Altered Mitochondrial Opa1-Related Fusion in Mouse Promotes Endothelial Cell Dysfunction and Atherosclerosis

Flow (shear stress)-mediated dilation (FMD) of resistance arteries is a rapid endothelial response involved in tissue perfusion. FMD is reduced early in cardiovascular diseases, generating a major risk factor for atherosclerosis. As alteration of mitochondrial fusion reduces endothelial cells’ (ECs) sprouting and angiogenesis, we investigated its role in ECs responses to flow. Opa1 silencing reduced ECs (HUVECs) migration and flow-mediated elongation. In isolated perfused resistance arteries, FMD was reduced in Opa1+/− mice, a model of the human disease due to Opa1 haplo-insufficiency, and in mice with an EC specific Opa1 knock-out (EC-Opa1). Reducing mitochondrial oxidative stress restored FMD in EC-Opa1 mice. In isolated perfused kidneys from EC-Opa1 mice, flow induced a greater pressure, less ATP, and more H2O2 production, compared to control mice. Opa1 expression and mitochondrial length were reduced in ECs submitted in vitro to disturbed flow and in vivo in the atheroprone zone of the mouse aortic cross. Aortic lipid deposition was greater in Ldlr−/--Opa1+/- and in Ldlr−/--EC-Opa1 mice than in control mice fed with a high-fat diet. In conclusion, we found that reduction in mitochondrial fusion in mouse ECs altered the dilator response to shear stress due to excessive superoxide production and induced greater atherosclerosis development.


Introduction
The vascular endothelium has a central role in vascular homeostasis, and an endothelial dysfunction is a hallmark of cardiovascular disorders [1]. Blood flow generates shear stress on the intraluminal side of the endothelial cells (ECs) and activates the production of vasodilator agents such as nitric oxide (NO) [2,3]. This flow-mediated dilation (FMD) is In all protocols, littermate wild-type mice were used as control. Each mouse was genotyped before entering a protocol. All procedures were performed in accordance with the principles and guidelines established by the National Institute of Medical Research (INSERM) and were approved by the local Animal Care and Use Committee (APAFIS#3577-201601141342229, APAFIS#2018011217209, APAFIS#30385-2021031010145750). The investigation conforms to the directive 2010/63/EU of the European parliament.
At the end of each protocol, the arterial segment was bathed in a Ca 2+ -free PSS containing ethylene-bis-(oxyethylenenitrolo) tetra-acetic acid (2 mmol/L) and sodium nitroprusside (10 µmol/L) before increasing pressure by step from 10 to 125 mmHg (no flow) in order to determine arterial diameter and wall thickness [32].
As these experiments involved living arteries some experiments were halted before the end of a protocol. Thus, 10 to 12 mice were used per group depending on litter size and on the number of KO and WT mice per litter. Because of tissue viability, only 2 or 3 arterial segments could be collected per mouse. The main reasons for stopping an experiment were: (1) breaking of a glass cannula, breaks during the mounting of the artery or during a washout, (2) perfusion pressure could not be maintained due to a leak (small collateral arteriole not visible when mounting the artery), (3) the artery does not pass a test: no contractility (smooth muscle damaged during dissection or mounting) or no dilation (endothelium damaged during dissection or mounting), (4) disconnection of the artery from one glass cannula before the end of the experiment, (5) formation of gas bubbles in the tubing and one bubble crosses the artery. This is enough to damage the endothelium and to stop the experiment. Consequently, the number of mice used per group was not homogenous. Number of mice included is given in the figure legends.

Perfused Isolated Mouse Kidney
In another group of experiments the right renal artery was cannulated in anesthetized mice. The right kidney was then excised and perfused as described previously [34]. Perfusion rate was 600 µL/min and the perfusion pressure was monitored using a PM-4 pressure monitor (LSI, Burlington, VT, USA). Endothelium-dependent dilation was assessed through the perfusion of acetylcholine (1 µmol/L) after precontraction (phenylephrine, 1 µmol/L). The perfusion flow was then increased by step in order to determine the flow-pressure relationship [34].
The PSS perfusing the kidney was collected when the flow was 600 µL/min. The perfusate was immediately frozen in liquid N 2 and then stored at −80 • C.
As these experiments involved living kidneys some experiments were halted before the end of a protocol. Thus, 8 mice were used per group and one kidney could be used per mouse. The main reasons for stopping an experiment were: (1) gas bubble moving from the tubing to the kidney, (2) no contractility or no dilation if the dissection or cannulation was too long, (3) disconnection of the kidney from the cannula. Consequently, the number of mice included per group was not homogenous. The number of mice included is given in the figure legend.

