Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

Contents lists available at ScienceDirect

Journal of Molecular and Cellular Cardiology

j ourna l homepage: www.e lsev ie r.com/ locate /y jmcc

Original article

Cardiac interstitial cells express GATA4 and control dedifferentiation and cell cyclere-entry of adult cardiomyocytes

Tania Zaglia a, Arben Dedja d,e, Cinzia Candiotto a, Emanuele Cozzi d,e,f,Stefano Schiaffino a,b,c,⁎, Simonetta Ausoni a,⁎a Department of Biomedical Sciences, University of Padua, Padua, Italyb CNR Institute of Neurosciences, Padua, Italyc Venetian Institute of Molecular Medicine (VIMM), Padua, Italyd Department of Medical and Surgical Sciences, University of Padua, Padua, Italye CORIT (Consorzio per la Ricerca sul Trapianto d'Organi), Padua, Italyf Direzione Sanitaria, Padua General Hospital, Padua, Italy

⁎ Corresponding authors.E-mail addresses: stefano.schiaffino@unipd.it (S. Sch

simonetta.ausoni@unipd.it (S. Ausoni).

0022-2828/$ – see front matter © 2008 Elsevier Inc. Adoi:10.1016/j.yjmcc.2008.12.010

a b s t r a c t

a r t i c l e i n f o

Article history:

Interstitial cells of the adult Received 28 July 2008Received in revised form 5 November 2008Accepted 11 December 2008Available online 31 December 2008

Keywords:GATA4Cardiac fibroblastsDedifferentiationCardiomyocytesCell cycle re-entry

rat heart were characterized with respect to i) expression of cardiac markers ofcommitment and differentiation, ii) myogenic potential in vitro and iii) ability to modulate cardiomyocytedifferentiation state. We demonstrate for the first time that fibroblasts and a proportion of pericytes in theadult rat heart express the transcription factor GATA4. This appears to be a peculiar property of the heart.Fibroblasts that are also derived from the splanchnopleuric mesoderm, such as those of the gut, or fibroblastsof different embryological origin, such as those of skin and skeletal muscle, lack this property. Of note, anestin+/GATA4+ putative stem cell population is also detected in the adult heart. GATA4+ cardiacinterstitial cells do not display myogenic potential in vitro. However, cardiac fibroblasts, but not skinfibroblasts, stimulate dedifferentiation of adult cardiomyocytes and their re-entry into the cell cycle in vitro,as demonstrated by the high number of cardiomyocytes expressing Ki67, phosphorylated histone H3 (H3P)and incorporating 5-bromodeoxiuridine (BrdU) in the co-cultures. In conclusion, cardiac fibroblasts havepeculiar expression of myogenic transcription factors, a property that may have an impact forreprogramming these cells to the myogenic differentiation. In addition, they are able to modulate thebehavior of adult cardiomyocytes, a property that may be used to promote dedifferentiation and proliferationof cardiac cells in the damaged myocardium.

© 2008 Elsevier Inc. All rights reserved.

1. Introduction

Cardiomyocytes represent approximately 75% of the normal heartvolume, but only 30–40% of the total number of cells of an adult heart.The remaining tissue accounts for cardiac interstitial cells, mainlyfibroblasts and vascular cells that provide the adequate scaffold formyocardial contractility and the proper perfusion. Cardiac fibroblastsare important determinants of structure and function of themyocardium inasmuch as they contribute to structural, biochemicaland mechanical properties of the cardiomyocytes [1]. In addition,cardiac fibroblasts generate paracrine signals that directly influencecardiomyocyte trophism, contractility and survival in vivo and in vitro[1–3]. The embryological origin of cardiac fibroblasts is still discussed.During embryogenesis the epicardium contributes to cardiac fibro-blasts [4], but it remains questioned whether other cell sources,

iaffino),

ll rights reserved.

included the bone marrow, are involved in the postnatal fibroblastrecruitment, either in normal [4] or diseased heart [5,6].

GATA4 is a zinc-finger transcription factor that acts as a criticalregulator of the cardiac differentiation-specific gene program. Duringembryogenesis GATA4 is required for myocardial and coronary vesseldevelopment and GATA4 null mice are embryonic lethal due tomultiple abnormalities, including hemodynamic dysfunctions [7]. Inaddition to regulating cardiogenesis and transcription of numerouscardiac genes in the heart [8,9], GATA4 controls survival and stressresponsiveness of adult cardiomyocytes and hypertrophy [10–13].Unexpectedly, during the hypertrophic response, up-regulated cardi-omyocyte GATA4 promotes angiogenesis, presumably via activation ofthe entire programs of angiogenic cytokines and growth factors [14].More recently, GATA4 has been also identified in stem and progenitorcells of the heart in combination with stemness markers [15–18].

In this study we looked for interstitial cells of the heartexpressing cardiac markers with the aim to identify cell populationswith wide tissue distribution and potentially suitable for cardiacregeneration. We demonstrate here, for the first time, that fibroblasts

654 T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

and pericytes, but not endothelial and smooth muscle cells of theadult heart express GATA4 and that this expression is organ-specific.These cells lack myogenic properties in vitro. However, in co-cultureexperiments they stimulate dedifferentiation and cell cycle re-entryof adult cardiomyocytes, a potential additional strategy for promo-ting cardiac regeneration.

2. Methods

2.1. Tissue samples

Two month-old adult (∼200 g) and 2, 5 and 11 day-old neonatalSprague Dawley rats (Harlan Italia, Italy) and adult transgenic rats ofcomparable weight expressing the green fluorescent protein (GFP)under the control of the cytomegalovirus enhancer and the chickenbeta-actin promoter [19], were used. The heart, stomach and intestinewere either directly frozen in liquid nitrogen or immediately fixed in1% or 2% paraformaldehyde at room temperature for 1 or 2 h,equilibrated in sucrose gradient, frozen in liquid nitrogen, cryosec-tioned and processed.

