Activation of AMP-activated protein kinase by humanEGF receptor 2/EGF receptor tyrosine kinaseinhibitor protects cardiac cellsNeil L. Spector*, Yosef Yarden†, Bradley Smith‡, Ljuba Lyass§, Patricia Trusk§, Karen Pry§, Jason E. Hill§, Wenle Xia*,Rony Seger†, and Sarah S. Bacus§¶

*Duke Comprehensive Cancer Center, Duke University Medical Center, 106 Research Drive, Medical Science Research Building II, Durham, NC 27710;†Department of Biological Regulation, Weizmann Institute, Room 302, Candiotty Building, 1 Hertzl Street, 76100 Rehovot, Israel; ‡Cell SignalingTechnologies, 166B Cummings Center, Beverly, MA 01945; and §Targeted Molecular Diagnostics, 610 Oakmont Lane, Westmont, IL 60559

Edited by Michael Sela, Weizmann Institute of Science, Rehovot, Israel, and approved May 8, 2007 (received for review February 12, 2007)

The human EGF receptor (HER) 2 receptor tyrosine kinase is asurvival factor for human cardiomyocytes, and its inhibition mayexplain the increased incidence of cardiomyopathy associated withthe anti-HER2 monoclonal antibody trastuzumab (Genentech,South San Francisco, CA), particularly in patients with prior expo-sure to cardiotoxic chemotherapies e.g., anthracyclines. Here, weshow that GW2974 (HER2/EGF receptor tyrosine kinase inhibitor),but not trastuzumab, activates AMP-activated protein kinase(AMPK), initiating a metabolic stress response in human cardio-myocytes that protects against TNF�-induced cell death. GW2974stimulates calcium dependent fatty acid oxidation in vitro and inthe myocardium of GW2974-treated rodents. Calcium chelation orsiRNA-targeted AMPK knockdown blocks GW2974 induced fattyacid oxidation. In addition, inhibition of AMPK by a specificinhibitor resulted in increased killing of cardiomyocytes. Elucidat-ing the effects of HER2-targeted therapies on AMPK may predictfor risk of cardiomyopathy and provide a novel HER2-targetedstrategy designed to protect myocardium from the pro-apoptoticeffects of pro-inflammatory cytokines released in response tocardiac injury by chemotherapy or acute ischemia.

monoclonal antibodies

The human EGF receptor (HER) 2 receptor tyrosine kinaseplays an essential role in cardiac development during embryo-

genesis and as a survival factor in adult myocardium (1–4), asevidenced by the increased incidence of cardiomyopathy associatedwith trastuzumab (Herceptin), a humanized anti-HER2 monoclo-nal antibody used to treat HER2-overexpressing breast cancers (5,6). Although inhibition of HER2 causes mitochondrial dysfunctionin cardiomyocytes (3, 7, 8), the exact mechanism responsible fortrastuzumab-induced cardiomyopathy is unknown. Other HER2-targeted therapies, with different mechanisms of action, do notappear to have the same risk of cardiomyopathy. For example,lapatinib (GW572016) (GlaxoSmithKline, King of Prussia, PA), asmall molecule inhibitor of the HER2 and EGF receptor (EGFR)tyrosine kinases, is clinically efficacious in heavily pretreatedHER2-overexpressing breast cancers (9), yet it appears to have lessrisk of cardiomyopathy compared with trastuzumab (10). Interest-ingly, a non-HER receptor tyrosine kinase inhibitor, imatinib(Novartis, East Hanover, NJ), which targets members of the c-ablfamily, was recently shown to induce heart toxicity (11–13), raisingconcerns over the cardiac safety of other small molecule tyrosinekinases.

Under normal physiologic conditions, the adult myocardiumutilizes fatty acid or glucose oxidation as its main energy source,with fatty acid oxidation transcriptionally regulated by members ofthe nuclear receptor superfamily, including the peroxisome prolif-erator-activated receptors (PPARs) and their coactivators, PPAR�coactivators-1� (PGC-1�) (14–17). Orphan nuclear receptors havealso been implicated in regulating cardiac energy metabolism,including the three members of the estrogen-related receptor

(ERR) family ERR�, ERR�, and ERR� (18). In adults, ERR� andERR� expression is enriched in tissues that rely primarily onmitochondrial oxidative metabolism for energy, such as heart,brown adipose, and slow-twitch skeletal muscle. ERR� likely playsa role in lipid catabolism in heart, consistent with its functionalinteraction with PGC-1�, making it a critical regulator of energymetabolism (18–24).

Under stress conditions, such as hypoxia, ischemia, glucosedeprivation, and starvation, an increase in the intracellularAMP:ATP ratio allosterically activates AMP-activated proteinkinase (AMPK), a response designed to maintain cellular energybalance (23, 25). AMPK was initially discovered as an activitythat inhibited preparations of acetyl-CoA carboxylase (ACC)and 3-hydroxy-3-methylglutaryl-CoA reductase. Activation ofAMPK initiates a series of downstream phosphorylation eventsthat switch cells from active ATP consumption (e.g., fatty acid,cholesterol, and protein biosynthesis) to ATP production (e.g.,fatty acid and glucose oxidation) (26, 27). Stress-induced acti-vation of AMPK occurs after its phosphorylation at threonine172 on the � subunit by one or more upstream AMPK kinases,including calmodulin-dependent kinase kinase �, a calcium-activated protein kinase (28), and LKB1, a serine/threoninekinase encoded by the Peutz–Jegher’s syndrome tumor suppres-sor gene (29–32). Activation of AMPK in skeletal muscle andheart leads to the phosphorylation and inhibition of ACC, whichin turn reduces the level of malonyl-CoA, itself an inhibitor ofcarnitine palmitoyltransferase l (CPT l). Derepression of CPT lresults in the concomitant increase in �-oxidation of fatty acid,leading to increased mitochondrial production of ATP (26, 27,33). Stress-induced activation of AMPK also inhibits proteinsynthesis by inhibition of mTOR and directly modulating elon-gation factor (eEF)2, a translation elongation factor known to beassociated with cardiac protection (31, 34–36). Importantly,alteration in mitochondrial function has been reported to lead tocardiomyocyte death by imatinib (11). Moreover, inhibition ofcap-dependent translation by AMPK-mediated TSC2 phosphor-ylation appears to be critical for cell survival in response to ATPdepletion (32). Increased biosynthesis of (rather than consump-

