Inhibition of autophagy blocks cathepsins–tBid–mitochondrial apoptotic signaling pathway via stabilization of lysosomal membrane in ischemic astrocytes

Our previous study and others have demonstrated that autophagy is activated in ischemic astrocytes and contributes to astrocytic cell death. However, the mechanisms of ischemia-induced autophagy remain largely unknown. In this study, we established a rat's model of permanent middle cerebral artery occlusion (pMCAO) and an in vitro oxygen and glucose deprivation (OGD) model. Autophagy was inhibited by either pharmacological treatment with 3-methyladenine (3-MA) and wortmannin (Wort) or genetic treatment with knockdown of Atg5 in primary cultured astrocytes and knockout of Atg5 in mouse embryonic fibroblast (MEF) cells, respectively. We found that pharmacological or genetic inhibition of autophagy reversed pMCAO or OGD-induced increase in LC3-II, active cathepsin B and L, tBid, active caspase-3 and cytoplastic cytochrome c (Cyt-c), and suppressed the injury-induced reduction in mitochondrial Cyt-c in ischemic cortex, in injured astrocytes and MEF cells. Immunofluorescence analysis showed that 3-MA or Wort treatment reversed OGD-induced release of cathepsin B and L from the lysosome to the cytoplasm and activation of caspase-3 in the astrocytes. Furthermore, treatment of 3-MA or Wort decreased OGD-induced increase in lysosomal membrane permeability and enhanced OGD-induced upregulation of lysosomal heat shock protein 70.1B (Hsp70.1B) in astrocytes. Inhibition of autophagy by 3-MA or Wort reduced infarction volume in rats and protected OGD-induced astrocytic cell injury. A non-selective caspase inhibitor z-VAD-fmk or a specific caspase-3 inhibitor Q-DEVD-OPh also rescued OGD-induced astrocytic cell injury. In conclusion, our presenting data suggest that inhibition of autophagy blocks cathepsins–tBid–mitochondrial apoptotic signaling pathway via stabilization of lysosomal membranes, possibly due to upregulation of the lysosomal Hsp70.1B in ischemic astrocytes.

Historically, three main morphological types of programmed cell death have been identified: type I apoptotic cell death, type II autophagic cell death and type III, which includes necrosis and cytoplasmic cell death. 1 Currently, there is no approved neuroprotective agent for acute ischemic stroke. One of the reasons may be due to the multiplicity of cell death mechanisms in which inhibition of a particular mechanism leaves the brain vulnerable to the alternative ones. 2 Therefore, it is essential to understand the different cell death mechanisms and their interactions. 2 Autophagy is a highly regulated process involving the bulk degradation of cytoplasmic macromolecules and organelles in mammalian cells via the lysosomal system, and is essential to the long-term health of cells, including neurons. 3 Autophagy contributes to both cell survival and cell death. 3 In recent years, the importance of autophagy in some human diseases has received much attention. [4][5][6] In the context of cerebral ischemia, it is proposed that autophagy is protective. 7,8 But increasing evidence indicates that autophagy is activated and involved in neuronal death in different animal models of ischemic brain injury, including hypoxia, 9 global 10 and focal ischemia. 11 Accumulating reports have shown that autophagy and apoptosis appear to interact with each other either positively or negatively under certain conditions. [12][13][14] Lysosomal proteases associated with autophagy, such as cathepsin B, have a role in apoptosis via cleavage of Bid, release of cytochrome c (Cyt-c) and activation of caspases in both neurons and non-neural cells. 15,16 Therefore, cathepsins may have important roles in the crosstalk between apoptosis and autophagy. 12 Stroke leads to the death or injury of both neurons and astrocytes. Astrocytes are involved in a number of activities that profoundly influence the consequences of ischemic brain damage, including the maintenance of water balance and the blood-brain barrier, the regulation of cerebral blood flow, brain energetics, glutamate and ion homeostasis, inflammation, trophic factor production, neurogenesis and vasculogenesis. [17][18][19][20][21][22][23] Therefore, the protection of astrocytes is likely required for neuronal survival and for functional recovery after ischemic stroke. 23,24 Astrocyte has been considered as a promising target for novel therapeutic approaches in ischemic brain protection. 12,24 Astrocyte apoptosis and autophagy may contribute to the pathogenesis of cerebral ischemia. 12,[24][25][26][27][28] The crosstalk between apoptosis and autophagy is largely unknown in ischemic astrocyte injury.
