|Home | About | Journals | Submit | Contact Us | Français|
Mitofusin-2 (Mfn-2) is a dynamin-like protein that is involved in the rearrangement of the outer mitochondrial membrane. Research using various experimental systems has shown that Mfn-2 is a mediator of mitochondrial fusion, an evolutionarily conserved process responsible for the surveillance of mitochondrial homeostasis. Here, we find that cardiac myocyte mitochondria lacking Mfn-2 are pleiomorphic and have the propensity to become enlarged. Consistent with an underlying mild mitochondrial dysfunction, Mfn-2-deficient mice display modest cardiac hypertrophy accompanied by slight functional deterioration. The absence of Mfn-2 is associated with a marked delay in mitochondrial permeability transition downstream of Ca2+ stimulation or due to local generation of reactive oxygen species (ROS). Consequently, Mfn-2-deficient adult cardiomyocytes are protected from a number of cell death-inducing stimuli and Mfn-2 knockout hearts display better recovery following reperfusion injury. We conclude that in cardiac myocytes, Mfn-2 controls mitochondrial morphogenesis and serves to predispose cells to mitochondrial permeability transition and to trigger cell death.
Mitochondria from a variety of organisms and tissues have been described as dynamic organelles that change their shape and size and remodel their internal membranes or move to distinct cellular locations (11, 36, 54). These morphological transitions are greatly influenced by fusion and fission of the mitochondrial membranes and have been referred to as mitochondrial dynamics (29, 46). In adult cardiac myocytes, mitochondria do not display significant motility and they are in close contact with each other (10, 81). Their morphological variability is confined and depends upon the myocyte compartment that they occupy (e.g., interfibrillar versus subsarcolemmal mitochondria [IFM and SSM, respectively]) (2, 53, 76). Furthermore, it has been recognized that cardiac mitochondria are arranged in a highly organized pattern and under localized stress conditions can coordinate their membrane potential and propagate depolarizing events throughout the cell, suggesting the existence of interorganellar communication mechanisms (4, 14, 15, 85). Therefore, questions remain as to what are the unique features of mitochondrial dynamics in fully differentiated cardiac myocytes and what is their impact on mitochondrial structure and energetics.
Mitochondrial fusion requires membrane potential, GTP hydrolysis, and the assembling activities of mitofusins 1 and 2 (Mfn-1 and Mfn-2, respectively) and optic atrophy protein 1 (Opa-1) (17, 18, 21, 43, 55, 74). Mfn-1 and Mfn-2 are integral to the outer mitochondrial membrane (OMM), whereas Opa-1 can be integral or associated with the inner mitochondrial membrane (IMM) (50, 62). Mitochondrial fission requires dynamin-related protein 1 (Drp-1), which is detected primarily in the cytosol but translocates to the OMM after interacting with fission protein 1 (Fis-1) (78, 84). All of these mitochondrion-shaping proteins are expressed in the mammalian heart (28, 32, 42, 73), but their roles in regulating organelle structure and function in this tissue remain to be elucidated.
Mfn-2 is a large GTPase that is essential for mitochondrial fusion during embryonic development and neuronal differentiation (16, 18, 19). In the human population, mutations in the MFN-2 locus are linked to Charcot-Marie-Tooth type 2a (CMT2a) neuropathy (86). Mfn-2 is robustly expressed in the heart (5), and Mfn-2 insufficiency and associated fragmentation of the mitochondrial network in cultured neonatal cardiac myocytes have been reported to promote early apoptotic events (66). In a different experimental setting, however, Mfn-2 is reported to induce death in neonatal cardiomyocytes and in H9C2 cells through the intrinsic mitochondrion-dependent pathway (75). This apparent controversy may be due to cell type-specific effects or may be reflective of the multiple roles ascribed to Mfn-2 (27). More recently, Mfn-2, in addition to its targeting on mitochondria, was shown to reside on endoplasmic reticulum (ER) membranes, and this dual localization is thought to facilitate transfer of Ca2+ from the ER into the adjacent mitochondria (26). This could potentially expose mitochondria to very high local Ca2+ concentrations, as suggested by the Ca2+-microdomain hypothesis (70, 71).
In parallel with their crucial role in fuel oxidation and energy conversion, cardiac mitochondria are also centrally involved in cell death cascades (37). The mitochondrial permeability transition pore (MPTP), classically activated by Ca2+ and reactive oxygen species (ROS), is an important determinant of myocyte loss, especially in the context of ischemia and reperfusion injury (7, 30, 38), but its molecular composition and regulation remain controversial (40). The initial working model suggested that the pore is made of the outer mitochondrial membrane voltage-dependent anion channel (VDAC), the inner mitochondrial membrane adenine nucleotide translocase (ANT), and the matrix protein cyclophilin D (Cyp-D) (23). However, genetic studies have challenged this model and showed that only Cyp-D is a critical member of the pore (8, 60), whereas ANT appears to perform a regulatory rather than a structural role (49). Finally, the outer mitochondrial membrane component of the pore remains elusive, as all isoforms of VDAC were shown to be dispensable for MPTP function (9).
In the present study, we find that the conditional deletion of Mfn-2 increases the proportion of enlarged mitochondria in cardiac myocytes but does not lead to a major impairment of cardiac function. In addition, Mfn-2-depleted mitochondria were found to be more tolerant to Ca2+-induced MPTP opening, and isolated Mfn-2-knockout myocytes were protected from local generation of ROS and subsequent MPTP activation. Finally, Mfn-2 knockout hearts were able to develop higher pressures during postischemic reperfusion and exhibited diminished cell death following in vivo regional ischemia and reperfusion injury. These data illustrate that Mfn-2 not only serves to maintain mitochondrial morphology in cardiac myocytes but also promotes MPTP opening in the heart under conditions of stress.
All procedures that involved animal handling were approved by the Institutional Animal Care and Use Committee at the Boston University School of Medicine or the University of Maryland School of Medicine. The mice were housed in a 12-hour light/dark cycle and temperature-controlled room with access to water and food ad libitum. Genotyping was performed to differentiate between the wild-type Mfn-2 allele (Mfn-2+) and the conditional Mfn-2flox (Mfn-2F) allele as previously described (19). Similarly, to differentiate between mice with the alpha myosin heavy chain (α-MHC) transgene (cre+) present and absent (cre0), genotyping was performed as previously described (1). Mice with the genotype Mfn-2flox/flox; cre+ are termed F/F;cre, whereas their littermates with the genotype Mfn-2F/F; cre0 are termed F/F;−. In addition, mice with the genotype Mfn-2+/+; cre+ were derived from different crossings within the same lineage and are termed +/+;cre. The mice examined in this study are on a mixed genetic background (129S/C57BL6/Black Swiss). The macrophage-specific Mfn-2-deficient strain was generated by crossing the Mfn-2F/F line to a strain expressing cre under the control of the endogenous promoter of lysozyme M (LyzM-cre strain; Jackson Laboratory). These mice are referred as F/F;creLysM.
Mice were anesthetized with 1% isoflurane and fixed in the supine position on a heating pad equipped with electrocardiogram (ECG) surface lead II, and preheated sonographic gel (Aquasonic) was applied to their shaved chests. The Vevo 770 system with probe 707 was used to obtain long-axis parasternal views of the left ventricle (LV). The endocardium and epicardium were traced in long-axis frames to calculate the LV mass according to a built-in formula that utilizes the average LV wall thickness, the epicardial and endocardial dimensions in the diastole, and the specific gravity constant of myocardium (1.05 g/ml). Using the ECG trace, we identified the LV frames that corresponded to the end diastole (LVd) and end systole (LVs). These measurements from the long axis were combined with measurements from four different levels from the short axis in order to calculate the end-systolic and end-diastolic volumes (ESV and EDV, respectively) according to Simpson's equation. Subsequently, the ejection fraction (EF) was calculated as the stroke volume (SV = EDV − ESV) over the EDV. Peak flow velocities at the pulmonic and aortic valves (PPF and PAF, respectively) were also recorded from the long-axis position using the pulsed-wave (PW) Doppler transducer. The short-axis M-mode view at the level of the papillary muscles was used to evaluate cardiac contractility in terms of fractional shortening (FS) based on the dimensions of the LV in diastole (LVIDd) and systole (LVIDs). At the four-chamber apical view, we recorded mitral flow velocities, and relevant waveforms were used to measure early (E) and late (A) filling velocities and ejection time (ET). From the same anatomic position, we also recorded the motion of the mitral annulus using tissue Doppler (TD), and E′ and A′ values were measured.
