Enhancing Macroautophagy Protects against Ischemia/Reperfusion Injury in Cardiac Myocytes*

Cardiac myocytes undergo programmed cell death as a result of ischemia/reperfusion (I/R). One feature of I/R injury is the increased presence of autophagosomes. However, to date it is not known whether macroautophagy functions as a protective pathway, contributes to programmed cell death, or is an irrelevant event during cardiac I/R injury. We employed simulated I/R of cardiac HL-1 cells as an in vitro model of I/R injury to the heart. To assess macroautophagy, we quantified autophagosome generation and degradation (autophagic flux), as determined by steady-state levels of autophagosomes in relation to lysosomal inhibitor-mediated accumulation of autophagosomes. We found that I/R impaired both formation and downstream lysosomal degradation of autophagosomes. Overexpression of Beclin1 enhanced autophagic flux following I/R and significantly reduced activation of pro-apoptotic Bax, whereas RNA interference knockdown of Beclin1 increased Bax activation. Bcl-2 and Bcl-xL were protective against I/R injury, and expression of a Beclin1 Bcl-2/-xL binding domain mutant resulted in decreased autophagic flux and did not protect against I/R injury. Overexpression of Atg5, a component of the autophagosomal machinery downstream of Beclin1, did not affect cellular injury, whereas expression of a dominant negative mutant of Atg5 increased cellular injury. These results demonstrate that autophagic flux is impaired at the level of both induction and degradation and that enhancing autophagy constitutes a powerful and previously uncharacterized protective mechanism against I/R injury to the heart cell.

Autophagy involves processes for the turnover of long lived macromolecules and organelles via the lysosomal degradative pathway (1,2). Macroautophagy (referred to hereafter as autophagy) is a specific mode of autophagy in which isolation membranes envelop a portion of the cytosol, containing nonspecific cytosolic components, selectively targeted toxic protein aggregates (3), intracellular pathogens (4), or organelles such as mitochondria (5,6). The autophagosomes are then delivered to the lysosome, forming the autophagolysosome, for subsequent degradation of their contents by lysosomal hydrolases (Fig. 10).
Interest in autophagy has increased recently, because of the recognition of its involvement in caspase-independent programmed cell death (PCD 2 type II) and its regulation by components of the apoptotic death pathway (PCD type I) (7)(8)(9). Anti-apoptotic Bcl-2 and Bcl-x L have been linked to the autophagic pathway via an interaction with Beclin1, a key mediator of autophagic activity (9,10).
Autophagy is a vital process in the heart, presumably participating in the removal of dysfunctional cytosolic components and serving as a catabolic energy source during times of starvation. For example, autophagy in cardiac myocytes has been suggested to provide a necessary source of energy between birth and suckling (11), and in a GFP-LC3 transgenic mouse, cardiac myocytes from starved animals displayed high numbers of autophagosomes, some of which contained mitochondria (12). On the other hand, impaired autophagy may play a causative role in cardiac disease. Incomplete autophagic removal of mitochondria may be the source of lipofuscin, a toxic waste product that builds up during the life span (13), and chronic impairment of the lysosome results in reduced myocardial function (14). Furthermore, disruption of the autophagic pathway may contribute to cardiac cell death under conditions where lysosomal integrity is lost and lysosomal proteases are released into the cytosol (15).
In the study presented here, we investigated the role and regulation of autophagy during ischemia/reperfusion (I/R) injury. Following a bout of ischemia (a reduction of blood flow resulting in oxygen and nutrient starvation), reperfusion must be achieved in order to rescue affected tissue. However, reperfusion can activate pathways that either preserve cell viability (preconditioning) or lead to cell death (I/R injury). Autophagy may be a protective response to I/R injury, as increased prevalence of autophagosomes has been documented in response to sub-lethal ischemia in the perfused heart (15). Moreover, it was recently reported that increased Beclin1 expression in the heart correlated with the onset of protection in an in vivo model of myocardial stunning (16).
The cardiac HL-1 cell line was subjected to simulated I/R (sI/R) as an in vitro model of I/R injury to the heart. Using three-dimensional high resolution fluorescence imaging, we analyzed the autophagic response to sI/R. Our results indicate that in HL-1 cardiac myocytes subjected to sI/R, autophagic flux is impaired at the level of both induction and degradation, yet remains a vital underlying protective response against sI/R injury. Moreover, increasing autophagic capacity of the cardiac myocyte is protective against sI/R injury.
Cell Culture and Transfections-Cells of the atrially derived cardiac cell line HL-1 (17) were plated in gelatin/fibronectincoated culture vessels and maintained in Claycomb medium (JRH Biosciences) supplemented with 10% fetal bovine serum, 0.1 mM norepinephrine, 2 mM L-glutamine, 100 units/ml penicillin, 100 units/ml streptomycin, and 0.25 g/ml amphotericin B. Cells were transfected with the indicated vectors using the transfection reagents Effectene (Qiagen) or Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions, achieving at least 40 and 60% transfection efficiency, respectively.
