Cardioprotection of Ischemia/Reperfusion Injury by Cholesterol-Dependent MG53-Mediated Membrane Repair

Xianhua Wang,* Wenjun Xie,* Yi Zhang,* Peihui Lin, Liang Han, Peidong Han, Yanru Wang, Zheng Chen, Guangju Ji, Ming Zheng, Noah Weisleder, Rui-Ping Xiao, Hiroshi Takeshima, Jianjie Ma, Heping Cheng

Rationale: Unrepaired cardiomyocyte membrane injury causes irreplaceable cell loss, leading to myocardial fibrosis and eventually heart failure. However, the cellular and molecular mechanisms of cardiac membrane repair are largely unknown. MG53, a newly identified striated muscle-specific protein, is involved in skeletal muscle membrane repair. But the role of MG53 in the heart has not been determined. Objective: We sought to investigate whether MG53 mediates membrane repair in cardiomyocytes and, if so, the cellular and molecular mechanism underlying MG53-mediated membrane repair in cardiomyocytes. Moreover, we determined possible cardioprotective effect of MG53-mediated membrane repair. Methods and Results: We demonstrated that MG53 is crucial to the emergency membrane repair response in cardiomyocytes and protects the heart from stress-induced loss of cardiomyocytes. Disruption of the sarcolemmal membrane by mechanical, electric, chemical, or metabolic insults caused rapid and robust translocation of MG53 toward the injury sites. Ablation of MG53 prevented sarcolemmal resealing after infrared laser–induced membrane damage in intact heart, and exacerbated mitochondrial dysfunction and loss of cardiomyocytes during ischemia/reperfusion injury. Unexpectedly, the MG53-mediated cardiac membrane repair was mediated by a cholesterol-dependent mechanism: depletion of membrane cholesterol abolished, and its recovery restored injury-induced membrane translocation of MG53. The redox status of MG53 did not affect initiation of MG53 translocation, whereas MG53 oxidation conferred stability to the membrane repair patch. Conclusions: Thus, cholesterol-dependent MG53-mediated membrane repair is a vital, heretofore unappreciated cardioprotective mechanism against a multitude of insults and may bear important therapeutic implications. (Circ Res. 2010;107:76-83.) Key Words: membrane repair MG53 cholesterol ischemia/reperfusion injury heart

n eukaryotic cells, the plasma membrane partitions a various pathophysiological stresses.3 Progressive necrotic and IϷ10 000-fold Ca2ϩ gradient and prevents loss of vital intra- apoptotic cell death causes onset of myocardial fibrosis and cellular constituents, thus representing the last line of defense for undermines cardiac contractile and electrophysiological perfor- cell integrity, homeostasis, and function. Physical, chemical or mance, ultimately leading to heart failure.4,5 metabolic disruption of the plasma membrane leads to a repair- Ironically, little was known about cardiac membrane repair or-die emergency of the cell. Although the natural tendency to until recently when several molecular components in striated reseal the lipid biomembrane acts constitutively, recent studies muscles were identified.6–10 In particular, 2 muscle-specific indicate that plasma membrane disruption requires active emer- proteins, dysferlin7 and MG53,9 have been implicated in the gency response mechanisms to mend the broken membrane.1 In repair of sarcolemmal membrane in skeletal muscles. Similar to the heart, plasma membrane repair is of particular importance the skeletal muscle, dysferlin-deficient mice display defective because cardiomyocytes are terminally differentiated cells, dis- cardiomyocyte membrane repair ability that is linked to in- playing only very limited self-renewal capability.2 Cardiomyo- creased susceptibility to cardiomyopathy,6 whereas the molecu- cytes undergo transient membrane injuries that occur as acci- lar function of dysferlin in membrane repair has not been fully dents under physiological conditions and can be exacerbated by elucidated. For example, it is unknown whether the formation of

Original received December 30, 2009; revision received April 23, 2010; accepted April 30, 2010. In March 2010, the average time from submission to first decision for all original research papers submitted to Circulation Research was 13.3 days. From the Institute of Molecular Medicine (X.W., W.X., Y.Z., L.H., P.H., Y.W., M.Z., R.-P.X., H.C., J.M.), State Key Laboratory of Biomembrane and Membrane Biotechnology, Peking University, Beijing, China; Department of Physiology and Biophysics (P.L., N.W., J.M.), Robert Wood Johnson Medical School, Piscataway, NJ; Institute of Biophysics (Z.C., G.J.), Chinese Academy of Sciences, Beijing, China; and Department of Biological Chemistry (H.T.), Kyoto University Graduate School of Pharmaceutical Sciences, Japan. *These authors contributed equally to this work. Correspondence to Xianhua Wang, PhD, Institute of Molecular Medicine, Peking University, Beijing 100871, China (E-mail [email protected]); or to Jianjie Ma, PhD, Department of Physiology and Biophysics, Robert Wood Johnson Medical School, 675 Hoes Ln, Piscataway, NJ 08854 (E-mail [email protected]). © 2010 American Heart Association, Inc. Circulation Research is available at http://circres.ahajournals.org DOI: 10.1161/CIRCRESAHA.109.215822 76 Wang et al Cardioprotection by MG53-Mediated Membrane Repair 77