Determination of ATP and H 2 O 2 Levels in the Kidney Perfusate
Flow-induced ATP and H 2 O 2 release were measured on perfusate collected from each kidney (500 µL) and concentrated as previously described [34]. Hydrogen peroxide (H 2 O 2 ) concentration was then measured using the Hydrogen Peroxide Assay Kit Colorimetric/Fluorometric NbAb102500, Abcam, Cambridge, UK) and the concentration of ATP was determined using the ATP Determination Kit (Nb A22066, Invitrogen TM Molecular Probes, Eugene, OR, USA) [34].

RNA Interference
RNA interference experiments were performed by transfection of a silencer scrambled negative control (Ambion, cat# AM4611, 50 nM, Warrington, UK) and pre-validated silencer selected pre-designed siRNAs against OPA1 (Ambion cat# s9851, 50 nM, Warrington, UK) using the Lipofectamine™ transfection reagent (Invitrogen cat# 18324012, Eugene, OR, USA), according to the manufacturer's instructions. HUVECs were seeded at a density of 750,000 cells/well in a 6-well plate. The lipofectamine and the siRNA were diluted in a reduced serum Opti-MEM (Gibco, Grand Island, NY, USA), mixed in a 1:1 v/v, and incubated for 30 min at room temperature. Then, 250 µL of the mixture was added to each well of the 6-well plate with 500 µL of Endothelial Cell Growth Basal Medium-2 (EBM TM -2) from Lonza (Basle, Switzerland), for 48 h. To verify siRNA efficiency, a western blot assay of Opa1 was carried out. Transfected cells were trypsinized, washed three times in ice-cold PBS, and harvested using SDS lysis buffer (SDS 1% v/v, Tris 10 mM pH7.4, halt protease and phosphatase inhibitor cocktail, 5 mM EDTA solution).

Cell Migration Assay
For the migration assay, Ibidi ® culture inserts 2-well in µ-well (cat# 81176, Ibidi, Munich, Germany) were used. Briefly, the inserts consist of two chambers separated by a 0.5 mm divider, each chamber with a growth area of 0.22 cm 2 . The inserts were planted into 35 mm tissue culture dishes using sterile tweezers. HUVECs were transfected using the previously described protocol, cell suspensions were prepared at 7-9 × 105 cells/mL Antioxidants 2022, 11, 1078 5 of 24 in medium, of which 70 µL was transferred to each chamber, allowing cell adhesion and growth to~100% confluence. After removing the strip between dividing the chambers image acquisition was performed along the cell-free zone from 0 to 8 h. Cell migration was quantified using the ImageJ-Fiji software.

In Vitro Exposure of Endothelial Cells to Shear Stress
The Ibidi ® pump system (Fluidic Unit: Cat No. 10902; Perfusion set 15 cm, ID 1.6 mm: Cat No. 10962; µ-slide I 0.4 Luer: Cat No. 80176; Ibidi GmbH, Munich, Germany) was used to submit ECs to flow. The pump system was constructed according to the instructions (Ibidi GmbH, AN 13: HUVECs under Perfusion, Munich, Germany). 350,000 primary HUVEC cells were seeded onto a µ-slide I 0.4 Luer and incubated for 24 h at 37 • C and 5% CO 2 to form a monolayer. The next day, the experiment was conducted under the following conditions: 31.9 mbar pressure, 21.70 mL/min flow rate, 0.007 dyn x s/cm 2 viscosity and a shear stress of 20 dyn/cm 2 . Static conditions were also tested on the same Ibidi ® µ-slide I 0.4 Luer (Ibidi, Munich, Germany).

Measurement of Endothelial Cells Alignment and Elongation
Endothelial cells' alignment was assessed using the Directionality plugins in Image J-Fiji (https://imagej.net/Directionality, available online https://imagej.net/software/fiji/ accessed on 28 June 2012). Quantification of cell orientation, measured by elongation factor (cell length along flow direction divided by cell width) [35] was performed on the same images.

Circular Flow Assay
In order to submit a larger number of cells to flow, orbital shaker experiments were performed using the MS1 (MILE SVEN1) cell line (ATCC ® CRL2279TM, Les Ullis, France). Cells were maintained in a DMEM culture medium (Lonza, Basle, Switzerland) and supplemented with 5% v/v fetal bovine serum, 2 mM glutamine, 100 U/mL penicillin and 100 µg/mL streptomycin, in a humid atmosphere of 37 • C with 5% CO 2 . Cells were then placed at 37 • C as control and cells under flow were placed under agitation of our laboratory digital orbital shaker for 24 or 72 h at 210 rpm (12 dyn/cm 2 ) or 260 rpm [36].
In order to verify that cells were submitted to laminar (protective) flow we measured Nos3 (encoding for eNOS) and Edn1 (encoding for ET-1) gene expression. Laminar flow has been shown to increase eNOS epression and to decrease Edn1 expression whereas a disturbed flow has the opposite effect [37][38][39][40][41].
Cells were then collected for transcriptomic analyses as described below.