Author contributions: S.S.B. designed research; Y.Y., L.L., P.T., K.P., J.E.H., W.X., and R.S.performed research; B.S. contributed new reagents/analytic tools; N.L.S. and S.S.B. ana-lyzed data; and N.L.S. and S.S.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Abbreviations: ACC, acetyl-CoA carboxylase; AMPK, AMP-activated protein kinase; eEF,elongation factor; EGFR, epidermal growth factor receptor; ERR, estrogen-related recep-tor; HER, human epidermal growth factor receptor; HMCs, human myocardiocytes; PGC,peroxisome proliferator-activated receptors coactivator.

¶To whom correspondence should be addressed. E-mail: sbacus@tmdlab.com.

This article contains supporting information online at www.pnas.org/cgi/content/full/0701286104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0701286104 PNAS � June 19, 2007 � vol. 104 � no. 25 � 10607–10612

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020

tion of) ATP following AMPK activation may also protectcardiomyocytes against ischemic injury (36).

Here, we demonstrate that GW2974, a potent small moleculeHER2/EGFR tyrosine kinase inhibitor with a similar profile tolapatinib (37) [supporting information (SI) Table 1], activatesAMPK and its downstream substrates, and stimulates fatty acidoxidation, which in turn increases ATP production in HER2-expressing human cardiomyocytes, protecting against apoptosisinduced by TNF�, a known cytokine detected in cardiac failure(38). Conversely, trastuzumab does not activate AMPK. Instead, itleads to enhanced cardiomyocyte cell death in response to TNF�.The effects of specific HER2-targeted therapies on AMPK andconsequently energy production may predict for the risk associatedcardiomyopathy and provide a novel HER2-directed therapeuticstrategy to protect myocardium from the killing effects of TNF� orother proapoptotic stimuli, following acute ischemic injury.

ResultsGW2974 Modulates the Activity of Genes Involved in Regulating FattyAcid Metabolism and Calcium Channels in HER2-Expressing CancerCells and Human Myocardiocytes (HMCs). To better understand thecardiac sparing effects of HER2 kinase inhibitors like lapatinib andGW2974, we assessed the effects of GW2974 on gene expression inthe HER2-overexpressing Au565 breast cancer cell line, a cell linesensitive to HER2 kinase inhibition. We found that GW2974modulated the activity of a number of genes associated withcalcium/ion channels and fatty acid metabolism (SI Tables 2–4).The increased activity of genes involved in calcium-mediated ac-tivities and ion channels led us to investigate the effects of GW2974on intracellular calcium levels in HER2 expressing cells. HMCs,which express physiological levels of HER2 and Au565 cells, werepreloaded with Fluro-4 dye, treated with GW2974, and thenscanned by using fluorescent microscopy. Intracellular calciumincreased rapidly in GW2974-treated cells (Fig. 1).

GW2974 is an equipotent inhibitor of the HER2 and EGFRtyrosine kinases similar to lapatinib (37) (SI Table 1). To determinewhether the effect of GW2974 on calcium flux was HER2 orEGFR-dependent, we compared intracellular calcium levels inparental MCF7 breast cancer cells, which express EGFR and onlylow levels of HER2 and are resistant to the anti-tumor effects ofHER2 kinase inhibitors (39). We also studied MCF7/HER2 cells,a stably transfected MCF7 line that overexpresses HER2. Incontrast to parental MCF7 cells where intracellular calcium levelsessentially remained unchanged, treatment with GW2974 resulted

in a 5- to 8-fold increase in intracellular calcium in MCF7/HER2cells (SI Fig. 7). Increased intracellular calcium levels in responseto GW2974 appear to require HER2 expression, although a sup-portive role of EGFR or HER3 as heterodimeric partners to HER2cannot be excluded.

GW2974, but Not Trastuzumab, Activates AMPK and Its DownstreamSubstrates in Breast Cancer Cells and HMCs. GW2974 affected theactivity of genes involved in cellular metabolism, calcium/ion chan-nels, and calcium/calmodulin-dependent proteins (SI Tables 2–4).Furthermore, calmodulin-dependent kinase kinase �, a calcium-regulated serine/threonine kinase, activates AMPK, a master reg-ulator of cell metabolism and energy production that is induced bya variety of stress stimuli, including an increase in the AMP:ATPratio. We therefore postulated that GW2974 activates AMPK in acalcium-dependent manner and that AMPK activation protectsHER2 expressing HMCs from apoptotic stimuli.

To test this hypothesis, we compared the activation state ofAMPK and its downstream substrates in HMCs and Au565 cellstreated with either GW2974 or trastuzumab. Importantly, GW2974,but not trastuzumab, increased steady-state protein levels of acti-vated, phosphorylated AMPK in HMCs and Au565 cells (Fig. 2a).As a consequence of AMPK activation, eEF2 and ACC werephosphorylated in GW2974 but not in trastuzumab-treated cells.Inhibition of PI3K-Akt signaling is a potent activator of AMPK.However, AMPK was activated in GW2974-treated HMCs despitevery little change in the steady-state protein levels of phosphory-lated Akt (Fig. 2a) indicating that AMPK activation in response toGW2974 is Akt-independent. The lack of inhibition of phosphor-ylated Akt in GW2974-treated HMCs compared with Au565 cellsis consistent with the dependence of Akt activation on HER2signaling in the HER2-overexpressing Au565 cells compared withHMCs, which express physiological levels of HER2 and where Aktactivation is likely maintained by redundant survival pathways. Inaddition, we were able to show that survival of HMCs depends

Fig. 1. GW2974 stimulates increased intracellular calcium levels in GW2974-treated HMCs and Au565. HMCs and Au565 were preloaded with Fluro-4 dye andthen treated with 25 �M GW2974. Images were then taken by confocal micros-copy after treatment with GW2974. Cells treated with vehicle alone served ascontrols and baseline autofluorescence was assessed in untreated cells. Theseresults are representative of three independent experiments.