Non-standard Abbreviations and Acronyms

⌬⌿ m mitochondrial membrane potential DTT dithiothreitol EBD Evans blue dye GFP green fluorescent protein I/R ischemia/reperfusion KO knockout M-␤-CD methyl-␤-cyclodextrin TMRM tetramethyl rhodamine methyl ester WT wild type

and protocols were approved by the Institutional Animal Care and Use Committee of Peking University accredited by AAALAC International. Ϫ Ϫ The mg53 / mice were generated as described9 and wild-type (WT) littermate mice were used as controls. Single ventricular myocytes were enzymatically isolated from the hearts as described previously.12 Ade- novirus containing mg53 or mg53(C242A) mutant gene was used to Figure 1. Membrane repair in intact hearts from mg53-null infect cardiomyocytes. Plasma membrane injury was elicited mechani- (KO) and wild type (WT) mice. A, Representative examples for FM1-43 ﬂuorescence accumulation in myocardium after cally, electrically, or chemically in cardiomyocytes. The whole-heart 2-photon laser irradiation (650 mW at 910 nm for 2000 scans in membrane repair assay using two-photon excitation microscopy was as 57 sec) in WT (top) and KO (bottom) hearts. Arrows mark described by Han et al,6 with minor modifications, and the I/R protocol laser-wound bands of 2ϫ250 ␮m2. Scale bars: 50 ␮m. B, Aver- used was as described previously.13 age time courses of laser-wound induced accumulation of An expanded Methods section is available in the Online Data FM1-43 ﬂuorescence. AU indicates arbitrary unit. Data are Supplement at http://circres.ahajournals.org. expressed as meansϮSEM; nϭ7 laser-wound bands from 4 hearts of KO mice and nϭ10 laser-wound bands from 4 hearts Results of WT mice. MG53 Mediates Membrane Repair in Intact Heart a membrane repair patch is associated with translocation of MG53 protein expression in mouse heart was detectable during dysferlin toward the injury site in cardiac muscle. Our previous embryonic development and was progressively increased during studies have shown that MG53 is an essential component of the first month after birth (Online Figure I, A). At the age of 12 membrane repair machinery in skeletal muscle, as MG53 abla- to 14 weeks, abundant expression of MG53 was found in tion results in defective membrane repair and progressive skel- muscular walls of all 4 chambers (right and left atria and etal myopathy.9 In addition, we have found that MG53 can ventricles) (Online Figure I, B). To test whether MG53 partici- interact with dysferlin and is required for dysferlin movement to pates in membrane repair in cardiac muscle, we used a mem- the acute membrane injury sites.11 However, the role of MG53 in brane repair assay modified from Han et al6 (Figure 1) and a the heart has not been determined and cardiac membrane repair mouse model with genetic ablation of mg53.9 Langendorff- remains largely elusive. perfused beating hearts from mg53Ϫ/Ϫ and WT littermate mice Here, we investigate the cellular and molecular mechanism of were subjected to infrared laser irradiation applied to a band MG53-mediated membrane repair in the heart and demonstrate of 500 ␮m2 (910 nm, 650 mW, 57 seconds). The process of its cardioprotective role against various insults. We show that membrane resealing was visualized by local accumulation of genetic ablation of mg53 prevents membrane resealing and FM1-43, a membrane-impermeable dye that becomes fluores- increases susceptibility to ischemia/reperfusion (I/R)-induced cent when it partitions into a membrane. In WT heart, time-lapse myocardial damage. Rapid and robust MG53 translocation two-photon imaging revealed that local FM1-43 fluorescence toward the injury sites occurs in cardiomyocytes subjected to leveled off after an initial rise following laser irradiation, either local or global plasma membrane disruption, whereas indicative of a halt in entry of FM1-43 after membrane resealing. physiological process of excitation-contraction coupling or Ϫ Ϫ In contrast, mg53 / mouse hearts exhibited a continuous, membrane deformation is unable to trigger such MG53 translo- high-rate elevation of FM1-43 fluorescence over much broader cation. Furthermore, we show that unfurled membrane choles- bands at the injury site (Figure 1), indicating an impairment of terol is indispensable in initiating MG53-mediated membrane repair, uncovering a novel signaling role of membrane choles- membrane resealing. These results establish an important role of terol. As an important cardioprotective response to stress, the MG53 in membrane repair in intact heart. cholesterol-dependent MG53-mediated membrane repair may provide a valuable therapeutic target for the treatment of heart Deficiency of MG53 Exacerbates I/R-Induced Loss disease. of Cardiomyocytes We next sought to determine the possible cardioprotective function of MG53-mediated membrane repair in the heart. In Methods Ϫ/Ϫ Animals were treated in compliance with the Guide for the Care and this regard, mg53 mice at young age (12 to 14 weeks old) did Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996), not exhibit detectable changes in cardiac morphometrics and 78 Circulation Research July 9, 2010