Analysis of mRNA Levels by RT-qPCR
We used quantitative polymerase chain reaction after reverse transcription of total RNA (RT-qPCR) in order to measure gene expression in the mouse aorta. Segments of the aorta were stored at −20 • C in RNAlater Stabilization Reagent (Qiagen, Valencia, CA, USA). In each segment of aorta, the RNA was extracted (RNeasy ® Micro Kit, Qiagen, Valencia, CA, USA) according to the instructions provided by the manufacturer. 500 ng) of the RNA extracted from the aortic segments was then used to synthesize cDNA using the QuantiTect ® Reverse Transcription Kit (Qiagen, Valencia, CA, USA). RT-qPCR was thenperformed using the Sybr ® Select Master Mix (Applied Biosystems Inc., Lincoln, CA, USA) in a LightCycler 480 Real-Time PCR apparatus (Roche, Branchburg, NJ, USA).
Primers were validated by testing PCR efficiency using standard curve as MIQE guidelines [42]. Mouse Primers (Eurogentec, Liège, belgium) were presented in Table 1, Hprt, Gapdh and Gusb were used as housekeeping genes.

Mitochondrial Shape Measurement in Endothelial Cells
The mitochondrial shape was analyzed using the Image J-Fiji software (version: 2.1.0/1.53c) as previously described [43,44] on the endothelial cells stained as described above. Images were acquired using Metamorph software and imported into ImageJ. Images were then processed using the MINA (mitochondria network analysis) plug in, providing mitochondria area and number [45]. The mitochondrial fission count (MFC) was then measured as it represents the mitochondria dynamics. MFC was calculated as the total number of mitochondria in a cell divided by the mitochondrial area in the same cell [44].

Analysis of Protein Expression Levels by Western Blot
HUVECs submitted to flow were extracted in RIPA lysis buffer. Homogenates were centrifuged at 13000 rpm at 4 • C for 20 min, and the resulting supernatant was collected. Protein concentration was determined using the Micro BCA protein assay kit (cat#23227, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Equal amounts of proteins (10 µg) were solubilized under denaturating conditions in 25 µL of Laemmli sample buffer containing 2.5% v/v β-mercaptoethanol, boiled 5 min at 90 • C, separated by 4-15% polyacrylamide gel electrophoresis (BioRad, Marnes la Coquette, France) and transferred to a nitrocellulose membrane (BioRad). Membranes were incubated overnight at 4 • C with the primary antibody followed by the appropriate peroxidase-labeled secondary antibody (Table 2) for 1 h. Reactions were visualized by ECL detection according to the manufacturer's instructions (Bio-Rad, Marnes-la-Coquette, France) and membranes were stripped at 50 • C for 30 min in the presence of beta-mercaptoethanol before re-blotting.

Endothelial Cell Isolation from Mouse Mesenteric Arteries
In order to validate the extinction of Opa1 in EC-Opa1 mice, we isolated endothelial cells from mesenteric arteries. A total of two mice were used to obtain enough endothelial cells for one experiment. Mice were anesthetized and exsanguinated. The mesenteric bed was then ligated below the duodenum and above the caecum and excised. All the arteries from the first to the fourth order of the mesenteric bed were dissected. They were placed in a digestion solution composed of Endothelial Cell Growth Basal medium 2 (Lonza, Basle, Switzerland) supplemented with 1000 IU type 2 collagenase. The arteries were cut into small segments using sterilized micro-scissors and left to stir in a 37 • C and 5% CO 2 incubator for 45 min. The homogenized suspension obtained was filtered through a 30 µm Miltenyi filter and then centrifugated (10 min, 300 g). Depending on the number of cells counted, a sufficient quantity of CD146-labeled Miltenyi Microbeads was added to the cell pellet, and left to incubate for 15 min at 4 • C. The cells were then washed before performing a magnetic separation on an MS Miltenyi column. Cells not retained by the column were seeded after washout in DMEM F12 10% v/v FCS in a well of a 6-well plate. They were identified as the non-endothelial cells. Cells retained by the column were flushed and then washed before being seeded into a well of a 12-well plate in Endothelial Growth medium 20% v/v FCS. After 24 to 48 h of adhesion, the cells were washed and allowed to grow until confluence in EGM-2 medium (Lonza, Basle, Switzerland).