2.2. Experimental skin and skeletal muscle lesion

Skin lesion was induced in the posterior legs of anesthetized2month-old Sprague Dawley rats (200 g) by application for 2min. of ametal block frozen in liquid nitrogen. Skeletal muscle injury wasinduced by bupivacaine as previously described [20]. TERRAMYCIN® L.A. (Pfizer, Milan, Italy) antibiotic (60 mg/kg) was administeredsubcutaneously and the analgesic Tramadol (5 mg/kg) was injectedintramuscularly in the first two postoperative days. Five days afterinjury regenerating skin and muscles were removed, frozen in liquidnitrogen, cryosectioned and processed.

2.3. Langendorff perfusion, cardiac interstitial cell culture andco-cultures

Langendorff perfusion was applied to hearts from both wild-type(n=10) and GFP+ transgenic rats (n=10) to isolate cardiacinterstitial cells and cardiomyocytes. GFP+ rats were specificallyused for the purpose to set up co-cultures of cardiomyocytes andinterstitial cells in which cardiomyocytes or interstitial cells could bedistinguished using a genetic marker. Briefly, the heart was excisedfrom euthanized, heparinated rats, quickly rinsed in ice-cold perfusionbuffer (123 mM NaCl, 5.4 mM KCl, 20 mM NaHCO3, 1 mM NaH2PO4,0.2% glucose, 1.7 mM Mg2+, 20 mM C4H7NO2) and immediatelycanulated through the aorta. Perfusionwith oxygenated buffer at 37 °Cwas applied to remove blood cells. Cell dissociation was obtained byperfusing the heart with B collagenase (1.5 mg/ml) (Roche, Milan,Italy) in perfusion solution containing 0.25 µM Ca2+ at 37 °C for40 min. using a peristaltic pump. The enzymatic solutionwas replacedwith 0.25 µM Ca2+ solution and the ventricles were minced.Cardiomyocytes were collected by centrifugation at 200 rpm for2 min. and the supernatant was filtered sequentially with 40, 20 and15 µm nylon mesh. Cardiac interstitial cells were collected bycentrifugation and maintained for up to 10 days in culture to testtheir myogenic potential. Different culture conditions were tested:

1) fetal cardiac cell medium containing 2.5% FBS and 5% HS (Gibco,Milan, Italy), 17% Medium199 (Sigma, Milan, Italy) and 75% DMEM(Sigma, Milan, Italy);2) fetal cardiac cell medium supplemented with 1% fetal heartextract;3) medium for adult cardiomyocytes containing DMEM supple-mented with 2% FBS, 1× insulin–transferrin–selenate cocktail(Invitrogen, Milan, Italy), 5 mM taurine and 5 mM creatine(Sigma, Milan, Italy);

4) enriched medium containing DMEM supplemented with 10%FBS, 1× insulin–transferrin–selenate cocktail (Invitrogen, Milan,Italy), 20 ng/μl FGF (PEPROTECH, Florence, Italy) and 10 mMdexamethasone [21];5) treatment with 3 µM 5-azacytidine (Sigma, Milan, Italy) for3 days, followed by replacement of freshmedium andmaintenancefor 3, 7 and 10 additional days [16];6) different coatings, such as fibronectin (25 µg/ml), laminin(20 µg/ml), gelatine (10 mg/ml) and poly-D-lysine (20 µg/ml);7) co-cultures of cardiac interstitial cells and adult cardiomyocytes,using bothGFP+cardiomyocytes andGFP− cardiac interstitial cells;8) co-cultures of GFP+ cardiomyocytes and primary adult GFP−skin fibroblasts maintained in non conditioned medium, condi-tioned medium derived from cardiac interstitial cells alone(conditioned medium A) and conditioned medium from cardiacinterstitial cells+cardiomyocytes (conditioned medium B).

In 7) and 8) cardiomyocytes and other cell types were plated at a1:100 ratio, but other ratios, such as 1:50, 1:200 and 1:400, weretested in initial experiments to optimize the protocol. Assays 1), 2)and 3), which had not been used previously to differentiatepluripotent cells, were validated using P19CL6 cell line.

2.4. Immunofluorescence, immunohistochemistry, BrdU treatment anddetection

Cryosections were processed for immunofluorescence and ana-lyzed by confocal microscope essentially as previously described[22,23]. Cultured cells were fixed in 4% paraformaldehyde for 30 min.at 4 °C, permeabilized with 0.1% Triton X-100 and processed forimmunofluorescence. Primary antibodies used in this study are listedin Table 1 Supplementary data. For BrdU incorporation in vitro, 5 daysafter plating co-cultures were added with 30 μMBrdU andmaintainedup to day 7 or added at day 7 and maintained up to day 10. BrdUdetection in cultured cells and in tissue sections was performed usingan anti-BrdU antibody (Sigma, Milan, Italy), essentially as described inthe manufacturer's instructions.

2.5. RNA extraction and RT-PCR

Total RNA was extracted from different tissues and cells usingTrizol reagent (Invitrogen, Milan, Italy), as recommended by themanufacturer. cDNA was synthesized using the Superscript III kit(Invitrogen, Milan, Italy) and the following PCR fragments wereamplified using appropriate primers: NKx2.5; GATA4; alpha myosinheavy chain (alpha-MyHC) and beta-actin, as housekeeping gene.Primers used in this study are listed in Table 2 Supplementary data.

