Mitochondria-localized caveolin in adaptation to cellular stress and injury

Abstract

We show here that the apposition of plasma membrane caveolae and mitochondria (first noted in electron micrographs >50 yr ago) and caveolae-mitochondria interaction regulates adaptation to cellular stress by modulating the structure and function of mitochondria. In C57Bl/6 mice engineered to overexpress caveolin specifically in cardiac myocytes (Cav-3 OE), localization of caveolin to mitochondria increases membrane rigidity (4.2%; P<0.05), tolerance to calcium, and respiratory function (72% increase in state 3 and 23% increase in complex IV activity; P<0.05), while reducing stress-induced generation of reactive oxygen species (by 20% in cellular superoxide and 41 and 28% in mitochondrial superoxide under states 4 and 3, respectively; P<0.05) in Cav-3 OE vs. TGneg. By contrast, mitochondrial function is abnormal in caveolin-knockout mice and Caenorhabditis elegans with null mutations in caveolin (60% increase free radical in Cav-2 C. elegans mutants; P<0.05). In human colon cancer cells, mitochondria with increased caveolin have a 30% decrease in apoptotic stress (P<0.05), but cells with disrupted mitochondria-caveolin interaction have a 30% increase in stress response (P<0.05). Targeted gene transfer of caveolin to mitochondria in C57Bl/6 mice increases cardiac mitochondria tolerance to calcium, enhances respiratory function (increases of 90% state 4, 220% state 3, 88% complex IV activity; P<0.05), and decreases (by 33%) cardiac damage (P<0.05). Physical association and apparently the transfer of caveolin between caveolae and mitochondria is thus a conserved cellular response that confers protection from cellular damage in a variety of tissues and settings.—Fridolfsson, H. N., Kawaraguchi, Y., Ali, S. S., Panneerselvam, M., Niesman, I. R., Finley, J. C., Kellerhals, S. E., Migita, M. Y., Okada, H., Moreno, A. L., Jennings, M., Kidd, M. W., Bonds, J. A., Balijepalli, R. C., Ross, R. S., Patel, P. M., Miyanohara, A., Chen, Q., Lesnefsky, E. J., Head, B. P., Roth, D. M., Insel, P. A., Patel, H. H. Mitochondria-localized caveolin in adaptation to cellular stress and injury.

In addition to sensing extracellular signals that modulate cellular function, the plasma membrane helps protect intracellular structures and activities from insults by the external environment. Caveolae (“little caves”), cholesterol- and sphingolipid-enriched invaginations of the plasma membrane (1), are a subset of lipid (membrane) rafts (2) that facilitate such sensing and contribute to cellular responses to environmental perturbations. Caveolins, scaffolding proteins of caveolae, contribute to the regulation of numerous cellular functions, including endocytosis, calcium homeostasis, and compartmentalization of signaling components (receptors, effectors) (3). Three isoforms, caveolins 1–3 (Cav-1, Cav-2, and Cav-3), function both inside and outside of caveolae (4).

In highly metabolic organs (e.g., heart, brain, and liver), mitochondria play a key role in the adaptation to cellular stress. Lack of oxygen inhibits mitochondrial function, including blockade of ATP synthesis and loss of membrane potential (5). Disruption of mitochondria also occurs during injury as a consequence of increased calcium and reactive oxygen species (ROS; ref. 6), which both contribute to opening of the mitochondrial permeability transition pore (mPTP) and apoptosis (7, 8). Mitochondrial dysfunction is a hallmark of the pathophysiology of highly metabolic organs and occurs in ischemic injury and other settings that include heart failure, diabetes mellitus, Alzheimer's disease, Parkinson's disease, cancer, and aging.

Caveolin-deficient stromal cells have compromised mitochondrial function (9) and mitochondria from Cav-1-knockout (KO) fibroblasts accumulate cholesterol and have severe dysfunction; such cells adapt poorly to nutrient starvation and are predisposed to apoptosis (10). Loss of caveolin alters mitochondrial function in adipose tissue, suggesting a link between caveolins and metabolism (11). Knockdown of caveolin accelerates neuronal aging and decreases adaptation to cerebral and cardiac injury (12–14). By contrast, cardiac-specific overexpression of Cav-3 protects the heart from injury and pressure overload-promoted failure (15, 16). Such findings suggest a link between caveolin and mitochondrial function but whether the two are directly related has not been defined.

Tissue protection from injury may be associated with the formation of plasma membrane signaling microdomains (signalosomes) that can interact with mitochondria (17, 18). For example, disruption of caveolae inhibits s-nitrosylation of mitochondrial proteins, a mechanism that contributes to the protective phenotype (17, 18). In addition, caveolae can form contacts that can communicate membrane-derived signals to other cellular components, e.g., “nanocontacts” between caveolae and the endoplasmic reticulum in smooth muscle cells (19).

Here, we tested the hypothesis that plasma membrane caveolae interact with mitochondria and that this interaction has important functional consequences. We used a variety of experimental approaches and cellular systems, including cardiac and cancer cells to define general features of this interaction. Use of Caenorhabditis elegans as a model organism provided further support for this hypothesis. The results thus identify caveolae-mitochondria interaction and mitochondrial caveolins as previously unrecognized, critical regulators in the adaptation of eukaryotic cells to stress.

MATERIALS AND METHODS

Ethics statement

Animals were treated in compliance with the U.S. National Academy of Science Guide for the Care and Use of Laboratory Animals, and protocols were approved by the Veterans Affairs San Diego Healthcare System Animal Care and Use Committee.