Analyses of Atherosclerosis Lesions
Male and female Ldlr −/− -Opa +/− , Ldlr −/− -Opa +/+ , Ldlr −/− -EC-Opa1 and Ldlr −/− -EC-WT mice aged ten-weeks old were subjected to a hypercholesterolemic atherogenic diet (1.25% w/w cholesterol, 7.7% w/w fat, no cholate, U8959P version 0022, SAFE, Lyon, France). After 4 months, lipid deposition size was estimated as previously described [46]. Briefly, the abdominal cavity was opened, and the internal organs were removed. The aortic sinus and the aorta from the arch to the iliac bifurcation were carefully cleaned of peri-adventitial connective tissue. Then the aorta was opened longitudinally by an incision along its ventral aspect. The thoracic and abdominal parts of the aorta were pinned out flat (using 10-mm tips of dental root-canal needles). The aorta and the sinus were stained with oil red O (cat#O0625, Merck, Darmastadt, Germany) and counterstained with Mayer's hematoxylin. The percentage of the areas with lesions was evaluated in each sample [46].

Lipids and Glucose Blood Level in Mice
The concentrations of glucose, total cholesterol, HDL cholesterol, LDL cholesterol, and triglycerides in the sera collected from the mice were assayed on ADVIA ChemistryXPT (Siemens) system, using Liquid Assayed Multiqual control sera, according to the manufacturer's recommendations.

Statistical Analyses
For FMD and the measurements of pressure-dependent increases in diameter and wall thickness, a 2-way ANOVA for repeated measurements (flow or pressure steps) was performed. When a significant interaction was observed between 2 factors, the effect of the gene deletion was studied using a Bonferroni post-test.
In the other experiments, the two-tailed Mann-Whitney test was used when 2 groups were compared and the Kruskal-Wallis test was used when more than 2 groups were involved. The test used is indicated in each figure legend. A probability value lower than 0.05 was considered significant.

In Vitro Endothelial Cell Migration
OPA1 expression was reduced by 76% in HUVECs treated with the siOPA1 compared to control cells treated with the scrambled siRNA (siCONT) ( Figure S1A). The siOPA1 induced a reduction in endothelial cell migration, compared to the scrambled siRNA ( Figure 1A-E).
A positive and a negative control were performed with VEGF and a high concentration of glucose in the medium, respectively ( Figure S2).

In Vitro Flow-Mediated Endothelial Cell Alignment and Elongation
HUVECs subjected to laminar shear stress were similarly aligned in the direction of flow after treatment with siOPA1 compared to cells treated with scrambled siCONT ( Figure 1F-J). Nevertheless, the elongation factor (ratio of cell length/cell width) was lower in siOPA1-treated cells (OPA1 silencing = 89%, Figure S1B) compared to ECs treated with scrambled siCONT (Figure 1K-M).

OPA1 Silencing Altered the Response of Endothelial Cells to Laminar Flow
OPA1 silencing in HUVECs submitted to laminar flow induced a 78% reduction in OPA1 expression (Figure 2A), without affecting eNOS level ( Figure 2B). Nevertheless, OPA1 silencing reduced Klf2, PFKFB3, SOD2, and Cu/Zn SOD expression levels ( Figure 2C-E), while increasing p22phox level, without affecting gp91 expression ( Figure 2G,H). These results suggest that the absence of OPA1 increased oxidative stress in ECs.

In Vitro Flow-Mediated Dilation in Opa1 +/− Mouse Resistance Arteries
The occurrence of Opa1 haploinsufficiency in mesenteric arteries isolated from Opa1 +/− mice was demonstrated in our previous work showing a 69% Opa1 reduction in mesenteric resistance arteries [27]. Stepwise increases in flow rate in mesenteric resistance arteries induced a progressive dilation or FMD ( Figure 3A). FMD was significantly reduced in resistance arteries isolated from male Opa1 +/− mice, compared to littermate Opa1 +/+ male controls ( Figure 3A). Arterial diameter ( Figure 3B) and wall thickness ( Figure 3C) were equivalent in male Opa1 +/− and Opa1 +/+ mice. Acetylcholine-dependent dilation was not significantly different between Opa1 +/− and Opa1 +/+ male mice ( Figure 3D). A similar pattern was observed in female mice ( Figure 3E-H).