RNA Interference-Sequences with 100% homology to regions within the open reading frame of mouse Beclin1 (gi: 27764874) were generated using the BLOCK-iT TM RNAi Designer, a construct that embeds a small hairpin RNA within a micro RNA fold, which is then processed by the endogenous RNAi machinery (18). The obtained target sequences, 5Ј-tgaaacttcagacccatctta-3Ј (a) and 5Ј-taatggagctgtgagttcctg-3Ј (b), showed no significant homology to other mouse proteins as determined by Blast analysis. The sequence was used to generate oligonucleotide pairs, which were inserted into the pcDNA TM 6.2-GW/EmGFP-miR, which has co-cistronic expression of EmGFP, allowing for determination of transfection efficiency by fluorescence microscopy. The vectors, pcDNA TM 6.2-GW/EmGFP-miR-Beclin1 (a) and -Beclin1 (b), were sequence-verified, and cells were co-transfected with both vectors to achieve maximal knockdown. To control for nonspecific RNAi effects, the construct pcDNA TM 6.2-GW/EmGFP-miR-LacZ (targeting ␤-galactosidase) was used as a control.
Simulated Ischemia/Reperfusion (sI/R)-Cells were plated in 14-mm diameter glass bottom microwell dishes (MatTek), and ischemia was introduced by a buffer exchange to ischemia-mimetic solution (in mM: 125 NaCl, Wide Field Fluorescence Microscopy-Cells were observed through a Nikon TE300 fluorescence microscope (Nikon) equipped with a ϫ10 lens (0.3 N.A., Nikon), a ϫ40 Plan Fluor, and a ϫ60 Plan Apo objective (1.4 N.A. and 1.3 N.A. oil immersion lenses; Nikon), a Z-motor (ProScanII, Prior Scientific), a cooled CCD camera (Orca-ER, Hamamatsu), and automated excitation and emission filter wheels controlled by a LAMBDA 10 -2 (Sutter Instrument) operated by MetaMorph 6.2r4 (Universal Imaging). Fluorescence was excited through an excitation filter for fluorescein isothiocyanate (HQ480/ϫ40) and Texas Red (D560/ϫ40). Fluorescent light was collected via a polychroic beam splitter (61002bs) and an emission filter for fluorescein isothiocyanate (HQ535/50m) and Texas Red (D630/60m). All filters were from Chroma. Acquired wide field Z-stacks were routinely deconvolved using 10 iterations of a three-dimensional blind deconvolution algorithm (AutoQuant) to maximize spatial resolution. Unless stated otherwise, representative images shown are maximum projections of Z-stacks taken with 0.3-m increments capturing total cellular volume.
Quantification of Cellular Autophagosome Content-LC3 forms I and II are known to be differentially recognized by the LC3 antibodies (19). Furthermore, in our hands immunodetection of endogenous LC3 in HL-1 cells was inconclusive (data FIGURE 1. sI/R-specific GFP-Bax activation as a parameter of sI/R-induced cellular injury. GFP-Bax-transfected HL-1 cells were incubated in either KH solution or ischemia/mimetic solution for 2 h in normoxic (sI/R Normox) or hypoxic (sI/R) conditions followed by 5 h of reperfusion in normoxic KH solution. A, Z-stacks of representative cells were acquired at ϫ60 magnification with 0.3-m increments. Shown are the maximum projections of total cellular GFP-Bax fluorescence. B, cells were classified as displaying diffuse cytosolic or punctate mitochondrial GFP-Bax localization at 5 h of reperfusion. Represented is the percentage of cells with punctate GFP-Bax distribution per total cells scored (*, p Ͻ 0.05). C, GFP-Bax transfected cells were co-transfected with Bcl-2 or Bcl-x L , and overexpression of Bcl-2/Bcl-x L was verified by immunodetection of cell lysates. D, GFP-Bax redistribution was assessed under both control and sI/R conditions as under B (*, p Ͻ 0.05). not shown). Therefore, cellular contents of autophagosomal structures were quantified via fluorescence imaging of GFP-LC3 (20) or mCherry-LC3. To generate pmCherry-LC3, mCherry was amplified from the pRSET-mCherry vector (21) and swapped with enhanced GFP of the vector pEGFP-LC3 (20); HL-1 cells were transfected with (mCherry-/)GFP-LC3, and 48 h after transfection, cells were subjected to sI/R as indicated. Cells were fixed with 4% formaldehyde in phosphatebuffered saline, pH 7.4, for 15 min. To quantify the autophagic response in a population of cells, cells were inspected at ϫ60 magnification and classified as either having predominantly diffuse (mCherry-/)GFP-LC3 fluorescence or as having numer-ous punctate (mCherry-/)GFP-LC3 structures, representing autophagic vacuoles, AVs. At least 150 cells were scored in each of three or more independent experiments. For quantification of the autophagic response of single cells, Z-stacks of (mCherry-/)GFP-LC3 fluorescence of 7-10 representative cells per condition in three separate experiments were acquired through the ϫ60 oil immersion lens with 0.3-m increments through the entire volume of the cell. Z-stacks were thresholded, and total number and volume of the autophagosome per cell were determined (AutoQuant).
Determination of LC3-II Degradation-To analyze autophagic flux, (mCherry-/)GFP-LC3-expressing cells were subjected to the indicated experimental conditions with and without a mixture of the cellpermeable lysosomal inhibitors bafilomycin A 1 (100 nM, vacuolar H ϩ -ATPase inhibitor) to inhibit autophagosome-lysosome fusion (22), E64D (5 g/ml, inhibitor of cysteine proteases, including cathepsin B), and pepstatin A methyl ester (5 g/ml, cathepsin D inhibitor) to inhibit lysosomal protease activity. Fluorescence microscopy of GFP-LC3 was used to determine cellular autophagosomal content as described above.