2.6. Electrophoresis and Western blotting

Total extracts were obtained from rat cultured fetal cardiomyo-cytes, 10 days cultured cardiac interstitial cells and cultured skinfibroblasts. Equal amounts of proteins were separated on 8% SDS/PAGE, transferred onto nitrocellulose and processed with goat andrabbit anti-GATA4 antibodies, anti-troponin (Tn) T and anti-lamin A/Cantibody for 3 h at room temperature. The blots were then incubatedwith secondary antibodies conjugated to horseradish peroxidase andthe reactivity was revealed by enhanced chemiluminescence (Pierce,Milan, Italy).

2.7. Cell counts and statistical analysis

To estimate the overall number of GATA4+ cardiac interstitial cellswe used 3 adult hearts and a total of 15 cryosections, 5 from eachanatomical level: atrium, apex and mid-portion of the ventricle.

655T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

Positive cells were counted in 12 randomly chosen fields of69,000 μm2 per section, corresponding to a total of 180 fields/heart.The number of GATA4+ fibroblasts and pericytes was estimated in 3additional adult hearts using the same procedure. To count thenumber of GATA4+ cardiac interstitial cells in the neonatal heart, weused 2 hearts for each stage and analyzed six non-consecutivecryosections at the mid-ventricle.

For cell counts in the co-culture experiments comparison betweengroups was performed by Student's t test. Results are alwayspresented as means±SD.

3. Results

3.1. Cardiac interstitial cells express GATA4

We hypothesized that the adult heart contains interstitial cellspotentially suitable for cardiac regeneration. For this reason weinvestigated the expression of GATA4, one of the transcription factorsactivated at the onset of cardiogenesis, using immunofluorescence,Western blotting and RT-PCR. GATA4 was detected not only incardiomyocytes and endocardial cells, but also in a large number of

Fig. 1. GATA4+ interstitial cells are present in the adult heart. (A–D) Cryosections at the mrabbit anti-GATA4 antibody, counterstained with either anti-TnI antibody (A) or anti-laminterstitial cells. (E–G) Cryosection at the mid-level of the ventricle double stained with rabb50 μm. (H)Western blotting incubated with goat or rabbit anti-GATA4 antibodies specificallywas incubated with anti-TnT antibody to ensure for lack of contaminating cardiomyocytes inprotein loading. Molecular weight markers are indicated on the left side of the panel.

interstitial cells (Fig. 1) as determined on adult hearts (15 samples)using three anti-GATA4 antibodies (monoclonal, rabbit polyclonal andgoat polyclonal). All antibodies gave comparable results, thus wechose the rabbit- and the goat-polyclonal sera for further analysesbecause they could be used conveniently in double immunofluores-cence. GATA4+ interstitial cells were observed in both mid-ventricle(Fig. 1A), atrium (Fig. 1B), apex (Fig. 1C) and septum (Fig. 1D) andwere estimated to be in the range of 13.1±0.8 every 100 cardiomyo-cytes. This value corresponds to 36.7±13.4% GATA4+ cells over thetotal number of interstitial cells, with a high standard deviation beingdue to a lower amount of GATA4+ cells in the atrium. GATA4+interstitial cells did not express Nkx-2.5, which, on the other hand,was detected in adult cardiomyocytes (Figs. 1E–G).

Specificity of the rabbit and goat anti-GATA4 antibodies used forimmunofluorescence was validated by Western blotting using totalextracts from adult heart, cardiac interstitial cells and skin fibroblasts.Both antibodies identified a 46 kDa protein in the adult heart and incardiac interstitial cells, but not in skin fibroblasts (Fig. 1H). Detectionof GATA4 in cardiac interstitial cells was not due to contaminationwith adult cardiomyocytes because an antibody specific for cardiacTnT did not detect any elecrophoretic band in this extract. An anti-

id-level of the ventricle (A), atrium (B), apex (C) and septum (D) were incubated withinin (B–D) and analyzed by confocal microscope. Arrows in (A–D) indicate GATA4+it anti-GATA4 (E) and anti-Nkx2.5 (F) antibodies. Merged image is shown in (G). Bars:detected a 46 kDa protein (asterisk) corresponding to GATA4. Control Western blottingthe preparation of cardiac interstitial cells and with anti-lamin A/C to ensure for equal

Fig. 2. Fibroblasts and pericytes of the adult heart express GATA4. (A–J) Cryosections of normal adult rat heart were processed for immunofluorescence and analyzed by confocalmicroscope. (A, B) Heart sections from GFP+ transgenic rat hearts were incubated with rabbit anti-GATA4 antibody and counterstained with DAPI. Merged images are shown. (C–J)Cryosections incubated with rabbit or goat anti-GATA4 antibody and antibodies to: P4H (C), vimentin (D), NG2 (E), CD31 (F), Smooth Muscle(SM)-MyHC (G), c-kit (H), Sca1 (I) andnestin (J). Bars: (A, B, D, H–J) 20 μm; (C, E–G) 50 μm.

656 T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

lamin antibody, which detected two bands corresponding to lamin Aand C in all samples, proved equal loading of the three extracts.

More than the 96% of GATA4+ cardiac interstitial cells wereidentified as fibroblasts and pericytes, based on tissue distribution(Figs. 2A and B), expression of the fibroblast-specific marker prolil-4-hydroxilase (P4H) (Fig. 2C) [24], vimentin (Fig. 2D) and theproteoglycan NG2 (Fig. 2E), a marker of pericytes. GATA4 wasdetected in 97.1±0.3% cardiac fibroblasts (3233 cells counted) and in40.0±3.6% cardiac pericytes (2155 cells counted), but was notdetected in vascular endothelial (Fig. 2F) and smooth muscle cells(Fig. 2G). We did not find GATA4+ cells expressing c-kit (Fig. 2H) orSca-1 (Fig. 2I), possibly because c-kit+ cells are extremely rare in thenormal adult heart. A very minor fraction of GATA4+ cells (less than4% of total interstitial cells) expressed nestin (Fig. 2J), a type VIintermediate filament protein detected in adult neural and cardiacstem cells [25].