Materials

Antibodies for Cav-3 (monoclonal) were from BD Biosciences (San Jose, CA, USA); antibodies for Cav-3, Cav-1 (polyclonal), and voltage-dependent anion channel (VDAC) were from Abcam (Cambridge, MA, USA); antibodies for prohibitin, adenine nucleotide translocase (ANT), and actin were from Santa Cruz Biotechnology (Santa Cruz, CA, USA); and antibodies for cytochrome c were from Imgenex (San Diego, CA, USA). FITC and Alexa-conjugated secondary antibodies were from Invitrogen (Carlsbad, CA, USA). Other secondary antibodies were obtained from Santa Cruz Biotechnology. Other chemicals and reagents were obtained from Sigma (St. Louis, MO, USA) unless otherwise stated.

Animals

Animals were kept on a 12-h light-dark cycle in a temperature-controlled room with ad libitum access to food and water. Cav-3-overexpressing (OE) mice were produced in a C57BL/6 background as described previously (16). Cav-3-KO mice were created as reported previously and backcrossed 10 generation in the C57BL/6 background (20). Transgene-negative (TGneg) siblings in the C57BL/6 background served as controls for Cav-3-OE and Cav-3-KO mice. Sprague-Dawley rats (250–300 g, male) were used for some studies.

CM preparation

Adult male Sprague-Dawley rats were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg), hearts were excised and retrograde-perfused with medium containing collagenase II, as described previously (21) to isolate ventricular CMs. A similar procedure was used for adult mouse ventricular myocyte isolation with slight variations (14).

Immunofluorescence and deconvolution microscopy of CMs

Samples were prepared for immunofluorescence microscopy, and images were deconvolved as described previously (21).

Mitochondrial isolation

Mice were sacrificed, and hearts were removed. Ventricles were placed in ice-cold mitochondrial isolation medium (MIM: 0.3 M sucrose, 10 mM HEPES, 250 uM EDTA), minced, and homogenized with a Tissuemiser (Fisher Scientific, Waltham, MA, USA). Homogenates were rinsed in MIM. Samples were centrifuged at 600 g to clear nuclear/membrane debris. The resulting supernatant was spun at 8000 g for 15 min. The resulting pellet was resuspended in MIM in the presence of 1 mM BSA, followed by another 8000-g spin for 15 min. The resulting pellet was resuspended in isolation buffer with BSA and spun again at 8000 g. Metabolically active mitochondria were then suspended in 150 μl MIM for functional studies.

To isolate pure mitochondria for biochemical and electron microscopy (EM) analysis, the washing steps were repeated with MIM in a final 2-ml resuspension of the pellet in mitochondrial resuspension buffer (MRB; 500 μM EDTA, 250 mM mannitol, and 5 mM HEPES). The mitochondria were layered on top of a 30% Percoll/70% MRB solution. The Percoll gradient was spun at 95,000 g for 30 min. The mitochondrial band was removed from the gradient, and volume was increased 10-fold with MRB to remove the Percoll by centrifugation at 8000 g for 15 min. The mitochondrial pellet was resuspended in 50–150 μl of MRB and subjected to further analysis.

Purification of subsarcolemmal mitochondria (SSM) and interfibrillary mitochondria (IFM)

Cardiac SSM and IFM populations were isolated using the method of Palmer et al. (22), modified as described previously by the use of Chappell–Perry buffer (100 mM KCl, 50 mM Mops, 1 mM EGTA, 5 mM MgSO₄, and 1 mM ATP, pH 7.4) for mitochondrial isolation (23).

Membrane fractionation of mitochondria

Purified mitochondria were lysed in buffer containing 150 mM Na₂CO₃ (pH 11.0) and 1 mM EDTA, and then sonicated on ice with 3 cycles of 20-s bursts. Approximately 1 ml of homogenate was mixed with 1 ml of 80% sucrose in 25 mM 2-(N-mopholino)ethanesulfonic acid (MES) and 150 mM NaCl (MBS; pH 6.5) to form 40% sucrose and loaded at the bottom of an ultracentrifuge tube. A discontinuous sucrose gradient was generated by layering 6 ml of 35% sucrose prepared in MBS followed by 4 ml of 5% sucrose (also in MBS). Gradients were centrifuged at 175,000 g using a SW41Ti rotor (Beckman Coulter, Fullerton, CA, USA) for 3 h at 4°C. Samples were removed in 1-ml aliquots to form 12 fractions and fractions probed for specific proteins.

EM

Cells or tissues were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer for 2 h and postfixed in 1% OsO₄ in 0.1 M cacodylate buffer (1 h) at room temperature, and then embedded as monolayers in LX-112 (Ladd Research, Williston, VT, USA), as described previously (24). Sections were stained in uranyl acetate and lead citrate prior to EM (Jeol 1200 EX-II; Jeol Ltd., Akishima, Japan; or Philips CM-10; Philips, Amsterdam, The Netherlands). Caveolae were quantified on random images per length of membrane. For immunogold labeling, purified mitochondria were fixed in 4% paraformaldehyde in 10 mM phosphate buffer (pH 7.4), cryoprotected with 2.3 M sucrose, and frozen in liquid nitrogen. Ultrathin cryosections were cut at −100°C using a Leica Ultracut UCT Microtome with an EMFCS cryoattachment (Leica Microsystems, Wetzlar, Germany), placed on glow-discharged nickel grids, stored on 2% gelatin and PBS at 4°C, and incubated with primary antibodies, followed by 5 or 10 nm gold, and then goat anti-rabbit or anti-mouse IgG in PBS with 10% fetal calf serum. Mitochondrial cross-reactions with either mouse or rabbit antibodies are commonly observed. Therefore, antibodies were also tested on mitochondria isolated from Cav-3-KO mice to test specificity. Only antibodies with limited background label were used in the analysis. Grids were absorption stained with 0.2% neutral uranyl acetate and 0.2% methyl cellulose and viewed directly on the microscope.