Activity of the Lysosomal Compartment-LysoTracker Red is a cell-permeable acidotropic probe that selectively labels vacuoles with low internal pH and thus can be used to label functional lysosomes (23). Following sI/R and control experiments, cells were loaded with 50 nM LysoTracker Red for 5 min in KH solution; the medium was then exchanged with dye-free KH solution, and cells were analyzed by fluorescence microscopy. Activity and intracellular distribution of cathepsin B, a predominant lysosomal protease, were assessed using (z-RR) 2 -MagicRed-Cathepsin B substrate (B-Bridge). MagicRed cathepsin B substrate was added to the cells during the last 30 min of an experiment according to the manufacturer's instructions.
Quantification of Cellular Injury-GFP-Bax (24) or mCherry-Bax (25) distribution was used as a parameter to quantify irreversible cellular injury. Cells were cotransfected with (mCherry-/)GFP-Bax and the indicated vectors and allowed to express for 48 h. Cells were then subjected to 2 h of ischemia in hypoxic pouches followed by 5 h of reperfusion, and live cells were analyzed by fluorescence microscopy. Cells were classified as cells with either diffuse or punctate mitochondrial (mCherry-/)GFP-Bax fluorescence. Approximately 300 transfected cells per condition were scored at ϫ60 magnification in each of three independent experiments.
Immunoblotting-Cells were harvested by scraping and centrifugation at 550 ϫ g for 5 min at 4°C and washed once with cold phosphate-buffered saline, pH 7.4. To prepare whole cell lysates, cell pellets were suspended in cold RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM EDTA, 1 mM Na 3 VO 4 , 1 mM NaF, and 1ϫ complete protease inhibitor mixture (Roche Applied Science)) and left on ice for 20 min. The cell extracts were centrifuged at 20,000 ϫ g for 5 min to remove cellular debris. After addition of sample buffer and reducing agent (Bio-Rad), samples were incubated at 95°C for 5 min, electrophoresed on SDS-polyacrylamide gels, and transferred to nitrocellulose membranes (Bio-Rad). Immunodetection was performed using antibodies against actin (clone AC-40; Sigma), Bcl-2 (C-2, Santa Cruz Biotechnology), Bcl-x L (H-5, Santa Cruz Biotechnology), Beclin1 (D-18; Santa Cruz Biotechnology), and fluorescent protein (BD Biosciences). Attempts to detect endogenous LC3-I and -II using F-14 and H-50 (Santa Cruz Biotechnology) and A0973 (Biosignatures) antibodies were unsatisfactory because of inconsistencies in immunoreactivity and nonspecificity, perhaps due to the fact that these are mouse antibodies being used against mouse proteins. Blots shown are representative of at least three independent experiments.
Statistics-The probability of statistical differences between experimental groups was determined by the Student's t test. Values are expressed as mean Ϯ S.E. of at least three independent experiments unless stated otherwise.

sI/R Induces Programmed Cell Death in HL-1 Cardiac
Myocytes-The HL-1 cell line is an excellent model for studying many aspects of cardiac cell physiology (26). In our hands HL-1 cells reproducibly underwent PCD in response to simulated I/R via pathways resembling in vivo cardiac I/R injury (25). One key feature of sI/R-induced cell death is the participation of the pro-apoptotic Bcl-2 protein Bax in the mitochondrial death pathway. Bax activation, a point-of-noreturn in the PCD pathway, is reflected by a redistribution from the cytosol to punctate clusters at the mitochondria and can be quantified via fluorescent imaging of a GFP-Bax fusion protein (Fig. 1A) (24). Using GFP-Bax redistribution as an index to monitor activation of PCD, we found that sI/R induced PCD in a hypoxia/reoxygenation-dependent manner (Fig. 1B). Overexpression of both Bcl-2 and Bcl-x L , known protectors against cardiac I/R injury (27)(28)(29), significantly reduced sI/R-induced GFP-Bax redistribution, further demonstrating suitability of the model (Fig. 1, C and D).
Cellular Autophagosomal Content Is Increased during the Early Phase of sI/R Injury-We used HL-1 cells to explore the role of autophagy during sI/R injury. To determine whether autophagic activity is modulated in response to sI/R, we first characterized changes in cellular autophagosomal content using high resolution three-dimensional imaging of GFP-LC3. During the initiation of autophagy, cytosolic LC3 (LC3-I) is cleaved and lipidated to form LC3-II (19,30). LC3-II is then recruited to the autophagosomal membrane (31). Thus, punctate GFP-LC3-labeled structures represent autophagosomes, also referred to as autophagic vacuoles (AVs). Importantly, overexpression of (GFP-)LC3 does not affect autophagic activity, and transgenic mice expressing GFP-LC3 display no detectable abnormalities (12,32).