To test whether GATA4 is expressed in fibroblasts of other tissueswe analyzed normal and wounded skin (Figs. 3A and B) and normaland regenerating skeletal muscle (Figs. 3C and D). GATA4 was neverdetected in fibroblasts and pericytes of these tissues. Next, we askedwhether fibroblasts of related embryological origin expressed GATA4and for this purpose we analyzed gut fibroblasts, which are alsoderived from the splanchnopleuric mesoderm. Anti-GATA4 antibodylabeled epithelial cells of the stomach mucosa (Fig. 3E), but not sub-mucosal cells (Figs. 3F and F′), indicating that GATA4 detection infibroblasts is somehow a peculiar property of the heart.

To establish whether GATA4 is already detectable in cardiacfibroblasts during organ maturation or is specifically expressed inthe adult life, we investigated the early postnatal stages, which

correspond to the timing when fibroblasts accumulate in the organ(Fig. 4). In the 2, 5, 11-day old neonatal hearts numerous interstitialcells expressed GATA4 (Figs. 4A–C). Initially, fibroblasts wererelatively rare (Fig. 4D), whereas numerous GATA4+ endothelialand smooth muscle cells were observed (Fig. 4G). Subsequently, therewas a progressive increase in GATA4+ cardiac fibroblasts (Figs. 4E andF) and a decrease of GATA4+ endothelial and smooth muscle cells(Figs. 4H and I), which were no longer detectable in the 11-day old ratheart. Overall, the percentage of GATA4+ interstitial cells, calculatedat themid-ventricular level, was 25.2±7.9, 36.4±4.1 and 51.1±8.2 inthe 2, 5 and 11 day-old rat heart, respectively, this increasepresumably reflecting the rapid accumulation of interstitial cellspostnatally.

3.2. Cultured GATA4+ interstitial cells do not differentiate intocardiomyocytes in vitro

We hypothesized that interstitial cells of the heart may have thepotential to transdifferentiate into muscle cells under appropriatedifferentiation stimuli. Therefore we isolated adult cardiac interstitialcells from adult normal and GFP+ rat hearts by Langendorffprocedure and maintained them in culture for 4, 7 or 10 days. Platedinterstitial cells (Fig. 5A) showed a high proliferation rate with adoubling time of approximately 24 h (Fig. 5B), reaching confluenceafter 7 days in culture. RT-PCR analysis confirmed expression ofGATA4, but not Nkx2.5, in these cells, both immediately after isolation(T=0) and following 10-days in culture (T=10). Lack of contaminat-ing cardiomyocytes in the preparation was confirmed by RT-PCR foralpha-MyHC (Fig. 5C). Immunofluorescence analyses showed that

Fig. 3. Fibroblasts and pericytes of other tissues do not express GATA4. Cryosections ofnormal (A) and regenerating (B) skin, normal (C) and regenerating (D) skeletal musclewere stained with rabbit anti-GATA4 antibody, revealed by anti-rabbit peroxidase andcounterstained with hematoxilin. m, mucosa; sm, submucosa; mm, muscularis mucosa.(E–F′) Cryosections of adult rat stomach were processed for immunoperoxidase usingrabbit anti-GATA4 antibody (E) or double stained with rabbit anti-GATA4 antibody andanti-P4H antibody (F) and counterstained with DAPI. High magnification in (F′) showsGATA4−/P4H+ fibroblasts of the submucosa. Magnification in (F′) corresponds to theinset in (F). Bars: 50 μm.

657T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

most of these cells were fibroblasts, based on morphology andexpression of P4H (Fig. 5D), whereas cells reactive for CD31, SM-MyHCand NG2 accounted for less than 5% total cells. By immunofluores-cence GATA4 was detectable in ∼70% of interstitial cells at T=0 and∼80% at T=10 (Fig. 5E). Skin fibroblasts never expressed GATA4under the same culture conditions (Supplementary Fig. 1). Cardiacinterstitial cells never showed spontaneous cardiac differentiation andaccumulation of alpha sarcomeric MyHC and Nkx2.5 over the timeperiod analyzed (Supplementary Figs. 2A and B), consistent with PCRexperiments.

The differentiation potential of cardiac interstitial cells waschallenged using different culture conditions. We used both well-established differentiation procedures, such as 3 μM 5-azacytidine

[16] and dexamethasone [21] and different media for fetal and adultcardiomyocytes and in no case could we induce cardiac differentia-tion. Media for fetal and adult cardiomyocytes were tested in parallelon P19CL6 cell line. Fetal, not adult, media induced cardiacdifferentiation, as demonstrated by expression of sarcomeric actinin(Supplementary Fig. 3) and beating after 8 days in culture (Supple-mentary Movies A and B). Different coatings, which were usedextensively to induce cardiogenic differentiation of stem andprogenitor cells, such as laminin, poly-D-lysine and fibronectin failedto differentiate cardiac interstitial cells, as demonstrated by lack ofNkx2.5 and sarcomeric protein expression in all culture conditions(not shown).

