Recombinant Thrombomodulin Exerts Anti-autophagic Action in Endothelial Cells and Provides Anti-atherosclerosis Effect in Apolipoprotein E Deficient Mice

Stress-induced alteration in endothelial cells (ECs) integrity precedes the development of atherosclerosis. Previous studies showed that the soluble recombinant thrombomodulin (rTM) not only increases ECs proliferation but also exerts anti-apoptotic activity in ECs. However, the functional significance of soluble rTM on autophagy-related apoptosis in ECs is still undetermined. Implicating a cytoprotective role for rTM in persistent serum starvation (SS)-induced autophagy in cultured ECs, we found that treatment of rTM decreased the expression of SS-induced autophagy-related proteins, ATG5 and LC3, and the formation of autophagosomes through activation of AKT/mTOR pathway. In addition, treatment of rTM decreased SS-induced EC apoptosis, but this effect of rTM could not be recapitulated by co-treatment with a potent autophagy inducer, rapamycin and in ECs with ATG5 knockdown. In human atherosclerosis specimens, expression of autophagy markers, ATG13 and LC3, were more abundant in aortic intimal ECs with severe atherosclerosis than those without atherosclerosis. Moreover, compared to saline treatment group, administration of rTM reduced LC3 and ATG13 expression, intimal EC apoptosis, and atherosclerotic lesion severity in the aorta of apolipoprotein E deficient mice. In conclusion, treatment with rTM suppressed stress-induced autophagy overactivation in ECs, provided ECs protective effects, and decreased atherosclerosis in apolipoprotein E deficient mice.

consisting autophagy related proteins (ATGs) that initiate autophagosome formation 6 . Recent study found that transient exposure of ECs to hypoxia may induce autophagy to promote ECs survival and growth, but if the hypoxia stress was prolonged, autophagy became mTOR dependent and persistently activated autophagy induced ECs apoptosis 7 . Vascular ECs are exposed to circulating inflammatory mediators that may trigger autophagy. During atherosclerotic process, deprivation of oxygen and nutrient from insufficient vascularization of growing neointima or atherosclerotic plaque causes persistent autophagy activation in the arterial wall 8 . Hence, targeting ECs death induced by overactivated autophagy and restoring endothelial dysfunction might be considered to be a potential therapeutic strategy for preventing atherosclerosis development.
Thrombomodulin (TM) is a cell membrane-bound glycoprotein mainly expressed on ECs. It consists of 5 domains including a highly charged N-terminal lectin-like domain (D1), a domain with six epidermal growth factor (EGF)-like structures (D2), a serine and threonine-rich domain (D3), a transmembrane domain (D4) and a cytoplasmic tail (D5). TM binds to thrombin via its EGF-like domain and activates protein C (APC) which degrades coagulation factors VIIIa and Va 9 . After binding thrombin, TM decreases the thrombin's procoagulant effects. Thrombin activates protease-activated receptor 1 (PAR-1)/sphingosine-1 phosphate receptor 3 pathway in ECs and causes junction disruption, adhesion molecule expression and monocyte adhesion on ECs 10 . All the proinflammatory responses induced by thrombin are also reduced by TM. The membrane-bound TM is released from ECs during stress and becomes soluble form in systemic circulation. Soluble TM has the similar ability to inhibit thrombin and produce APC as its membrane-bound form. Originally, soluble TM is considered only as a marker of endothelial damage. Later studies showed that soluble recombinant TM (rTM) confers direct ECs protective effect by decreasing stress-induced ECs apoptosis and increasing ECs proliferation 11,12 . Our previous studies demonstrated that the cytoprotective effects of rTM are independent of its thrombin inhibition or APC producing functions, and works by direct binding of rTM to fibroblast growth factor receptor 1 (FGFR1) on ECs with subsequent PI3K/AKT pathway activation [12][13][14] which is also important in inhibition of autophagy. In the current study, we hypothesized that: (1) rTM is an autophagy inhibitor by activating PI3K/AKT/mTOR complex 1 signaling pathway; (2) rTM suppresses stress-induced autophagy overactivation in ECs to provide ECs protective effects; and (3) rTM treatment may suppress overactivated autophagy in ECs during the process of atherosclerosis formation and decrease atherosclerosis severity in animal atherosclerosis model. The following experiments were designed to prove our hypotheses.

