International Journal of Cardiology 151 (2011) 18–33

Contents lists available at ScienceDirect

International Journal of Cardiology

j ourna l homepage: www.e lsev ie r.com/ locate / i j ca rd

Fibrosis in endstage human heart failure: Severe changes in collagen metabolism andMMP/TIMP profiles

Victoria Polyakova a, Ivo Loeffler a, Stefan Hein b, Shigeru Miyagawa b,c, Izabela Piotrowska d,Sebastian Dammer a, Juha Risteli e, Jutta Schaper a, Sawa Kostin a,⁎a Core Lab for Molecular and Cellular Biology, Max-Planck-Institute for Heart and Lung Research, Bad Nauheim, Germanyb Kerckhoff Clinic, Department of Cardiac Surgery, Bad Nauheim, Germanyc Department of Cardiac Surgery, Osaka University Graduate School of Medicine, Osaka, Japand Molecular Haematology and Cancer Biology Unit, Institute of Child Health, University College London, 30 Guilford Street, WC1N 1BL, London, UKe University of Oulu, Oulu, Finland

Abbreviations: ANP, atrial natriuretic peptide; DCMejection fraction; ECM, extracellular matrix; FGF, fibrofailure; ICH, immunohistochemistry; ICM, ischemic carzone of the mycardium in patients with ICM; ICM-BZ,cardiomyocytes and scar tissue in patients with ICM; PLVEDP, left ventricular end-diastolic pressure; MMP, mmyocarditis; TIMP, tissue inhibitor of matrix metallopropropeptide of type I procollagen (newly synthesized collIII collagen peptide (newly synthesized collagen III); ICT(degradation product of cross-linked collagen I); IIIN(cross-linkedmature collagen III); RT-PCR, real-time polytransforming growth factor β1.⁎ Corresponding author.Max-Planck-Institute for Heart

Molecular and Structural Biology, Parkstraße 1, 61231 Bad6032 705 420; fax: +49 6032 705 419.

E-mail address: sawa.kostin@mpi-bn.mpg.de (S. Kos

0167-5273/$ – see front matter © 2010 Elsevier Irelanddoi:10.1016/j.ijcard.2010.04.053

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 9 October 2009Received in revised form 1 March 2010Accepted 17 April 2010Available online 23 May 2010

Keywords:Heart failureFibrosisCollagenMMPTIMPImmunoconfocal microscopy

Objectives: We studied fibrosis, collagen metabolism, MMPs/TIMPs and cytokine expression in various forms ofhuman heart failure (HF) by quantitative immunofluorescent microscopy, Western blot, zymography, RT-PCRand in situ hybridization. In explanted human hearts with HF due to either dilated (DCM, n=6) or ischemic(ICM-BZ-borderzone, ICM-RZ-remote zone, n=7) or inflammatory (myocarditis, MYO, n=6) cardiomyopathyand 8 controlsMMP2, 8, 9, 19, and TIMP1, 2, 3, 4 aswell as procollagens I and III (PINP, PIIINP),mature collagen III(IIINTP) and the cross-linked collagen I degradation product (ICTP) were measured.Results: In comparison with controls, MMPs and TIMPs were significantly upregulated ranging (from highest tolowest) from ICM-BZ, DCM, ICM-RZ, MYO for all MMPs with the exception of MMP9 (highest in DCM), and forTIMPs from ICM-BZ, ICM-RZ, DCM andMYO.MMP2 and 9were activated in all groups. The TIMP/MMP ratio was1.3 for control, 1.9 in ICM-BZ (TIMPNMMP) and lowered to 1.0 in the other groups. Collagen I/collagen III ratiocorrelated significantlywith the decrease in LVEDP. PINPwas higher than ICTP in all groups. PIIINP elevationwaspresent inDCMand ICM-RZ and IIINTPwasup to4-fold augmented in all groups. FibrosinmRNAwas upregulated

in ICM-BZ, activin A in MYO but FGF1 and FGF2 remained unchanged. ANP mRNA was increased in all groups.Conclusions: Although different degrees of severity of collagen metabolism, MMP/TIMP imbalance and cytokineexpression in diverse forms of HF are present, the end product is collagen deposition. These findings suggestmultiple mechanisms acting alone or in concert in fibrosis development in HF.

© 2010 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The main collagens in the myocardial extracellular matrix (ECM)are type I (50–85%) and type III (10–45%) collagens [1]. These fibrillarcollagens are secreted into the extracellular space as procollagens

, dilated cardiomyopathy; EF,blast growth factor; HF, heartdiomyopathy; ICM-RZ, remotebordere zone between viableBS, phosphate-buffered saline;atrix metalloproteinase; MYO,teinases; PINP, aminoterminalagen I); PIIINP, N-terminal typeP, carboxyterminal telopeptideP, aminoterminal telopeptidemerase chain reaction; TGFβ1,

and Lung Research, Core Lab forNauheim, Germany. Tel.: +49

tin).

Ltd. All rights reserved.

with large propeptides at both ends. After secretion, the propeptidesare cleaved by specific tissue endopeptidases and the collagenmolecules aggregate to form the fibril. They are stabilized by intra-and intermolecular cross-linking [2]. The propeptides are differenti-ated into either N-terminal propeptides (PINP and PIIINP) orcarboxyterminal telopeptides (ICTP) or aminoterminal telopeptides(IIINTP). PINP and PIIINP characterize the newly synthesized type Iand III collagens, IIINTP is a marker of cross-linked mature collagen III,whereas ICTP detects a degradation product of cross-linked type Icollagen [3,4].

Collagens are extremely stable and highly resistant to degradationby all proteinases except for matrixmetalloproteinases (MMP). MMPsare known modulators of myocardial remodeling underlying heartfailure progression (reviewed in [5]). The interplay and balance ofMMPs and their tissue inhibitors (TIMP) determine the maintenanceof interstitial tissue homeostasis.

Different specific proteolysis/antiproteolysis changes of the ECMwere reported in various heart diseases [6–10], however, more data areneeded to substantiate the consistency of these results, especially inhuman patients. Furthermore, it appears to be interesting to correlate

Table 1Clinical data.

