Translational Science

Atypical G Protein b5 Promotes CardiacOxidativeStress, Apoptosis, and Fibrotic Remodeling inResponse to Multiple Cancer ChemotherapeuticsSreemoyee Chakraborti1, Arnab Pramanick1, Sudipta Saha2, Somnath Singha Roy1,Arnab Ray Chaudhuri3, Madhusudan Das4, Sujoy Ghosh4, Adele Stewart5,and Biswanath Maity1,4

Abstract

The clinical use of multiple classes of cancer chemotherapeuticsis limited by irreversible, dose-dependent, and sometimes life-threatening cardiotoxicity. Though distinct in their mechanismsof action, doxorubicin, paclitaxel, and 5-FU all induce rapidand robust upregulation of atypical G protein Gb5 in the myocar-dium correlating with oxidative stress, myocyte apoptosis, andthe accumulation of proinflammatory and profibrotic cytokines.In ventricular cardiac myocytes (VCM), Gb5 deficiency providedsubstantial protection against the cytotoxic actions of chemother-apeutics, including reductions in oxidative stress and simultaneousattenuation of ROS-dependent activation of the ATM and CaMKIIproapoptotic signaling cascades. In addition, Gb5 loss allowedfor maintenance of Dym, basal mitochondrial calcium uniporterexpression, andmitochondrial Ca2þ levels, effects likely to preservefunctional myocyte excitation–contraction coupling. The deleteri-ous effects of Gb5 are not restricted to VCM, however, as Gb5knockdown also reduces chemotherapy-induced release of proin-flammatory cytokines (e.g., TNFa), hypertrophic factors (e.g.,

ANP), and profibrotic factors (e.g., TGFb1) from both VCM andventricular cardiac fibroblasts, with the most dramatic reductionoccurring in cocultured cells. Our experiments suggest that Gb5facilitates the myofibroblast transition, the persistence of whichcontributes to pathologic remodeling and heart failure. The con-vergence of Gb5-mediated, ROS-dependent signaling pathways inboth cell types represents a critical etiological factor in thepathogenesis of chemotherapy-induced cardiotoxicity. Indeed,intracardiac injection of Gb5-targeted shRNA allowed for heart-specific protection against the damaging impact of chronicchemotherapy. Together, our results suggest that inhibition ofGb5 might represent a novel means to circumvent cardiotoxicityin cancer patients whose treatment regimens include anthra-cyclines, taxanes, or fluoropyrimidines.

Significance: These findings suggest that inhibiting an atypicalG-protein might provide a strategy to limit the cardiotoxicity incancer patients treated with anthracyclines, taxanes, or fluoropyr-imidines. Cancer Res; 78(2); 528–41. �2017 AACR.

IntroductionAccording to the World Health Organization, cancer caused 1

in 6 deaths globally in 2015 with cancer-related deathsexpected to rise by 70% in the next 20 years. Despite decadesof research, cancer remains a leading cause of morbidity andmortality worldwide. Current therapeutic regimens for cancertreatment combine surgical tumor resection with radiotherapy,

chemotherapy, and/or additional targeted, hormonal, orimmunological therapies to remove cancerous tissue and pre-vent further dissemination of malignant cells. Though provento improve patient survival over a 20-year period (1), the utilityof many chemotherapeutics, including anthracyclines, taxanes,and fluoropyrimidines, is limited by adverse effects, amongthem, dose-dependent and often irreversible cardiac damagethat can result in eventual heart failure (2).

Anthracyclines, including the widely used drug doxorubicin,induce dilated or restrictive cardiomyopathy, left ventriculardysfunction, and congestive heart failure (3). Indeed, the risk ofcardiac death in childhood cancer survivors receiving high-doseanthracycline chemotherapy is increased 4-fold (4). Antimicro-tubule agents such as paclitaxel, in contrast, are associated withdysregulations of the cardiac conduction system includingsinus bradycardia, atrioventricular block, and ventricular tachy-cardia, in addition to chronic congestive heart failure (3).Cardiotoxicity induced by 5-fluorouracil (5-FU), though lessfrequently observed, is characterized by arrhythmias, myocar-dial infarction, and risk for sudden cardiac death (5). In manycases, the only course of action to prevent further heart damageis dose limitation or complete cessation of chemotherapy.Though specific mechanisms have been described for individ-ual drugs, a unifying theory that could lead to the development

1Centre of Biomedical Research, Sanjay Gandhi Post-Graduate Institute ofMedical Sciences, Lucknow, Uttar Pradesh, India. 2Department of Pharmaceu-tical Sciences, Babasaheb Bhimrao Ambedkar University, Vidya Vihar, Lucknow,Uttar Pradesh, India. 3Department of Molecular Genetics, Erasmus MedicalCentre, Rotterdam, the Netherlands. 4Department of Zoology, University ofCalcutta, Kolkata, West Bengal, India. 5Department of Biomedical Science,Charles E. Schmidt College of Medicine, Florida Atlantic University, Jupiter,Florida.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Author: B. Maity, Centre of Biomedical Research, SGPGI Cam-pus, Raebareli Road, Lucknow, Uttar Pradesh 226014, India. Phone: 91-9674575238; Fax: 91-522-2668215; E-mail: bmaity28@gmail.com

doi: 10.1158/0008-5472.CAN-17-1280

�2017 American Association for Cancer Research.

CancerResearch

Cancer Res; 78(2) January 15, 2018528

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

of versatile therapeutics aimed at ameliorating the cardiotoxiceffects of cancer chemotherapy remains elusive. It is possible, infact, that the very mechanisms that contribute to the cancercell killing actions of these drugs also lead to off target tissuedamage. Among these mechanisms are the activation of theDNA damage signaling cascade and mitochondrial dysfunc-tion–triggered oxidative stress that can lead to cellular apopto-sis and immunological reactions (6, 7).

The cardiotoxic actions of doxorubicin are best characterized.Though many mechanisms have been described, the accumu-lation of reactive oxygen species (ROS) within cardiomyocytesappears to be a key pathogenic nexus for doxorubicin-inducedcardiac damage. Production of free radicals can occur follow-ing doxorubicin exposure through dysregulation of the mito-chondrial electron transport chain (8, 9), direct interaction withendothelial nitric oxide synthase (10), and activation ofNADPH oxidase (Nox) complexes (11, 12). ROS scavengersattenuate cardiac activation of the proapoptotic ATM/p53 path-way following doxorubicin exposure (13). p53 activity is criticalfor doxorubicin-induced contractile dysfunction as genetic dis-ruption or inhibition of p53 protects against doxorubicin-induced myocardial cell apoptosis (14, 15). Dox-induced,Ca2þ/Calmodulin-dependent protein kinase (CaMKII)–depen-dent leak of Ca2þ from the sarcoplasmic reticulum also con-tributes to cardiomyocyte Ca2þ overload, actions that requireROS accumulation (16). Given the demonstrated importance ofCaMKII in multiple models of cardiac injury (17), it seemslikely that CaMKII upregulation, possibly through oxidativemeans (18), functions as a central mechanism for myocytedysfunction and death following doxorubicin treatment. Fol-lowing death of cardiac myocytes, a complex adaptation processbegins whereby resident fibroblasts attempt to repair structuraldamage to the myocardial wall and maintain cardiac output. Inthe face of extrinsic stress, such as repeated chemotherapyexposure, this remodeling can often turn maladaptive, whereincytokines released by fibroblasts feedback to further promoteoxidative stress in myocytes (19). The challenge then, fortherapeutics aimed at amelioration of chemotherapy-inducedcardiac damage, lies in identifying upstream targets responsiblefor activation of a number of parallel signaling cascades allcontributing to and dependent on chemotherapy-induced car-diac oxidative stress.

