Reactive Oxygen Species Mediate Amplitude-DependentHypertrophic and Apoptotic Responses to Mechanical

Stretch in Cardiac MyocytesDavid R. Pimentel, Jay K. Amin, Lei Xiao, Thomas Miller, Jason Viereck, Jennifer Oliver-Krasinski,

Ragavendra Baliga, Jing Wang, Deborah A. Siwik, Krishna Singh, Patrick Pagano,Wilson S. Colucci, Douglas B. Sawyer

Abstract—Oxidative stress stimulates both growth and apoptosis in cardiac myocytes in vitro. We investigated whetheroxidative stress mediates hypertrophy and apoptosis in cyclically stretched ventricular myocytes. Neonatal ratventricular myocytes cultured on laminin-coated silastic membranes were stretched cyclically (1 Hz) at low (nominal5%) and high (nominal 25%) amplitudes for 24 hours. Stretch caused a graded increase in superoxide anion productionas assessed by superoxide dismutase (SOD)-inhibitable cytochromec reduction or electron paramagnetic resonancespectroscopy. The role of reactive oxygen species (ROS) was assessed using the cell-permeable SOD/catalase mimeticsMn(II/III)tetrakis(1-methyl-4-peridyl) (MnTMPyP) and EUK-8. Stretch-induced increases in protein synthesis (3H-leucine incorporation) and cellular protein content were completely inhibited by MnTMPyP (0.05 mmol/L) at both lowand high amplitudes of stretch. In contrast, while MnTMPyP inhibited basal atrial natriuretic factor (ANF) mRNAexpression, the stretch-induced increase in ANF mRNA expression was not inhibited by MnTMPyP. In contrast tohypertrophy, only high-amplitude stretch increased myocyte apoptosis, as reflected by increased DNA fragmentation ongel electrophoresis and an'3-fold increase in the number of TUNEL-positive myocytes. Similarly, only high-amplitudestretch increased the expression of bax mRNA. Myocyte apoptosis and bax expression stimulated by high-amplitudestretch were inhibited by MnTMPyP. Both low- and high-amplitude stretch caused rapid phosphorylation of ERK1/2,while high-, but not low-, amplitude stretch caused phosphorylation of JNKs. Activation of both ERK1/2 and JNKs wasROS-dependent. Thus, cyclic strain causes an amplitude-related increase in ROS, associated with differential activationof kinases and induction of hypertrophic and apoptotic phenotypes.(Circ Res. 2001;89:453-460.)

Key Words: oxidant stressn remodelingn apoptosisn hypertrophy

Hemodynamic overload is associated with an increase inmyocardial wall stress, which has been implicated as a

stimulus for myocardial remodeling. In vitro, both tonic andcyclic mechanical stretch have been used to simulate in-creased wall stress. In cardiac myocytes in vitro, mechanicalstretch stimulates growth,1 alterations in gene expression,2

and apoptosis,3,4 cellular events that are observed in thefailing heart.

In animal models of hemodynamic overload leading tomyocardial remodeling and failure, there is a chronic increasein myocardial oxidative stress,5 which may contribute tomyocardial remodeling.6 Interestingly, tonic mechanicalstretch of rat papillary muscle increases the production ofreactive oxygen species (ROS), which appear to be involvedin mediating myocyte apoptosis in that model.3 An increase inoxidative stress caused by direct addition of ROS also

induces myocyte apoptosis.7 We found that a small increasein myocyte oxidative stress caused by partial inhibition ofCuZn-superoxide dismutase (SOD) resulted in myocyte hy-pertrophy, whereas a higher level of oxidative stress due tomore complete inhibition of SOD caused apoptosis.8 Angio-tensin, tumor necrosis factor-a, a1-adrenergic agonists, andouabain also appear to increase myocardial ROS that mediatemyocyte growth.9–11

These observations led us to hypothesize that oxidativestress mediates the effects of mechanical stretch on myocytegrowth and survival. Using an in vitro system to subjectcardiac myocytes to cyclic mechanical stretch, we testedwhether (a) mechanical stretch causes an increase in theformation of ROS, (b) oxidative stress mediates the effects ofmechanical stretch on myocyte hypertrophy and survival, and(c) stretch, acting via ROS, causes activation of mitogen-ac-

Original received July 26, 2000; resubmission received March 15, 2001; revised resubmission received July 6, 2001; accepted July 25, 2001.From the Cardiovascular Section (D.R.P., J.K.A., L.X., T.M., J.O.-K., R.B., J.W., D.A.S., W.S.C., D.B.S.), Department of Medicine, Boston University

Medical Center; The Myocardial Biology Unit (D.R.P., J.K.A., L.X., T.M., R.B., J.W., D.A.S., W.S.C., D.B.S.), Boston University School of Medicine;the Department of Biophysics (J.V.), Boston University Medical Center; and the Boston Veterans Administration Medical Center (K.S.), Boston, Mass.Current affiliation for P.P. is Henry Ford Hospital, Detroit, Mich.

Correspondence to Douglas B. Sawyer, MD, PhD, Cardiovascular Section, Boston University Medical Center, 88 E Newton St, Boston, MA 02118.E-mail douglas.sawyer@bmc.org

© 2001 American Heart Association, Inc.

