Role of tumor necrosis factor-� in myocardial dysfunction andapoptosis during hindlimb ischemia and reperfusion

Xiangru Lu, MD; Joel A. Hamilton, MSc; Ji Shen, MD, MSc; Theresa Pang, BSc; Douglas L. Jones, PhD;Richard F. Potter, PhD; J. Malcolm O. Arnold, MD; Qingping Feng, MD, PhD

Prolonged limb ischemia fol-lowed by reperfusion (I/R) in-duces systemic inflammatoryresponse syndrome (SIRS) (1,2). Clinical studies suggest that the heartis at risk in the setting of SIRS. Patientsundergoing complex vascular surgerythat involves prolonged clamping of theaorta have been shown to develop SIRS(3–6) and exhibit a four-fold increasedrisk of cardiac events including heart fail-ure following surgery (7, 8). Given thatthe major clinical consequences of SIRS

include multiple organ dysfunctionand/or death, it is important to under-stand the underlying mechanisms re-sponsible for these phenomena. Tumornecrosis factor (TNF)-� is considered oneof the primary mediators of SIRS (9–12).Although the main sources of TNF-� areresident macrophages found within or-gans, recent evidence indicates that car-diomyocytes themselves are a rich sourceof TNF-� production in the heart (13–15). The cardiovascular effects of TNF-�include myocardial depression, hypoten-

sion, and a decrease in systemic vascularresistance (16–18). In addition, TNF-�also induces cardiomyocyte apoptosis(19–22).

The goal of the present study was toinvestigate the role of TNF-� in myocar-dial dysfunction and apoptosis resultingfrom noncardiac injury. We used a mu-rine model of SIRS induced by bilateralhindlimb I/R. Previous studies haveshown that hindlimb I/R increases circu-lating TNF-� levels and causes remoteorgan injuries including lungs and liver(11, 23). We hypothesized that TNF-�production contributes to cardiac dys-function and myocardial apoptosis fol-lowing hindlimb I/R. We showed thathindlimb I/R injury caused myocardialdysfunction only in the presence ofTNF-� and that animals lacking TNF-� ortreated with a TNF-� blocker were pro-tected. Furthermore, the myocardial dys-function seen following hindlimb I/R wastemporally correlated with increasedmyocardial apoptosis, indicating that ap-

Objective: Peripheral vascular surgery involving limb ischemia/reperfusion is associated with tumor necrosis factor-� productionand an increased risk of cardiac complications. The objective of thisstudy was to investigate the role of tumor necrosis factor-� inmyocardial apoptosis and dysfunction following hindlimb ischemia/reperfusion.

Design: Randomized perspective animal study.Setting: Research laboratory.Subjects: Adults male tumor necrosis factor-��/� and litter-

mate wild-type mice.Interventions: Bilateral hindlimb ischemia/reperfusion was in-

duced in wild-type and tumor necrosis factor-��/� mice usingtourniquet occlusion. After 2 hrs of hindlimb ischemia, the tour-niquets were released, allowing reperfusion for 0.5–24 hrs.

Measurements and Main Results: In wild-type mice, hindlimbischemia/reperfusion resulted in myocardial depression earlyduring the reperfusion period (p < .05). These effects weretemporally correlated with enhanced levels of myocardial andplasma tumor necrosis factor-�. All variables were restored tobaseline levels by 24 hrs of reperfusion. Myocardial apoptosis,

assessed by cell death enzyme-linked immunosorbent assay,terminal deoxynucleotidyl transferase-mediated biotin-dUTPnick-end labeling staining, and caspase-3 activity, was also sig-nificantly higher at 6 hrs of reperfusion (p < .05) but returned tobaseline levels by 24 hrs. Interestingly, cardiac dysfunction andmyocardial apoptosis were abolished in tumor necrosis factor-��/� mice subjected to the same degree of hindlimb ischemia/reperfusion as the wild-type mice. Treatment of etanercept re-stored cardiac function in wild-type mice.

Conclusions: Tumor necrosis factor-� contributes significantlyto myocardial dysfunction and apoptosis in hindlimb ischemia/reperfusion. Although a causal link between myocardial apoptosisand cardiac dysfunction is not established, our study does sug-gest that tumor necrosis factor-� may be a potential therapeutictarget for cardiac injury in clinical situations involving prolongedremote ischemia/reperfusion. (Crit Care Med 2006; 34:484–491)

KEY WORDS: hindlimb ischemia and reperfusion; cardiac dys-function; tumor necrosis factor-�; myocardial apoptosis; sys-temic inflammatory response syndrome

From the Cardiology Research Lab (XL, JAH, JS, TP, QF)and Centre for Critical Illness Research (XL, JAH, JS, TP, RFP,QF), Lawson Health Research Institute, London Health Sci-ences Centre, Victoria Hospital; and the Departments ofMedicine (DLJ, JMOA, QF), Biophysics (RFP), and Physiologyand Pharmacology (JAH, JS, TP, DLJ, JMOA, QF), Universityof Western Ontario, London, Ontario, Canada.

Supported, in part, by grant MGP-14653 from theCanadian Institutes of Health Research (QF) and grantT-4923 from the Heart & Stroke Foundation of Ontario(QF). Dr. Feng is a HSFO Career Investigator and arecipient of a Premier’s Research Excellence Award

(PREA) from the Province of Ontario.Xiangru Lu, Joel A. Hamilton, and Ji Shen contrib-

uted equally to this work.There are no conflicts of interest to disclose.Address requests for reprints to: Dr. Qingping

Feng, Cardiology Research Lab, London Health Sci-ences Centre, Victoria Hospital, 800 CommissionersRoad, London, Ontario, Canada N6A 4G5.

