Autophagy protein 5 controls flow-dependent endothelial functions

Abstract

Dysregulated autophagy is associated with cardiovascular and metabolic diseases, where impaired flow-mediated endothelial cell responses promote cardiovascular risk. The mechanism by which the autophagy machinery regulates endothelial functions is complex. We applied multi-omics approaches and in vitro and in vivo functional assays to decipher the diverse roles of autophagy in endothelial cells. We demonstrate that autophagy regulates VEGF-dependent VEGFR signaling and VEGFR-mediated and flow-mediated eNOS activation. Endothelial ATG5 deficiency in vivo results in selective loss of flow-induced vasodilation in mesenteric arteries and kidneys and increased cerebral and renal vascular resistance in vivo. We found a crucial pathophysiological role for autophagy in endothelial cells in flow-mediated outward arterial remodeling, prevention of neointima formation following wire injury, and recovery after myocardial infarction. Together, these findings unravel a fundamental role of autophagy in endothelial function, linking cell proteostasis to mechanosensing.

Keywords: Autophagy; Endothelium; Flow-mediated dilatation; Mechanosensing; VEGFR2; eNOS.

Fig. 1

ATG5 deficiency promotes mitochondrial dysfunction and metabolic changes in endothelial cells. A-B Selected canonical pathways identified by Ingenuity Pathway Analysis (IPA) performed on differentially expressed genes (A) and proteins (B) between primary lung endothelial cells isolated from Atg5^lox/lox and Cdh5.Cre Atg5^lox/lox mice. n = 3 replicates. C Western blot analysis of the expression of ATG5 and LC3B showing decreased autophagic flux in HUVECs transduced with a lentivirus coding for shRNA against ATG5 (ATG5 KD). GAPDH was used as a loading control. Representative of n = 4 replicates. Lanes #1 and #2 represent 2 different protein extracts from 2 different experiments. D Immunofluorescence of P62 (green) and mitotracker (red) in control (GFP) and ATG5 KD HUVECs. Scale bar 25 μm. E Quantification of the mitochondrial footprint, P62 expression, and Mander’s correlation of the mitochondria staining colocalizing with P62 staining in control and ATG5 KD HUVECs. ATG5 KD HUVECs present P62 accumulation, altered mitochondria architecture, and increased number of mitochondria colocalizing with P62. Each dot represents a different culture well. F Western blot analysis of the expression of mitochondrial proteins: SDHA, SDHB, COX1, COX4, ATP5A and UQCRC2. GAPDH was used as a loading control. Representative of n = 4 replicates. Lanes #1 and #2 represent 2 different protein extracts from 2 different experiments. G–H Flow cytometry analysis of the active pool of mitochondria in control and ATG5 KD HUVECs using mitotracker Red CMXRos (G) and TRMR (H) stainings. n = 3 experimental replicates. I Metabolomic detection of TCA intermediates in control and ATG5 KD HUVECs. n = 9 experimental replicates. J Graphical summary

Fig. 2

Endothelial autophagosome content. A Volcano plot of proteins enriched in autophagosomes of lung endothelial cells isolated from GFP-LC3 mice. B Selected canonical pathways identified by Ingenuity Pathway Analysis (IPA) performed on proteins contained into autophagosomes isolated from primary lung endothelial cells from GFP-LC3 mice. C Heatmap representation of some membrane and extracellular proteins contained in autophagosomes. n = 3 replicates. *p < 0.05, **p < 0.01, ***p < 0.001 GFP-LC3 vs. NEG (D) Fold-change in protein content of some cell surface and extracellular proteins in primary lung endothelial cells isolated from Cdh5.Cre Atg5^lox/lox mice compared to protein content in primary lung endothelial cells from Atg5^lox/lox. E Immunoblot analysis of VEGFR2 and LC3B expression in autophagosomes isolated by GFP immunoprecipitation without or with VEGFA treatment. Lanes #1 and #2 represent two different protein extracts from two different experiments. F Immunofluorescence of VEGFR2 and GFP in primary lung endothelial cells isolated from GFP-LC3 mice with or without VEGFA treatment. Scale bar 5 μm. n = 6 mice. G Immunofluorescence of VEGFR2, LC3B and ATG16L1 in primary lung endothelial cells from WT mice after VEGFA treatment. Cell surface VEGFR2 only was stained to analyze VEGFR2 internalization. n = 3 replicates. Scale bar 30 μm. H Immunofluorescence of VEGFR2 and LC3B in HUVECs. Cell surface VEGFR2 only was stained to analyze VEGFR2 internalization. n = 3 replicates. Scale bar 15 μm. G,H Colocalization was quantified using Mander’s coefficient