Circulation Researchis available at http://www.circresaha.org

453

tivated protein kinases (MAPKs) that have been implicated inthe regulation of myocyte hypertrophy and survival.

Materials and Methods

Preparation of Neonatal Rat VentricularMyocytes (NRVMs)Ventricular myocytes were isolated from neonatal rats by a modifi-cation of Kasten’s technique, as previously described.12 Unlessotherwise stated, the cells were plated at 800 cells/mm2 on 6-wellBioFlex (Flexcell International) plates precoated with 0.02 mmol/Llaminin (GibcoBRL). After 24 hours in DMEM (GibcoBRL) con-taining 7% (vol/vol) inactivated fetal bovine serum (GibcoBRL) and1% (vol/vol) penicillin-streptomycin (GibcoBRL), the medium waschanged to serum-free DMEM for an additional 24 hours beforetreatment. Approximately 650 to 700 cells/mm2 remained platedafter this second medium change.

Mechanical StretchNRVMs were subjected to cyclic stretch by means of the Flexercellcomputer-driven vacuum system as has been previously described.13

NRVMs plated on BioFlex culture plates were placed on a gasketedbaseplate in a 37°C, 5% CO2 incubator. The plates were subjected toa vacuum of25 or 221 kPa, low and high stretch, respectively, ata frequency of 1 Hz applied in a square-wave pattern of strain lasting0.5 seconds per cycle for up to 24 hours. Using these parameters, thissystem produces a gradient of deformation across the membrane,with a maximal deformation of 5% and 25% at25 and221 kPa,respectively. Parallel BioFlex culture plates not subjected to stretchserved as controls.

Measurement of Superoxide ReleaseSuperoxide production was measured as the SOD-inhibitable reduc-tion of cytochromec.14 NRVMs were stretched or not stretched for24 hours. The cells were washed with phosphate-buffered saline(PBS; 137 mmol/L NaCl, 2.68 mmol/L KCl, 1.47 mmol/L KH2PO4,8.1 mmol/L Na2HPO4, 0.9 nmol/L CaCl2, 0.5 nmol/L MgCl2, and2 mmol/L EDTA [pH 7.4]). Cytochromec (9.9 mmol/L, with 1200U/mL of catalase and 0.1 mmol/L EDTA in phenol-free medium)with or without 300 U/mL SOD was added for 1 hour at 37°Cwithout mechanical stretch. Cytochromec reduction was measuredas the absorbance at 550 nm. The difference in A550 between wellswith and without SOD was calculated.

Electron Paramagnetic Resonance (EPR)Measurement of ROS ProductionEPR measurements were performed as described by Sorescu et al.15

Myocytes were washed with ice-cold PBS and scraped in a50 mmol/L phosphate buffer (pH 7.4; pretreated with Chelex [5g/100 mL] for 2 hours) containing aprotinin (10mg/mL), pepstatin(0.7 mg/mL), and PMSF (0.5 mmol/L). The cells were sonicated for30 seconds on ice followed by centrifugation at 50 000g for 30minutes at 4 C. The membrane pellets were resuspended in 25mL oflysis buffer and aliquots containing 25mg of protein (Bradfordmethod) were resuspended with 0.1 mmol/L diethylenetriaminepen-taacetic acid, 300mmol/L NADH or NADPH, and 25 mmol/LDEPMPO (OXIS International). The assay mixture was incubated at37°C for 30 minutes and was stopped by snap-freezing. Afterthawing the sample at room temperature, EPR measurements wereperformed at room temperature using a Bruker EMX spectrometerwith an ER4119HS high-sensitivity resonator cavity. The instrumentsettings were as follows: modulation amplitude 3 G; time constant 82ms; modulation frequency 100 kHz; microwave power 20 mW; andmicrowave frequency 9.40 GHz. Five scans were accumulated, andthe double integral of the second peak of the composite spectrum wasused for comparison of conditions.

Measurement of3H-Leucine IncorporationIncorporation of3H-leucine was measured over the final 4 hours ofa 24-hour period of cyclic stretch, and total cellular protein contentwas determined by the Bradford method as previously described.16

After cell lysis, samples were resuspended in a buffer composed of100 mmol/L Tris, 10 mmol/L EDTA, 1 mol/L NaCl, and 10mg ofthe fluorochrome H33258 (Calbiochem Corp). In all experiments,cellular DNA content was determined by a fluorometric quantifica-tion of the H33258 in comparison to a standard curve with herringsperm DNA, and3H-leucine incorporation is reported as CPM/mgDNA.16

Changes in Total Cellular ProteinAfter precipitation of myocyte lysates with 10% TCA and resuspen-sion in 0.4 mol/L NaOH buffer, aliquots were removed for determi-nation of cellular protein. Total cellular protein was determined bythe Bradford method using Coomassie Blue (BioRad Laboratories).Cellular protein content was normalized to DNA content as de-scribed above.

Northern HybridizationAfter 24 hours of cyclical stretch or nonstretch control, cells werecollected and total RNA isolated as previously described.12 Northernhybridizations with32P-labeled rat cDNAs for prepro-atrial natriuret-ic peptide (ANP) (courtesy of C. Seidman, Brigham and Women’sHospital, Boston, Mass) and bax (courtesy of J. Tilly, MassachusettsGeneral Hospital, Boston, Mass) were performed and quantified aspreviously described.12 mRNA levels were normalized to 18Sribosomal RNA determined by reprobing blots with32P-labeledoligonucleotide complementary to 18S rRNA.