In 2010, we provided the first evidence that autophagy is activated in ischemic astrocytes and contributes to astrocytic Figure 1 3-MA treatment inhibits ischemia-induced activation of cathepsin B or cathepsin L-tBid-mitochondrial apoptotic signaling pathway in the ischemic cortex. 3-MA (150, 300 and 600 nmol) or vehicle was administrated icv 10 min after ischemia induced by pMCAO. (a) Brain slice stained with 2, 3, 5-triphenyltetrazolium chloride (TTC). The region within the blue line indicates the part of the ischemic brain cortex collected as samples for western blotting analysis. (b-g) Representative western blotting images of the protein levels of active cathepsin B (b) at 6 h or cathepsin L (c) at 3 h, tBid (d), mitochondrial (e) and cytoplastic Cyt-c (f) and active caspase-3 (g) at 24 h after ischemia. (h-m) Columns represent quantitative analysis of immunoblots in (b-g), respectively (means ± S.D., n = 3). Cyt-c oxidase IV (COX IV), which is located in the inner mitochondrial membrane acts as a mitochondrial marker. β-Actin or HSP-60 was used as a loading control. ## Po0.01 versus Sham group; *Po0.05, **Po0.01 versus ischemic control group (IC) cell death. 12 More recently, we further found that the activation of the cathepsin B and L and their movement into the cytosol contribute to ischemia-induced astrocytic cell injury via tBidmitochondrial apoptotic signaling pathways. 24 It is established that the movement of cathepsin B or L into the cytosol can be used as a measure of lysosomal membrane permeability (LMP) in neurons or in astrocytes. 24,29 In this study, we tested the hypothesis that autophagy activation may induce cathepsins-tBid-mitochondrial apoptotic signaling pathway via increasing the LMP and that this activation may involve the lysosomal Hsp70.1B-dependent regulating mechanism.

Results
Inhibition of autophagy reduces infarct volume and blocks ischemia-induced activation of cathepsin B and cathepsin L-tBid-mitochondrial apoptotic signaling pathway in ischemic cortex. Our previous data and others showed that 3-methyladenine (3-MA) treatment at 300-600 nmol (intracerebroventricular (icv)) could reduce infarct volume and improve neurological deficits in rat models of permanent middle cerebral artery occlusion (pMCAO). 11,12 In the current study, we confirmed that 3-MA treatment at 300-600 nmol (icv) could decrease infarct volume at 24 h after pMCAO in rats (Supplementary Figure S1). Our previous study further demonstrated that 3-MA treatment at 300-600 nmol (icv) could protect astrocytes injury in the ischemic cortex. 12 Owing to the ability to induce Bak and Bax, Bid has a key role in apoptotic signaling pathway and results in Cyt-c release. 30 Our previous study demonstrated that cathepsin B and L were activated in the ischemic cortex after pMCAO, and led to the activation of tBid-mitochondrial apoptotic signaling pathway. 24 The peak for cathepsin B and L activation was at 6 and 3 h postischemia, respectively. The activation of cathepsin B and L led to a maximal increase in tBid, cytoplastic Cyt-c and active caspase-3, and conversely, a maximal reduction in mitochondrial Cyt-c during 12-24 h post-ischemia period.
In this study, we found that 3-MA treatment at 300-600 nmol (icv) inhibited ischemia-induced increase in active cathepsin B Inhibition of autophagy protects astrocytes and mouse embryo fibroblast cells against OGD-induced injury. We next determined whether inhibition of autophagy could protect astrocytes against oxygen and glucose deprivation (OGD)-mediated injury. Our previous study and others found that a higher dose of 3-MA (10 mM) could inhibit OGD, TNF and palmitic acid-induced autophagy in astrocytes, 12 in FADD-deficient Jurkat cells 31 and in endothelial cells. This dose also exhibited a mild protection against OGD-induced injury to astrocytes and a significant protection against palmitic acid-induced injury to endothelial cells. In this study, we further found that lower doses of 3-MA (0.1, 0.5 or 1 mM) also could protect astrocytes against OGD-induced injury. Figure 2 showed that the number of astrocytes appeared to increase ( Figure 2a) and the leakage of LDH was decreased ( Figure 2c) in the OGD+3-MA group, compared with these in the OGD group. Moreover, wortmannin (Wort), another autophagy inhibitor, also appeared to increase the number of astrocytes ( Figure 2b) and to decrease leakage of LDH ( Figure 2d). 3-MA (0.1, 0.5 or 1 mM) or Wort (25, 50 or 100 nM) alone did not alter the number of astrocytes (data not shown) and the leakage of LDH (data not shown). These results indicate that 3-MA or Wort at these concentrations are not toxic in cultured astrocytes. The above results show that the inhibition of autophagy protects astrocytes against OGD-induced injury.