The atypical G protein Gb5 forms costabilizing complexes withR7 subfamily regulators of G protein signaling (RGS) proteins(20–22). One member of this family, RGS6, upregulated follow-ing doxorubicin exposure in cancer cells and heart (23, 24), hasrecently been identified as a mediator of doxorubicin-inducedROS generation, activation of p53, andmyocardial cell apoptosis(23). Because deletion of Gb5 results in complete elimination ofRGS6 expression in heart (25), it is likely that Gb5 also plays a keyrole in mediating doxorubicin-induced cardiac damage. Thereremain, however, fundamental gaps in our understanding of howRGS6/Gb5 heterodimers promote doxorubicin-induced, ROS-dependent cardiac damage. For example, though RGS6 appearsto be a critical mediator of doxorubicin-induced p53 activation,inhibition of the p53 pathway is insufficient to protect againstdoxorubicin-dependent cardiac fibrosis (26). Manipulation ofGb5 expression, then, represents an opportunity to investigatethe importance of the RGS/Gb5 complexes in cardiac damageinduced by multiple chemotherapeutic drugs and the specificmechanism(s) involved. In this work, we provide evidence impli-

cating Gb5 in chemotherapy-induced cardiac damage. By pro-moting ROS generation following drug exposure, Gb5 facilitatesactivation of multiple proapoptotic signaling cascades in myo-cytes while simultaneously promoting pathologic cross-talkbetween cardiomyocytes and surrounding fibroblasts contribut-ing to heart hypertrophy and fibrosis. Given these findings, wepostulate that targeting of the RGS/Gb5 interaction interfacemight be a viable means to protect the heart from the deleteriousimpact of chemotherapeutic drugs.

Materials and MethodsMaterials

Doxorubicin hydrochloride, paclitaxel, and 5-FU were pur-chased from Abcam as were kits for detection of cleaved cas-pase-3, myeloperoxidase (MPO), annexin V/propidium iodide–positive cells, superoxide dismutase (SOD) and calcium (Ca2þ)and antibodies for Gb5, CaMKII, p-CaMKII, ATM, p-ATM, p53, p-p53, b myosin heavy chain (b-MHC), atrial natriuretic peptide(ANP), and b-actin. Diphenyleneiodonium chloride (DPI), N-acetyl cysteine (NAC), dimethylthiourea (DMTU),N-Nitro-L-argi-nine methyl ester (L-NAME), Ruthenium 360 (RU360), cyclo-sporine A, KN93, KU55933, CM-H2DCFDA, Oil Red O stain,Masson trichrome stain, cycloheximide, polyethylene glycol-SOD(Peg-SOD), and polyethylene glycol-catalase (Peg-Cat) were pro-cured from Sigma. The ELISA Kit for analyses of TGFb1, TNFa,and IL10was obtained from Thermo Fisher Scientific. The TUNELKit was from Biovision, GW788388 was from Tocris Biosciences,and the hydrogen peroxide (H2O2) assay Kit was from CaymanChemicals. Antibodies for mitochondrial calcium uniporter(MCU), Bax, and phosphorylated (p)-STAT3 were procured fromCell Signaling Technology, and Ox-CaMKII antibody was pur-chased from GeneTex. The cell death detection Kit was fromRoche. Lentiviral vectors encoding small hairpin RNA (shRNA)targeting Gb5, ATM, Nox4, CaMKIId, and STAT3 were purchasedfrom Santa Cruz Biotechnology.

Animals and in vivo Gb5 knockdownMice experiments were performed at the Department of

Pharmaceutical Sciences, Babasaheb Bhimrao Ambedkar Uni-versity in association with S.D. College of Pharmacy andVocational Studies, Uttar Pradesh, India. Animals were pro-cured after obtaining clearance from the Animal Ethics Com-mittee (SDCOP& VS/AH/CPCSEA/01/0021) and were handledfollowing International Animal Ethics Committee Guidelines.Male Swiss albino mice (25–30 g) were reared on a balancedlaboratory diet as per NIN, Hyderabad, India, and were kept at20 � 2�C, 65% to 70% humidity, on a 12/12 hour day/nightcycle. Animals were given tap water and food ad libitumthroughout the study. Twelve- to 14-week-old animals weresubjected to both acute and chronic chemotherapeutic drugregimens wherein they received multiple doses of doxorubicin(cumulative dose of 45 mg/kg i.p.; 9 mg/kg every other week),paclitaxel (cumulative dose of 85 mg/kg i.p.; 17 mg/kg everyother week), 5-FU (cumulative dose of 200mg/kg i.p.; 40mg/kgevery other week), or saline over the period of 10 weeksor a single dose of doxorubicin (20 mg/kg i.p.), paclitaxel(50 mg/kg i.p.), 5-FU (150 mg/kg i.p.), or saline administratedvia i.p. injection. In a separate study, 1-week-old WT micereceived an intracardaic injection of 4.5 � 108 lentiviral vectorscontaining control or Gb5-targeted shRNA according to a

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 529

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

previously published protocol (27) before being returned totheir mothers until weaning. These animals were then exposedto chemotherapeutic drugs according to the protocols outlinedabove. For all animal experiments, analyses were performed byan observer blinded to drug and repeated at least 3 times.

Ventricular cardiomyocyte and fibroblast isolation andculture

Primary ventricular cardiomyocytes (VCM) and fibroblasts(VCF) were isolated from 8- to 10-week-old adult mice accord-ing to a published protocol (28). Twenty-four hours afterisolation, cells were treated with doxorubicin (3 mmol/L),paclitaxel (100 nmol/L), or 5-FU (500 mmol/L) for differenttime points in the presence or absence of DPI (1 mmol/L), NAC(5 mmol/L), RU360 (50 mmol/L), cyclosporine A (0.2 mmol/L),KN93 (50 mmol/L), KU55933 (10 mmol/L), or GW78838(2 mmol/L) where indicated. Cells were transduced with lenti-viral vectors encoding shGb5 or control shRNA according to themanufacturer's instructions.

Measurement of ROS generation and apoptosisROS generation was estimated in the tissues and primary

cells using the cell-permeable oxidation-sensitive probe, CM-H2DCFDA, as described previously (23). Briefly, cells wereharvested by centrifugation, washed 3 times with ice-cold PBS,resuspended in PBS, and incubated with 5 mmol/L CM-H2DCFDA for 20 minutes at 37�C. After incubation, cells wereagain washed and lysed in PBS with 1% Tween 20. ROS levelwas determined as the ratio of dichlorofluorescein excitationat 480 nm to emission at 530 nm. H2O2 was measured viaELISA. Cell lysates and cardiac tissue lysates were used tomeasure apoptosis and caspase-3 activity using ELISA kitsaccording to the manufacturer's protocols. Apoptotic cells werealso detected by measuring phosphatidylserine exposure inthe outer membrane using an annexin V/propidium iodidekit according to the manufacturer's protocol. Annexin V–pos-itive cells relative to total number of cells were counted using afluorescent microscope.

Masson trichrome, terminal deoxynucleotidyl transferase–mediated dUTP nick end, and hematoxylin–eosin staining

Formalin-fixed, paraffin-embedded heart tissue sectionswere prepared. Masson trichrome staining kit was purchasedfrom Sigma, and staining was performed according to themanufacturer's protocol. Terminal deoxynucleotidyl transfer-ase–mediated dUTP nick end (TUNEL) and hematoxylin &eosin staining were performed as described previously (27).

ImmunoblottingTissues were rapidly dissected from mice and flash frozen in

liquid nitrogen. Tissue homogenates and cell lysates were pre-pared in RIPA buffer containing protease and phosphatase inhi-bitors (Sigma), quantified and probed as previously described(29). Twenty micrograms of protein per sample was subjected toSDS-PAGE and immunoblotting using standard techniques.Immunoblots were developed using a chemiluminescence meth-od with horseradish peroxidase–labeled secondary antibodies.Densitometric quantification of Western blots was performedutilizing Image J software (NIH). Protein expression was normal-ized to loading controls and expressed relative to controlconditions.

Assays of enzyme activity, oxidative stress, and cytokine levelsSOD and MPO activity as well as levels of TGFb1, TNFa, IL10,

and Ca2þ were assayed from cardiac tissue homogenates andcellular lysates using ELISA kits according to the manufacturer'sprotocol. GPX activity was measured using the method of Floheand Gunzler (30).

Statistical analysesData were analyzed by the Student t test or one or two-way

ANOVA with the Bonferroni post hoc adjustment as appropriate.Statistical analyses were performed using Origin software. Resultswere considered significantly different at P < 0.05. Values areexpressed as mean � SEM.