Hemodynamic analysis of the heart in situ was performed on isoflurane-anesthetized and constantly ventilated mice as previously described (65). Briefly, the 1.4F SPR-839 catheter (Millar), connected to a PowerLab/8SP (ADInstruments), was inserted in the LV via the right carotid artery. Pressure-volume (PV) loops at baseline and after vena cava occlusion were recorded and subsequently analyzed offline using PVAN 3.2 software. The volume signal was corrected for parallel conductance using the saline bolus injection technique and was converted to μl according to a standard curve generated with a calibration cuvette. In a separate set of experiments, the jugular vein was cannulated for isoproterenol infusions and the catheter was inserted in the LV of anesthetized mechanically ventilated mice via an apical stub. Following stabilization, baseline recordings were taken and isoproterenol (Calbiochem) dissolved in normal saline (0.2 μg/ml) was continuously infused for 5 min with a syringe pump (Harvard Apparatus) at a rate of 5 ng/kg/min. At the end of the infusion, PV loops were recorded to calculate the various hemodynamic parameters offline.
Mice were heparinized and anesthetized with sodium pentobarbital (150 mg/kg of body weight intraperitoneally [i.p.]), and their hearts were quickly removed. Hearts were perfused in the Langendorff mode with phosphate-free Krebs-Henseleit buffer containing 118 mM NaCl, 25 mM NaHCO3, 5.3 mM KCl, 2.0 mM CaCl2, 1.2 mM MgSO4, 0.5 mM EDTA, 5 mM glucose, and 0.5 mM pyruvate at 37.5°C as previously described (57). The perfusate was equilibrated with 95% O2 and 5% CO2 (pH 7.4). All hearts were stabilized for 25 min at a constant perfusion pressure of 80 mm Hg. A water-filled balloon was inserted into the LV to record ventricular pressure and heart rate. After stabilization, balloon volume was adjusted to achieve the end diastolic pressure (8 to 10 mm Hg). Baseline tracings were acquired for 10 min, and hearts were subjected to no-flow global ischemia for 10 min and reperfused for 20 min (57).
The mice were anesthetized with intraperitoneal injection of 50-mg/kg sodium pentobarbital (Nembutal), mechanically ventilated at 130 breaths/min, and connected to a PowerLab/8SP for constant monitoring of heart rate and ECG patterns as previously described (65). Core body temperature was monitored with a rectal probe connected to a temperature controller (Harvard Apparatus) and maintained between 37.0 and 37.2°C. The chest was opened, and the left anterior descending (LAD) coronary artery was tied in line with a snare occluder using a monofilament suture (8-0; S&T) to produce regional ischemia. Reperfusion injury was induced by removing the occluder 30 min later. During manipulations of the LAD coronary artery, the ECG pattern was used to confirm ischemia and reperfusion. The animals were allowed to recover, and 2 h after reperfusion, the LAD coronary artery was reoccluded and 0.15 ml Evans blue (5%, wt/vol) was injected into the jugular vein to delineate the ischemic from the nonischemic area. The hearts were then collected and cut into 5 slices, which were weighed and subsequently incubated in triphenyl-tetrazolium-chloride (TTC; 1%, wt/vol) at 37°C for 5 min to highlight the infarct areas (IA). Imaging and calculations of the different portions of the heart (i.e., area at risk [AAR] and IA) were carried out as previously described (65). In addition, some slices were fixed in formalin, and 5-μm-thick sections were collected from paraffin-embedded heart samples and were stained to detect double-stranded DNA breaks using the terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assay according to manufacturer's specifications (Roche).
For electron microscopy, the hearts were perfused through the apex with Sorensen's phosphate buffer (0.2 M, pH 7.2) and excised and the LV free wall was dissected and cut into 4 longitudinal rod-shaped pieces (~1 mm in diameter). The pieces were placed in Karnovsky fixative (2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer; Electron Microscopy Sciences) for 2 h. Following postfixation in 2% OsO4, the pieces were dehydrated through ethanol and embedded in Epon plastic. Toluidine blue staining of semithin sections was used to evaluate the orientations of the sections and to select the areas containing predominantly longitudinal myofibers. Preparation of ultrathin sections (~70 nm) perpendicular to the long axis of the tissue was performed with a diamond knife microtome (Sorvall). Sections were stained with lead citrate and uranyl acetate and visualized using a Philips CM12 transmission electron microscope. Ten to 18 fields containing longitudinally arrayed myofibrils, excluding nuclei, were photographed from each section at the ×6,300 or ×35,000 magnification. Mitochondrial and myofibrillar volume density was assessed directly from the micrographs using the grid method (59). Using ImageJ (Wayne Rasband, NIH), a mask was generated for each field photographed and cross-sectional areas and maximum/minimum Feret's diameters of individual mitochondria were measured. Kolmogorov-Smirnov analysis was used to evaluate differences in the diameter distributions between the two genotypic groups.
In some experiments, cardiac myocytes were isolated from adult mice and seeded on glass bottom dishes (MatTek) that had been coated with laminin (BD Biosciences; 240 μg/dish) in minimum essential medium (MEM; Invitrogen) supplemented with 20 mM butanedione monoxime (BDM; Sigma) and 5% (vol/vol) calf serum (Invitrogen) as previously described (65). After an initial plating step for at least 30 min, cells were exposed to fresh medium containing 2 nM tetramethylrhodamine ethyl ester (TMRE; Molecular Probes) for 15 min at 37°C. The medium was replaced with dye-free medium, and cells were visualized with a Zeiss LSM 710 confocal microscope. To analyze mitochondrial morphology, we used the 100× oil immersion lens (Plan-Apochromat; numerical aperture [NA], 1.4), TMRE was excited with the 543-nm laser set at 0.2% power, and myocytes were scanned in their z axis in 15 to 20 slices spaced by 0.38 μm. Volume renderings were performed with the three-dimensional (3D) viewer plug-in of ImageJ (M. Abramoff). In other experiments, 100 nM tetramethylrhodamine methyl ester (TMRM) was loaded onto myocytes for 15 min at 37°C, and after the monocytes were switched to dye-free medium, they were exposed to 200 μM H2O2 at time zero (t0) and imaged at 1-min intervals with the 63× objective (1.40 oil differential inference contrast [DIC]) and with the 543-nm excitation laser set at 1.8% power, for a total duration of 25 min. The images were analyzed offline for changes in fluorescence intensity over time from multiple regions of interest (ROI) (8 by 8 μm) per myocyte using ImageJ. To assess mitochondrial membrane potential (ΔΨm), freshly prepared myocytes were incubated for 30 min at 37°C in medium containing 5 μM JC-1, which was subsequently replaced by dye-free medium. Stained myocytes were individually imaged using the 100× objective; the monomeric form of JC-1 was excited using the 488-nm laser, and the aggregate form was excited using the 543-nm laser. Acquired images were analyzed offline using ImageJ, and the fluorescence intensity for each JC-1 form was determined for multiple regions of interest per myocyte. In other experiments, fresh myocytes were exposed to 500 nM MitoTracker Red (CMXRos; Molecular Probes) or 100 nM TMRM for 15 min at 37°C, after which they were switched to dye-free medium. The ΔΨm-dependent accumulation of fluorophore molecules in mitochondria was determined by exciting with the 561-nm (MitoTracker Red) or 543-nm (TMRM) laser, and the fluorescence intensity was determined from multiple ROI per myocyte and analyzed offline with ImageJ software.