Figure 2. MG53 deﬁciency exacerbates I/R-induced loss of mitochondrial membrane potential (⌬⌿m) and cardiomyocyte death. A, Simulta- neous confocal imaging of ⌬⌿m (TMRM, shown in green) and cardiomyocyte death (EBD uptake, shown in red)in Langendorff-perfused heart after I/R injury. Note that all ⌬⌿m- intact cells are EBD-negative, and all EBD-positive cells show loss of ⌬⌿m but not vice versa. Scale bars: 50 ␮m. B, Statis- tics. From left to right, percentage of ⌬⌿m-lost cells before and after I/R, EBD-positive cells after I/R, and EBD-positive cells among ⌬⌿m-lost cells. Data are expressed as meansϮSEM; nϭ6 hearts with 220ϳ250 image frames for each group. #PϽ0.01 before vs after I/R; **PϽ0.01 KO vs WT after I/R; *PϽ0.05 KO vs WT after I/R. C, MG53 immunostaining in heart slices. Note the translocation of endogenous MG53 to the sarcolemma in WT heart after I/R injury and the absence of MG53 immunoﬂuorescence in KO heart. Scale bars: 10 ␮m. contractile functions when compared to WT littermate controls that MG53 retards the transition from reversible to irreversible (Online Figure II). Histological analysis failed to reveal any cardiomyocyte damage. significant myocardial pathology, either (Online Figure III). The Furthermore, we used conventional assessments to evaluate absence of overt cardiac phenotypes in young mg53Ϫ/Ϫ mice the cardioprotective effect of MG53 in response to I/R injury. As raised under normal laboratory conditions is consistent with the shown in Online Figure IV, MG53 ablation increased infarct size Ϫ/Ϫ finding in dysferlin-deficiency mice6 and supports the hypothe- (42.32Ϯ3.06% in mg53 versus 31.73Ϯ2.11% in WT, as- sis that membrane repair serves as an emergency response sessed in Langendorff-perfused heart subjected to global I/R against detrimental stimuli to the heart. injury, PϽ0.05, nϭ6 hearts for each group), TUNEL-positive Ϫ/Ϫ To assess potential MG53-mediated cardioprotection under nuclei (15.64Ϯ1.60% in mg53 versus 9.14Ϯ1.17% in WT, stress conditions, we subjected the Langendorff-perfused mouse PϽ0.01, nϭ6 hearts for each group) and also lactate dehydro- hearts to an I/R protocol (30 minutes/30 minutes) and assessed genase release by 2-fold. Importantly, plasma membrane trans- cardiac damage in the presence and absence of MG53-mediated location of endogenous MG53 did occur when the intact heart membrane repair. Because loss of mitochondrial membrane was subjected to I/R insult (Figure 2C), suggesting that MG53- ⌬⌿ 13,14 ⌬⌿ mediated repair mechanism is activated by this stress. potential ( m) is an early event of cell death, we used m as a sensitive readout of cardiomyocyte damage and performed MG53 Translocation During Acute simultaneous confocal measurement of tetramethyl rhodamine Membrane Injury methyl ester (TMRM) fluorescence (index for ⌬⌿ ) and Evans m As shown in Figure 2C, MG53 translocated and formed exten- blue dye (EBD) uptake (index for loss of membrane integrity) sive patches on the sarcolemma in clusters of cardiomyocytes in (Figure 2A). Before the I/R treatment, there was a trend of ⌬⌿ WT heart after the I/R injury, supporting that MG53 plays a increased number of cardiomyocytes showing a loss of m in crucial role in mending broken membrane. To investigate the Ϫ/Ϫ Ϯ mg53 compared to WT mouse hearts (10.15 0.95% in underlying mechanism, we used adenoviral expression of a Ϫ/Ϫ Ϯ Ͼ ϭ mg53 versus 5.54 2.00% in WT, P 0.05, n 6 hearts with MG53 fusion protein with an amino-terminal GFP (GFP-MG53) ϳ 220 250 image frames for each group), but the difference was in cultured rat cardiomyocytes (Online Figure V). Under the not statistically significant (Figure 2B). Confocal microscopy present experimental conditions, GFP-MG53 was expressed at a revealed that, after the I/R injury, local multicellular areas level similar to that of endogenous MG53 (Online Figure V, A), ⌬⌿ showing complete loss of m were interspersed with areas of and high-resolution confocal microscopy allowed for real-time ⌬⌿ intact m, giving rise to a sharply delimited mosaic pattern visualization of intracellular GFP-MG53 trafficking. Figure 3A (Figure 2A). Importantly, whereas the I/R injury caused a shows that focal electroporation of the sarcolemmal membrane ⌬⌿ significant increase in the number of m-lost cardiomyocytes caused rapid enrichment of GFP-MG53 at the wound site. Local Ϫ/Ϫ ⌬⌿ in both mg53 and WT mouse hearts, the number of m-lost GFP-MG53 fluorescence rose biexponentially with time con- Ϫ Ϫ cardiomyocytes in mg53 / hearts was greater than that in WT stants of 12.0 and 166 s, respectively (Figure 3B and Online controls by 140% (Figure 2B), indicating a cardioprotective role Video I). Mechanical injury by microelectrode penetration of the of MG53. Detailed analysis further revealed that a higher plasma membrane also triggered GFP-MG53 translocation to ⌬⌿ percentage of cardiomyocytes that had lost m was EBD- the injury site, and repetitive injuries at different locations Ϫ Ϫ positive in the absence of MG53 (69.62Ϯ5.94% in mg53 / resulted in multiple patches of enriched GFP-MG53 (Online versus 44.23Ϯ8.18% in WT, PϽ0.05, nϭ6 hearts with Figure VI), indicating that MG53 translocation ensues wherever 220ϳ250 image frames for each group) (Figure 2B), suggesting and whenever the plasma membrane is disrupted. Wang et al Cardioprotection by MG53-Mediated Membrane Repair 79