ResultsPharmacologically distinct cancer chemotherapeuticsdamage cardiac tissue via common pathologic mechanismscorrelating with upregulation of Gb5

We first sought to investigate the impact of cardiotoxic che-motherapeutics on Gb5 expression in heart. All three drugs(doxorubicin, paclitaxel, and 5-FU) disruptedmyofilament archi-tecture following chronic exposure (Fig. 1A) with detectablefibrotic remodeling, increased myocyte area, and heart hypertro-phy (Fig. 1B). These pathologic changers were associated withearly elevations in the inflammatory cytokine TNFa (Fig. 1C) anda later accumulation of TGFb1 (Fig. 1D). Multiple reports havesuggested that biphasic changes in cytokine levels, particularlyTGFb1, are seen acrossmultiple heart failure and cardiomyopathymodels. A large increase in apoptotic nuclei in TUNEL stainedcardiac tissue sections detectable as soon as 24 hours after acutedrug exposure and further increased 3 and 7 days after injection(Fig. 1E). Similarly, total ROS levels exhibited a time-dependentelevation following chemotherapy challenge (Fig. 1F). By day 5after drug administration, we observed a 3-fold increase in Gb5protein (Fig. 1G). Protein levels of Gb5 were reduced butremained elevated at day 5 (Fig. 1H). Importantly, at no timedid we observe robust increases in Gb5 mRNA (SupplementaryFig. S1A), but we were able to block drug-induced Gb5 upregula-tion using cycloheximide, an inhibitor of translation (Supple-mentary Fig. S1B), indicating that Gb5 induction likely requiresnew protein synthesis. The apparent correlation between Gb5expression and pathologic sequelae that follow chemotherapyexposure in the heart led us to test whether Gb5 represents apreviously undiscovered mediator of cardiac damage induced bychemotherapeutic drugs.

Gb5 is required for chemotherapy-induced, ROS-dependentapoptosis

We tested the efficacy of multiple Gb5-targeted lentiviralshRNA constructs, identifying two capable of reducing bothbasal and chemotherapeutic-dependent Gb5 expression inVCM (Supplementary Fig. S2A and S2B). Viral knockdown ofGb5 led to a reduction in doxorubicin-induced ROS (Fig. 2A).The impact of Gb5 knockdown was comparable with theamelioration of ROS generation resulting from uncouplingof mitochondrial electron transport using the drug rotenoneor inhibition of active NADPH complexes with DPI with noadditive effects observed for the various drug combinations(Fig. 2A). Similar results were obtained in cells treated witheither paclitaxel or 5-FU (Supplementary Fig. S3A–S3E).

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research530

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Further, we were able to replicate results obtained via pharma-cologic inhibition with shRNA-specific knockdown of Nox4(Fig. 2B) implicating this particular isoform in chemotherapy-induced cardiac ROS generation. Together, these data indicatethat Gb5-mediated ROS generation likely involves Nox activa-tion and/or production of mitochondrial-dependent ROS.

Because there are several limitations to the use of CM-H2DCFDA fluorescent probes in detecting total ROS fromcells or tissue (31), among them an inability to differentiatebetween the various reactive species, we extended our analysisof Gb5-dependent ROS generation to include direct mea-surement of H2O2. We observed a dramatic reduction in

Figure 1.

Pharmacologically distinct cancer chemotherapeutics damage cardiac tissue via common pathologic mechanisms correlating with upregulation of Gb5.WT mice were treated chronically with doxorubicin (cumulative dose of 45 mg/kg i.p.), paclitaxel (cumulative dose of 85 mg/kg i.p.), 5-FU (cumulativedose of 200 mg/kg i.p.), or saline control over a period of 10 weeks. On the second day following the final injection, tissues were harvested for subsequentanalyses. A, Hematoxylin and eosin (H&E) staining (myofibrillar architecture) of cardiac tissue (day 7; scale bar, 20 mm; black arrows, areas of myofilamentdisarray), masson staining (fibrosis) of cardiac tissues (day 5; scale bar, 20 mm; black arrows, areas of collagen deposition), or representative hematoxylin andeosin–stained heart sections (day 10; scale bar, 50 mm) depicting myocyte hypertrophy. B, Heart weight normalized to body weight in animals (n ¼ 6) treatedwith chemotherapeutics as in A. In a separate study, WT mice were given a single dose of doxorubicin (20 mg/kg i.p.), paclitaxel (50 mg/kg i.p.), 5-FU (150mg/kg i.p.), or saline control. On days 1, 3, 5, 7, or 10 after injection, tissues were harvested for subsequent analyses. TNFa (C; n ¼ 6) or TGFb1 (D; n ¼ 6)levels in serum collected on days 1, 3, and 7 after injection. E, TUNEL staining was performed using formalin-fixed, paraffin-embedded cardiac tissuesections (n ¼ 6/group). Apoptotic cell nuclei were quantified from 10 random microscope fields. F, ROS generation (CM-H2DCFDA fluorescence) wasmeasured in cardiac tissue lysates (n ¼ 6/group) and expressed relative to control. G and H, Representative Western blot probing for Gb5 expression (day 5)and quantified in H (n ¼ 3; days 3, 5, and 10). � , P < 0.05; �� , P < 0.01; ��� , P < 0.001 compared with untreated control group. Data were analyzed by one-wayANOVA, with post hoc tests where appropriate. Data are presented as mean � SEM.

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 531

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

chemotherapeutic-induced H2O2 production (Fig. 2C). In addi-tion to direct toxic effects, superoxide can also react with nitricoxide (NO) to produce damaging reactive nitrogen species.Gb5-deficient VCM were largely protected from increased NOsynthase activity (Fig. 2D) and accumulation of the NO precursornitrite (Fig. 2E), actions likely to lessen overall production of

toxic peroxynitrite. In the presence of increasing ROS accumula-tion, multiple antioxidant mechanisms are also recruited toreduce oxidative stress. Gb5 knockdown reduced chemothera-py-induced upregulation of antioxidant enzymes glutathioneperoxidase (GPX; Supplementary Fig. S4A), involved in reductionof H2O2, and SOD, responsible for dismutation of superoxide

Figure 2.

Gb5 promotes chemotherapeutic drug–induced oxidative stress in isolated VCM. Isolated primary VCM were transduced with lentivirus shRNA targetingthe Gb5 gene (GNB5). Gb5 knockdown (Gb5KD) and control cells were then treated with doxorubicin (3 mmol/L, 24 hours), paclitaxel (100 nmol/L, 24 hours),or 5-FU (500 mmol/L, 24 hours). A, Total ROS generation (CM-H2DCFDA fluorescence) measured from cellular lysates (n ¼ 4) treated withchemotherapeutic drugs with or without a 1-hour pretreatment with DPI (1 mmol/L) or rotenone (200 nmol/L) to inhibit NADPH oxidase or decouplethe mitochondrial electron transport chain, respectively. VCM treated with ROS inducer H2O2 (200 mmol/L, 24 hours) were used as positive control. B, ROSgeneration was measured in VCM (n ¼ 6) transduced with viral constructs encoding shRNA targeting Gb5 and/or Nox4 and treated with chemotherapeuticdrugs as in A. Hydrogen peroxide (H2O2) levels (C; n ¼ 4); NO synthase activity (D; n ¼ 3), and nitrite levels (E; n ¼ 5) in VCM treated as above withdoxorubicin, paclitaxel, and 5-FU in the presence or absence of Gb5 knockdown. F, Isolated primary VCM transduced with lentivirus containing Gb5 orcontrol shRNA were pretreated with the ROS scavenger NAC (5 mmol/L) prior to challenge with chemotherapeutic drugs as in A. Cellular lysates (n ¼ 6)were then processed to measure apoptosis via detection of Annexin V–positive cells or caspase-3 activity. G, Control or Gb5 knockdown VCM werechallenged with doxorubicin � ROS scavenger pretreatment (Peg-SOD, 1,000 m/mL; Peg-Cat, 200 m/mL; L-NAME, 1 mmol/L; DMTU, 1 mmol/L) for 1 hour.Lysates (n ¼ 6) were processed to measure apoptosis expressed as the fold increase in cytoplasmic histone-associated DNA fragments (enrichmentfactor). � , P < 0.05; �� , P < 0.01; ��� , P < 0.001 compared with untreated control group; #, P < 0.05; †, P < 0.01; z, P < 0.001 compared with drug-treated group.Data were analyzed by one or two-way ANOVA, with post hoc tests where appropriate. Data are presented as mean � SEM.

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research532

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

(Supplementary Fig. S4B). The diminution in antioxidant mobi-lization in cardiomyocytes lacking functional Gb5 expression islikely due to a reduction in overall ROS levels.