In other experiments, adult mouse cardiomyocytes were isolated using previously described methods (77). In brief, mice were heparinized and then anesthetized using a lethal dose of i.p. pentobarbital (100 mg/kg). The heart was quickly excised and placed in an ice-cold, Ca2+-free physiologic buffer solution. Following aortic cannulation, the heart was perfused with Ca2+-free physiological saline and then with a solution containing 1 mg/ml collagenase and 0.05 μM Ca2+. The ventricles were cut, minced, and treated with a second enzymatic solution containing 0.67 mg/ml collagenase, 0.13 mg/ml protease, 0.05 μM Ca2+, and 16.67 mg/ml bovine serum albumin (BSA). Ventricular tissue was lightly triturated with a coated glass pipette. Cells were washed through a large mesh filter (300 μm) and lightly centrifuged at 200 rpm for 2 min. The cells were gradually reintroduced to physiologic Ca2+ concentration and stored in physiological saline at room temperature.
Freshly isolated cardiomyocytes were bathed at room temperature with a physiological saline solution that contained 100 nM TMRM for 20 min. Following the loading period, cells were transferred to a TMRM-free physiological solution and stored until use. Imaging was carried out in living TMRM-loaded cardiomyocytes placed on laminin-coated glass coverslips in a custom-designed perfusion chamber. A laser scanning confocal microscope (Zeiss LSM 510, 100× oil immersion lens; NA, 1.3) was used to image the cells with 543-nm excitation, and cells were viewed through a 560LP emission filter. Repetitive imaging was carried out at 0.7 Hz with constant illumination intensity and a constant size (30 by 35 μm) of the region of interest.
To measure cell contractility, isolated myocytes were loaded with the Ca2+-sensitive indicator fluo-4 AM (10 μM) for 15 min before they were washed and stored in physiological saline. Cells were imaged using a 488-nm excitation laser and imaged with a model 505LP optical filter. Longitudinal line scan images along the full length of each cardiomyocyte were acquired at a rate of 520 scans per second. While being imaged, cardiomyocytes were stimulated by field shocks (MyoPacer [IonOptix], 20 to 40 V for 1 to 2 ms). Cells were stimulated at a 1-Hz frequency for 10 s to ensure proper sarcoplasmic reticulum (SR) loading before images were acquired for 10 s, during which time cells were stimulated at 1 Hz for 5 s, followed by 2.5 s of rest.
Neonatal rat cardiac myocytes (NRCMs) were isolated as previously described (68) and treated with 90 nM Mfn-2-specific or unrelated small interfering RNA (siRNA) (Dharmacon) using the Lipofectamine reagent (Invitrogen) under serum-free conditions for 48 h. For mitochondrial morphology, cells were loaded with 50 nM MitoTracker Red and incubated at 37°C for 30 min, after which cells were washed and fixed. Imaging was performed using the 63× lens of a Nikon inverted microscope. Additionally, NRCMs were loaded with 1 nM TMRM for 30 min and then switched to imaging medium (Dulbecco's modified Eagle's medium [DMEM] with 25 mM HEPES and without phenol red and pyruvate). Oxidative stress was induced with 200 μM H2O2, and imaging was initiated immediately. Images were captured every 90 s for 60 min using a Nikon deconvolution wide-field epifluorescence microscope system controlled with Nikon software. Similarly, peritoneal macrophages isolated from F/F;creLyzM and F/F;− mice were loaded with 1 nM TMRM for 30 min, and the medium was replaced with dye-free medium. Oxidative stress was induced with 200 μM H2O2, and imaging was initiated immediately. Cells were visualized with a Zeiss LSM 226 710 confocal microscope. The TMRM dye was excited using the 543-nm laser set at 1.8% power. Images were captured at 1-min intervals.
Subsarcolemmal mitochondria (SSM) and interfibrillar mitochondria (IFM) were isolated from adult mouse hearts as previously described (48). Briefly, hearts were minced in ice-cold isolation buffer (IB; 100 mM KCl, 50 mM MOPS [morpholinepropanesulfonic acid], 5 mM MgSO4·7H2O, 1 mM EGTA, 1 mM ATP, pH 7.4), and the suspension was disrupted by Polytron treatment and homogenized with a Potter-Elvehjem tissue grinder. The homogenate was centrifuged at 600 × g, and then the SSM were collected from the supernatant at 3,000 × g and washed/resuspended in KME buffer (100 mM KCl, 50 mM MOPS, 0.5 mM EGTA) at a final concentration of 25 mg mitochondrial protein/ml. To release IFM, the pellet was disrupted by Polytron treatment in the presence of trypsin (5 mg/ml) and centrifuged at 600 × g. The IFM were collected from the supernatant at 3,000 × g and were washed/resuspended in KME buffer at a final concentration of 25 mg mitochondrial protein/ml.
In a separate set of experiments, total cardiac mitochondrial populations were isolated using methods previously described (39, 61). Briefly, hearts were harvested after CO2 asphyxiation and homogenized in 4 ml buffer A (67 mM sucrose, 50 mM Tris-HCl, 2 mM EGTA, 50 mM KCl, and 0.2% fatty acid-free BSA, pH 7.4) using an ice-cold glass homogenizer (Kontes). The homogenate was centrifuged at 2,000 × g, and the mitochondrion-containing supernatant was centrifuged at 10,000 × g. This pellet was washed twice with 1 ml of buffer B (buffer A without BSA), resuspended in 1 ml of buffer B, loaded onto a 19% (vol/vol) Percoll solution, and centrifuged at 14,000 × g in 4°C. The mitochondrial fraction was resuspended in 0.2 ml of mitochondrial swelling buffer (67 mM sucrose, 50 mM Tris-HCl, and 50 mM KCl, pH 7.4). Mitochondrial protein concentration was determined using the Bradford method.
To determine mitochondrial diameter, a flowmetric assay was used as previously reported (24). Briefly, isolated SSM and IFM were stained with MitoTracker Deep Red 633 (Molecular Probes) and assessed using a BD LSR I flowmeter (BD Biosciences). The mean output from the forward scatter detector was used as an index of mitochondrial size. The adjustment to actual μm was performed using calibration microspheres (0.5 to 6.0 μm; Invitrogen), and mitochondrial volume was calculated based on the diameter values.
SSM and IFM were kept in respiration buffer containing 100 mM KCl, 50 mM MOPS, 5 mM KH2PO4, 1 mM EGTA, and 0.1% fatty acid-free BSA, pH 7.0. Oxygen consumption in mitochondrial subpopulations was assessed using a Clark-type electrode and the substrate combinations (i) 10 mM pyruvate plus 5 mM malate, (ii) 40 μM palmitoylcarnitine, and (iii) 20 mM succinate plus 3.75 μM rotenone as previously described (48, 64, 72). State III respiration was measured in the presence of 200 μM ADP, and state IV was measured after ADP consumption began. Activities of citrate synthase (CS), isocitrate dehydrogenase (IDH), and medium-chain acyl coenzyme A (acyl-CoA) dehydrogenase (MCAD) were measured from heart homogenates as previously described (64).
MPTP opening in SSM and IFM was performed as previously described (64). In short, mitochondria were resuspended in 2.0 ml assay medium containing 100 mM KCl, 50 mM MOPS, 5 mM KH2PO4, 5 μM EGTA, 1 mM MgCl2, 5 mM glutamate, and 5 mM malate and assayed for Ca2+ uptake in a fluorescence spectrophotometer at 37°C. CaCl2 (5 mM) was infused at a rate of 2 μl/min, and the concentration of free Ca2+ in the medium was calculated by monitoring the fluorescence of the Ca2+-bound and Ca2+-free Fura-6-F (0.1 M; Molecular Probes) (the excitation wavelengths for the Ca2+-bound and Ca2+-free fluorophores were 340 and 380 nm, respectively, and the emission wavelength was 550 nm). Calibrations of Fura-6-F fluorescence were performed at the end of the experiment using 0.1 Μ EGTA and 0.1 M CaCl2 to establish a zero level and a Ca2+-saturated level of the fluorophore, respectively (64). MPTP opening was inferred from the sudden and large increase of fluorescence. Mitochondrial Ca2+ tolerance was defined as the cumulative Ca2+ load that was required to induce the abrupt increase in extramitochondrial Ca2+ from a semi-log plot (47).