Calcium pulse

Opening of the MPT pore after in vitro Ca²⁺ overload was assessed by following changes in the membrane potential (Δψ_m) by using the fluorescent dye rhodamine 123 (50 nM; Invitrogen, Carlsbad, CA, USA) in the presence of pyruvate and malate (5 mM) (25). Fluorescence was monitored with an Infinite M200 plate reader (Tecan Group Ltd., Männedorf, Switzerland). Excitation and emission wavelengths were set to 503 and 527 nm, respectively. Isolated mitochondria (0.5 mg/ml) were suspended in 1.0 ml recording buffer containing 220 mM sucrose, 10 mM 4-[2-hydroxyethyl] piperazine-1-ethanesulfonic acid, and 10 mM KH₂ PO₄ (pH 7.3). At the end of the preincubation period, pulses of 10 μM CaCl₂ were administered at 60-s intervals. After sufficient Ca²⁺ loading, MPT pore opening results in a sudden collapse of Δψ_m. To achieve complete mitochondrial depolarization, 1 μM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) was added to the buffer at the end of the experiment. The amount of Ca²⁺ necessary to trigger this sudden collapse of Δψ_m was used as an indicator of the susceptibility of the MPT pore to Ca²⁺ overload.

Calcium swelling

Calcium swelling was measured on an Infinite M200 plate reader at 540 nm over a span of 20 min. Crude mitochondria (0.5 μg/μl) in the absence of calcium were loaded onto a clear flat bottom 96-well plate and challenged with 250 μM calcium, with absorbance measured every 10 s. Change at 540 nm was compared between samples.

Mitochondrial respiration

Mitochondrial respiratory function was studied according to published protocols (26). Oxygen consumption was measured using a Clark-type oxygen electrode (Oxygraph; Hansatech, Norfolk, UK) during the sequential additions of substrates and inhibitors to purified mitochondria. Purified mitochondria (∼100–200 μg protein) were added to the oxymetry chamber in a 300 ml solution containing 100 mM KCl, 75 mM mannitol, 25 mM sucrose, 5 mM H₃PO₄, 0.05 mM EDTA, and 10 mM Tris-HCl, pH = 7.2 at 37°C. After 2 min equilibration, 5 mM pyruvate and 5 mM malate were added and oxygen consumption followed for ∼1–2 min (state 4). ADP (250 μM) was added to measure state 3 (phosphorylating) respiration. To switch from NAD⁺- to FAD⁺- linked respiration, we first eliminated complex I through the inhibition of the back electron transfer using 0.5 mM rotenone and triggered complex II activity by the addition of 10 mM succinate. Next, we inhibited complex III by the addition of 5 mM anti-mycin A. Complex IV activity was measured in the presence of 0.5 mM 2,2,4-trimethyl-1,3-pentanediol (TMPD) and 2 mM ascorbate. Finally, to test the integrity of the mitochondrial outer membrane, we followed the oxygen consumption rate on the addition of 10 mM cytochrome c, which is expected to enhance oxygen consumption if the outer membrane was compromised during the isolation. Oxygen utilization traces and rate determinations were obtained using Oxygraph software and normalized to protein.

Reactive oxygen species (ROS) assessment using dihydroethidium (DHE) staining and electron paramagnetic resonance (EPR)

For DHE assay, isolated adult mouse CMs were incubated with DHE (5 μM) for 30 min at 37°C in a humidified incubator. After incubation, cells were washed with PBS, hypoxia medium lacking glucose was added, and cells were placed in a hypoxia chamber for 60 min. Following hypoxia, fresh medium was added to cells. After 20 min in fresh medium, the cells were imaged for DHE fluorescence. As DHE is oxidized by ROS, it becomes ethidium and accumulates in the nucleus, so we quantified nuclear specific fluorescence between the groups. For EPR studies, immediately after mixing mitochondria (0.1–0.2 mg of protein) with 70 mM 5-(diisopropoxyphosphoryl)-5-ethyl-1-pyrroline-N-oxide (DEPMPO) and appropriate combinations of the substrates, the mixture was loaded into 500-μl glass capillary tubes and introduced into the EPR cavity of a MiniScope MS300 benchtop spectrometer (Magnettech GmbH, Berlin, Germany). We confirmed that the detected EPR signals are substrate specific, and not due to redox cycling in the studied mixtures, by lack of signals when DEPMPO was mixed with combinations of substrates and inhibitors in the absence of mitochondria. Assignment of the observed signals from mitochondria was confirmed through computer-assisted spectral simulation using the WinSim software (http://epr.niehs.nih.gov/pest.html). In most cases, a mixture of signals due to DEPMPO-OOH and DEPMPO-OH adducts, with occasional contribution from a carbon-centered radical, was detected, but the complete removal of these signals on the inclusion of SOD confirmed that superoxide radical was the exclusive source of the observed EPR-active species. Signals were quantified by measuring the peak amplitudes of the observed spectra and normalized by mitochondrial protein concentrations.

