TECs exhibit elevated autophagy levels. A-C: Representative immunofluorescence images showing the expression of autophagy-related protein LC3B (red) in CD31⁺ vessels (green) in tumor tissues from patients with hepatocellular carcinoma (HCC), glioblastoma (GBM), and non-small cell lung cancer (NSCLC). Scale bars = 20 μm. n = 3 patients per group. D: Dot plot visualization of autophagy-related gene expression in NECs and TECs. The x-axis represents the tissue source (NEC or TEC), whereas the y-axis lists the autophagy-related genes. Dot size indicates the percentage of samples in which each gene was detected, and dot color represents the average gene expression level across samples. The color scale ranges from blue (low expression) to red (high expression). E: Comparison of autophagy gene expression scores between NECs and TECs, revealing a statistically significant increase in autophagy-related gene expression in TECs (P-value = 1.9e-15). F-K: Representative immunofluorescence images showing LC3B (red) in endothelial cells (indicated by arrows) within both adjacent normal and tumor tissues from patients with colon and gastric cancer. Box plots (H and K) quantifying the number of LC3B-positive puncta in CD31⁺ cells from adjacent normal (A) and tumor (T) tissues in colon and gastric cancer samples. Box plots display the maximum and minimum values, medians, and 25th/75th percentiles. P values were calculated using two-tailed Student's t-tests. ****P < 0.0001. Scale bars = 50 μm. n = 3 patients per group.

Endothelial autophagy is preferentially induced in dysfunctional tumor vessels. A: Pathway enrichment analysis based on the DEGs (top 500 highly expressed genes) in endothelial cells with high- versus low-autophagy gene scores. B: Dot plot comparing the expression of autophagy-related genes in normoxic (HIF-1α-negative) versus hypoxic (HIF-1α-positive) TECs. The color scale indicates expression levels, whereas dot size reflects the percentage of samples in which the gene was detected. C and D: Multiplex immunofluorescence analysis of autophagic activity in the vasculature of hypoxic and normoxic regions in ICC tissues. CA-IX (magenta) labels hypoxic regions, CD31 (red) identifies endothelial cells in the tumor vasculature, and LC3B (green) indicates autophagic activity. The zoomed-in section highlights a specific region comparing LC3B density in the tumor vasculature between the hypoxic and normoxic areas. The corresponding mask image shows the precise localization of the LC3B puncta (autophagic vesicles) within the zoomed region. The right panel (D) presents a box plot quantifying LC3B puncta density (puncta/μm²) in TECs from hypoxic and normoxic regions, showing significantly higher autophagic activity in hypoxic areas (*P < 0.05). Scale bar = 100 μm. E and F: In vivo 3D imaging of autophagy in the zebrafish tumor xenograft vasculature. Flk-GFP (green) visualizes the vessel structure, while LC3-mCherry (red) indicates autophagy in normal brain vessels, as well as functional tumor vessels and dysfunctional tumor vessels in CT26 cells at 4 dpi. Quantification of autophagic puncta showed a significant increase in dysfunctional tumor vessels compared to that in normal and functional vessels (***P < 0.001). Scale bar = 50 μm. G: Discrimination between autophagic and perfused endothelia in B16-F10 melanoma mouse tumor xenografts. Double immunostaining for CD31 (red) and LC3B (white) was used to label the endothelial cells and autophagy, respectively, with DAPI (blue) counterstaining of the nuclei. Perfused vessels were detected using fluorescein isothiocyanate (FITC)-dextran angiography (green). Autophagic endothelium (LC3B⁺ CD31⁺) was predominantly observed in non-perfused tumor vessels, with minimal overlap between FITC-dextran-positive (perfused) vessels and autophagic cells. Scale bar = 100 μm.