Fig. 3

ATG5 deficiency induces defective VEGFR2 signaling. A Immunofluorescence of VEGFR2 (red) and VE-cadherin (green) expressions on flat mount aortas from Atg5^lox/lox and Cdh5.Cre Atg5^lox/lox mice. n = 5 mice per genotype. Scale bar 20μ m. B Quantification of VE-cadherin discontinuity and VEGFR2 area per field in flat-mount aortas from Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice. 4 fields per mouse were analyzed for VE-cadherin quantification, and 2 fields per mouse for VEGFR2 quantification. C Immunofluorescence of cell surface VEGFR2 (red) expression in control and ATG5 KD HUVECs. Nuclei were stained with Hoechst. n = 3 replicates. Scale bar 25 μm (D) Quantification of the cell surface area of VEGFR2 per cell area in control and ATG5 KD HUVECs. E Western blot analysis of phospho-VEGFR2, VEGFR2, phospho-eNOS and phospho-P38 in control and ATG5 KD HUVECs after VEGFA treatment. Tubulin was used as a loading control. n = 4 replicates. F Western blot analysis of phospho-VEGFR2, VEGFR2, phospho-NOS3 and NOS3 expression in primary lung endothelial cells from Atg5^lox/lox and Cdh5.cre Atg5^lox/lox mice after VEGFA treatment. n = 4 replicates. G Immunofluorescence of VEGFR2 (red) and EEA1 (green) expression in control and ATG5 KD HUVECs after 20 min of VEGFA treatment. Cell surface VEGFR2 was stained to visualize internalization upon VEGFA treatment. Nuclei were stained with Hoechst. n = 3 replicates. Scale bar 25 μm. H Quantification of VEGFR2 staining overlapping EEA1 staining and EEA1 staining overlapping VEGFR2 staining in control and ATG5 KD HUVECs after 20 min of VEGFA treatment

Fig. 4

Endothelial autophagy deficiency impairs angiogenesis. A Representative images of aortic explants in 4 to 6-week-old Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice. Scale bar 400 mμ. B Representative images of TRITC-coupled isolectin showing endothelial cells forming tubes. Scale bar 200 mμ. Higher magnification is shown in the inset. C Aortic ring assay quantification in Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice showing decreased vascular sprouting in Cdh5.Cre.Atg5^lox/lox aortic explants. D Representative images of choroid explants in 4 to 6-week-old Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice. Scale bar 400 mμ. E Choroid sprouting angiogenesis assay quantification in Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice showing decreased vascular sprouting in Cdh5.Cre.Atg5^lox/lox choroidal explants. Values are individual values and means ± SEM of 10–14 mice. *p < 0.05. F In vivo DIVAA angiogenesis assay. Angioreactors containing FGF2 or VEGFA were implanted subcutaneously for 15 days. After removal, endothelial cells into angioreactors were stained with a fluorescent isolectin staining and fluorescence was quantified. Quantification of angiogenesis in angioreactors after implantation in Atg5^lox/lox and Cdh5.Cre.Atg5^lox/lox mice showed decreased FGF- and VEGF-induced angiogenesis in Cdh5.Cre.Atg5^lox/lox mice. n = 4 angioreactors per condition. Values are individual values and means ± SEM. 2 way-ANOVA: genotype effect: *p < 0.05, post-hoc Fisher LSD test *p < 0.05

