Cardiomyocyte-specificIκB kinase (IKK)/NF- B …Cardiomyocyte-specificIκB kinase (IKK)/NF-κB...

6
Cardiomyocyte-specicIκB kinase (IKK)/NF-κB activation induces reversible inammatory cardiomyopathy and heart failure Harald J. Maier a , Tobias G. Schips a,1 , Astrid Wietelmann b , Marcus Krüger b , Cornelia Brunner a , Martina Sauter c , Karin Klingel c , Thomas Böttger b , Thomas Braun b , and Thomas Wirth a,2 a Institute of Physiological Chemistry, University of Ulm, 89081 Ulm, Germany; b Max Planck Institute for Heart and Lung Research, 61231 Bad Nauheim, Germany; and c Department of Molecular Pathology, University Hospital of Tübingen, 72076 Tübingen, Germany Edited by Peter K. Vogt, The Scripps Research Institute, La Jolla, CA, and approved June 6, 2012 (received for review October 9, 2011) Inammation is a major factor in heart disease. IκB kinase (IKK) and its downstream target NF-κB are regulators of inammation and are activated in cardiac disorders, but their precise contributions and targets are unclear. We analyzed IKK/NF-κB function in the heart by a gain-of-function approach, generating an inducible transgenic mouse model with cardiomyocyte-specic expression of constitu- tively active IKK2. In adult animals, IKK2 activation led to inamma- tory dilated cardiomyopathy and heart failure. Transgenic hearts showed inltration with CD11b + cells, brosis, fetal reprogramming, and atrophy of myocytes with strong constitutively active IKK2 ex- pression. Upon transgene inactivation, the disease was reversible even at an advanced stage. IKK-induced cardiomyopathy was depen- dent on NF-κB activation, as in vivo expression of IκBα superrepres- sor, an inhibitor of NF-κB, prevented the development of disease. Gene expression and proteomic analyses revealed enhanced expres- sion of inammatory cytokines, and an IFN type I signature with activation of the IFN-stimulated gene 15 (ISG15) pathway. In that respect, IKK-induced cardiomyopathy resembled Coxsackievirus-in- duced myocarditis, during which the NF-κB and ISG15 pathways were also activated. Vice versa, in cardiomyocytes lacking the regu- latory subunit of IKK (IKKγ/NEMO), the induction of ISG15 was at- tenuated. We conclude that IKK/NF-κB activation in cardiomyocytes is sufcient to cause cardiomyopathy and heart failure by inducing an excessive inammatory response and myocyte atrophy. transcription factors | transgenic mice I nammation is a major pathogenetic factor for cardiovascular diseases. In the heart, inammation is most prominently ob- served in myocarditis, but also in other cardiac pathologic pro- cesses, such as postinfarct remodeling or dilated cardiomyopathy. The IKK/NF-κB signaling pathway is an important regulator of inammation, but also of many other biological processes, in- cluding cell growth and cell survival. IKK/NF-κB activation has been implicated in myocarditis, cardiac hypertrophy, ischemia/ reperfusion damage and myocardial infarction, and heart failure (1, 2). However, the signicance of its activation remains unclear: depending on the context, IKK/NF-κB activation may be pro- tective or damaging, or a mere epiphenomenon. Although loss-of- function studies in genetically modied mice have provided some insight into IKK/NF-κB signaling in the heart, many aspects are still a matter of controversy (3). NF-κB is a dimeric transcription factor, consisting of different combinations of the subunits p50, p52, RelA, RelB, and c-Rel. In most cell types, NF-κB is kept inactive in the cytoplasm by in- hibitory κB proteins (i.e., IκBs). Upon upstream eventsoften related to some form of physical, chemical, or biological stressthe IκB kinase (IKK) complex, consisting of the catalytic sub- units IKK1 (IKKα) and IKK2 (IKKβ) and the regulatory subunit NEMO (IKKγ), is activated. This results in the phosphorylation of IκB, followed by its ubiquitination and proteasomal degra- dation. Thus, NF-κB dimers are released and translocate to the nucleus, where they regulate the transcription of a diverse range of target genes (4). IKK/NF-κB is not required for normal heart development (57). However, for maintaining the adult heart in a healthy state, a baseline NF-κB activity may be required for antioxidant pro- tection (7). The function of IKK/NF-κB during ischemia/reper- fusion injury, myocardial infarction, and subsequent remodeling processes is controversial (1, 2). As to myocyte growth, IKK/NF- κB activation in vivo is required for certain forms of hypertrophy (811), but not for others (6, 7). In respect to myocarditis, IKK/NF-κB is likely of importance, but its function may depend on the etiology of the disease (1215). One of the main causes of myocarditis in humansand, presumably, its sequela, dilated cardiomyopathyis infection with enteroviruses. This disease can be modeled in mice by in- fection with Coxsackievirus B3, which leads to acute myocarditis and dilated cardiomyopathy in susceptible mouse strains (16). However, the activation status and the contribution of IKK/ NF-κB in this model have not been studied thus far. In light of these controversial results, we decided to analyze cardiac IKK/NF-κB by a gain-of-function approach. We report here that IKK/NF-κB activation in cardiomyocytes induces myo- carditis, inammatory dilated cardiomyopathy, and muscle ber atrophy, with a strong IFN type I response similar to the one observed in viral myocarditis. Results Expression of Constitutively Active IKK2 in the Heart Induces Cardio- myopathy. Activation of IKK/NF-κΒ in cardiomyocytes was achieved by crossing mice expressing the tetracycline trans- activator (tTA) under the control of the α-myosin heavy chain (MyHC) promoter (17) with mice bearing a tTA-responsive constitutively active IKK2 (IKK2-CA) allele (18), thus generat- ing double-transgenic IKK MyHC mice (Fig. S1A). The transgenic system was activated in adult animals by removing doxycycline from the drinking water (Fig. S1B). Expression of IKK2-CA was detected in all areas of the heart, but not in other organs (Fig. 1 A and B and Fig. S1C). IKK-specic kinase activity was strongly enhanced in the hearts of IKK MyHC mice and led to a strong increase in NF-κBspecic DNA-binding activity (Fig. 1 C and D). Whereas the expression levels of most NF-κB subunits were unchanged, RelB expression was up-regulated (Fig. S1D). Six to 8 wk after doxycycline discontinuation, IKK MyHC mice began to show signs of heart failure. Hearts were enlarged strongly, with dilation of ventricles and atria (Fig. 1E). In some Author contributions: H.J.M., T. Braun, and T.W. designed research; H.J.M., T.S., A.W., M.K., C.B., M.S., and T. Böttger performed research; H.J.M., K.K., T. Braun, and T.W. analyzed data; and H.J.M. and T.W. wrote the paper. The authors declare no conict of interest. This article is a PNAS Direct Submission. 1 Present address: Howard Hughes Medical Institute, Cincinnati Childrens Hospital Med- ical Center, Cincinnati 45229, OH. 2 To whom correspondence should be addressed. E-mail: [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1116584109/-/DCSupplemental. 1179411799 | PNAS | July 17, 2012 | vol. 109 | no. 29 www.pnas.org/cgi/doi/10.1073/pnas.1116584109 Downloaded by guest on January 29, 2020