We transfected HL-1 cardiac myocytes with GFP-LC3 and compared the abundance of AVs in cells subjected to sI/R to normoxic control cells. Under normoxic conditions in KH solu-  panel). B, inhibitor-sensitive autophagy at 1.5 h of reperfusion was quantified by determining the percentage of cells with numerous GFP-LC3-labeled AVs in each condition without (steady-state AVs) and with (cumulative AVs) lysosomal inhibitors present during reperfusion (cumulative AVs: *, p Ͻ 0.001 (KH versus ischemia); **, p Ͻ 0.01 (KH versus ISCH Normox)). sI/R Normox, ischemia/mimetic solution in the presence of oxygen. C, GFP-LC3-expressing cells were fixed immediately following 2 h of ischemia without and with presence of lysosomal inhibitors, and autophagic flux was determined as described above (*, p Ͻ 0.01, steady-state AVs (KH versus sI/R); **, p Ͻ 0.05, cumulative AVs (KH versus sI/R)). Lysosomal inhibitors were present throughout ischemia. ISCH, ischemia; ISCH normoxic, ischemia/mimetic solution in the presence of oxygen. tion, GFP-LC3 was diffusely distributed throughout the cell, with very few detectable AVs ( Fig. 2A, upper panel). Cells subjected to sI/R, however, displayed increased numbers of AVs ( Fig. 2A, bottom panel). In addition, in control cells the few pre-existing AVs were randomly distributed, whereas AVs in cells subjected to sI/R were typically more clustered at the center of the cell. This distinctive distribution contrasts with the autophagic response to starvation in hepatocytes, where no such clustering was observed (33).
To quantify the increase in GFP-LC3-labeled AVs, the percentage of cells displaying numerous punctate GFP-LC3 structures was determined. Only a small fraction of cells displayed punctate GFP-LC3 fluorescence when incubated in fully supplemented medium or KH solution (Fig. 2B). In cells subjected to sI/R, however, the number of cells with numerous AVs was significantly increased (Fig. 2B). Quantitative analysis performed on Z-stacks of GFP-LC3 fluorescence revealed that sI/R significantly increased the number of AVs per cell and, likewise, the total autophagosomal volume (Fig. 2C).
Changes in Autophagic Activity during Ischemia and Reperfusion-Our results demonstrate that cellular AV content was increased early in the reperfusion period. We subsequently addressed the effect of sI/R on actual autophagic activity. Autophagy involves the delivery of the autophagosomes and their contents to lysosomes that contain the degradative enzymes needed to complete the catabolic processes of autophagy (1). Therefore, the increased presence of AVs may reflect enhanced formation of AVs, impaired fusion of AVs with lysosomes to generate autophagolysosomes, or a combination of the two. Moreover, LC3-II may be removed by lysosomal degradation at a rate that exceeds our imaging capabilities, i.e. the transit is so rapid and/or the AVs so small that only a few AVs can be detected at any given time. Accordingly, a low number of GFP-LC3-labeled AVs may be due to either low or high autophagic activity. To characterize autophagic activity, we therefore determined two relevant parameters of autophagosome-lysosome fusion, the index of LC3-II degradation and downstream lysosomal activity.
Flux of LC3-II Degradation during sI/R-Using an approach based on the inhibition of downstream lysosomal degradation of AVs and their cargo, we determined whether the increase in cellular AVs during sI/R was indicative of increased or impaired autophagy. Cells were subjected to various experimental conditions and treated with a mixture of lysosomal inhibitors to inhibit autophagolysosome formation (with bafilomycin A 1 ) and lysosomal protease activity (with E64D and pepstatin A). By analyzing the lysosomal inhibitormediated increase in GFP-LC3-II (AV) accumulation within a cell population, we were able to obtain a quantitative index of the flux of AV formation and degradation. Bar graphs with offset superimposed bars depict the percentage of cells exhibiting high AV levels in the absence and presence of lysosomal inhibitors, per condition. The difference between the two bars (see values in graphs) is a measure of the percentage of cells demonstrating high autophagic activity, or flux.
We found that in KH solution AV content was dramatically increased in the presence of inhibitors (Fig. 3, A and B). Thus, under control conditions in KH solution, which lacks the serum and amino acid component of full medium, autophagy was strongly active. Notably, this response is only revealed through the use of inhibitors; based on GFP-LC3 imaging alone (Fig. 2), low autophagic activity in KH solution would be incorrectly assumed.
sI/R augmented the number of cells with increased numbers of AVs (Fig. 3B). Under lysosomal inhibition the number of cells with high AV content was increased only slightly more, indicating that the previously described increase in cellular AV content in sI/R (Fig. 2 and 3B) is a reflection of an accumulation of AVs, presumably due to impairment in the autophagic pathway at a point(s) following AV formation and before AV degradation. As the level of AV accumulation was substantially smaller than the inhibitor-mediated response seen in KH solution for the same period of time, it can be concluded that autophagy is also impaired at the level of AV formation.
Most AVs were formed during the reperfusion period, as cells fixed immediately after the ischemic period were essentially devoid of AVs, either with or without lysosomal inhibitors, indicating a complete blockage of autophagy during the ischemic period (Fig. 3C). Interestingly, hypoxia was a necessary component of the insult, as cells incubated in ischemia/ mimetic solution alone, under normoxic conditions, exhibited only a minor reduction of autophagic flux, which recovered completely upon reperfusion (Fig. 3B).