In the face of overwhelming cellular stress, including accu-mulation of toxic ROS levels, damaged cells undergo apoptosis.We observed a significant reduction in chemotherapeutic drug–induced apoptosis and activation of the executioner caspase-3(Fig. 2F) in VCM lacking Gb5. These actions of Gb5 were entirelydependent on ROS generation as the ROS scavenger NACfailed to provide additive protection against chemotherapeu-tic-induced apoptotic cell death (Fig. 2F) implicating Gb5 as acritical mediator of ROS-dependent apoptosis in heart. GivenGb5 knockdown appeared to affect multiple ROS pathways, thequestion became which pathways were critical to chemother-apy-induced cell death. We found that scavenging of eithersuperoxide or H2O2 but not hydroxyl radicals or NO reduceddoxorubicin-induced apoptosis in a Gb5-dependent manner(Fig. 2F and G) consistent with activation of NADPH oxidase–and mitochondrial-dependent ROS generation leading to celldeath following drug treatment. Similar results were obtainedin VCM exposed to paclitaxel (Supplementary Fig. S3D) or 5-FU (Supplementary Fig. S3E).

Gb5 promotes chemotherapeutic-induced activation of DNAdamage signaling

Multiple apoptotic signaling cascades are believed to medi-ate ROS-induced cardiac damage including the ATM/p53pathway (32). Given that Gb5 binding partner RGS6 hasbeen implicated in doxorubicin-induced ATM/p53 activation(23, 24), we investigated the impact of Gb5 knockdown onmobilization of ATM/p53 in VCM. In cells where drug-induced Gb5 upregulation was blocked via shRNA-mediatedknockdown, phosphorylation-dependent activation of ATMand p53 as well as upregulation of p53 and the proapoptoticBcl-2 family member Bax were largely absent (Fig. 3A and B).Despite evidence that p53 is required for doxorubicin-induced heart damage (14), we observed only a modestreduction in chemotherapy-induced VCM apoptosis in cellstreated with the ATM inhibitor KU55933 (Fig. 3C) or cells inwhich ATM had been knocked down with targeted shRNA(Fig. 3D). Importantly, combined inhibition of ATM and Gb5produced no additional protection against chemotherapeutictoxicity than Gb5 knockdown alone (Fig. 3C and D), indi-cating these proteins function in the same pro-apoptoticpathway in VCM.

Figure 3.

Gb5 is required for chemotherapeutic drug–induced initiation of proapoptotic signaling cascades in heart. A, Isolated primary VCM were transduced withlentiviral vectors carrying control or Gb5-targeted shRNA. Cell lysates were processed to determine expression of Gb5, p-ATM, ATM, p-p53, p53, andBax by Western blotting 24 hours after treatment with doxorubicin (3 mmol/L), paclitaxel (100 nmol/L), or 5-FU (500 mmol/L). B, Densitometricquantification of p53 phosphorylation was performed from three independent experiments. C, Isolated primary VCM transduced with lentiviruscontaining Gb5 or control shRNA were pretreated with the ATM inhibitor KU55933 (10 mmol/L) prior to challenge with chemotherapeutics as in A. Cellularlysates (n ¼ 6) were then processed to measure apoptosis via detection of Annexin V–positive cells or caspase-3 activity. D, Apoptosis (fold increase incytoplasmic histone fragments) was measured in VCM (n ¼ 6) transduced with viral constructs encoding shRNA targeting Gb5 and/or ATM and treated withchemotherapeutic drugs as in A. �� , P < 0.01 and ��� , P < 0.001 compared with untreated control group; #, P < 0.05; †, P < 0.01; z, P < 0.001 compared with drug-treated group. Data were analyzed by one or two-way ANOVA with post hoc tests where appropriate. Data are presented as mean � SEM.

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 533

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Gb5 mediates mitochondrial dysfunction in VCM exposed tochemotherapeutic drugs

Because ATM activity accounted for only a small portion ofchemotherapeutic drug–induced apoptosis, we sought toinvestigate additional mechanisms that might be involvedin Gb5-dependent myocyte death. One potential pathway isthe intrinsic mitochondrial-dependent apoptotic signalingcascade. We observed a dramatic loss of mitochondrialmembrane potential (Dym) in cells exposed to all threechemotherapeutic agents that was restored following Gb5

knockdown (Fig. 4A). Gb5 knockdown was equally as effi-cacious in maintaining Dym as inhibition of the mitochon-drial permeability transition pore (mPTP) with cyclosporinA or blockade of MCU with Ru360 (Fig. 4A). Consistent withmaintenance Dym, Gb5 deficiency completely prevented che-motherapeutic-induced mitochondrial calcium accumula-tion (Fig. 4B), results similar to those obtained throughMCU inhibition. Indeed, MCU upregulation following che-motherapeutic exposure was abolished in VCM lacking Gb5(Fig. 4C).

Figure 4.

Gb5 facilitates mitochondrial dysfunction in VCM following chemotherapeutic drug exposure. Isolated primary VCM transduced with Gb5-targetedshRNA were pretreated with Ru360 (50 mmol/L) or cyclosporin A (0.2 mmol/L) for 45 minutes to selectively block mitochondrial Ca2þ uptake or opening ofthe mPTP, respectively. Cells were then challenged with doxorubicin (3 mmol/L), paclitaxel (100 nmol/L), or 5-FU (500 mmol/L) for 24 hours. Celllysates were used to measure mitochondrial membrane potential (Dym; n ¼ 3; A) and mitochondrial Ca2þ content (n ¼ 4; B). Immunoblots depicting proteinexpression of MCU and Gb5 (C) or Gb5, total CaMKII, ox-CaMKII, and p-CaMKII (D) from isolated primary VCM transduced with control or Gb5-targetedshRNA, followed by exposure to chemotherapeutics as in A. For all Western blots, actin served as a loading control and images depicted are representativeof three independent experiments. E, Isolated primary VCM transduced with lentivirus containing Gb5 or control shRNA were pretreated with theCaMKII inhibitor KN-93 (50 mmol/L) prior to drug challenge as in A. Cellular lysates (n ¼ 4) were then processed to measure apoptosis via detection ofAnnexin V–positive cells or caspase-3 activity. F, Apoptosis (fold increase in cytoplasmic histone fragments) was measured in VCM (n ¼ 6) transduced withviral constructs encoding shRNA targeting Gb5 and/or CaMKIId and treated with chemotherapeutic drugs as in A. ���, P < 0.001 compared with untreatedcontrol group; #, P < 0.05; †, P < 0.01; z, P < 0.001 compared with drug-treated group. Data were analyzed by one or two-way ANOVA, withpost hoc tests where appropriate. Data are presented as mean � SEM.

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research534

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

CaMKII has emerged as a critical regulator of mitochondrialcalcium, MCU activity, and mPTP in multiple models of cardiacinjury (33, 34). Because it was recently discovered that CaMKIIcan be directly activated by oxidation of the enzyme's regula-tory domain (35), we hypothesized that the link between Gb5and mitochondria might lie in ROS-dependent CaMKII acti-vation. Indeed, Gb5 knockdown prevented chemotherapeutic-induced CaMKII oxidation and phosphorylation in VCM(Fig. 4D). Further, inhibition of CaMKII via either pharmaco-logic (Fig. 4E) or genetic (Fig. 4F) means produced no addi-tional protection against chemotherapeutic-mediated VCMapoptosis as compared with Gb5 knockdown alone, indicatingthat CaMKII-dependent apoptosis requires Gb5. Together, thesedata identify Gb5 as a critical regulator of CaMKII-mediatedmitochondrial dysfunction.

The proapoptotic actions of ATM and CaMKII followingchemotherapy exposure require Gb5 and ROS

Based on the recent emergence of oxidative mechanisms ofactivation for both CaMKII (35) and ATM (36), we suspected thatthe ability of Gb5 to promote activation of both kinases requiredGb5-dependent ROS generation. Indeed, inhibition of Nox-derived ROS prevented chemotherapeutic-induced ATM phos-phorylation (Supplementary Fig. S5A) and CaMKII oxidation(Supplementary Fig. S5B) in VCM. The protective effects of Noxblockade were equivalent to and nonadditive with resultsobtained following Gb5 knockdown (Supplementary Fig. S5Cand S5D), indicating that Gb5 participates in a pathway linkingNox-dependent ROS to ATM and CaMKII. In keeping with loss ofkinase activation in the absence of ROS accumulation, ATM(Supplementary Fig. S6A) or CaMKII (Supplementary Fig. S6B)inhibition in the presence of the ROS scavenger NAC failed toreduce cellular apoptosis following chemotherapy exposure.Thus, it is likely that oxidative stress–induced myocyte toxicityin cells chemotherapeutic-treated requires bothATMandCaMKII.Of particular note, though singular inhibition of either ATM orCaMKII provided minimal protection against chemotherapeutic-induced apoptosis, combinedblockade of bothpathways reducedmyocyte death to levels observed in cells transduced with shGb5(Supplementary Fig. S6C). In addition, Gb5 knockdown in cellspretreated with CaMKII and ATM antagonists failed to furtherreduce VCM apoptosis (Supplementary Fig. S6C), implicatingGb5 as an upstream modulator of both pathways.