DNA LadderingAfter 24 hours of cyclical stretch or nonstretch control, total DNAfrom NRVMs in a single Flexcell plate (6 wells) was isolated andelectrophoresed as previously described17 to visualize the pattern ofDNA fragmentation.

In Situ Nick-End Labeling of DNA StrandBreaks (TUNEL)NRVMs plated on BioFlex plates were stretched or not stretched for24 hours as indicated. The cells were TUNEL-labeled and nucleistained with 10mg/mL Hoechst as previously described.8 After threewashes in PBS, a glass coverslip was attached to the underside of theBioFlex silastic membrane, which was cut to the size of thecoverslip. The membranes were then mounted onto glass slides andviewed with an epifluorescent microscope. Slides were viewed in ablinded fashion and at least 250 total nuclei (Hoechst stained) werecounted from each coverslip (a total of 4 coverslips per experimentwere viewed), and the number of TUNEL-positive cells was deter-mined for each field.

Immunoblots for Activated MAPK and BaxThe activation of ERKs, JNK, and total bax was examined by usingWestern blot analysis using specific antibodies.18 Total cellularhomogenates from rat myocytes were prepared from one Flexcellplate (6 wells), and equal amounts (50mg) of the denatured proteinswere loaded and separated on 10% or 12% SDS-polyacrylamide gels(Mini Protean II, BioRad) and transferred to polyvinylidene difluo-ride membranes (Hybond-P, Amersham Life Science). The mem-brane was blocked with 5% nonfat dry milk (BioRad) in PBS for 1hour. The membranes were incubated with 1:200 rabbit polyclonalantibody to phospho-p44/42 MAPK, phospho-JNK (Cell Signaling),or bax (Santa Cruz Biotechnology) overnight in 4% BSA in TBS at4°C. The membranes were washed six times with TBS followed byincubation for 1 hour with a horseradish peroxidase–labeled goatanti-rabbit antibody (Santa Cruz) in 4% BSA in TBS. Afterward, themembranes were exposed to a chemiluminescent reagent (Pierce)and autoradiographed for 1 to 2 minutes.

454 Circulation Research August 31, 2001

MaterialsSOD, catalase, and cytochromec were from Sigma. Mn(II/III)tetrakis(1-methyl-4-peridyl)porphyrin (MnTMPyP)19 was fromCalbiochem, Inc. EUK-8 was gift from Eukarion (Bedford, Mass).32P-CTP and3H-leucine were from New England Nuclear, Inc. Allother reagents were from Sigma.

Statistical AnalysisAll data are reported as mean6SEM. Significance was determinedwith the InStat program for student’s unpairedt tests or ANOVAwhere appropriate. A value ofP,0.05 was considered to besignificant.

ResultsMechanical Stretch Increases MyocyteO2

2 ProductionIsolated NRVMs plated onto silastic membranes coated withlaminin were subjected to cyclic mechanical stretch (1 Hz) atlow (5%) and high (25%) amplitudes for 24 hours. SOD-inhibitable cytochromec reduction in medium was assessedas a measure of myocyte O2

2 production. There was a gradedincrease in O22 production in stretched cells, with a 230%increase in O22 production at low-amplitude stretch and a405% increase at high-amplitude stretch (Figure 1A).

To confirm the increased production of ROS and examinepotential sources, we used EPR. Using DEPMPO as a spintrap with membrane fractions of myocytes incubated withNADPH, we observed the characteristic hydroxyl adduct(DEPMPO-OOH). Formation of the DEPMPO-OOH adductwas increased in myocytes subjected to stretch (Figures 1Band 1C). DEPMPO-OOH adduct formation was completelyinhibited with the addition of MnSOD. There was greaterformation of DEPMPO-OOH in the presence of NADPH thanNADH at baseline, and with stretch the absolute increase wasgreater with NADPH (versus NADH, Figure 1C). Additioneither of the endothelin receptor antagonist (BQ123, 1mmol/L) or the angiotensin receptor antagonist (losartan, 10mmol/L) had no effect on baseline levels of ROS productionbut significantly inhibited the formation of DEPMPO-OOHin stretched myocytes by 42% and 41%, respectively (n53;P,0.05 for both).

Stretch-Induced Myocyte Hypertrophy IsMediated by ROS3H-leucine incorporation and total protein were assessed asmeasures of myocyte hypertrophy. Cyclic stretch for 24 hourscaused a graded increase in3H-leucine incorporation, with a13% increase at low amplitude and a 80% increase at highamplitude (Figures 2A and 2B). Likewise, stretch caused anamplitude-dependent increase in total cellular protein (Fig-ures 2C and 2D).