In support of the above results, we found that inhibition of autophagy also protected mouse embryo fibroblast cells against OGD injury. Compared with these in the OGD+ Atg5+/+ group, the number of mouse embryo fibroblast cells was significantly increased and the leakage of LDH was markedly decreased in the OGD+Atg5 − / − group (Supplementary Figures S2a and b).
Inhibition of autophagy blocks OGD-induced activation of cathepsin B and cathepsin L-tBid-mitochondrial apoptotic signaling pathway in astrocytes and mouse embryo fibroblast cells. We next tested the effects of pharmacological or genetic inhibition of autophagy on OGDinduced activation of cathepsin B and cathepsin L-tBidmitochondrial apoptotic signaling pathway. First, we confirmed that 3-MA (0.1, 0.5 or 1 mM) or Wort (25, 50 or  Our recent study demonstrated that cathepsin B and L were activated after OGD-induced astrocyte injury, resulting in the activation of tBid-mitochondrial apoptotic signaling pathway. 24 The peak for cathepsin B or L activation was at 6 or 3 h post-OGD, respectively; and the maximal increase in tBid, cytoplastic Cyt-c, active caspase-3 and the maximal reduction in mitochondrial Cyt-c were at 12 h post-OGD. In this study, we found that 3-MA (0.1, 0.5 or 1 mM), Wort (25,50  Inhibition of autophagy reduces OGD-mediated release of cathepsin B and L from the lysosome into the cytoplasm and activation of caspase-3 in astrocytes. We next tested the effects of 3-MA or Wort on the release of cathepsin B and L from the lysosome into the cytoplasm and the activation of caspase-3 induced by OGD in astrocytes with immunofluorescence. As shown in Figures 4 and 5, there were less fine, granular, perinuclear cathepsin B ( Figure 4) and L ( Figure 5) immunostaining, which colocalized with Lamp 1-positive lysosomes in non-OGD astrocytes. This finding was consistent with the predominantly lysosomal location of these proteases. 15,24,[32][33][34] In astrocytes treated with OGD, cathepsin B and L granules became larger and irregular, formed aggregates or showed diffuse cytoplasmic staining at 6 h (    We further tested the effects of 3-MA on OGD-induced activation of caspase-3 in astrocytes with immunostaining. The results showed that much less active caspase-3 immunoreactivity was seen in non-OGD astrocytes (Supplementary Figure S5). In astrocytes treated with OGD, the active caspase-3-positive astrocytes increased over time and peaked at 12 h after OGD (Supplementary Figure S5). In contrast, 3-MA reduced active caspase-3-positive astrocytes at 12 h after OGD ( Figure 6).
In addition, we confirmed the role of caspase-3, z-VAD-fmk (nonspecific caspase inhibitor) and Q-DEVD-OPh (a specific inhibitor of caspase-3) both reduced the protein levels of caspase-3 ( Supplementary Figures S6a and c, b and d), suggesting that caspase-3 is activated in our OGD model system. To further confirm the role of caspase-3, the LDH leakage was measured. Both z-VAD-fmk and Q-DEVD-OPh at 25 and 50 μM markedly decreased the leakage of LDH in astrocytes 12 h post-OGD (Supplementary Figures S6e and   g, f and h), indicating that inhibition of caspases or caspase-3 has a protective effects on ischemic astrocytes. These data further suggest that the protective effects of autophagy inhibition on ischemic astrocytes are potentially mediated by inhibiting the activation of caspase-3.
Inhibition of autophagy decreases OGD-induced LMP in astrocytes. Excessive autophagy induces LMP 35,36 and it is possible that LMP mediates cathepsin B and L cytosolic translocation. Hence, we evaluated LMP formation by Acridine Orange (AO) and Lyso-Tracker Red staining assays. Normally, AO, a metachromatic fluorophore cloistering inside of the lysosome, exhibits a high level of red fluorescence and a low level of green fluorescence. When lysosomes are disrupted, AO relocates to the cytosol from the lysosomes and manifests a reduced red fluorescence and an increased green fluorescence. 36 As shown in Figures 7a and c, OGD induced a reduction in red fluorescence in astrocytes. In contrast, treatment with 3-MA or Wort markedly inhibited OGD-induced reduction in red granular fluorescence of AO staining. Lyso-Tracker Red uptake images in astrocytes  (Figures 7b and d).