Lysosomal Activity during sI/R-One possible explanation for the observed accumulation of AVs during sI/R was a nonfunctional lysosomal compartment. To investigate downstream lysosomal activity, HL-1 cells were incubated in LysoTracker Red, which labels the highly acidic lysosomal vacuoles and thus reports activity of the vacuolar H ϩ -ATPase (v-ATPase). Before and after sI/R, we observed similar patterns of LysoTracker Red fluorescence, indicating that, consistent with its importance in cell survival during I/R (34), activity of the v-ATPase is maintained during the reperfusion period (Fig. 4A).
Furthermore, we determined the activity and subcellular localization of cathepsin B, a predominant lysosomal protease, using a MagicRed substrate that fluoresces when cleaved by cathepsin B (35). We did not detect a decrease in cathepsin B activity following sI/R, as MagicRed fluorescence was still punctate (lysosomal) and displayed an intensity comparable with the normoxic control (Fig. 4B). Moreover, we found that cathepsin B activity was not detected in the cytosol, indicating that cathepsin B is not released from the lysosomes following sI/R. Together, these results indicate a functional lysosomal compartment during our experiments.
Unraveling the Role of Autophagy in sI/R Injury-The above results demonstrate that formation of AVs is suppressed during ischemia and that autophagic flux only partially recovers during reperfusion. As such, to determine whether autophagy represents a protective response or a component of injury, we altered the autophagic pathway using pharmacological and molecular perturbations at the levels of induction and/or formation of AVs.
Pharmacological Perturbation of Autophagy Induction Influences sI/R Injury-Phosphatidylinositol 3-kinases (PI3K) exert opposing actions on the autophagic pathway, with class I PI3K inhibiting and class III PI3K stimulating autophagy (36). The PI3K inhibitors 3-methyladenine (3-MA) and wortmannin (37,38) are used as inhibitors of autophagy at the sequestration step because of their inhibition of class III PI3K (39,40). Conversely, the inhibition of autophagy by class I PI3K is mediated through its downstream action on mammalian target of rapamycin (mTOR), which negatively controls autophagy. Inhibition of mTOR with the immunosuppressant rapamycin results in the induction of autophagy (41)(42)(43).
Cells were treated with either 10 mM 3-MA or 100 nM wortmannin, to inhibit autophagy, or with 1 M rapamycin, to enhance autophagy. To verify the effect on autophagic flux, GFP-LC3-expressing cells were incubated with the indicated inhibitors, and autophagic flux was evaluated in KH solution. Both 3-MA and wortmannin completely abolished autophagic flux in KH solution (Fig. 5A). During sI/R this was reflected in a complete lack of AV accumulation in cells treated with 3-MA and wortmannin (Fig. 5B). Treatment with rapamycin on the other hand increased autophagy in KH solution at the level of AV formation (Fig. 5A). It can be concluded that rapamycininduced AV formation exceeds the capacity for AV degrada-  OCTOBER 6, 2006 • VOLUME 281 • NUMBER 40 tion, because although a maximal percentage of cells showed active autophagy as revealed by lysosomal inhibition, the percentage of cells with high steady-state AVs increased significantly. During sI/R, rapamycin-treated cells still displayed numerous AVs (Fig. 5B). Determination of autophagic activity during sI/R revealed that rapamycin did not rescue autophagy during ischemia but increased autophagic flux at 1.5 h of reperfusion as fewer cells displayed steady-state accumulation of AVs (Fig. 5C).

Macroautophagy Protects against Ischemia/Reperfusion Injury
The effect of pharmacological interference with the activation of autophagy on sI/R-induced cellular injury was evaluated based on mitochondrial translocation of GFP-Bax (Fig. 6A). Under normoxic control conditions, over the duration of the experiment both 3-MA and wortmannin slightly increased basal levels of GFP-Bax translocation. Incubation with 3-MA and wortmannin prior to and during sI/R resulted in a significant aggravation of cellular injury.
Similarly, disruption of lysosome function increased sI/R injury. Under normoxic control conditions lysosomal inhibition did not affect cell viability for up to 7 h, as assessed by GFP-Bax clustering (Fig. 6B). However, lysosomal inhibition significantly increased GFP-Bax clustering following sI/R, indicating that lysosomal proteases do not exert pro-apoptotic functions during sI/R but rather participate in the maintenance of cell survival. In contrast, cells treated with rapamycin were significantly protected against sI/R injury, suggesting that autophagy may serve a pro-survival role during sI/R injury.
Beclin1 Protects from sI/R-activated Cell Death and Increases Autophagic Flux-The above results implicate autophagy as a protective response to I/R injury. However, the pharmacological reagents employed are known to also exert effects unrelated to autophagy (44 -46). We therefore sought to confirm these findings by targeting specific proteins in the autophagic pathway.
Beclin1, which is endogenously expressed in HL-1 cells (Fig.  7A), is a necessary participant in AV formation and is also believed to take part in the trafficking of lysosomal proteins (47,48). We examined the effect of overexpression of Beclin1 on sI/R-induced cellular injury. Similar to the protective effect of rapamycin, cells that had been transiently transfected with Beclin1 were significantly protected against sI/R-induced GFP-Bax translocation (Fig. 7B). The interaction between Beclin1 and the class III PI3K, hVps34, is a prerequisite for the induction of autophagy (48), through the generation of phosphatidylinositol 3-phosphate (PI3P), which functions as a second messenger to activate autophagy (36). However, the presence and distribution of PI3P did not change as a function of sI/R injury (based on imaging with the fluorescent PI3P reporter GFP-2xFYVE; data not shown).