Previous studies demonstrated that co-cultures with cardiomyo-cytes can force transdifferentiation of undifferentiated or multipotentcells. To test this hypothesis we co-cultured cardiomyocytes andcardiac interstitial cells isolated by Langendorff perfusion. Cells wereplated at the 1:100 ratio, which we identified as the optimal conditionto balance proliferation of fibroblasts and number of cardiomyocytesto follow in long-term cultures. Immediately after dissociation ∼ 60%cardiomyocytes were beating and rod-shaped (Fig. 5F). As soon asinterstitial cells attached to the plate, cardiomyocytes spread onto thisfeeder layer. After 7 and 10 days in culture we observed cells with thetypical shape and sarcomeric protein distribution of neonatalcardiomyocytes (Figs. 5G and H). When we co-cultured GFP+cardiomyocytes with GFP− cardiac interstitial cells, all cardiomyo-cytes with immature phenotype were, with no exception, GFP+, thustheywere not transdifferentiated interstitial cells, but dedifferentiatedadult cardiomyocytes (Fig. 5I).

3.3. Cultured GATA4+ interstitial cells promote dedifferentiation of adultcardiomyocytes and their re-entry into the cell cycle

Dedifferentiation of cardiomyocytes co-cultured with interstitialcells was investigated further. Desmin (Fig. 6A) and myotilin, themyofibrillar protein with titin-like Ig domains [26] (Fig. 6B), showeddisorganized distribution and punctate pattern in dedifferentiatedcells instead of the typical Z-disc association observed in normalcardiomyocytes. In dedifferentiated cells we also detected SMA, atypical marker of immature cardiomyocytes (Fig. 6C). To determinewhether interstitial cells derived from other tissues equally stimulateddedifferentiation of adult cardiomyocytes we used primary skinfibroblasts. After 10 days in culture 35.8±10.0% cardiomyocytes(6308 cardiomyocytes analyzed) spread and dedifferentiated whenplated with cardiac interstitial cells. This value dropped to 3.0±1.3%(5618 cardiomyocytes analyzed) in the presence of skin fibroblasts(Fig. 6D).

To test whether cell–cell contact or humoral factors released bycardiac fibroblasts promoted cardiomyocyte dedifferentiation, wecultured adult cardiomyocytes and skin fibroblasts with conditionedmedium derived either from cardiac interstitial cells alone (condi-tioned medium A), or cardiac interstitial cells with cardiomyocytes(conditioned medium B). Under these conditions, 4.8±1.7% and0.64±0.52% cardiomyocytes (5000 and 4360 cardiomyocytesanalyzed), underwent dedifferentiation, respectively (Fig. 6D),indicating that the presence of cardiac interstitial cells, not justfactors released in the medium, is required to promote dediffer-entiation. In our experiments and with the time course of analysischosen for this study dedifferentiation of cardiomyocytes culturedalone was less than 2%.

Next we aimed to establish whether dedifferentiation of cardio-myocytes promoted activation of the cell cycle. This questionwas askedbecause previous studies have shown that cell cycle re-entry can bemodulated in cultured adult cardiomyocytes by periostin, a proteinsynthesized and released by fibroblasts [27]. Proliferationmarkers Ki67and H3P and BrdU incorporation were tested. Ki67 reactivity wasdetected in a large number of dedifferentiated cardiomyocytes after 7

Fig. 4. Cardiac interstitial cells express GATA4 in the postnatal development. Cryosections of 2 (A, D, G), 5 (B, E, H) and 11 day-old (C, F, I) neonatal rat hearts were incubated withgoat or rabbit anti-GATA4 antibody and antibodies to: alpha-actinin (A–C), P4H (D–F) and SM-MyHC (G–I) and analyzed by confocal microscope. Arrows in (A–C) and (D–F)indicate GATA4+ interstitial cells and GATA4+ cardiac fibroblasts, respectively. Bars: (A–C) 20 μm; (D–I) 50 μm.

658 T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

and 10 days in culture, whereas H3P was rarely detected (Figs. 6E–G).BrdU incorporation was observed in 12.5±3.8% dedifferentiatedcardiomyocytes co-cultured with cardiac interstitial cells. This valuedropped to 2.1±1.0%, 0.9±1.0% and 1.9± 0.5% when cardiomyocyteswere plated in the presence of skin fibroblasts, skin fibroblasts withconditioned medium A and B, respectively (Fig. 6H). Overall, thesedata indicate that cardiac interstitial cells modulate cardiomyocytedifferentiation state.

4. Discussion

4.1. GATA4 expression in cardiac interstitial cells and its significance

This paper demonstrates for the first time that fibroblasts andpericytes of the rat heart express the transcription factor GATA4. In theearly postnatal heart GATA4 is detectable in cardiomyocytes and cellsof the vascular network and, very rarely, in the interstitium.Progressively after birth, endothelial and smooth muscle cells becomeGATA4−whereas a large amount of GATA4+ fibroblasts and pericytesaccumulate, concomitant with organ maturation. The presence ofGATA4 in cardiac fibroblasts is consistent with their embryologicalorigin from the epicardium, since epicardial cells also express GATA4in the developing and adult mouse [28,29]. On the other hand, gutfibroblasts, which are also derived from the splanchnopleuricmesoderm [30], are GATA4−, suggesting that the peculiar propertyof cardiac fibroblasts cannot be merely explained by their embry-ological origin. Why cardiac fibroblasts and pericytes express GATA4pre- and postnatally remains an open issue. Other transcription

factors, such as Isl1 and Nkx2.5 accumulate in the proepicardial cells[31], but they are no longer detectable after birth. It is likely thatGATA4 acts as an organ-specific marker that controls cardiac and non-cardiac cell fate in the developing heart. In the adulthood, GATA4 maybe re-employed for additional functions, such as protection fromapoptosis, control of chemical, mechanical and stress-signals. PartialGata4 gene targeting in cardiac and non-cardiac cells indicates thateven modest variations in Gata4 gene level or activity can have a rolein maintenance of normal cardiac function [32]. GATA4 is also a directtranscriptional activator of cyclin D2 and Cdk4 [33], a property thatmay control both fibroblast proliferation during organ maturation andthe fibrogenic response after tissue damage. In addition to anintracellular function, GATA4 is also implicated in intercellular cross-talk. GATA4 produced by cardiomyocytes induces hypertrophy-associated neoangiogenesis via VEGF-release and targeting to theendothelium [14]. Similarly, it is likely that GATA4 orchestrates othersignaling pathways, including signals from fibroblasts and pericytes tocardiomyocytes.