OPA1 Silencing Altered the Response of Endothelial Cells to Laminar Flow
OPA1 silencing in HUVECs submitted to laminar flow induced a 78% reduction in OPA1 expression (Figure 2A), without affecting eNOS level ( Figure 2B). Nevertheless, OPA1 silencing reduced Klf2, PFKFB3, SOD2, and Cu/Zn SOD expression levels ( Figure  2C-E), while increasing p22phox level, without affecting gp91 expression ( Figure 2G,H). These results suggest that the absence of OPA1 increased oxidative stress in ECs.

Flow-Mediated Dilation in Mouse Lacking Opa1 in Endothelial Cells
Flow-mediated dilation was then measured in isolated mesenteric resistance arteries isolated from mice lacking Opa1 in endothelial cells (EC-Opa1) compared to controls (EC-WT) ( Figure 4A). Flow-mediated dilation was significantly reduced in male EC-Opa1 mice compared to EC-WT, while arterial diameter ( Figure 4B) and wall thickness ( Figure 4C) were not affected by Opa1 absence in EC-Opa1. Acetylcholine-and SNP-dependent dilation ( Figure 4D,E) were not significantly different between EC-Opa1 and EC-WT mice. Similarly, contraction induced by phenylephrine (1 µmol/L, Figure 4F) or KCl (80 mmol/L, Figure 4E) was not affected by the absence of Opa1 in ECs. A similar pattern was observed in female mice ( Figure 4H-N).
Blockade of NO synthesis with L-NNA significantly reduced FMD in both male and female EC-Opa1 and EC-WT mice ( Figure 4A,H).
Blockade of Opa1 with the Opa1 inhibitor MYSL22 significantly reduced FMD in WT mice ( Figure 4O).
Opa1 silencing was 61% in endothelial cells from resistance mesenteric arteries isolated from EC-Opa1 mice. No change in Opa1 level was observed in smooth muscle cells isolated from the same vessels ( Figure 4P). Whole blots are shown in Figures S5 and S6.
Superoxide reduction with SOD and catalase restored FMD in male and female EC-Opa1 mice so that no difference between EC-Opa1 and EC-WT was observed after incubation of the arteries with SOD and catalase ( Figure 5A,B). Similarly, after blocking superoxide production by the mitochondria with MitoTempo, no difference in FMD between EC-Opa1 and EC-WT mice was observed ( Figure 5C,D).
FMD was also measured after the addition of BH4 and L-Arg. In this condition, FMD was equivalent in EC-Opa1 and EC-WT mice ( Figure 5E,F).

Flow-Pressure Relationship, ATP and H 2 O 2 Production in Ex-Vivo Perfused Kidney
As the kidney is a well autoregulated organ with a dense microvascular network, we investigated flow-mediated responsiveness in perfused kidneys isolated from EC-Opa1 and EC-WT mice ( Figure 6A). First, the flow-pressure relationship was shifted upward in EC-Opa1 compared to EC-WT male and female mice ( Figure 6B,C), suggesting that endothelial responsiveness to flow was reduced, in agreement with the reduced FMD observed in mesenteric arteries. Then, we measured ATP and H 2 O 2 production in the kidney perfusate. Interestingly, ATP production measured in the kidney perfusate was reduced ( Figure 6D,E) whereas H 2 O 2 production was higher ( Figure 6F,G) in EC-Opa1 compared to EC-WT male and female mice.