DCM ICM MYO Control

Number 6 7 6 8Age (years) 49.3±6.7 54.8±2.4 43.4±5.8⁎ 51.5±6.3Sex (male/female) 6/0 7/0 4/2 6/2NYHA class, n IV (6/6) IV (7/7) IV (6/6) –

EF (%) 15±2.1 21±2.2 23±4.3 N60%

Medical history:Hypertension (n) – 3 –

Diabetes mellitus (n) 3 4 2 –

Family history of CAD (n) 2 – – –

Family history ofcardiomyopathy (n)

2 – 1 –

ASD (n) – – – 2MVS (n) – – – 2MVI (n) 5 5 4Atrial fibrillation (n) 2 3 3 1LBBB (n) 1 3 1 –

SVT (n) 1 1 2 –

NSVT (n) 3 4 3 –

Medications:ACE inhibitors (n) 3 3 3 –

Beta-blockers (n) 2 2 2 –

CC blockers (n) 2 2 2 –

Digitalis (n) 5 5 4 –

Diuretics (n) 5 6 5 –

ASD— atrial septumdefect, CAD— coronary artery disease, CC— Calcium channel, LBBB—

left bundle branch block, MVS — mitral valve stenosis, MVI — mitral valve insufficiency,NSVT — nonsustained ventricular tachycardia, SVT — sustained ventricular tachycardia.⁎ pb0.05 MYO vs ICM.

19V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

changes in MMP/TIMP concentrations with alterations of the collagenmetabolism.

The circulating collagen propeptides are commonly used asmarkers of collagen metabolism. Serum appearance of collagen Iand III fragments were demonstrated after myocardial infarction,hypertrophic cardiomyopathy and other heart diseases [11–15].However, only a direct examination of cardiac tissues represents themost reliable method for the evaluation of collagen metabolism andthe quantification of myocardial fibrosis [16] because collagenmarkers in serum may be elevated due to liver fibrosis, systemicsclerosis, cancer, bone and joint diseases, atherosclerosis and otherpathologies [17–21]. There is, however, very limited informationconcerning changes of collagen metabolism in failing human hearts.

Several factors such as TGFβ, angiotensin II and others regulate myo-cardial fibrosis [22,23]. Fibroblast growth factor 1 (FGF1) and 2 (FGF2),atrial natriuretic peptide (ANP), activin A and fibrosin are cytokines withmany important functions including anti- and profibrogenic activity andthe ability to activate different types of cells. It is therefore reasonable tohypothesize that these factors may also be involved in myocardial ECMremodeling.

There are no comprehensive studies comparing changes in collagenmetabolism, the levels of MMPs and TIMPs and the degree of cytokineexpression in failing human hearts with different cardiac diseases.Therefore, we tested the hypothesis that each pathogenetic entitypossesses its own pattern of proteolysis/antiproteolysis, but that thecollagen turnover is similar in all patientsfinally resulting in remodelingof the ECM and fibrosis. Similarly, we postulated that changes incollagen metabolism, the levels of MMPs and TIMPs and the degree ofcytokine expression interrelate with each other. This process ofmyocardial remodeling reduces cardiac compliance and significantlycontributes to diastolic dysfunction in the failing heart.

2. Materials and methods

2.1. Patients

Left ventricular (LV) samples were collected from the explanted hearts of patientswho underwent orthotopic transplantation because of endstage heart failure. Allpatients were severely symptomatic (NYHA grade IV). Patients either had normalcoronary arteries with histologically proven MYO (n=6) according to the Dallascriteria, or were diagnosed as idiopathic DCM (n=6), or ischemic ICM (n=7) cardio-myopathy. The duration of myocarditis was 3.5±1.8 years. Clinical data aresummarized in Table 1. All patients before transplantation had a standard medicationregimen including diuretics, digitalis, ACE inhibitors and ß-blockers. LV myocardiumfrom four donor hearts that for technical reasons were not used for transplantation andintraoperative myocardial biopsies from 2 patients with atrial septal defect and 2patients with a stenosed mitral valve but with normal LV function served as controltissues. The institutional Ethical Committee approved the study, and all patients gavewritten informed consent. The investigation conforms to the principles outlined in theDeclaration of Helsinki.

2.2. Tissue sampling

At the time of transplantation, left ventricular specimens were removed, dissectedinto smaller pieces, immediately frozen in liquid nitrogen and stored at −80 °C.Myocardium from patients with ICM was divided into remote (ICM-RZ) and borderzone (ICM-BZ) regions.

2.3. Immunolabeling and fluorescent microscopy

The tissue samples were mounted in tissue Tec and cryosections 5 μm thick wereprepared. Cryosections were air dried and fixed for 10 min in acetone (−20 °C),4% paraformaldehyde, or Carnoys' solution (room temperature). After washing inphosphate-buffered saline (PBS) sections were incubated with 1% bovine serumalbumin for 30 min to block non-specific binding sites. After rinsing in PBS, the sampleswere incubated overnight with primary antibodies against collagens I and III, MMP2and MMP9 (Biotrend), MMP8 and MMP19 (Sigma), TIMP1 and TIMP2 (Calbiochem),TIMP3 and TIMP4 (Sigma), smoothmuscleα-actin (Sigma), vimentin (Sigma), CD3 andCD68 (Dako), and TGFß (R&D).

Polyclonal antibodies to ICTP, PINP, PIIINP and IIINTP were prepared and used aspreviously described [4,24]. After incubation with the primary antibodies, the sectionswere thoroughly washed in PBS and incubated with secondary biotin-conjugated anti-

mouse, anti-rabbit or anti-sheep IgG antibodies (Dianova) for 1 h followed by incubationwith Cy2 or Cy3 conjugated streptavidin. Nuclei were stained blue with DRAQ5 (Alexis).

Tissue sections were examined by laser scanning confocal microscopy (Leica TCSSP2). Series of confocal optical sections were taken using a Leica Planapo x40/1.00 orx63/1.32 objective lens. Each recorded image was taken using dual-channel scanningand consisted of 1024×1024 pixels. To improve image quality and to obtain a highsignal to noise ratio, each image from the series was signal-averaged. After dataacquisition, the images were transferred to a Silicon Graphics workstation (SiliconGraphics) for restoration and three-dimensional reconstruction using Imaris 4.5multichannel image processing software (Bitplane, Zürich, Switzerland).

2.4. Quantitative immunofluorescent measurements

Cryosections from at least two different tissue blocks in each case were used. Allsamples were immunolabeled simultaneously with identical conditions of fixation anddilutions of primary and secondary antibodies. Sections exposed to PBS instead ofprimary antibodies served as negative controls. For each patient at least 10 randomfields of vision per each tissue block were analyzed with a fluorescent microscope Leica(Leitz DMRB) using a x40 Planapo objective (Leica). Ten randomly fields of vision fromonly one tissue block were analyzed in control tissues obtained from 2 patients withmitral valve stenosis and 2 patients with atrial septum defect. Immunolabeledcryosections were studied using image analysis using “Quantification” and “VoxelShop”options of Imaris 4.5 and Imaris 2 (Bitplane, Zürich, Switzerland) and Image J software.For each protein a specific setting was established and kept constant in allmeasurements. The area of specific labeling for MMPs, TIMPs and collagen markerswas calculated as percent of positive labeling per tissue area.