Copyright © 2006 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000199079.64231.C1

484 Crit Care Med 2006 Vol. 34, No. 2

optosis may in part contribute to the de-velopment of cardiac dysfunction posthindlimb I/R.

METHODS

Animals. Breeding pairs of TNF-��/� andC57BL/6 mice were purchased from the Jack-son Laboratory. TNF-��/� mice were back-crossed to C57BL/6 background for at least sixgenerations. Male adult TNF-��/� and wild-type littermates (age 2–5 months) were usedfor the experiments. All mice were genotypedby polymerase chain reaction using genomicDNA extracted from the tail. All animals weremaintained on normal mouse chow and givenwater ad libitum in a 12/12-hr light-dark cy-cle. Animals used in this study were handledin accordance with the guidelines of the Ca-nadian Council on Animal Care, and studyprotocols were approved by the Animal UseSubcommittee at the University of WesternOntario, London, Canada.

Hindlimb Ischemia and Reperfusion. TNF-��/� and wild-type mice were anesthetized bysodium pentobarbital injection (50 mg/kg in-traperitoneally) and were placed on a heatingpad to maintain body temperature at 37°Cthroughout the procedure. Ischemia wasachieved using a tourniquet (#2 silk suture)above the greater trochanter of each hindlimb.After 2 hrs of ischemia, the tourniquets werereleased to allow hindlimb reperfusion. Anal-gesic (buprenorphine, 0.03 mg/kg subcutane-ously) was administered along with 1.5 mL ofsaline for fluid resuscitation. Reperfusion ofthe hindlimbs occurred for various periods oftime (0.5, 3, 6, 12, 18, and 24 hrs). Shamligations were used as controls, where animalsreceived similar anesthetic treatment to I/Ranimals but no hindlimb I/R was induced.Some wild-type mice were treated with etan-ercept (2 mg/kg intraperitoneally), a TNF-�blocking protein containing the extracellularbinding portion of the p75 TNF-� receptor(recombinant human p75 receptor Fc fusionprotein, TNFR/Fc) immediately after induc-tion of hindlimb ischemia. The dose of etan-ercept was chosen from a previous study thatshowed improvements in cardiac function inendotoxemic mice (24).

In Vivo Measurements of Cardiac Func-tion. Following reperfusion of various times,mice were reanesthetized, intubated, and me-chanically ventilated (SAR830, CWE, Ard-more, PA) with room air for catheter place-ments. A Millar tip pressure transducer (1.4-Fr, Millar Instruments) was inserted into theright carotid artery to obtain mean arterialblood pressure and heart rate. The catheterwas further advanced to the left ventricle formeasurement of left ventricular systolic pres-sure, left ventricular end-diastolic pressure,maximal rate of pressure development (�dP/dtmax), and maximal rate of relaxation of leftventricle (�dP/dtmax). All animals were kepton a proportional heating pad throughout theexperiment to maintain the body temperature

at 37°C. All measurements were recorded us-ing a Gould recorder (2400S). After the mea-surement of hemodynamic variables was com-plete, whole blood samples, left ventricle (LV)tissue, and skeletal muscle tissue were takenand stored at �70°C before further studies.

Isolated Heart Preparation. Mouse heartswere isolated and perfused in a Langendorff-system with Krebs-Henseleit buffer at 2 mL/min constant flow. The perfusion buffer wasmaintained at 37°C and bubbled continuouslywith a mixture of 95% oxygen and 5% CO2.Myocardial function was assessed according toour recent report (25). Briefly, a 6–0 silksuture was placed through the apex of the leftventricle and threaded through a lightweightrigid coupling rod, which was connected to aforce-displacement transducer (FT03) torecord tension and heart rate. The heart workwas calculated by multiplying the force (g) bythe heart rate (beats/min). Maximal and min-imal first derivative of force (�dF/dtmax and�dF/dtmin) as the rate of contraction and re-laxation were analyzed by PowerLab Chartprogram (ADInstruments, Mountain View,CA).

Enzyme Immunoassay for CytoplasmicHistone-Associated DNA Fragments. Apopto-sis was quantified by measuring cytoplasmichistone-associated DNA fragments (mono- andoligonucleotides) using a photometric enzymeimmunoassay (Cell Death Detection ELISA,Roche Molecular Biochemicals, Indianapolis,IN) according to our previous report (26).

In Situ Detection of Apoptotic Cells. Todetermine the cell type undergoing apoptosis,apoptotic cells in the myocardium were iden-tified using a terminal deoxynucleotidyl trans-ferase (TdT)-mediated dUTP nick end labeling(TUNEL), which incorporates fluorescein intothe DNA strand breaks (Roche Molecular Bio-chemicals) (26). The signal of TUNEL wasthen detected by an antifluorescein antibodyconjugated with peroxidase using diamino-benzidine as a chromogen. Sections werecounterstained with hematoxylin. The num-ber of apoptotic nuclei was quantified by twoblinded observers as we previously described(27).

Caspase-3 Activity. Caspase-3 activity wasmeasured using a caspase-3 fluorescent assaykit according to the manufacturer’s protocol(BIOMOL Research Laboratories, PA). This as-say is based on fluorescent intensity of thechromophore 7-amino-4-methylcoumarin af-ter cleavage from the C-terminus of the pep-tide substrate (26). Caspase-3 activity wasquantified using a fluorescent spectrophotom-eter (excitation at 355 nm, emission at 460nm) and normalized using inhibitor-treatedsamples as background. Caspase-3 activity wasexpressed as nmol 7-amino-4-methylcouma-rin cleaved per mg sample protein per hour.