Expression of GATA4 in cardiac adult pericytes and in a populationof nestin+ cells deserves an additional comment. Postnatal pericytesof skeletal and cardiac muscle have been shown to display myogenicpotential [34,35]. Our experiments did not establish whether cardiacpericytes of the adult heart maintain this property, because cellsexpressing the pericyte marker NG2 were rare in the interstitial cellfraction isolated from the adult heart. Additionally, the finding thatonly 40% of cardiac adult pericytes express GATA4 indicates hetero-geneity of this cell population and will require cloning analysis toaddress the question of their myogenic potential in vivo. Regard co-

Fig. 5. Cardiac interstitial cells do not differentiate into cardiomyocytes in vitro. (A) Bright field of isolated cardiac interstitial cells and (B) proliferation curve. (C) RT-PCR of beta-actin(internal control), alpha-MyHC, Nkx2.5 and GATA4 of control tissues and cardiac interstitial cells immediately after dissociation (T=0) and after 10 days in culture (T=10). (D, E)Cardiac interstitial cells were incubated with anti-P4H (D) and rabbit anti-GATA4 (E) antibody and counterstained with DAPI. c. int. cells, cardiac interstitial cells. (F, G) Bright field ofisolated adult cardiomyocytes immediately after dissociation (F) or co-cultured with cardiac interstitial cells for 10 days (G). The arrow in (G) indicates a differentiatedcardiomyocyte. (H) Ten-day co-cultures were stained with anti-sarcomeric actinin and counterstained with DAPI to detect striated cells. (I) Co-cultures of GFP+ cardiomyocytes andGFP− cardiac interstitial cells were stained after 10 days with anti-sarcomeric actinin antibody (left) and counterstained with DAPI. Reacting cells were all GFP+ (right). Bars: (A, D,E, G) 50 μm; (F, H, I) 20 μm.

659T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

expression of GATA4 and nestin in some interstitial cells, it is likelythat they represent a stem cell population. According to thishypothesis and consistent with other reports [36], we observed anincreased number of these cells in the damaged heart (ourunpublished observation).

The identification of cardiac markers in cardiac fibroblasts andpericytes has a potential impact in the light of the recent discoverythat adult fibroblasts can be reprogrammed to a pluripotent state andthereof to a novel differentiation program by the introduction ofOct4, Sox2, Klf4 and c-Myc transcription factors [37,38]. The epige-netic mechanisms of such a process are still under investigation andmany questions related to the potential of reprogrammed cells fortissue regeneration need to be addressed. For example, it is unlikelythat reprogramming is equivalent in all cell types. Organ-specificfibroblasts may have a “leg up” in being more susceptible toreprogramming and also display a different kinetic when inducedto differentiate into a specialized cell phenotype. The evidence thatcardiac fibroblasts differ from other fibroblasts with respect to therepertoire of genes of cardiac differentiation opens to the investiga-tions of their potential towards the myogenic lineage. Although theexperiments reported here indicate that manipulation of the cultureconditions are not sufficient to induce transdifferentiation of fibro-

blasts to cardiomyocytes, genetic manipulations may be morerewarding in this respect.

4.2. Control of adult cardiomyocyte dedifferentiation and cell cyclere-entry by GATA4+ cardiac interstitial cells

A second major finding of this study is that cardiac fibroblastspromote dedifferentiation of adult cardiomyocytes in co-cultureexperiments. In agreement with previous reports [39,40,41] weshow that cell–cell contact with cardiac fibroblasts is crucial forcardiomyocyte dedifferentiation in vitro and cannot be reproduced byhumoral factors. In contrast with previous reports [41], however, weshow that both survival and dedifferentiation of adult cardiomyocytesare significantly higher in the presence of cardiac fibroblasts,compared to skin fibroblasts, indicating that the feeder layer doesnot simply provide a physical contact for cardiomyocyte anchorageand stretching. One possible explanation of this discrepancy is thatour culture conditions, in which we plated cardiomyocytes at a lowdensity, probably reduced progression speed and rate of dediffer-entiation, thereby allowing visualization of differences that could notbe observed when cardiomyocytes and fibroblasts are plated at a 1:1ratio.

Fig. 6. Cardiac interstitial cells promote adult cardiomyocyte dedifferentiation and cell cycle re-entry. (A–C) Double staining with anti-sarcomeric actinin (A, B, left) and either anti-desmin (A, middle) or anti-myotilin (B, middle) antibody showed disorganized desmin or myotilin distribution in dedifferentiated cardiomyocytes, but normal pattern indifferentiated cardiomyocytes (A, B, right). (C) Double staining with anti-sarcomeric actinin (left) and anti-SMA showed re-expression of SMA in dedifferentiated (middle), but notdifferentiated (right) cardiomyocytes. (D) Quantification of dedifferentiated cardiomyocytes co-cultured with skin fibroblasts, cardiac interstitial cells, and skin fibroblasts in thepresence of conditioned medium A or conditioned medium B. Card., cardiomyocytes; skin f, skin fibroblasts, c. int. cells, cardiac interstitial cells. Bars: 20 μm. (E–G) Adultcardiomyocytes were co-cultured with cardiac interstitial cells and double stained with: anti-Ki67 (E), anti-H3P (F) anti-BrdU (G) and either anti-alpha actinin (E, F) or anti alpha-MyHC (G). Merged images are shown. (H) BrdU+ cardiomyocytes over the number of dedifferentiated cardiomyocytes were quantified in co-cultures with skin fibroblasts, cardiacinterstitial cells and skin fibroblasts in the presence of conditioned medium A or conditioned medium B. Card., cardiomyocytes; skin f, skin fibroblasts, c. int. cells, cardiac interstitialcells. Bars: 20 μm.