EPR for membrane fluidity

Hydrocarbon chain mobility was measured using fatty acid spin labeling EPR analysis using 5-nitroxyl stearate (5-DSA) as a spin probe (27, 28). The number designation indicates the relative position of the nitroxide on the stearic acid relative to the polar carboxylic group. In the case of 5-DSA, the spin probe is firmly held in place by the head groups of the lipids, which is reflected in broad EPR lines. Plasma membrane or purified mitochondria was incubated for 15 min with 5-DSA (1 mM final concentration) at 25°C. The mixture was then loaded into a 50-μl glass capillary tube and inserted into the EPR cavity of a MiniScope MS200 benchtop spectrometer (Magnettech), maintained at 37°C, where the EPR spectra registered. EPR conditions were the following: microwave power, 5 mW; modulation amplitude, 2 G; modulation frequency, 100 kHz; sweep width, 150 G centered at 3349.0 G; scan rate, 7.5 G/s, with each spectrum representing the average of 5 scans. The fluidity parameters T_‖ and T_⊥ were used to calculate the order parameter, as described previously (27, 28).

C. elegans strains

C. elegans were cultured using standard conditions, and N2 was used as wild type (WT; ref. 29). The cav-1(ok2089) and cav-2(hc191) strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the U.S. National Institutes of Health National Center for Research Resources (NCRR).

ATP assay

Mitochondria ATP production was measured by the ATP Determination Kit (Invitrogen, Carlsbad, CA, USA) as described by the manufacturer. Functional mitochondria were normalized to 5 μg/μl, and 10 μl was added to 90 μl of the reaction mix. Samples were loaded into a Tecan Infinite M200 plate reader, and luminescence was quantified over 10 min.

Generation of stable cell lines

We purchased human colon cancer cell lines, HCT116 and HT29, from American Type Culture Collection (ATCC; Rockville, MD, USA). Both cell lines were cultured in McCoy's 5a medium containing 10% fetal bovine serum and 1% penicillin/streptomycin in a 95% air, 5% CO₂ humidified atmosphere at 37°C. Stably transfected HT29 cells were supplemented with 800 μg/ml G418. For generation of stable cell lines, mouse Cav-1 (456 bp) sequence was cloned into the pTurboRFP-mito vector (Evrogen, Moscow, Russia). The original vector containing red fluorescent protein (RFP) was used as a control. Cells (5×10⁵/well) were seeded on 24-well plates without antibiotics and cultured overnight. On the next day, 0.8 μg of pTurbo-mito RFP or pTurbo-mito Cav-1 plasmid was transfected into HT29 cells using 2 μl of Lipofectamine 2000 in Opti-MEM Reduced Serum Medium following the manufacturer's instructions. Transfected cells were selected with G418 at 800 μg/ml for 4 wk, and the expression of Cav-1 protein in selected cell clones was determined by Western blotting. The mitochondrial-targeted caveolin scaffolding domain (mito-CSD) was generated by direct cloning of synthesized DNA with selective restriction enzyme ends into the pTurboRFO vector with RFP deletion. Transient transfections were performed in HCT116 cells using Lipofectamine 2000.

Apoptosis assay

After 24 h treatment with TNF-related apoptosis-inducing ligand (TRAIL), apoptosis was quantified by nucleosomal fragmentation (Cell Death Detection ELISA plus; Roche Applied Science, Indianapolis, IN, USA). The absorbance values were normalized to those from control-treated cells to derive a nucleosomal enrichment factor at all concentrations as per the manufacturer's protocol.

Mitochondrial membrane potential

JC-1, a lipophilic cationic dye that enters mitochondria and shifts its fluorescence from green to red when Δψ_m increases, was used to assess Δψ_m in the setting of TRAIL stress. Following treatment, 5 × 10⁵ cells were washed with PBS and incubated with 2 μM JC-1 fluorescent probe in dark for 15 min at 37°C. Cells were washed again with PBS and analyzed immediately by flow cytometry. Detection setting was JC-1 green fluorescence at 530 nm and JC-1 red fluorescence at 590 nm. The ratios of green to red fluorescence intensity values were calculated, and the percentage increase in these ratios were expressed as measures of percentage increase in ΔΨ_m.

Generation of adenoassociated virus 9 (AAV9) vector and TaqMan assay

The construct used to express inner mitochondrial membrane (IMM)-targeted Cav-3 was generated by amplifying mouse Cav-3 cDNA by PCR and inserting it into the BamHI and XbaI sites of the pTurboRFP-mito vector (Evrogen). This removed the RFP sequence and replaced it with Cav-3. AAV9.IMM.Cav-3 was generated by integration of this construct into AAV9. To determine viral copy number, TaqMan probes were designed to bind the viral backbone of AAV9, and a TaqMan assay was performed on various tissues.

In vivo ischemia-reperfusion (I/R) injury

Pentobarbital (80 mg/kg)-anesthetized mice were mechanically ventilated, and ischemia was produced by occluding the left coronary artery with a 7-0 silk suture on a tapered BV-1 needle (Ethicon, Somerville, NJ, USA) for 30 min (30). After 30 min occlusion, the ligature was released and the heart was reperfused for 2 h. Ischemic preconditioning (IPC) was induced by occlusion of the left coronary artery for 5 min followed by 15 min reperfusion just prior to ischemia. The area at risk (AAR) was determined by staining with 1% Evans blue (1.0 ml, Sigma) (30). The heart was immediately excised and cut into 1 mm slices (McIlwain tissue chopper; Brinkmann Instruments, Westbury, NY, USA). Left ventricle was counterstained with 1% 2,3,5,-triphenyltetrazolium chloride (Sigma). Images were analyzed by Image-Pro Plus (Media Cybernetics, Inc., Bethesda, MD, USA), and infarct size was determined by planimetry.