Measurement of TNF-�. TNF-� levels inthe left ventricle and present in the blood weremeasured using a solid-phase sandwich en-zyme-linked immunosorbent assay (BiosourceInternational, Camarillo, CA) as we previously

described (24). For the measurement of myo-cardial TNF-�, excised LV tissue was mechan-ically homogenized for 1 min in ice-cold phos-phate-buffered saline. For measurement ofcirculating TNF-�, 50 �L of plasma was addedto the microtiter plates. The amount of TNF-�in each sample was determined from a linear-ized standard curve and was expressed asng/mg protein and pg/mL for LV tissues andplasma, respectively.

Immunohistochemistry. Sections of for-malin-fixed and paraffin-embedded myocardialtissue were analyzed by an indirect immuno-peroxidase technique (27). Briefly, 5-�m sec-tions were incubated with goat anti-TNF-�antibody (Santa Cruz Technology, Santa Cruz,CA) overnight at 4°C and then with peroxi-dase-conjugated donkey anti-goat immuno-globulin G antibody. The reaction was devel-oped with diaminobenzidine and H2O2.Sections were counterstained with hematoxy-lin, dehydrated, and mounted by routinemethods.

Lactate Dehydrogenase (LDH) Assay. LDHis a constitutively expressed enzyme present inmuscle tissue, which on cell membrane rup-ture is released into the circulation. For de-termination of skeletal leg muscle tissue dam-age, we measured residual LDH enzymaticactivity (units/�g protein) in leg muscle sam-ples using a photometric enzyme assay (SigmaChemical, St. Louis, MO) as previously de-scribed (28). Briefly, leg muscle samples werehomogenized. After protein concentrationswere determined using the Bradford assay(29), samples (2 �g protein) were reacted withexcess LDH substrates (i.e., pyruvate, reducednicotinamide adenine dinucleotide, and H�)and changes in absorbance were measured at340 nm using a spectrophotometer.

Western Blot Analysis. Cardiac tissue sam-ples (30 �g protein) were subjected to sodiumdodecyl sulfate polyacrylamide gel electro-phoresis using 15% gels, followed by electro-transfer to nitrocellulose membranes. Bothintact and cleaved troponin-I bands were de-termined similarly to a recent report (30) byprobing the blots using goat anti-troponin-Iantibody (sc-8118, Santa Cruz Technology,1:1000), followed by enhanced chemilumines-cence detection.

Data Analysis. Data are expressed as mean� SEM. All graphs were generated using theGraphPad Prism computer program (San Di-ego, CA). Data were analyzed using analysis ofvariance followed by Student-Newman-Keuls’test. We considered p � .05 as statisticallysignificant.

RESULTS

Assessment of Cardiac Function. Car-diac function was assessed in vivo in atotal of 92 animals (53 wild-type [WT]and 39 TNF-��/� mice). LV �dP/dtmaxwas significantly lower in WT mice after 2hrs of hindlimb ischemia followed by

485Crit Care Med 2006 Vol. 34, No. 2

0.5–6 hrs of reperfusion (Table 1, p �.05). However, no significant changes ofLV �dP/dtmax were seen at any timepoints of reperfusion in TNF-��/� ani-mals (p � nonsignificant). As a result, LV�dP/dtmax was significantly increased at3 and 6 hrs of reperfusion in TNF-��/�

compared with WT mice (p � .01). Meanarterial blood pressure was significantlylower in WT animals during the first 6hrs of hindlimb reperfusion (Table 1, p �.05). By 12 hrs of hindlimb reperfusion,all hemodynamic variables were recov-ered to baseline levels. By contrast, nochanges in mean arterial blood pressurewere seen in TNF-��/� animals duringthe course of reperfusion.

To eliminate reflex influences andloading conditions during in vivo mea-surements, cardiac function was alsomeasured in isolated hearts (Table 2).Mice underwent 2 hrs of hindlimb isch-emia followed by 3 hrs of reperfusion.Consistent with in vivo measurements,the maximal rate of force development(�dF/dt) and heart work were signifi-cantly decreased at baseline in WT butnot in TNF-��/� mice. Response to a�1-adrenergic agonist dobutamine (2 �gadded to the perfusate) was significantlydecreased following hindlimb I/R in WTbut not in TNF-��/� mice. Treatmentwith a recombinant soluble TNF-� recep-tor, etanercept (2 mg/kg intraperitone-

ally), immediately after induction ofhindlimb ischemia significantly improvedbasal and dobutamine-stimulated cardiacfunction in WT (p � .05). Heart rate wasnot significantly altered in TNF-��/�,WT, or etanercept treatment group. Insummary, both in vivo and isolated heartmeasurements showed that cardiac func-tion was higher in TNF-��/� comparedwith WT mice, and inhibition of TNF-�restored cardiac function following hind-limb I/R in WT mice.