Figure 3. MG53 translocates to injured sarcolemma in cardiomyocytes. A, Patchy GFP-MG53 translocation induced by focal electroporation via a pair of microelectrodes (inset). B, Time course of GFP-MG53 accumulation at the electroporation site. Arrowhead indicates time of electroporation. Normalized ﬂuo-

rescence F/F0, where F0 refers to the level before electroporation, is expressed as meanϮSEM; nϭ9 cells. C, Subcellular distribution of GFP and GFP-MG53 before and after treatment with 50 ␮g/mL saponin. D, Sarcolemmal and cytosolic GFP-MG53 during saponin permeabilization. Arrowhead denotes time of saponin administration. Data are expressed as meansϮSEM; nϭ5 cells. E, MG53 immunostaining in Ad-GFP– or Ad-GFP-MG53–infected cells after saponin treatment. Note that GFP fluorescence (shown in green) is completely lost in Ad-GFP infected cells. In cells expressing GFP-MG53, distributions of GFP fluorescence from GFP-MG53 and of TRITC immunofluorescence (shown in red) from native and GFP-fused MG53 are virtually identical. Scale bars: 10 ␮m.

Global membrane permeabilization by a chemical detergent, MG53-mediated membrane repair protects heart from I/R in- saponin (50 ␮g/mL), induced robust GFP-MG53 translocation duced injury. over the entire sarcolemmal membrane (Figure 3C and Online Video II). The gain of GFP-MG53 at the membrane mirrored a MG53 Translocation Requires near-complete loss of GFP-MG53 in the cytoplasm, both dis- Membrane Disruption playing a time constant of Ϸ4.4 seconds (Figure 3D). This result What is the exact signal that is sensed by MG53 and promotes indicates that most of the GFP-MG53 molecules are available its plasma membrane translocation? To this end, the above 2ϩ and responsive when damage occurs at the sarcolemmal mem- experiments in Figure 3 were performed in a Ca -free external brane of the cardiomyocytes. In GFP-expressing cells, however, solution (with the benefit of preventing cell contracture after membrane damage), suggesting that Ca2ϩ entry is not a prereq- complete loss of GFP following chemical permeabilization uisite for MG53 translocation in cardiomyocytes, as is the case occurred without any appreciable staining of the sarcolemmal ϩ in skeletal muscle.9 To further test whether intracellular Ca2 membrane (Figure 3C and Online Video III). Similar results release participates in regulating MG53 translocation, we used were obtained with another membrane disruptive reagent, Triton caffeine (20 mmol/L) to deplete the intracellular sarcoplasmic X-100 (0.1% vol/vol) (Online Figure VII). Notably, immuno- ϩ reticulum Ca2 store before cell wounding. Figure 4A shows fluorescence of native MG53 was concentrated at the damaged that GFP-MG53 translocation induced by saponin treatment plasma membrane in a manner similar to that of GFP-MG53 remained intact even in the absence of intracellular Ca2ϩ release. (Figure 3E). Furthermore, in GFP-MG53-expressing cells, the Conversely, we also found no accumulation of GFP-MG53 at intrinsic fluorescence from GFP-MG53 overlapped with the the sarcolemmal membrane despite physiological Ca2ϩ tran- anti-MG53 immunofluorescence with a colocalization coef- sients and cell contraction elicited by electric pacing at 1 or 5 Hz Ϯ ϭ ϩ ficient of 0.91 0.02 (n 9 cells). Similarly, immunofluores- (Figure 4B). Taken together, we conclude that neither Ca2 cence of endogenous MG53 was also found to concentrate at entry nor intracellular Ca2ϩ transients are necessary or sufficient membrane injury sites wounded by electroporation (Online for GFP-MG53 translocation. This finding reinforces the idea Figure VIII, A) and microelectrode penetration (Online Fig- that MG53 translocation is an early event before Ca2ϩ- ure VIII, B). These results corroborate that GFP-MG53 and dependent vesicle trafficking and membrane fusion steps in the endogenous MG53 behave similarly during the process of membrane repair cascade.15–18 membrane translocation. Further studies showed that membrane stretching or altered Furthermore, consistent with Figure 2C, hypoxia-reoxygena- membrane curvature does not result in MG53 translocation to tion treatment on cardiomyocytes induced membrane transloca- the cell membrane. Here we used a patch pipette to create the tion of GFP-MG53 (Online Figure IX), substantiating that “⍀” shaped membrane deformation by negative pressure, as is 80 Circulation Research July 9, 2010