Disturbed Flow In Vivo and In Vitro Reduced Opa1 Level and Mitochondrial Length
In the mouse aorta, Opa1 gene expression was higher in the greater curvature (GC) than in the lower curvature (LC) ( Figure 7A). On the opposite, Dnm1l gene expression was lower in the GC than in the LC of the aortic cross ( Figure 7B). This was observed in mice receiving a normal diet ( Figure 7A,B) and in mice receiving a Western diet ( Figure 7C,D).
The mitochondrial network density in the two aortic zones (LC in Figure 7E and GC in Figure 7F) was assessed by the measurement of the number of mitochondria ( Figure 7G) and by the percentage of mitochondria on the cell surface ( Figure 7H). These two parameters were similar in the LC and in the GC of the aortic cross. Nevertheless, mitochondrial fission count was lower ( Figure 7I) and branch length was greater in the GC compared to the LC ( Figure 7J), suggesting an alteration in mitochondrial dynamics in the LC.  Panel A: * EC-Opa1 vs EC-WT: p = 0.0213, interaction p = 0.0045, **** LNNA in EC-WT: p < 0.0001, interaction p < 0.0001; * LNNA in EC-Opa1: p = 0.0198, interaction p = 0.0065. Panel H: *** EC-Opa1 vs EC-WT p = 0.00006, interaction p < 0.0001; *** LNNA in EC-WT: p = 0.0007, interaction p < 0.0001; ** LNNA in EC-Opa1: p = 0.0028, interaction p < 0.0001. Panel O: * p = 0.0221, interaction p = 0.0228. NS: Two-way ANOVA for repeated measurements (B,C,I,J). NS, two-tailed Mann-Whitney test, panels D-G and K-N.  As the LC of the mouse aortic cross is submitted in vivo to disturbed blood flow, we subjected mouse endothelial cells (MS1) in vitro, to a laminar or to a disturbed flow using an orbital shaker, submitting cells to a circular flow. Laminar flow was associated with an elevation in eNOS (Nos3) gene expression ( Figure 7K) and a decrease in endothelin-1 (Edn1) gene expression ( Figure 7L). In these conditions, Opa1 and Dnm1l expression levels were not modified by laminar flow (Figure 7M,N, respectively). On the other hand, under conditions of disturbed flow, evidenced by a decrease in Nos3 gene expression ( Figure 7O) and an increase in Edn1 gene expression ( Figure 7P), Opa1 expression level was decreased ( Figure 7Q) while Dnm1l expression level was increased ( Figure 7R).

Lipid Deposits in the Aorta of Opa1 +/− Mice
After 4 months of high-fat diet, mice had lipid deposits in the aorta (Figure 8). Lipid deposit was greater in male Opa1 +/− mice compared to male Opa1 +/+ mice ( Figure 8A) without significant differences in mouse body weight ( Figure 8B), total cholesterol (Figure 8C), triglycerides ( Figure 8D) and glycemia ( Figure 8E) between the two groups. In mice submitted to a standard diet, no lipid deposit could be observed (Pictures on the left). reduced ( Figure 6D,E) whereas H2O2 production was higher ( Figure 6F,G) in EC-Opa1 compared to EC-WT male and female mice.

Lipid Deposits in the Aorta of Opa1 +/− Mice
After 4 months of high-fat diet, mice had lipid deposits in the aorta (Figure 8). Lipid deposit was greater in male Opa1 +/− mice compared to male Opa1 +/+ mice ( Figure 8A) without significant differences in mouse body weight ( Figure 8B), total cholesterol ( Figure 8C), triglycerides ( Figure 8D) and glycemia ( Figure 8E) between the two groups. In mice submitted to a standard diet, no lipid deposit could be observed (Pictures on the left).

Lipid Deposits in the Aorta of Mice Lacking Opa1 in Endothelial Cells
Lipid deposition was greater in male EC-Opa1 mice compared to male EC-WT mice ( Figure 9F). Bodyweight ( Figure 9G  After 4 months of high-fat diet, mice had lipid deposits in the aorta (Figure 8). Lipid deposit was greater in male Opa1 +/− mice compared to male Opa1 +/+ mice ( Figure 8A) without significant differences in mouse body weight ( Figure 8B), total cholesterol ( Figure 8C), triglycerides ( Figure 8D) and glycemia ( Figure 8E) between the two groups. In mice submitted to a standard diet, no lipid deposit could be observed (Pictures on the left).

Figure 8.
Consequence of Opa1 deficiency in endothelial cells on plaque formation. Opa1 +/+ and Opa1 +/− mice that were three months old were fed with an atherogenic diet (Western diet) and compared to Opa1 +/+ and Opa1 +/− mice fed with a standard diet. After 4 months the aorta was collected from the sinus to the iliac bifurcation and stained with Oil red-O. Lipid deposition was quantified in the aortic cross, (A), in the descending aorta (B) and in the aortic sinus (C). Bodyweight (D), plasma cholesterol (E), triglycerides (F), and glycemia (G) were measured in each group of mice.

Lipid Deposits in the Aorta of Mice Lacking Opa1 in Endothelial Cells
Lipid deposition was greater in male EC-Opa1 mice compared to male EC-WT mice ( Figure 9F). Bodyweight ( Figure 9G

Discussion
The present study demonstrates that the mitochondrial fusion protein Opa1 is involved in ECs' response to acute changes in flow (FMD), with a protective effect against atherosclerosis. This effect was equivalent in male and female mice.
Blood flow induces shear stress at the surface of ECs leading to the activation of protective pathways [4]. Laminar shear stress is a stabilizing stimulus for ECs leading to a nonproliferative, nonmigratory and antithrombotic phenotype [47,48]. An acute increase

