Ubiquitously expressed Trpm2 channel limits oxidative stress and preserves mitochondrial function. We first demonstrated that intracellular Ca2+ concentration increase after Trpm2 activation was due to direct Ca2+ influx and not indirectly via reverse Na+/Ca2+ exchange. To elucidate whether Ca2+ entry via Trpm2 is required to maintain cellular bioenergetics, we injected adenovirus expressing green fluorescent protein (GFP), wild-type (WT) Trpm2, and loss-of-function (E960D) Trpm2 mutant into left ventricles of global Trpm2 knockout (gKO) or WT hearts. Five days post-injection, gKO-GFP heart slices had higher reactive oxygen species (ROS) levels but lower oxygen consumption rate (OCR) than WT-GFP heart slices. Trpm2 but not E960D decreased ROS and restored OCR in gKO hearts back to normal levels. In gKO myocytes expressing Trpm2 or its mutants, Trpm2 but not E960D reduced the elevated mitochondrial superoxide (O2.−) levels in gKO myocytes. After hypoxia-reoxygenation (H/R), Trpm2 but not E906D or P1018L (inactivates Trpm2 current) lowered O2.− levels in gKO myocytes and only in the presence of extracellular Ca2+, indicating sustained Ca2+ entry is necessary for Trpm2-mediated preservation of mitochondrial function. After ischemic-reperfusion (I/R), cardiac-specific Trpm2 KO hearts exhibited lower maximal first time derivative of LV pressure rise (+dP/dt) than WT hearts in vivo. After doxorubicin treatment, Trpm2 KO mice had worse survival and lower +dP/dt. We conclude 1) cardiac Trpm2-mediated Ca2+ influx is necessary to maintain mitochondrial function and protect against H/R injury; 2) Ca2+ influx via cardiac Trpm2 confers protection against H/R and I/R injury by reducing mitochondrial oxidants; and 3) Trpm2 confers protection in doxorubicin cardiomyopathy.
- ischemic cardiomyopathy
- doxorubicin cardiomyopathy
- voltage-independent Ca2+ channels
- cardiac Trpm2 currents
- mitochondrial superoxide
in mammalian cells, the transient receptor potential (Trp) protein superfamily is a diverse group of voltage-independent, cation-permeable channels organized into six subfamilies based on amino acid sequence homology (9, 30). Monomeric Trp proteins have six putative transmembrane (TM) domains and intracellular NH2 and COOH termini. To form a functional channel, Trp proteins assemble into either homo- or hetero-tetramers, with the putative pore formed by loops between the fifth and sixth TM domains (21, 23). The Trp-melastatin (Trpm) subfamily consists of eight mammalian members (Trpm1-Trpm8)(30), of which Trpm2 (27), -m4, -m5, and -m6 (49) are expressed in the heart. Although Trpm4 is associated with conduction abnormalities and cardiac arrhythmias (1, 36), there is little information on the physiological and pathophysiological function of Trpm2 in the heart.
Trpm2 is expressed in the sarcolemma and transverse (t) tubules in adult mouse ventricular myocytes (27). In adult cardiac myocytes, Trpm2 is activated by H2O2 and intracellular adenosine diphosphate-ribose (ADPR), inhibited by clotrimazole and flufenamic acid, does not inactivate, and has a conductance for Ca2+ that is ∼50% of Na+ (26, 27). After ischemia/reperfusion (I/R), global Trpm2 knockout (gKO) hearts exhibited reduced contractility when compared with wild-type (WT)-I/R hearts (27). Although results from single cell studies suggest that mitochondrial dysfunction is a result of Trpm2 deficiency in gKO myocytes (26), the use of the whole body knockout mouse reduces the ability to specifically define the role of the cardiac Trpm2 in protecting the heart from I/R injury. In addition, whether Ca2+ influx through activated Trpm2 is required for ameliorating mitochondrial dysfunction was not addressed in previous studies. The present study was undertaken to 1) determine whether the H2O2-induced [Ca2+]i increase in cardiac myocytes is due to Ca2+ entry via activated Trpm2 channels or through reverse Na+/Ca2+ exchange; 2) evaluate whether rescue of mitochondrial dysfunction requires Ca2+ influx through Trpm2 channels by expressing Trpm2, loss-of-function (E960D), or inactivating (P1018L) mutants in gKO myocytes; 3) assess whether cardiac-specific Trpm2 KO hearts suffered worse contractile dysfunction post-I/R compared with WT hearts; and 4) explore whether Trpm2 is cardio-protective in another model of oxidative injury: doxorubicin (Doxo) cardiomyopathy. Our findings reveal that Ca2+ entry via Trpm2 is necessary for proper cardiac function through modulation of mitochondrial oxidative signals, especially after I/R. Trpm2 is also protective of doxorubicin cardiomyopathy.
Generation of global and cardiac-specific Trpm2 KO mice.
Global Trpm2 KO mice were generated as described previously (27). To generate cardiac-specific Trpm2 KO mice, exons 21 and 22 encoding TM domains 5 and 6 and the putative Ca2+ pore of Trpm2 gene (45) were flanked by loxP recombination sites. Trpm2fx/fx mice (C57BL/6) were mated with αMHC-Cre (C57BL/6) mice to generate Trpm2fx/+Cre+/− mice, which were mated with Trpm2fx/fx mice to generate the final desired progeny Trpm2fx/fxCre+/− mice and Trpm2fx/fxCre−/− mice served as WT. αMHC-Cre is expressed in the ventricle and atrium at around 1 to 2 days after birth. Adult mice (8 to 10 wk old) were used in this study. Mice were housed and fed on a 12-h:12-h light/dark cycle at either the Temple University or The Pennsylvania State University Animal Facility supervised by full-time veterinarian staff members. Standard care was provided to all mice used for experiments. All protocols and procedures applied to the mice in this study were approved by the Institutional Animal Care and Use Committees of Temple University and The Pennsylvania State University. For brevity, throughout this report, global Trpm2 KO and cardiac-specific KO are abbreviated as gKO and cKO, respectively, whether applied to mice, hearts, myocytes, or left ventricular (LV) slices.
Construction of Trpm2 and mutants.