Figure 4. Effects of Ca2؉, membrane stretch and deformation on MG53 translocation. A, MG53 translocation in the absence of external and internal store Ca2ϩ. Cardiomyocytes were sus- pended in extracellular solution containing 5 mmol/L EGTA and depletion of the sarcoplasmic reticulum Ca2ϩ store was achieved by inclusion of 20 mmol/L caffeine for 3 minutes. Saponin (50 ␮g/mL) was used for membrane disruption. B, Intracellular Ca2ϩ transients and contractile activity did not cause MG53 translocation. Top, Traces of Ca2ϩ transients and cell contraction. Bottom, Subcellular distribution of GFP-MG53 before and after pacing at 1 or 5 Hz. Similar results were obtained in 5 cells. C, Effects of membrane stretch and membrane deformation on MG53 translocation. Deforma- tion of membrane in a patch pipette under gigaseal conditions did not trigger GFP-MG53 translocation, but subsequent membrane rupture did. Scale bars: 10 ␮m.

the practice during the formation of a gigaseal for electrophys- exerts differential effects at different phases of MG53 translo- iological recording. No GFP-MG53 localization was evident in cation in cardiomyocytes. Specifically, application of the mem- spite of the sharp membrane deformation and local stretch stress brane permeable oxidant H2O2 with different concentrations (Figure 4C). Only after the membrane was ruptured by suction (0.1, 0.5, 1 mmol/L) failed to recruit GFP-MG53 to the plasma did GFP-MG53 translocate to membrane at the patch pipette tip membrane (Online Figure X), even at a high dose (5 mmol/L) (Figure 4C). Thus, these data show that plasma membrane that caused cell damage and contracture (Figure 5A). Inclusion disruption is a prerequisite for membrane translocation of of the reducing reagent dithiothreitol (DTT) did not affect the MG53. initial GFP-MG53 translocation, either; nevertheless, the nascent GFP-MG53 matrix was quickly dissociated from the membrane Oxidation Confers Stability to the MG53 Repair in the presence of DTT (Figure 5B), suggesting that oxidation is Patch in Cardiomyocytes required to stabilize the MG53 repair patch. To further investi- Our previous study in skeletal muscle has shown that entry of gate this phenomenon, we expressed GFP-MG53(C242A) mu- extracellular oxidants is necessary for MG53 translocation, and tant (the cysteine of 242 amino acid was mutated to alanine), the cysteine residue of 242 amino acid of MG53 serves a which lacks the redox-sensing ability,9 to a MG53-null back- Ϫ Ϫ redox-sensor for MG53 oligomerization or crosslinking.9 In this ground of cardiomyocytes from mg53 / mouse heart (to avoid study, we extended this finding by showing that redox regulation possible interference of endogenous MG53). In response to

Figure 5. Redox regulation of MG53

translocation. A, H2O2 (5 mmol/L) failed to trigger membrane translocation of GFP-MG53, while causing severe cell contracture. B, DTT (5 mmol/L) did not affect the initial time course of GFP- MG53 translocation but subsequently dissolved the nascent GFP-MG53 matrix. Data are expressed as meansϮSEM; nϭ9 cells. C, GFP- MG53(C242A) mutant protein translocated to membrane in response to membrane permeabilization but was unable to maintain a stable repair patch. Cardi- omyocytes were isolated from mg53Ϫ/Ϫ mouse hearts. Similar results were obtained in 6 cells. Images a, b, and c in the top panel were obtained at time points indicated in the bottom panel. Arrowheads denote time of saponin administration. Scale bars: 10 ␮m. Wang et al Cardioprotection by MG53-Mediated Membrane Repair 81

Figure 6. Cholesterol is required for membrane translocation of MG53. A, Colocalization of GFP- MG53 (top) and cholesterol (filipin fluorescence) (bottom) at the mechanically injured plasma membrane site. Enlarged views of the boxed region are shown to the right. Differing from localization of GFP-MG53 only to the injury site, cholesterol shows enrichment to the entire surface membrane and to perinuclear regions (marked by arrows). B, Effects of membrane cholesterol manipulation on MG53 translocation. For cholesterol depletion, cardiomyocytes were treated with 5 mmol/L M-␤-CD (1 hour at 37°C). For cholesterol recovery, cholesterol-depleted cardiomyocytes were further treated with 0.25 mmol/L cholesterol/2.5 mmol/L M-␤-CD mixture (1 hour at 37°C). For lipid raft/caveolar disruption without cholesterol deprivation, cardiomyocytes were treated with 2 ␮g/mL filipin (1 hour at 37°C). Membrane injuries were inflicted by focal electroporation (left) (arrowheads mark sites of injury) or Triton X-100 treatment (0.1% vol/vol) (right). C, Time courses of local GFP-MG53 accumulation on electroporatic injury under various experimental conditions. Data are expressed as meansϮSEM; nϭ8 to 10 cells for each group. Scale bars: 10 ␮m.