2.5. Western blot

Western blot was performed as previously described [8,24]. In brief, frozen tissuewas homogenized in RIPA buffer buffer (containing 20 mmol/L Tris–HCl at pH 7.4,100 mmol/L NaCl, 5 mmol/L thylene-diamine tetraacetic acid, 1% Triton X-100, 10%glycerol, 0.1% sodium dodecylsulfate, 1% deoxycholate, 50 mmol/L NaF, 10 mmol/LNa4P2O7, 1 mmol/L Na3VO4, 1 mmol/L phenylmethylsulfonylflouride) and mammalianprotease inhibitor cocktail (Sigma) at pH 7.4 and centrifuged at 2000×g at 4 °C for10 min. LVmyocardial extracts were loaded onto 12% polyacrylamide gel and separatedunder the reducing conditions. Proteins were electrotrasferred onto nitrocellulosemembrane (Invitrogen) and blocked with 5% non-fat dry milk in Tris-buffered salineTween-20 (TBST) at 4 °C. After washing with TBST, proteins were exposed overnight at4 °C to antibodies against MMP2 and MMP9 (Biotrend) diluted in TBS with 5%powdered milk. Bound antibodies were detected by peroxidase-conjugated anti-mouse

20 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

IgG horseradish peroxidase-conjugated and SuperSignal WestFemto (Pierce) detectionsystem and exposed to X-ray film. Quantification of immunoblots was done byscanning on a STORM 860 (Amersham Pharmacia Biotech) using ImageQuant software.The immunoblotting values for the investigated proteints were normalized per actin(clone HHF-35, Sigma).

2.6. Zymography

Zymographywas doneasdescribed before [8]. Inbrief, LV sampleswerehomogenizedin ice-cold extraction buffer and then centrifuged at 2000×g for 30 min at 4 °C. Themyocardial extracts were loaded onto electrophoretic gels (SDS-PAGE) containing agelatine (0.1%) substrate under non-reducing conditions. After electrophoresis the gelwasincubated twice in renaturating buffer for 30 min each (25 °C), rinsed in water, and

Fig. 1. Representative confocal micrographs of collagen type III (green) and type I (red) in coincreased accumulation of fibrillar collagens in diseased myocardium. Nuclei are stained bl

incubated for 20 h indevelopingbuffer (25 °C). Sampleswere incubated in the presenceof1 mM of PMSF (inhibitor of serine proteinase) and 5 mM iodacetamide (inhibitor ofcysteine protease) and an identical gel was incubated in reaction buffer containing 20 mMEDTA, an inhibitor ofMMPs. After the reaction, the gel was stainedwith 0.25% Commassiebrilliant blue R250. Zones of lysis were visualized as clear bands against the bluebackground. The zymogramswere digitized using a constant light intensity. MMP activitywas determined by quantitative image analysis.

2.7. Quantitative RT-PCR

Total RNA was extracted from cardiac tissue using TRIzol reagent (Invitrogen LifeTechnologies, USA) according to the manufacturer's instructions. After DNAse treatment

ntrol (A–C), MYO (D–F), and in RZ (G–I) and BZ regions (J–L) in ICM demonstrating anue with DAPI.

21V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

(TURBO DNA-free Ambion) 100 ng RNA were reverse transcribed (Superscript II,Invitrogen). cDNA was amplified using gene specific human primers

FGF1: forward 5′-AATGGGAGCTGCAAACGCGGTCC-3′,reverse 5′-TCAACCAGGTGAGGACCCCTCGA-3′;FGF2: forward 5′-TCAAGGACCCCAAGCGGCTGTAC-3′,reverse 5′-AGCTTGATGTGAGGGTCGCTC-3′;Fibrosin 1: forward 5′-AGCACGAACCCTGAGCTGCCACCA-3′,reverse TCATGTAAGCCACACGCACGTG-3′;Activin A: forward 5′-CATGATACAAGAGCCGGCTGGTGG-3′,reverse 5′-GCGCAGGTTTTTTTGTGTGTGTGG-3′

on an icycler real-time PCR system as previously described [25]. Samplesweremeasuredin triplicate with a minimum of 2 independent experiments. The relative amount oftranscript normalized to 18 S RNA was calculated following standard procedures [26].

Fig. 2. Confocal images showing vimentin (red) in control (A), DCM (B), ICM-RZ (C), ICM-BZnuclei are labeled blue with DAPI. The insert in Panel F shows representative vimentin-positelevated in diseased myocardium, especially in the BZ myocardium of patients with ICM an

2.8. In situ hybridization

ANPwas analyzed using in situ hybridization. First strand cDNAwas synthesized fromtotal humanmyocardial tissue RNA. Reverse transcription reactionswere performedusingSuperscript II (Invitrogen) and oligodT(15) (Promega), followed by PCR reaction with0.2 pmol of each primer targeting ANP: (forward: 5′-CTCCTTCTCCATCACCAAGG-3′;reverse: 5′-CTCTGGGCTCCAATCCTGTC′) and 5U Taq polymerase (Eppendorf). The PCRamplified sequences were purified, cloned into the pCRII-TOPO vector (Invitrogen), andsequenced. Probes were synthesized on TOPO-ANP linearized with HindIII (antisenseprobe) and EcoRI (sense probe) plasmid using T7 and Sp6 RNA polymerase, respectively(Promega) and labeledwith digoxigenine. In situ hybridizationwas carried out in paraffinembedded human heart samples using an alkaline phosphatase conjugated antidigox-igenine antibody as described [27,28]. The ANP intensity was determined using Image Jand expressed as intensity units per myocardial area.

(D) andMYO (E). Myocytes are stained greenwith phalloidin-conjugated with FITC, andive cells used for quantification. F: The number of vimentin-positive cells is significantlyd in patients with MYO.

22 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

2.9. Statistical analysis

All data are presented as means±SD. For multiple comparisons we used ANOVA,followed by analysis with the Bonferroni t-test. Associations between variables wereassessed using Spearman's rank correlation. Differences between groups wereconsidered significant at pb0.05.