660 T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

Cardiomyocyte dedifferentiation occurs in vivo in different patho-logical conditions, including chronic hibernation [42], chronic atrialfibrillation [43] and ischemia [44] and it has been shown thatdedifferentiated cardiomyocytes better tolerate acute ischemia insults.

It remains to be established whether fibroblasts are involved indedifferentiation under these conditions andwhat is the role of GATA4in this cross-talk. Short range signals, such as hormones and growthfactors need to be investigated. A number of pathological stimuli

661T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

increase GATA4 phosphorylation and activity in cardiomyocytes, andpossibly in cardiac interstitial cells. To define the role of GATA4 incardiac fibroblasts it will be essential to generate transgenic mousemodels in which GATA4 is selectively over-expressed or inactivated inthese cells.

The most interesting aspect of cardiomyocyte dedifferentiationinduced by cardiac fibroblasts is its association with cell cycle re-entry.Inour study this is documented byexpressionof specificmarkers of DNAsynthesis, such as H3P, and by BrdU incorporation. A similar effect waspreviously described in periostin-induced cardiomyocyte dedifferentia-tion where, in addition, completion of cell cycle was proved [27]. Thisand other findings indicate that cell cycle can be reprogrammed in adultcardiomyocytes and may be considered an additional option forregeneration. In the newt heart, efficient cardiac regeneration occursthrough dedifferentiation of pre-existing cardiomyocytes and subse-quent proliferation [45]. In mammalian neonatal cardiomyocytes, celldivision is accompanied by sequential myofibrillar disassembly andreassembly [46]. However, this capacity is lost at subsequent develop-mental stages presumably due to both down-regulation of cell cycleproteins and acquisition of a compact myofibrillar cytoarchitecture. Atthe moment it is not clear whether pathological conditions that inducecardiomyocyte dedifferentiation also promote cell cycle re-entry. In theborder zone of the infarcted myocardium only a very minor fraction ofcardiomyocytes exhibits cell cycle re-entry [47], but blockade of p53activity enables border zone cardiomyocytes, not cardiomyocytes fromother regions, to re-enter and conclude the cell cycle [48], suggestingthat these cells become more permissive to cell division. We ourselvesobserved a high amount of dedifferentiated cardiomyocytes in rat heartsheterotopically transplanted,mostofwhichexpressed cell cyclemarkers(our unpublished observations). Having established a simple co-culturesystem that combines dedifferentiation and cell cycle re-entry of adultcardiomyocytes will allow to overcome some limitations of in vivostudies and to investigate treatments and signaling molecules topromote cardiac cell division.

In conclusion, we have demonstrated that fibroblasts, pericytesand a putative stem cell population of the heart express GATA4, afinding that opens to the possibility, yet unexplored, that this proteinis also involved in molecular interactions between myogenic andnon-myogenic cells of the heart. The observation made in this studygives us a valuable starting point to unravel the biological function ofGATA4 in cardiac fibroblasts and to investigate how cardiacfibroblasts influence differentiation stage of cardiomyocytes andcontrol of cell division.

Fundings

This work was supported by the European Union (HeartRepairLSHM-CT-2005-018630 to S.S.), the University of Padova (Progetto diRicerca di Ateneo 2005 CPDA054298 to S.A.) and the Consorzio per laRicerca sul Trapianto d'Organo (CORIT). Tania Zaglia is a post-docfellow of the European Union (HeartRepair).

Conflict of interestThe authors declare no conflict of interest.

Acknowledgments

We are grateful to Fabio Di Lisa and Roberta Menabò for help withLangendorff perfusion, Olli Carpen and Lars-Eric Thornell for the anti-myotilin antibody and Giulio Bassi for help with some immunofluo-rescence experiments.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.yjmcc.2008.12.010.

References

[1] Camelliti P, Borg TK, Kohl P. Structural and functional characterisation of cardiacfibroblasts. Cardiovasc Res 2005;65:40–51.

[2] LaFramboise WA, Scalise D, Stoodley P, Graner SR, Guthrie RD, Magovern JA, et al.Cardiac fibroblasts influence cardiomyocyte phenotype in vitro. Am J Physiol CellPhysiol 2007;292:C1799–808.

[3] Manabe I, Shindo T, Nagai R. Gene expression in fibroblasts and fibrosis:involvement in cardiac hypertrophy. Circ Res 2002;91:1103–13.

[4] Norris RA, Borg TK, Butcher JT, Baudino TA, Banerjee I, Markwald RR. Neonatal andadult cardiovascular pathophysiological remodeling and repair: developmentalrole of periostin. Ann N Y Acad Sci 2008;1123:30–40.

[5] Haudek SB, Xia Y, Huebener P, Lee JM, Carlson S, Crawford JR, et al. Bone marrow-derived fibroblast precursors mediate ischemic cardiomyopathy in mice. Proc NatlAcad Sci U S A 2006;103:18284–9.

[6] Haudek SB, Trial J, Xia Y, Gupta D, Pilling D, Entman ML. Fc receptor engagementmediates differentiation of cardiac fibroblast precursor cells. Proc Natl Acad SciU S A 2008;105(29):10179–84.

[7] Pu WT, Ishiwata T, Juraszek AL, Ma Q, Izumo S. GATA4 is a dosage-sensitiveregulator of cardiac morphogenesis. Dev Biol 2004;275:235–44.