Assessment of Myocardial Apoptosis.The presence of apoptotic cell death inthe LV myocardium following hindlimbI/R was demonstrated by DNA fragmen-tation and caspase-3 activation, hall-

Table 1. In vivo assessment of cardiovascular function after 2 hrs of hindlimb ischemia followed by 0.5–24 hrs of reperfusion in wild-type (WT) and tumornecrosis factor (TNF)-��/� mice

Variables Genotype Sham 0.5 Hrs 3 Hrs 6 Hrs 12 Hrs 18 Hrs 24 Hrs

HR Wild-type 438 � 27 433 � 24 468 � 24 434 � 19 465 � 24 436 � 9 426 � 10TNF-��/� 421 � 8 410 � 19 452 � 8 444 � 12 432 � 10 416 � 10 430 � 13

MAP Wild-type 77 � 3 66 � 2 62 � 4a 67 � 3 70 � 3 67 � 2 66 � 3TNF-��/� 79 � 4 69 � 3 76 � 6b 72 � 3 76 � 4 76 � 4 71 � 4

LVEDP Wild-type 4.8 � 0.6 5.8 � 0.7 5.2 � 0.5 5.6 � 0.5 4.3 � 0.6 6.4 � 0.5 4.8 � 0.6TNF-��/� 5.1 � 0.8 6.1 � 0.8 5.1 � 0.6 5.7 � 0.9 5.6 � 1.8 6.5 � 1.5 3.1 � 1.2

LVSP Wild-type 97 � 3 87 � 2 89 � 4 90 � 3 94 � 2 94 � 3 90 � 2TNF-��/� 104 � 5 94 � 4 104 � 6b 98 � 4 97 � 1 110 � 9c 113 � 3b

LV �dP/dtmax Wild-type 7035 � 269 4968 � 145a 5050 � 392a 5027 � 199a 6531 � 265 6211 � 263 5683 � 358

TNF-��/� 6621 � 482 6050 � 254 6600 � 485b 6550 � 278c 6050 � 413 6300 � 713 6750 � 712LV �dP/dtmin Wild-type 6785 � 258 5031 � 99

d 5100 � 405a 4944 � 355d 6250 � 183 5846 � 237 5733 � 325TNF-��/� 6786 � 594 6050 � 320 6900 � 367b 6350 � 145c 6300 � 496 6650 � 840 6750 � 523

HR, heart rate (beats/min); MAP, mean arterial pressure (mm Hg); LVEDP, left ventricular end diastolic pressure (mm Hg); LVSP, left ventricularsystolic pressure (mm Hg); LV �dP/dtmax, maximum rate of left ventricular pressure development (mm Hg/sec); �dP/dtmin, maximum rate of leftventricular relaxation (mm Hg/sec).

ap � .01 vs. corresponding sham; bp � .01 vs. wild-type; cp � .05 vs. wild-type; dp � .05 vs. corresponding sham. Results are expressed as mean � SEM.n � 5–7 and 7–8 for each time point in TNF-��/� and wild-type, respectively.

Table 2. Ex vivo assessment of cardiac function using an isolated heart preparation after 2 hrs of hindlimb ischemia followed by 3 hrs of reperfusion (I/R)in wild-type mice (WT, n � 7), tumor necrosis factor (TNF)-��/� mice (n � 6), and WT mice treated with etanercept (n � 6)

Parameters Genotype Sham I/RSham 2 �gDobutamine

I/R 2 �gDobutamine

HR, beats/min WT 354 � 25 329 � 26 460 � 20 424 � 34TNF-��/� 375 � 23 337 � 24 458 � 15 476 � 36WT � etanercept 361 � 14 480 � 42

Force, g WT 3.8 � 0.35 2.9 � 0.28 5.3 � 0.41 3.7 � 0.47TNF-��/� 3.5 � 0.33 4.0 � 0.27a 4.4 � 0.29 5.2 � 0.25a

WT � etanercept 3.63 � 0.28 5.4 � 0.2a

�dF/dtmax, g/sec WT 99 � 9 58 � 5b 205 � 24 109 � 16c

TNF-��/� 99 � 11 94 � 14a 161 � 18 191 � 20d

WT � etanercept 82 � 6a 166 � 12a

�dF/dtmin, g/sec WT 107 � 9 77 � 4c 186 � 16 126 � 16c

TNF-��/� 106 � 12 105 � 13 156 � 13 191 � 18a

WT � etanercept 93 � 5a 160 � 9Heart work, g � beats/min WT 1356 � 184 916 � 74 2443 � 227 1535 � 213c

TNF-��/� 1322 � 185 1352 � 151a 2034 � 154 2492 � 256d

WT � etanercept 1300 � 94a 2609 � 293a

HR, heart rate; �dF/dtmax, maximum rate of force development; �dF/dtmin, maximum rate of relaxation.ap � .05 vs. WT; bp � .01 vs. corresponding sham; cp � .05 vs. corresponding sham; dp � .01 vs. WT. Dobutamine (2 �g) was added to the perfusate

to measure contractile response to dobutamine stimulation. Etanercept (2 mg/kg intraperitoneally) was administered immediately after hindlimb ischemia.Results are expressed as mean � SEM.

486 Crit Care Med 2006 Vol. 34, No. 2

marks of apoptosis. First, the cytosolicDNA fragments in the myocardium weredetermined using cell death detection en-zyme-linked immunosorbent assay. Cyto-solic DNA fragments were significantlyincreased at 3 and 6 hrs of reperfusion inWT mice (Fig. 1, p � .05) and graduallyreturned to control levels after 12 hrs ofreperfusion. By contrast, animals lackingTNF-� showed no significant increase inthe level of myocardial DNA fragmenta-tion at any time points following hind-limb reperfusion (p � nonsignificant).