In a separate set of experiments, Ca2+ tolerance was evaluated according to previous reports (22, 61). Briefly, mitochondria resuspended in swelling buffer (2 mg/ml protein) were warmed to room temperature in a 96-well plate and CaCl2 (200 μM) was added. The time-dependent decrease in absorbance at 520 nm is indicative of the tendency for MPTP opening and mitochondrial swelling. For the mitochondrial shrinkage assay, mitochondria were preswollen by incubating them at 2 μg protein/ml at room temperature for 10 min in the mitochondrial swelling buffer containing 500 μM CaCl2. Pores were closed by addition of 1 mM EGTA to chelate the added Ca2+. Five percent polyethylene glycol (PEG) was added to swollen mitochondria, and the absorbance at 520 nm was monitored for 20 min using a spectrophotometer.
Cardiac myocytes were isolated from adult mice using aortic cannulation and Liberase dissociation as described above. Myocytes were seeded on laminin-coated dishes in the presence of a normoxia medium that consists of MEM supplemented with 1.2 mM CaCl2, 12 mM NaHCO3, 2.5% (vol/vol) fetal bovine serum (FBS), 1% (vol/vol) penicillin-streptomycin, and 25 μM blebbistatin (44). Hypoxia was induced by replacing the normoxia medium with a hypoxia medium that consisted of 118 mM NaCl, 16 mM KCl, 24 mM NaHCO3, 1 mM NaHPO4, 2.5 mM CaCl2·2H2O, 1.2 mM MgCl2, 20 mM sodium lactate, 10 mM deoxyglucose, and 10 mM HEPES, pH 6.2 (25) and placing the trays in a Billups-Rothenberg modular incubator chamber saturated with 95% N2, 5% CO2, and 1% O2. One hour later, the normoxia medium was used to replace the hypoxia medium, and myocytes were incubated for two more hours in normoxic conditions. In parallel experiments, myocytes were exposed to 20 μM H2O2 or were left untreated for 2 h. To determine myocyte death either after hypoxia/reoxygenation (H/R) or H2O2 treatment, cells were exposed to 0.04% trypan blue-containing medium for 10 min, and photographs of myocytes were obtained in a systematic fashion using the ×10 magnification of a light microscope. Cell counting was performed with ImageJ software. Following counting, the cells were collected in lysis buffer and total myocyte protein was isolated for Western blotting.
Cardiac samples weighing 30 mg were flash frozen in liquid nitrogen, and protein was extracted in tissue lysis buffer (tissue protein extraction reagent [T-PER buffer; Pierce] containing EDTA-free protease inhibitor [Roche]). Protein concentration in the lysates was quantified using the bicinchoninic acid (BCA) assay (Thermo Scientific) according to the manufacturer's specifications. Twenty micrograms of protein from each sample was resolved on 10% SDS-PAGE gels (Lonza) and transferred to polyvinylidene difluoride (PVDF) membranes (Amersham). After semidry transfer of proteins at 400 mA for 60 min at 4°C, the membranes were blocked in 3% nonfat milk in phosphate-buffered saline (PBS) containing 0.5% Tween 20 (PBS-T) for 1 h. Primary antibodies for mitofusin-1 (molecular mass, 80/85 kDa; Abcam), mitofusin-2 (80 kDa; Sigma), VDAC-porin (30 kDa; Abcam), C-recombinase (40 kDa; Novus Biologicals), cyclophilin D (17 kDa; Thermo Scientific), ANT1/2 (35 kDa; Santa Cruz), α-tubulin (55 kDa; Calbiochem), cytochrome c oxidase subunit IV (COX-IV; 18 kDa; Abcam), Bcl-2 (25 kDa; BD Transduction Laboratories), lactate dehydrogenase (LDH; 35 kDa; Cell Signaling), complex V subunit α (50 kDa; Molecular Probes), Bax (20 kDa; Cell Signaling), caspase-9 (49/39 kDa; Cell Signaling), poly(ADP-ribose) polymerase (PARP-1) (fragments of 116, 89, and 24 kDa; Cell Signaling), Drp-1 (75 to 80 kDa; BD Transduction Laboratories), Opa-1 (80 to 90 kDa; Abcam), and GAPDH (glyceraldehyde-3-phosphate dehydrogenase; 37 kDa; Cell Signaling) were diluted to a 1:1,000 ratio in 3% blocking solution and incubated with the membrane overnight at 4°C. Detection of immunoreactive bands was performed with the appropriate secondary antibodies conjugated with horseradish peroxidase activity (HRP) using the ECL reagent (Amersham).
Hearts were perfused through the apex with normal saline, harvested, and cut through the long axis. Sagittal slices were fixed in 10% buffered formalin, passed through graduated concentrations of ethanol, and embedded in paraffin according to standard protocols. Four-micrometer-thick sections were processed with Harris hematoxylin and eosin (H&E) or Masson's trichrome reagents (Sigma). Sections were photographed under a light microscope equipped with a digital camera, and images were quantified for myocyte cross-sectional area or collagen content as previously described (65).
Total cardiac mRNA was extracted from frozen samples using the Qiagen fibrous tissue minikit according to the manufacturer's specifications. Eight hundred fifty nanograms of RNA was reverse transcribed into cDNA using the Thermoscript reverse transcription-PCR (RT-PCR) system (Invitrogen). The amounts of different cDNAs were quantified using the SYBR green reagent and the StepOne real-time system (Applied Biosystems). These quantities were expressed relatively to that of the GAPDH gene, which was used as the housekeeping gene. All primer sequences for genes shown in Table 3 are available upon request.
All values shown are means ± standard errors of the means (SEM) unless otherwise specified. When two groups were compared, the Student two-tailed t test was applied (unpaired). For comparisons between 3 or more groups, we used one-way analysis of variance (1-way ANOVA), and if statistically significant differences were detected, we utilized Bonferroni's post hoc test to further identify groups with different means. Differences were considered significant for P values of less than 0.05.
Because the heart is a mitochondrion-rich tissue, we used the α-MHC-Cre transgenic mouse line (1, 34) to disrupt the Mfn-2 locus selectively in cardiac myocytes of Mfn-2loxP mice (19). Cre-mediated excision of Mfn-2 exon 6 is associated with a predictable loss of Mfn-2 protein (Fig. (Fig.11 A), which in the case of α-MHC-Cre × Mfn-2F/F mice reaches ~90% efficiency (Fig. (Fig.1A,1A, right panel). We refer to these mice as F/F;cre and utilize their cre-negative F/F;− littermates or cre-only (+/+;cre) mice as age-matched controls. The protein levels of Mfn-1 did not appear to change significantly in these heart samples. Furthermore, the genetic recombination is found to be specific to the heart and is not detectable in any other tissue of F/F;cre mice, nor does it appear to occur in the absence of the cre transgene (results not shown).
