-way ANOVA. ***P < 0.001; ****P < 0.0001. C TCM-exposed BMDM were treated or not with 25-HC, Troglitazone, or the combination of 25-HC and Troglitazone for 48 h, washed, and incubated with OT-1 effector cells for 48 h. T-cell phenotype was analyzed by flow cytometry. Data were representative of two independent experiments with n = 6 replicates each. Data were presented as mean values ± SD. One-way ANOVA. *P < 0.05; ***P < 0.001; ***P < 0.0001. D LEC^ΔCh25h and LEC^WT mice were inoculated with B16F10-OVA VEGF-C cells mixed with either WT or PPAR-γ deficient (PPAR-γ^KO) BMDM and vaccinated with OVA protein and CpG-B on day 4 and day 8. Tumor growth was followed. Data are presented as mean values ± SEM. Two-way ANOVA, **P < 0.01. Data were representative of two independent experiments with n = 4–5 mice/group each.