3. Results

3.1. Collagens I and III

Collagens I and III were used as indicators of fibrosis and weredetermined separately; the typical localization is shown in Fig. 1 andin Fig. S1. It is evident that the amount of collagen III in all tissues wasmuch higher than that of collagen I, with the exception of collagen I inICM-BZ which was similar to collagen III. Vimentin immunolabelingwas used as a marker of fibroblasts. As shown in Fig. 2, as compared

Fig. 3. Quantitative data of collagen I (A), collagen III (B) and the collagen III/collagen I ratiocollagen I (E), and the collagen III/collagen I ratio (F).

with control, the number of vimentin-positive cells were apparentlyincreased in all groups, especially in ICM-BZ and MYO. These findingswere confirmed by the quantitative determination of the number ofvimentin-positive cells (Fig. 2F). Morover, the number of vimentin-positive cells significantly correlated with both, collagen I andcollagen III (Fig. S2).

Quantitative determination of the amount of collagen I and collagenIII is shown in Fig. 3. Collagen Iwas3% innormal tissueand itwas10-foldincreased in ICM-BZ and 2–3 times elevated in the other 3 groups(Fig. 3A). Collagen III was 9% in normal tissue and significantly increasedin all groupswith ICM-BZexhibitinga peakvalue of almost 30%(Fig. 3B).The ratio collagen III/collagen I was 3.3 in normal tissue; it was lower inall disease groups (Fig. 3C). In ICM-BZ the collagen III/I ratio was 1.0indicating that equal amounts of both types of collagen were presentwhich was due to the tremendous increase of collagen I (Fig. 3C). Asignificant correlation was found to exist between the amount of

in control and in the failing heart (C). Relationship between LVEDP and collagen III (D),

23V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

collagen I and the increase of LVEDP in all patients (Fig. 3E). This wasabsent for collagen III (Fig. 3D) but the ratio between both correlatedclosely with an elevated LVEDP (Fig. 3F).

3.2. Markers of collagen turnover

Fig. 4 shows that PINP (newly synthesized collagen I) is confined tothe interstitial space and perivascular areas (Fig. 4D). In comparisonwith control myocardium, this marker of collagen I synthesis wasfound to be increased in all groups. Similar localization anddistribution was observed using immunolabeling for PIIINP (Fig. 5A–D and Fig. S3).

Fig. 4. Representative immunofluorescent confocal images of PINP (green) in control (A), DCMaccumulation of newly synthetized collagen I in diseased myocardium. Note the perivasculabeled red with TRITC-phalloidin, nuclei are stained blue with DAPI.

Quantitative analsysis showed that PINP was almost 5 foldupregulated in ICM-BZ and 3–4 times higher than control in the otherthree groups (Fig. 5E). ICTP indicating collagen degradation was threetimes higher than control in all groups (Fig. 5F). PIIINP was especiallyincreased in DCM and ICM-RZ; it was also elevated in MYO but notdifferent fromcontrol in ICM-BZ (Fig. 5G).All groupswere characterizedby a significant elevation of IIINTP indicating mature cross-linkedcollagen III (Fig. 5H).

The ratio between PINP and ICTP was 1.83 in control. It wassignificantly elevated in ICM-BZ and MYO but decreased in ICM-RZand DCM. The ratio of PIIINP/IIINTP was decreased in ICM-BZ incomparison to control. The PINP/PIIINP ratio was significantly

(B), in RZ (C,D) and BZ regions (E) in ICM, and inMYO (F) demonstrating an increasedlar localization (V — vessels) of PINP in RZ from patients with ICM (D). Myocytes are

Fig. 5. Representative immunofluorescent images of aminoterminal telopeptide of type III collagen (IIINTP) in control (A), DCM (B), and in RZ (C) and BZ regions (D) in ICMdemonstrating an increased accumulation of mature collagen III in diseased myocardium. IIINTP is green, myocytes are labeled red with TRITC-phalloidin, and nuclei are stained bluewith DAPI. Quantitative data of markers of collagen type I (PINP in Panel E and ICTP in Panel F) and collagen type III (PIIINP in Panel G and IIINTP in Panel H) in control and in thefailing heart.

24 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

increased in ICM-BZ while the ICTP/IIINTP ratio remained unchanged(Table 2).

3.3. MMP expression

In all tissues, MMP labeling was confined to the interstitial spaceand the perivascular area (Fig. 6). It was obvious that in failing

myocardium (Fig. 6B–E), the different MMPs were significantlyincreased in comparison with control (Fig. 6A).

In different groups of patients, MMPs were produced by differentinterstitial cells, i.e. macrophages and CD3 lymphocytes in MYO(Figs. 7A and 6B) and in ICM-BZ by myofibroblasts (Figs. 7C and 6D).In contrast, in DCM and in ICM-RZ, the main cells in the MMP areawere fibroblasts (Fig. 7E and F).

Table 2Ratios of collagen metabolism markers and of total MMPs/total TIMPs.

Control DCM ICM-RZ ICM-BZ Myo

PINP/ICTP 1.83±0.40 1.46±0.10b 1.53±0.14 2.56±0.40a 2.08±0.10PIIINP/IIINTP 1.76±0.2 1.82±0.32b 1.48±0.14 0.67±0.07a 1.13±0.17PINP/PIIINP 0.78±0.16 1.07±0.29 0.98±0.1b 2.29±0.4a 1.21±0.23ICTP/IIINTP 1.07±0.48 1.16±0.15 0.89±0.05 0.61±0.18 0.75±0.15TIMPs/MMPs 1.29±0.10 1.08±0.19b 1.41±0.18 1.90±0.12 0.74±0.13b

a vs control.b vs ICM-BZ.

Fig. 6. Representative micrographs of MMP2 (green) staining patterns in control (A), DCM (B(E). Note the perivascular localization (V— vessels) of MMP2 in a patient with MYO (E). Panearea. Myocytes are stained red with phalloidin-conjugated with TRITC. Nuclei are stained b

25V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

Immunofluorescent quantitative analysis showed that all MMPswere increased in all groups, especially in ICM-BZ, with the exceptionof MMP9 (Fig. 8A–D). IncreasedMMP2 levels in diseasedmyocardiumas determined by quantitative immunofluorescent analysis wereconfirmed by Western blot (Fig. 8E). Similar results were obtained inother previously published studies [8,13,29]. In addition, zymographyshowed that the activity of both, MMP2 and MMP9 was significantlyupregulated in all groups in comparison to control (Fig. 8F and G).