V5-tagged human Trpm2 in pAdTrack was used to generate the loss-of-function (E960D), current-inactivating (P1018L), and enhanced Ca2+ permeability (Q981E/P983Y) mutants using Altered Sites II in vitro mutagenesis system (Promega). To verify the function of Trpm2 and its mutants, HEK-293 cells were transfected with control pAdTrack-CMV vector alone (3 μg) or vector (2 μg) + pAdTrack-CMV- Trpm2 (WT or mutant, 1 μg), trypsinized 24 h after transfection using trypsin-EDTA, transferred to 35-mm dishes containing sterile glass coverslips, and incubated for an additional 24 h before measurement of Trpm2 currents with patch-clamp (26). Authenticated Trpm2 and its mutants in pAdTrack were used to construct recombinant, replication-deficient adenovirus (Adv) expressing the exogenous gene as previously described (52).
Isolation, adenoviral infection, and culture of adult murine cardiac myocytes.
Cardiac myocytes were isolated from the septum and LV free wall of WT and gKO mice (8 to 10 wk old) according to the protocol of Zhou et al. (53), and plated on laminin-coated glass coverslips (44). Two hours after isolation, myocytes were infected with Adv-GFP (4.4 × 108 particles/ml), Adv-GFP- Trpm2 (1.6 × 108 particles/ml), or Adv-GFP- Trpm2 mutants (E960D 2.0 × 108; P1018L 4.5 × 108; Q981E/P983Y 3.75 × 108 particles/ml) in 5 ml of fetal bovine serum (FBS)-free Eagle minimal essential medium (MEM) containing 0.2% bovine serum albumin, creatine (5 mM), carnitine (2 mM), taurine (5 mM), NaHCO3 (4.2 mM), penicillin (30 mg/l), gentamicin (4 mg/l), insulin-transferrin-selenium supplement, and 2,3-butanedione monoxime (10 mM) for 3 h. An additional 5 ml of MEM (with same supplements) was then added, and myocytes were cultured for 24 h before measurements. We have previously demonstrated that under our culture conditions, adult mouse myocytes cultured for up to 48 h maintained rod-shape morphology, t-tubule organization, and normal contractile function (39).
Immunolocalization of Trpm2 and its mutants in cardiac myocytes.
Adv-infected and cultured gKO myocytes were washed three times with PBS containing 2 mM EGTA (PBS-EGTA). Myocytes were fixed for 30 min in 4% paraformaldehyde in PBS-EGTA. After two rinses with PBS-EGTA, myocytes were permeabilized for 2 min with 0.05% Triton X-100. Myocytes were rinsed two times with PBS-EGTA and once with Blotto (5% nonfat dry milk, 0.1 M NaCl, and 50 mM Tris·HCl; pH 7.4). Primary rabbit polyclonal anti-V5 antibody (1:500; Sigma) diluted in Blotto was added to the cells, incubated at room temperature in the dark for 60 min, and rinsed three times with Blotto. Secondary antibodies (Alexa Fluor 594 goat anti-rabbit; 1:250) diluted in Blotto were added to the cells, incubated in the dark for 30 min, and followed by three PBS-EGTA rinses. Coverslips were mounted to slides with Prolong gold antifade mounting solution (Invitrogen). Confocal images (×63 oil objective, 510 Meta; Carl Zeiss) were acquired at 594 nm excitation and 617 nm for emission.
The protocol of Despa et al. (12) was followed with few modifications. Briefly, myocytes were exposed to 10 μM sodium-binding benzofuran isophthalate (SBFI) AM in the presence of Pluronic F127 (0.05% wt/vol) for 2 h at 37°C. Media was changed and myocytes were incubated for a further 30 min to allow de-esterification of the AM ester. Measurements of [Na+]i were performed in myocytes incubated in Tyrode's solution containing (in mM) 140 NaCl, 4 KCl, 1 CaCl2, 10 glucose, and 5 HEPES, pH 7.4. Dual excitation wavelengths (340 ± 15 and 380 ± 15 nm; alternating at 5 Hz) were directed to a single myocyte via a Nikon 40×/1.30 NA UV oil objective situated in a Nikon TE200U inverted microscope. Emission from a small area of the myocyte was recorded at 510 ± 40 nm (Ionoptix, Milton, MA). All measurements were performed at 37°C. To minimize SBFI photobleaching and myocyte photodamage, emission data were collected at 60-s intervals. Under our measurement conditions, the average autofluorescence from 10 myocytes collected at both excitation wavelengths was <10% of SBFI signals and was ignored.
In another series of experiments, myocytes loaded with SBFI were exposed to different extracellular Na+ concentrations ([Na+]o: 0, 10, 20, and 40 mM) in the presence of gramicidin D (10 μM) and strophanthidin (100 μM). Myocytes were exposed to each [Na+]o for ≥10 min to allow equilibration before SBFI signals were measured. The composition of Na+ solutions were identical to those detailed by Despa et al. (12), with the exception that 2,3-butanedione monoxime (BDM; 10 mM) was included to minimize myocyte hypercontracture.
Measurement of Trpm2 current in HEK293 cells and cardiac myocytes.
Trpm2 currents (ITrpm2) were measured at 30°C in transfected HEK293 cells and Adv-infected LV myocytes (cultured for 24 h) with whole cell patch-clamp (26, 27, 39, 44, 47). Fire-polished pipettes (tip diameter 2 to 3 μm for HEK293 cells and 4–6 μm for myocytes) were used. Pipette solution contained (in mM) 110 CsCl, 20 TEA·Cl, 10 HEPES, 10 EGTA, and 5 MgATP, pH 7.2; and bathing solution contained (in mM) 127 NaCl, 5.4 CsCl, 2 CaCl2, 1.3 MgSO4, 4 4-aminopyridine, 10 HEPES, 10 Na-HEPES, 15 glucose, and 0.001 verapamil, pH 7.4. Our solutions are designed to minimize L-type Ca2+ current, Na+-K+-ATPase current, Na+/Ca2+ exchange current, and potassium currents. After break-in, cells were subjected to voltage ramp (+100 to −100 mV; 500 mV/s). In some experiments, ADPR (300 μM) was included in pipette solutions to activate Trpm2 channels (22, 26). In other experiments, after full activation of Trpm2 channels by ADPR, flufenamic acid (0.5 mM) was added to the extracellular medium to inhibit ITrpm2 (19, 26).
To detect ITrpm2 inactivation (P1018L) and to estimate relative Ca2+ to Na+ conductance (GCa/GNa) (Q981E/P983Y), cultured gKO myocytes expressing Trpm2 or its mutants were voltage-clamped at −80 mV. Pipette solution was identical to that used above and contained ADPR. Extracellular solution contained (in mM) 140 NaCl, 10 HEPES, and 15 glucose, pH 7.4. After break-in, inward Na+ currents were obtained in gKO myocytes expressing Trpm2, P1018L, or Q981E/P983Y at −80, −90, and −100 mV (Fig. 2). Extracellular solution was then changed to one containing (in mM) 110 CaCl2, 10 HEPES, and 15 glucose, pH 7.4. Inward Ca2+ currents were obtained at −80, −90, and −100 mV from the same myocyte (Fig. 2). GCa/GNa is calculated from the ratio of the slopes of I–V plots with Ca2+ or Na+ as the permeant ion.