Results
rTM suppresses stress-induced autophagy in cultured ECs. We stressed the human umbilical vein endothelial cells (HUVECs) with serum starvation (SS, 0.5% fetal bovine serum [FBS]) for 24 hr. SS decreased the phosphorylation of AKT, mTOR and the downstream effector protein, p70 ribosomal S6 kinase (S6K). Autophagy-related proteins, including ATG5 and microtubule-associated protein 1 light chain 3 (LC3) expression were increased during starvation (Fig. 1A). Treatment with rTM 100 ng/mL restored AKT, mTOR and S6K phosphorylation and inhibited ATG5 and LC3 expression at 24 hr after starvation (Fig. 1A). AKT inhibitor, LY294002, significantly reduced rTM effect on mTOR and S6K phosphorylation indicating that rTM effect on mTOR works through AKT activation (Fig. 1B). Monodansylcadaverine (MDC) (Fig. 1C) and LC3 (Fig. 1D) staining showed that persistent SS for 24 hr significantly induced autophagosomes formation in ECs and rTM 100 ng/mL treatment could decrease the number of autophagosomes in ECs. These results demonstrated that rTM treatment suppressed stress-induced autophagy overactivation in ECs.
Autophagy plays a role in rTM effect on ECs. Cell death detection ELISA showed that SS for 24 hr induced ECs apoptosis. There was a significant reduction of apoptosis in rTM-treated ECs after SS. Rapamycin (a potent autophagy activator) treatment abolished the rTM protective effect on SS-induced ECs apoptosis ( Fig. 2A). Flow cytometry study with annexin V and propidium iodide staining reconfirmed the findings that rTM treatment reduced SS-induced EC apoptosis and rapamycin abolished the rTM effect ( Fig. 2B and C). ATG 5 knockdown ECs do not have significant autophagy after stress 15 . To investigate whether inhibition of autophagy could abolish rTM cytoprotective effect on ECs, ATG5 knockdown ECs were used. The efficacy of ATG5 knockdown was checked by western blot (Fig. 2D). In ATG 5 knockdown ECs, we found SS could induce cell apoptosis, but rTM or rapamycin treatment could not influence SS-induced cell apoptosis shown by cell death ELISA (Fig. 2E) and flow cytometry ( Fig. 2G and H). Treatment with rTM induced EC proliferation. Autophagy overactivation caused by rapamycin also inhibited rTM-induced ECs proliferation (Fig. 2F). Wound healing and BrdU assays showed that ECs with ATG5 knockdown had higher migration and proliferation ability than ECs without ATG5 knockdown ( Fig. 2I and J). These results indicated that autophagy played an important role in rTM protective effect on stress-induced EC apoptosis and EC behaviors through AKT/mTOR pathway.
Autophagy is activated in intimal ECs in human atherosclerosis. In human aortic specimens, we investigated the expression of autophagy-related proteins, LC3 and ATG13 in aortic ECs in patients with or without severe coronary atherosclerosis. The signals of LC3 and ATG13 were low in intimal CD31-positive ECs in patient without significant atherosclerosis ( Fig. 3A-F); while the p-S6K signal was found ( Fig. 3G-I). But in patient with severe coronary atherosclerosis, double immunofluorescent staining revealed that LC3 and ATG13 were highly expressed in aortic intimal ECs (Fig. 3J-O) and no p-S6K signal can be found (Fig. 3P-R). The findings indicated that autophagy was activated in intimal ECs in human atherosclerosis. rTM suppresses ECs autophagy activation in apolipoprotein (apo)E deficient mice. In order to determine whether treatment of rTM inhibits ECs autophagy overactivation in vivo, we used a well-established mouse model, apoE deficient mice fed with high fat diet, to recapitulate the development of human atherosclerosis 16 . In apoE deficient mice without rTM treatment, double immunofluorescent staining revealed that LC3 and ATG13 were highly expressed in aortic intimal ECs. Treatment with rTM reduced the signals of LC3 and ATG13 in aortic intimal ECs and increased the signal of mTOR downstream effector protein p-S6K (Fig. 4A). TUNEL assay was used to evaluate cell apoptosis. The number of TUNEL-positive cells in intimal layer was significantly reduced in rTM-treated mice (Fig. 4B). ApoE deficient mice with rTM treatment had less atherosclerotic plaque lesions (with vs. without rTM treatment, Oil Red stained aortic plaque area: 11 ± 4% vs. 57 ± 7%, p < 0.05) (Fig. 4C).
FGFR1 mediates rTM effect on autophagy. Finally, we looked at the upstream signaling pathway of rTM effect on autophagy. Our previous study showed that rTM directly interacted with FGFR1 and induced the activation of FGFR1 14 . Figure 5A shows that rTM activated FGFR1 and FGFR1 substrate, FGFR substrate 2α (FRS2α), a major adapter protein bound to FGFR juxtamembrane domains, with subsequent AKT/mTOR pathway activation and suppressed autophagy activity (Fig. 5A).