Gb5 deficiency protects cardiac cells from chemotherapy-induced cytokine accumulation and inflammation

Thus far, our investigation into the role of Gb5 in chemo-therapy-induced cardiotoxicity focused exclusively on myo-cyte-intrinsic proapoptotic mechanisms. However, followingpathologic stress maladaptive remodeling occurs in the myo-cardium characterized by myocyte hypertrophy, dysregulatedadaptations in extracellular matrix composition, inflamma-tion, and fibrosis. In addition to the reduction in cell deathobserved in VCM lacking Gb5 expression, the activity of MPO, abiomarker of inflammation, was decreased in Gb5-deficientmyocytes (Fig. 5A). We also observed a dramatic ameliorationof TNFa (Fig. 5B) release from cultured VCM lacking Gb5.Though implicated in transcriptional regulation of multiplecytokines and growth factors, knockdown of STAT3 in VCMfailed to affect levels of TNFa (Fig. 5B). In contrast, knockdownof either STAT3 or Gb5 reduced secretion of TGFb1 (Fig. 5C)

and IL10 (Fig. 5D) from VCF with the combined manipulationproviding no further reduction in cytokine levels followingchemotherapeutic drug treatment. The later result indicatesthat Gb5 functions in the same pathway as STAT3 to promoteVCF cytokine release. Similarly, inhibition of Nox failed toaffect TGFb1 levels in Gb5-deficient VCF (Fig. 5E), indicatingthat a combined mechanism involving ROS, Gb5, and STAT3promotes TGFb1 release from VCF exposed to chemotherapeu-tic drugs.

Gb5 promotes myocyte–fibroblast cross-talk followingchemotherapeutic drug exposure

Irrespective of Gb5 levels, STAT3 knockdown failed to affectchemotherapeutic drug–induced VCM apoptosis consistent withthe idea that STAT3-dependent factors contribute not to celldeath, but rather inflammation and fibrosis following doxorubi-cin, paclitaxel, or 5-FU treatment (Fig. 6A). Drug-induced releaseof TGFb1 was magnified in cocultured VCM and VCF with acorresponding increase in the impact of Gb5 loss (Fig. 6B),suggesting that communication between VCMand VCF promotesTGFb1 secretion and that these processes are regulated by Gb5.Interestingly, although knockdown of p53 in myocytes failed toaffect doxorubicin-dependent myofibroblast differentiation(Supplementary Fig. S7A), a TGFb1-dependent process wherebyfibroblasts take on characteristics of neighboringmyocytes detect-able utilizing myofibroblast markers such as a smooth muscleactin (aSMA; ref. 37), VCF-specific p53 knockdown reducedaSMA levels in cocultured cells (Supplementary Fig. S7B). Incontrast, drug-induced expression of aSMA was lost in cells inwhich Gb5 was knocked down in vitro (Fig. 6C) or in vivo (Sup-plementary Fig. S8A), and Gb5 loss in either cell type preventedaSMA induction (Supplementary Fig. S7A and S7B). Together,these data suggest that, unlike p53, Gb5 functions in both VCMand VCF to promote fibrosis implicating Gb5 as a modulator ofpathologic myocyte–fibroblast cross-talk in the chemotherapy-exposed myocardium.

In vivo, profibrotic and prohypertrophic factors upregulatefetal isoforms of genes responsible for regulation of cardiaccontractility in VCM including b-MHC and lead to the releaseof ANP, a peptide used as a diagnostic and prognostic markerin heart failure. Interestingly, all three chemotherapeutic drugstested failed to increase expression of ANP or b-MHC inisolated VCM (Fig. 6D; Supplementary Fig. S8B). However,in the presence of supportive fibroblasts, ANP, b-MHC, andTGFb1 levels were elevated, indicating that paracrine releaseof factors from neighboring fibroblasts is required for acqui-sition of this pathologic phenotype following chemotherapeu-tic drug exposure, a process facilitated by Gb5 as it was largelyprevented in cell lacking Gb5 expression (Fig. 6D; Supplemen-tary Fig. S8B and S8C). The TGFb1 antagonist GW788388also prevented chemotherapeutic-induced ANP upregulationto a similar extent as Gb5 knockdown (Fig. 6E; SupplementaryFig. S8D). Intriguingly, though STAT3 inhibition failed toaffect VCM apoptosis induced by chemotherapeutics, inhibi-tion of TGFb1 signaling resulted in a modest reduction in celldeath with no additive impact of Gb5 loss (Fig. 6F), indicatingthat TGFb1-dependent cell death is, again, mediated by Gb5.Thus, Gb5 appears to function as a critical modulator offibroblast–myocyte cross-talk, actions that likely facilitatepathologic hypertrophy and remodeling following chemother-apy exposure.

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 535

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Gb5 deficiency protects against cardiac damage andhypertrophy induced by chemotherapeutics in vivo

Because constitutive, global Gb5 knockout results in a host ofneurological abnormalities, altered brain development, andrunting (38–40), we employed a targeted approach to knock-down Gb5 specifically in heart: intracardiac injection of viralparticles carrying Gb5 shRNA. Both basal (Fig. 7A) and che-motherapeutic drug–induced Gb5 expressions (Fig. 7B) werereduced in animals infected with Gb5-targeted shRNA com-pared with control animals. Our methodology was specific toheart as we could detect loss of Gb5 protein and a correspond-ing reduction in activation of the ATM/p53 axis following drugtreatment in heart tissue and not breast (Fig. 7C). Importantly,elimination of cardiac Gb5 expression prevented STAT3 activa-tion (Fig. 7D), myofilament disarray (Fig. 7E), and cardiachypertrophy (Fig. 7F) following chronic chemotherapy expo-sure underscoring the importance of Gb5 in mediating thecardiotoxic actions of doxorubicin, 5-FU, and paclitaxel in anintact animal.

DiscussionDespite their proven capacity to improve long-term cancer

patient survival (1), the utility of anthracycline, taxane, andfluoropyrimidine chemotherapeutics is limited by irreversible

cardiac toxicity (2). Though the ability of these drugs toprevent the perpetuation of malignant cells relies on distinctcytotoxic mechanisms, our work provides evidence for sharedpathologic sequelae involved in the pathogenesis of chemother-apy-induced cardiomyopathy including accumulation of oxida-tive stress, myocyte apoptosis, and release of proinflammatoryand profibrotic cytokines from VCM and adjacent fibroblasts.Now, we have identified a novelmolecule, the atypical G proteinGb5, as an initiator of multiple pathologic signaling pathwaysresponsible for chemotherapy-induced heart damage (Fig. 7G).For all pathologic endpoints, doxorubicin, by far the mostcardiotoxic compound, was the most potent. Differences indose, duration of treatment, drug half-life, and, perhaps mostimportantly, the bioavailability of the drug at the site of actionmay explain the relative severity of doxorubicin-induced cardiacdamage compared with other compounds. Indeed, althoughdoxorubicin is cleared rapidly from the circulation, it accumu-lates in tissue where its half-life is anywhere from 20 to 48 hours.Both 5-FU and paclitaxel, conversely, are eliminated in minutesto hours. Of note, both 5-FU and paclitaxel are known to inducedysregulation in the cardiac conduction system, which couldresult from transient Gb5 induction as Gb5–RGS complexes havebeen identified as critical mediators of parasympathetic stimu-lation of heart whose loss results in increased heart rate vari-ability (25, 41–43).

Figure 5.

Gb5 promotes myocardial cytokine release from VCM and VCF following chemotherapeutic drug treatment. Isolated primary VCM were transduced withlentivirus containing Gb5 or STAT3-targeted shRNA where indicated. Cells were then treated with doxorubicin (3 mmol/L, 24 hours), paclitaxel (100 nmol/L,24 hours), or 5-FU (500 mmol/L, 24 hours). VCM lysates were used to measure the level of MPO (n ¼ 6; A) or TNFa (n ¼ 4; B). Isolated primaryVCF were transduced with viral vectors targeting Gb5 or STAT3 where indicated. Cells were then exposed to doxorubicin (3 mmol/L, 24 hours),paclitaxel (100 nmol/L, 24 hours), or 5-FU (500 mmol/L, 24 hours), and levels of TGFb1 (n ¼ 4; C) and IL10 (n ¼ 3; D) measured from cellular lysates.E, In a separate experiment, VCF (n ¼ 3) were pretreated with DPI (1 mmol/L) for 1 hour prior to chemotherapeutic drug exposure and TGFb1 levels measured.�� , P < 0.01 and ��� , P < 0.001 compared with untreated control group; #, P < 0.05; †, P < 0.01; z, P < 0.001 compared with drug-treated group.Data were analyzed by one or two-way ANOVA, with post hoc tests where appropriate. Data are presented as mean � SEM.