To estimate percent Trpm2 knockdown in cKO myocytes independent of qPCR method, freshly isolated WT and cKO were subjected to patch-clamp studies, using similar solutions and voltage ramp protocol for transfected HEK293 cells. ADPR (300 μM) was present in pipette solutions to activate Trpm2 channels in WT and cKO myocytes (if present). The percentage of cKO myocytes whose ITrpm2 exceeded means + 2 SD of ITrpm2 measured in gKO myocytes was determined.
Confocal mitochondrial superoxide measurements.
Adult WT and gKO myocytes expressing GFP, Trpm2, or its mutants were loaded with the mitochondrial superoxide (O2.−) sensitive fluorophore MitoSOX Red (Invitrogen; 22 μM) in extracellular media (ECM) containing 2% BSA, 0.06% pluronic acid, and 20 μM sulfinpyrazone at 37°C for 30 min. Cells were then washed, resuspended in ECM containing 0.25% BSA, and imaged using a Carl Zeiss Meta 510 confocal microscope with a 40× oil objective with 1.7× digital zoom at 561 nm for MitoSOX Red (26, 29). In some experiments, gKO myocytes expressing GFP, Trpm2, or its mutants were subjected to 30 min of hypoxia (1% O2-5% CO2) followed by 30 min of reoxygenation (21% O2-5%CO2) before O2.− measurements.
In vivo adenovirus-mediated gene transfer.
After the skin was cleansed with Betadine solution, the left chest of an anesthetized (2% inhaled isoflurane) WT or gKO mouse was opened, the heart was exteriorized, and 30 μl (total volume) of Adv-GFP (4.4 × 1011 particles/ml), Adv- Trpm2 (4.0 × 1010 particles/ml), or Adv-E960D (5.0 × 1010 particles/ml) was directly injected into the anterior and posterior LV wall and apex (47). The heart was returned to the chest cavity and the wound sutured. The entire surgical procedure took <45 s. Typically >95% of animals survived the procedure. Survivors were allowed to recover for 5 days before experiments.
Measurement of ROS and oxygen consumption rate in heart slices and isolated myocytes.
Regions of hearts successfully infected with adenovirus expressing GFP ± Trpm2 or its mutants were detected by GFP fluorescence (47). LV cross-sections (2 mm) were obtained from GFP-positive regions, equilibrated in Krebs Henseleit bicarbonate (KHB) buffer (30 min; 37°C), stained with dihydroethidium (DHE; 30 μM), and incubated in the dark (30 min) with gentle rotation (8). DHE-stained LV sections were imaged with a Carl Zeiss 710 multi-photon confocal microscope (10× objective; 561 nm ex). ZEN software was used to collect and analyze confocal images of each section and generate 2.5-dimensional heatmap plots of mean DHE intensity (26). To assess the oxygen consumption rate (OCR) in hearts, heart slices were generated from GFP-positive regions and suspended in XF media, and OCR was measured at 37°C in an XF96 extracellular flux analyzer (Seahorse Bioscience) (26).
In another series of experiments, LV myocytes were isolated from regions of hearts successfully infected with adenovirus expressing GFP ± Trpm2 or its mutants. Freshly isolated myocytes were initially placed in Ham's F-10 supplemented with penicillin-streptomycin (100 units/ml) and 5% FBS at 37°C in 5% CO2-humidified atmosphere for 1 h. The media was changed to XF Assay Medium supplemented with 25 mM glucose and 1 mM sodium pyruvate, incubated at 37°C in CO2-free chamber for 1 h, and followed by the XF flux assay for OCR measurement.
I/R surgery in mice.
I/R surgery was performed as previously described (14, 26, 27). Briefly, WT and cKO mice (8 to 10 wk) were anesthetized with 2% isoflurane, and the heart was exposed through a left thoracotomy at the 5th intercostal space. The slipknot was tied around the left anterior descending (LAD) coronary artery 2 to 3 mm from its origin, and the heart was immediately returned to the chest cavity followed by evacuation of pneumothorax and closure of muscle and skin layers. The slipknot was released after 30 min of ischemia to allow reperfusion. Sham-operated animals were subjected to the same surgical procedure except that the slipknot was not tied. Animals recovered from anesthesia within 5 min after the completion of surgery and received ibuprofen (10 mg/50 ml drinking water) for 48 h as post-surgery analgesia. Studies on survivors were performed on day 3 post-surgery.
In vivo hemodynamics analyses of cardiac function.
For in vivo hemodynamic measurements, a 1.4-Fr micromanipulator-tipped catheter (SPR-671; Millar Instruments) was inserted into the right carotid artery and advanced into the LV of lightly anesthetized (tribromoethanol/amylene hydrate; Avertin; 2.5% wt/vol, 8 μl/g ip) mice with spontaneous respirations and placed on a heated (37°C) pad. Hemodynamics including heart rate, LV end-diastolic (LVED) pressure, and maximal first time derivative of LV pressure rise (+dP/dt) and fall (−dP/dt) were recorded in closed-chest mode, both at baseline and in response to increasing doses of isoproterenol (Iso; 0.1, 0.5, 1, 5, and 10 ng) (27, 47).
Infarct size measurement.
The myocardium was stained with 2% 2,3,5-triphenyltetrazolium (TTC) to measure infarct size as previously described (14, 27). Briefly, 72 h after I/R, the slipknot around the LAD was retied followed by injection of 2% Evans blue dye (0.2 ml). Hearts were excised, and the LV was sliced into five equally thick sections perpendicular to the short axis of the heart and incubated in PBS containing TTC. After 15 min at room temperature, slices were digitally photographed. The Evans blue-stained area (area not at risk), TTC-negative area (infarcted myocardium), and area at risk (AAR; including both TTC-negative and -positive areas) were measured with computer-based image analyzer SigmaScan Pro 5.0 (SPSS Science, Chicago, IL). The AAR was expressed as percentage of total LV, whereas the infarcted myocardium was expressed as a percentage of the AAR.
Doxorubicin cardiomyopathy model.