Discussion
The present study demonstrates that the mitochondrial fusion protein Opa1 is involved in ECs' response to acute changes in flow (FMD), with a protective effect against atherosclerosis. This effect was equivalent in male and female mice.
Blood flow induces shear stress at the surface of ECs leading to the activation of protective pathways [4]. Laminar shear stress is a stabilizing stimulus for ECs leading to a nonproliferative, nonmigratory and antithrombotic phenotype [47,48]. An acute increase in flow induces a rapid vasodilation (FMD) in vitro and in vivo [31,34,[49][50][51].
FMD measurement in humans has revealed a close relationship between the occurrence of risk factors and endothelial dysfunction [4].
Although ECs contain very few mitochondria [52] and do not depend on mitochondrial function for energy production [10,15,53], recent studies have shown that angiogenic factors such as VEGF activate endothelial glycolytic metabolism in ECs and that fatty acid and amino acid oxidation is involved in angiogenic sprouting [15]. Moreover, a recent study has shown that the mitochondrial fusion protein Opa1 limits NFkβ signaling in response to angiogenic stimuli, thus promoting angiogenesis [18]. This latter study highlights the role of Opa1 in angiogenesis and uncovered a major role of ECs' mitochondria, independent of energy production. It further shows that Opa1 silencing in HUVECs and extinction in mouse ECs does not affect ECs' ATP content, respiration and mitochondrial membrane potential. In this respect, our results showing a reduction in ECs migration and flowmediated elongation in ECs submitted to Opa1 silencing, agree with these observations.
Mitochondrial fusion and fission are sensitive to flow (shear stress) in ECs [16] and flow-dependent Ca 2+ mobilization relies on mitochondria-dependent activation of endoplasmic reticulum channels [17]. We observed that FMD was reduced in resistance arteries isolated from Opa1 +/− mice, the mouse model of ADOA, and from mice lacking Opa1 in ECs. These findings suggest a selective defect in the response of ECs to laminar shear stress. FMD is a rapid dilatory response of blood vessels depending on the ability of ECs to produce vasodilator agents such as nitric oxide (NO) [54]. Interestingly, FMD was also reduced in vitro by the specific Opa1 inhibitor MYLS22 [18] after a short (30 min) period of incubation, suggesting that Opa1 is involved dynamically in FMD rather than through changes in gene expression of the actors of flow-mediated mechanotransduction. It is most likely that Opa1 is involved selectively in the endothelial response to flow as the dilation induced by acetylcholine was not affected by Opa1 haploinsufficiency or by the absence of Opa1 in ECs. Similarly, SNP-dependent dilation which is independent of the endothelium and arterial contractility were not affected. Thus, the deficiency in endothelial Opa1 seems to affect mainly the endothelial response to flow.
We found that FMD was mainly dependent on the production of NO as L-NNA reduced FMD greatly. This observation agrees with previous studies in mouse and rat mesenteric arteries [31,55,56] as well as human coronary arteries in vitro [7] and in vivo as assessed at the level of the brachial artery [50]. Although FMD was reduced in mice lacking Opa1 in ECs, NO synthesis blockade with L-NNA remained able to further reduce FMD in these mice. In parallel, we found that the endothelium remained fully efficient in response to acetylcholine, which also activates NO production [57]. Acetylcholine and flow activate the production of NO by endothelial cells through different pathways. Acetylcholine activates eNOS phosphorylation of Ser1177 through the activation of Ca 2+ /calmodulindependent protein kinase II whereas flow acts through the activation of Akt [57]. This selective reduction in FMD without change in receptor-dependent dilation suggests an involvement of Opa1 in the pathway activated by flow (shear stress) only.
The decrease in FMD observed in mice lacking Opa1 in ECs could be the consequence of an excessive ROS production, at least in part of mitochondrial origin, as superoxide reduction by the combination of SOD and catalase improved FMD in these arteries. In agreement, we found that Opa1 silencing in ECs submitted to laminar flow reduced SOD expression and increased p22phox expression, suggesting that Opa1 is involved in the reduction in oxidative stress in ECs submitted to laminar flow. This observation is further confirmed by our results showing that in mice lacking Opa1 in ECs, the kidney produced less ATP and more H 2 O 2 in response to flow. ATP is released by ECs in response to flow to activate purinergic receptor-dependent NO production [58,59]. On the other hand, H 2 O 2 results from excessive oxidative stress as shown in arteries of patients suffering from coronary artery disease, which produce superoxide and H 2 O 2 , instead of NO in response to flow [7]. Other studies suggested that mitochondrial dynamics modulate ECs functions in pathological conditions, favoring