Histological examination of F/F;cre hearts detected cardiac enlargement that was not accompanied by overt ventricular dilatation (Fig. (Fig.1B).1B). Microscopic analysis revealed the presence of myocyte hypertrophy in F/F;cre hearts without significant increases in the collagen content (Fig. (Fig.1C).1C). We analyzed the cardiac function of adult F/F;cre and F/F;− or +/+;cre mice using noninvasive echocardiography or cardiac catheterization (Tables (Tables11 and and2).2). There were no significant differences in chamber dimensions, systolic function, or hemodynamic performance between the two groups, except for the detection of increased left ventricle (LV) mass in the F/F;cre group (Table (Table1).1). To examine the heart function under conditions of β-adrenergic stress, we acutely infused isoproterenol (5 ng/kg/min) and monitored the hemodynamic response using LV catheterization. As shown in Fig. Fig.22 A and B and Table Table3,3, this approach revealed small but statistically significant differences between F/F;cre and F/F;− or +/+;cre hearts in terms of systolic function (i.e., reductions in the end-systolic pressure and the maximum rate of LV pressure rise [dP/dtmax, where P is pressure and t is time] in the F/F;cre group [Table [Table3]).3]). Examination of contractility in isolated cardiac myocytes identified a small reduction in fractional shortening in the F/F;cre myocytes, while the intracellular Ca2+ ([Ca2+]i) transients appear to be similar between F/F;cre and F/F;− myocytes (Fig. 2C to F). Furthermore, the mRNA levels of various genes associated with stress, metabolism, and mitochondrial biogenesis or function were normal or showed small changes in expression in F/F;cre hearts, with the exception of the atrial natriuretic peptide (ANP) mRNA, which was upregulated by 2.9-fold (Table (Table4).4). Taken together, these data suggest that the lack of Mfn-2 from the heart is associated with modest myocyte hypertrophy accompanied by mild deterioration of left ventricular function.
Morphological analysis using electron microscopy of the LV wall revealed mostly round or rectangular mitochondria with diameters ranging from 0.5 to 2 μm in F/F;− hearts (Fig. (Fig.33 A). In F/F;cre samples, however, some regions had mitochondria with normal morphology (Fig. (Fig.3B)3B) while other areas contained enlarged mitochondria with diameters sometimes up to 3 or 4 μm and, more rarely, up to 5 or 6 μm that tended to form into clusters (Fig. 3C and D). In some cases, the enlarged mitochondria in the F/F;cre tissue displayed further abnormalities in their internal structure, such as loss of cristae and formation of inner membrane vesicles (Fig. (Fig.3E).3E). However, as shown in Fig. Fig.3F,3F, the gross mitochondrial cross-sectional area remains unchanged in the F/F;cre hearts, suggesting the maintenance of overall mitochondrial mass. The copy numbers of the mitochondrial gene for NADH dehydrogenase subunit 1 were not found to differ significantly between the two groups (results not shown), further suggesting normal mitochondrial biogenesis in F/F;cre myocytes. The myofibrillar compartment appears largely intact in F/F;cre sections (Fig. (Fig.3F,3F, myofibrils). Nevertheless, the cross-sectional area per individual mitochondrion is found to be, on average, significantly increased in the knockout group (Fig. (Fig.3G),3G), in agreement with the presence of enlarged mitochondria. Furthermore, the distribution of the maximum and minimum mitochondrial diameters (Fig. 3H and I, respectively) was found to be significantly altered in the Mfn-2 knockout group as it shifted to the right, indicative of mitochondria with increased diameters (4 to 5 μm in the major axis and 2 to 3 μm in the minor axis). This analysis also revealed that the number of detectable mitochondria per equal area analyzed was reduced in the knockout group (2,042 versus 1,186). Mfn-2 has been recently implicated in the bridging of the outer mitochondrial membrane with the endoplasmic reticulum (26). Using electron microscopy on heart samples with or without Mfn-2, we examined the organization of the “Ca2+ release domains” that include the T-tubule, the junctional sarcoplasmic reticulum (jSR) and the outer mitochondrial membrane (13). As shown in Fig. Fig.3J,3J, the distance between the center of the T-tubule and the outer mitochondrial membrane does not appear to change significantly in the F/F;cre group, indicating that the gross distance between the jSR and the outer mitochondrial membrane is not likely to be affected by the absence of Mfn-2. Collectively, the electron microscopic analysis identified the propensity in Mfn-2-deficient mitochondria to become fewer and enlarged without changing their overall mass, indicating a defect in mitochondrial patterning and distribution rather than in biogenesis.
To further examine mitochondrial morphology, we performed confocal microscopy on intact myocytes isolated from adult hearts. Mitochondria visualized with the membrane potential-sensitive dye TMRE have a rectangular shape and display the typical striated appearance in myocytes expressing normal levels of Mfn-2 (Fig. (Fig.44 A). In contrast, the mitochondria from isolated F/F;cre myocytes are more heterogeneous in shape, often spherical, enlarged, and less precisely organized within the myocyte (Fig. (Fig.4B).4B). Three-dimensional representation of mitochondria in the two groups further illustrates the presence of enlarged mitochondria within a given area of the myocyte (Fig. 4C and D). We examined, in addition to these morphological defects, the levels of the mitochondrial membrane potential (ΔΨm) in the absence of Mfn-2 using three different dyes that sequester in polarized mitochondria (Fig. (Fig.4E).4E). As shown in the left panel of Fig. Fig.4E,4E, the ratio of the JC-1 aggregate fluorescence to the JC-1 monomer fluorescence in F/F;cre myocytes is decreased, indicating a lower ΔΨm in mitochondria lacking Mfn-2. Consistently, fluorescence intensity due to accumulation of MitoTracker red or TMRM into mitochondria was found to be lower in myocytes without Mfn-2, again suggesting a partial decrease in ΔΨm (Fig. (Fig.4E,4E, middle and right panels). Taken together, results of the confocal microscopy analysis of isolated myocytes indicate that in the absence of Mfn-2, mitochondria lose their strict structural organization within the myocyte and display an increase in size which coincides with a partial reduction in membrane potential.
The effect of Mfn-2 on mitochondrial morphology was also examined in cultured neonatal rat cardiac myocytes treated with Mfn-2-specific siRNAs. This approach led to significant reductions in Mfn-2 mRNA and protein levels (results not shown). Mfn-2 downregulation was associated with fragmentation of the elongated and interconnected mitochondria into numerous smaller spherical mitochondria (Fig. (Fig.4F,4F, upper panels). However, this was not the case in adult myocytes, where Mfn-2 ablation led to increased mitochondrial size (Fig. (Fig.4F,4F, lower panels). These data suggest that the effects of Mfn-2 on mitochondrial morphology can be greatly affected by the cellular context.
To assess the effects of Mfn-2 ablation on mitochondrial function, we examined the activities of mitochondrial enzymes in whole-heart preparations or the respiratory activity of isolated interfibrillar mitochondria (IFM) and subsarcolemmal mitochondria (SSM) (Fig. (Fig.5).5). Deletion of Mfn-2 did not affect the activities of citrate synthase (CS), isocitrate dehydrogenase (IDH), or medium-chain acyl-CoA dehydrogenase (MCAD) in whole tissue (Fig. (Fig.5A)5A) and in IFM and SSM (results not shown), consistent with normal biogenesis of mitochondrial mass in the absence of Mfn-2. The sizes of isolated mitochondria were assessed using a flowmetric approach. As shown in Fig. Fig.5B,5B, the mitochondrial volume does not change significantly upon deletion of Mfn-2 when comparisons are made between IFM. Nevertheless, the absence of Mfn-2 is associated with a significant increase in volume in SSM (Fig. (Fig.5C),5C), in agreement with the notion that loss of Mfn-2 results in the formation of a subset of enlarged mitochondria that coexist with structurally normal mitochondria. Using a number of different substrates, the rates of ADP-driven (state III) oxygen consumption in isolated IFM and SSM were found to be similar, regardless of the presence/absence of Mfn-2 (Fig. 5D and E). Furthermore, levels of state IV respiration were found to be similar between the two groups, and the respiratory control ratio was unaffected (results not shown), indicating that Mfn-2 is dispensable for normal coupling of the respiratory chain.