Fig. 2. GW2974 stimulates calcium-dependent activation of AMPK and itsdownstreamsubstrates that regulate fattyacidoxidationandprotein synthesis inresponse to HER2 kinase inhibition by GW2974. (a) HER2-positive HMCs and theHER2-overexpressingAu565breast cancercell lineweretreatedwithGW2974(25�M) or trastuzumab (50 �g/ml) for 15 min. Equal amounts of protein from totalcell lysate were separated by SDS/PAGE, and steady-state protein levels wereassessed by Western blot as indicated. Steady-state protein levels of phospho-AMPK� and its downstream substrates, ACC and eEF2, increased in GW2974 butnot trastuzumab-treated HMCs and Au565 cells. Activation of AMPK� in HMCsoccurred despite steady-state phosphorylated Akt (p-Akt) protein levels remain-ing unchanged. Actin steady-state protein levels served as a control for equalloading of protein. These results are representative of three independent exper-iments. The HMCs or Au565 cells were treated with or without GW2974 andlapatinib (25 �M or 50 �M) and were tested. (b) HMCs cells were treated with theintracellular calcium chelator BAPTA/AM (30 �M) for 30 min before addingGW2974 (25 �M) for an additional 15 min. Phospho-AMPK� steady protein levelswere then determined in cells treated with vehicle alone (lane 1), GW2974 (lane2), BAPTA/AM alone (lane 3), or BAPTA/AM and the GW2974 (lane 4). Pretreat-ment with BAPTA/AM blocked the induction of phosphorylated AMPK� (p-AMPK�) protein in GW2974-treated cells. Results are representative of threeindependent experiments.

10608 � www.pnas.org�cgi�doi�10.1073�pnas.0701286104 Spector et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020

AMPK activation as treatment with compound C, a specific inhib-itor of AMPK, induced cell death even in the presence of GW2974(SI Fig. 8). The activation of AMPK depended on HER2 in Au565but not in HMCs cells (SI Fig. 9 a and b), because the addition ofheregulin abolished AMPK activation only in Au565 cells.

To further define the calcium-dependent activation of AMPK inresponse to GW2974, HMCs and HER2-overexpressing breastcancer cells were pretreated with BAPTA/AM, a calcium chelator.Treatment with BAPTA/AM alone did not activate AMPK. How-ever, pretreatment with BAPTA/AM completely blocked the ac-tivation of AMPK by GW2974 (Fig. 2b). These findings indicatethat inhibition of HER2 kinase by GW2974 is calcium-dependent.

Activation of Fatty Acid Oxidation and Change in Lipid Content inGW2974-Treated HMCs Is Calcium and AMPK-Dependent. Activationof AMPK stimulates increased fatty acid oxidation and ATPproduction. We compared the oxidation of fatty acids in HMCstreated with GW2974 or vehicle alone. GW2974 markedly reducedacylcarnitine C18 and C16 with a consequential increase in smallerlipid oxidation products (C2) (SI Table 5). We next determined theeffects of GW2974 on cellular lipid content in HMCs. Using OilRed-O staining to assess cellular lipid content, we found that lipidcontent was markedly reduced in GW2974 (but not trastuzumab)treated HMCs (Fig. 3a). Increased fatty acid oxidation and reduc-tion in cellular lipid content in response to GW2974 was calcium-dependent as these biological effects were blocked by pretreatmentwith BAPTA/AM (Fig. 3a).

Although these results implicate the activation of AMPK in theinduction of fatty acid oxidation and reduction in cellular lipidcontent in GW2974-treated cells, they do not demonstrate a directcause and effect relationship. To directly demonstrate the func-tional role of AMPK in fatty acid oxidation and changes in lipidcontent in GW2974-treated HER2 expressing cells, we selectivelyknocked down AMPK, using siRNA transfection. Steady-stateAMPK protein levels were effectively reduced by AMPK siRNA(SI Fig. 10). AMPK knockdown blocked the induction of fatty acidoxidation in response to GW2974 compared with cells transfectedwith a control (scrambled) siRNA construct (Fig. 3b).

In addition to the changes in lipid content, treatment withGW2974 increased steady-state protein levels of PGC-1 and ERR�,molecules associated with mitochondrial fatty acid oxidation, andMCAD, an inhibitor of fatty acid synthesis (Fig. 3c), whereastrastuzumab failed to do so. Thus, GW2974 triggers a cascade ofmetabolic events mediated by calcium-dependent activation ofAMPK, resulting in downstream events that conserve cellular

energy by inhibiting fatty acid synthesis while activating fatty acidoxidation as an alternative source of energy.

GW2974 Induces Fatty Acid Oxidation and Changes in Lipid Contentand Protect Cells from Apoptosis in the Myocardium of GW2974-Treated Rats. We next determined whether the effects of GW2974on cellular lipid content in HMCs also occur in vivo. Eight-week-oldSprague–Dawley rats were first starved for 12 h and then admin-istered a single dose of GW2974 (100 mg/kg) or vehicle by oralgavage. Eight hours after treatment, the animals were killed, theirhearts were harvested by snap freezing, and then myocardial lipidcontent was assessed by Oil Red-O staining.