Autophagy inhibitors can selectively prune dysfunctional tumor vessels, optimizing the structure and function of the tumor's vascular network. A: Experimental timeline for evaluating the effects of autophagy inhibitors on zebrafish CT26 tumor vasculature. Fluorescently labelled tumor cells were injected into zebrafish 72 h post-fertilization (hpf), followed by screening 12 h post-injection (hpi). Treatment with autophagy inhibitors (chloroquine, 50 μmol/L; ULK-101, 4 μmol/L; MRT68921, 50 μmol/L) was administered from 1 d post-injection (dpi) to 5 dpi, with subsequent assessment of tumor vessels using confocal microscopy. B and C: Confocal 3D projection images of CT26 tumor vessels in the control and autophagy inhibitor-treated groups, with vessels highlighted in green. Blood vessel diameters were quantified using ImageJ software. Quantitative graph showing blood vessel diameter comparisons among the different groups (C). Quantitative data are shown as the mean ± SD. D and E: Confocal images of transgenic zebrafish expressing the endothelial marker flk:GFP and erythrocyte marker Gata1:dsRed in CT26 tumor vessels for the control and autophagy inhibitor-treated groups. Boxplot graph showing the persistence of blood flow quantified as a percentage of the luminated area. Boxplot showing the maximum and minimum values, medians, and 25/75 percentiles. F and G: Confocal images showing injection of low-molecular-weight FITC-dextran (20 kDa) in CT26 tumor vessels for the control and autophagy inhibitor-treated groups, with tumor vessels labelled in red and dextran in green. Bar graph depicting FITC signal intensity within tumor vessels. Quantitative data are shown as the mean ± SD. P values (versus control) were generated using one-way analysis of variance, followed by Tukey's post-hoc tests for multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars = 50 μm.

Endothelial autophagy blockade remodels tumor vasculature and enhances perivascular immune niche formation in Atg7^iECKO mice. A: Schematic of the experimental timeline for the B16-F10 lung metastasis model in Atg7^iECKO mice. On day 2 after the intravenous injection of B16-F10 cells, tamoxifen (20 mg/ml in corn oil) was administered intraperitoneally (2 mg/20 g body weight) for five consecutive days to induce endothelial cell-specific Atg7 knockout. Tumor tissues were harvested 21 d post-transplantation. B: Representative mIF images of tumor sections from control and Atg7^iECKO mice, stained for CD31 (endothelial cells, red), α-SMA (pericytes, green), CD4 (helper T cells, white), CD8 (cytotoxic T cells, yellow), and DAPI (nuclei, blue). Right panel: Enlarged view showing T cell aggregates in the perivascular immune niche of Atg7^iECKO tumor tissue. Scale bar = 100 μm. C: Quantification of functional tumor vessels, defined as the percentage of CD31⁺ vessels co-stained with α-SMA. Data are presented as the mean ± SD (n = 4 mice per group; **P < 0.01, Student's t-test). D-G: Quantification of CD4⁺ and CD8⁺ T cell density in the tumor stroma and the number of these cells within a 10-μm radius around tumor blood vessels. Data are shown as box plots displaying the minimum, maximum, median, and 25/75 percentiles (n = 4 mice per group; *P < 0.05, **P < 0.01, Student's t-test). H: Schematic of the experimental timeline for the B16-F10 subcutaneous tumor model in Atg7^iECKO mice. On day 2 following the subcutaneous injection of B16-F10 cells, tamoxifen (20 mg/ml in corn oil) was administered intraperitoneally (2 mg/20 g body weight) for five consecutive days to induce endothelial cell-specific Atg7 knockout. Tumor tissues were collected on day 21 post-transplantation. I: Representative mIF images of tumor sections from control and Atg7^iECKO mice, stained for CD31 (endothelial cells, red), α-SMA (pericytes, green), CD4 (helper T cells, white), CD8 (cytotoxic T cells, yellow), and DAPI (nuclei, blue). Right panel: Enlarged view showing T cell aggregates in the perivascular immune niche of Atg7^iECKO tumor tissue. Scale bar = 100 μm. J: Quantification of functional tumor vessels, defined as the percentage of CD31⁺ vessels co-stained with α-SMA. Data are presented as the mean ± SD (n = 4 mice per group; **P < 0.01, Student's t-test). K-N: Quantification of CD4⁺ and CD8⁺ T cell density in the tumor stroma and the number of these cells within a 10-μm radius around tumor blood vessels. Data are shown as box plots displaying the minimum, maximum, median, and 25/75 percentiles (n = 4 mice per group; *P < 0.05, **P < 0.01, Student's t-test).