WT and gKO mice received Doxo (3 mg/kg ip q wk) or saline injections, starting at 8 wk and continued to 16 wk. In one cohort of animals, survival was followed for 18 wk. In another cohort of mice, echocardiography (38, 48) and in vivo hemodynamics were evaluated at 7 and 8 wk, respectively, after initiation of Doxo injection.
Heart homogenates from WT and gKO mice (both basal and 72h post-I/R) were prepared as previously described (44). Proteins were separated by SDS-PAGE (10, 12 or 15%) followed by transfer to Hybond-C Extra nitrocellulose (Amersham). Blots were blocked for 1 h with 5% milk and probed overnight with anti-NADH dehydrogenase (ubiquinone) 1α subcomplex 4-like 2 (NDUFA4L2) (1:1,000; Abcam), anti-Bcl2/adenovirus E1B 19 KDa-interacting protein 3 (BNIP3)(1:750; Abcam), anti-calsequestrin (1:15,000; Fitzgerald), anti-superoxide dismutase 1 (SOD1)(1:3,000; Calbiochem), anti-SOD2 (1:400,00; Abcam), or thioredoxin2 (Trx2)(1:10,000; Abcam) antibodies. Blots were washed and incubated with the appropriate secondary antibody conjugated to horseradish peroxidase. Enhanced chemiluminescence (Amersham) was used for the detection of signals. Intensity of the bands was quantitated with densitometry and normalized to that of calsequestrin (loading control).
All results are expressed as means ± SE. For analysis of ITrpm2 as a function of group and voltage, +dP/dt as a function of group and isoproterenol, two-way ANOVA was used. For analysis of [Na+]i, ROS levels, fractional shortening, protein abundance, and O2 consumption, one-way ANOVA was used. A commercially available software package (JMP version 7; SAS Institute, Cary, NC) was used. In all analyses, P < 0.05 was taken to be statistically significant.
H2O2 exposure did not increase [Na+]i in adult cardiac myocytes.
Trpm2 is permeable to both Ca2+ and Na+ (26). H2O2-induced [Ca2+]i increase in WT myocytes (27) could either be from direct Ca2+ entry or secondarily from reverse Na+/Ca2+ exchange (3 Na+ out: 1 Ca2+ in) due to elevated [Na+]i. Treatment of WT and gKO myocytes with H2O2 did not change the F340/F380 signal of the ratiometric Na+-sensitive dye SBFI (Fig. 1), indicating no detectable changes in [Na+]i with H2O2. This result indicates that the H2O2-induced increase in [Ca2+]i was due to direct Ca2+ influx via activated Trpm2 channels.
Functional characterization of electrophysiological phenotype of Trpm2 and its mutants.
To manipulate Ca2+ influx via activated Trpm2, we engineered loss-of-function (E960D), current-inactivating (P1018L), and enhanced Ca2+ permeability (Q981E/P983Y) mutants from V5-tagged WT human Trpm2. When transfected into HEK293 cells, Trpm2 but not E960D displayed the characteristic linear I–V relationship with reversal potential close to 0 mV when the channel was activated by intracellular ADPR (300 μM)(Fig. 2). In addition, the ADPR-elicited current was abolished by flufenamic acid (0.5 mM)(data not shown), indicating it was ITrpm2. P1018L showed I–V curve indistinguishable from that of Trpm2 (data not shown). Expressed in HEK293 cells, the relative Ca2+ to Na+ conductance (GCa/GNa) was 0.65 ± 0.08 (n = 5) for Trpm2 and 1.12 ± 0.17 (n = 5)(P < 0.035) for Q981E/P983Y, respectively.
We next measured ITrpm2 in gKO myocytes expressing Trpm2, P1018L, E960D, or Q981E/P983Y by adenovirus-mediated gene transfer (>90% infection efficiency; only myocytes that exhibited GFP fluorescence were analyzed)(52). In agreement with previous findings (26), ITrpm2 due to Trpm2 expression did not inactivate with time (Fig. 2). By contrast, P1018L ITrpm2 inactivated with time, whereas E960D displayed no appreciable current elicited by intracellular ADPR (Fig. 2). GCa/GNa in gKO myocytes expressing Trpm2 was 0.56 ± 0.02 (n = 4)(Fig. 2) and 1.06 ± 0.03 (n = 3)(P < 0.0001) in myocytes expressing Q981E/P983Y.
Ca2+ influx via Trpm2 was required to reduce elevated mitochondrial O2.− in gKO myocytes.
To evaluate whether lack of Trpm2-mediated Ca2+ influx was the culprit in mitochondrial dysfunction in KO myocytes, we expressed WT Trpm2, loss-of-function E960D, inactivating P1018L, or enhanced Ca2+ permeability Q981E/P983Y in gKO myocytes by adenovirus-mediated gene transfer. As expected, immunolocalization studies with anti-V5 antibodies indicated that Trpm2 and its mutants were targeted to sarcolemma and t-tubules in Adv-infected gKO myocytes (Fig. 3), similar to the distribution of endogenous Trpm2 in WT myocytes (27). By contrast, GFP fluorescence was distributed in the cytoplasm (Fig. 3). When compared with WT-GFP myocytes (Fig. 4A), gKO-GFP myocytes (Fig. 4B) had ∼5× mitochondrial O2.− levels (Fig. 4G), similar to previous observations in freshly isolated WT and gKO myocytes (26). Elevated mitochondrial O2.− in gKO-GFP myocytes (Fig. 4B) were substantially reduced by expression of Trpm2 (Fig. 4F) but not by the loss-of-function E960D (Fig. 4C). Overexpression of Trpm2 in WT myocytes had no effects on mitochondrial O2.− (Fig. 4E). Remarkably, the inactivating P1018L partially reduced mitochondrial O2.− (Fig. 4D). Collectively, these data suggest that Ca2+ entry via Trpm2 was necessary to maintain mitochondrial function, as reflected by reduced mitochondrial O2.− production by WT but not E960D mutant in gKO myocytes (Fig. 4G).
Ca2+ influx via Trpm2 was required to reduce cardiac oxidants and enhance bioenergetics in gKO hearts in vivo and in isolated gKO myocytes in vitro.