Discussion
The major findings of this study were: (1) rTM suppressed stress-induced autophagy overactivation in ECs; (2) autophagy modulated ECs cell behavior; (3) autophagy was activated in aortic ECs in human atherosclerosis and apoE deficient mice; (4) rTM treatment suppressed aortic ECs autophagy and decreased atherosclerosis in apoE deficient mice and (5) rTM effect on autophagy was achieved through mTOR activation via the FGFR1/ FRS2α-mediated PI3K/AKT pathway.
Endothelial dysfunction is one of the most critical initiating and perpetuating events during atherosclerotic plaque formation. Any treatment that could protect vascular endothelium may lead to atherosclerosis alleviation. We and other studies showed that rTM treatment could inhibit atherosclerosis in various animal models [17][18][19] . Initially, we considered the rTM's thrombin inhibitory effects with thrombin-activated PAR-1 signal downregulation and reduced endothelial inflammation were the major mechanisms 19 . Recently, rTM was found to have direct endothelial protective effects. Treatment with rTM not only increases EC proliferation, migration and tubule formation but also reduces stress-induced EC apoptosis [11][12][13]20 . In the current study, we further demonstrated that rTM suppressed stress-induced autophagy overactivation which could modulate ECs cell behaviors, and this function was one of the mechanisms that explain the rTM's anti-atherosclerosis effect. Autophagy could either promote or suppress ECs migration or survival in different environmental conditions. Transient exposure (4 hr) to nutrient deprivation induced ECs autophagy and inhibition of this autophagy activation significantly reduced ECs tubule formation, migration and angiogenesis 21 . Similarly, stress with short-term hypoxia (30 min to 2 hr) induced autophagy in endothelial progenitor cells (EPCs) and inhibition of autophagy increased cell apoptosis 22 . These data indicated that transient stress-induced autophagy plays a protective role for cell proliferation and survival. However, autophagy could have the opposite effects for ECs if the stress is overwhelming or persistent. Chen et al. found that transient hypoxia-induced autophagy promoted ECs survival and growth, and this autophagy was regulated by hypoxia-inducible factor-1α (HIF-1α); but if the stress was prolonged (24 to 72 hr), autophagy overactivation became mTOR-dependent and caused EC death 7 . Brief hypoxia (2 hr)-induced autophagy could increase EPCs migration and tubule formation 23 . But under normal condition, EPCs or other cells without autophagy with ATG5 knockdown showed an increased migratory ability 24,25 and this finding was also observed in our studies. All the data showed that the environmental conditions play an important role in the cell response to autophagy. In human and mouse atherosclerosis, autophagy activation in aortic ECs was found in our study. In vascular ECs, sustained autophagy triggers, such as oxidized low-density lipoprotein, advanced glycation end product, or reactive oxygen species, could come from circulating blood or within the subendothelial layer of atherosclerotic plaque 26 . The persistent autophagy overactivation leads to ECs apoptosis and endothelial dysfunction which may accelerate atherosclerotic process. Our in vivo experimental results demonstrated that rTM treatment could suppress ECs autophagy, decrease ECs apoptosis and atherosclerotic plaque area in the apoE deficient mice. Treatment with rTM could protect ECs by modulating mTOR-dependent autophagy and restrict atherosclerosis development. Other compound that activated mTOR and suppressed autophagy also showed anti-atherosclerotic effect in apoE deficient mice 27 . These data implied that manipulation of autophagy during atherosclerosis formation may carry therapeutic potential. However, the results showed in our animal experiments were the summation effect of rTM. The rTM effects on autophagy in other vascular cell types, such as smooth muscle cell and macrophage, need further studies. Activated mTOR is the major regulator to downregulate autophagy. We proved that rTM could inhibit autophagy via the activation of AKT/mTOR pathway and decrease expression of autophagy-related proteins and autophagosome formation in ECs. Our recent study showed that rTM protective effect on ECs was through FGFR1 activation. There was direct interaction between rTM and FGFR1 and the rTM effect can be inhibited by an FGFR1-specific tyrosine kinase inhibitor, PD173074, or by knockdown of FGFR1 using siRNA 14 . FRS2α is a membrane-anchored adaptor protein that is required in the FGF/FGFR1 signal transduction pathway. Previous studies showed that FRS2α is necessary for FGF to activate mTOR and suppress the autophagic activity in fibroblasts and cardiac progenitor cells 28,29 . In the current study, we reconfirmed that rTM works through the similar transduction pathway by activating FGFR1/ FRS2α and AKT/mTOR to inhibit autophagy in ECs. A schematic overview of the transduction pathway of rTM and autophagy was proposed in the Fig. 5B.