Figure 1. Effect of cyclic mechanical stretch on myocyte super-oxide anion production. NRVMs maintained in serum-freemedium for 24 hours were subjected to cyclic stretch (1 Hz) for24 hours at low (5%) and high (25%) amplitudes. A, SOD-inhibitable cytochrome c reduction was measured spectropho-tometrically as a measure of O2

2 production and showed agraded increase in O2

2 production in stretched cells, with a230645% increase in O2

2 production at low-level stretch (n53;**P50.02 vs control) and a 4056133% increase in O2

2 produc-tion at high-level stretch (n55; *P50.005 vs control, P50.02 vslow-level stretch). B and C, ESR detection of ROS from myocyte

membranes after 24 hours of high-amplitude stretch. B, Repre-sentative DEPMPO-OOH spectra (see Materials and Methods) ofmyocyte membranes that were stimulated with NADH (top),NADPH (middle), and inhibited by SOD (NADPH stimulated andpretreated with SOD) (bottom). C, Quantitative comparison ofEPR signals (see Materials and Methods). NADPH-dependentEPR signal increased significantly (n53; P,0.01) in myocytesafter high-amplitude stretch. NADH-dependent EPR signal waslower at baseline and did not increase significantly in stretchedmyocytes (n53; P.0.1).

Pimentel et al ROS Mediate Stretch-Induced Phenotypes 455

To examine whether stretch-induced O22 production me-

diates myocyte hypertrophy, we measured3H-leucine incor-poration in the presence and absence the ROS scavengersEUK-8 and MnTMPyP.19 Both ROS scavengers abolishedthe increases in3H-leucine uptake (Figures 2A and 2B) andtotal protein (Figures 2C and 2D) in response to low- andhigh-amplitude stretch. The ROS scavengers alone had noeffect on3H-leucine incorporation or total protein content.

Myocyte hypertrophy is often associated with increasedexpression of fetal genes such as atrial natriuretic factor(ANF). Stretch at low and high amplitude increased ANFmRNA by 62% and 104%, respectively (Figure 3). WhileMnTMPyP inhibited baseline ANF expression, MnTMPyPdid not decrease the stretch-induced increases in ANFexpression.

Stretch-Induced Myocyte Apoptosis Is Mediatedby ROSDNA isolated from myocytes stretched at high amplitude for24 hours showed a ladder pattern on agarose gel electrophore-sis indicative of apoptosis (Figure 4A). Minimal DNA lad-dering was evident in static cells (Figure 4A) or cellsstretched at low amplitude (data not shown). Pretreatmentwith MnTMPyP or EUK-8 prevented DNA laddering withhigh-amplitude stretch. TUNEL staining of adherent myo-cytes in separate experiments was performed to quantify thepercent of apoptotic myocytes. Low-amplitude stretch tendedto decrease the number of myocytes staining positive by theTUNEL method, whereas high-amplitude stretch caused an'3-fold increase that was inhibited by MnTMPyP (Figure4B), suggesting that apoptosis with high-amplitude stretchwas ROS-dependent.

The expression of bax is increased in response to apoptoticlevels of oxidative stress in cardiac myocytes.8,7 We thereforeexamined whether bax was increased by mechanical stretch.Low-amplitude stretch had no effect on bax mRNA levels,whereas high-amplitude stretch increased bax mRNA expres-sion. Stretch-induced bax expression was prevented byMnTMPyP (Figures 5A and 5B). Likewise, bax protein levelsassessed by immunoblotting increased with high-amplitudestretch, and the increase was inhibited by MnTMPyP orEUK-8 (Figure 5C).

Stretch-Induced Signaling Is Mediated by ROSROS can stimulate MAPK signaling in cardiac myocytes.20,21

We examined the effects of low- and high-amplitude stretchon activation of ERK1/2 and JNKs. ERK activation after 8minutes of stretch, as assessed by immunoblotting for phos-phorylated ERK1/2, was increased to similar levels at bothlow and high amplitudes of myocyte stretch compared withcontrol myocytes. MnTMPyP inhibited the activationERK1/2 in both amplitudes of stretch (Figures 6A and 6B).Phosphorylation of JNKs was also detected in myocytes after

Figure 2. Effect of the ROS scavengers MnTMPyP and EUK-8on stretch-induced myocyte 3H-leucine incorporation and totalprotein. Myocytes underwent cyclic mechanical stretch with andwithout the SOD mimetic MnTMPyP (0.05 mmol/L) and EUK-8(0.05 mmol/L) for 24 hours. A, 3H-leucine incorporation (CPM/mgDNA) for low-level stretch. There is an increase in 3H-leucineincorporation in myocytes subjected to 5% stretch. MnTMPyPand EUK-8 completely abolished the increase in 3H-leucineincorporation. (n56; *P,0.01 vs control; # and @P,0.01 vsstretch alone). B, 25% stretch increased leucine incorporationinto cardiac myocytes to a greater magnitude than 5% stretch(P,0.001 for 25% vs 5%). This effect was inhibited by MnTMPyPand EUK-8. Neither MnTMPyP nor EUK-8 alone had an effecton leucine incorporation (n56; **P,0.01 vs control; @@P,0.01vs stretch alone; ##P,0.05 vs stretch alone). C, Low-levelstretch increased total cellular protein content (n56,

*P,0.01 vs control; @ and #P,0.01 vs stretch alone). Similarly,MnTMPyP and EUK-8 inhibited stretch-induced increases intotal protein. D, High-level stretch increased total cellular pro-tein. MnTMPyP and EUK-8 inhibited this effect. Neither MnT-MPyP nor EUK-8 alone had any effect on total protein content(n54, **P,0.01 vs control; @@ and ##P,0.01 vs stretch alone).

456 Circulation Research August 31, 2001

8 minutes of high-, but not low-amplitude stretch, and thiswas inhibited by MnTMPyP (Figures 6C and 6D).