Fig. 5

Mice with selective-endothelial autophagy deficiency present decreased peripheric blood flow velocity and endothelial dysfunction. A-E Changes in mesenteric artery contractility in mice with endothelial autophagy deficiency. A Wall force was measured in response to phenylephrine (Phe) in isolated mesenteric arteries from Atg5^lox/lox and Cdh5.cre-Atg5^lox/lox mice. B, C Mesenteric arteries were precontracted with Phe. Acetylcholine (Ach) (B) or NO donor sodium nitroprusside (SNP) (C) was cumulatively added after the contraction had reached a steady-state level. Data are shown as percentage of relaxation of the steady-state preconstriction. D Myogenic tone. E Flow-mediated dilation (FMD). Autophagy deficiency strongly impairs FMD in isolated mesenteric arteries. F Intrarenal pressure flow-relationship measured in isolated and perfused kidneys from Atg5^lox/lox and Cdh5.cre-Atg5^lox/lox mice. A–F Values are means ± SEM of n = 9 Atg5^lox/lox mice and n = 7 Cdh5.cre-Atg5^lox/lox mice. ***p < 0.001 (G–J) Echo-Doppler measurements of mean blood flow velocity (mBFV). G, I Representative BFV waveforms in the renal artery (G) and the basilar trunk artery (I) in 12 weeks-old Atg5^lox/lox and Cdh5.cre.Atg5^lox/lox mice at steady state. H, J Echo-Doppler measurements of mean blood flow velocity (mBFV) in the renal artery (H) and the basilar trunk artery (BT) (J). Endothelial autophagy deficiency is associated with a significant decrease in renal and cerebral mBFV velocity. *p < 0.05; ** p < 0.01. Values are individual values and means ± SEM. n = 10 mice per group except for (J) where n = 10 Atg5^lox/lox mice and n = 5 Cdh5.cre.Atg5^lox/lox mice. K, L Representative immunoblot (K) and quantifications (L) of AKT, phospho-AKT, eNOS (NOS3) and phospho-eNOS expressions in freshly isolated renal arteries from Atg5^lox/lox and Cdh5.cre.Atg5^lox/lox mice. β Actin (ACTB) was used as a loading control. In K, lanes #1 and #2 represent 2 different protein extracts from 2 different mice. Values are individual values and mean ± SEM of n = 9 Atg5^lox/lox mice and n = 11 Cdh5.cre-Atg5^lox/lox mice. ***p < 0.001; **p < 0.01. (M,N) Hypercapnia-induced vasodilatation is impaired in the BT from Cdh5.cre.Atg5^lox/lox mice. M mBFV in BT from Atg5^lox/lox and Cdh5.cre.Atg5^lox/lox mice after air or CO₂ inhalation. CO2 inhalation induces a rise in mBFV in Atg5^lox/lox mice but not in Cdh5.cre-Atg5^lox/lox mice. (N) Hypercapnia-induced relative vasoreactivity in BT from Atg5^lox/lox and Cdh5.cre-Atg5^lox/lox mice. # CO₂ vs. air in Atg5^lox/lox mice, * Atg5^lox/lox vs. Cdh5.cre-Atg5^lox/lox. #p < 0.05; **p < 0.01 ***p < 0.001. Values are individual values and means ± SEM. n = 10 Atg5^lox/lox mice and n = 5 Cdh5.cre.Atg5^lox/lox mice. O NOx measurement in the plasma of Atg5^lox/lox and Cdh5.cre.Atg5^lox/lox mice. NOx level was always below detection threshold in the plasma of Cdh5.cre.Atg5^lox/lox mice. Values are individual values and means ± SEM. n = 10 Atg5^lox/lox mice and n = 6 Cdh5.cre.Atg5^lox/lox mice. **p < 0.01

Fig. 6

Endothelial autophagy deficiency impairs flow-mediated vascular remodeling. A–C Cdh5.cre Atg5^lox/lox mice have an abnormally elevated rise in blood pressure after angiotensin II perfusion. Day measurements of systolic (A) and diastolic (B) blood pressure and heart rate (C) were recorded by radiotelemetry. Values are means ± SEM of n = 4 mice per group. D Endothelial autophagy deficiency impairs flow-mediated outward remodeling of mesenteric arteries. Arterial diameter was measured 1 week after arterial ligation in high-flow (HF) and normal-flow (NF) arteries isolated from Atg5^lox/lox control and Cdh5.cre-Atg5^lox/lox mice in response to increasing intraluminal pressure. Values are means ± SEM of n = 7–10 mice. 2-way ANOVA: pressure p < 0.0001, genotype p < 0.001. Fisher’s LSD test: * # $ p < 0.05, ** ##p < 0.001, ***p < 0.001. * NF vs. HF in Atg5^lox/lox, $ NF vs. HF in Atg5^lox/lox, # HF Atg5lox^/lox vs. HF Cdh5.cre Atg5^lox/lox. E Representative images of Masson’s trichrome (top panel), picrosirius (middle panel) and GSA-TRITC (red)/WGA-FITC (green) (lower panel) staining of hearts from Cdh5.Cre Atg5^lox/lox and Atg5^lox/lox mice, 10 days after MI. Scale bar 50μ m. F–H Quantifications of the infarcts size (F), collagen content (G) and capillary density (H). Values are individual plots and means ± SEM of n = 8 mice. *p < 0.05