Statistics

All data are presented as means ± sem. GraphPad Prism 4 software (GraphPad Software, Inc., San Diego, CA, USA) was used for all statistical analysis. Statistical analyses were performed by unpaired Student's t test (2-tailed testing) or 1-way ANOVA followed by a Bonferroni's post hoc test.

RESULTS

Caveolae are closely apposed to mitochondria

Immunohistochemical staining revealed that Cav-3 is expressed in a punctate pattern along the sarcolemmal membrane and in transverse striations within the interior of adult CMs. Cav-3 colocalizes with cytochrome c, a mitochondrial marker, along the sarcolemmal membrane (Fig. 1A, ∼70% colocalization within 2 μm of the surface). By contrast, Cav-3 and cytochrome c stain in parallel, transverse patterns with minimal colocalization in the cellular interior. EM reveals that CMs have abundant caveolae closely apposed to SSM (Fig. 1B), which in turn are in proximity to regions of the plasma membrane that contain caveolae (Fig. 1C): ∼50% of total caveolae per micrometer of sarcolemma associate with SSM (Fig. 1D). A subset of caveolae on the sarcolemmal membrane is thus closely juxtaposed to mitochondria, suggesting functional interaction between caveolae and mitochondria.

Figure 1.

Caveolae are closely apposed to mitochondria, and IPC is associated with the formation of tethers between caveolae and mitochondria. A) Caveolae (Cav-3) and mitochondria (Cyto C) show close association at the sarcolemmal membrane but not internal regions of adult rat CMs. B) EM image shows close apposition of caveolae and mitochondria. C, D) Nearly all SSM are associated with caveolae (C), but only about half of the membrane caveolae are associated with mitochondria (D), suggesting the existence of mitochondrial-associated and -unassociated pools of caveolae. E) IPC shows evidence of increased association of caveolae-mitochondria. Arrows denote caveolae; triangles represent unidentified structures that link mitochondria to caveolae. F) Higher magnification of the micrograph shown in E. G) No such structures are evident in controls hearts. Scale bars = 50 nm. H) EM quantification after IPC showed increased association of caveolae and mitochondria.

IPC promotes connections between caveolae and mitochondria and the transfer of Cav-3 protein to mitochondria

Caveolae contribute to signal transduction, including the protection of CMs from I/R injury (3). Cellular signaling events that protect from I/R injury converge on the mitochondria (31). Given the close apposition of caveolae and mitochondria in CMs and their importance in protection from I/R injury, we investigated whether caveolae and mitochondria have direct connections during ischemic stress. We thus subjected mice to 5 min of ischemia by occluding the coronary circulation (i.e., a protective preconditioning stimulus involving sublethal stress; ref. 32) and then allowed recovery for 15 min. We identified structures between caveolae and mitochondria in the preconditioned hearts (Fig. 1E, F); i.e., in response to the IPC stimulus, but these structures are not found in untreated hearts (Fig. 1G). Preconditioning increased the association of caveolae with mitochondria (Fig. 1H). IPC thus induces the formation of caveolae and increases their association with mitochondria.

Exposure of adult CMs to sublethal stress (10 min oxygen/glucose deprivation) also generates a connection between caveolae and mitochondria; this occurs 15 min after the stress but is lost by 60 min, at which time vesicles are observed in SSM (Fig. 2A). These results suggest that a transient connection forms between caveolae and mitochondria but progresses to a fusion with mitochondrial membranes and an intermingling of constituents in caveolar and mitochondrial membranes.

Figure 2.

IPC modifies the association of caveolae and mitochondria and Cav-3 localization to mitochondria. A) Adult CMs under basal conditions show close apposition, which transitions to a physical association with sublethal stress and finally to mitochondrial internalized structures at 60 min poststress (denoted by arrowheads). B) Cav-3 localizes to the IMM, mitochondrial matrix, and potentially the outer mitochondrial membrane. C) The majority of Cav-3 in mitochondria is present in SSM but not IFM. D) IPC increased the mitochondrial localization of Cav-3 (normalized to ANT); n = 4.

Immunogold labeling revealed that Cav-3 localizes in the IMM, matrix, and perhaps also on the outer mitochondrial membrane (Fig. 2B). Probing for Cav-3 protein in SSM and IFM revealed that most mitochondrial Cav-3 is in SSM (Fig. 2C), implying its preferential transfer from plasma membrane caveolae to SSM but not to all mitochondria.

This transfer of caveolin protein occurs in response to cellular stress. Mice subjected to a preconditioning stimulus and allowed to recover for 60 min have increased mitochondrial localization of Cav-3 (Fig. 2D; protein levels normalized to ANT, an IMM protein). In view of the important role of mitochondria in limiting I/R injury (33, 34), the association of caveolae and mitochondria following sublethal ischemia and the transfer of caveolin to mitochondria is a previously unappreciated cellular response and suggests that plasma membrane caveolae “sense” stress, perhaps helping mitochondria in proximity to the plasma membrane resist damage and maintain cellular function.

Mitochondria from Cav-3-OE mice have increased Cav-3 protein, improved Ca²⁺ tolerance and respiration, and altered mitochondrial membrane structure

Transgenic mice with CM-specific Cav-3 overexpression have increased sarcolemmal caveolae and improved functional recovery following I/R injury; e.g., decreased infarct size and apoptosis than do TGneg controls (16). CM-specific Cav-3-OE mice thus have cardioprotection akin to that of WT mice undergoing IPC (16); mitochondria of Cav-3-OE mice have greater Cav-3 protein levels than do TGneg mitochondria (Fig. 3A).

Figure 3.