To evaluate if the beneficial effects of Trpm2 on cultured gKO myocytes could be translated in vivo, we expressed Trpm2 or E960D in intact WT or gKO hearts by adenovirus-mediated gene transfer. Five days after Adv injection, significant portions of the LV exhibited GFP fluorescence, indicating successful transfer and expression of the exogenous gene. Heart slices (50 mg tissue/well, sliced to ∼150 μm) that fluoresced green (Fig. 5A) were isolated, and in situ ROS and OCR were measured. Similar to our previous observations on WT and gKO hearts (26), gKO-GFP had much higher ROS levels (Fig. 5B) and reduced OCR (Fig. 5C) than WT-GFP heart slices. Trpm2 but not the loss-of-function E960D mutant rescued mitochondrial dysfunction in gKO hearts, as evidenced by reduced ROS (Fig. 5B) and increased OCR (Fig. 5C) back to WT levels. Intriguingly, when compared with that of WT-GFP hearts, ROS levels were modestly (P < 0.05) reduced in WT heart slices overexpressing Trpm2 (Fig. 5B). In addition, the pattern of OCR differences observed in isolated gKO myocytes expressing GFP, Trpm2, or E960D mutant (Fig. 5D) was in excellent agreement with that obtained from gKO heart slices expressing identical constructs (Fig. 5C). The advantage of using LV slices is that both OCR and ROS could be measured in situ.
Effects of Trpm2 knockout on expression of selected proteins associated with oxidant defense and mitochondria.
Our results demonstrated that absence of Trpm2 in cardiac myocytes is associated with increased O2.− and decreased mitochondrial respiration; we therefore evaluated expression of proteins associated with mitochondrial and cellular oxidant defense. Under basal conditions, there were no differences in SOD1 expression between WT and gKO hearts (Fig. 6A). By contrast, SOD2, Trx2 and BNIP3 expression were significantly decreased in gKO hearts. Expression of NDUFA4L2 appeared to be lower in gKO hearts, but the difference did not reach statistical significance.
Post-I/R, expression of SOD1, SOD2, NDUFA4L2, and BNIP3 was significantly reduced in gKO when compared with WT hearts (Fig. 6B). By contrast, Trx2 expression was significantly increased in gKO hearts (Fig. 6B).
Effects of I/R on infarct size and cardiac performance in WT and cardiac-specific Trpm2 KO mice.
Global Trpm2 KO mice have significant lower +dP/dt 3 days post-I/R when compared with WT mice, despite similar infarct sizes and areas-at-risk (27). To ascertain whether the deleterious effects observed are due to absence of Trpm2 in the cardiac myocyte or other tissues such as phagocytes, we generated a cardiac-specific Trpm2 KO mouse. At 2 mo of age, 64.0 ± 9.7% of floxed Trpm2 gene was deleted in cardiac-specific KO hearts as evaluated by qPCR (5 cardiac-specific KO; 8 WT). This likely represents the lower bound of Trpm2 deletion in cardiac myocytes since the whole heart contains fibroblasts, vascular endothelial cells, and blood cells. In contrast, Trpm2 expression in cardiac-specific KO mice was not affected in bone marrow (P < 0.88), brain (P < 0.94), and liver when compared with WT mice. Because preliminary studies indicate that our anti-Trpm2 antibodies gave very weak signals for endogenous Trpm2 on Western blots, we used electrophysiological measurements to estimate the percentage of myocytes in which Trpm2 was eliminated by the Cre recombinase system. A similar approach was used to estimate the percentage of myocytes in which Na+/Ca2+ exchange current was absent in the cardiac-specific Na+/Ca2+ exchanger knockout mice (32). Electrophysiological studies (Fig. 7) indicate that cKO myocytes exhibited much lower ITrpm2 than WT myocytes when exposed to intracellular ADPR. When stimulated with ADPR, the means ± SD of ITrpm2 (at +100 mV) for 6 WT, 7 gKO, and 19 cKO myocytes were 4.51 ± 0.83, 0.62 ± 0.44, and 1.06 ± 0.91 pA/pF, respectively. The range of ITrpm2 (at +100 mV) for 19 cKO myocytes was 0.16 to 3.24 pA/pF. With the use of the cut-off of 1.5 pA/pF (means + 2 SD for ITrpm2 in gKO myocytes), there were 4 out of 19 cKO myocytes whose ITrpm2 exceeded 1.5 pA/pF, giving an estimated knock-out rate of 79% for cKO myocytes. This compares very well with the 64.0 ± 9.7% knock-out rate measured with qPCR.
In agreement with observations in gKO-I/R hearts (27), cKO-I/R hearts exhibited significantly lower +dP/dtmax when compared with WT-I/R hearts (Fig. 7), despite similar areas-at-risk and infarct sizes (Fig. 7). In addition, baseline systolic pressure, maximal systolic pressure, and baseline and maximal −dP/dt were lowest in cKO-I/R hearts (Table 1). These observations strongly indicate that the beneficial effects were directly mediated by cardiac Trpm2 channels, rather than by Trpm2 inhibition of NOX-mediated ROS production in WT phagocytes (13), thereby reducing the inflammatory response and subsequent cardiac injury.
Ca2+ influx via Trpm2 was required for reduction of mitochondrial O2.− levels post-H/R.
To establish whether Ca2+ influx through Trpm2 is required for its protection against I/R injury, and as an independent approach to confirm the beneficial effects of cardiac Trpm2 on myocardial function post-I/R, we expressed GFP, Trpm2, E960D, P1018L, or Q981E/P983Y in gKO myocytes and then subjected myocytes to H/R. Mitochondrial O2.− levels were measured as an indicator of mitochondrial malfunction and by inference, cell injury. After 30 min of hypoxia followed by 30 min of reoxygenation, gKO-GFP myocytes exhibited high mitochondrial O2.− levels (Fig. 8A). Trpm2 (Fig. 8B) or Q981E/P983Y (Fig. 8E) but not E960D (Fig. 8C) or P1018L (Fig. 8D) significantly lowered mitochondrial O2.− levels in gKO myocytes post-H/R (Fig. 8F), indicating sustained Ca2+ influx through non-inactivating Trpm2 channels was required for protection against H/R injury. As a parallel approach to evaluate the absolute requirement for Ca2+ entry, we conducted the H/R experiments in the absence of extracellular Ca2+ (no added Ca2+ + 4 mM EGTA). Absence of extracellular Ca2+ abolished the beneficial effects of Trpm2 on myocyte H/R injury (Fig. 8F), further supporting our hypothesis that Ca2+ entry via activated Trpm2 channels is essential for cell survival during cardiac I/R injury.
Trpm2 improved survival and limited doxorubicin-induced cardiac dysfunction.