Transcript of Cardiomyocyte-specificIκB kinase (IKK)/NF- B …Cardiomyocyte-specificIκB kinase (IKK)/NF-κB...

Cardiomyocyte-specific IκB kinase (IKK)/NF-κBactivation induces reversible inflammatorycardiomyopathy and heart failureHarald J. Maiera, Tobias G. Schipsa,1, Astrid Wietelmannb, Marcus Krügerb, Cornelia Brunnera, Martina Sauterc,Karin Klingelc, Thomas Böttgerb, Thomas Braunb, and Thomas Wirtha,2

aInstitute of Physiological Chemistry, University of Ulm, 89081 Ulm, Germany; bMax Planck Institute for Heart and Lung Research, 61231 Bad Nauheim,Germany; and cDepartment of Molecular Pathology, University Hospital of Tübingen, 72076 Tübingen, Germany

Edited by Peter K. Vogt, The Scripps Research Institute, La Jolla, CA, and approved June 6, 2012 (received for review October 9, 2011)

Inflammation is a major factor in heart disease. IκB kinase (IKK) andits downstream target NF-κB are regulators of inflammation and areactivated in cardiac disorders, but their precise contributions andtargets are unclear. We analyzed IKK/NF-κB function in the heartby a gain-of-function approach, generating an inducible transgenicmouse model with cardiomyocyte-specific expression of constitu-tively active IKK2. In adult animals, IKK2 activation led to inflamma-tory dilated cardiomyopathy and heart failure. Transgenic heartsshowed infiltrationwith CD11b+ cells, fibrosis, fetal reprogramming,and atrophy of myocytes with strong constitutively active IKK2 ex-pression. Upon transgene inactivation, the disease was reversibleeven at an advanced stage. IKK-induced cardiomyopathywasdepen-dent on NF-κB activation, as in vivo expression of IκBα superrepres-sor, an inhibitor of NF-κB, prevented the development of disease.Gene expression and proteomic analyses revealed enhanced expres-sion of inflammatory cytokines, and an IFN type I signature withactivation of the IFN-stimulated gene 15 (ISG15) pathway. In thatrespect, IKK-induced cardiomyopathy resembled Coxsackievirus-in-duced myocarditis, during which the NF-κB and ISG15 pathwayswere also activated. Vice versa, in cardiomyocytes lacking the regu-latory subunit of IKK (IKKγ/NEMO), the induction of ISG15 was at-tenuated. We conclude that IKK/NF-κB activation in cardiomyocytesis sufficient to cause cardiomyopathy and heart failure by inducingan excessive inflammatory response and myocyte atrophy.

transcription factors | transgenic mice

Inflammation is a major pathogenetic factor for cardiovasculardiseases. In the heart, inflammation is most prominently ob-

served in myocarditis, but also in other cardiac pathologic pro-cesses, such as postinfarct remodeling or dilated cardiomyopathy.The IKK/NF-κB signaling pathway is an important regulator ofinflammation, but also of many other biological processes, in-cluding cell growth and cell survival. IKK/NF-κB activation hasbeen implicated in myocarditis, cardiac hypertrophy, ischemia/reperfusion damage and myocardial infarction, and heart failure(1, 2). However, the significance of its activation remains unclear:depending on the context, IKK/NF-κB activation may be pro-tective or damaging, or amere epiphenomenon. Although loss-of-function studies in genetically modified mice have provided someinsight into IKK/NF-κB signaling in the heart, many aspects arestill a matter of controversy (3).NF-κB is a dimeric transcription factor, consisting of different

combinations of the subunits p50, p52, RelA, RelB, and c-Rel. Inmost cell types, NF-κB is kept inactive in the cytoplasm by in-hibitory κB proteins (i.e., IκBs). Upon upstream events—oftenrelated to some form of physical, chemical, or biological stress—the IκB kinase (IKK) complex, consisting of the catalytic sub-units IKK1 (IKKα) and IKK2 (IKKβ) and the regulatory subunitNEMO (IKKγ), is activated. This results in the phosphorylationof IκB, followed by its ubiquitination and proteasomal degra-dation. Thus, NF-κB dimers are released and translocate to thenucleus, where they regulate the transcription of a diverse rangeof target genes (4).