Cav-3-OE mice show cardiac protection; their mitochondria have increased Cav-3 and exhibit Ca²⁺ tolerance. A) Mitochondria from Cav-3-OE mice have increased Cav-3 expression (normalized to prohibitin and VDAC loading). B) Mitochondria from Cav-3-OE mice have increased tolerance to calcium pulsing. n = 4–6. C) Mitochondria isolated from Cav-3-OE mouse hearts are more resistant to calcium swelling. n = 4–6. D) Mitochondria from Cav-3-OE mouse hearts show increased respiratory rates during state 3, as triggered by malate + pyruvate + ADP, and when complex IV is artificially stimulated by TMPD + ascorbate (n=5–6). E) 5-DSA probes membrane rigidity close to the hydrophilic surface of the membrane, while 16-DSA probes the fluidity in the core of the phospholipid bilayer. F) Cav-3-OE mouse hearts have increased order parameter at the membrane interface in both the plasma and mitochondrial membranes, as revealed by the analysis of 5-DSA spectra. n = 5. *P < 0.05.

To test whether the increased Cav-3 in mitochondria of Cav-3-OE mice contributes to enhanced cardiac protection and improves mitochondrial function, we challenged purified mitochondria with calcium overload. Calcium overload in the mitochondrial matrix can occur during I/R injury and open the mPTP, depolarize mitochondrial membrane potential, release cytochrome c, and produce apoptosis (7, 8). We monitored mitochondrial membrane potential in TGneg and Cav-3-OE mice in the presence of increasing calcium concentration and found that mitochondria from Cav-3-OE mice have greater recovery and maintain membrane potential at higher calcium concentrations than do mitochondria from TGneg mice (Fig. 3B). Cav-3-OE mitochondria are also more resistant to calcium swelling than are TGneg mitochondria (Fig. 3C). Thus, mitochondrial Cav-3 may help maintain calcium homeostasis in mitochondria and prevent mitochondria-mediated apoptosis during I/R.

Mitochondria from the hearts of Cav-3-OE mice (compared to those from TGneg mice) have greater respiratory rates during state 3 in response to addition of malate, pyruvate, and ADP (Fig. 3D), indicating more efficient function of complex I in the electron transport chain. Such mitochondria also have increased complex IV oxygen consumption in response to TMPD and ascorbate (Fig. 3D) but no changes with succinate, suggesting limited effects on complex II. The heart is almost entirely dependent on mitochondria-generated ATP for contractile energy; thus, improvement in mitochondrial oxidative phosphorylation implies better contractile function during reperfusion. Increased respiratory rates in mitochondria from Cav-3-OE mice is predicted to increase ATP production and may account for the enhanced functional recovery of those mice during I/R injury.

We used EPR and phospholipid spin probes to assess membrane fluidity and determine whether mitochondria-localized caveolin alters mitochondrial structure (Fig. 3E). We found that membrane order parameter, indicative of the ratio of membrane rigidity to fluidity, increases in the cardiac sarcolemma and mitochondria of Cav-3-OE mice (Fig. 3F), suggesting that an increase in expression of caveolin alters membrane structure and may modulate mitochondrial function.

Cav-3 overexpression reduces ROS generation in CMs and mitochondria, whereas loss of caveolin enhances mitochondrial dysfunction

Reperfusion leads to increased utilization of cellular oxygen for generation of ROS, rather than energy production. ROS generation along with mitochondrial dysfunction contributes to myocardial I/R injury (35). ROS have destructive potential because of their high reactivity with molecules, including lipids, proteins, and DNA. CMs from Cav-3-OE mice have decreased generation of superoxide in response to oxygen/glucose deprivation for 1 h followed by 20 min recovery (Fig. 4A, B); by contrast, CMs from Cav-3-KO mouse hearts have increased superoxide production, which may contribute to cardiac pathology in these animals (13, 36). Consistent with this idea, mitochondria from TGneg mouse hearts produced greater superoxide signals with the addition of malate and pyruvate in state 3 (with ADP) and 4 (no ADP) compared to mitochondria from Cav-3-OE mice (Fig. 4C, D). This result implies that complex I is more tightly coupled in Cav-3-OE mitochondria and is likely a source of ROS leakage in mitochondria from TGneg mice. No changes were observed with addition of succinate, suggesting that complex II is not involved in the response. These results also confirm that the increased oxygen consumption of mitochondria from Cav3-OE mice produces energy and not increased amounts of ROS.

Figure 4.

Cav-3 overexpression reduces the generation of ROS in CMs and mitochondria. A) DHE was used to probe superoxide with nuclear-localized ethidium. Hypo, hypoxia/reoxygenation. B) Increased expression of Cav-3 suppresses superoxide formation, while knockdown of Cav-3 increases superoxide formation; n = 6. C, D) EPR with DEPMPO spin probe on mitochondria from Cav-3-OE hearts have reduced ROS production during NAD-linked, but not FAD-linked, respiration; n = 5–6. E) EM reveals that mitochondria from Cav-3-KO, but not Cav-3-OE, mice have morphological abnormalities. F) ATP synthesis in C. elegans caveolin mutants under state 3 conditions was suppressed; n = 5. G, H) EPR traces (G) and normalized results (H) show increased ROS generation in C. elegans caveolin mutants; n = 6. *P < 0.05.

Mitochondria in cardiac muscle are arranged in filamentous networks; changes in mitochondrial morphology have been noted in apoptosis (37). Fragmentation of mitochondria is greater in Cav-3-KO hearts, but mitochondrial structure is maintained in TGneg and Cav-3-OE hearts (Fig. 4E). Fragmentation of mitochondria may contribute to increased ROS production and cardiac pathology in Cav-3-KO mice.