To extend the potential translational implications of Trpm2 protection of the heart, we used a second model of cardiac injury: doxorubicin-induced cardiomyopathy (16), a model that involves oxygen free radicals (37) via Doxo's effects on mitochondrial electron transport chain (7, 15). Survival of gKO mice treated with Doxo was much worse than WT mice treated similarly (Fig. 9). By contrast, both WT and gKO mice injected with saline had no mortality. At 7 wk after initiation of Doxo or saline injection, both WT-Doxo and gKO-Doxo mice exhibited significantly (P < 0.04) depressed fractional shortening (FS) compared with their respective littermates treated with saline (Fig. 9), as expected. Importantly, gKO-Doxo hearts had significantly (P < 0.02) lower FS compared with WT-Doxo hearts. At 8 wk after initiation of Doxo or saline injection, we measured in vivo hemodynamics in closed-chest anesthetized mice injected with escalating doses of isoproterenol (38, 47). When compared with WT and gKO mice injected with saline, +dP/dt was lower (P < 0.0001) in both WT and gKO mice treated with Doxo (Fig. 9). In agreement with echocardiographic results, when compared with WT-Doxo hearts, gKO-Doxo hearts had significantly lower (P < 0.03) +dP/dt across the range of Iso examined. Our results suggest that in addition to salutatory effects post-I/R, Trpm2 channels exerted protective effects in Doxo-induced cardiomyopathy.
Trp channels are involved in many fundamental cell functions and are associated with many disease states (30). Trpm channels are a subgroup of Trp channel superfamily. Trpm2 is expressed in many tissues including heart, vasculature, hematopoietic cells, and brain (18, 28). Trpm2 is activated by ADP-ribose (ADPR) and H2O2 and mediates Ca2+ influx into the cell (41). Trpm2 has an essential role in susceptibility to oxidative stress (17, 28, 34). The conventional wisdom regarding the role of Trpm2 in disease states is that activation of Trpm2 induces cell death by sustained increases in [Ca2+]i (17, 41) or mediates enhanced chemokine production in hematopoietic cells thereby aggravating inflammatory response and tissue injury (42). However, recent reports suggest a novel paradigm that Ca2+ entry via Trpm2 channels is protective, rather than deleterious, in pathophysiological conditions. For example, in WT mice subjected to intra-peritoneal injection of endotoxin, survival is ∼5× higher than Trpm2 KO mice. This is due to Ca2+ entry via Trpm2 channels, thereby depolarizing plasma membrane and resulting in decreased NOX-mediated ROS production in WT phagocytes (13). Similarly, in pyramidal neurons subjected to oxidant injury, inhibition of Trpm2 results in enhanced cellular damage (2), confirming that Trpm2 reduces oxidative stress.
We have previously demonstrated that exposure to H2O2 (200 μM) resulted in significantly higher [Ca2+]i increase in WT when compared with gKO myocytes (27). The H2O2-induced [Ca2+]i increase was dependent on extracellular Ca2+ and inhibited by clotrimazole (50 μM), suggesting mediation by activated Trpm2. Because cardiac Trpm2 is permeable to both Na+ and Ca2+ (26), the H2O2-induced [Ca2+]i increase may be due to direct Ca2+ entry through activated Trpm2, Ca2+ influx through reverse Na+/Ca2+ exchange, or a combination of both pathways. Thus the first major finding is that in freshly isolated adult mouse LV myocytes, H2O2 treatment did not result in appreciable [Na+]i increase (Fig. 1), suggesting that H2O2-induced [Ca2+]i increase was largely due to direct Ca2+ entry through activated Trpm2. Lack of [Na+]i increase may be due to robust Na+-K+-ATPase activity, leading to rapid removal of Na+ that has entered via Trpm2 channels. By contrast, in cultured neonatal rat ventricular myocytes, H2O2 treatment resulted in increases in both [Ca2+]i and [Na+]i (50). An explanation for the discrepancy may relate to the fact that neonatal rat ventricular myocytes express α1- and α3-isoforms (24), whereas adult mouse cardiac myocytes express α1- and α2-isoforms of Na+-K+-ATPase (4). When expressed in Sf-9 insect cells, the α3-isoform exhibits greater than twofold less affinity for Na+ when compared with α2-isoform (5) and may account for lower Na+ transport rate. In addition, adenovirus-mediated overexpression of α2-isoform in adult rat cardiac myocytes leads to greater affinity for Na+ and higher Na+-K+-ATPase activity compared with control myocytes expressing β-galactosidase (10). Finally, neonatal rat ventricular myocytes do not possess the highly organized t-tubules in which α2-isoform is preferentially localized in adult mouse cardiac myocytes (4). These differences in physiology may account for the observations in this study compared with a previous report (50).
Loss of Trpm2 in cardiac myocytes resulted in mitochondrial dysfunction as indicated by lower Δψm, reduced mitochondrial Ca2+ uptake due to both lower driving force and mitochondrial Ca2+ uniporter (MCU) activity, impaired oxygen consumption resulting in lower ATP levels, and elevated cardiac ROS and mitochondrial O2.− levels when compared with WT myocytes (26). The second major finding is that mitochondrial dysfunction in gKO myocytes was rescued by Trpm2 but not by the loss-of-function E960D (Fig. 4), indicating that Ca2+ entry via Trpm2 was necessary to maintain basal mitochondrial bioenergetics. Interestingly, the inactivating P1018L (Fig. 2) afforded partial rescue (Fig. 4), suggesting the amount of Ca2+ influx via Trpm2 was important in ameliorating mitochondrial dysfunction in gKO myocytes. In intact gKO hearts, oxygen consumption rate was improved and cardiac ROS levels were reduced by expressing Trpm2 but not E960D (Fig. 5). The observations on intact gKO hearts (Fig. 5) thus corroborate those on isolated gKO cardiac myocytes (Fig. 4). The requirement for Ca2+ entry via Trpm2 to improve mitochondrial function in gKO myocytes is in agreement with the recent demonstration that constitutive, low level mitochondrial Ca2+ uptake is essential in maintaining cellular bioenergetics (6). Trpm2 may provide the necessary Ca2+ for mitochondrial bioenergetics in the heart. In this context, it is important to note that low levels of H2O2 emission, which occurs in normal cardiac mitochondria (40), may tonically activate Trpm2 channels, thereby realizing a virtuous cycle of bioenergetics maintenance in WT but not KO myocytes (Fig. 10A). In addition, because adenovirus expressing Trpm2 or the E960D was directly injected in gKO hearts, the beneficial effects were due to direct actions of Trpm2 on gKO cardiac muscle (cardiac-specific), as opposed to indirect effects (e.g., dampening NOX-mediated ROS production by Trpm2 in WT but not KO phagocytes) (13).