Lipid staining was markedly reduced in the myocardium ofGW2974-treated rats compared with those administered vehiclealone (Fig. 4a). Because AMPK is also activated in response tonutrient deprivation (starvation) and can affect mitochondrial TCAbased metabolism, we tested the apoptotic activity in the myocar-dium of starved rats administered GW2974 or vehicle alone.

Fig. 3. Activation of AMPK by GW2974 in turn activates mediators of fatty acid oxidation, resulting in reduced cellular lipid content, which is calcium-dependent. (a) HMCs cells were subjected to the indicated treatment conditions (see Fig. 2a) for 3 days, and then cellular lipid content was determined by OilRed-O staining and light phase microscopy (100X). GW2974, but not trastuzumab, reduced cellular lipid content, which was blocked by pretreatment withBAPTA/AM. These results are representative of three independent experiments. (b) Selective AMPK knockdown blocked the effects of GW2974 on cellular lipidcontent compared with scrambled siRNA controls. Cells were untreated or treated with GW2974 together with either AMPK siRNA or scrambled siRNA andstained with Oil Red-O to assess cellular lipid content after two doses. (c) Steady-state protein levels of ERR�, PGC-1, and MCAD were increased in GW2974-treatedcells; HMCs were treated for 2 days as described in Fig. 2a, and then steady-state protein levels of the indicated proteins were assessed by Western blot. GW297,but not trastuzumab, increased steady-state protein levels of PGC-1 and ERR�, molecules associated with mitochondrial fatty acid oxidation and MCAD, aninhibitor of fatty acid synthesis. These results are representative of three independent experiments.

Fig. 4. Myocardial lipid content is reduced in GW2974-treated rats, which isassociated with protection against metabolic stress-induced apoptosis. (a) Eight-week-old female Sprague–Dawley rats were treated as described in Methods.Briefly, rats were fasted for 12 h before treatment with 100 mg/kg of GW2974 for8 h or left untreated and then were killed. Their hearts were harvested and snapfrozen, and sections were stained for lipid by Oil Red-0 and examined by lightmicroscopy (�40). Notice the abundant lipid in untreated rats compared withthose treated with GW2974. (b) Frozen sections of hearts from untreated orGW2974-treatedratswerestainedbyhematoxylinandeosinandforapoptosisbyTUNEL staining. Notice the brown stained apoptotic nuclei (TUNEL positivemarked with an arrow) in the untreated rats (5%) compared with the absence ofsignificant apoptosis (�1%) in the myocardium of rats treated with GW2974.

Spector et al. PNAS � June 19, 2007 � vol. 104 � no. 25 � 10609

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020

Apoptosis (TUNEL staining) was 5 times higher in the ratsadministered vehicle alone versus rats that were treated withGW2974 (�5% versus �1%, respectively), consistent with thecardioprotective effect of short-term treatment with GW2974(Fig. 4b).

GW2974, but Not Trastuzumab, Increases ATP Production and ProtectsAgainst TNF�-Induced HMCs Cell Death. Pro-inflammatory cytokines(e.g., TNF�) are frequently elevated in cancer patients, a potentialcontributing factor to the cardiotoxic effects of trastuzumab. In thisregard, we speculated that AMPK activation and stimulation oflipid oxidation in GW2974-treated HMCs might protect againstTNF� induced cell death.

To test this hypothesis, HMCs was subjected to the followingtreatment conditions: (i) GW2974; (ii) trastuzumab; (iii) TNF�; (iv)TNF� � GW2974; or (v) TNF� � trastuzumab. After 2 days oftreatment, cell viability was determined. In contrast to HMCstreated with trastuzumab and TNF� alone, which caused moderatecell killing, cell viability remained unchanged in GW2974-treatedcells (Fig. 5a). Combining trastuzumab with TNF� resulted inenhanced cell death compared with either treatment alone. Con-versely, concomitant treatment with GW2974 and TNF� protectedHMCs from TNF� killing.

Activation of AMPK reduces energy consumption while stimu-lating alternate metabolic pathways that lead to increased ATPproduction. There were marked differences in cellular ATP levelsin GW2974-treated HMCs compared with cells treated with tras-tuzumab or TNF�. In three independent experiments with a totalof 11 data points, ATP levels were significantly increased inGW2974-treated HMCs (P � 0.002) compared with reduced ATPlevels in trastuzumab (P � 0.007) and TNF� treated cells (P �0.0001) (Fig. 5b). The ability to activate fatty acid oxidation andconsequently maintain or increase intracellular ATP content maybe a critical factor mediating the cardioprotective effects of short-term exposure to GW2974.

In addition to their disparate effects on AMPK, GW2974 andtrastuzumab differ in their ability to activate the anti-apoptoticNF�B survival pathway in HMCs. TNF� is a potent activator ofNF�B/RelA, as evidenced by the increased steady-state proteinlevels of phospho-NF�B in HMCs (SI Fig. 11). Concomitanttreatment with GW2974 did not affect the activation of NF�B byTNF�. However, steady-state phospho-NF�B protein levels weremarkedly reduced in cells treated concomitantly with TNF� andtrastuzumab. The inhibition of protein synthesis by phosphorylationof eEF2 and the activation of the prosurvival factor NF�B as well

as up-regulation of ATP levels may provide a mechanism by whichGW2974 protects HMCs from TNF� mediated cell death.

DiscussionOncogenic tyrosine kinases, including HER2, are attractive targetsfor cancer therapeutics. Unfortunately, the increased incidence ofcardiomyopathy associated with targeted agents such as trastu-zumab (40) or more recently imatinib (11) (Gleevec), the smallmolecule inhibitor of c-abl kinase family members, may limit the useof many targeted therapies, particularly in the minimal diseasesetting, where patients are otherwise healthy and symptomaticheart failure might not be even an acceptable risk.