saponin-induced membrane disruption, GFP-MG53(C242A) ro- gated whether such substructures are required for cholesterol bustly translocated to the membrane but was unable to form a regulation of MG53 translocation. By filipin disruption of the stable repair patch (Figure 5C), as was the case with DTT lipid rafts and caveolae without depleting membrane cholesterol, treatment. Hence our data suggest that initial movement of we found that GFP-MG53 translocation to membrane damage MG53 on membrane disruption is redox-independent, whereas sites remained unaffected (Figure 6B and 6C), suggesting that oxidative regulation of MG53 at cysteine of 242 amino acid MG53 recognizes unfurled membrane cholesterol regardless of plays a critical role in stabilizing the MG53 repair patch. the integrity of lipid rafts and caveolae. Thus, membrane cholesterol is obligatory for MG53 translocation, and exposure Membrane Cholesterol Is Necessary to Signal of membrane cholesterol likely acts as an initial step of the MG53 Translocation MG53–mediated membrane repair. Given that MG53 discriminates between intact and injured membrane, we hypothesized that MG53 interacts with the Discussion hydrophobic core of the lipid bilayer membrane, which is exposed only in the event of membrane disruption. In this Cardioprotection by MG53-Mediated regard, cholesterol came into the picture because it is embedded Membrane Repair in the hydrophobic core and accounts for Ϸ40% of the lipid As a muscular pump undergoing constant contractile activity, molecules of the plasma membrane in animal cells.19 Besides, the heart is an organ under intense mechanical stress with the majority (Ϸ90%) of it is enriched in plasma versus organelle high metabolic demands. As a result, cardiomyocytes are membranes.19,20 We therefore tested the hypothesis that mem- under constant threat of mechanical injury, oxidative stress, brane cholesterol, when unfurled, signals MG53 to translocate to and metabolic insult. Because cardiomyocytes cannot be the membrane. replaced in large numbers, lost cardiomyocytes are replaced Staining cholesterol with filipin uncovered a dense colocal- by fibroblasts, resulting in myocardial fibrosis in response to ization of GFP-MG53 and cholesterol to the membrane site a number of etiologies, including myocardial infarction injured by electrode penetration, but an absence of GFP-MG53 caused by coronary heart disease. Thus, the heart must in intact cholesterol-rich plasma membrane (Figure 6A). Strik- develop intrinsic defensive mechanism to survive such inju- ingly, when membrane cholesterol was depleted with methyl-␤- ries by preventing cardiomyocytes death. In this regard, the cyclodextrin (M-␤-CD), the GFP-MG53 translocation after fo- present study has identified MG53 as a potent cardiac cal electroporation was completely abolished (Figure 6B and protector: MG53 ablation exacerbated the I/R-induced myo- 6C). The effects of Triton X-100 as a membrane-disrupting cardial damage, as manifested by markedly worsened mito- reagent were similar to that of electroporation (Figure 6B). chondrial dysfunction and loss of cardiomyocytes. Restoration of cholesterol by incubation the cells with M-␤-CD Several lines of evidence have demonstrated that cardio- complexed with cholesterol largely revived the GFP-MG53 protection by MG53 is mediated, at least in part, by its translocation to damaged membranes (Figure 6B and 6C). membrane repair function. First, we have shown that local or Because cholesterol is particularly enriched in the membrane global membrane disruption evokes MG53 to translocate microdomains called lipid rafts and caveolae,21,22 we investi- exclusively to the injured membrane in cardiomyocytes. 82 Circulation Research July 9, 2010