Despite the relatively normal metabolic function of Mfn-2-deficient mitochondria, there was a pronounced resistance to Ca2+-induced mitochondrial permeability transition (MPT). As shown in Fig. Fig.66 A, the incremental infusion of Ca2+ in isolated SSM and IFM can lead to a marked increase in extramitochondrial Ca2+. This is attributed to the formation of the high-conductance MPT pore (MPTP), which mediates an exponential release of the previously accumulated Ca2+ (41, 51). This assay shows that the loss of Mfn-2 can significantly delay the MPTP, as judged by the comparison of the Ca2+ release curves in control and Mfn-2-deficient mitochondria, an effect that can be observed in both SSM and IFM (Fig. (Fig.6A).6A). The cumulative Ca2+ load required for MPTP opening (mitochondrial Ca2+ tolerance) was also increased in Mfn-2 SSM and IFM (Fig. (Fig.6B).6B). This comparison shows that Mfn-2-deficient mitochondria required approximately twice the Ca2+ load applied to wild-type mitochondria to induce MPTP opening.
MPTP opening was also assessed by measuring the gradual swelling of isolated mitochondria in the presence of Ca2+. Mitochondria with or without Mfn-2 were exposed to 200 μM Ca2+, and swelling was monitored as a decrease in absorbance over time. As shown in Fig. Fig.77 A and B, the change in absorbance (relative to the baseline absorbance) is more pronounced in wild-type mitochondria than in mutant mitochondria, indicating that the absence of Mfn-2 is associated with an attenuated MPT response. In the presence of cyclosporine (CsA), mitochondria from both groups maintained their optical density throughout the assay (results not shown), suggesting that the decrease in absorbance seen here is attributable to Ca2+-induced MPTP opening. As an alternative way to assess MPT, we also examined the ability of mitochondria, previously swollen by Ca2+, to undergo shrinkage after being exposed to polyethylene glycol (PEG). As shown in Fig. Fig.7C,7C, the addition of PEG in F/F;− mitochondria is associated with a gradual increase in absorbance that signifies mitochondrial shrinkage facilitated by MPTP opening. On the other hand, the addition of PEG to preswollen Mfn-2-depleted mitochondria resulted in a lower rate of mitochondrial shrinkage, indicating a less-than-optimal MPTP opening.
To directly assess the effect of Mfn-2 deletion on the expression of proteins previously associated with the function of the MPTP, we analyzed whole-heart extracts by Western blotting. As shown in Fig. Fig.7D,7D, the levels of the MPTP-regulatory component Cyp-D were not different between F/F;cre and F/F;− extracts. This was also the case for the other purported MPTP components, such as the voltage-dependent anion channel (VDAC)-porin, and the two isoforms of the adenine nucleotide translocase (ANT1/2). Identical results were obtained with extracts from isolated mitochondria (results not shown). These data indicate that the loss of Mfn-2 is sufficient to affect MPT in the absence of significant changes in the levels of candidate MPTP components.
To assess the consequences of Mfn-2 ablation on stress-induced MPTP opening in intact cells, myocytes were isolated and examined for permanent mitochondrial depolarization under conditions of ROS generation, which is known to promote loss of membrane potential via MPTP activation (85). As shown in Fig. Fig.88 A, polarized mitochondria are detected as bright rectangles due to the accumulation of TMRM. Laser illumination of TMRM-loaded mitochondria, leading to a tightly controlled local generation of ROS (4), allows the assessment of MPTP activation downstream of ROS. Representative time points (1, 5, and 9 min [t1, t5, and t9]) are shown in the middle panels of Fig. Fig.8A,8A, where the gradual depolarization of the mitochondrial population is depicted. As shown in Fig. Fig.8B8B (also see the movie in the supplemental material), the time course of mitochondrial depolarization is significantly delayed in myocytes lacking Mfn-2 (blue tracing), compared to the depolarization in myocytes with normal levels of Mfn-2 (purple tracing). To further determine the critical involvement of MPTP in the outcome of this assay, we analyzed the effects of CsA pretreatment (a potent inhibitor of MPTP) on mitochondrial depolarization. As shown in Fig. Fig.8B,8B, the addition of CsA significantly delays the rate of depolarization in wild-type mitochondria (orange tracing), while it appears to act additively with the Mfn-2 deficiency to induce further delays in loss of membrane potential (green tracing). These actions are also observed when calculating the time to half depolarization (T50) in the different groups (Fig. (Fig.8C).8C). We further examined the response of adult myocytes to exogenous H2O2 as an alternative source of ROS. As shown in Fig. 8D, H2O2 exposure is able to induce time-dependent mitochondrial depolarization and, eventually, hypercontracture in isolated myocytes. In agreement with the observations made above, the loss of Mfn-2 is associated with a reduced rate of mitochondrial depolarization (Fig. 8E and F), providing additional evidence for the involvement of Mfn-2 in MPT.
We also examined ROS-induced MPTP in neonatal rat cardiac myocytes that were depleted of Mfn-2 via siRNA knockdown. As shown in Fig. Fig.99 A, mitochondrial depolarization and TMRM fluorescence decay in response to H2O2 exposure are accelerated upon Mfn-2 knockdown, which is in contrast to findings for adult myocytes. The enhanced depolarizing effect in the absence of Mfn-2 in NRCMs was also associated with an increased release of lactate dehydrogenase (LDH), a marker of cell death (Fig. (Fig.9B).9B). To examine the role of Mfn-2 deficiency in an independent system, mice lacking Mfn-2 in macrophages were constructed by breeding the Mfn-2loxP mice with the LyzM-cre transgenic mouse line. In macrophages recruited to the peritoneum by thioglycolate treatment, Mfn-2 deficiency is protective against mitochondrial depolarization (Fig. (Fig.9C),9C), in agreement with the observations made with adult cardiac myocytes. The delay in mitochondrial depolarization in Mfn-2 macrophages was also associated with a reduced release of LDH into the culture medium (Fig. (Fig.9D).9D). Taken together, these data show that the effect of Mfn-2 insufficiency on MPTP activation is cell type dependent.
Based on the observation that F/F;cre mice have normal cardiac function at baseline but contain mitochondria that exhibit resistance to MPTP opening, we examined their response to reperfusion injury using the isolated heart configuration. As shown in Fig. Fig.10,10, F/F;− and F/F;cre hearts had similar systolic and developed pressures at baseline (Fig. 10A and B) and global ischemia for 10 min led to the expected reduction in cardiac function in both groups. However, upon reperfusion, Mfn-2-ablated hearts were able to produce higher systolic pressures than control hearts, indicating a protection from the injurious effects of the reflow (Fig. 10A). Similar observations were also made when developed pressures were measured in the two groups (Fig. 10B). Therefore, these experiments demonstrate that loss of Mfn-2 can alleviate some of the detrimental effects of postischemic reperfusion, the period that is known to coincide with the induction of MPT (35). To obtain further molecular details on the Mfn-2-associated cardioprotection during ex vivo ischemia/reperfusion (I/R), Western blot analysis of purified mitochondrial protein was performed. As shown in Fig. 10C, the antiapoptotic protein Bcl-2 is found to be more abundant on Mfn-2-depleted mitochondria.
To examine the impact of Mfn-2 loss on cell death after I/R in greater detail, an in vitro hypoxia/reoxygenation assay was employed. In this experiment, cardiac myocytes purified from wild-type and Mfn-2-deficient hearts were exposed to normoxia only or were exposed to hypoxic conditions (1% O2, 5% CO2) and then returned to normoxic conditions (reoxygenation). As shown in Fig. Fig.1111 A and B, the percentage of myocytes undergoing necrotic cell death (indicated by the number of trypan blue-stained cells) decreases in the absence of Mfn-2. In fact, Mfn-2-deficient cardiac myocytes were able to better tolerate the stress induced by the isolation process, as there were fewer cells from this genotype staining positive for trypan blue under normoxic conditions. Consistently, the Mfn-2-deficient myocytes retained their resistance to necrosis under the hypoxia/reoxygenation conditions (Fig. 11B). It has been previously reported that Mfn-2 can also promote cardiomyocyte apoptosis by activating the intrinsic/mitochondrial pathway (75). We therefore examined the activation of apoptosis in F/F;− and F/F;cre myocytes from the above-described assay by Western blotting. As shown in Fig. Fig.1111 C, the abundance of cleaved caspase-9 and cleaved PARP-1 is decreased in F/F;cre samples compared to F/F;− samples, both at normoxia and upon hypoxia/reoxygenation, suggesting inhibition of the apoptotic pathway in the absence of Mfn-2. The tolerance of Mfn-2-deficient myocytes to cell death was further demonstrated using H2O2 treatment. As shown in Fig. 11D, the proportion of dead cells in myocyte preparations left untreated or exposed to 20 μM H2O2 is consistently lower in the absence of Mfn-2. Taken together, these observations show that loss of Mfn-2 is associated with improved cell survival in the face of death-inducing stimulation.