3.4. TIMP expression

TIMP1, TIMP2, TIMP3 and TIMP4 were all found in the ECM (Fig. 9).TIMPswere increased in all groups butMYO showed the lowest amount

), and in the RZ (C) or BZ (D)myocardium of patients with ICM and in patient with MYOl F is the identical image as shown in E to show the overall morphology of a perivascularlue with DAPI.

Fig. 7. Serial sections stained for MMP8 (green, A), CD68 (red, B) and CD3 (green, B) in a patient with MYO demonstrating co-localization of MMP8 with macrophages andlymphocytes. Serial sections of a BZ region in a patient with ICM stained for MMP2 (green, C) and vimentin (red, D) and smooth muscle α-actin (green, D) showing that MMP2 is co-localized with myofibroblasts (arrows). Serial sections stained for MMP9 (green, E) and vimentin (red, F) in a patient with DCM showing co-localization of MMP9 with fibroblasts(arrows). In all micrographs nuclei are stained blue with DAPI.

26 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

of TIMP1 and TIMP2 and only TIMP3 was significantly elevated incomparison with control (Fig. 9). The expression levels of the differentTIMPs significantly correlated with each other (Fig. S4). From differentTIMPs, only TIMP4 correlated positively withmarkers of collagen type Iand type III synthesis (PINP, r=0.51, pb0.001 and PIIINP, r=0.32,pb0.05).

The ratio between total MMP/total TIMP was 1.29 in control; it washighest in ICM-BZ and slightly decreased in the other groups (Table 2).

3.5. Expression of cytokines

TGF-ß1-labeling was confined to fibroblasts and macrophages. Inthe control myocardium we found only few TGF-ß1-positive cells inthe interstitium (Fig. 10A) but in all disease groups a significantelevation of TGF-ß1-containing cells was observed (Fig. 10B–F).

RT-PCR showed significantly elevated levels of fibrosin mRNA inICM-BZ; in MYO a markedly increased level of activin A mRNA wasdocumented but FGF1 and FGF2 mRNAs remained unchanged in allpatients in comparison with control tissues (Fig. 11).

In situ hybridization showed that in comparison with controltissues, the ANP mRNA was significantly increased in all groups(Fig. 12).

4. Discussion

The major results of the study are the following: All values forcollagens and markers of collagen metabolism, as well as MMPs andTIMPs are low in control tissue and all including themarkers for newlysynthetized and mature collagens are elevated in failing hearts. Inaddition, the failing myocardium is characterized by increasednumbers of fibroblasts and TGF-ß1-positive cells. Based on thesefindings, a schematic representation of different mechanisms involv-ing ECM cells, MMPs, TIMPs, and collagen metabolism resulting inprogression fibrosis that is typical is depicted in Fig. 13.

It is noteworthy that the range of collagen increasewas found to behighest in ICM-BZ and lowest in MYO. The same applies to changes inMMPs and TIMPs. In DCM and ICM-RZ all these parameters aremoderately increased. Taken together, the findings of the presentwork and our previous studies suggest that structural alterations ofthe ECM are rather specific for each type of cardiac disease thansimply due to heart failure.

4.1. Alterations in ICM-BZ and ICM-RZ

In the ICM-BZ the changes of collagen composition were the mostobvious with a 10-fold increase in collagen I and a 3-fold increase of

Fig. 8. Quantitative immunofluorescent data of MMP2 (A), MMP8 (B), MMP9 (C) and MMP19 (D) expression (percent per myocardial tissue area) in control and diseasedmyocardium. Panel E shows a typical Western blot for MMP2 and quantitative protein data in control and in the DCM, ICM-RZ, ICM-BZ and MYO groups. Data are expressed aspercentage relative to control. Typical MMP9 andMMP2 gelatinolytic activities and their quantitative analysis in control and in the DCM, ICM-RZ, ICM-BZ andMYO groups (F and G).Data are expressed as percentage relative to control.

27V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

collagen III which resulted in a high level of fibrosis of almost 30%. Thiswas accompanied by a moderate increase in MMP concentrationwhereas the TIMPs showed the highest values of all patients studiedindicating an elevated antiproteolytic control resulting finally in

prevalence of ECM synthesis over its degradation and therefore infibrosis. An increased collagen turnover was evident from the elevatedlevels of both, PINP indicating newly synthetized collagen and ICTP as amarker of collagen degradation, which contains peptide material from

Fig. 9. Representative micrographs of TIMP4 (green) staining patterns in a control (left panel) and in the BZ myocardium of a patient with ICM (right panel). Note the markedincreased TIMP in the latter case. Myocytes are stained red with phalloidin-conjugated with TRITC. Nuclei are stained blue with DAPI. The diagrams represent quantitativeimmunofluorescent data of TIMPs expression in control and diseased myocardium.

28 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

three different polypeptide chains [30]. The elevation of PINP prevailedindicating increased synthesis over degradation. The turnover ofcollagen I seemed to be more important than that of collagen III.However, an up-regulation of IIINTP was found in all patients, whereasPIIINP was significantly increased only in ICM-RZ. The up-regulation ofIIINTP indicates ongoing collagen type III production and cross-linking.These data are consistent with experimental reports demonstrating anactivation of collagen synthesis not only in the infarcted area but also inthe RZ and BZ [5,31,32]. These ischemic regions can therefore not beregarded as static tissue. Since the mature collagens I and III werestrongly increased, we hypothesize that they persist because of theinhibition of MMP2 and MMP8 by the elevated TIMP levels.

MMP2 is themost universalMMP and is responsible for the varying-size degradation products of native type I collagen. Up-regulation ofMMP2 found in this study confirmed results of Etoh et al. who reportedde novo synthesis and release of MMP2 in pigs with coronarysyndromes [33]. Numerous studies have been carried out in experi-mental animal models and most of these showed increased MMP2release from acutely infarcted tissue (see [5]). Especially interesting isthe work by Wilson [34] who presented a comprehensive overview ofMMP changes with time after infarction. MMP2was upregulated in thezone of transition 10–12 days after infarction which might becomparable to the up-regulation of MMP2 observed in the ICM-BZ inthe present study.

Fig. 10. Confocal images showing TGF-β1 expression in control (A), DCM (B), ICM-RZ (C), ICM-BZ (D) andMYO (E). Myocytes are stained red with phalloidin-conjugated with TRITC.The insert in Panel E is a higher magnification of the boxed region showing a perivascular accumulation of TGF-β1-positive cells in a patient with MYO. F: The number of TGF-β1-positive cells is significantly elevated in diseased myocardium, especially in the BZ myocardium of patients with ICM.