Autophagy inhibitor (SBI-0206965) combined with anti-mouse PD-1 (CD279) induces perivascular immune niches formation and improves the TiME. A: Schematic representation of the treatment regimen used in the CT26 murine subcutaneous tumor model. Mice were treated with the autophagy inhibitor SBI-0206965 (10 mg/kg, intraperitoneally, daily starting from day 7) and an anti-PD-1 antibody (5 mg/kg, intraperitoneally, every 3 d starting on day 11 post-tumor inoculation). Tumor tissues were harvested on day 21 for further analyzes. B: Representative mIF images of tumor sections from the control, PD-1 inhibitor-treated, and combination (SBI-0206965 + PD-1 inhibitor)-treated groups. TiME was stained for CD4⁺ helper T cells (white), CD8⁺ cytotoxic T cells (yellow), granzyme B⁺ cells (GrzB, magenta), endothelial cells (CD31, red), and pericytes (α-SMA, green). Nuclei were stained with DAPI (blue). The zoomed-in regions highlight immune cell distribution and formation of perivascular immune niches in the combination therapy group. Scale bars = 100 μm. C-F:Quantitative analysis of immune cell densities in the tumor stroma, including CD4⁺ T cells (C), CD8⁺ T cells (D), CD4⁺ granzyme B⁺ T cells (E), and CD8⁺ granzyme B⁺ T cells (F). G and H: Quantification of the number of CD4⁺ granzyme B⁺ T cells (G) and CD8⁺ granzyme B⁺ T cells (H) within a 10-μm radius around tumor vessels. Data are presented as box plots showing the maximum and minimum values, medians, and 25/75 percentiles (n = 7 mice per group; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; One-way ANOVA). I: Tumor volume changes in each treatment group starting on the day of tumor grafting. Tumor growth was significantly reduced in the combination therapy group (**P < 0.01, two-way ANOVA with Bonferroni post-hoc test). Data are represented as the mean ± SEM.

Inhibition of autophagy activates p53 and induces apoptosis in endothelial cells under hypoxia and nutrient deprivation. A and B: Immunofluorescent staining of HUVECs cultured under different conditions: control, hypoxia (1% O₂), starvation (glucose and serum starvation in MEM), and combined hypoxia and starvation (HS). LC3B (red) marks autophagic activity, whereas DAPI (blue) stains the nuclei. Quantification of LC3B puncta-positive cells showed a significant increase in autophagy under combined hypoxia and nutrient deprivation compared to that in the control (****P < 0.0001). Scale bar = 50 μm. C: Volcano plot displaying DEGs in HUVECs cultured under hypoxia and nutrient deprivation versus controls. Upregulated genes are highlighted in red, downregulated genes in blue, and stable genes in grey. MAP1LC3B2 expression was significantly upregulated, indicating elevated autophagic activity. D: GO analysis showing the enriched pathways in HUVECs cultured under hypoxia and nutrient deprivation. E: Volcano plot comparing gene expression in HUVECs treated with hypoxia and nutrient deprivation combined with 4 μmol/L ULK-101 (autophagy inhibitor) versus hypoxia and nutrient deprivation alone. Apoptosis-related genes, including FADD, BAX, CASP9, and FAS, were significantly upregulated, while TP53 expression was restored by ULK-101 treatment. F: GO analysis of HUVECs treated with ULK-101 under hypoxia and nutrient deprivation conditions, showing significant enrichment of apoptosis-related pathways, angiogenesis, and positive regulation of immune responses. G and H: Flow cytometry analysis of apoptosis in HUVECs under control, hypoxia and nutrient deprivation (HS), 4 μmol/L ULK-101, and combined ULK-101 and HS treatments. Apoptosis was measured using Annexin V/PI staining. Quantification (H) showed a significant increase in apoptosis in HUVECs treated with ULK-101 under hypoxic and nutrient-deprived conditions (****P < 0.0001; **P < 0.01). I and J: Immunofluorescence staining of HUVECs treated with control, hypoxia and nutrient deprivation (HS), 4 μmol/L ULK-101, and combined ULK-101 and HS treatments, with and without the p53 inhibitor pifithrin-α (PFTα) HBr (10 μmol/L). P53 (red) and cleaved caspase-3 (green) indicate apoptosis, with DAPI staining of the nuclei (blue). Quantification revealed a significant increase in nuclear p53 translocation after ULK-101 treatment (J). Scale bar = 50 μm. K: Western blot analysis of p53 and BAX protein expression in HUVECs under control, hypoxia and nutrient deprivation (HS), 4 μmol/L ULK-101, and combined treatment with 10 μmol/L PFTα. The 4 μmol/L ULK-101 treatment significantly increases p53 and BAX expression under hypoxia and nutrient deprivation.