Autophagosome immunofluorescence staining. Monodansylcadaverine (MDC) is an autofluores-
cent compound that can label autophagosome directly. Cells were stained with 50 μM MDC (Sigma-Aldrich) for 30 min at 37 °C and the fluorescent signals were captured using a confocal microscope (FV1000, Olympus, Japan). LC3 is located on the autophagosomal membrane and can be regarded as a specific autophagosome marker. To assess LC3 localization, the cells were incubated with LC3 antibody (1:500, MBL, Nagoya, Japan) and then incubated with Alexa Fluor 488 goat anti-rabbit IgG (1:200, Invitrogen). Cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) to identify the nuclei. Images were captured with the same confocal microscope. For quantitating autophagosomes, ten cells were randomly selected on the microscope (600X) and the number of autophagosome (MDC-positive dot) or LC3 puncta in each cell was counted in ECs migration and proliferation assay. For wound healing assay, cells treated with ATG5 siRNA or control siRNA were seeded and grew until nearly 100% confluence. A linear scratch was made by using 200 μL pipette tip. Then the cells were incubated for 24 hr and images were taken at 0 and 24 hr. For proliferation assay, cells were seeded in the 96 well plate (5000 cells per well) and transfected with ATG5 siRNA for 72 hr. After transfection, 10 μM BrdU was added to the cells and followed by incubation for 2 hr. The nuclear incorporation of BrdU was measured with a cell proliferation ELISA kit (Roche Diagnostics). Immunofluorescent staining of human aortas. Human aorta specimens were obtained from a recipient of cardiac transplantation with ischemic cardiomyopathy and severe coronary artery disease. Normal human aorta specimens obtained from a donor of cardiac transplantation with normal coronary artery were used for comparison. Human tissue collection was approved by our Institutional Review Board (IRB approval number: A-ER-103-043). For double immunofluorescent staining, the sections were stained for anti-human CD31 (1:50; GeneTex, Irvine, CA) and anti-LC3 (1:500; MBL) or anti-ATG13 (1:200; Sigma-Aldrich) antibodies, followed by incubation with respective fluorescent secondary antibodies (1:200; Invitrogen). Isotype IgG control antibodies were used as negative controls to confirm the specificity of respective primary antibodies (Calbiochem, San Diego, CA) in human and mouse aortic specimens (Supplement Figure II). Nuclei were counterstained with DAPI. The fluorescent signals were captured using a fluorescent microscope (BX61, Olympus). ApoE deficient mice. ApoE deficient (Jackson Laboratory, stock number: 002052; ApoE −/− knockout C57BL/6 background, 8-week-old) mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and fed with a high cholesterol diet (PMI LabDiet 40097, Richmond, IN) from 8 to 28 weeks of age. Initially, osmotic pumps (Alzet 2004; Durect, Cupertino, CA) containing rTM (145 μg/kg/day) were implanted subcutaneously on the back of mice for 4 weeks (from 8 to 12-week-old). At 28 week, the mice were sacrificed, the periaortic tissue was trimmed around the aorta and the whole aorta was obtained. The atherosclerotic plaques in aortas were stained with Oil-red O (Sigma-Aldrich). The images of the aortas were taken and analyzed by use of ImagePro Plus. The atherosclerotic severity was expressed as a percentage of atherosclerotic plaque area to the total aortic surface area. All animal experiments were performed in accordance with the institutional guidelines and approved by the Institutional Animal Care and Use Committee, National Cheng Kung University (Approval number:104055). Immunofluorescent staining of mouse tissue. The aortas isolated from mice were embedded in optimum cutting temperature compound. Frozen sections (5 μm thick) were fixed and blocked with 5% goat serum. For double-immunofluorescent staining, the sections were stained for anti-mouse CD31 (1:100; BD Biosciences) and anti-LC3 (1:500; MBL), anti-ATG13 (1:200; Sigma-Aldrich) or anti-pS6K (1:100; Cell Signaling) antibodies, followed by incubation with fluorescent secondary antibodies (1:200; Invitrogen). Nuclei were counterstained with DAPI. Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay was used to evaluate cell apoptosis with the In Situ Cell Death Detection kit (Roche Diagnostics). Images were captured with a fluorescence microscope (BX61, Olympus) and images were analyzed using Image J software. The intensities of immunofluorescence staining images were analyzed by AlphaImager 2200 software.
Statistical Analysis. Data were presented as mean ± standard error. Comparisons between 2 groups were made by Mann Whitney test. For multiple comparisons of groups, one-way ANOVA was used and followed by Bonferroni post hoc analysis. All statistical analyses were performed using SPSS 12.0 (SPSS Inc. Chicago, Il, USA). A p value < 0.05 was considered statistically significant.