Mitochondrial respiration controls neoangiogenesis during wound healing and tumour growth

Abstract

The vasculature represents a highly plastic compartment, capable of switching from a quiescent to an active proliferative state during angiogenesis. Metabolic reprogramming in endothelial cells (ECs) thereby is crucial to cover the increasing cellular energy demand under growth conditions. Here we assess the impact of mitochondrial bioenergetics on neovascularisation, by deleting cox10 gene encoding an assembly factor of cytochrome c oxidase (COX) specifically in mouse ECs, providing a model for vasculature-restricted respiratory deficiency. We show that EC-specific cox10 ablation results in deficient vascular development causing embryonic lethality. In adult mice induction of EC-specific cox10 gene deletion produces no overt phenotype. However, the angiogenic capacity of COX-deficient ECs is severely compromised under energetically demanding conditions, as revealed by significantly delayed wound-healing and impaired tumour growth. We provide genetic evidence for a requirement of mitochondrial respiration in vascular endothelial cells for neoangiogenesis during development, tissue repair and cancer.

Fig. 1. Endothelial cox10 is required for embryonic development.

a Genotypes of weaned mice or embryos at different stages of gestation obtained from crossing Tie2-Cre cox10^fl/wt to cox10^fl/fl. b Representative images of cox10^EC+/+ (cox10^fl/fl) and cox10^EC−/− E12.5 embryos. c Whole-mount yolk sacs stained with anti-CD31 as an endothelial marker and d quantification of CD31+ yolk sac vasculature (n = 4/genotype). e Quantification of blind ends (n = 4/genotype). f Whole embryos (E12.5) stained with anti-CD31 in 5-fold (upper panel) and 10-fold (lower panel) magnification. g Quantification of embryo CD31 area density (n = 3/genotype). Data are presented as mean ± SD. Individual data points in (d, e, g) represent analysed mice per genotype (d, e n = 4, g n = 3). Exact p-values (unpaired students t-test, two-tailed): d 0.0003, e 0.0003, g 0.0010.

Fig. 2. Loss of endothelial cox10 changes EC metabolic phenotype.

a Oxygen consumption rates (OCR) of COX proficient cox10^fl/fl control primary ECs vs. COX-deficient cox10^KO ECs before and after sequential injection of oligomycin, FCCP and a mixture of rotenone/antimycin A by a Seahorse XF96 analyser. b–e Corresponding calculated parameters of mitochondrial respiration. f Extracellular acidification rate (ECAR) of cox10^fl/fl vs. cox10^KO ECs after sequential injection of glucose, oligomycin and 2-DG (2-deoxy-D-glucose). g, h corresponding calculated parameters expressing glycolytic capacity of ECs. i Calculation of drop in OCR or gain in ECAR of isolated murine ECs (cox10^fl/fl control primary ECs vs. COX-deficient cox10^KO ECs) in response to different glucose concentrations. j Metabolomic analysis of pathway intermediates, grouped into citrate cycle, glycolysis and nucleotides, normalised to cox10^fl/fl controls. Data are presented as mean ± SD. Individual data points in (a)–(j) represent technical replicates within a representative experiment, sample size: a–h n = 6, i n = 5, j n = 3. Exact p-values (unpaired students t-test, two-tailed): b, c, d, e, g, h <0.0001; i gain from bottom to top: **<0.0001, **=0.0024, **=0.0009, **=0.0034. Drop from bottom to top: ##: 0.0022, n.s.: 0.1202, ##: 0.0066, n.s.: 0.0515. j Citrate cycle from left to right: 0.0339, 0.1097, 0.0068, 0.0191, 0.0019; glycolysis from left to right: 0.0019, 0.0052, 0.0116, 0.1247, 0.9148. Nucleotides from left to right: 0.0006, 0.0534, 0.1033, 0.0348, 0.3952, 0.0024, 0.0055, 0.0015, 0.0041.

Fig. 3. Cox10 is essential for EC function.

a Time course of cell death of cox10^fl/fl ECs and cox10^KO ECs in the presence of different glucose concentrations measured by yoyo-1 uptake. b Representative images of scratch wound assay of cox10^fl/fl and cox10^KO ECs after 16 h cultured in medium containing 5.5-mM glucose. c Quantification of scratch wound closure of monolayer cultured control cox10^fl/fl and cox10^KO ECs (% of wound closure) at indicated glucose levels after 16 h. d Representative bright field images of sprouting spheroids. e Quantification of sprouts per spheroid. f Maximum projections of eGFP-expressing ECs generated from multiphoton microscopy z-stacks of murine ear tissue of cox10^wt (EndSCLCreERT/R26mTmG) or cox10^ECKO (EndSCLCreERT/cox10^fl/fl/R26mTmG) mice after tamoxifen treatment. GFP expression indicates EC-specific Cre recombination. g Kaplan–Meier curve showing survival probability of cox10^{wt or fl/fl} mice vs. cox10^ECKO mice. h Representative fluorescent images (maximum projection of confocal stacks) of aortic rings derived from cox10^wt (EndSCLCreERT/R26mTmG) or cox10^ECKO (EndSCLCreERT/cox10^fl/fl/R26mTmG) mice after tamoxifen treatment. i Quantification of GFP+ sprouts and j junctions. Data are presented as mean ± SD. Individual data points in (a), (c) and (e) represent technical replicates within a representative experiment. Sample size: a n = 3, c n = 3, e n = 4, except cox10^KO 11 mM: n = 5. Individual data points in (i) and (j) represent mean values of individual mice of the respective genotype. Sample size: i n = 4 (wt) vs. n = 6 (ECKO), j n = 3 (wt) vs. n = 7 (ECKO). Exact p-values (unpaired students t-test, two-tailed): a 0 mM: <0.0001, 5 mM: 0.0021, 11 mM: <0.0001, 22 mM: 0.00057 (comparing time point 24 h), e 0.0025 (22 mM), 0.0016 (11 mM), i 0.0008, j 0.0003. Log-rank test was used to determine differences between survival curves in (g), exact p-value: 0.8607. P-values (one-way ANOVA followed by Bonferroni post hoc test): c **<0.0001.