To determine the cell type (myocytevs. nonmyocyte) in the myocardium un-dergoing apoptosis, TUNEL staining wasused. Myocytes were identified by theirdistinct morphology under bright field.Since the TUNEL signal was diaminoben-zidine, which was visualized under brightfield, troponin-I double staining was notneeded. Representative images are shownin Figure 2A. Very few TUNEL-positivenuclei were observed in sham-operatedanimals. After 2 hrs of hindlimb ischemiafollowed by 6 hrs of reperfusion, WT miceshowed a significant increase in TUNEL-positive nuclei compared with sham con-trols (p � .05, Fig. 2B). The number of

TUNEL-positive nuclei was significantlydecreased in TNF-��/� compared withWT mice (p � .05, Fig. 2B). Almost all ofthe TUNEL-positive nuclei observed werewithin cardiac myocytes, suggesting thatmyocytes are the primary cell type under-going apoptosis in the LV myocardiumduring hindlimb I/R.

Further confirmation of the presenceof apoptosis was provided through thedetermination of caspase-3 activity.Caspases represent a family of aspartate-associated proteins that are involved innuclear DNA cleavage. Caspase-3 activitywas significantly increased in the LV

myocardium of WT animals after 6 hrs ofhindlimb reperfusion vs. sham controls(p � .05). Furthermore, myocardialcaspase-3 activity after 6 hrs of hindlimbreperfusion was significantly decreased inTNF-��/� compared with WT mice (p �.05, Fig. 3).

Myocardial and Plasma TNF-� Levels.The levels of TNF-� in the LV myocar-dium and plasma were investigated in anattempt to correlate levels of TNF-� tothe degree of myocardial dysfunction fol-lowing hindlimb I/R. Figure 4A showsthat plasma TNF-� levels peaked at 30mins of hindlimb reperfusion in WT mice

Figure 1. Accumulation of cytoplasmic DNA frag-ments in the left ventricular (LV) myocardiumduring hindlimb ischemia and reperfusion inwild-type (WT) and tumor necrosis factor (TNF)-��/� mice. Animals were subjected to 2 hrs ofhindlimb ischemia followed by 0.5–24 hrs ofreperfusion. Sham animals (time 0) underwentanesthesia but without hindlimb ischemia/reperfusion. Cytoplasmic DNA fragments fromLV myocardium were determined by cell deathdetection enzyme-linked immunosorbent assay,and percent change of control is shown followingvarious times of reperfusion. Cytoplasmic DNAfragments in the LV myocardium were signifi-cantly increased after 3 and 6 hrs of reperfusionin WT mice (*p � .05, **p � .01 vs. sham,respectively). Myocardial DNA fragments weresignificantly decreased in TNF-��/� comparedwith WT mice (†p � .05). n � 5–7 and 7–8 foreach time point in TNF-��/� and WT mice, re-spectively.

Figure 2. Myocardial apoptosis as measured by terminal deoxynucleotidyl transferase-mediated biotin-dUTP nick-end labeling (TUNEL) staining. Animals were subjected to sham operation or 2 hrs ofhindlimb ischemia followed by 6 hrs of reperfusion (I/R) in wild-type (WT) and tumor necrosis factor(TNF)-��/� mice. TUNEL staining was performed on paraffin-embedded heart sections. A, represen-tative images (original magnification 400). Sham-operated mice showed very few TUNEL-positivenuclei in WT and TNF-��/� mice. A marked increase in the number of TUNEL-positive nuclei (arrows)was evident in the left ventricle (LV) myocardium of WT compared with that of TNF-��/� mice. B,quantitative analysis of myocyte apoptosis. Apoptotic nuclei in the LV myocardium were significantlyincreased after I/R compared with sham operation in WT mice (**p � .01). Apoptotic nuclei after I/Rwere significantly decreased in TNF-��/� compared with WT mice (†p � .05). n � 4 per group.

487Crit Care Med 2006 Vol. 34, No. 2

(p � .05) and returned to control levelsby 3 hrs of reperfusion. Interestingly,myocardial TNF-� levels also signifi-cantly increased at 30 mins of hindlimbreperfusion relative to sham levels (p �.01). However, these levels persisted over12 hrs following reperfusion (Fig. 4B).Immunohistochemical staining con-firmed myocardial TNF-� production.Following 0.5, 3, and 6 hrs of hindlimbreperfusion, TNF-� staining was mark-edly increased in the myocardium of WTmice and mainly present in cardiomyo-cytes (Fig. 5, C, E, and G). No TNF-�staining was present in TNF-��/� hearts(Fig. 5, D, F, and H).

Skeletal Muscle Injury. The level ofskeletal leg muscle injury was investi-gated to determine whether there weredifferences between TNF-��/� and WTmice. Wet/dry ratio, which measures tis-sue edema, was significantly increased inthe leg muscle following hindlimb I/R,peaking at the 6-hr time point by abouttwo-fold compared with sham controls(Fig. 6). This response was similar be-tween TNF-��/� and WT mice (p � non-significant). Skeletal leg muscle injurywas further assessed by measuring theactivity of LDH, a constitutively ex-pressed enzyme present in muscle tissue(Fig. 7A). As expected, LDH activity levelsin the tissue were high in sham animalsand significantly decreased after 6 hrs ofreperfusion (p � .05). However, therewas no significant difference betweenTNF-��/� and WT mice (p � nonsignif-icant). Thus, the extent of skeletal muscleinjury, as measured by both wet/dry ratioand residual LDH activity, was similarthroughout the time course of reperfu-sion for both TNF-��/� and WT animals.