Second, a membrane repair assay using FM1-43 fluorescence lipid rafts has unraveled an important role of cholesterol in revealed that MG53 deficiency impaired membrane resealing assembling these membrane microdomains critical for the reg- after two-photon irradiation damage in intact heart. That ulation of signal transduction and membrane trafficking.21,22 MG53 is inert to physiological Ca2ϩ transients and cardiac Here we have identified yet another fundamental role of mem- contraction is also consistent with the idea that MG53 is brane cholesterol in signaling membrane repair. Depletion of normally a bystander but plays a key role in the emergency membrane cholesterol abolished, and its recovery restored response to rescue the cell from disastrous membrane MG53 translocation on membrane electroporation or chemical disruption. permeabilization. Remarkably, cholesterol regulation of MG53 One other muscle specific protein thought to contribute to translocation remained intact even after filipin disruption of this membrane repair response in striated muscle is dysfer- membrane lipid rafts or caveolae, suggesting that cholesterol- lin.6–8 Dysferlin has been shown to be essential for membrane mediated hydrophobic interaction between MG53 and the dis- repair in skeletal muscle7; however it requires the function of rupted membrane is not confined to specialized membrane MG53 to translocate to sites of membrane injury.11 Thus, microdomains. Our previous study in skeletal muscle showed MG53 can nucleate the assembly of repair machinery at sites that MG53 mediates membrane repair through direct interaction of membrane damage, whereas dysferlin likely operates at a with membrane phosphatidylserine,9 here we further found that downstream stage of the membrane resealing process. Our cholesterol, as another membrane lipid, is obligatory for MG53- findings that MG53 is important in protecting the heart from mediated membrane repair. Because MG53 can interact with I/R injury is in contrast to previous report that dysferlin is phosphatidylserine directly but not cholesterol in the lipid unable to protect the heart from cardiomyopathy induced by profiling assay,9 cholesterol-dependent secondary membrane coronary ligation in mice.6 The contrasting phenotypes of structure or other yet-to-be identified partner may contribute to MG53 and dysferlin deficiencies suggest that MG53 and this role of cholesterol. dysferlin play different cardioprotective roles, or that the In determining whether the cholesterol-dependent MG53 dysferlin-mediated Ca2ϩ-dependent membrane repair mech- translocation is sensitive to redox regulation as promoted by anism is partly redundant of MG53-mediated repair mecha- studies in skeletal muscle,9 we demonstrated 2 steps for forma- nism in the heart. Future investigation is required for better tion of the MG53 repair patch in cardiomyocytes: trigger of understanding the relationship and relative contributions of MG53 translocation and stabilization of the translocated MG53. these distinctive coexisting repair mechanisms. The initial MG53 translocation to membrane injury sites is It is also noteworthy that MG53 ablation did not induce redox-independent, but MG53 oxidation at cysteine of 242 any abnormality for heart structure and function in young amino acid appears to be necessary for stabilizing the nascent adult mice raised under normal laboratory conditions. This MG53 repair patch. In skeletal muscle, the initial MG53 trans- finding is in agreement with cardiac phenotypes of dysferlin location phase appears to be very brief or the MG53 repair patch deficiency: myocardial fibrosis does not begin in young adult is very unstable, such that MG53 translocation was seen as only mice (less than 32 weeks) and is mild in senescent animals a small blip in the time course plot under reducing conditions ϩ (Ն1 year old).6 These previous and present findings under- (see figure 4f and 4g in the article by Cai et al9). Also, the Ca2 score the idea that membrane repair mechanisms may not be independence of MG53 translocation in cardiomyocytes is essential in young, healthy cardiomyocytes. In fact, mem- consistent to that in skeletal muscle which showed that extra- ϩ brane deformation that does not rupture the membrane, cellular Ca2 entry is not essential for MG53 translocation to intracellular Ca2ϩ release or contraction induced by electric membrane injury sites (see figure 6e and 6f in the article by Cai pacing, and even membrane deformation and stretching were et al9). not sufficient to trigger the MG53 translocation process. Intuitively, this cholesterol dependence of MG53 transloca- These results are in general agreement with the present tion serving to initiate cardiac membrane repair would be robust thought in this field that membrane repair process consists of and fool-proof, because membrane cholesterol is exposed if and molecular components that act as an “emergency repair” only if the membrane is disrupted. Given the abundance of response system.1 As such, MG53 safeguards the heart from membrane cholesterol, it confers one additional advantage of both cumulative loss of cardiomyocytes under normal wear- efficiency to membrane repair. Both robustness and effective- and-tear conditions and from massive acute cardiomyocyte ness are important features in a situation where speed and death in the event of myocardial I/R. fidelity of the response is a matter of survival of the cell. In summary, we have shown that MG53 membrane translo- Role of Membrane Cholesterol in MG53-Mediated cation constitutes the initial step of repairing disrupted cardio- Membrane Repair myocyte plasma membrane in the heart. Mechanistically, mem- In searching for the exact signal that triggers MG53 transloca- brane cholesterol exposed in the disrupted membrane signals tion, we have demonstrated, for the first time, that membrane MG53 translocation in a redox-independent manner, whereas cholesterol is an indispensable molecular player for the initiation MG53 oxidation confers stability to the repair patch. As a result, of MG53 translocation in cardiac membrane repair. As highly cholesterol-dependent MG53-mediated membrane repair pro- hydrophobic, compactly ring-rigid structure, cholesterol is an tects heart from loss of mitochondrial function and subsequent important component of the plasma membrane. It is embedded irreplaceable loss of cardiomyocytes under stress conditions in the hydrophobic core of the lipid bilayer, and thus acts to such as myocardial I/R injury. These findings uncover an rigidify the membrane, reduce the passive permeability, and important heretofore unappreciated cardioprotective mechanism increase the mechanical durability.23 Recent investigation of that involves cardiac membrane repair, and might help to Wang et al Cardioprotection by MG53-Mediated Membrane Repair 83 develop therapeutic strategies to ameliorate cardiomyopathy and 8. Han R, Campbell KP. Dysferlin and muscle membrane repair. Curr Opin to retard, or potentially reverse, the progression of heart failure. Cell Biol. 2007;19:409–416. 9. Cai C, Masumiya H, Weisleder N, Matsuda N, Nishi M, Hwang M, Ko JK, Lin P, Thornton A, Zhao X, Pan Z, Komazaki S, Brotto M, Acknowledgments Takeshima H, Ma J. MG53 nucleates assembly of cell membrane repair We thank I. C. Bruce, A. W. Cheng, and C. L. Wei for critical machinery. Nat Cell Biol. 2009;11:56–64. discussion and C. M. Cao, Y. Zhang, H. Q. Fang, H. L. Zhang, and 10. McNeil PL. Membrane repair redux: redox of MG53. Nat Cell Biol. N. Hou for technical assistance. 2009;11:7–9. 11. Cai C, Weisleder N, Ko JK, Komazaki S, Sunada Y, Nishi M, Takeshima H, Ma J. Membrane repair defects in muscular dystrophy are linked to Sources of Funding altered interaction between MG53, caveolin-3, and dysferlin. J Biol This work was supported by the Major State Basic Research Develop- Chem. 2009;284:15894–15902. ment Program of China (2007CB512100) and the National Natural 12. Cheng H, Lederer WJ, Cannell MB. Calcium sparks: elementary events Science Foundation of China (30630021, 30628009, and 30900264). underlying excitation-contraction coupling in heart muscle. Science. 1993;262:740–744. 13. Matsumoto-Ida M, Akao M, Takeda T, Kato M, Kita T. Real-time Disclosures 2-photon imaging of mitochondrial function in perfused rat hearts sub- None. jected to ischemia/reperfusion. Circulation. 2006;114:1497–1503. 14. Crompton M. The mitochondrial permeability transition pore and its role References in cell death. Biochem J. 1999;341(Pt 2):233–249. 15. Steinhardt RA, Bi G, Alderton JM. Cell membrane resealing by a 1. McNeil PL, Kirchhausen T. An emergency response team for membrane vesicular mechanism similar to neurotransmitter release. Science. 1994; repair. Nat Rev Mol Cell Biol. 2005;6:499–505. 263:390–393. 2. Pasumarthi KB, Field LJ. Cardiomyocyte cell cycle regulation. Circ Res. 16. Bi GQ, Alderton JM, Steinhardt RA. Calcium-regulated exocytosis is 2002;90:1044–1054. required for cell membrane resealing. J Cell Biol. 1995;131:1747–1758. 3. Clarke MS, Caldwell RW, Chiao H, Miyake K, McNeil PL. Contraction- 17. Miyake K, McNeil PL. Vesicle accumulation and exocytosis at sites of induced cell wounding and release of fibroblast growth factor in heart. plasma membrane disruption. J Cell Biol. 1995;131:1737–1745. Circ Res. 1995;76:927–934. 18. McNeil PL, Steinhardt RA. Loss, restoration, and maintenance of plasma 4. Nishida K, Kyoi S, Yamaguchi O, Sadoshima J, Otsu K. The role of membrane integrity. J Cell Biol. 1997;137:1–4. autophagy in the heart. Cell Death Differ. 2009;16:31–38. 19. Maxfield FR, Wustner D. Intracellular cholesterol transport. J Clin Invest. 5. Kostin S, Pool L, Elsasser A, Hein S, Drexler HC, Arnon E, Hayakawa Y, 2002;110:891–898. Zimmermann R, Bauer E, Klovekorn WP, Schaper J. Myocytes die by 20. Liscum L, Munn NJ. Intracellular cholesterol transport. Biochim Biophys multiple mechanisms in failing human hearts. Circ Res. 2003;92:715–724. Acta. 1999;1438:19–37. 6. Han R, Bansal D, Miyake K, Muniz VP, Weiss RM, McNeil PL, Campbell 21. Simons K, Ikonen E. Functional rafts in cell membranes. Nature. 1997; KP. Dysferlin-mediated membrane repair protects the heart from stress- 387:569–572. induced left ventricular injury. J Clin Invest. 2007;117:1805–1813. 22. Simons K, Toomre D. Lipid rafts and signal transduction. Nat Rev Mol 7. Bansal D, Miyake K, Vogel SS, Groh S, Chen CC, Williamson R, McNeil Cell Biol. 2000;1:31–39. PL, Campbell KP. Defective membrane repair in dysferlin-deficient 23. Simons K, Ikonen E. How cells handle cholesterol. Science. muscular dystrophy. Nature. 2003;423:168–172. 2000;290:1721–1726.