superoxide production [60][61][62]. Thus, Opa1 is likely to facilitate the response of ECs to laminar flow through a reduction in oxidative stress, thus allowing a better NO bioavailability and consequently a better FMD. Indeed, flow (shear stress) simultaneously activates the production of NO and a reduction in the production of ROS and inflammatory agents [19]. The present study suggests that Opa1 is rather involved in the second arm of the response to flow, thus leading to flow-mediated reduction in ROS production. The resulting effect on NO bioavailability being thus secondary to this effect on ROS. Consequently, the reduction in Opa1 level in mice led to a decrease in FMD. In agreement, we found that ROS reduction with SOD and catalase restored FMD in EC-Opa1 mice and that counteracting the effect of ROS with L-Arginine and BH4 had a similar effect. These observations further support that Opa1 might be involved in flow mediated reduction in ROS production rather that in the direct activation of the NO pathway by flow. Finally, we also found that MitoTempo restored FMD to control levels in arteries isolated from EC-Opa1 mice, thus suggesting that flow reduced the production of ROS by the mitochondria. This agrees with our previous work showing that flow-mediated reduction in ROS involves the non-nuclear estrogen receptor alpha through a reduction in mitochondrial ROS production [33,34]. Nevertheless, the link between flow (shear stress), Opa1 and mitochondrial ROS production remains to be further investigated. Similarly, the interaction between this pathway and the non-nuclear estrogen receptor alpha remains unknown. Of note, we found no difference between male and female mice in the response to flow in mice lacking Opa1.
These findings led us to investigate the role of Opa1 in pathological conditions. Shear stress due to flow is the major regulator of ECs homeostasis. Nevertheless, flow is disturbed in zones of branching, bifurcations and curvatures, leading to atherosclerosis susceptibility [47]. Indeed, shear stress due to laminar or pulsatile flow induces a protective signal in ECs, whereas a low or oscillatory shear stress induces a deleterious signal [63]. We investigated Opa1 expression in mouse ECs in vitro and in vivo. Opa1 expression level was reduced in mouse ECs exposed in vitro to disturbed flow. Compared to laminar flow, the disturbed flow was characterized by reduced eNOS and increased ET-1 expression level. In agreement, Opa1 expression was decreased in the lower curvature of the mouse aortic cross which is an atheroprone zone compared to the greater curvature which is less susceptible to atherosclerosis. This was associated with reduced mitochondrial length and increased fission count in ECs of the mouse aortic cross.
Consequently, disturbed shear stress is likely to reduce Opa1 levels leading to increased oxidative stress, thus decreasing the protective effect of shear stress on ECs. This was confirmed by our results showing that both the mouse model of ADOA (Opa1 +/− mice) and mice lacking Opa1 in ECs were more susceptible to atherosclerosis when fed a high-fat diet. Interestingly we observed a progressive protective effect of Opa1 along the aorta with a maximal protection in the descending aorta. We observed large areas of lipid deposits in the aortic sinus without effect of the absence of Opa1 (Opa1 +/− mice) whereas lipid deposition in the aortic cross was 36% greater in Opa1 +/− mice compared to Opa1 +/+ mice. Finally, in the descending aorta the lipid deposition was increased by 113% in the absence of Opa1. A similar pattern was observed in EC-Opa1 mice. Thus, it is most likely that the protective effect of Opa1 was more pronounced in areas of the aorta with a more laminar flow as we also found that the expression level of Opa1 was reduced in ECs submitted to disturbed flow and increased in ECs submitted to laminar flow.
In conclusion, we found that Opa1 facilitates the acute response of arteries to flow (FMD) through a reduction in ECs oxidative stress. This finding suggests that a defect in FMD due to reduced flow-signaling of ECs might affect ADOA patients. This reduced FMD could explain, at least in part, the reduced retinal vascular density observed in ADOA patients [23,24]. These findings suggest that targeting Opa1 or more generally mitochondrial dynamics could provide new therapeutic tools against endothelial disorders which are involved in a handful of cardiovascular and cerebrovascular diseases.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antiox11061078/s1, Figure S1: Validation of Opa1 silencing in HUVECs. Figure S2: Positive and negative controls of ECs migration. Figure S3: Whole blots corresponding to the graphs shown in Figure 2A,C,F,G. Figure S4: Whole blots corresponding to the graphs shown in Figure 2B,D,E,H. Figure S5: Validation of Opa1 extinction in mesenteric artery endothelial cells isolated from EC-Opa1 or EC-WT mice. Figure S6: Validation of Opa1 extinction in mesenteric artery smooth muscle cells isolated from EC-Opa1 or EC-WT mice.