In global Trpm2 KO hearts, ROS levels (as detected by dihydroethidium DHE) are elevated when compared with that of WT hearts (26). In addition, O2.− (as detected by MitoSox Red; Fig. 4) but not H2O2 levels [as detected by 5-(and 6-)chloromethyl-2′,7′-dichlorofluorescein; DCF] (27) were elevated in gKO when compared with those WT myocytes. Expression of proteins associated with oxidant defense such as SOD2 and Trx2 was lower in gKO than in WT hearts (Fig. 6A). In addition, expression of BNIP3 and perhaps NDUFA4L2 was lower in gKO myocytes (Fig. 6A). NDUFA4L2 expression has been shown to reduce ROS production, although the exact mechanism is not known (43), and BNIP3 is necessary for mitochondrial autophagy, which also limits ROS production (35, 51). These observations suggest that elevated ROS levels in gKO hearts were due to both increased production and decreased scavenging.
The picture is more complex post-I/R (or H/R). ROS levels are highest in gKO-I/R than WT-I/R, gKO-sham, or WT-sham hearts (26). Protein levels of SOD1, SOD2, NDUFA4L2, and BNIP3 were reduced in gKO-I/R (Fig. 6B), whereas Trx2 (Fig. 6B) and NOX4 expression (27) was elevated when compared with that of WT-I/R hearts. When compared with those of WT-H/R myocytes, O2.− levels were significantly elevated in gKO-H/R myocytes (Fig. 8). Interestingly, despite increases in Trx2 expression (Fig. 6B), H2O2 levels were still much higher in gKO-H/R than in WT-H/R myocytes (27), suggesting that the tremendous increase in O2.− generation may overwhelm defenses by myocyte Trx2. These observations suggest that the large increase in ROS levels in gKO-I/R (or H/R) myocytes is due to increased mitochondrial O2.− and H2O2 production, increased ROS generation by NOX4, and decreased O2.− and H2O2 scavenging.
To ascertain whether the protective effects by Trpm2 were exerted directly on the heart, we generated a cardiac-specific Trpm2 KO mouse in which 64% to 79% of TRPM2 gene was deleted in the heart but not in bone marrow, brain, or liver. Thus, the third major finding is that post-I/R, despite similar infarct sizes and areas-at-risk, cardiac-specific Trpm2 KO hearts had significantly lower myocardial performance compared with that of WT hearts (Fig. 7 and Table 1). Consistent with previous observations on gKO-I/R hearts (27), this unequivocally demonstrated that the beneficial effects were due to direct actions of Trpm2 in the heart. In addition, with the use of elevated mitochondrial O2.− levels as a biomarker for myocyte H/R injury, the fourth major finding is that Ca2+ entry via Trpm2 channels was essential for the protection afforded by Trpm2 during I/R or H/R. This is because 1) unlike WT Trpm2, the loss-of-function E960D (Fig. 2) did not reduce mitochondrial O2.− levels in gKO myocytes post-H/R (Fig. 8) and 2) protection by Trpm2 against H/R injury in gKO myocytes was lost when extracellular Ca2+ was removed (Fig. 8).
The precise mechanism by which Trpm2, a sarcolemmal ion channel, affects cardiac mitochondrial function and decreases ROS levels is unknown and warrants additional studies. Under basal conditions, tonic activation of Trpm2 by low levels of H2O2 emitted by normal cardiac mitochondria (40) may provide the necessary Ca2+ for bioenergetics maintenance (6) in WT but not KO myocytes (Fig. 10A). This is consistent with our finding of lower expression of BNIP3 and perhaps NDUFA4L2 (Fig. 6A), higher ROS levels (Fig. 5B), and reduced OCR (Fig. 5, C and D) in Trpm2 KO myocytes. Post-I/R, when compared with gKO-I/R hearts, WT-I/R hearts had higher SOD1, SOD2, BNIP3, and NDUFA4L2 (Fig. 6B) but lower NOX4 levels (27). We postulate that Ca2+ entry via activated Trpm2 channels stimulates calcineurin, which dephosphorylates receptor for activated C kinase 1 (RACK1) and blocks RACK1 dimerization (Fig. 10B). This leads to increased HIF-1α levels by inhibiting its ubiquitination and degradation. Consistent with this view is that HIF-1α levels are higher in WT-I/R than gKO-I/R hearts (27). HIF-1α is a transcription factor that regulates a large number of genes including those involved in energy and redox homeostasis (forhead box transcription factor 3a or FoxO3a and SOD2), electron transport chain (NDUFA4L2: Complex I), and mitochondrial autophagy (BNIP3). SOD2, BNIP3, and physiological Complex I activity may act in concert to reduce mitochondrial ROS levels.
A recent study using an independent global Trpm2 KO mouse (C57BL/6 background) reported very different results (20). Specifically, after 45 min of ischemia followed by 24 h of reperfusion in vivo, neutrophil infiltration was less, infarct size was smaller, and +dP/dt was higher in KO than in WT hearts. The authors speculated that increased neutrophil adhesion to endothelial cells mediated by Trpm2 channels caused increased damage post-I/R. The proposed mechanism, however, is not compatible with the results from a recent study that Ca2+ entry via activated Trpm2 channels decreases NOX-mediated ROS production in phagocytes (13), thereby lessening the inflammatory response. The proposed mechanism would also have predicted similar post-I/R myocardial dysfunction in WT and cardiac-specific Trpm2 KO mice since Trpm2 channels are present in bone marrow cells. This prediction is not supported by our observations on cardiac-specific Trpm2 KO mice post-I/R (Fig. 7). The reasons for the different results between our previous (26, 27) and current studies and those of Hiroi et al. (20) may include 1) Trpm2 was knocked out by targeting different exons; 2) 45 vs. 30 min of ischemia resulted in much larger infarcts (45% vs. 27% AAR); 3) cardiac function was examined at 24 h vs. 72 h of reperfusion; 4) different anesthesia (pentobarbital vs. isoflurane); 5) different surgical techniques used in measuring in vivo hemodynamics (opening the chest followed by LV puncture vs. catheterizing the right carotid artery in a closed-chest preparation); and 6) increased heat dissipation in open-chest mice compared with closed-chest mice. Our current results with cardiac-specific Trpm2 KO hearts, together with the positive rescue experiments by Trpm2 but not by loss-of-function E960D in global Trpm2 KO myocytes, strongly argue in favor of beneficial, rather than detrimental, effects of Trpm2 on cardiac bioenergetics and function, under both basal and stressed conditions. In addition, our current observations on the cardiac-specific Trpm2 KO hearts suggest the role played by Trpm2 channels in neutrophils on cardiac I/R injury is likely to be small. Our conclusion is consistent with the weight of evidence arguing against a pivotal role of neutrophils as a causative factor in most forms of I/R injury in the heart and brain (3).