DiscussionMechanical stretch of cardiac myocytes, either tonic orcyclic, leads to hypertrophic growth1,22 as well as apopto-sis.3,4 Individually, these effects of mechanical stretch havebeen studied in some detail. To our knowledge, this is the firstreport to show an amplitude-dependent shift in myocytephenotype in response to mechanical stretch. We have pre-viously shown that graded increases in the level of myocyteoxidative stress induce a graded phenotype shift in cardiacmyocytes, from hypertrophy and fetal gene expression at lowlevels of oxidative stress, to apoptosis at high levels ofoxidative stress.8 The response of cardiac myocytes to lowand high levels of mechanical stretch is similar to the effect

of directly increasing myocyte oxidative stress. First, there isa stretch amplitude-dependent increase in ROS production.Second, there is an amplitude-dependent phenotype shift,from hypertrophy at low levels of stretch, to apoptosis at highlevels of mechanical stretch. Finally, the phenotype shift ateach level of stretch was inhibited by pharmacologicantioxidants.

Stretch-Induced ROS ProductionCyclic stretch increased myocyte ROS production in anamplitude-dependent manner. Static stretch of isolated pap-illary muscles causes a similar amplitude-dependent increasein myocardial ROS production.3 We found that ROS produc-tion in stretched myocytes was at least in part mediated byangiotensin and endothelin receptor activation, consistentwith prior work implicating paracrine release of angiotensinand endothelin in stretch-induced hypertrophy23 and apopto-sis.3,11 Inhibition of either receptor caused a similar reductionin superoxide production consistent with prior reports sug-gesting that angiotensin acts through endothelin in cardiacmyocytes.24

The cellular source of increased ROS production instretched myocytes remains to be determined, but it mayinvolve an NAD(P)H oxidase system and/or changes inmitochondrial production of ROS. Mitochondria are a recog-nized source of ROS in the myocardium.25 Mechanical stretchincreases myocardial oxygen consumption26,27 and may thus

Figure 3. Effect of the SOD mimetic on stretch-induced myo-cyte ANF mRNA expression. A, Representative Northern blotshowing induction of ANF expression by 5% and 25% stretch.MnTMPyP lowers the basal ANF expression, but ANF expres-sion increased from this lower basal level after either low- orhigh-amplitude stretch. B, Expression of ANF was normalized to18S rRNA and expressed as percent change versus control.Analysis by two-way ANOVA shows that stretch at low and highamplitudes increased ANF mRNA by 62.2622.2% (n56;*P50.009) and 104.3640.8% (n56; **P50.01). MnTMPyP inhib-ited the basal expression of ANF mRNA (P,0.05 for both 5%and 25% stretch). There was no significant effect ofMnTMPyP on stretch-induced ANF mRNA expression.

Figure 4. Effect of ROS scavengers on mechanical stretch–in-duced myocyte apoptosis. A, Gel electrophoresis of DNA frommyocytes after 25% stretch for 24 hours shows a pattern ofDNA laddering consistent with the induction of myocyte apopto-sis. Pretreatment with the SOD mimetic MnTMPyP or EUK-8prevented the induction of DNA degradation. Gel is representa-tive of 4 separate experiments. B, Quantification of TUNEL-positive myocytes after stretch with and without the SODmimetic MnTMPyP. There is an increase in the number ofTUNEL-positive myocytes with 25% stretch (n55; *P50.02),which was inhibited by pretreatment with MnTMPyP (n53;#P50.03); 5% stretch caused a trend toward a reduction inTUNEL-positive cells (n53; P50.09).

Pimentel et al ROS Mediate Stretch-Induced Phenotypes 457

lead to increased ROS formation by mitochondria.28 Indeed,in failing myocardium, mitochondria appear to be a source ofincreased hydroxyl radical (OH·) production, possibly due toa decrease in complex I activity.29,30However, given that thesubstrate NADPH produced much higher EPR signals thanNADH in stretched myocytes, an NADPH-dependent oxidasesuch as the plasmalemmal NAD(P)H oxidase31 or nitric oxidesynthase32 seems more likely to be the source of increasedROS in mechanically stretched myocytes. Further work isneeded to elucidate the sources of ROS that are involved inthe effects of mechanical stretch on myocyte growth andsurvival.

Role of ROS in Stretch-InducedMyocyte HypertrophyThe magnitude of the growth response we observed withcyclic stretch is similar to that reported by others using tonicstretch.1 Both 3H-leucine incorporation and cellular proteincontent showed a graded increase with the amplitude ofstretch. Protein synthesis was associated with increased

expression of ANF, suggestive of a fetal phenotype that istypical of myocyte hypertrophy. Stretch-induced protein syn-thesis (but not ANF mRNA expression) was abolished byantioxidants. These results add to a growing literature sug-gesting that ROS can act as signaling mediators of growth.

In vascular smooth muscle cells, angiotensin increasesROS production resulting in cell growth through the activa-tion of an NAD(P)H oxidase system.33 Angiotensin, tumor

Figure 5. Effect of the SOD mimetic MnTMPyP on stretch-induced increases in bax expression. A, Representative North-ern blot showing bax mRNA and 18S rRNA expression in myo-cytes. Bax expression is increased in myocytes after 24 hoursof 25% stretch. This effect is inhibited by the SOD mimeticMnTMPyP. B, Bax expression was normalized to 18S rRNAexpression. Whereas 5% stretch caused no change in baxexpression, 25% stretch increased bax expression by123.8620.2% (n54; *P50.009), which was inhibited byMnTMPyP (n54; #P50.04 vs stretch). C, Bax protein increaseswith high-level stretch and is inhibited with both MnTMPyP andEUK-8. Immunoblot is representative of 3 separate experiments.