IKK/NF-κB is not required for normal heart development (5–7). However, for maintaining the adult heart in a healthy state,a baseline NF-κB activity may be required for antioxidant pro-tection (7). The function of IKK/NF-κB during ischemia/reper-fusion injury, myocardial infarction, and subsequent remodelingprocesses is controversial (1, 2). As to myocyte growth, IKK/NF-κB activation in vivo is required for certain forms of hypertrophy(8–11), but not for others (6, 7).In respect to myocarditis, IKK/NF-κB is likely of importance,

but its function may depend on the etiology of the disease (12–15). One of the main causes of myocarditis in humans—and,presumably, its sequela, dilated cardiomyopathy—is infectionwith enteroviruses. This disease can be modeled in mice by in-fection with Coxsackievirus B3, which leads to acute myocarditisand dilated cardiomyopathy in susceptible mouse strains (16).However, the activation status and the contribution of IKK/NF-κB in this model have not been studied thus far.In light of these controversial results, we decided to analyze

cardiac IKK/NF-κB by a gain-of-function approach. We reporthere that IKK/NF-κB activation in cardiomyocytes induces myo-carditis, inflammatory dilated cardiomyopathy, and muscle fiberatrophy, with a strong IFN type I response similar to the oneobserved in viral myocarditis.

ResultsExpression of Constitutively Active IKK2 in the Heart Induces Cardio-myopathy. Activation of IKK/NF-κΒ in cardiomyocytes wasachieved by crossing mice expressing the tetracycline trans-activator (tTA) under the control of the α-myosin heavy chain(MyHC) promoter (17) with mice bearing a tTA-responsiveconstitutively active IKK2 (IKK2-CA) allele (18), thus generat-ing double-transgenic IKKMyHC mice (Fig. S1A). The transgenicsystem was activated in adult animals by removing doxycyclinefrom the drinking water (Fig. S1B). Expression of IKK2-CA wasdetected in all areas of the heart, but not in other organs (Fig. 1A and B and Fig. S1C). IKK-specific kinase activity was stronglyenhanced in the hearts of IKKMyHC mice and led to a strongincrease in NF-κB–specific DNA-binding activity (Fig. 1 C andD). Whereas the expression levels of most NF-κB subunits wereunchanged, RelB expression was up-regulated (Fig. S1D).Six to 8 wk after doxycycline discontinuation, IKKMyHC mice

began to show signs of heart failure. Hearts were enlargedstrongly, with dilation of ventricles and atria (Fig. 1E). In some

Author contributions: H.J.M., T. Braun, and T.W. designed research; H.J.M., T.S., A.W., M.K.,C.B., M.S., and T. Böttger performed research; H.J.M., K.K., T. Braun, and T.W. analyzeddata; and H.J.M. and T.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Howard Hughes Medical Institute, Cincinnati Children’s Hospital Med-ical Center, Cincinnati 45229, OH.

2To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1116584109/-/DCSupplemental.

11794–11799 | PNAS | July 17, 2012 | vol. 109 | no. 29 www.pnas.org/cgi/doi/10.1073/pnas.1116584109

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020

cases, intracardial thrombi were present (Fig. S1E). Many miceexperienced dyspnea, generalized edema, and end-organ damagesecondary to heart failure (Fig. 1F and Fig. S1 F–H). A third ofthe animals died or were terminally ill within 12 wk, and twothirds within 20 wk, of doxycycline discontinuation (Fig. 1G). Ofthose animals that initially survived, the heart weight steadilyincreased, with IKKMyHC hearts weighing ∼60% more thancontrol hearts 12 wk after doxycycline discontinuation (Fig. 1H)and the heart weight/body weight ratio increasing from 5.4 ×10−3

to 8.0 × 10−3 (Fig. S1I). The development of heart failure wasreflected by an increasing expression of natriuretic peptides (Fig.1I), markers also induced in heart failure in human patients.

IKKMyHC Hearts ShowMyocyte Atrophy and Hypertrophy, Inflammation,and Fibrosis. IKKMyHC hearts showed a mixture of cardiomyocyteatrophy and hypertrophy and an inflammatory infiltrate (Fig. 2A).Staining for inflammatory cell markers CD45, CD11b, and F4/80was strongly, for T lymphocyte markers CD4 and CD8, focally in-creased (Fig. 2B and Fig. S2A). Quantification by FACS analysisrevealed that the number of hematopoietic (i.e., CD45+) cells wasincreased more than 11-fold compared with controls, with morethan 82% of infiltrating (i.e., CD45+) cells expressing the myeloidmarker CD11b (Fig. S2B). A small percentage of CD11b+ cellsstained positive for dendritic (5%; CD11b+/CD11c+/MHC-II+)and granulocyte cell markers (1%; CD11b+/Gr1+), suggesting thatthe large majority consisted of macrophages/monocytes. This im-pression was supported by the up-regulation of Emr1 (F4/80)mRNA, a marker for macrophages (Fig. 2C). mRNA transcripts ofintercellular adhesion molecule-1 (ICAM-1), a marker typicallyseen in inflammatory cardiomyopathy/myocarditis, were up-regu-lated as well (Fig. 2D). Moreover, IKKMyHC hearts exhibited mildfibrosis (Fig. 2E).

Cardiomyocyte-Specific IKK/NF-κB Activation Induces Fetal Repro-gramming, Remodeling, and Myocyte Atrophy. Heart disease is of-ten associated with the reexpression of fetal markers in the adultheart. Indeed, the expression of α-smooth muscle actin, an

isoform of actin not normally expressed in adult striated muscle,was increased (Fig. 3A). We also detected atrial natriureticpeptide (ANP) in ventricles, where it is normally absent in adultmice. Destrin, an actin-depolymerizing enzyme, was enhanced aswell, suggesting ongoing remodeling (Fig. 3B). Between the twomain isoforms of myosin heavy chain in mice, the α-isoform(α-MyHC) predominates in adult mice, whereas the β-isoform(β-MyHC) is strongly expressed in embryos. We observed a down-regulation of α-MyHC transcripts, whereas β-MyHC transcriptswere up-regulated, leading to a strong (16-fold) reduction in theα/β-MyHC ratio (Fig. 3C).IKK expression in individual cardiomyocytes was variable,

and, curiously, cardiomyocyte caliber correlated inversely withtransgene expression: cells with high IKK2-CA expression wereatrophic, whereas the majority of low- or nonexpressing cellswere normal or hypertrophic (Fig. 3D and Fig. S2C). Apoptosisand autophagy were not prominent (Fig. S2 D and E). Thus, inaddition to the induction of a general inflammatory state andfetal reprogramming, IKK/NF-κB promoted cardiomyocyte at-rophy, but not cell death. This atrophy-promoting function ofIKK2 was dependent on its kinase activity (Fig. S2 F and G).