To determine whether protection conferred by Beclin1 overexpression correlated with increased autophagic flux, LC3 processing was quantified in cells expressing Beclin1 following sI/R. Fluorescence analysis of GFP-LC3 co-transfected cells revealed a significant increase in autophagic flux after sI/R, as the percentage of cells with steady-state accumulation of AVs was decreased (Fig. 7C).
We subsequently reduced Beclin1 expression using the Block-iT miR RNAi vector that bicistronically encodes for GFP as an expression marker. We found that 96 h of expression were necessary in order to achieve a significant knockdown of Beclin1 (Fig. 8A). Cells transfected with Beclin1 RNAi had a reduced autophagic capacity when compared with control cells, as demonstrated using nutrient deprivation (Fig. 8B). Beclin1 silencing aggravated cellular injury after sI/R (Fig. 8, C  and D), revealing that autophagic activity following sI/R, although somewhat impaired, is a crucial component of the survival response of the cell.
Beclin1 Protection against sI/R Injury Requires a Functional Bcl-2/-x L -binding Domain-Beclin1 contains a Bcl-2/-x L -binding domain that has been shown to interact with the anti-apoptotic Bcl-2 family members Bcl-2 and Bcl-x L (10,49). To determine whether the protection conferred by Beclin1 following sI/R was regulated by the Bcl-2 family, we made use of a Beclin1 deletion mutant, Beclin1⌬Bcl2BD, which lacks the Bcl-2 binding domain (10) (Fig. 7A) but still can function in autophagy induction (49,50). Beclin1⌬Bcl2BD reduced autophagic activity during the reperfusion period as reflected in a decreased percentage of cells with steady-state accumulation of AVs as well as a complete absence of autophagic flux (Fig. 7C). Unlike overexpression of Beclin1, Beclin1⌬Bcl2BD expression did not confer protection against sI/R injury (Fig. 7B), confirming a cytoprotective role for autophagy.
Inhibition of Atg5 Aggravates Injury and Counteracts Protection Conferred by Beclin1-We next sought to determine whether the protective effect exerted by Beclin1 was specific to autophagy or rather was the result of a perturbation in the ratio of pro-and anti-apoptotic Bcl-2 family members, via binding of Beclin1 to Bcl-2 and/or Bcl-x L . To do so we targeted Atg5, a necessary component of the autophagic machinery downstream of Beclin1 activity (31). We utilized fusion proteins of the monomeric red fluorescent protein mCherry (21) and Atg5 or the dominant negative mutant of Atg5, Atg5 K130R (51).
Atg5 K130R is defective in its conjugation to Atg12, which is required for LC3 incorporation into the early autophagosomal structure, and thus inhibits autophagy at the level of AV formation (52, 53).
Autophagic flux at 1.5 h following sI/R was virtually the same for cells expressing mCherry-Atg5 and mCherry alone, suggesting that Atg5 is not a rate-limiting factor in the autophagic response to sI/R. Conversely, mCherry-Atg5 K130R significantly reduced autophagy during sI/R (Fig. 9A). To assess the consequence of reducing autophagy via expression of mCherry-Atg5 K130R , HL-1 cells were cotransfected with GFP-Bax and either mCherry-Atg5, mCherry-Atg5 K130R , or mCherry alone and subjected to sI/R (Fig. 9B). Consistent with the lack of an effect on autophagic flux, mCherry-Atg5 expression did not affect sI/R-induced GFP-Bax activation. Similar to the results obtained through silencing of Beclin1, mCherry-Atg5 K130R expression resulted in increased GFP-Bax clustering, indicating that residual autophagic activity represents an underlying protective response to sI/R. Finally, to determine whether Beclin1-mediated protection was because of its effect on autophagy, we blocked autophagy with mCherry-Atg5 K130R in Beclin1 overexpressing cells. Most strikingly, protection resulting from Beclin1 expression was lost when co-expressing mCherry-Atg5 K130R , indicating that Beclin1 protection is entirely linked to its role in increasing autophagy (Fig. 9C).

DISCUSSION
In this study we investigated the dynamics and role of the autophagic response to sI/R in the cardiac HL-1 cell line. We demonstrate that autophagic flux is null during the ischemic period and increases at reperfusion, but not to the same degree as under normoxic conditions. Autophagy plays a protective role in sI/R; pharmacologic inhibition of autophagy with wortmannin or 3-MA, RNAi knockdown of Beclin1, and overexpression of the dominant negative Atg5 K130R all sensitized cardiac cells to apoptosis after sI/R. Thus, residual levels of autophagy, even though not maximal, functioned to preserve cell viability following sI/R. Moreover, enhanced autophagy (through rapamycin treatment or Beclin1 overexpression)  OCTOBER 6, 2006 • VOLUME 281 • NUMBER 40 reduced apoptosis after sI/R. These results implicate autophagy as an important intracellular defense against sI/R injury and reveal new therapeutic targets.