We assessed C. elegans strains carrying null mutations in caveolin genes as a further means to determine whether caveolin deficiency can result in mitochondrial dysfunction. C. elegans express two isoforms of caveolin (38), but their role in mitochondrial function is not known. We found that under state 3, but not state 4 (data not shown) conditions, C. elegans mutants of Cav-1 and Cav-2 have compromised ATP synthesis (Fig. 4F). The Cav-2 mutant has enhanced generation of carbon-centered free radicals, as assessed by EPR (Fig. 4G, H). These data confirm a role for caveolin expression in mitochondrial function and protection and extend this conclusion to an organism evolutionarily distant from mammals.

Mitochondria-localized caveolin enhances, whereas targeted disruption of mito-CSD reduces, colon cancer cell survival in response to cellular stress

Caveolin has been considered both a tumor promoter and suppressor (39), and thus its role in cancer cell survival and cell death is controversial (40). We identified two colon cancer cell lines, HCT116 and HT29, which express high and low/no levels of caveolin, respectively (Fig. 5A), and used these cells as another system to test whether mitochondria-localized caveolin regulates adaptation to cellular stress. Mitochondria isolated from HCT116 cells are enriched in caveolin (Fig. 5A), a result confirmed by immunohistochemistry (Fig. 5B). We generated stable HT29 cell lines that have targeted expression of RFP or caveolin to mitochondria (mRFP or mCav-1; Fig. 5C–E). Treatment of these stable HT29 cells with TRAIL revealed that cells with mitochondria-targeted caveolin have greater resistance to TRAIL (Fig. 5F), more stable mitochondrial membrane potential (Fig. 5G), and increased biogenesis of mitochondria (Fig. 5H, I). In addition, transient transfection of high-expressing, mitochondria-localized caveolin HCT116 cells with a mito-CSD peptide vector (to compete for caveolin binding partners by disrupting mitochondria-specific caveolin signaling; Fig. 5J) enhances the sensitivity to TRAIL-induced apoptosis (Fig. 5K). Such data suggest that the level of mitochondrial expression of caveolin helps control cellular adaptation to stress.

Figure 5.

Mitochondria-localized and functional caveolin contribute to survival of colon cancer cells. A) Colon cancer cell lines can have high (HCT116) or no (HT29) caveolin expression; purified mitochondria show caveolin enrichment in HCT116 cells. B) Caveolin localize to mitochondria (as observed by costaining with prohibitin) in HCT116 but not HT29 cells. C) Schematic of mitochondria-targeted caveolin in HT29 cells (mCav-1 HT29, RFP served as control). D, E) Mitochondria from mCav-1 HT29 show enrichment of caveolin by immunoblot (D) and colocalization between caveolin and prohibitin (E). F, G) Mitochondria stable HT29 cells were stressed with TRAIL and show reduced apoptosis (F) and more stable mitochondria membrane potential (G); n = 4. H, I) EM reveals increased mitochondrial density in representative (H) and quantified (I) samples of mCav-1 HT29 cells; n = 4. J) A mito-CSD peptide was designed to displace putative mitochondria-localized caveolin binding partners that confer cytoprotection. K) Transient transfection of HCT116 cells with mito-CSD enhances apoptosis in response to TRAIL treatment; n = 4. *P < 0.05.

AAV9.IMM.Cav-3 increases cardiac mitochondrial localization of Cav-3 and Ca²⁺ tolerance of mitochondria, improves mitochondrial respiration, and reduces infarct size

The results thus far show that caveolin is found in mitochondria, predominantly localizing in the IMM and implicating a role for caveolin in stress adaptation by maintenance of mitochondrial function. To test the hypothesis that a cellular stress-induced transfer of caveolin protein from caveolae to mitochondria improves calcium tolerance and respiration and reduces production of ROS, ultimately leading to cell protection, we engineered AAV9 to selectively express Cav-3 in the IMM (AAV9.IMM.Cav-3) by utilizing the mitochondrial targeting sequence for subunit VIII of cytochrome c oxidase. We tested this vector in an in vivo model of I/R injury. AAV9 provides efficient cardiac gene transfer compared to other serotypes (41); indeed, AAV9.IMM.Cav-3 showed cardiac-selective gene transfer (relative to most organs) 14 d after jugular vein injection (Fig. 6A). Cardiac mitochondria isolated from AAV9.IMM.Cav-3-treated mice have increased mitochondrial Cav-3 protein levels, indicating proper expression and targeting (Fig. 6B). Mitochondria from mice treated with AAV9.IMM.Cav-3 withstand higher Ca²⁺ concentrations and show delayed depolarization (Fig. 6C), results consistent with delayed opening of the mPTP and stronger resistance to apoptosis. Mitochondria from those mice had increased oxygen consumption during all respiratory states (Fig. 6D), indicating more efficient function of complexes I, II, and IV.

Figure 6.

Treatment of mice with AAV9.IMM.Cav-3 increases cardiac mitochondrial Cav-3 expression and Ca²⁺ tolerance, improves mitochondrial respiration, and reduces infarct size 14 d after gene transfer. A) AAV9 engineered to express mitochondria-targeted Cav-3 to the IMM (AAV9.IMM.Cav-3) was injected via the jugular vein, and organs were harvested 14 d later to assess viral copy number (via TaqMan assay). The heart showed higher specific gene transfer with AAV9 relative to most other organs, although the liver showed the highest copy number. B) Isolated mitochondria from AAV9.IMM.Cav-3-treated mice have increased Cav-3 expression. C) Mitochondria from AAV9.IMM.Cav-3-treated mice have increased tolerance to calcium pulsing. D) Mitochondrial respiration, assessed using an Oxygraph, was improved in mitochondria from AAV9.IMM.Cav-3-treated mice relative to controls. E) Mice were subjected to 30 min occlusion and 2 h reperfusion with and without gene transfer of AAV9.IMM.Cav-3. Infarct size was reduced by mitochondria-targeted Cav-3; n = 5. *P < 0.05. F) Model showing transfer of caveolin to mitochondria from caveolae and the resultant facilitation of stress adaptation by preservation of mitochondrial function.