29V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

MMP8 or collagenase 2 cleaves the collagens type I, II, and III andmay exert a coordinate action with MMP2. It was upregulated in bothzones of ICMwhich comfirmsWilson's report on elevatedMMP8 levelsin the periinfarct zone [35]. Similarly, a specific temporal profile ofMMP8 and MMP9 up-regulation was found in post-MI patients [35].

In our study, MMP9 seems not to be essential neither in border norin remote ischemic areas. This is in contrast to other work reportingattenuation of left ventricular enlargement and collagen accumulation

by targeted MMP9 deletion [36]. Furthermore, in a recent study, up-regulation of active MMP9 in rat heart infarcts was observed [37].Both reports, however, described elevation of MMP9 in acutelyinfarcted rodent myocardium most probably reflecting an acuteinflammatory reaction whereas the present study investigated tissuefrom border and remote zones of longstanding old infarcts in humanpatients with cardiac failure. This difference in models and tissuesstudied might explain the differing results.

Fig. 11. Expression levels of fibrosin 1 (A), activin A (B), FGF1 (C) and FGF2 mRNAs (D) in the myocardium of patients with DCM, ICM, MYO and in the control group obtained byRT-PCR analysis.

30 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

In contrast toMMP9,MMP19, one of the recently discoveredMMPsrepresenting a novel subclass [38,39], was significantly augmented,which has not yet been described in the human heart. MMP19 has ahighly potent cleaving activity on awide variety of ECM substrates andhas a tendency to autoactivation [39]. Such MMPs cannot be inhibitedby tissue inhibitors because TIMPs are only able to inhibit the activeenzyme but not the autoactivation. In the present study, we found asignificant up-regulation of MMP19 in ICM-BZ as compared to controlreflecting its important role in the myocardial remodeling process.

4.2. Alterations in DCM

In tissue frompatientswithDCMcollagens I and IIIwere significantlyelevated resulting in a ratio of 1.9 and fibrosis was evident. Collagen IIIwas especially high and we hypothesize that the accumulation andmaturation of collagen type III, with its relatively smaller diameter andmore elasticity than collagen type I fibers, is most probably an adaptivereaction tohighmyocardial stiffness causedby thepresenceof the large-diameter, cross-linked fibers of collagen type I. The increase in collagenIII is reflected also in the elevated levels of PIIINP, which resulted finallyin a 3fold increase ofmature cross-linked collagen III (IIINTP).While thecollagen metabolism was active, the MMPs 2, 8 and 19 were onlymoderately increased while MMP9 exhibited the highest value of allgroups studied. TIMP1 and 2 were slightly increased but TIMP 3 andTIMP4 were significantly elevated. These findings confirm earlierstudies describing anelevationofMMP2andMMP9and thedistribution

of MMP and TIMP in myocardium from DCM patients [5,40]. It is ofinterest that Sawicki et al. not only foundMMP2 andMMP9 elevated inDCM, but that they identifiedMMP2within themyocytes andwere ableto show that myosin can be degraded by this protease [41]. This mightbe an explanation for the disappearance of contractile proteins frommyocytes in DCM, which our group reported earlier [42]. As comparedto the ICM-BZ region, the TIMPs showed a rather low expression, whichis in accordancewith reports byHtunet al.where theprotein abundanceof all TIMPs was significantly reduced in DCM [43]. Since in the presentstudy the MMPs were only moderately increased, the net result of theMMP/TIMP balance would favor matrix remodeling resulting in anincreased degree of fibrosis. In addition, oncotic death of myocytes, aphenomenon commonly observed in endstage DCM most probably isfacilitated by the breakdown of the basal lamina of myocytes by MMP9and further contributes to the formation of fibrosis and therefore toadverse remodeling [22,44].

4.3. Alterations in MYO

InMYO the collagen contentwasonly slightly higher than in controlsbut themarkers of collagenmetabolismwere increased indicating rapidturnover of fibrotic material. The MMPs showed a low abundance andonly TIMP3 and TIMP4 were significantly enhanced. Given thatexpression levels of TIMP4 directly correlated with PINP and PIIINP, itis tempting to speculate that the inhibitory effect of the cardiac specificTIMP4 may play an important role in collagen synthesis.

Fig. 12. In situ hybridization of ANP mRNA using sense (A) and antisense (AS) probes (A–F) in the myocardium of control (C) and in patients with DCM (D), ICM-RZ (E) and ICM-BZ(F) and MYO (G). In panel F, MI - myocardial infarct. Note the absence of the ANP mRNA signal using sense probes (A) and in control patients (C). Human atrial tissue was used as apositive control (B). Panel H shows quantitative data of the ANP mRNA expression. Data are expressed as fold change relative to control.

31V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

In the present study, we found that in patients with MYO, as wellas in ICM-BZ, the number of fibroblasts is significantly higher than incontrol tissues. Although, it is for a long time believed that fibroblastreaction is a secondary phenomenon following any cardiac insult,recent studies indicate that fibroblasts may play a primary role inECM remodeling and myocardial diseases [45]. Therefore, it istempting to predict that targeting the fibroblast activity will be anew attractive therapeutical approach for treating the detrimentalECM remodeling.

Most studies on myocarditis have been carried out in mice. In arecent publication, McManus and his group showed that lack of MMP9has deleterious effects on viral infected myocardium whereas targeted

ablation of MMP8 did not exhibit any additional injurious action [46].Pauschinger et el reported an increase in MMP and a decrease in TIMP,which is in contrast to our results [47]. Most probably, the difference iscaused by the fact that most studies investigated the phase of acutemyocarditis whereas our patients were in a stage of chronic long-standing healing or healed inflammation.

4.4. mRNA expression of regulatory factors

Fibrosin 1 is a fibrogenic cytokine which modulates the expressionand differentiation of myofibroblasts [48]. The increased number ofmyofibroblasts in ICM-BZ may have caused the increased expression

Fig. 13. Schematic representation of different mechanisms active in collagenmetabolism.Up-regulation of both, MMPs, and TIMPs results in progressive fibrosis by interactionwithECM cells and cytokines (not shown).

32 V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

of fibrosin 1 whereas in the ICM-RZ, where an increased number ofmyofibroblasts was not evident, fibrosin 1 remained at the controllevel.

One of the important and novel finding of this study concerns the up-regulation of activin A, which was found only in patients with MYO.Activin A is a known activator of inflammatory cells [49]. It may cause anaccumulation of lymphocytes and macrophages that in turn areimportant cellular sources of different MMPs [21,50]. Given that allgroups were characterized by different MMP/TIMP profiles and changesin collagen metabolism, the regulatory factors analyzed here probablytake part in the regulation of ECM proteolysis but not in the consecutivecollagen reorganization. The fact that each of the patient groups ischaracterizedby specific changes in expressionof different cytokinesmaypartially explain the divergent MMP portfolios between these groups, atleast in ICM-BZ and MYO.