MRI Reveals Severely Compromised Heart Function in IKKMyHC Mice.For functional characterization, a cohort of mice was examinedby cardiac MRI. In IKKMyHC mice, global heart function wasindeed severely compromised (Fig. 4, Fig. S3A, and Movies S1,S2, and S3). Both ventricles and atria were dilated, with an in-crease in end-diastolic and especially end-systolic volumes. Sys-tolic heart failure was obvious, with an average left ventricularejection fraction reduced from 59% to 16% and a decrease instroke volume from 41 to 16 μL.

IKK-Induced Cardiomyopathy Is Reversible. Surprisingly, when thetransgene was shut off in mice with IKK-induced cardiomyopa-thy, the disease process was almost completely reversed, with fullrecovery of heart function (Fig. 4, Fig. S3A, and Movies S1, S2,and S3). As shown in Fig. S3B and Movie S4, mice with an

Fig. 1. Expression of IKK2-CA in hearts of transgenic miceleads to cardiomyopathy and heart failure. (A) In double-transgenic IKKMyHC mice, IKK2-CA is expressed under a bi-directional promoter together with luciferase, which can beused as a marker for the activation of the system. Afteradministration of luciferin, luciferase activity can be detec-ted by in vivo imaging in the left thoracal region of IKKMyHC

mice. (B) Western blot of heart extracts with antibodiesdetecting human IKK (transgene-specific, Upper) and mu-rine and human IKK (Middle). GAPDH was used as loadingcontrol. (C) IKK complex of heart extracts from control andIKKMyHC mice was precipitated with an antibody againstNEMO and analyzed in an in vitro kinase assay by using GST-IκBα as substrate. (D) An electrophoretic mobility shift assayof nuclear heart extracts with NF-κB– and Oct-1–specificprobes shows strong activation of NF-κB in the hearts ofIKKMyHC mice. Heart extracts from WT mice after i.p. in-jection of TNF-α, lipopolysaccharide (LPS), or PBS solutionare shown as controls. (E) Left: Cardiomegaly of an IKKMyHC

heart. (Scale bar: 5 mm.) Right: Ventricular dilation (H&Estain). (Scale bar: 200 μm.) (F) Liver congestion (Upper) andgeneralized edema (Lower) in an IKKMyHC mouse. (G) Dis-ease-free survival of IKKMyHC and control mice, weeks afterdoxycycline removal (n = 30 transgenic, n = 15 control mice).(H) Heart weight of IKKMyHC mice, weeks after doxycyclinewithdrawal or during decompensated heart failure at anytime point (HF), as a percentage of age-matched controlmice (n ≥ 4 per group). (I) Expression of brain natriureticpeptide and ANP mRNA (quantitative PCR in ventricularheart tissue; fold up-regulation in comparison with age-matched control mice; n ≥ 4 per group).

Maier et al. PNAS | July 17, 2012 | vol. 109 | no. 29 | 11795

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020

obvious clinical decompensation could also be treated success-fully. The recovery did not only concern heart function, but alsomolecular parameters: markers of inflammation, fetal reprog-ramming, remodeling, and heart failure normalized almostcompletely (Fig. 5 and Fig. S3 C and D).

Gene Expression Analysis of IKKMyHC Cardiomyocytes. To gain insightinto proximal events after activation of IKK2/NF-κB, we isolatedcardiomyocytes and activated the transgene in vitro, thusavoiding interference of other cell types (Fig. S4 A and B). Geneexpression analysis revealed up-regulation of transcripts of che-motactic cytokines, adhesion molecules, and a striking number ofIFN-regulated genes in IKKMyHC cardiomyocytes (Table S1). Inaddition, apoptotic and antiapoptotic genes were up-regulated.

Role of Chemokines, Cytokines, Adhesion Molecules, and MyD88Signaling in IKK-Induced Cardiomyopathy. The enhanced expres-sion of chemokines by cardiomyocytes was validated by quanti-tative PCR (Fig. 6A). In a MS-based secretome analysis, the

chemokines CCL7, CXCL10, and CX3CL1, as well as galectin-3–binding protein, a lectin mediating interaction with macro-phages, were detected in the supernatant of IKKMyHC cells (Fig.6B and Table S2). The enhanced expression of chemotacticfactors and their receptors was also detected in vivo (Fig. S5A).A cytokine that has been associated with cardiomyopathy in miceis TNF-α. We suspected that IKK/NF-κB might induce cardio-myopathy at least in part by enhanced expression of TNF-α (Fig.S5B). However, IKK-induced cardiomyopathy was not rescued inmice deficient in CCR2 or in TNFR1 (p55), a critical receptorfor TNF-α (Fig. S5 C–E). Another important pathway in in-flammatory cardiomyopathy signals via the adapter proteinMyD88. IKK-induced cardiomyopathy was modulated by abla-tion of MyD88, with a significant reduction of heart weight inIKKMyHC/MyD88−/− animals (Fig. S5 C–E). However, thesemice still progressed to heart failure.

NF-κB Is the Mediator of IKK-Induced Cardiomyopathy. We nextwanted to determine by which specific pathway activation ofIKK2 induces cardiomyopathy in IKKMyHC mice. Expression ofan IκBα superrepressor in the hearts of IKKMyHC animals com-pletely abrogated myocarditis and cardiomyopathy, thus clearlyestablishing that IKK induces its effects via classical NF-κB sig-naling. Heart weight/body weight ratio, expression of ANP,amount of inflammatory cells in the heart, as well as chemokine,chemokine receptor, collagen, and Myh6 mRNA transcripts werecompletely normal in IKKMyHC mice with a blocked NF-κBpathway (Fig. 6 C–F and Fig. S6A).