Cardiac Injury. To verify cardiac in-jury, both TNF-��/� and WT mice were

subjected to 2 hrs of hindlimb ischemiafollowed by 3 hrs of reperfusion. Tropo-nin-I cleavage in the LV myocardium, anindex of cardiac injury, was clearly seenin WT mice but barely detectable in TNF-��/� mice (Fig. 7B). There was no tropo-nin-I cleavage in sham animals. An LVsample from a WT mouse 30 days aftermyocardial infarction served as a positivecontrol (31). Our data showed that car-diac injury following hindlimb I/R wasattenuated in TNF-��/� mice.

DISCUSSION

In the present study, we demonstratedthat hindlimb I/R induced myocardialTNF-� expression, cardiomyocyte apo-ptosis, and cardiac dysfunction in WTmice. Interestingly, these changes wereabsent or significantly decreased in TNF-��/� mice with the same degree of hind-limb skeletal muscle injury as the WTmice. Our study suggests that TNF-�plays an important role in myocardialdysfunction and apoptosis during hind-limb I/R.

TNF-� has been implicated as a pri-mary mediator of various cardiac pathol-ogies, including heart failure, myocardialI/R, and sepsis. In the present study, weshowed that hindlimb I/R induced a re-versible reduction in cardiac functionthat coincided with a time-dependent in-crease in plasma and myocardial TNF-�levels. The sharp peak of TNF-� levels inthe plasma occurring 30 mins after hind-limb reperfusion has been observed inprevious studies (32, 33). These studiesindicate that the plasma TNF-� spikecomes from the TNF-� made duringhindlimb ischemia and flushed out afterreperfusion (32, 33). he fact that theelevated systemic TNF-� levels were notmaintained beyond 3 hrs of hindlimbreperfusion suggests that TNF-� iscleared rapidly from the circulation. In-terestingly, myocardial TNF-� levels alsopeaked at 30 mins but persisted over 12hrs following hindlimb reperfusion. Onepotential explanation for the rapid reduc-tion in systemic TNF-� levels is thatTNF-� binds to its receptors in organsand tissues including the heart, thus re-moving it from circulation. The rapid in-crease in plasma levels of TNF-� maytrigger a sustained local production ofTNF-� in the heart, possibly via reactiveoxygen species (34). In fact, cardiomyo-cytes are the major site of TNF-� produc-tion in the heart following hindlimb I/Ras demonstrated by our immunohisto-chemical staining.

Apoptosis is a highly regulated physi-ologic process that is crucial to tissuedevelopment and/or remodeling. How-ever, excessive apoptosis has been shownto contribute to various pathologies in-cluding sepsis and heart failure (35–37).In the present study, we provided strongevidence for an increased myocardial ap-optosis following hindlimb I/R in WTmice. Myocardial levels of cytoplasmic hi-stone-associated DNA fragments andcaspase-3 activity were significantly in-creased in WT mice subjected to hind-limb I/R. We also observed TUNEL-positive staining mainly present incardiomyocytes. Taken together, all threemeasures of apoptosis indicated thathindlimb I/R induced a significant in-crease in myocardial apoptosis in WTmice. Interestingly, myocardial apoptosiswas almost abrogated in TNF-��/� mice,suggesting that TNF-� plays a criticalrole in cardiomyocyte apoptosis followinghindlimb I/R.

It is well known that TNF-� impairsmyocardial function and has been impli-

Figure 3. Myocardial caspase-3 activity after 2 hrsof hindlimb ischemia followed by 6 hrs of reper-fusion (I/R) in wild-type (WT) and tumor necrosisfactor (TNF)-��/� mice. Hindlimb I/R signifi-cantly increased myocardial caspase-3 activitycompared with sham operation in WT mice (*p �.05). However, myocardial caspase-3 activity wassignificantly decreased in TNF-��/� comparedwith WT mice (†p � .05). n � 5–7 per group.

Figure 4. Effects of hindlimb ischemia and reper-fusion (I/R) on plasma (A) and myocardial (B)tumor necrosis factor (TNF)-� levels in wild-type(WT) mice. Animals were subjected to 2 hrs ofischemia followed by 0.5–24 hrs of reperfusion orsham operation (time 0). Plasma and myocardialTNF-� levels were determined by enzyme-linkedimmunosorbent assay. *p � .05 vs. sham, **p �.01 vs. sham, n � 4–7 for each time point pergroup.

488 Crit Care Med 2006 Vol. 34, No. 2

cated in clinical pathologies includingheart failure and sepsis (16, 17). In thepresent study, we demonstrated thatmyocardial TNF-� expression was mark-edly increased and was mainly producedin cardiomyocytes following hindlimbI/R. Increased myocardial TNF-� expres-sion was associated with decreased car-diac function in WT mice. Decreased car-diac function was demonstrated in bothin vivo whole animal models and isolatedhearts. To study the specific role ofTNF-� in cardiac dysfunction during

hindlimb I/R, we employed TNF-��/�

mice. We showed that cardiac function inboth in vivo and isolated heart prepara-tions was well preserved in TNF-��/�

mice. Furthermore, inhibition of TNF-�by etanercept restored cardiac function inWT mice. These results suggest an im-portant role of TNF-� in cardiac dysfunc-tion post hindlimb I/R.

In the present study, skeletal muscleinjury following hindlimb I/R was similarbetween TNF-��/� and WT mice as de-termined by wet/dry ratio and residual

LDH activity. However, cleavage of car-diac troponin-I was barely detectable inTNF-��/� mice. The fact that TNF-��/�

mice had much less cardiac troponin-Icleavage but similar skeletal muscle inju-ries with the WT further supports thenotion that TNF-� is pivotal in the devel-opment of myocardial apoptosis and dys-function post hindlimb I/R.