Increased expression of ANP which has antifibrogenic properties[51] was found in all groups. This can be regarded as a compensatorymechanism to inhibit the progression of fibrosis, which, however, isapparently ineffective in failing myocardium.

5. Conclusions

The present study shows that although different degrees ofseverity of collagen metabolism, MMP/TIMP imbalance and cytokineexpression in diverse forms of HF are present, the end product iscollagen deposition. In addition, the failing myocardium is character-ized by discordant changes in the content of collagen types I and III.This observation implies that multiple mechanisms acting alone or inconcert are active in fibrosis development. This finding is importantbecause the properties of the interstitial collagen matrix canessentially affect the stiffness of the myocardium contributing to thedevelopment of diastolic and systolic dysfunction.

Acknowledgments

We thank Dr. Svetlana Oustanina and Dr. Kerstin Troidl for theirexpert technical assistance in performing RT-PCR. This study wassupported by grants from the Max-Planck-Gesellschaft, München,Germany, for cooperation with the Kerckhoff Clinic, Bad Nauheim,Germany (PFOR404 to V.P.; PFOR371 to V.P. and S.K.; and PFOR401 toS.H., J.S and S.K).

The authors of this manuscript have certified that they complywith the principles of ethical publishing in the International Journal ofCardiology [52].

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi: 10.1016/j.ijcard.2010.04.053.

References

[1] Weber KT, Brilla CG, Janicki JS. Myocardial fibrosis: functional significance andregulatory factors. Cardiovas Res 1993;27:341–8.

[2] Jensen LT, Host NB. Collagen: scafold for repair or execution. Cardiovasc Res1997;33:535–9.

[3] Risteli J, Risteli L. Analysing connective tissue metabolites in human serum.Biochemical, physiological and methodological aspects. J Hepatol 1995;22:77–81.

[4] Bode MK, Soini Y, Melkko J, Satta J, Risteli L, Risteli J. Increased amount of type IIIpN-collagen in human abdominal aortic aneurysms: evidence for impaired type IIIcollagen fibrillogenesis. J Vasc Surg 2000;32:1201–7.

[5] Spinale FG. Myocardial matrix remodeling and the matrix metalloproteinases:influence on cardiac form and function. Physiol Rev 2007;87:1285–342.

[6] Pauschinger M, Knof D, Petschauer S, et al. Dilated cardiomyopathy is associatedwith significant changes in collagen type I/III ratio. Circulation 1999;99:2750–6.

[7] Peterson JT, Li H, Dillon L, Bryant JW. Evolution of matrix metalloprotease andtissue inhibitor expression during heart failure progression in the infarcted rat.Cardiovasc Res 2000;46:307–15.

[8] Polyakova V, Hein S, Kostin S, Ziegelhoeffer T, Schaper J. Matrix metalloproteinasesand their tissue inhibitors in pressure-overloaded human myocardium duringheart failure progression. J Am Coll Cardiol 2004;44:1609–18.

[9] Anne W, Willems R, Roskams T, et al. Matrix metalloproteinases and atrialremodeling in patients with mitral valve disease and atrial fibrillation. CardiovascRes 2005;67:655–66.

[10] Deten A., Marx G., Briest W., Christian Volz H., Zimmer H.G. Heart function andmolecular biological parameters are comparable in young adult and aged rats afterchronic myocardial infarction. Cardiovasc Res 2005;2005:364-373.

[11] Schwartzkopff B, Fassbach M, Pelzer B, Brehm M, Strauer BE. Elevated serummarkers of collagen degradation in patients with mild to moderate dilatedcardiomyopathy. Eur J Heart Fail 2002;4:439–44.

[12] Lombardi R, Betocchi S, Cacace A, Losi MA, Chiariello M. Myocardial interstitialfibrosis and diastolic dysfunction in hypertrophic cardiomyopathy. Ital Heart JSuppl 2003;4:645–50.

[13] Pauschinger M, Rutschow S, Chandrasekharan K, et al. Carvedilol improves leftventricular function in murine coxsackievirus-induced acute myocarditis associ-ation with reduced myocardial interleukin-1beta and MMP-8 expression and amodulated immune response. Eur J Heart Fail 2005;7:444–52.

[14] Villarreal F, Omens J, Dillmann W, Risteli J, Nguyen J, Covell J. Early degradationand serum appearance of type I collagen fragments after myocardial infarction.J Mol Cell Cardiol 2004;36:597–601.

[15] Garcia-Bolao I, Lopez B, Macias A, Gavira JJ, Azcarate P, Diez J. Impact of collagentype I turnover on the long-term response to cardiac resynchronization therapy.Eur Heart J 2008;29:898–906.

[16] Eriksen HA, Satta J, Risteli J, Veijola M, Vare P, Soini Y. Type I and type III collagensynthesis and composition in the valve matrix in aortic valve stenosis. Atheroscle-rosis 2006;189:91–8.

[17] Hunzelmann N, Risteli J, Risteli L, et al. Circulating type I collagen degradationproducts: a new serum marker for clinical severity in patients with scleroderma?Br J Dermatol 1998;139:1020–5.

[18] Ebert W, Muley T, Herb KP, Schmidt-Gayk H. Comparison of bone scintigraphywith bone markers in the diagnosis of bone metastasis in lung carcinoma patients.Anticancer Res 2004;24:3193–201.

[19] Maxwell PR, Flisiak R. Changes in serological biomarkers of liver function andconnective tissue turnover in chronic hepatitis B during lamivudine therapy.Biomarkers 2005;10:475–84.

[20] Jinga DC, Blidaru A, Condrea I, et al. MMP-9 and MMP-2 gelatinases and TIMP-1and TIMP-2 inhibitors in breast cancer: correlations with prognostic factors. J CellMol Med 2006;10:499–510.

[21] Pistol G, Matache C, Calugaru A, et al. Roles of CD147 on T lymphocytes activationand MMP-9 secretion in systemic lupus erythematosus. J Cell Mol Med 2007;11:339–48.

[22] Hein S, Arnon E, Kostin S, et al. Progression from compensated hypertrophy tofailure in the pressure-overloaded human heart: structural deterioration andcompensatory mechanisms. Circulation 2003;107:984–91.