To extend the potential translational implications of Trpm2 protection on the heart, we used the doxorubicin cardiomyopathy model: a model that involves oxygen free radicals (37) via Doxo's effects on mitochondrial electron transport chain (7, 15). Similar to observations on both gKO-I/R (27) and cKO-I/R hearts (Fig. 7), gKO-Doxo hearts had significantly decreased contractile function when compared with WT-Doxo hearts (Fig. 9, middle and bottom). More importantly, animal survival was much worse in gKO-Doxo mice than in WT-Doxo mice (Fig. 9, top). These observations not only reinforce the importance of Trpm2 channels in protecting the heart against oxidative stress/injury, but are clinically relevant in that because Trpm2 channels are involved in cell proliferation and differentiation, they have increasingly become novel targets for cancer therapy (25). Targeting Trpm2 channels in cancer cells without adequate cardiac protection or sparing cardiac Trpm2 channels, or using a cocktail of Trpm2 inhibitors and anthracyclines developed for cancer chemotherapy, may have serious untoward cardiac side effects including increased mortality. Therefore, the study of Trpm2 in the heart is timely and significant in the emerging field of onco-cardiology.
There are some caveats to the present study. The first is that the XF96 extracellular flux analyzer technology was originally developed for cells or isolated mitochondria adherent to the bottom of the plate. More recently, bioenergetic analysis of cell suspensions such as hematopoietic cells and lymphocytes using XF analyzer has been published (31, 46). A recent study used a similar protocol for bioenergetic profiling of adult mouse cardiac myocytes seeded on laminin-coated plates (33). Although immobilization is recommended for isolated cells, for large cells such as cardiac myocytes, we have performed the analysis in suspension as well as in immobilized cells and obtained similar results (26). We have standardized this protocol and have been routinely using it for measuring OCR in cardiac myocytes. Similarly, tissue slices (50 mg tissue/well, sliced to ∼150 μm) were used to perform OCR measurements, and we obtained reliable and reproducible basal OCR data (26). Indeed, the pattern of OCR differences observed in gKO heart slices expressing either Trpm2, GFP, or E960D mutant (Fig. 5C) is in excellent agreement with that observed in isolated gKO myocytes expressing identical constructs (Fig. 5D). The second is that we measured OCR in intact cardiac myocytes (Fig. 5) rather than in permeabilized myocytes or isolated mitochondria. The rationale is that our current results show that Trpm2-dependent Ca2+ signals regulate mitochondrial function; therefore, it is unlikely that either permeabilized myocytes or isolated mitochondria will replicate the physiological trans-cytosolic relationship. We recognize that changes in basal O2 consumption in intact cardiac myocytes may not necessarily reflect mitochondrial dysfunction. This is because O2 consumption in intact myocytes is influenced by many factors including but not limited to substrate delivery to mitochondria, mitochondrial capacity and function, and ATP turnover rate by ion pumps and contractile apparatus. We hasten to add, however, that when compared with WT myocytes, we have shown in gKO myocytes increased mitochondrial superoxide levels (Figs. 4 and 8); decreased Complex I, III, and IV expression by proteonomics (26); reduced NDUFA4L2 protein levels (Fig. 6); lower mitochondrial membrane potential (26); decreased mitochondrial Ca2+ uptake (26); and smaller mitochondrial Ca2+ uniporter current (26). The data, when viewed in its entirety, suggest the most logical explanation for decreased O2 consumption in intact gKO myocytes is mitochondrial dysfunction.
In summary, [Ca2+]i increase on Trpm2 activation was due to direct Ca2+ entry through Trpm2 channel rather than indirectly through reverse Na+/Ca2+ exchange. Ca2+ entry via Trpm2 channels was essential for maintenance of mitochondrial function in the heart. Ca2+ entry via Trpm2 channels was required to lower the elevated mitochondrial O2.− levels in global Trpm2 knockout myocytes subjected to hypoxia-reoxygenation. After I/R, contractility in WT hearts was significantly better than cardiac-specific Trpm2 knockout hearts, indicating that the role played by Trpm2 channels in neutrophils on cardiac I/R injury is likely to be small. Chronic doxorubicin administration resulted in significantly worse cardiac contractility and increased mortality in global Trpm2 knockout mice when compared with WT mice. We suggest that Trpm2 protects the heart from oxidative stress and injury. Our findings advise caution in the combined use of Trpm2 inhibitors and anthracyclines in cancer chemotherapy.
This work was supported in part by National Institutes of Health Grants RO1-DK46778 (BAM); RO1-HL56205, RO1-HL61690, RO1-HL85503, PO1-HL-75443 and PO1-HL-91799 (WJK); RO1-HL86699 (MM); PO1-HL91799 (Project 2) and RO1-HL123093 (AMF); RO1-HL58672, RO1-HL74854 and RO1-HL123093 (JYC); and in part by independent PA CURE grants to AMF (Project 4; SAP#4100062220) and JYC (Project 5; SAP#4100062220).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: N.E.H., B.A.M., J.W., J.W.E., W.J.K., A.M.F., M.M., and J.Y.C. conception and design of research; N.E.H., B.A.M., J.W., S.R., E.G., J.S., X.-Q.Z., I.H.-L., S.S., and M.M. performed experiments; N.E.H., B.A.M., J.W., J.W.E., S.R., E.G., J.S., X.-Q.Z., I.H.-L., S.S., M.M., and J.Y.C. analyzed data; N.E.H., B.A.M., J.W., J.W.E., S.R., E.G., J.S., X.-Q.Z., I.H.-L., S.S., W.J.K., A.M.F., M.M., and J.Y.C. interpreted results of experiments; N.E.H., S.R., X.-Q.Z., I.H.-L., M.M., and J.Y.C. prepared figures; N.E.H., B.A.M., J.W.E., E.G., I.H.-L., W.J.K., A.M.F., M.M., and J.Y.C. edited and revised manuscript; N.E.H., B.A.M., J.W., J.W.E., S.R., E.G., J.S., X.-Q.Z., I.H.-L., S.S., W.J.K., A.M.F., M.M., and J.Y.C. approved final version of manuscript; B.A.M. and J.Y.C. drafted manuscript.
- Copyright © 2015 the American Physiological Society