Fig. 2. Cardiomyocyte-specific IKK/ NF-κB activation induces inflammationand fibrosis. H&E stain (A) and immunofluorescent staining (B) of ventricularheart cryosections from control and IKKMyHC mice show infiltration withinflammatory cells. (Scale bar: 100 μm.) (C) Quantitative PCR for Emr1/ F4/80mRNA in ventricular tissue of IKKMyHC and control hearts as a marker for mac-rophage infiltration (fold up-regulation vs. control; n ≥ 8 mice per group). (D)Quantitative PCR for mRNA transcripts encoding the inflammatory adhesionmolecule ICAM-1 in ventricular tissue of IKKMyHC and control hearts (fold up-regulation vs. control; n ≥ 8 mice per group). (E) Sirius red stain on paraffinsections of ventricular tissue of control (Left) and IKKMyHC (Right) mice for thedetection of fibrosis (counterstain, fast green). (Scale bar: 100 μm.)

Fig. 3. Cardiomyocyte-specific IKK/NF-κB activation induces fetal reprog-ramming and myocyte atrophy. (A) Heart cryosections of control and IKKMyHC

animals were stained for α-smooth muscle actin (α-SMA) and dystrophin, andwith DAPI. (Scale bars: 100 μm.) (B) Western blot of ventricular extracts ofIKKMyHC and control mice (8 wk after doxycycline withdrawal) against ANP,α-SMA, destrin, and GAPDH (loading control). (C) Quantitative PCR for Myh6andMyh7 transcripts encoding the α- and β-isoforms ofMyHC (shown are foldregulation vs. age-matched control mice and calculated Myh6/ Myh7 ratio;n ≥ 8 mice per group). (D) Control and IKKMyHC heart paraffin sections werestained against the transgene (human IKK2, red), and with wheat germagglutinin (WGA, green) and DAPI (blue). (Scale bars: 100 μm.)

11796 | www.pnas.org/cgi/doi/10.1073/pnas.1116584109 Maier et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020

IKK Induces Strong IFN Response and Activation of ISG15 Pathway inan NF-κB–Dependent Manner. Among the IFN-regulated genes up-regulated upon IKK activation, we analyzed IFN-regulated gene15 (Isg15) in more detail. Its gene product, ISG15, is a ubiquitin-like modifier, which can be covalently attached to a range of pro-teins, a process referred to as ISGylation. Isg15 transcripts, as wellas those for its E3 ligase and the protease removing ISG15 fromother proteins, were increased in cardiomyocytes expressing IKK2-CA (Fig. S6B). On IKKMyHC heart sections, ISG15 was expressedin infiltrating cells and in cardiomyocytes (Fig. S6C). Immunoblotsrevealed widespread protein ISGylation in cardiomyocytes (Fig.S6D and E), which was dependent on NF-κB (Fig. 6G). Vice versa,when cardiomyocytes deficient for NEMO (and thus, also classicalNF-κB signaling) were stimulated with IFN-β, the induction ofISG15 was reduced significantly, but not abolished (Fig. S6F).

Significance of IKK/NF-κB and ISG15 Pathways in a Mouse Model ofViral Myocarditis. Finally, wewanted to examine whethermolecularpathways active in our mouse model were relevant in an

establishedmodel of virus-induced inflammatory heart disease. AsNF-κB and ISG15 are central mediators of the defense againstviruses, we analyzed their activation status in acute murine CVB3myocarditis. In CVB3-infected mice, NF-κB–specific DNA bind-ing was strongly activated in the heart (Fig. 6H and Fig. S6G).Expression of ISG15 protein was enhanced, with strong andwidespread ISGylation of target proteins similar to that observedin IKK-induced cardiomyopathy (Fig. 6I and Fig. S6G). Thus,IKK-induced cardiomyopathy recapitulates molecular pathwaysrelevant in other mouse models of cardiac disease.

DiscussionIKK/NF-κB is a master regulator of inflammation in many celltypes and organs. We show here that, in the heart, activation ofIKK/NF-κB leads to myocarditis and inflammatory dilated car-diomyopathy, which is associated with atrophy in IKK-expressingmyocytes and a severe reduction in contractility.IKK activation has been linked to inflammation in several

organ systems, but does not necessarily lead to inflammation inall cell types and tissues: in skeletal muscle, no inflammation wasobserved upon IKK2 activation (19). In our cardiac model, IKKactivation led to strong NF-κB activation and inflammation.Cardiomyopathy was mediated by NF-κB, as disease was pre-vented in mice expressing an IκBa superrepressor. We suspectedthat some of the inflammatory effects of NF-κB could be me-diated by its target gene TNF-α, which by itself can induce car-diomyopathy (20). However, the phenotype could not be rescuedby ablation of TNFR1, the receptor considered the mediator ofthe deleterious effects of TNF-α (21). Vice versa, NF-κB maystill mediate some of the detrimental effects of TNF-α (22, 23)Importantly, KO of TNFR1 improves heart function after cor-onary artery ligation, which correlates with decreased NF-κBactivation (22).Other target genes of NF-κB encode chemokines, which also

were up-regulated in our model. However, ablation of the che-mokine receptor CCR2 did not modulate our phenotype, pos-sibly because of the high redundancy of the chemokine recep-tor system.IKK-induced cardiomyopathy displayed molecular features of

viral myocarditis, an important cause for dilated cardiomyopa-thy. These comprised the transcriptional up-regulation of Toll-like receptors as well as the induction of IFN type I-dependentgene expression. Downstream of TLR signaling, the adapterprotein MyD88 is critical in different types of inflammatorycardiomyopathy (24, 25). Ablation of MyD88 modulated ourphenotype by attenuating the excessive gain in heart weight;however, it did not prevent heart failure.