However, not all HER2 targeted therapies have the same risk ofcardiomyopathy as trastuzumab. Lapatinib, a dual HER2 andEGFR tyrosine kinase inhibitor appears to have a significantlyreduced risk of cardiotoxicity in women with HER2-overexpressingbreast cancer compared with trastuzumab (10).

To better understand the mechanism of cardiotoxicity associatedwith certain HER2-targeted therapies, we have identified AMPK,a master regulator of cell metabolism, as a key factor. Here, we showthat GW2974, a tyrosine kinase inhibitor with a similar activityprofile to lapatinib activates AMPK in HMCs, resulting in fatty acidoxidation, ATP production, up-regulation of nuclear receptorERR� and coactively PGC-1, molecular events associated withcardiac energy metabolism. It also provided protection againstTNF� induced cell death. In contrast, trastuzumab neither activatesAMPK nor stimulates fatty acid oxidation, and importantly doesnot protect HMCs against TNF� mediated killing.

Activation of AMPK occurs in response to perturbations in theAMP:ATP ratio in cells. The ensuing cascade of events that areinitiated by AMPK phosphorylation triggers a switch in cell me-tabolism that serves to produce rather than consume ATP. One ofthe hallmarks of this protective metabolic stress response is theoxidation of fatty acids for the purpose of increasing ATP produc-tion. GW2974, but not trastuzumab, phosphorylates ACC, whichresults in the inhibition of fatty acid synthesis and also modulatesother enzymes involved in promoting fatty acid oxidation. Inaddition, activation of AMPK in HMCs by GW2974 and lapatanibresulted in phosphorylation of eEF2, inhibiting protein synthesis,and protected cardiac cells from TNF�-induced killing.

Moreover, we demonstrated that induction of fatty acid oxidationby GW2974 results in a reduction of cellular lipids and protectionfor cardiac cell death not only occurs in HMCs in vitro but also inthe myocardium of animals administered a single dose of GW2974.Although a number of upstream kinases are capable of phosphor-ylating and activating AMPK, the calcium-dependent activation ofAMPK by GW2974 suggests this process is likely mediated bycalmodulin-dependent kinase kinase �.

HER2 expression appears to be important for the activation ofAMPK by GW2974. The explanation as to why GW2974 but nottrastuzumab activates AMPK and subsequent fatty acid oxidationand ATP production is unknown.

One possibility is that small molecule kinase inhibitors likeGW2974 and lapatinib have more potent inhibitory effects onHER2 signaling, with a more pronounced effect on HER2 trans-activation of HER3 (41). HER2:HER3 heterodimers are the mostpotent activators of survival pathways, because HER3 has sixbinding sites for subunits of PI3K (42, 43). Trastuzumab wasrecently shown to inhibit signaling via EGFR:HER2 heterodimersbut not HER2:HER3 complexes (44). The increased intracellularcalcium levels resulting in subsequent activation of AMPK and fattyacid oxidation in GW2974-treated cells, but not in trastuzumab-treated cells, may reflect the more potent effects of a small moleculeHER2 tyrosine kinase inhibitor on HER2:HER3 complex activa-tion compared with an antibody-based therapy like trastuzumab(Fig. 6). Preliminary results in our laboratory favor this hypothesis.The involvement of protein tyrosine kinases in ion and calciumchannels are reported in ref. 45.

Fig. 5. HER2 kinase inhibition by GW2974 increases cellular ATP and protectsHMCs from TNF� induced apoptosis. (a) HMCs cells were treated as previouslydescribed in Fig. 2a. Briefly, cells were treated with GW2974 (25 �M), TNF�

(300 ng/ml), trastuzumab (50 �g/ml) or a combination TNF� with GW2974 ortrastuzumab. Cell viability was determined by staining, using 1% methyleneblue, elution of dye with 0.1 M HCl, and then measuring absorbance OD 640nm by spectrophotometer. (b) Cells were treated as indicated in Methods inthe presence or absence of GW2974; trastuzumab and TNF� and ATP cellularcontent was quantified by using an ATP Bioluminescence Assay Kit HS 2 (seeMethods). ATP cellular content increased in cells treated with GW2974 anddecreased in those treated with trastuzumab or TNF�. These results arerepresentative of three independent experiments.

10610 � www.pnas.org�cgi�doi�10.1073�pnas.0701286104 Spector et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020

Importantly, GW2974 protects human cardiomyocytes againstTNF� induced cell death, which has potentially significant clinicalimplications, because circulating proinflammatory cytokines, suchas TNF�, are frequently increased in patients with cancer andparticularly patients receiving chemotherapy.

Anti-cytokine therapies, including the soluble TNF� receptorantagonist etanercept, have even been proposed as strategies toprevent anthracycline-induced cardiotoxicity. Inhibition of HER2kinase, increased intracellular calcium levels and consequentialactivation of AMPK by GW2974-protected HMCs against TNF�-mediated cell killing. In addition to the differential effects ofGW2974 and trastuzumab on the AMPK protective metabolicstress response, specifically GW2974, but not trastuzumab, activatesthe anti-apoptosis molecule NF�B. In fact, trastuzumab blocked theactivation of NF�B by TNF�, which might contribute to its lack ofcytoprotective effect in HMCs compared with the small moleculetyrosine kinase inhibitor. Increases in intracellular calcium havebeen shown to activate NF�B (46, 47), providing a potentialexplanation for its activation by GW2974 but not trastuzumab. Themechanism(s) responsible for NF�B activation and the increasedintracellular [Ca2�] is currently being investigated.