Macroautophagy Protects against Ischemia/Reperfusion Injury
Autophagic Flux during sI/R-To quantify autophagy in our experimental system, we inhibited lysosomal degradation and analyzed the accumulation of GFP-LC3-positive AVs by fluorescence microscopy. This powerful methodology allowed us to monitor both subtle and robust changes in autophagic activity in single cells as well as in the population. We found that AV content did not correlate with autophagic activity; high steadystate AV content was indicative of the inability of the cell to degrade formed AVs, because of either excessive AV formation or a disruption of downstream degradation.
Importantly, unlike other assays that measure the degradation of long lived proteins, the technique we employed is specific to macro-autophagy. This precision is relevant as the parallel pathway of chaperone-mediated autophagy is also activated by oxidative stress (54) and starvation (55).
The accumulation of GFP-LC3-labeled AVs during sI/R is complex, resulting from simultaneous alterations in the induction and formation of AVs, their transit to and fusion with lyso-somes, and their ultimate degradation within the autophagolysosomes (see Fig. 10). During ischemia, the formation of AVs was blocked. Rapamycin treatment was unable to increase autophagy during the ischemic period, whereas autophagic flux was robust in cells incubated with the ischemic solution under normoxic conditions. As the conjugation steps performed by E1-like Atg7 during the formation of the AV require ATP (56), it is likely that energetic constraints of hypoxia preclude autophagy during the ischemic period.  . Inhibition of Atg5 increases sI/R injury and abolishes protection through Beclin1. Cells were transfected with either GFP-LC3 (A) or GFP-Bax (B), together with mCherry-Atg5, mCherry-Atg5(K130R), or mCherry alone (Control) and allowed to express for 48 h. Co-expression of GFP and mCherry was found to be Ͼ95%. A, autophagic flux at 1.5 h of reperfusion was determined via scoring of mCherry-positive cells with numerous GFP-LC3 vacuoles without (steady-state AVs) and with (cumulative AVs) lysosomal inhibitors. Represented is the percentage of cells with numerous GFP-LC3 vacuoles per total cells scored (*, p Ͻ 0.01). B, cells were subjected to simulated ischemia followed by 5 h of reperfusion, and cellular injury was assessed by scoring mCherry-positive cells with mitochondrially translocated GFP-Bax. Represented is the percentage of cells with punctate GFP-Bax distribution per total cells scored (*, p Ͻ 0.01). C, cells were co-transfected with GFP-Bax and Beclin1 plus mCherry, Beclin1 plus mCherry-Atg5(K130R) and subjected to sI/R 48 h after transfection. mCherry-positive cells with mitochondrial GFP-Bax were scored and quantified as described above (*, p Ͻ 0.001; **, p Ͻ 0.05).
Reperfusion led to an increase in the formation and, to a lesser extent, degradation of AVs. However, as the autophagic response was smaller than that seen in KH solution, it is likely that autophagy is impaired at the level of induction. This interpretation is supported by our finding that Atg5 overexpression did not increase autophagic flux, indicating that the autophagic machinery downstream of Beclin1-dependent induction was functional and not a rate-limiting factor.
The accumulation of AVs in a significant percentage of cells indicates a second disruption to the autophagic pathway downstream of AV formation. One potential mechanism would be lysosomal dysfunction. During the onset of some forms of PCD, lysosomal proteases can be released into the cytosol where they trigger apoptosis via cleavage of Bid and perhaps other targets (57,58). Furthermore, decreased activity of the predominant lysosomal protease cathepsin B has been reported in neuronal apoptosis (59). However, our experiments did not reveal a disruption of lysosomal function at the level of the v-ATPase or cathepsin B activity indicating that autophagic flux was impaired upstream of lysosomes. Thus, the observed accumulation of AVs during sI/R likely reflects impaired delivery to (60) or fusion with (61) the lysosomes (Fig. 10).
Interestingly, both rapamycin and Beclin1 overexpression increased autophagic flux by decreasing the percentage of cells with steady-state accumulation of AVs, yet without changing the total percentage of cells demonstrating active autophagy. In addition to their role in the initiation of AV formation, Beclin1 and/or hVps34 have been shown to play a role in the sorting of lysosomal proteins (48, 62, 63). Rab7, a member of the family of Rab GTPases that regulate transport and tethering/docking of vesicles (64), mediates the fusion of AVs with lysosomes (61,65). Rab7 also interacts with the Beclin1-interacting partner hVps34 (66), and it has been suggested that the localization of Rab7 to AVs is negatively regulated by mTOR (61). We propose that the protection conferred by rapamycin and Beclin1 overexpression is linked to rescued autophagolysosome formation. Further studies will be needed to support this hypothesis.
Protection, Bcl-2, and Autophagy-Cellular survival following I/R depends in large part on the interactions between anti-apoptotic (e.g. Bcl-2 and Bcl-x L ) and pro-apoptotic (e.g. Bid and Bax) Bcl-2 family members. Bcl-2 and Bcl-x L are known to protect against I/R injury (27,28) and were likewise protective in our model. Both Bcl-2 and Bcl-x L interact with Beclin1 (10). Importantly, the ability of Atg5 K130R to block the Beclin1-mediated cytoprotection demonstrates that the protective effect of Beclin1 is because of its enhancement of autophagy rather than a perturbation of Bcl-2/-x L homeostasis.