Fig. 4. Cardiac MRI of IKKMyHC mice and reversibility of IKK-induced car-diomyopathy. A cohort of control and IKKMyHC mice was examined in a firstMRI under doxycycline to assess normal heart function, in a second MRI after12 wk of doxycycline withdrawal to assess heart function during IKK-inducedcardiomyopathy, and in a third MRI after 3 mo of doxycycline readminis-tration. EDV, end-diastolic volume; ESV, end-systolic volume; EF, ejectionfraction; LV, left ventricular (n = 8 mice for first and second MRI, n = 5 micefor third MRI; Fig. S3A and Movies S1, S2, S3, and S4).

Fig. 5. Reversibility of IKK-induced cardiomyopathy on thecellular and molecular level. Doxycycline in drinking waterwas discontinued for 12 wk to activate transgene expressionand induce cardiomyopathy. Mice were then either analyzed(diseased) or given doxycycline for 3 mo and then analyzed(reversed). (A and B) Quantitative PCR of ventricular tissuesamples for F4/80 (macrophage infiltration), ICAM-1 (in-flammation), ANP (ventricular remodeling), and brain natri-uretic peptide (BNP; heart failure) mRNA transcripts (n ≥ 8diseased, n ≥ 5 reversed state). Shown is the fold regulation vs.control mice at diseased time point. (C) Western blot for IKK,α-SMA, ANP, destrin, and GAPDH (loading control). (D)Quantitative PCR for α- and β-MyHC mRNA transcripts asa marker for the reversal of fetal reprogramming (n ≥ 8 dis-eased, n ≥ 5 reversed state).

Maier et al. PNAS | July 17, 2012 | vol. 109 | no. 29 | 11797

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020

IKK activation alone was sufficient to activate the ISG15pathway, an important antiviral defense mechanism, and elicitedwidespread ISGylation of proteins in an NF-κB–dependentmanner. This was also found in CVB3-induced myocarditis,which showed a strong activation of the NF-κB and ISG15pathways. Vice versa, in the absence of the IKK subunit NEMO,the induction of ISG15 by IFN-β was attenuated. These resultsmay hint at an important role of IKK/NF-κB signaling in or-chestrating the antiviral response in the heart.The question whether IKK/NF-κB is protective or damaging in

heart disease has been a matter of ongoing debate (3). IKK/NF-κB can be protective by reducing oxidative stress (6, 7). Consis-tent with that, apoptosis was not prominent upon IKK/NF-κBactivation in our model. However, our finding that IKK/NF-κBactivation can lead to a dramatically impaired systolic functionmay be of interest for cardiac diseases with strong activation ofIKK/NF-κB. In viral myocarditis, for example, myocyte damageand reduced contractility may not only result from direct viralaction, but also from the host response to the virus, whichinvolves IKK/NF-κB activation.Establishing a balance between desired protective functions

and unwanted side effects may also apply to the IFN type I re-sponse. In a recent study on dermatomyositis, an autoimmunedisorder involving skeletal muscle, the ISG15 pathway was as-sociated with atrophy (26). In light of our results, further in-vestigation into the link between type I IFN response andmyocyte damage may prove fruitful.

Previous studies have established that NF-κB is required forcardiac hypertrophy, but is dispensable for normal heart struc-ture and function (8, 10, 11, 27). Our finding that activation ofIKK/NF-κB in the heart primarily induces cardiomyocyte atro-phy rather than hypertrophy is not necessarily a contradiction tothese studies, as the biological function of IKK/NF-κB may de-pend on the strength of activation and/or on context, and on thecomponent of the IKK/NF-κB system that is manipulated. Forexample, IKK2 and NEMO KO mice showed an increase inoxidative stress and apoptosis when challenged by aortic band-ing, but hypertrophy was not inhibited (6, 7). Interestingly, anatrophy-promoting function of IKK/NF-κB is well established inskeletal muscle (19, 28).For therapeutic interventions in heart disease, it is of interest

that normal heart function and morphology were almost com-pletely restored in IKK-induced cardiomyopathy by shutting offtransgene expression. Obviously, the processes regulated by IKK/NF-κB are remarkably plastic. However, IKK/NF-κB signalingcannot be simply labeled protective or damaging, as it displaysantiapoptotic and antioxidative functions as well as proin-flammatory, atrophy-promoting effects in heart muscle, and canlead to a dramatic reduction in contractility. This complexityshould be kept in mind when considering targeted pharmaco-logical interventions involving IKK/NF-κB in heart disease.

Materials and MethodsMice.Micewere kept under specific pathogen-free conditions. The tetO.IKK2-CA mouse line EE-5 has been described previously (18); the phenotype was

Fig. 6. Mechanisms, consequences, and significance of IKK-induced cardiomyopathy. (A) Cardiomyocytes were isolated from mice kept under doxycycline,cultivated without doxycycline (i.e., transgene induction) for 48 h, and analyzed by quantitative PCR for chemokine transcripts. Shown is fold regulation vs.control (n = 3 controls, n = 6 transgenic; Mann–Whitney test). (B) Proteomic analysis of supernatant of IKKMyHC cardiomyocytes shows presence of factorsinvolved in inflammatory cell recruitment. G-3-BP, galectin-3 binding protein; n.d., not detectable in control supernatant (Table S2). (C) In vivo blockade of NF-κB by transgenic expression of an IκBα superrepressor (3M+) prevents IKK-induced cardiomyopathy in IKKMyHC mice. Shown is the heart weight/body weightratio of mice in the acute disease state (n ≥ 3 mice per group). (D) Western blot of heart extracts of control and IKKMyHC mice in the acute disease stateexpressing or not expressing IκBα-3M. (E) Immunofluorescent staining for CD45+ cells in heart cryosections (acute disease state) reveals absence of infiltrates inIKKMyHC hearts expressing IκBα superrepressor (IKK-MyHC/3M+). (F) Quantitative PCR of mRNA isolated from heart tissue of control and IKKMyHC animals, inthe absence (3M−) or presence (3M+) of IκBα-3M (acute disease state; n ≥ 3 mice per group). (G) Western blot with an antibody against ISG15 reveals strongexpression of ISG15 and widespread ISGylation of other proteins in the hearts of IKKMyHC animals, both dependent on NF-κB; GAPDH, loading control. (H) WTmice of the susceptible ABY/SnJ strain were infected with CVB3, and nuclear heart extracts were obtained from the acute phase (8 d post infection) andanalyzed in an NF-κB– and an Oct-1–specific EMSA. (I) Western blot against ISG15 and GAPDH (loading) with heart extracts from noninfected control andCVB3-infected mice (8 d post infection).