We have previously shown (48) that combining small moleculeHER2 kinase inhibitors with trastuzumab results in enhancedanti-tumor activity in HER2-overexpressing breast cancer cells. Inaddition, preliminary results in our laboratory show in vitro cellularprotection of HMCs treated by both therapies. Our findings raisethe possibility that combining small molecule HER2 kinase inhib-itors or other therapies that activate AMPK might protect againstthe cardiotoxic effects of trastuzumab. A clinical trial is currentlyassessing the clinical activity of the combination of lapatinib andtrastuzumab. This trial will also closely examine the effects of thiscombination on cardiac function. Nonetheless, activators of AMPKmight provide cardioprotection against cardiotoxic therapies.

Although activation of AMPK and its inhibitory effects onprotein synthesis might be protective of normal cardiac myocytes,this effect appears to be lethal to sensitive HER2-overexpressingbreast cancer cells, which are ‘‘addicted’’ to glycolysis (49). Despiteactivating AMPK, GW2974 induces HER2-overexpressing breastcancer cells to undergo apoptosis (data not shown). When HER2-overexpressing breast cancers are withdrawn from their ‘‘glycolyticaddicted’’ state after inhibition of HER2 activation, their ability to

switch from glucose to fatty acids as an alternative bioenergeticsubstrate is compromised, leading to energy failure and cell death.Thus, the very mechanism that protects HMCs from TNF� inducedkilling, may contribute to the mechanism of anti-tumor activity byGW2974, lapatanib, and related compounds.

In summary, the activation of AMPK and its consequentialeffects on metabolic pathways and energy production in humancardiac myocytes may explain the apparent reduced risk of cardio-toxicity associated with certain HER2 small molecule tyrosinekinase inhibitors compared with trastuzumab. We propose thefollowing model by which HER2 kinase inhibitors like GW2974protect HMCs from TNF� mediated apoptosis. First, inhibition ofHER2 (and EGFR) kinase activity by GW2974 blocks the trans-activation of HER3 and leads to increased intracellular [Ca2�](yellow). Second, increased [Ca2�] activates calmodulin-dependentkinase kinase, which activates AMPK. Third, activation of AMPKinhibits (i) mTOR and eEF2, and (ii) ACC blocking protein andfatty acid synthesis, respectively. And finally, AMPK induces fattyacid oxidation by activating mitochondrial ERR� and PGC-1�. Thecumulative effect of this signaling cascade is to reduce consumptionand increase ATP production, which protects HMCs from apopto-sis. In light of the protective effect of GW2974 and lapatanib,screening HER2 and other tyrosine kinase inhibitors therapies thatactivate rather than inhibit AMPK (50) and using other therapiesthat activate AMPK, such as pharmacological agents like thiazo-lidinediones, metformin, and 5-aminoimidazole-4-carbox-amide-1-�-D-ribofuranoside together with tyrosine kinase inhibitors mightprovide novel targeted approaches to protect myocardium fromapoptosis induced by proinflammatory cytokines released or otherstresses such as acute ischemia injury and warrants furtherinvestigation.

Materials and MethodsWestern Blot Analysis. Equal amounts of total protein (25 �g perlane) were electrophoresed on either an 8% or 10% SDS-polyacrylamide gel, transferred to PVDF membranes (Millipore,Billerica, MA) and blotted, using the following primary anti-bodies: pAkt, pErk1/2, pAMPK�, pACC, pNF�B, and peEF2(Cell Signaling, Beverly, MA); ERR�, ERR� (R & D Systems,Minneapolis, MN); PGC-1 (Chemicon International, Temecula,CA); MCAD (Cayman Chemicals, Ann Arbor, MI), and actin(Sigma, St. Louis, MO). Secondary antibodies were either ECLanti-mouse HRP conjugated or ECL anti-rabbit HRP conju-gated (Amersham Biosciences, Pittsburgh, PA). Proteins werevisualized with the SuperSignal West Pico ChemiluminescentSubstrate (Pierce, Rockford, IL).

Cell Culture and Treatments. Au565 breast cancer cell line wasobtained from American Type Culture Collection (Manassas, VA)and cultured in RPMI medium 1640 with 15% heat-inactivatedFBS. HMCs (embryonic human primary cardiomyocytes) wereobtained from ScienCell Research Laboratories (San Diego, CA)and cultured in the media provided by the same company specif-ically for the cell line, on plastic wear coated with (poly)L-lysine.Cell cultures were maintained in a humidified atmosphere of 5%CO2 at 37°C. Treatments included: 30 �M BAPTA/AM (Calbio-chem, San Diego, CA); 25 �M/ml GW2974 (Sigma); 50 �g/mltrastuzumab (Herceptin) (Genentech, South San Francisco, CA);200 �M Compound C (Calbiochem); 300 ng/ml (short treatment)or 100 ng/ml (long treatment) TNF� (Sigma); heregulin 50 ng/ml(Santa Cruz Biotechnologies, Santa Cruz, CA). Charcoal-strippedFBS (charcoal, dextran coated) (Sigma) was used for all experi-ments analyzing fatty acid metabolism.

Lipids Staining. Cells or 5-�m frozen sections were fixed in 10%neutral buffered formalin, incubated in the working solution of OilRed-O (Sigma) for 30 min, differentiated in 60% 2-Propanol(Sigma), and mounted in Glycerol Gelatin (Sigma). Oil Red-O

Fig. 6. A schematic illustration of the two proposed mechanisms of cardiacprotection by GW2974. Inhibition of EGFR and HER2 phosphorylation byGW2974, which also blocks the transactivation of HER3, leads to increased [Ca2�](yellow) in HMCs. Calcium activates calmodulin-dependent kinase kinase, whichin turn activates AMPK resulting in the inhibition of lipid (ACC) and proteinsynthesis (eEF2, mTOR) and increased fatty acid oxidation through activation ofmitochondrial ERR� and PGC-1. The cumulative effect of a decrease in lipid andprotein synthesis combined with an increase in fatty acid oxidation leads toincreased production rather than consumption of ATP.