[23] Kai H, Mori T, Tokuda K, et al. Pressure overload-induced transient oxidative stressmediates perivascular inflammation and cardiac fibrosis through angiotensin II.Hypertens Res 2006;29:711–8.

[24] Polyakova V, Miyagawa S, Szalay Z, Risteli J, Kostin S. Atrial extracellular matrixremodelling in patients with atrial fibrillation. J Cell Mol Med 2008;12:189–208.

[25] Troidl C, Troidl K, Schierling W, et al. Trpv4 induces collateral vessel growth duringregeneration of the arterial circulation. J Cell MolMed 2008, doi:10.1111/j.1582-4934.2009.00707.x.

[26] PfafflMW.Anewmathematicalmodel for relative quantification in real-timeRT-PCR.Nucleic Acids Res 2001;29:e45.

[27] Schneider M, Kostin S, Strøm C, et al. S100A4 is upregulated in injured myocardiumand promotes growth and survival of cardiac myocytes. Cardiovasc Res 2007;75:40–50.

[28] Loch T, VakhrushevaO, Piotrowska I, et al. Different extent of cardiacmalfunction andresistance to oxidative stress in heterozygous and homozygous in manganese-dependent superoxide dismutase-mutant mice. Cardiovasc Res 2009;82:448–57.

[29] Li J, Schwimmbeck PL, Tschope C, et al. Collagen degradation in amurinemyocarditismodel: relevance of matrix metalloproteinase in association with inflammatoryinduction. Cardiovasc Res 2002;56:235–47.

[30] Sassi M, Jukkola A, Riekki R, et al. Type I collagen turnover and cross-linking areincreased in irradiated skin of breast cancer patients. Radiother Oncol 2001;58:317–23.

33V. Polyakova et al. / International Journal of Cardiology 151 (2011) 18–33

[31] Gaertner R, Jacob MP, Prunier F, Angles-Cano E, Mercadier JJ, Michel JB. Theplasminogen-MMP system is more activated in the scar than in viable myocardium3 months post-MI in the rat. J Mol Cell Cardiol 2005;38:193–204.

[32] Sun Y, Zhang JQ, Zhang J, Lamparter S. Cardiac remodeling by fibrous tissue afterinfarction in rats. J Lab Clin Med 2000;135:316–23.

[33] Etoh T, Joffs C, Deschamps AM, et al. Myocardial and interstitial matrixmetalloproteinase activity after acute myocardial infarction in pigs. Am J PhysiolHeart Circ Physiol 2001;281:H987–94.

[34] Wilson EM, Gunasinghe HR, Coker ML, et al. Plasma matrix metalloproteinase andinhibitor profiles in patients with heart failure. J Card Fail 2002;8:390–8.

[35] Webb CS, Bonema DD, Ahmed SH, et al. Specific temporal profile of metallpro-teinase release occurs in patients after myocardial infarction: relation to leftventricular remodeling. Circulation 2006;114:1020–7.

[36] DucharmeA, Frantz S,AikawaM, et al. Targeteddeletionofmatrixmetalloproteinase-9attenuates left ventricular enlargement and collagen accumulation after experimentalmyocardial infarction. J Clin Invest 2000;106:55–62.

[37] Abbate A, Salloum FN, Vecile E, et al. Anakinra, a recombinant human interleukin-1receptor antagonist, inhibits apoptosis in experimental acute myocardial infarction.Circulation 2008;117:2670–83.

[38] Murphy G, Knauper V, Cowell S, et al. Evaluation of some newer matrixmetalloproteinases. Ann NY Acad Sci 1999;878:25–39.

[39] Stracke JO, Hutton M, Stewart M, et al. Biochemical characterization of thecatalytic domain of human matrix metalloproteinase 19. Evidence for a role as apotent basement membrane degrading enzyme. J Biol Chem 2000;275:14,809–16.

[40] Yokoseki O, Yazaki Y, Suzuki J, Imamura H, Takenaka H, Isobe M. Association ofmatrix metalloproteinase expression and left ventricular function in idiopathicdilated cardiomyopathy. Jpn Circ J 2000;64:352–7.

[41] Sawicki G, Leon H, Sawicka J, et al. Degradation of myosin light chain in isolated rathearts subjected to ischemia-reperfusion injury: a new intracellular target formatrix metalloproteinase-2. Circulation 2005;112:544–52.

[42] Schaper J, Froede R, Hein S, et al. Impairment of the myocardial ultrastructure andchanges of the cytoskeleton in dilated cardiomyopathy. Circulation 1991;83:504–14.

[43] Htun P, ItoWD, Hoefer IE, Schaper J, Schaper W. Intramyocardial infusion of FGF-1mimics ischemic preconditioning in pig myocardium. J Mol Cell Cardiol 1998;30:867–77.

[44] Kostin S, Pool L, Elsasser A, et al. Myocytes die by multiple mechanisms in failinghuman hearts. Circ Res 2003;92:715–24.

[45] Thum T, Gross C, Fiedler J, et al. MicroRNA-21 contributes to myocardial disease bystimulating MAP kinase signalling in fibroblasts. Nature 2008;456:980–4.

[46] Cheung C, Marchant D, Walker EK, et al. Ablation of matrix metalloproteinase-9increases severity of viral myocarditis in mice. Circulation 2008;117:1574–82.

[47] Pauschinger M, Chandrasekharan K, Schultheiss HP. Myocardial remodeling inviral heart disease: possible interactions between inflammatory mediators andMMP-TIMP system. Heart Fail Rev 2004;9:21–31.

[48] Prakash S, Paul WE, Robbins PW. Fibrosin, a novel fibrogenic cytokine, modulatesexpression of myofibroblasts. Exp Mol Pathol 2007;82:42–8.

[49] Phillips DJ, Jones KL, Scheerlinck JY, Hedger MP, de Kretser DM. Evidence foractivin A and follistatin involvement in the systemic inflammatory response. MolCell Endocrinol 2001;180:155–62.

[50] Anghelina M, Moldovan L, Zabuawala T, Ostrowscki MC, Moldovan NL. Asubpopulation of peritoneal macrophages form capillary-like lumens andbranching patterns in vitro. J Cell Mol Med 2006;10:708–15.

[51] Fujisaki H, Ito H, et al. Natriuretic peptides inhibit angiotensin II-inducedproliferation of rat cardiac fibroblasts by blocking endothelin-1 gene expression.J Clin Investig 1995;96:1059–65.

[52] Coats AJ. Ethical authorship and publishing. Int J Cardiol 2009;131:149–50.