11798 | www.pnas.org/cgi/doi/10.1073/pnas.1116584109 Maier et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020

confirmed in a second mouse line (EE-7; Fig. S7). For most experiments, micewere administered doxycycline (1 g/L in drinking water) until adult age andanalyzed at the indicated time point after doxycycline withdrawal. For acuteactivation of the system (designated as acute disease state), mice were keptunder doxycycline (0.1 g/ L) until the age of 4 wk, and analyzed 10 d later.α-MyHC.tTA mice (17), IκBα-3M mice (29), IKK2-DN mice (18), and mice witha floxed Ikbkg allele (30) have been described previously. For induction ofviral myocarditis, 4- to 5-wk-old ABY/SnJ mice (H-2b) were infected in-traperitoneally with 1 × 105 pfu of purified CVB3 as described (16). Allexperiments were in accordance with institutional guidelines and Germananimal protection laws.

In Vivo Bioluminescence Imaging. Mice were anesthetized by continuous inha-lation of 2% (vol/vol) isoflurane. At 5 min after i.p. injection of 3.5 mg luciferin,luminescence was detected and quantified in an IVIS 200 system (Caliper).

Western Blotting. Tissue was snap-frozen in liquid nitrogen and lysed ina buffer containing 4% SDS. Lysates were separated by SDS/PAGE and blottedonto membranes, which were blocked, incubated in primary and HRP-cou-pled secondary antibodies, and developed by chemiluminescence.

Kinase Assay. A total of 500 μg of protein lysate were precipitated withNEMO antibody. The kinase reaction was performed in the presence of 32P-γ-ATP and recombinant GST-IκBα as substrate, followed by boiling inLaemmli buffer and SDS/PAGE.

Electrophoretic Mobility Shift Assay. Nuclear extraction was done as described(5). For electrophoretic mobility shift assays (EMSAs), 4 μg of nuclear extractwas incubated with radiolabeled DNA probes containing a consensus κB orOct-1 site and separated on a native polyacrylamide gel.

Histology and Immunofluorescence Microscopy. Cryosections and paraffinsections were prepared as described in SI Materials and Methods. First an-tibody incubation was done overnight at 4 °C with a diluent (Dako). Sec-ondary antibodies coupled with Alexa Fluor 488 or 596 were incubated for1 h at room temperature.

Cardiac MRI. Isoflurane-anesthetized mice were examined on a BrukerPharmascan 7.0 T equipped with a 300 mT/m gradient system. Multiplecontiguous short-axis slices consisting of six to eight slices were acquired for

complete coverage of the left ventricle. MRI data were analyzed with Qmassdigital imaging software (Medis).

Adult Murine Cardiomyocyte Isolation. Excised hearts were digested by ret-rograde Langendorff perfusion. The cell suspension was filtered througha 100-μm mesh, centrifuged, and seeded on laminin-coated wells.

RNA Extraction and Quantitative PCR. RNA extraction and cDNA synthesiswere done with kits (Qiagen and Roche). Quantitative PCR was done ona LightCycler 480 system with use of the Universal Probe Library (Roche).Rpl13 was used as reference gene for relative quantification. Primersequences are available on request.

Gene Expression Profiling. A total of 200 ng total RNA was amplified andlabeled with the Whole Transcript Sense Target Labeling Assay (Affymetrix).Labeled samples were hybridized to Affymetrix GeneChip Mouse Gene 1.0 STArray, processed, scanned, and analyzed (RMA algorithm with AffymetrixExpression Console and GeneSifter microarray data analysis system).

Secretome Analysis. Supernatant (serum-free medium) of cardiomyocytes wascollected after 24 h, filtered, and concentrated. Protein samples were pro-cessed and mass spectrometry was performed as described in SI Materials andMethods by using an LTQ-Orbitrap (Thermo Fisher Scientific) combined witha Proxeon nanoflow HPLC system.

Statistical Analysis. Values are given as arithmetic mean ± SD. Means of twogroups were compared by the Student t test or, when indicated, the Mann–Whitney test. Welch correction was applied for unequal variances. Means ofmultiple groups were compared by ANOVA. All tests were two-tailed. Pvalues <0.05 were deemed significant. Detailed procedures are described inSI Materials and Methods.

ACKNOWLEDGMENTS. The authors thank Melanie Gerstenlauer and UtaManfras for excellent technical assistance, Marion Wiesnet for cardiomyo-cyte isolation, Ursula Hofmann for conducting cardiac MRI, Ute Leschik forhelp with the kinase assay, Marc Schmidt-Supprian (Max Planck Institute ofBiochemistry, Martinsried, Germany) for providing mice carrying floxedIkbkg alleles, Keith Jones (University of Cincinnati, Cincinnati, OH) for pro-viding IκBα-3M mice, and Klaus-Peter Knobeloch (University of Freiburg,Freiburg, Germany) for providing ISG15 antibody. This study was supportedin part by Deutsche Forschungsgemeinschaft Grant SFB/TR19 (to K.K.).