Fig. 4. EC OxPhos is required for neoangiogenesis in wound healing and tumour growth.

a Representative macroscopic wound images of three different mice per genotype (tamoxifen-treated cox10^fl/fl and EndSCLCreERT/cox10^fl/fl (cox10^ECKO) mice). b CD31 and haematoxylin stained wound sections derived from these mice at day 7 post injury. d: dermis; gt: granulation tissue; he: hyperproliferative epithelium (scale bar 500 µm). c Proportion of closed wounds 7 days post injury. d Quantification of CD31+ vessels within the granulation tissue. e Lewis lung cell carcinoma (LLC) subcutaneous tumour growth (mean ± SEM) in tamoxifen-treated cox10^fl/fl mice and EndSCLCreERT/cox10^fl/fl (cox10^ECKO) mice. f Representative fluorescent images of CD31+ tumour vessels in LLC tumours. g Quantification thereof. h Quantification of tumour necrosis in LLC tumours. i Melanoma (B16F10) subcutaneous tumour growth (mean ± SEM). j Representative fluorescent images of CD31⁺ vessels in B16F10 tumours. k Quantification thereof. l Quantification of B16F10 tumour necrosis. m Representative fluorescent images of CD31+ vessels in dextran perfused LLC tumours. n Quantification of dextran leakage and o vessel perfusion, p Representative H&E-stained images of metastatic lungs from LLC-injected (s.c.) mice. q Quantification of the metastatic index. Data are presented as mean ± SD except when indicated otherwise above. Individual data points in (d), (g), (h), (k), (l), (n), (o) and (q) represent mean values of individual mice of the respective genotype. Sample sizes: d n = 4, e n = 9 (cox10^fl/fl) vs. n = 7 (cox10^ECKO), g n = 6 vs. n = 7, h n = 6 vs. n = 6, i n = 7 (cox10^fl/fl) vs. n = 6 (cox10^ECKO), k n = 5 vs. n = 5, l n = 4 vs. n = 3, n n = 3 vs. n = 3, o n = 3 vs. n = 3, q n = 6 vs. n = 5. Exact p-value (one sample t-test, two-tailed): c 0.0424. Exact p-values (unpaired students t-test, two-tailed): d 0.0004, g 0.0248, h 0.0087, k 0.0363, l 0.0096, n 0.0435, o 0.0086, q 0.0351. Exact p-values (two-way ANOVA, Bonferroni post hoc): e <0.0001 (day 21), i <0.0001 (day 27).

Fig. 5. Lactate fuels EC respiration.

a Calculated glycolysis rate of murine lung ECs, murine Lewis Lung Cell Carcinoma (LLC) cells and murine B16F10 melanoma cells. b Analysis of L-lactate medium levels secreted by ECs and murine tumour cell lines B16F10 and LLC. c Basal respiration of isolated murine ECs with or without lactate supplementation. d Calculated OxPhos coupled ATP levels of primary murine lung ECs in the absence or presence of lactate. e Representative bright field images of sprouting spheroids exposed to lactate. f Quantification of sprouting response. g Representative aortic ring explants in the presence and absence of lactate. h Quantification of the sprouting response. Data are presented as mean ± SD. Individual data points in (a), (b), (c), (d) and (f) represent technical replicates within a representative experiment. Individual data points in (h) represent mean values of individual mice of the respective genotype. Sample sizes: a LLC, B16: n = 3, ECs n = 6, b n = 3 for all cells, c n = 4 vs. 4, d n = 4 for both conditions, f n = 4 vs. 4, h n = 3 vs. 3 (−lactate); n = 5 vs. 5 (+lactate). Exact p-values (unpaired students t-test, two-tailed): c <0.0001, d <0.0001, f 0.0044, h *=0.0305, **=0.0107, n.s. = 0.2864, **=0.0018. Exact p-values (unpaired students t-test, two-tailed): ECs compared with both tumour cells (**: ECs vs LLC, ^#: ECs vs B16F10): a **^,#: <0.0001, b **: 0.0025, ^#: 0.0030.