Novelty and Significance What Is Known? myocardial fibrosis and eventually heart failure. However, the ● Plasma membrane repair is an active emergency response to mend cellular and molecular mechanisms of cardiac membrane repair broken membrane. are largely unclear. In this study, we demonstrate that MG53, a ● MG53 is an essential component of membrane repair machinery in newly identified striated muscle-specific protein, constitutes an skeletal muscle. essential component of the membrane repair machinery in ● Deficiency of dysferlin-mediated membrane repair leads to heart. MG53-mediated membrane repair serves as a powerful cardiomyopathy. cardioprotection mechanism in preventing ischemia-reperfusion injury. Furthermore, we find that the redox status of MG53 does What New Information Does This Article Contribute? not affect initiation of MG53 translocation, whereas MG53 ● MG53 plays crucial roles in membrane repair response of heart. oxidation confers stability to the membrane repair patch. ● MG53-mediated membrane repair protects the heart from ischemia/ Unexpectedly, we reveal a crucial role of membrane cholesterol reperfusion injury. in signaling MG53 to mend the broken membrane. We uncover ● Cholesterol is necessary for signaling MG53-mediated membrane a new role of plasma cholesterol and, for the first time, link repair response. cholesterol, MG53, and membrane repair with endogenous mechanisms defending against the life-threatening cardiac Cardiomyocytes undergo membrane injury, which happens accident. As an important cardioprotective response to stress, under physiological conditions and this injury could be exacer- the cholesterol-dependent MG53-mediated membrane repair bated by pathophysiological stress. Unrepaired cardiomyocyte may provide a valuable therapeutic target for the treatment of membrane injury causes irreplaceable cell loss, leading to heart disease.