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research536

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Irrespective of the compound tested, we have demonstratedthat cardiac-specific Gb5 knockdown is sufficient to reducehypertrophy and maintain cardiac structure following chronicchemotherapy. Gb5 participates in multiple signaling cascadesto promote chemotherapy-induced heart damage. At this stage,it is unclear what upstream factor(s) lead to Gb5 accumulationin heart, but our data suggest that Gb5 facilitates both acutemyocyte death and oxidative stress that immediately follow

chemotherapy exposure as well as the process of adaptivefibrotic remodeling recruited thereafter to repair structuraldamage and maintain cardiac output and associated withrelease of cytokines from resident fibroblasts (19). Whether ornot oxidative stress precedes Gb5 upregulation, our in vitro datasuggest that Gb5 plays a critical role in promoting Nox- andmitochondrial-derived ROS generation following chemothera-peutic drug exposure in VCM. In so doing, Gb5 triggers

Figure 6.

Gb5 promotes pathologic myocyte–fibroblast cross-talk activated upon chemotherapeutic drug exposure. A, Cocultures of primary VCM and VCF weretransduced with control, Gb5, and/or STAT3 shRNA and then exposed to doxorubicin (3 mmol/L), paclitaxel (100 nmol/L), or 5-FU (500 mmol/L) for 24 hours.Lysates (n ¼ 6) were processed to measure apoptosis expressed as the fold increase in cytoplasmic histone-associated DNA fragments (enrichmentfactor). B, Isolated VCM or cocultured VCM and VCF were transduced with lentiviral shGb5 and treated with doxorubicin (3 mmol/L), paclitaxel (100 nmol/L),or 5-FU (500 mmol/L) for 48 hours. TGFb1 levels were measured from cell lysates (n ¼ 3–6) via ELISA. C, VCF were transduced with control or Gb5 shRNA andchallenged with doxorubicin (3 mmol/L), paclitaxel (100 nmol/L), or 5-FU (500 mmol/L) for 36 hours. Immunoblotting was then performed probing forinduction of the myofibroblast marker aSMA. D, VCM in isolation or cocultures of primary VCM and VCF were transduced with control or Gb5 shRNAand exposed to chemotherapeutics as in B. The expressions of hypertrophy markers ANP and b-MHC were determined by Western blotting. E, Control- orshGb5-transduced cocultures of primary VCM and VCF were pretreated with the TGFb1R inhibitor GW788388 (2 mmol/L) for 1 hour, challenged withdrugs as in A, and lysates used to probe for ANP protein levels. F, Apoptosis was measured in lysates (n ¼ 6) from shGb5-transduced cocultures ofprimary VCM and ventricular VCF pretreated with the TGFb1R inhibitor GW788388 (2 mmol/L) for 1 hour and then challenged with drugs as in A. AllWestern blots are representative of at least three independent experiments. Actin is shown as a loading control. �� , P < 0.01 and ��� , P < 0.001 comparedwith untreated control group; #, P < 0.05; †, P < 0.01; z, P < 0.001 compared with drug-treated group. Data were analyzed by one or two-way ANOVA, with posthoc tests where appropriate. Data are presented as mean � SEM.

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 537

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

activation of the kinases ATM and CaMKII—consistent withreports implicating Nox as an upstream modulator of bothkinases (18, 44). Thus, in the absence of Gb5, chemotherapy-induced, ATM/CaMKII-dependent mobilization of proapopto-tic machinery in cardiomyocytes is crippled. In addition toprotecting against apoptosis, Gb5 loss allows for maintenance

of Dym and retains basal MCU and mitochondrial Ca2þ levels,effects likely to preserve excitation–contraction coupling in themyocardium. Gb5 knockdown also reduces the release ofproinflammatory cytokines (TNFa) and profibrotic factors(TGFb1, IL10) from both VCM and VCF while simultaneouslypreventing the accumulation of myofibroblast and hypertrophy

Figure 7.

Gb5 knockdown protects against cardiac apoptosis, fibrosis, and hypertrophy induced by cancer chemotherapeutics in vivo. One-week-old WT micereceived a single intracardiac injection of 4.5 � 108 lentiviral vectors containing control or Gb5 targeted or control shRNA, followed by the chronic treatmentof either doxorubicin (cumulative dose of 45 mg/kg i.p.), paclitaxel (cumulative dose of 85 mg/kg i.p.), 5-FU (cumulative dose of 200 mg/kg i.p.), orsaline control over a period of 10 weeks. On the second day after the last injection, tissues were harvested for subsequent analyses. On day 5 after terminationof the chemotherapy regimen, cardiac tissue lysates were collected to measure the protein expression of Gb5 in either drug na€�ve (A) or drug or saline-treated(B) animals. In a separate experiment, intracardiac injections were performed in female WT mice, followed by a single acute treatment with doxorubicin (20mg/kg i.p.), paclitaxel (50 mg/kg i.p.), 5-FU (150 mg/kg i.p.), or saline control. On day 5 after injection, breast and cardiac tissue lysates were immunoblottedfor Gb5, p-p53, and p-ATM (C) or p-STAT3 (D). Actin served as a loading control for all immunoblots. Images shown are representative of at least threeindependent experiments. E, Representative hematoxylin and eosin (H&E)-stained heart sections (scale bar, 50 mm) depicting myocyte hypertrophy.Hematoxylin and eosin staining (myofibrillar architecture) of cardiac tissue (scale bar, 20 mm; black arrows, areas of myofilament disarray). F, Followingchronic drug exposure, hearts were collected and weighed with total cardiac tissue weight normalized to animal body weight. G, Mechanistic diagramdepicting the role of Gb5 in chemotherapy-induced cardiac damage. � , P < 0.05 and �� , P < 0.01 compared with untreated control group; #, P < 0.05 and†, P < 0.01 compared with drug-treated group. Data were analyzed by one or two-way ANOVA, with post hoc tests where appropriate. Data are presented asmean � SEM.

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research538

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

markers. Though necessary to maintain structural integrityfollowing cell loss, the persistence of these secretory factorscauses progressive, pathologic remodeling, a hallmark of thefailing heart (45). The overall protective impact of Gb5 defi-ciency in chemotherapy-exposed cardiac tissue, thus, involvesthe convergence of signaling in and between VCF and VCM(Fig. 7G).

Gb5 appears unique in its capacity to not only act upstream ofmultiple pathologic factors, but also participate in profibroticsignaling in multiple cardiac cell types. In contrast, althoughcardiomyocyte-specific deletion of p53 protects against acutedoxorubicin-induced myocyte loss (14), inhibition of p53 failsto affect doxorubicin-induced cardiac fibrosis (26) and actuallyexacerbates late-stage cardiac damage (46). Although our datasuggest that p53 acts in fibroblasts to promote myofibroblastdifferentiation, it is clear that the predominant action of myocytep53 is in mediation of initial cell death following doxorubicintreatment. Cardioprotective strategies aimed at amelioration ofchemotherapy-induced cardiotoxicity that focus solely on myo-cyte intrinsic signaling ignore the role of VCF, which also con-tribute substantially to the malignant secretome established fol-lowing sustained chemotherapeutic drug exposure. A recentobservation that ATM deletion in VCF and not VCM providedsubstantial protection against doxorubicin cardiotoxicity (47)underscores this point. Conversely, in our study, Gb5-dependentSTAT3-mediated release of profibrotic factors fromVCFpromotedmyofibroblast differentiation, but failed to affect myocyte apo-ptosis. Theobservation thatGb5-dependentmodulationof STAT3activity is restricted to cardiac fibroblasts is important as cardio-myocyte-specific ablation of STAT3 exacerbates doxorubicin-induced myocyte apoptosis and worsens cardiac function (48)while overexpression is protective (49).