To further assess the role of Mfn-2 in cardiac myocyte apoptosis, we examined the expression of proapoptotic and antiapoptotic proteins that are known to be associated with mitochondrial morphogenesis (45). As shown in Fig. Fig.1212 A, the proapoptotic protein Bax was found to be increased in nonischemic heart extracts lacking Mfn-2. The antiapoptotic Bcl-2 appeared to be upregulated in the same experimental group, in agreement with our previous observations (Fig. 10C). We also examined the expression of the mitochondrial fission factor Drp-1, which has also been implicated in ischemia/reperfusion injury and cell death (63). As shown in Fig. 12A, the levels of Drp-1 protein were decreased in hearts lacking Mfn-2. Finally, the levels of the protein Opa-1, which is implicated in inner mitochondrial membrane fusion and also in the regulation of cytochrome c release, did not appear to change significantly upon genetic deletion of Mfn-2.
Finally, we subjected Mfn-2 F/F;− and F/F;cre mice to in vivo ischemia/reperfusion injury by surgically closing and reopening the LAD coronary artery for 30 min and 2 h, respectively. As shown in representative images (Fig. 12B) and the accompanying histogram (Fig. 12C), the area at risk (AAR) to the left ventricle (LV) area is the same between the two groups, indicating similar magnitudes of ischemic stress. However, the ratio of the infarct area (IA) (shown as the white band in Fig. 12B) to the AAR (IA/AAR) was found to be lower in F/F;cre hearts, indicating a diminished cell death response. In support of the above conclusions, the percentage of TUNEL-positive nuclei was found to be significantly lower in F/F;cre hearts than in F/F;− hearts (Fig. 12D and E), indicating that the apoptotic response under these conditions is mitigated in the absence of Mfn-2.
In adult cardiac myocytes, the compact placement of mitochondria between the myofibrils sets significant constraints in terms of their positioning and structure. Thus, mitochondria in this cell type have been described as independent units, with very little spatial freedom and precise arrangement into a “crystal-like” pattern (10, 81). Despite these limitations, the possibility that a functional cross talk takes place among mitochondria in cardiac myocytes has been documented (4, 14, 15, 85), although the molecular mediators of this process remain largely unknown. The pore that mediates mitochondrial permeability transition (MPTP) is generally linked to cell death events, but recent work has suggested that it may also be involved in physiologically relevant processes, such as mitochondrial Ca2+ efflux (31) and ROS signaling between mitochondria (82), thereby making it an attractive candidate mediator of intermitochondrial cross talk.
Mfn-2 resides on the outer mitochondrial membrane and has been previously shown to regulate mitochondrial fusion, a process that involves the exchange of content among adjacent mitochondria in a variety of cell types (29). Mfn-2 is highly expressed in the adult mammalian heart, yet its roles in this tissue remain to be determined. In the present study, we have ablated the expression of the Mfn-2 gene specifically in murine cardiac myocytes and analyzed the impact of this perturbation on mitochondrial and cardiac functions under baseline and stress conditions.
The deletion of Mfn-2 led to detectable aberrations in the mitochondrial compartment, the most prominent of them being an increase in mitochondrial size that was reflected by differences in mitochondrial cross-sectional area and volume. However, not all of the mitochondria were enlarged in a given myocyte or field, and it was found that the enlarged mitochondria occurred in the subsarcolemmal rather than the interfibrillar compartment. These findings for cardiac myocytes are similar to observations with other mouse models of Mfn-2 ablation. The deletion of Mfn-2 in Purkinje cells produces mitochondria that are enlarged and spherical and appear in clusters (19). Mfn-2-deficient mouse embryo fibroblasts contain enlarged mitochondria that are spherical (18). Likewise, the deletion of the two Mfn-2 alleles and one Mfn-1 allele from skeletal muscle is reported to lead to the formation of a few unusually large mitochondria (20). These features resemble the mitochondrial phenotype identified in the present study, although their extent appears to be less severe here. Despite these structural abnormalities, the total mitochondrial mass in cardiac myocytes was not affected by the deletion of Mfn-2. Furthermore, the respiratory activity of isolated mitochondria was found to be normal, and the levels of expression of genes associated with mitochondrial biogenesis and functions were similar or displayed small differences between wild-type and Mfn-2-depleted hearts.
Mfn-2-deficient cardiac myocytes underwent modest hypertrophy, leading to an increase in the heart weight/body weight ratio that was associated with elevated transcript levels of ANP. However, the function of the intact Mfn-2 null heart appeared normal in most physiological analyses, although isoproterenol stimulation was able to unmask a mild systolic dysfunction. Individual Mfn-2 null myocytes also exhibited a small decrease in contractility when paced ex vivo. In agreement with reports of other systems (5, 18, 67), a reduction in baseline mitochondrial membrane potential could be detected in Mfn-2-deficient cardiac myocytes. Collectively, these analyses identify a modest reduction in function, likely to be attributed to an underlying mitochondrial defect. This partial phenotype may indicate that compensatory mechanisms operate, perhaps through the ability of Mfn-1, to functionally counterbalance the loss of Mfn-2 (20).
In addition to the above-described features at baseline, multiple lines of evidence in the present study suggest that Mfn-2 deficiency can delay MPTP opening in stressed adult cardiac myocytes. First, isolated mitochondria that lack Mfn-2 are capable of higher loads of Ca2+ uptake. Second, the time course of mitochondrial swelling in response to Ca2+ exposure in Mfn-2-depleted mitochondria is distinct from that in wild-type mitochondria. Third, water extrusion through the MPTP due to PEG treatment is diminished in purified mitochondria lacking Mfn-2. Fourth, the mitochondrial depolarization in cardiac myocytes lacking Mfn-2 is resistant to photon stress or free radical-induced MPTP activation. Furthermore, Mfn-2-deficient myocytes are protected from death in hypoxia/reoxygenation and H2O2 exposure assays, an effect likely to be at least in part attributable to MPTP inhibition. In addition, hearts isolated from knockout mice display improved function following global ischemia/reperfusion (I/R) injury ex vivo, and the cardiac cell death response to in vivo regional ischemia and reperfusion injury is attenuated in Mfn-2 knockout mice.
Despite this evidence that Mfn-2 deletion delays MPTP opening in adult cardiac myocytes, the opposite outcome was observed in cultured neonatal myocytes treated with siRNA to ablate Mfn-2. In the neonatal system, Mfn-2 knockdown facilitated the loss of mitochondrial membrane potential in response to ROS stress. Mfn-2-deficient neonatal myocytes also displayed changes in mitochondrial morphology that were strikingly different from those seen after Mfn-2 ablation in adult myocytes (compare panels in Fig. Fig.4F),4F), suggesting that the behavior of mitochondria in the neonatal myocyte system may not be predictive of that of adult myocytes or the intact heart. To further test the effects of Mfn-2 ablation in an independent system, mice lacking Mfn-2 in peritoneal macrophages were constructed and analyzed. The loss of Mfn-2 in this cell type resulted in a delay in mitochondrial membrane depolarization due to ROS stress, in agreement with observations with adult cardiac myocytes. Collectively, these experiments indicate that the actions of Mfn-2 on MPTP are likely to be cell type specific, perhaps depending on the extent of cellular differentiation.