The above data suggest that 3-MA or Wort can stabilize OGD-induced lysosomal membrane instability in astrocytes.
Inhibition of autophagy enhances OGD-induced upregulation in lysosomal heat shock protein 70.1B (Hsp70.1B) in astrocytes. Hsp70.1B is known to stabilize lysosomal membrane by recycling damaged proteins and protect cells from various insults such as heat, ischemia and other oxidative stresses. [37][38][39] The chaperone function and inhibition of lysosomal membranes permeabilization or rupture are the major mechanisms by which Hsp70.1B protects cells. [39][40][41] We found that OGD induced a significant increase in Hsp70.1B level during the period of 3-12 h post-OGD in astrocytes (Figures 8a  and b). Double immunofluorescence staining of Hsp70.1B and Lamp 1 showed that in non-OGD astrocytes, there was less immunoreactive colocalization of Hsp70.1B with Lamp 1 (Figures 8c-e). After OGD, the immunoreactivities of Hsp70.1B  (Figures 8c-e). Surprisingly, Hsp70.1B colocalized with Lamp 1 was more intense when 3-MA or Wort was added to the astrocytes (Figures 8c-e). These data indicate that the inhibition of autophagy upregulates the lysosomal Hsp70.1B, possibly contributing to a reduction in OGD-induced lysosomal membrane instability in astrocytes.

Discussion
To date, it is well accepted that autophagy is a major mediator of neuronal cell death in cerebral ischemia. [9][10][11]28,42,43 In 2010, we first reported that autophagy is activated in ischemic astrocytes and contributes to astrocytic cell death. 12 Similarly, Pamenter et al. 44 found that astrocytes are more sensitive to conditions mimicking metabolic and ischemic stress of penumbral tissue than neurons and exhibit a stronger autophagic response to these stresses. Recent advances have elucidated that autophagy and apoptosis can share common regulators, [45][46][47][48] such as Bcl-2, which has been identified as a central regulator of autophagy and apoptosis by interacting with both Beclin-1 and Bax/Bak, respectively. Several apoptotic proteins (e.g., PUMA, Noxa, Nix, Bax, XIAP and Bim) are also believed to be regulators of autophagy. 48 However, the molecular mechanisms linking autophagy and apoptosis are not fully defined, especially in ischemic astrocytes. The novel aspect of the present work is that the inhibition of autophagy blocks the activation and release of cathepsin, and lead to the inhibition of tBid-mitochondrial apoptotic signaling pathway involving stabilization of the lysosomal membrane via upregulation of the lysosomal Hsp70.1B in ischemic astrocytes.
The inhibition of autophagy blocks cathepsins-tBid-mitochondrial apoptotic signaling pathway in ischemic cortex. Lysosomal proteases, such as cathepsin B, have important roles in apoptosis via cleavage of Bid, release of Cyt-c and activation of caspases in both neurons and non-neural cells. 15,16 Our prior studies demonstrated that cathepsin B and L are activated in the ischemic cortex after pMCAO, and lead to the activation of tBid-mitochondrial apoptotic signaling pathway. 24 The peak for cathepsin B or L activation was at 6 or 3 h post-ischemia, respectively. The maximal increase in tBid, cytoplastic Cyt-c and active caspase-3 and the maximal reduction in mitochondrial Cyt-c were at 12-24 h postischemia. Our present data and others showed that 3-MA treatment at 300-600 nmol (icv) reduced infarct volume and improved neurological deficits in rat models of pMCAO. 11,12 Our previous study also found that 3-MA treatment at 300-600 nmol (icv) could protect astrocytes in the ischemic cortex. 12 In the current studies, we further found that 3-MA treatment at 300-600 nmol (icv) could inhibit ischemia-induced increase in active cathepsin B or cathepsin L at 6 or 3 h post-ischemia, reverse ischemia-mediated increase in tBid, cytoplastic Cyt-c and active caspase-3, and ischemia-mediated reduction in mitochondrial Cyt-c at 24 h after ischemia. These data indicate that the ischemia-induced autophagy activation confers the activation of cathepsin B and L, the cleavage of Bid, the translocation of Cyt-c from the mitochondria to the cytosol and the activation of caspase-3 in the ischemic cortex.