Figure 6. Effect of stretch on activation of ERK1/2 and JNKs.Myocytes were stretched at 5% or 25% for 15 minutes, andlevels of phosphorylated ERK and JNK were assessed in celllysates using antibodies to phosphorylated active ERK1/2 orJNK. A, Representative immunoblots for phosphorylatedERK1/2. B, Densitometric analysis of ERK1/2 is expressed aspercent change versus control. Stretch at low and high levelsincreased levels of phospho-ERK1/2 by 158655.2% (n54;*P50.02) and 136.9634% (n56; **P50.03), respectively.MnTMPyP inhibited the increase in ERK activity at both 5%(n54; #P50.03) and 25% (n55; ##P50.04) stretch. MnTMPyPalone did not significantly alter ERK activation. C, Representa-tive immunoblots for phosphorylated JNKs. D, Densitometricanalysis of JNK phosphorylation is expressed as percentchange versus control. Low-level stretch did not change thelevel of JNK phosphorylation whereas high levels increased JNKby 193.1623.1% (n53; *P50.002). MnTMPyP inhibited theincrease in JNK expression at high-level stretch (n53;##P50.008) but had no effect on its own.

458 Circulation Research August 31, 2001

necrosis factor-a, anda-adrenergic stimulation likewise havebeen shown to cause myocyte hypertrophy through an ROS-dependent pathway.9,10Endothelin appears to modulate early-response gene expression through a ROS-dependent pathwayinvolving ras,34 and ouabain causes hypertrophy via ROS-dependent activation of aras/MAPK pathway. Stretch-induced hypertrophic signaling has also been shown toinvolve a rac-dependent pathway.35 Rac is part of theNAD(P)H oxidase complex,36 again supporting the thesis thatan NAD(P)H oxidase system may be an important source ofROS under these conditions.

Role of ROS in Stretch-Induced ApoptosisThe proapoptotic effects of high-amplitude stretch are com-plementary to the findings of Anversa and colleagues whoreported that stretch of a papillary muscle preparation led toparallel increases in ROS formation and myocyte apoptosis.3

ROS are well known to induce apoptosis in many cell typesincluding cardiac myocytes. Addition of extracellular sourcesof ROS, such as H2O2 or xanthine/xanthine oxidase, causedapoptosis in neonatal rat myocytes7 and a myocyte-derivedcell line.21 Similarly, we and others have shown that anincrease in intracellular ROS formation due to (a) inhibitionof SOD,8 (b) addition of an O2

2 generator (eg, anthracy-clines),17 or (c) conditions that favor the formation of intra-cellular peroxynitrite37,32 results in increased myocyteapoptosis.

Interestingly, while we found that both low- and high-amplitude stretch increased ROS production, apoptosis wasincreased only with high-amplitude stretch. Likewise, wepreviously found that graded inhibition of SOD resulting intwo levels of ROS caused hypertrophy at both ROS levels,but apoptosis only at the higher level.8 Taken together, thesefindings suggest that the quantity and/or quality of ROS is animportant determinant of the activation of the apoptoticcascade.

ROS-Dependent Activation of MAPKIn cardiac myocytes, H2O2 and superoxide anion activateMAPKs38 that have been implicated in the regulation of cellgrowth.39 We found that mechanical stretch caused activationof ERK1/2 at both low and high amplitude, whereas JNK wasactivated only at high-amplitude stretch. The activation ofboth ERK1/2 and JNK was ROS-dependent. ERK1/2 hasbeen implicated in mediating myocyte growth,40 whereasJNK has been implicated in mediating both growth andapoptosis.41,21 Interestingly, the activation of ERK1/2 inresponse to stretch was not amplitude-dependent. Thus, itwould appear that the graded increase in protein synthesis athigh- versus low-amplitude stretch requires additional signal-ing pathways (eg, JNKs). Thus, differential activation ofMAPK signaling pathways may be involved in the amplitude-dependent effects of stretch on myocyte phenotype.

ConclusionOur data demonstrate that the effects of cyclic mechanicalstretch on myocyte growth and death are amplitude-dependent and mediated by ROS. These findings furthersupport the thesis that oxidative stress mediates important

aspects of myocardial remodeling in response to hemody-namic overload. This in vitro system should provide theability to understand the molecular mechanisms that deter-mine the effects of mechanical overload on myocytephenotype.

AcknowledgmentsThis work was supported in part by NIH grants HL61639 (W.S.C.),HL057947 (K.S.), HL03878 (D.B.S.); a Grant-in-Aid from the AHA,Massachusetts Affiliate (D.B.S.); and a Merit grant from the Depart-ment of Veterans Affairs (K.S.). The authors would like to thankMargaret Gibbons for administrative assistance.

References1. Sadoshima J-i, Xu Y, Slayter HS, Izumo S. Autocrine release of angio-

tensin II mediates stretch-induced hypertrophy of cardiac muscles invitro. Cell. 1993;75:977–984.