1. Dhingra R, Shaw JA, Aviv Y, Kirshenbaum LA (2010) Dichotomous actions of NF-

kappaB signaling pathways in heart. J Cardiovasc Transl Res 3:344–354.2. Van der Heiden K, Cuhlmann S, Luong A, Zakkar M, Evans PC (2010) Role of nuclear

factor kappaB in cardiovascular health and disease. Clin Sci (Lond) 118:593–605.3. Gordon JW, Shaw JA, Kirshenbaum LA (2011) Multiple facets of NF-κB in the heart: To

be or not to NF-κB. Circ Res 108:1122–1132.4. Oeckinghaus A, Ghosh S (2009) The NF-kappaB family of transcription factors and its

regulation. Cold Spring Harb Perspect Biol 1:a000034.5. Dawn B, et al. (2001) Cardiac-specific abrogation of NF- kappa B activation in mice by

transdominant expression of a mutant I kappa B alpha. J Mol Cell Cardiol 33:161–173.6. Hikoso S, et al. (2009) The IkappaB kinase beta/nuclear factor kappaB signaling

pathway protects the heart from hemodynamic stress mediated by the regulation of

manganese superoxide dismutase expression. Circ Res 105:70–79.7. Kratsios P, et al. (2010) Antioxidant amelioration of dilated cardiomyopathy caused

by conditional deletion of NEMO/IKKgamma in cardiomyocytes. Circ Res 106:133–144.8. Freund C, et al. (2005) Requirement of nuclear factor-kappaB in angiotensin II- and

isoproterenol-induced cardiac hypertrophy in vivo. Circulation 111:2319–2325.9. Gupta S, Young D, Sen S (2005) Inhibition of NF-kappaB induces regression of cardiac

hypertrophy, independent of blood pressure control, in spontaneously hypertensive

rats. Am J Physiol Heart Circ Physiol 289:H20–H29.10. Young D, Popovic ZB, Jones WK, Gupta S (2008) Blockade of NF-kappaB using IkappaB

alpha dominant-negative mice ameliorates cardiac hypertrophy in myotrophin-

overexpressed transgenic mice. J Mol Biol 381:559–568.11. Zelarayan L, et al. (2009) NF-kappaB activation is required for adaptive cardiac hy-

pertrophy. Cardiovasc Res 84:416–424.12. Alter P, Rupp H, Maisch B (2006) Activated nuclear transcription factor kappaB in

patients with myocarditis and dilated cardiomyopathy—relation to inflammation and

cardiac function. Biochem Biophys Res Commun 339:180–187.13. O’Donnell SM, et al. (2005) Organ-specific roles for transcription factor NF-kappaB in

reovirus-induced apoptosis and disease. J Clin Invest 115:2341–2350.14. Yamamoto K, et al. (2003) Attenuation of virus-induced myocardial injury by in-

hibition of the angiotensin II type 1 receptor signal and decreased nuclear factor-

kappa B activation in knockout mice. J Am Coll Cardiol 42:2000–2006.15. Yokoseki O, et al. (2001) cis Element decoy against nuclear factor-kappaB attenuates

development of experimental autoimmune myocarditis in rats. Circ Res 89:899–906.

16. Klingel K, et al. (1992) Ongoing enterovirus-induced myocarditis is associated withpersistent heart muscle infection: Quantitative analysis of virus replication, tissuedamage, and inflammation. Proc Natl Acad Sci USA 89:314–318.

17. Yu Z, Redfern CS, Fishman GI (1996) Conditional transgene expression in the heart.Circ Res 79:691–697.

18. Herrmann O, et al. (2005) IKK mediates ischemia-induced neuronal death. Nat Med11:1322–1329.

19. Cai D, et al. (2004) IKKbeta/NF-kappaB activation causes severe muscle wasting inmice. Cell 119:285–298.

20. Kubota T, et al. (1997) Dilated cardiomyopathy in transgenic mice with cardiac-spe-cific overexpression of tumor necrosis factor-alpha. Circ Res 81:627–635.

21. Higuchi Y, et al. (2004) Tumor necrosis factor receptors 1 and 2 differentially regulatesurvival, cardiac dysfunction, and remodeling in transgenic mice with tumor necrosisfactor-alpha-induced cardiomyopathy. Circulation 109:1892–1897.

22. Hamid T, et al. (2009) Divergent tumor necrosis factor receptor-related remodelingresponses in heart failure: Role of nuclear factor-kappaB and inflammatory activa-tion. Circulation 119:1386–1397.

23. Kawamura N, et al. (2005) Blockade of NF-kappaB improves cardiac function andsurvival without affecting inflammation in TNF-alpha-induced cardiomyopathy. Car-diovasc Res 66:520–529.

24. Fuse K, et al. (2005) Myeloid differentiation factor-88 plays a crucial role in thepathogenesis of Coxsackievirus B3-induced myocarditis and influences type I in-terferon production. Circulation 112:2276–2285.

25. Blyszczuk P, et al. (2009) Myeloid differentiation factor-88/interleukin-1 signalingcontrols cardiac fibrosis and heart failure progression in inflammatory dilated car-diomyopathy. Circ Res 105:912–920.

26. Salajegheh M, et al. (2010) Interferon-stimulated gene 15 (ISG15) conjugates proteinsin dermatomyositis muscle with perifascicular atrophy. Ann Neurol 67:53–63.

27. Li Y, et al. (2004) NF-kappaB activation is required for the development of cardiachypertrophy in vivo. Am J Physiol Heart Circ Physiol 287:H1712–H1720.

28. Mourkioti F, et al. (2006) Targeted ablation of IKK2 improves skeletal musclestrength, maintains mass, and promotes regeneration. J Clin Invest 116:2945–2954.

29. Brown M, et al. (2005) Cardiac-specific blockade of NF-kappaB in cardiac patho-physiology: differences between acute and chronic stimuli in vivo. Am J Physiol HeartCirc Physiol 289:H466–H476.

30. Schmidt-Supprian M, et al. (2000) NEMO/IKK gamma-deficient mice model incon-tinentia pigmenti. Mol Cell 5:981–992.

Maier et al. PNAS | July 17, 2012 | vol. 109 | no. 29 | 11799

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

020