The inhibition of autophagy blocks cathepsins-tBid-mitochondrial apoptotic signaling pathway in ischemic astrocytes. Previous studies demonstrated that a higher dose of 3-MA (10 mM) could inhibit TNF-induced autophagy in FADDdeficient Jurkat cells, 31 and pre-treatment with 3-MA (10 mM) reduced staurosporine-induced neuronal death. 49 In the previous study, we also found a higher dose of 3-MA (10 mM) exhibits a mild protection against OGD-induced astrocytes injury. In the current study, we further demonstrated that low doses of 3-MA (0.1, 0.5, or 1 mM) or Wort also protected astrocytes against OGD-induced injury.
Previously, we reported that OGD induces an increase in activated cathepsin B and cathepsin L, tBid, activated caspase-3, and cytoplastic Cyt-c and a reduction in mitochondrial Cyt-c in astrocytes at 3-12 h post-OGD. Inhibition of cathepsin B or L confers protective effect on ischemic astrocytes via inhibiting the activation of tBid-mitochondrial apoptotic signaling pathway. In the current study, we further found that the pharmacological or genetic inhibition of autophagy reversed OGDinduced increase in active cathepsin B and L, tBid, active caspase-3 and cytoplastic Cyt-c and OGD-induced reduction in mitochondrial Cyt-c in astrocytes. Our above data suggest that the activation of autophagy in the ischemic astrocytes may be involved in apoptotic regulation via activating lysosomal proteases, leading to the cleavage of Bid, the release of the mitochondrial Cyt-c into the cytosol and the activation of caspase cascade. Atg5 is an autophagy-specific gene required for autophagosome precursor synthesis and its deletion in yeast, mammalian cells and mice effectively blocks autophagy. 50 In support of this finding, knockout of atg5 also protected OGD-induced mouse embryo fibroblast cells injury and inhibited OGD-induced activation of cathepsin B or cathepsin L-tBid-mitochondrial apoptotic signaling pathway.
The inhibition of autophagy blocks OGD-induced translocation of cathepsin B/L from the lysosome into the cytoplasm and the activation of caspase-3 in astrocytes. Along with others, we have found that cathepsin B or L is normally confined to the endolysosomal compartment in neuron and astrocyte. When ischemia occurs, cathepsin B or L translocates to the cytoplasm from the lysosome, and leads to the activation of tBid-mitochondrial apoptotic signaling pathway. 24,51 One of the novel finding of this study is that 3-MA or Wort reversed OGD-induced release of cathepsin B or cathepsin L from the lysosomes into the cytoplasm and the activation of caspase-3 in astrocytes. In addition, we confirmed that caspase-3 plays a role in ischemic astrocytic injury associating with autophagy activation in our model system.
The inhibition of autophagy decreases OGD-induced LMP in astrocytes. The movement of lysosomal cathepsin B or L into the cytosol can be used to measure the LMP in neurons or in astrocytes. 24,29 Excessive autophagy leads to LMP induction. 35,36 Another novel finding of this study is that the inhibition of autophagy by 3-MA or Wort can stabilize the OGD-induced lysosomal membrane instability in astrocytes.
The inhibition of autophagy enhances OGD-induced upregulation of lysosomal Hsp70.1B in astrocytes. Hsp70.1 is one major protein of human Hsp70 family, and mainly functions as a chaperone enabling the cell to deal with harmful aggregations of denatured proteins upon various insults such as heat, ischemia and other oxidative stresses. [37][38][39] In 2010, Sahara et al. 39 demonstrated that Hsp70.1 was upregulated at the lysosomal membranes of neuronal cells after ischemia-reperfusion injury and inhibited LMP. An important unexpected finding of this study is that the inhibition of autophagy by 3-MA or Wort enhanced OGDinduced upregulation of lysosomal Hsp70.1B, perhaps contributing to a reduction in OGD-induced lysosomal membrane instability in astrocytes. This finding confirmed the link between Hsp70.1 and autophagy, which was reported by Sisti. 52 However, the molecular mechanisms underlying the upregulation of lysosomal Hsp70.1B by 3-MA or Wort requires further investigation.