The significance of the interaction between Beclin1 and Bcl-2/Bcl-x L is unclear. Bcl-2 down-regulation has been shown to increase autophagy in HL-60 cells (7), whereas overexpression of Bcl-2 or Bcl-x L increased etoposide-activated autophagic cell death in mouse embryonic fibroblasts (9). Bcl-2 was also shown to reduce the presence of steady-state AVs during starvation in MCF7 cells (49).
Expression of Beclin1 Bcl-2 binding domain mutants has been reported to increase AV formation and autophagic cell death (49). However, our results using lysosomal inhibitors to quantify autophagic flux clearly indicate that expression of Beclin1⌬Bcl2BD decreased autophagy during sI/R and furthermore did not confer protection against sI/R. Beclin1⌬Bcl2BD may interfere with endogenous Beclin1 interaction with the PI3K, hVps34, as PI3P-induced GFP-2xFYVE clustering was decreased under control conditions and at 1.5 h of reperfusion (data not shown).
Moreover, expression of either wild-type Bcl-2 or Bcl-x L had no detectable effect on the autophagic response (data not shown) yet significantly protected against sI/R injury. This finding is in conflict with the previous report that wild-type Bcl-2 and, more robustly, ER-targeted Bcl-2 suppressed the formation of AVs as well as degradation of long lived proteins (49). However, the relationship between Bcl-2 and Beclin1 may be more complex, as ER-targeted Bcl-2 depletes ER calcium stores (67), whereas autophagy requires high ER calcium content (68). (1) induction of autophagy requires activity of Beclin1 and its interacting partner, class III PI3K (hVps34), resulting in the generation of PI3P; and it is negatively regulated by class I PI3K through mTOR (2). Formation of the phagophore requires conjugation of Atg12 to lysine 130 of Atg5 as a prerequisite for recruiting LC3-II (3). Sequestration of cytoplasmic material may be nonspecific or selective; mechanisms that may govern selectivity are incompletely understood (4). To accomplish degradation of the autophagosomes and its cargo, the autophagosomes are then transported to and fuse with the acidic lysosome, generating the autophagolysosome. Within the autophagolysosome lysosomal proteases degrade the inner autophagosomal membrane and cargo. During ischemia, autophagy was inhibited at the level of autophagosome formation. Upon reperfusion, autophagy partially recovered, with submaximal induction and impaired degradation. Enhancing autophagic flux was protective against sI/R injury. AA, amino acids.
Further studies will be needed to determine the relationship between Bcl-2 control of autophagy through its interaction with Beclin1 and/or mediation of sarcoendoplasmic reticulum calcium homeostasis.
The suppressive effect of Beclin1⌬Bcl2BD on autophagy indicates that an interaction between Beclin1 and Bcl-2/-x L is required for full autophagic activity, arguing for the possibility of coinciding pro-survival activities of both Bcl-2/-x L and autophagy. In fact, increased expression of both Bcl-2 and Beclin1 is correlated with protection during myocardial stunning (16,69).
Nature of Protection Exerted by Autophagy-Our model of sI/R conferred a rather mild insult as the cell death program was executed over a time course of several hours and therefore allowed us to study a heterogeneous cell population that, like the injured myocardium in vivo, included the following: (i) cells that had suffered irreversible damage and underwent cell death, and (ii) cells that had undergone sublethal stress and were able to eventually recover.
The autophagosomal engulfment of mitochondria (i.e. mitophagy) was a prominent feature in the autophagic response to sI/R (data not shown). It is conceivable that cells that have suffered a mild insult could be rescued through the selective elimination of damaged and pro-apoptotic mitochondria (those undergoing mitoptosis), whereas a more profound insult resulting in damage to numerous mitochondria would surpass the capacity for rescue, as removal of too many mitochondria would leave behind a cell with insufficient capacity to produce ATP and maintain calcium homeostasis. Recently we determined that the pro-apoptotic Bcl-2 family member Bnip3 is a key mediator of I/R injury (51). A key finding was that Bnip3-activated autophagy was protective and correlated with increased mitophagy. We suspect one function of autophagy during I/R is to sequester damaged mitochondria, in an effort to limit the spread of proapoptotic factors such as AIF, SMAC, and cytochrome c, as well as to remove mitochondria that consume ATP via reverse mode of the F 1 -ATPase.
In the neuron, a vital role of autophagy is to remove toxic protein aggregates, which are believed to be a causative component of many neurodegenerative diseases (70) and contribute to cell death following cerebral ischemia (71). In the heart, cytosolic proteins are known to aggregate (72) and may be causative in heart failure (73). Activity of the ubiquitin-proteasome system, which is responsible for protein degradation, is decreased during I/R (74), possibly because of inactivation by reactive oxygen species (75). Future studies will determine whether autophagy may compensate for decreased proteasome activity in the heart.
Conclusions-Our results demonstrate that autophagy is part of the physiologic response to I/R, and that up-regulation of autophagy (e.g. by rapamycin) can be accomplished in a short time frame, potentially compatible with therapeutic intervention after ischemia. To the best of our knowledge, the work presented here demonstrates for the first time that autophagy constitutes an underlying protective response against I/R injury in heart cells. Future experiments are needed to determine whether autophagy represents a means to reduce cardiac I/R injury in vivo.