Spector et al. PNAS � June 19, 2007 � vol. 104 � no. 25 � 10611

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020

stock solution was prepared by dissolving 300 mg of the dye in 100ml of 2-propanol by heating to 100°C with stirring. Working solutionwas prepared by combining six parts of stock solution with fourparts of water.

Apoptosis Analysis. Apoptosis staining was performed by theTUNEL staining procedure as described in ref. 9.

Calcium Detection. Au565, MCF7, MCF7/HER2, and HMCs cellswere seeded on glass bottom dishes (MatTek, Ashland, MA) theday before the experiment. On the day of experiment, a 1:1 mixtureof Ca2� indicator Fluro-4 (Molecular Probes, Carlsbad, CA) (5�M) and Pluoronic was diluted with regular growth medium andused to replace the original media. Incubation with Fluro-4 was 30min at 37°C. Afterward, cells were washed in 37°C PBS andincubated in Phenol Red-free medium for 15 min. Then 2� solutionof the drug in Phenol Red-free medium was made and mixed gentlynear the microscope (Yokogawa Nipkow spinning disk confocalbuilt on a Nikon TE-2000U microscope at Northwestern Univer-sity, Chicago, IL).

ATP Measurements. ATP determination was done by using an ATPBioluminescence Assay Kit HS 2 (Roche, Basel, Switzerland).Briefly, human cardiomyocytes were grown and treated with thedrugs on six-well culture dishes. On the day of the experiment, cellswere trypsinized, spun down, and resuspended in the dilutionbuffer. Equal amounts of the cell suspension and lysis buffer weremixed gently in Eppendorf (Boulder, CO) tubes and incubated for5 min, and aliquots were transferred into 96-well microplates (white,nontransparent) with luciferase reagent. ATP was measured im-mediately by using a Dynatech (Chantilly, VA) ML 1000 PlateReader. The ATP standard curve was prepared by using ATPprovided with the kit.

siRNA Transfection. siRNA to AMPK was purchased from Dhar-macon (Lafayette, CO). siRNA duplex for scrambled control

siRNA was synthesized (IDT, Coralville, IA) as follows: sense 5�rGrCrGrCrGrCrUrUrUrGrUrArGrGrArUrUrCrGTT and anti-sense 5� rCrGrArArUrCrCrUrArCrArArArGrCrGrCrGrCTT.The duplex was resuspended according to the manufacturer’srecommended protocol. Primary HMCss were transfected with 2ng of AMPK or scrambled siRNA, using Nucleofector (Amaxa,Gaithersburg, MD) and the Rat Cardiomyocyte Neo NucleofactorKit. After transfection the cells were seeded into (poly)L-lysinecoated chamber slides. The following day, the cells were either leftuntreated or treated with 25 �M GW2974 for 3 days and thenstained for lipids.

In Vivo Studies. Female Sprague–Dawley rats were obtained fromHarlan (Indianapolis, IN). At 8 weeks, the rats were treated. Beforetreatment, the rats were put under fast for 12 h, administeredGW2974 100 mg/kg by oral gavage, and killed 8 h later, at whichtime their hearts were harvested and snap frozen. GW2974 wasobtained from Sigma and dissolved in 0.5% hydroxypropylmethyl-cellulose/0.1%Tween 80/99.6% water for dosing. To prepare thedosing solution, the required amount of vehicle or GW2974 wasadded and mixed thoroughly. The final dosing suspension was abright yellow fine suspension with pH 4.8. The rats were housed inThoren (Hazleton, PA) microisolator caging with Bed-O’Cobsbedding. All procedures carried out in this experiment were ap-proved by the MIR Animal Care and Use Committee.

Statistical Analysis of ATP Levels. Analysis was performed by usingSPSS software version 13.0 (SSPS, Chicago, IL), using Fisher’s exacttest for ATP levels. We assumed the control value to be 100% inall cases. The percent difference from the control was calculated foreach data-point and the statistical significance of the differencebetween the treatment and the control was assessed.

We thank Adriana Greisman (medical writer) and Roberta Dvells (admin-istrative assistant to S.S.B.) for their assistance in preparation of thismanuscript.

1. Lee KF, Simon H, Chen H, Bates B, Hung MC, Hauser C (1995) Nature 378:394–398.2. Meyer D, Birchmeier C (1995) Nature 378:386–390.3. Crone SA, Zhao YY, Fan L, Gu Y, Minamisawa S, Liu Y, Peterson KL, Chen J, Kahn

R, Condorelli G, et al. (2002) Nat Med 8:459–465.4. Ozcelik C, Erdmann B, Pilz B, Wettschureck N, Britsch S, Hubner N, Chien KR,

Birchmeier C, Garratt AN (2002) Proc Natl Acad Sci USA 99:8880–8885.5. Cobleigh MA, Vogel CL, Tripathy D, Robert NJ, Scholl S, Fehrenbacher L, Wolter

JM, Paton V, Shak S, Lieberman G, et al. (1999) J Clin Oncol 17:2639–2648.6. Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A, Fleming T,

Eiermann W, Wolter J, Pegram M, et al. (2001) N Engl J Med 344:783–792.7. Grazette LP, Boecker W, Matsui T, Semigran M, Force TL, Hajjar RJ, Rosenzweig

A (2004) J Am Coll Cardiol 44:2231–2238.8. Gassmann M, Casagranda F, Orioli D, Simon H, Lai C, Klein R, Lemke G (1995)

Nature 378:390–394.9. Spector NL, Xia W, Burris, H, III, Hurwitz H, Dees EC, Dowlati A, O’Neil B,

Overmoyer B, Marcom PK, Blackwell KL, et al. (2005) J Clin Oncol 23:2502–2512.10. Geyer CE, Forster J, Lindquist D, Chan S, Romieu CG, Pienkowski T, Jagiello-

Gruszfeld A, Crown J, Chan A, Kaufman B, et al. (2006) N Engl J Med 355:2733–2743.11. Kerkela R, Grazette L, Yacobi R, Iliescu C, Patten R, Beahm C, Walters B, Shevtsov