2. Watson PA, Hannan R, Carl LL, Giger KE. Desmin gene expression incardiac myocytes is responsive to contractile activity and stretch.Am JPhysiol. 1996;270(4 pt 1):C1228–C1235.

3. Cheng W, Li B, Kajstura J, Li P, Wolin MS, Sonnenblick EH, Hintze TH,Olivetti G, Anversa P. Stretch-induced programmed myocyte cell death.J Clin Invest. 1995;96:2247–2259.

4. Leri A, Claudio PP, Li Q, Wang X, Reiss K, Wang S, Malhotra A,Kajstura J, Anversa P. Stretch-mediated release of angiotensin II inducesmyocyte apoptosis by activating p53 that enhances the local renin- an-giotensin system and decreases the Bcl-2-to-Bax protein ratio in the cell.J Clin Invest. 1998;101:1326–1342.

5. Dhalla AK, Singal PK. Antioxidant changes in hypertrophied and failingguinea pig hearts.Am J Physiol. 1994;266:H1280–H1285.

6. Dhalla AK, Hill MF, Singal PK. Role of oxidative stress in transition ofhypertrophy to heart failure.J Am Coll Cardiol. 1996;28:506–514.

7. von Harsdorf R, Li PF, Dietz R. Signaling pathways in reactive oxygenspecies-induced cardiomyocyte apoptosis.Circulation. 1999;99:2934–2941.

8. Siwik DA, Tzortzis JD, Pimental DR, Chang DL, Pagano PJ, Singh K,Sawyer DB, Colucci WS. Inhibition of copper-zinc superoxide dismutaseinduces cell growth, hypertrophic phenotype, and apoptosis in neonatalrat cardiac myocytes in vitro.Circ Res. 1999;85:147–153.

9. Nakamura K, Fushimi K, Kouchi H, Mihara K, Miyazaki M, Ohe T,Namba M. Inhibitory effects of antioxidants on neonatal rat cardiacmyocyte hypertrophy induced by tumor necrosis factor-a and angiotensinII. Circulation. 1998;98:794–799.

10. Amin JK, Xiao L, Pimental DR, Pagano PJ, Singh K, Sawyer DB,Colucci WS. Reactive oxygen species mediatea-adrenergic receptor-stimulated hypertrophy in adult rat ventricular myocytes.J Mol CellCardiol. 2001;33:131–139.

11. Xie Z, Kometiani P, Liu J, Li J, Shapiro JI, Askari A. Intracellularreactive oxygen species mediate the linkage of Na1/K1-ATPase to hy-pertrophy and its marker genes in cardiac myocytes.J Biol Chem. 1999;274:19323–19328.

12. Thaik CM, Calderone A, Takahashi N, Colucci WS. Interleukin-1b mod-ulates the growth and phenotype of neonatal rat cardiac myocytes.J ClinInvest. 1995;96:1093–1099.

13. Banes AJ, Gilbert J, Taylor D, Monbureau O. A new vacuum-operatedstress-providing instrument that applies static or variable duration cyclictension or compression to cells in vitro.J Cell Sci. 1985;75:35–42.

14. McCord JM, Fridovich I. Superoxide dismutase. An enzymic function forerythrocuprein (hemocuprein).J Biol Chem. 1969;244:6049–6055.

15. Sorescu D, Somers MJ, Lassegue B, Grant S, Harrison DG, GriendlingKK. Electron spin resonance characterization of the NAD(P)H oxidase invascular smooth muscle cells.Free Radic Biol Med. 2001;30:603–612.

16. Kaye D, Pimental D, Prasad S, Maki T, Berger HJ, McNeil PL, SmithTW, Kelly RA. Role of transiently altered sarcolemmal membrane per-meability and basic fibroblast growth factor release in the hypertrophicresponse of adult rat ventricular myocytes to increased mechanicalactivity in vitro. J Clin Invest. 1996;97:281–291.

17. Sawyer DB, Fukazawa R, Arstall MA, Kelly RA. Daunorubicin-inducedapoptosis in rat cardiac myocytes is inhibited by dexrazoxane.Circ Res.1999;84:257–265.

Pimentel et al ROS Mediate Stretch-Induced Phenotypes 459

18. Marshall CJ. Specificity of receptor tyrosine kinase signaling: transientversus sustained extracellular signal-regulated kinase activation.Cell.1995;80:179–185.

19. Faulkner KM, Liochev SI, Fridovich I. Stable Mn(III) porphyrins mimicsuperoxide dismutase in vitro and substitute for it in vivo.J Biol Chem.1994;269:23471–23476.

20. Clerk A, Fuller SJ, Michael A, Sugden PH. Stimulation of “stress-regulated” mitogen-activated protein kinases (stress-activated proteinkinases/c-Jun N-terminal kinases and p38-mitogen-activated proteinkinases) in perfused rat hearts by oxidative and other stresses.J BiolChem. 1998;273:7228–7234.

21. Turner NA, Xia F, Azhar G, Zhang X, Liu L, Wei JY. Oxidative stressinduces DNA fragmentation and caspase activation via the c-Jun NH2-terminal kinase pathway in H9c2 cardiac muscle cells.J Mol CellCardiol. 1998;30:1789–1801.