Targeting Cav-3 to mitochondria thus improves mitochondrial function. To test whether such mitochondria have improved cardiac protection, we subjected mice to 30 min ischemia and 2 h reperfusion in the absence or presence of AAV9.IMM.Cav-3 gene transfer. We found that mitochondria-targeted Cav-3 reduces infarct size (Fig. 6E). Direct targeting of Cav-3 to mitochondria thus mimics a preconditioning stimulus and induces stress adaptation in the heart.

DISCUSSION

These studies identify physical and functional associations between caveolae/caveolin and mitochondria that help cells adapt to stress via an evolutionarily conserved mechanism (Fig. 6F). Is there a precedent for this observation? Shortly after Palade identified caveolae (1), he published an image of a membrane caveola in close proximity to a mitochondrion (plate 14 in ref. 42). A similar apposition of caveolae and mitochondria is found in studies of multiple cell types, but such studies do not note this anatomic relationship (e.g., Fig. 5 in ref 43; Figs. 8, 18, and 32 in ref. 44; Figs. 5 and 6 in ref. 45; Fig. 12 in ref. 46; Fig. 5 in ref. 47; and Figs. 1 and 2 in ref. 48). The current study thus links to observations in the literature for over half a century that find caveolae and mitochondria can be closely apposed.

Caveolins can localize in mitochondria (46), but the functional role of this localization has not been defined. The current studies identify such a functional role—modulation of mitochondrial function in the adaptation to cellular stress—and lead us to postulate that the plasma membrane senses cellular stress and, in response, interacts with nearby mitochondria to facilitate cellular adaptation to maintain homeostasis. Optimal mitochondrial function may thus depend on interaction with the plasma membrane. Caveolae-mitochondria communication may also provide a unifying molecular explanation for preconditioning, a widely studied phenomenon but whose mechanism is ill-defined; in addition, this communication may be relevant to the function of cancer cells. Overall, our results underscore the importance of mitochondria-localized caveolin as a conserved mechanism for adaptation to cellular stress.

Highly metabolic organs require an ongoing supply of energy and oxygen. The high content of mitochondria in cardiac muscle provides this energy through oxidative phosphorylation and aerobic respiration. Cardiac I/R damages mitochondria, reducing oxidative phosphorylation while promoting release of cytochrome c, production of ROS, and mitochondrial calcium overload (49). The latter two effects contribute to mPTP opening, which induces myocyte death. By preserving the function of mitochondria, one can decrease myocardial injury during I/R (33, 34). Similar effects occur in other cell types, organ systems, and disease/injury states (50–53), but the mechanisms for mitochondrial dysfunction or that protect them from injury are poorly understood. The proximity of caveolae to subsarcolemmal mitochondria may facilitate a linkage between stress sensors in the plasma membrane and protective responses.

Loss of caveolins results in severe defects in various tissues (12, 54, 55), including decreased response to stimuli that protect tissues from cellular stress states (12, 14, 21). By contrast, protective stimuli enhance caveolae formation, and Cav-3-OE mice have a “protected” phenotype (16). Conditions such as ischemia, in which a mismatch occurs between energy production and utilization, are associated with decreased expression of caveolae and/or redistribution of caveolin (56, 57). However, no mechanism has linked caveolin/caveolae expression, oxidative damage, and mitochondria (58, 59). The current results identify the maintenance of mitochondrial function as a means whereby caveolin promotes protection from I/R injury and show that caveolin-mitochondria interaction also influences apoptosis of cancer cells.

Our findings lead to two major questions that require further study: the mechanism for translocation of caveolin to mitochondria; and how caveolin in mitochondria enhance adaptation to cellular stress. Cav-3-OE mice have basal adaptation to stress and also enhanced inhibitory heterotrimeric G protein (G_i) signaling; in addition, interventions that activate G_i signaling (i.e., IPC) yield an increase in caveolae expression (16). Moreover, pertussis toxin, which inhibits G_i, blocks adaptation to stress and decreases the localization of Cav-3 in mitochondria (unpublished results). Perhaps G_i signaling helps facilitate movement of Cav-3 to mitochondria. Once present in mitochondria, caveolin appears to influence components of the electron transport chain (changes oxygen consumption and ROS generation; Figs. 3D, 4, and 6D) and of mitochondrial membranes [alters mitochondrial membrane fluidity (Fig. 3F) and loss of membrane potential in response to stress (Fig. 5G) and increases Ca²⁺ tolerance (Fig. 3B, C)]. Future studies will need to define how caveolin influences mitochondrial components.

The numerous functions of caveolins have generally been thought to derive from their role in forming caveolae, but evidence for the presence of caveolins in cells that lack caveolae contradict this notion (4). The translocation of caveolins from the plasma membrane to mitochondria provides new evidence for localization and function of caveolins outside of the plasma membrane. Perhaps the caveolin content of mitochondria increases with cellular stress when mitochondria are vulnerable to damage. By loading mitochondria with caveolin, one can protect cells (and mitochondria) from injury. Therapies to alter mitochondria-localized caveolin may thus provide novel ways to influence cell death and survival.