S, Pesant S, Clubb FJ, et al. (2006) Nat Med 12:908–916.12. Mann DL (2006) Nat Med 12:881–882.13. Gottschalk S, Anderson N, Hainz C, Eckhardt SG, Serkova NJ (2004) Clin Cancer

Res 10:6661–6668.14. Vega RB, Huss JM, Kelly DP (2000) Mol Cell Biol 20:1868–1876.15. Vega RB, Kelly DP (1997) J Biol Chem 272:31693–31699.16. Mu J, Brozinick JT, Jr, Valladares O, Bucan M, Birnbaum MJ (2001) Mol Cell

7:1085–1094.17. Schreiber SN, Knutti D, Brogli K, Uhlmann T, Kralli A (2003) J Biol Chem

278:9013–9018.18. Kressler D, Schreiber SN, Knutti D, Kralli A (2002) J Biol Chem 277:13918–13925.19. Wu Z, Puigserver P, Andersson U, Zhang C, Adelmant G, Mootha V, Troy A, Cinti

S, Lowell B, Scarpulla RC, et al. (1999) Cell 98:115–124.20. Huss JM, Torra IP, Staels B, Giguere V, Kelly DP (2004) Mol Cell Biol 24:9079–9091.21. Huss JM, Kopp RP, Kelly DP (2002) J Biol Chem 277:40265–40274.22. Kamei Y, Ohizumi H, Fujitani Y, Nemoto T, Tanaka T, Takahashi N, Kawada T,

Miyoshi M, Ezaki O, Kakizuka A (2003) Proc Natl Acad Sci USA 100:12378–12383.23. Hardie DG (2004) J Cell Sci 117:5479–5487.24. Huss JM, Kelly DP (2005) J Clin Invest 115:547–555.25. Hardie DG, Scott JW, Pan DA, Hudson ER (2003) FEBS Lett 546:113–120.26. Kudo N, Barr AJ, Barr RL, Desai S, Lopaschuk GD (1995) J Biol Chem 270:17513–

17520.

27. Kudo N, Gillespie JG, Kung L, Witters LA, Schulz R, Clanachan AS, Lopaschuk GD(1996) Biochim Biophys Acta 1301:67–75.

28. Mercy L, Pauw A, Payen L, Tejerina S, Houbion A, Demazy C, Raes M, Renard P,Arnould T (2005) FEBS J 272:5031–5055.

29. Lizcano JM, Goransson O, Toth R, Deak M, Morrice NA, Boudeau J, Hawley SA,Udd L, Makela TP, Hardie DG, et al. (2004) EMBO J 23:833–843.

30. Shaw RJ, Kosmatka M, Bardeesy N, Hurley RL, Witters LA, DePinho RA, CantleyLC (2004) Proc Natl Acad Sci USA 101:3329–3335.

31. Shaw RJ, Bardeesy N, Manning BD, Lopez L, Kosmatka M, DePinho RA, CantleyLC (2004) Cancer Cell 6:91–99.

32. Inoki K, Zhu T, Guan KL (2003) Cell 115:577–590.33. Lehman JJ, Barger PM, Kovacs A, Saffitz JE, Medeiros DM, Kelly DP (2000) J Clin

Invest 106:847–856.34. Bolster DR, Crozier SJ, Kimball SR, Jefferson LS (2002) J Biol Chem 277:23977–23980.35. Horman S, Browne G, Krause U, Patel J, Vertommen D, Bertrand L, Lavoinne A,

Hue L, Proud C, Rider M (2002) Curr Biol 12:1419–1423.36. Terai K, Hiramoto Y, Masaki M, Sugiyama S, Kuroda T, Hori M, Kawase I, Hirota

H (2005) Mol Cell Biol 25:9554–9575.37. Lackey KE (2006) Curr Top Med Chem 6:435–460.38. Bryant D, Becker L, Richardson J, Shelton J, Franco F, Peshock R, Thompson M,

Giroir B (1998) Circulation 97:1375–1381.39. Konecny GE, Pegram MD, Venkatesan N, Finn R, Yang G, Rahmeh M, Untch M,

Rusnak DW, Spehar G, Mullin RJ, et al. (2006) Cancer Res 66:1630–1639.40. Chien KR (2006) N Engl J Med 354:789–790.41. Xia W, Liu LH, Ho P, Spector NL (2004) Oncogene 23:646–653.42. Prigent SA, Gullick WJ (1994) EMBO J 13:2831–2841.43. Fedi P, Pierce JH, di Fiore PP, Kraus MH (1994) Mol Cell Biol 14:492–500.44. Wehrman TS, Raab WJ, Casipit CL, Doyonnas R, Pomerantz JH, Blau HM (2006)

Proc Natl Acad Sci USA 103:19063–19068.45. Davis MJ, Wu X, Nurkiewicz TR, Kawasaki J, Gui P, Hill MA, Wilson E (2001) Am J

Physiol 281:H1835–H1862.46. Zhou X, Yang W, Li J (2006) J Biol Chem 281:31337–31347.47. Antonsson A, Hughes K, Edin S, Grundstrom T (2003) Mol Cell Biol 23:1418–

1427.48. Xia W, Gerard CM, Liu L, Baudson NM, Ory TL, Spector NL (2005) Oncogene

24:6213–6221.49. Elstrom RL, Bauer DE, Buzzai M, Karnauskas R, Harris MH, Plas DR, Zhuang H,

Cinalli RM, Alavi A, Rudin CM. et al. (2004) Cancer Res 64:3892–3899.50. Chan AY, Dyck JR (2005) Can J Physiol Pharmacol 83:24–28.

10612 � www.pnas.org�cgi�doi�10.1073�pnas.0701286104 Spector et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

9, 2

020