CONCLUSIONS

Hindlimb I/R induced myocardialTNF-� expression, cardiomyocyte apopto-sis, and cardiac dysfunction in WT mice.Deficiency in TNF-� protected the myocar-dium from apoptosis and improved cardiacfunction during the same degree of hind-limb injury. We conclude that TNF-� con-tributes significantly to myocardial dys-function and apoptosis in hindlimb I/R.Although a causal link between myocardialapoptosis and cardiac dysfunction is notestablished, our study does suggest thatTNF-� may be a potential therapeutic tar-

Figure 5. Immunohistochemical staining of tumor necrosis factor (TNF)-� in the left ventricle (LV)myocardium after 2 hrs of ischemia followed by 0.5, 3, and 6 hrs of reperfusion (I/R) in wild-type (WT)and TNF-��/� mice. Sham and TNF-��/� mice showed no TNF-� staining (A, B, D, F, H). MarkedTNF-� staining was seen in the myocardium after 0.5, 3, and 6 hrs of reperfusion in WT mice (C, E,G). Shown are representatives of three independent experiments. (Original magnification 500).

Figure 6. Changes of skeletal muscle wet/dry ra-tio after 2 hrs of hindlimb ischemia followed by0.5–24 hrs of reperfusion in wild-type (WT) andtumor necrosis factor (TNF)-��/� mice. Follow-ing reperfusion, the wet/dry ratio was signifi-cantly increased in both WT and TNF-��/� mice.However, there was no statistical difference be-tween the two groups. *p � .05, **p � .01, ***p� .001 vs. corresponding sham (time 0), respec-tively. n � 4–7 for each time point per group.

T umor necrosis fac-tor-� contributessignificantly tomyocardial dysfunction and

apoptosis in hindlimb

ischemia/reperfusion.

489Crit Care Med 2006 Vol. 34, No. 2

get for cardiac injury in clinical situationsinvolving prolonged remote I/R.

REFERENCES

1. Davies MG, Hagen PO: Systemic inflamma-tory response syndrome. Br J Surg 1997;84:920–935

2. Beal AL, Cerra FB: Multiple organ failuresyndrome in the 1990s: Systemic inflamma-tory response and organ dysfunction. JAMA1994; 271:226–233

3. Froon AH, Greve JW, Van der Linden CJ, etal: Increased concentrations of cytokines and

adhesion molecules in patients after repair ofabdominal aortic aneurysm. Eur J Surg1996; 162:287–296

4. Holzheimer RG, Gross J, Schein M: Pro- andanti-inflammatory cytokine-response in ab-dominal aortic aneurysm repair: A clinicalmodel of ischemia-reperfusion. Shock 1999;11:305–310

5. Juvonen J, Surcel HM, Satta J, et al: Elevatedcirculating levels of inflammatory cytokinesin patients with abdominal aortic aneurysm.Arterioscler Thromb Vasc Biol 1997; 17:2843–2847

6. Swartbol P, Norgren L, Parsson H, et al:

Endovascular abdominal aortic aneurysm re-pair induces significant alterations in surfaceadhesion molecule expression on donorwhite blood cells exposed to patient plasma.Eur J Vasc Endovasc Surg 1997; 14:48–59

7. Goldman L: Cardiac risk for vascular surgery.J Am Coll Cardiol 1996; 27:799–802

8. Baron JF, Mundler O, Bertrand M, et al:Dipyridamole-thallium scintigraphy andgated radionuclide angiography to assess car-diac risk before abdominal aortic surgery.N Engl J Med 1994; 330:663–669

9. Kumar A, Haery C, Parrillo JE: Myocardialdysfunction in septic shock. Crit Care Clin2000; 16:251–287

10. Shahani R, Marshall JG, Rubin BB, et al: Roleof TNF-alpha in myocardial dysfunction afterhemorrhagic shock and lower-torso isch-emia. Am J Physiol Heart Circ Physiol 2000;278:H942–H950

11. Seekamp A, Warren JS, Remick DG, et al:Requirements for tumor necrosis factor-alpha and interleukin-1 in limb ischemia/reperfusion injury and associated lung in-jury. Am J Pathol 1993; 143:453–463

12. White DJ, Maass DL, Sanders B, et al: Cardi-omyocyte intracellular calcium and cardiacdysfunction after burn trauma. Crit CareMed 2002; 30:14–22

13. Gurevitch J, Frolkis I, Yuhas Y, et al: Tumornecrosis factor-alpha is released from the iso-lated heart undergoing ischemia and reper-fusion. J Am Coll Cardiol 1996; 28:247–252

14. Kapadia S, Lee J, Torre-Amione G, et al:Tumor necrosis factor-alpha gene and pro-tein expression in adult feline myocardiumafter endotoxin administration. J Clin Invest1995; 96:1042–1052

15. Meldrum DR, Cleveland JC Jr, Cain BS, et al:Increased myocardial tumor necrosis factor-alpha in a crystalloid-perfused model of car-diac ischemia-reperfusion injury. Ann Tho-rac Surg 1998; 65:439–443

16. Kelly RA, Smith TW: Cytokines and cardiaccontractile function. Circulation 1997; 95:778–781

17. Meldrum DR: Tumor necrosis factor in theheart. Am J Physiol 1998; 274:R577–R595

18. Yokoyama T, Vaca L, Rossen RD, et al: Cel-lular basis for the negative inotropic effectsof tumor necrosis factor-alpha in the adultmammalian heart. J Clin Invest 1993; 92:2303–2312