22. Kira Y, Nakaoka T, Hashimoto E, Okabe F, Asano S, Sekine I. Effect oflong-term cyclic mechanical load on protein synthesis and morphologicalchanges in cultured myocardial cells from neonatal rat.Cardiovasc DrugsTher. 1994;8:251–262.

23. Sadoshima J-i, Xu Y, Slayter HS, Izumo S. Autocrine release of angio-tensin II mediates stretch-induced hypertrophy of cardiac muscles invitro. Cell. 1993;75:977–984.

24. Gray MO, Long CS, Kalinyak JE, Li HT, Karliner JS. Angiotensin IIstimulates cardiac myocyte hypertrophy via paracrine release of TGF-b1and endothelin-1 from fibroblasts.Cardiovasc Res. 1998;40:352–363.

25. Sawyer DB, Colucci WS, Mitochondrial oxidative stress in heart failure:“oxygen wastage” revisited.Circ Res. 2000;86:119–120.

26. Gunning JF, Cooper G, Harrison CE, Coleman HN III. Myocardialoxygen consumption in experimental hypertrophy and congestive heartfailure due to pressure overload.Am J Cardiol. 1973;32:427–436.

27. Strauer BE, Beer K, Heitlinger K, Hofling B. Left ventricular systolicwall stress as a primary determinant of myocardial oxygen consumption:comparative studies in patients with normal left ventricular function, withpressure and volume overload and with coronary heart disease.Basic ResCardiol. 1977;72:306–313.

28. Boveris A, Cadenas E, Stoppani AO. Role of ubiquinone in the mito-chondrial generation of hydrogen peroxide.Biochem J. 1976;156:435–444.

29. Ide T, Tsutsui H, Kinugawa S, Utsumi H, Kang D, Hattori N, Uchida K,Arimura Ki, Egashira K, Takeshita A. Mitochondrial electron transportcomplex I is a potential source of oxygen free radicals in the failingmyocardium.Circ Res. 1999;85:357–363.

30. Ide T, Tsutsui H, Kinugawa S, Suematsu N, Hayashidani S, Ichikawa K,Utsumi H, Machida Y, Egashira K, Takeshita A. Direct evidence forincreased hydroxyl radicals originating from superoxide in the failingmyocardium.Circ Res. 2000;86:152–157.

31. Griendling KK, Minieri CA, Ollerenshaw JD, Alexander RW. Angioten-sin II stimulates NADH and NADPH oxidase activity in cultured vascularsmooth muscle cells.Circ Res. 1994;74:1141–1148.

32. Xia Y, Dawson VL, Dawson TM, Snyder SH, Zweier JL. Nitric oxidesynthase generates superoxide and nitric oxide in arginine-depleted cellsleading to peroxynitrite-mediated cellular injury.Proc Natl Acad SciU S A. 1996;93:6770–6774.

33. Ushio-Fukai M, Zafari AM, Fukui T, Ishizaka N, Griendling KK. p22phox

is a critical component of the superoxide-generating NADH/NADPHoxidase system and regulates angiotensin II-induced hypertrophy invascular smooth muscle cells.J Biol Chem. 1996;271:23317–23321.

34. Cheng TH, Shih NL, Chen SY, Wang DL, Chen JJ. Reactive oxygenspecies modulate endothelin-I-induced c-fos gene expression in cardio-myocytes.Cardiovasc Res. 1999;41:654–662.

35. Aikawa R, Komuro I, Yamazaki T, Zou Y, Kudoh S, Zhu W, KadowakiT, Yazaki Y. Rho family small G proteins play critical roles inmechanical stress- induced hypertrophic responses in cardiac myocytes.Circ Res. 1999;84:458–466.

36. Kreck ML, Freeman JL, Abo A, Lambeth JD. Membrane association ofRac is required for high activity of the respiratory burst oxidase.Bio-chemistry. 1996;35:15683–15692.

37. Arstall MA, Sawyer DB, Fukazawa R, Kelly RA. Cytokine-mediatedapoptosis in cardiac myocytes: the role of inducible nitric oxide synthaseinduction and peroxynitrite generation.Circ Res. 1999; 85:829–840.

38. Aikawa R, Komuro I, Yamazaki T, Zou Y, Kudoh S, Tanaka M, ShiojimaI, Hiroi Y, Yazaki Y. Oxidative stress activates extracellular signal-regulated kinases through Src and Ras in cultured cardiac myocytes ofneonatal rats.J Clin Invest. 1997;100:1813–1821.

39. Force T, Bonventre JV. Growth factors and mitogen-activated proteinkinases.Hypertension. 1998;31:152–161.

40. Yue TL, Gu JL, Wang C, Reith AD, Lee JC, Mirabile RC, Kreutz R,Wang Y, Maleeff B, Parsons AA, Ohlstein EH. Extracellular signal-regulated kinase plays an essential role in hypertrophic agonists,endothelin-1 and phenylephrine-induced cardiomyocyte hypertrophy.J Biol Chem. 2000;275:37895–37901.

41. Ramirez MT, Sah VP, Zhao XL, Hunter J, Chien KR, Brown JH. TheMEKK-JNK pathway is stimulated bya1-adrenergic receptor and rasactivation and is associated with in vitro and in vivo cardiac hypertrophy.J Biol Chem. 1997;272:14057–14061.

460 Circulation Research August 31, 2001