In conclusion, the current study provides the first evidence that inhibition of autophagy blocks activation and release of cathepsins via stabilization of lysosomal membrane. This effect may result from upregulation of lysosomal Hsp70.1B, leading to inhibition of the tBid-mitochondrial apoptotic signaling pathway in ischemic astrocytes.

Materials and Methods
Animals. Male Sprague-Dawley rats weighing 280-320 g were purchased from the Center for Laboratory Animals, Soochow University,Suzhou, China (production license: XCYK-2002-0008). Animal procedures were performed according to a protocol approved by the Institutional Animal Care and Use Committee of Soochow University, Suzhou, China.
pMCAO model. pMCAO model was prepared as described previously. 12 Briefly, rats were anesthetized with intraperitoneal injection of 4% choral hydrate (350 mg/ kg). Permanent focal cerebral ischemia was induced by a 30 mm length of 4-0 nylon monofilament suture (Φ 0.22-0.24 mm) inserted from the right common carotid artery (CCA) to the internal carotid artery through a small incision in the CCA, and then advanced to the circle of Willis to occlude the origin of the right middle cerebral artery. Body temperature was maintained at 37°C by a heating pad during and after surgery until recovery from anesthesia. Sham-operated rats underwent the same procedures except for inserting a nylon monofilament suture to the artery. 3-MA (08592, Sigma-Aldrich, St. Louis, MO, USA) or vehicle was administrated icv 10 min after ischemia.
Western blotting analysis. The collected cortical tissue or cells was added in lysing buffer with protease inhibitor cocktail (Roche, Basel, Schweiz, 04693159001) and sonicated on ice. Protein concentrations were determined by a BCA protein assay kit (Pierce, Rockford, IL, USA). The proteins were separated using SDS-PAGE and transferred to a nitrocellulose membrane, and then blocked with 5% non-fat milk for 1 h. Blots were incubated with specific primary antibodies overnight at 4°C and corresponding secondary antibodies for 1 h at room temperature. Blots were captured by odyssey scanner (LI-COR, Bioscience, Lincoln, NE, USA). Densitometric analysis of the bands is quantitatively analyzed with Sigma Scan Pro 5 (Sigma-Aldrich, St. Louis, MO, USA). The antibodies used in this study are listed in the Supplementary Tables S1 and S2.
Immunofluorescence. Cells planted on 24-well plates were fixed with 4% paraformaldehyde for 5 min, permeabilized and blocked with 1% BSA containing 0.1% Triton X-100 for 1 h at room temperature, and incubated overnight at 4°C in specific primary antibodies. The cells were subsequently incubated (1 h, room temperature) with corresponding secondary antibodies. Then, the cells were incubated with DAPI (1 : 10 000, D9564, Sigma) or Hoechst (1 : 10 000, 33258, Sigma) solution for 10 or 30 min, respectively, to stain nuclei. Images were obtained by fluorescence or confocal microscope. The antibodies used in this study are listed in the Supplementary Tables S1 and S2.
Measurement of lysosomal stability. AO (318337, Sigma-Aldrich) and Lyso-Tracker Red (C1046, Beyotime) staining assays were widely used for evaluating the LMP. 36,53,54 AO is a lysosomotropic base and a metachromatic fluorophore. Normally, AO captures protons inside the acidic vacuolar compartment, preferentially in secondary lysosomes, and retains its charged form, which results in red fluorescence. When LMP is increased, AO relocates to the cytosol from the lysosomes, leading to cytoplasmic diffuse green fluorescence and reduced red fluorescence. Cells were suffered OGD treatment for 6 h, and then incubated with 5 μg/ml of AO in complete medium for 15 min at 37°C, or stained with Lyso-Tracker Red (75 nM) in the dark for 60 min at 37°C. 3-MA (1 mM) or Wort (100 nM) was added in medium 30 min or 2 h before OGD, respectively. Images were acquired using a confocal laser scanning microscopy (LSM 710, Carl Zeiss, Oberkochen, Germany).
Statistical analysis. Data are expressed as mean ± S.D., statistical analysis was carried out by one-way ANOVA followed by the Tukey's post-hoc test with Prism software (La Jolla ,CA, USA). Significant difference was set at Po0.05. Image-Pro Plus (Rockville, MD, USA) was used to calculate the colocalization coefficients.
Other Materials and Methods are available in the Supplementary Materials and Methods.

Conflict of Interest
The authors declare no conflict of interest.