19. Song W, Lu X, Feng Q: Tumor necrosis fac-tor-alpha induces apoptosis via inducible ni-tric oxide synthase in neonatal mouse cardi-omyocytes. Cardiovasc Res 2000; 45:595–602

20. Krown KA, Page MT, Nguyen C, et al: Tumornecrosis factor alpha-induced apoptosis incardiac myocytes. Involvement of the sphin-golipid signaling cascade in cardiac celldeath. J Clin Invest 1996; 98:2854–2865

21. Aikawa R, Nitta-Komatsubara Y, Kudoh S, etal: Reactive oxygen species induce cardiomy-ocyte apoptosis partly through TNF-alpha.Cytokine 2002; 18:179–183

22. Carlson DL, Willis MS, White DJ, et al: Tu-

Figure 7. Skeletal muscle and cardiac injury. A, changes of residual skeletal muscle lactate dehydro-genase (LDH) activity after 2 hrs of hindlimb ischemia followed by 0.5–24 hrs of reperfusion or shamoperation (time 0) in wild-type (WT) and tumor necrosis factor (TNF)-��/� mice. Following 6 hrs ofreperfusion, LDH activity was significantly decreased in both WT and TNF-��/� mice. However, therewas no statistical difference between the two groups. *p � .05 vs. corresponding sham. n � 4–7 foreach time point per group. B, representative Western blot analysis of cardiac troponin-I. Markedcleavage of troponin-I from the left ventricle (LV) myocardium was detected in WT mice after 2 hrs ofhindlimb ischemia followed by 3 hrs of reperfusion (I/R, lanes 2 and 3). The cleaved band was barelydetectable in TNF-��/� mice subjected to the same hindlimb I/R protocol (lanes 4 and 5). Thenoninfarcted LV myocardium from mice subjected to coronary artery ligation (MI) for 30 days servedas positive controls (lane 6). Each lane represents a sample from an individual mouse. n � 4 for eachI/R group. n � 2 for each sham group.

490 Crit Care Med 2006 Vol. 34, No. 2

mor necrosis factor-alpha-induced caspaseactivation mediates endotoxin-related car-diac dysfunction. Crit Care Med 2005; 33:1021–1028

23. Brock RW, Lawlor DK, Harris KA, et al: Ini-tiation of remote hepatic injury in the rat:Interactions between Kupffer cells, tumornecrosis factor-alpha, and microvascular per-fusion. Hepatology 1999; 30:137–142

24. Peng T, Lu X, Lei M, et al: Inhibition of p38MAPK decreases myocardial TNF-alpha ex-pression and improves myocardial functionand survival during acute endotoxemia inmice. Cardiovasc Res 2003; 59:893–900

25. Peng T, Lu X, Feng Q: Pivotal role ofgp91phox-containing NADH oxidase in lipopo-lysaccharide-induced tumor necrosis factor-alpha expression and myocardial depression.Circulation 2005; 111:1637–1644

26. Feng Q, Song W, Lu X, et al: Development ofheart failure and congenital septal defects inmice lacking endothelial nitric oxide syn-thase. Circulation 2002; 106:873–879

27. Moe GW, Naik G, Konig A, et al: Early andpersistent activation of myocardial apoptosis,

bax and caspases: Insights into mechanismsof progression of heart failure. Pathophysi-ology 2002; 8:183–192

28. Lazarus B, Messina A, Barker JE, et al: Therole of mast cells in ischaemia-reperfusioninjury in murine skeletal muscle. J Pathol2000; 191:443–448

29. Bradford MM: A rapid and sensitive methodfor the quantitation of microgram quantitiesof protein utilizing the principle of protein-dye binding. Anal Biochem 1976; 72:248–254

30. van der Velden J, Merkus D, Klarenbeek BR,et al: Alterations in myofilament functioncontribute to left ventricular dysfunction inpigs early after myocardial infarction. CircRes 2004; 95:e85–e95

31. Feng Q, Lu X, Jones DL, et al: Increasedinducible nitric oxide synthase expressioncontributes to myocardial dysfunction andhigher mortality after myocardial infarctionin mice. Circulation 2001; 104:700–704

32. Lawlor DK, Brock RW, Harris KA, et al: Cy-tokines contribute to early hepatic parenchy-mal injury and microvascular dysfunction af-

ter bilateral hindlimb ischemia. J Vasc Surg1999; 30:533–541

33. Sternbergh WC III, Tuttle TM, Makhoul RG,et al: Postischemic extremities exhibit imme-diate release of tumor necrosis factor. J VascSurg 1994; 20:474–481

34. Li JM, Fan LM, Christie MR, et al: Acutetumor necrosis factor alpha signaling viaNADPH oxidase in microvascular endothelialcells: Role of p47phox phosphorylation andbinding to TRAF4. Mol Cell Biol 2005; 25:2320–2330

35. Fauvel H, Marchetti P, Chopin C, et al: Dif-ferential effects of caspase inhibitors on en-dotoxin-induced myocardial dysfunction andheart apoptosis. Am J Physiol Heart CircPhysiol 2001; 280:H1608–H1614

36. Olivetti G, Abbi R, Quaini F, et al: Apoptosisin the failing human heart. N Engl J Med1997; 336:1131–1141

37. Razavi HM, Hamilton JA, Feng Q: Modula-tion of apoptosis by nitric oxide: implicationsin myocardial ischemia and heart failure.Pharmacol Ther 2005; 106:147–162

491Crit Care Med 2006 Vol. 34, No. 2