Unlike other G protein b subunits, Gb5 fails heterotrimerizewith Ga/Gg proteins coupled to membrane receptors. Instead,Gb5 forms costabilizing complexes with R7 family RGS proteins(20–22). RGS6 is the predominant R7 RGS protein expressed inheart (25, 43). As Gb5 expression is not completely absent in thehearts of RGS6�/� mice, another family member is likely alsoexpressed (25, 42). Thus, phenotypes observed in Gb5-deficientcardiomyocytes may be similar, but not identical to resultsobtained from cells lacking RGS6. Indeed, although RGS6 hasbeen heavily implicated in doxorubicin-induced activation of theDNA damage signaling pathway and subsequent apoptosis (23,24), no studies have investigated participation of RGS6 in addi-tional myocyte-intrinsic pathologic mechanisms, cardiac damageinduced by other chemotherapeutics, or myofibroblast recruit-ment and the release of proinflammatory and profibrotic cyto-kines. Thus, it is likely thatGb5modulates chemotherapy-inducedcardiotoxicity via a combination of RGS6-dependent and -inde-pendent mechanisms.

ThemechanismsdescribedhereinwherebyGb5 promotes ROS-dependent cellular signaling are a novel function that does notnecessarily involve canonical G protein signaling. Identificationof Gb5 as an essential signaling moiety in the ROS/apoptosispathway in cardiomyocytes as well as proinflammatory andprofibrotic pathways in adjacent fibroblasts advances our under-

stating of the pathogenesis of chemotherapy-induced cardiomy-opathy and suggests that drugs designed to inhibit or destabilizeGb5 may have some utility in adjuvant therapies designed tomitigate the cardiotoxic actions of chemotherapeutic regimens.It is important to note that whatever means might be employedto disrupt Gb5 action (e.g. blockade of the costabilizing RGSprotein/Gb5 interaction interface) would be most efficacious ifconfined to the heart as RGS6, a Gb5-binding partner, mediatesthe cancer killing actions of doxorubicin in certain cancer celltypes (24, 50). Indeed, we believe it likely that the heart isparticularly sensitive to the cytotoxic actions of chemotherapeu-tics as it exhibits the highest expression of R7 family RGS proteinsoutside of the brain (25, 43). Given the frequency with whichboth RGS6 and Gb5 are downregulated in multiple humanmalignancies (50–52), it is also possible that, for cancers lackingfunctional Gb5, global inhibition might be a viable means ofprotecting the heart against chemotherapy-induced damage.Future studies aimed at defining the exact spatial and temporalfacets of Gb5 action in chemotherapy-induced cardiotoxicity mayprovide additional insight into a therapeutic window in whichtargeting Gb5 may prove clinically efficacious in ameliorating thedamaging actions of chemotherapeutics onnonmalignant tissues.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: S. Chakraborti, M. Das, S. Ghosh, A. Stewart, B. MaityAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): S. Chakraborti, A. Pramanick, S. Ghosh, B. MaityAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): S. Chakraborti, A. Pramanick, S. Saha, A. Stewart,B. MaityWriting, review, and/or revision of the manuscript: S. Chakraborti,A. Pramanick, S. Saha, A. Ray Chaudhuri, A. Stewart, B. MaityStudy supervision: A. Ray Chaudhuri, B. Maity

AcknowledgmentsWe thank Dr. Sougata Roy Chowdhury (INSPIRE Faculty, University of

Calcutta), Department of Zoology, University of Calcutta and University ofKalyani and Centre of Biomedical Research for all the help. We also thankprofessor Ganesh Pandey (Director, Centre of Biomedical Research), Dr. Mee-nakshi Munshi, Department of Biotechnology and Department of Science andTechnology, Government of India. Finally, our sincere gratitude to professorMadhusudan Das (Dean of Science) and Dr. Sujoy Ghosh, Department ofZoology, University of Calcutta, for their generous help, inspiration, andcontinuous support. This work was supported by Ramalingaswami Fellowship(BT/RLF/Re-entry/15/2013), Department of Biotechnology, andDepartment ofScience and Technology, SERB (EMR/2016/006873), Government of India (toB.Maity). B.Maitywas previously employed at theUniversity of Calcutta asDBTRamalingaswami Fellow.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received May 4, 2017; revised September 8, 2017; accepted November 13,2017; published OnlineFirst November 15, 2017.

References1. Shakir DK, Rasul KI. Chemotherapy induced cardiomyopathy: pathogen-

esis, monitoring and management. J Clin Med Res 2009;1:8–12.2. Albini A, Pennesi G, Donatelli F, Cammarota R, De Flora S, Noonan

DM. Cardiotoxicity of anticancer drugs: the need for cardio-oncology

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 539

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

and cardio-oncological prevention. J Natl Cancer Inst 2010;102:14–25.

3. Yeh ET, Tong AT, Lenihan DJ, Yusuf SW, Swafford J, Champion C, et al.Cardiovascular complications of cancer therapy: diagnosis, pathogen-esis, and management. Circulation 2004;109:3122–31.

4. TukenovaM,GuiboutC,OberlinO,Doyon F,Mousannif A,HaddyN, et al.Role of cancer treatment in long-term overall and cardiovascular mortalityafter childhood cancer. J Clin Oncol 2010;28:1308–15.

5. Alter P, Herzum M, Soufi M, Schaefer JR, Maisch B. Cardiotoxicity of5-fluorouracil. Cardiovasc Hematol Agents Med Chem 2006;4:1–5.

6. Elliott P. Pathogenesis of cardiotoxicity induced by anthracyclines. SeminOncol 2006;33:S2–7.

7. Schimmel KJ, Richel DJ, van den Brink RB, Guchelaar HJ. Cardiotoxicity ofcytotoxic drugs. Cancer Treat Rev 2004;30:181–91.

8. Carvalho RA, Sousa RP, Cadete VJ, Lopaschuk GD, Palmeira CM, Bjork JA,et al. Metabolic remodeling associated with subchronic doxorubicin car-diomyopathy. Toxicology 2010;270:92–8.

9. Ichikawa Y, Ghanefar M, Bayeva M, Wu R, Khechaduri A, Naga Prasad SV,et al. Cardiotoxicity of doxorubicin is mediated through mitochondrialiron accumulation. J Clin Invest 2014;124:617–30.

10. Neilan TG, Blake SL, Ichinose F, Raher MJ, Buys ES, Jassal DS, et al.Disruption of nitric oxide synthase 3 protects against the cardiac injury,dysfunction, and mortality induced by doxorubicin. Circulation 2007;116:506–14.

11. Deng S, Kruger A, Kleschyov AL, Kalinowski L, Daiber A, Wojnowski L.Gp91phox-containing NAD(P)H oxidase increases superoxideformation by doxorubicin and NADPH. Free Radic Biol Med 2007;42:466–73.

12. Zhao Y, McLaughlin D, Robinson E, Harvey AP, Hookham MB, ShahAM, et al. Nox2 NADPH oxidase promotes pathologic cardiac remodel-ing associated with Doxorubicin chemotherapy. Cancer Res 2010;70:9287–97.

13. Yoshida M, Shiojima I, Ikeda H, Komuro I. Chronic doxorubicin cardio-toxicity is mediated by oxidative DNA damage-ATM-p53-apoptosis path-way and attenuated by pitavastatin through the inhibition of Rac1 activity.J Mol Cell Cardiol 2009;47:698–705.

14. Shizukuda Y, Matoba S, Mian OY, Nguyen T, Hwang PM. Targeteddisruption of p53 attenuates doxorubicin-induced cardiac toxicity inmice.Mol Cell Biochem 2005;273:25–32.

15. Liu X, Chua CC, Gao J, Chen Z, Landy CL, Hamdy R, et al. Pifithrin-alpha protects against doxorubicin-induced apoptosis and acutecardiotoxicity in mice. Am J Physiol Heart Circ Physiol 2004;286:H933–9.

16. Sag CM,Kohler AC, AndersonME, Backs J,Maier LS. CaMKII-dependent SRCa leak contributes to doxorubicin-induced impaired Ca handling inisolated cardiac myocytes. J Mol Cell Cardiol 2011;51:749–59.

17. Schulman H, Anderson ME. Ca/Calmodulin-dependent protein kinase IIin heart failure. Drug Discov Today Dis Mech 2010;7:e117–e22.

18. He BJ, JoinerML, SinghMV, Luczak ED, Swaminathan PD, Koval OM, et al.Oxidation of CaMKII determines the cardiotoxic effects of aldosterone.Nat Med 2011;17:1610–8.

19. Schunke KJ, Coyle L, Merrill GF, Denhardt DT. Acetaminophen attenuatesdoxorubicin-induced cardiac fibrosis via osteopontin and GATA4 regula-tion: reduction of oxidant levels. J Cell Physiol 2013;228:2006–14.