The mechanisms by which Mfn-2 promotes permeability transition in adult cardiac myocytes are likely to be pleiotropic. Opening of the MPTP has been linked to cataclysmic cell events, such as apoptosis and necrotic cell death (6, 52), yet the structure and regulation of the MPTP have not been fully unraveled (40). The initial working model suggested that the MPTP is made of VDAC, ANT, and Cyp-D (23). However, genetic studies have challenged this model, and all isoforms of VDAC, which is the candidate component on the outer mitochondrial membrane, were found to be dispensable for MPTP opening (9). On the other hand, only Cyp-D appears to function as a central regulator of the pore (8, 60), while ANT is likely to function as a peripheral regulator (49). In contrast to Mfn-2, an integral protein of the outer membrane, Cyp-D, is located in the matrix, where it participates in protein folding and is inhibited by CsA. Cyp-D is thought to facilitate MPTP opening by functioning as a Ca2+-sensing element that translates the Ca2+ stimulus into conformational changes of an MPTP component(s) on mitochondrial membranes (56). Like Cyp-D-deficient cardiac mitochondria, Mfn-2-deficient cardiac mitochondria display enhanced capacity for Ca2+ uptake. However, Cyp-D-null mitochondria are completely resistant to swelling induced by low concentrations of Ca2+, whereas Mfn-2-depleted mitochondria undergo swelling in response to this treatment but at substantially lower rates than wild-type mitochondria. Furthermore, Cyp-D-null fibroblasts displayed a robust resistance to H2O2-induced mitochondrial depolarization (8), while Mfn-2-null cardiac myocytes displayed a marked delay but still underwent depolarization upon H2O2 exposure. Finally, Mfn-2-null hearts display significant protection from ischemia/reperfusion injury an effect also observed in Cyp-D-null hearts (8, 60). Thus, whereas Mfn-2 exhibits a number of parallels with Cyp-D and could be a novel regulator of MPTP, it should be noted that the pore can still function in the absence of Mfn-2, ruling out the possibility that it alone is an essential structural component of this poorly defined complex. In this regard, it would be interesting to test whether a functional redundancy exists between Mfn-1 and Mfn-2 as potential regulators of MPTP.
An alternative hypothesis is that MPTP is influenced by Mfn-2 ablation due to diminished coupling between the mitochondria and the SR. Mfn-2 has been shown to regulate the efficiency of mitochondrial Ca2+ uptake from intracellular stores by controlling the distance between the ER and mitochondria (26). Therefore, it can be suggested that Mfn-2 promotes MPTP opening by enforcing mitochondrial Ca2+ uptake from the SR. However, this hypothesis is difficult to reconcile with the following observations: (i) the defect in MPTP activation could be detected on isolated mitochondria that were free from cytosolic contaminants and (ii) the gross structure of the “calcium release domain” was not significantly impaired in Mfn-2-deficient myocytes. Although it is possible that Mfn-2 can influence MPTP activation through subtle effects on SR/mitochondrial connections, this can only partly explain the phenotypes presented here. In fact, the findings from isolated mitochondria suggest that the ability of Mfn-2 to influence the MPTP is likely owing to its localization on mitochondria per se.
Another mechanism could be that Mfn-2 facilitates the permeability transition through its participation in OMM remodeling and fusion. Dynamin-like proteins can introduce destabilization of the lipid bilayer, allowing adjacent mitochondria to merge (58, 69). It is therefore possible that, under stress conditions, Mfn-2-dependent local OMM destabilization can lead to the formation of MPTP. The potential involvement of Mfn-2 in OMM disruption can be in agreement with the finding that Mfn-2 functionally interacts with Bax and Bak in healthy cells (45). Bax and Bak are thought to induce the formation of large pores on the OMM through a process termed mitochondrial outer membrane permeabilization (MOMP), a key step in mitochondrion-mediated apoptosis (3, 80). Although the interaction of Bax/Bak with Mfn-2 was linked to mitochondrial remodeling rather than membrane permeabilization (45), the possibility that Mfn-2 association with Bax/Bak promotes MOMP in cardiac myocytes remains an open possibility. In this regard, we find that genetic deletion of Mfn-2 in heart attenuates the apoptotic response downstream of different ischemic insults. Interestingly, the loss of Mfn-2 was associated with increased levels of Bax but also with a concomitant increase in Bcl-2. Our findings that loss of Mfn-2 leads to inhibition of apoptosis are in agreement with the study from Shen et al. which demonstrates that overexpression of Mfn-2 in cardiac myocytes activates the mitochondrion-dependent apoptotic pathway (75).
Finally, the alterations in the structure of mitochondria due to Mfn-2 ablation could potentially impact MPTP function indirectly, perhaps due to changes in the structure of the mitochondria. However, it should be noted that only a portion of mitochondria in Mfn-2 F/F;cre hearts displayed structural abnormalities, while their delayed response to MPTP activation appeared to be a general property exhibited by mitochondria with either normal or abnormal morphology. This can be seen in Fig. 5B and C, which show that Mfn-2 knockout IFM have the same volume as wild-type IFM but that the SSM become enlarged. Despite these differences in morphology, mitochondria from both compartments display a significant delay in MPTP activation (Fig. 6A and B). It could also be argued that a lower ΔΨm detected in Mfn-2 knockout mitochondria may have secondary effects on MPTP activation. However, it should be noted that a lower ΔΨm has been found to be a strong inducer of MPTP (12); thus, a decrease in ΔΨm in Mfn-2-deficient mitochondria should, if anything, decrease, not increase, their tendency for MPTP opening. Therefore, we conclude that the ability of Mfn-2-deficient mitochondria to undergo delayed MPTP activation is distinct from their baseline structural characteristics.
A close relation between mitochondrion-shaping proteins and apoptotic cell death has previously been recognized (79, 83). For example, Opa-1, which is required for inner mitochondrial membrane fusion, can independently regulate the release of cytochrome c from cristae and control apoptosis (33). Based on the present study, it would be of interest to test whether the ablation of other regulators of mitochondrial morphology/dynamics also affects MPTP and to what extent this may lead to apoptotic or necrotic cell death. In this regard, pharmacologic inhibition of Drp-1 in adult myocytes is reported to attenuate mitochondrial depolarization in response to hypoxia/reoxygenation and to protect hearts from I/R injury (63), and Mfn-2 knockout hearts exhibiting resistance to I/R had consistently lower levels of Drp-1.
In conclusion, Mfn-2 deficiency in cardiac myocytes leads to perturbations in mitochondrial morphology and to disruption of their normal spatial orientation. Despite these changes, mitochondrial and cardiac function is normal or minimally impaired. Surprisingly, Mfn-2-depleted mitochondria are more tolerant of Ca2+ overload, and Mfn-2 deficiency in myocytes protects isolated cells from ROS stress and hearts from ischemia-reperfusion injury, suggesting that Mfn-2 can function to control MPTP opening. Although MPTP opening is linked to cell death events, emerging evidence suggests that it is also involved in physiologically relevant processes, such as ROS signaling (82) and mitochondrial Ca2+ unloading (31). Because mitochondria within adult myocytes display very limited motility, Mfn-2 may primarily function as an MPTP mediator in this context to promote intermitochondrial communication, allowing these organelles to coordinate membrane potential under conditions of Ca2+ and ROS stress.
We thank David C. Chan for the provision of the Mfn-2flox mouse line, Michael D. Schneider for the α-MHC-Cre mouse line, Donald L. Gantz for assistance with electron microscopy, and Michael T. Kirber for help with confocal microscopy. We are also thankful to David R. Pimentel for providing neonatal rat cardiac myocytes and to Akiko Higuchi and Taina Rokotuiveikau for assistance with the animal colony.
This study was funded by National Institutes of Health grants HL061639, HL064750, and NO1-HV-28178 to W. S. Colucci, HL074237 to W. C. Stanley, and HL102874, AG34972, AG15052, and HL68758 to K. Walsh.
Published ahead of print on 18 January 2011.
†Supplemental material for this article may be found at http://mcb.asm.org/.