20. Witherow DS, Wang Q, Levay K, Cabrera JL, Chen J, Willars GB, et al.Complexes of theGprotein subunit gbeta 5with the regulators ofGproteinsignaling RGS7 and RGS9. Characterization in native tissues and intransfected cells. J Biol Chem 2000;275:24872–80.

21. Snow BE, Betts L, Mangion J, Sondek J, Siderovski DP. Fidelity of Gprotein beta-subunit association by the G protein gamma-subunit-likedomains of RGS6, RGS7, and RGS11. Proc Natl Acad Sci U S A 1999;96:6489–94.

22. Posner BA, Gilman AG, Harris BA. Regulators of G protein signaling 6and 7. Purification of complexes with gbeta5 and assessment of theireffects on g protein-mediated signaling pathways. J Biol Chem 1999;274:31087–93.

23. Yang J, Maity B, Huang J, Gao Z, Stewart A, Weiss RM, et al. G-proteininactivator RGS6mediates myocardial cell apoptosis and cardiomyopathycaused by doxorubicin. Cancer Res 2013;73:1662–7.

24. Huang J, Yang J, Maity B, Mayuzumi D, Fisher RA. Regulator of Gprotein signaling 6 mediates doxorubicin-induced ATM and p53 acti-

vation by a reactive oxygen species-dependent mechanism. Cancer Res2011;71:6310–9.

25. Posokhova E, Wydeven N, Allen KL, Wickman K, Martemyanov KA. RGS6/Gbeta5 complex accelerates IKACh gating kinetics in atrial myocytesand modulates parasympathetic regulation of heart rate. Circ Res 2010;107:1350–4.

26. Feridooni T, Hotchkiss A, Remley-Carr S, Saga Y, Pasumarthi KB. Cardi-omyocyte specific ablation of p53 is not sufficient to block doxorubicininduced cardiac fibrosis and associated cytoskeletal changes. PLoS One2011;6:e22801.

27. Der Sarkissian S, Grobe JL, Yuan L, NarielwalaDR,Walter GA, KatovichMJ,et al. Cardiac overexpression of angiotensin converting enzyme 2 protectsthe heart from ischemia-induced pathophysiology. Hypertension 2008;51:712–8.

28. Long CS, Henrich CJ, Simpson PC. A growth factor for cardiac myocytes isproduced by cardiac nonmyocytes. Cell Regul 1991;2:1081–95.

29. Stewart A,Maity B, Anderegg SP, Allamargot C, Yang J, Fisher RA. Regulatorof G protein signaling 6 is a critical mediator of both reward-relatedbehavioral and pathological responses to alcohol. Proc Natl Acad SciU S A 2015;112:E786–95.

30. Flohe L, Gunzler WA. Assays of glutathione peroxidase. Methods Enzymol1984;105:114–21.

31. Kalyanaraman B, Darley-Usmar V, Davies KJ, Dennery PA, Forman HJ,Grisham MB, et al. Measuring reactive oxygen and nitrogen species withfluorescent probes: challenges and limitations. Free Radic Biol Med2012;52:1–6.

32. Minotti G, Menna P, Salvatorelli E, Cairo G, Gianni L. Anthracyclines:molecular advances and pharmacologic developments in antitumor activ-ity and cardiotoxicity. Pharmacol Rev 2004;56:185–229.

33. AndersonME.Oxidant stress promotes disease by activating CaMKII. JMolCell Cardiol 2015;89:160–7.

34. Joiner ML, Koval OM, Li J, He BJ, Allamargot C, Gao Z, et al. CaMKIIdetermines mitochondrial stress responses in heart. Nature 2012;491:269–73.

35. Erickson JR, Joiner ML, Guan X, Kutschke W, Yang J, Oddis CV, et al. Adynamic pathway for calcium-independent activation ofCaMKII bymethi-onine oxidation. Cell 2008;133:462–74.

36. Guo Z, Kozlov S, Lavin MF, Person MD, Paull TT. ATM activation byoxidative stress. Science 2010;330:517–21.

37. Scharenberg MA, Pippenger BE, Sack R, Zingg D, Ferralli J, Schenk S,et al. TGF-beta-induced differentiation into myofibroblastsinvolves specific regulation of two MKL1 isoforms. J Cell Sci 2014;127:1079–91.

38. Wang Q, Levay K, Chanturiya T, Dvoriantchikova G, Anderson KL, BiancoSD, et al. Targeted deletion of one or two copies of the G protein betasubunit Gbeta5 gene has distinct effects on body weight and behavior inmice. FASEB J 2011;25:3949–57.

39. Zhang JH, PandeyM, Seigneur EM, Panicker LM, Koo L, SchwartzOM, et al.Knockout of G protein beta5 impairs brain development and causesmultiple neurologic abnormalities in mice. J Neurochem 2011;119:544–54.

40. Xie K, Ge S, Collins VE, Haynes CL, Renner KJ, Meisel RL, et al. Gbeta5-RGS complexes are gatekeepers of hyperactivity involved in controlof multiple neurotransmitter systems. Psychopharmacology 2012;219:823–34.

41. Posokhova E, Ng D, Opel A, Masuho I, Tinker A, Biesecker LG, et al.Essential role of the m2R-RGS6-IKACh pathway in controlling intrinsicheart rate variability. PLoS One 2013;8:e76973.

42. WydevenN, Posokhova E, Xia Z, Martemyanov KA,Wickman K. RGS6, butnot RGS4, is the dominant regulator of G protein signaling (RGS) mod-ulator of the parasympathetic regulation of mouse heart rate. J Biol Chem2014;289:2440–9.

43. Yang J, Huang J, Maity B, Gao Z, Lorca RA, Gudmundsson H, et al. RGS6, amodulator of parasympathetic activation in heart. Circ Res 2010;107:1345–9.

44. Weyemi U, Redon CE, Aziz T, Choudhuri R, Maeda D, Parekh PR, et al.NADPH oxidase 4 is a critical mediator in Ataxia telangiectasia disease.Proc Natl Acad Sci U S A 2015;112:2121–6.

45. Weber KT, Sun Y, Bhattacharya SK, Ahokas RA, Gerling IC. Myofibroblast-mediated mechanisms of pathological remodelling of the heart. Nat RevCardiol 2013;10:15–26.

Chakraborti et al.

Cancer Res; 78(2) January 15, 2018 Cancer Research540

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

46. Zhu W, Zhang W, Shou W, Field LJ. P53 inhibition exacerbates late-stageanthracycline cardiotoxicity. Cardiovasc Res 2014;103:81–9.

47. Zhan H, Aizawa K, Sun J, Tomida S, Otsu K, Conway SJ, et al. Ataxiatelangiectasia mutated in cardiac fibroblasts regulates doxorubicin-induced cardiotoxicity. Cardiovasc Res 2016;110:85–95.

48. Jacoby JJ, Kalinowski A, Liu MG, Zhang SS, Gao Q, Chai GX, et al.Cardiomyocyte-restricted knockout of STAT3 results in higher sensitiv-ity to inflammation, cardiac fibrosis, and heart failure with advancedage. Proc Natl Acad Sci U S A 2003;100:12929–34.

49. Kunisada K, Negoro S, Tone E, Funamoto M, Osugi T, Yamada S, et al.Signal transducer and activator of transcription 3 in the heart trans-duces not only a hypertrophic signal but a protective signal against

doxorubicin-induced cardiomyopathy. Proc Natl Acad Sci U S A2000;97:315–9.

50. Maity B, Yang J, Huang J, Askeland RW, Bera S, Fisher RA. Regulator of Gprotein signaling 6 (RGS6) induces apoptosis via a mitochondrial-depen-dent pathway not involving its GTPase-activating protein activity. J BiolChem 2011;286:1409–19.

51. Maity B, Stewart A,O'Malley Y, Askeland RW, Sugg SL, Fisher RA. Regulatorof G protein signaling 6 is a novel suppressor of breast tumor initiation andprogression. Carcinogenesis 2013;34:1747–55.

52. Yang J, Platt LT, Maity B, Ahlers KE, Luo Z, Lin Z, et al. RGS6 is an essentialtumor suppressor that prevents bladder carcinogenesis by promoting p53activation and DNMT1 downregulation. Oncotarget 2016;7:69159–72.

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 541

Gb5 Mediates Chemotherapy-Induced Cardiotoxicity

Cancer Research. by guest on September 2, 2020. Copyright 2017 American Association forhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from