[8] Charron F, Nemer M. GATA transcription factors and cardiac development. SeminCell Dev Biol 1999;10:85–91.

[9] Molkentin JD. The zinc finger-containing transcription factors GATA-4, -5, and -6.Ubiquitously expressed regulators of tissue-specific gene expression. J Biol Chem2000;275:38949–52.

[10] Hasegawa K, Lee SJ, Jobe SM, Markham BE, Kitsis RN. cis-Acting sequences thatmediate induction of beta-MyHCgene expression during left ventricular hyper-trophy due to aortic constriction. Circulation 1997;96:3943–53.

[11] Liang Q, De Windt LJ, Witt SA, Kimball TR, Markham BE, Molkentin JD. Thetranscription factors GATA4 and GATA6 regulate cardiomyocyte hypertrophy invitro and in vivo. J Biol Chem 2001;276:30243–5.

[12] Aries A, Paradis P, Lefebvre C, Schwartz RJ, Nemer M. Essential role of GATA4 in cellsurvival and drug-induced cardiotoxicity. Proc Natl Acad Sci U S A2004;101:6975–80.

[13] Oka T, Maillet M, Watt AJ, Schwartz RJ, Aronow BJ, Duncan SA, et al. Cardiac-specific deletion of Gata4 reveals its requirement for hypertrophy, compensation,and myocyte viability. Circ Res 2006;98:837–45.

[14] Heineke J, Auger-Messier M, Xu J, Oka T, Sargent MA, York A, et al. CardiomyocyteGATA4 functions as a stress-responsive regulator of angiogenesis in the murineheart. J Clin Invest 2007;117:3198–210.

[15] Beltrami AP, Barlucchi L, Torella D, Baker M, Limana F, Cimenti S, Kasahara H, RotaM, Musso E, Urbanek K, Leri A, Kjstura J, Nadal-Ginard B, Anversa P. Adult cardiacstem cells are multipotent and support myocardial regeneration. Cell2003;114:763–76.

[16] Oh H, Bradfute SB, GallardoTD, Nakamura T, Gaussin V, Mishina Y, Pocius J, MichaelLH, Behringer RR, Schwartz RJ, EntmanML, Schneider MD. Cardiac progenitor cellsfrom adult myocardium: homing, differentiation, and fusion after infarction. ProcNatl Acad Sci U S A 2003;100:12313–8.

[17] Urbanek K, Cesselli D, RotaM, Nascimbene A, De Angelis A, Hosoda T, Bearzi C, BoniA, Bolli R, Kajstura J, Anversa P, Leri A. Stem cell niches in the adult mouse heart.Proc Natl Acad Sci U S A 2006;103:9226–31.

[18] Bearzi C, Rota M, Hosoda T, Tillmanns J, Nascimbene A, De Angelis A, Yasuzawa-Amano S, Trofimova I, Siggins RW, Lecapitaine N, Cascapera S, Beltrami AP,D'Alessandro DA, Zias E, Quaini F, Urbanek K, Michler RE, Bolli R, Kajstura J, Leri A,Anversa P. Human cardiac stem cells. Proc Natl Acad Sci U S A 2007 Aug 28;104(35):14068–73.

[19] Okabe M, Ikawa M, Kominami K, Nakanishi T, Nishimune Y. ‘Green mice’ as asource of ubiquitous green cells. FEBS Lett 1997;407:313–9.

[20] Pallafacchina G, Calabria E, Serrano AL, Kalhovde JM, Schiaffino S. A proteinkinase B-dependent and rapamycin-sensitive pathway controls skeletal musclegrowth but not fiber type specification. Proc Natl Acad Sci U S A 2002;99:9213–8.

[21] Tateishi K, Ashihara E, Honsho S, Takehara N, Nomura T, Takahashi T, Ueyama T,Yamagishi M, Yaku H, Matsubara H, Oh H. Human cardiac stem cells exhibitmesenchymal features and are maintained through Akt/GSK-3β signalling.Biochem Biophys Res Commun 2007;352:635–41.

[22] Ausoni S, Zaglia T, Dedja A, Di Lisi R, Seveso M, Ancona E, et al. Host-derivedcirculating cells do not significantly contribute to cardiac regeneration inheterotopic rat heart transplants. Cardiovasc Res 2005;68:394–404.

[23] Dedja A, Zaglia T, Dall'Olmo L, Chioato T, Thiene G, Fabris L, et al. Hybridcardiomyocytes derived by cell fusion in heterotopic cardiac xenografts. FASEB J2006;20:2534–6.

[24] HarmsW, Rothamel T, Miller K, Harste G, GrassmannM, Heim A. Chracterization ofhuman myocardial fibroblasts immortalized by HPV16 E6–E7 genes. Exp Cell Res2001;268:252–61.

[25] Quaini F, Urbanek K, Beltrami AP, Finato N, Beltrami CA, Nadal-Ginard B, et al.Chimerism of the transplanted heart. N Engl J Med 2002;346:5–15.

[26] Mologni L, Salmikangs P, Fougerousse F, Beckmann JS, Carpen O. Developmentalexpression of myotilin, a genemutated in limb-girdle muscular dystrophy type 1A.Mech Dev 2001;103:121–5.

[27] Kühn B, del Monte F, Hajjar RJ, Chang YS, Lebeche D, Arab S, et al. Periostin inducesproliferation of differentiated cardiomyocytes and promotes cardiac repair. NatMed 2007;13:962–9.

[28] Rivera-Feliciano J, Lee KH, Kong SW, Rajagopal S, Ma Q, Springer Z, et al.Development of heart valves requires Gata4 expression in endothelial-derivedcells. Development 2006;133:3607–18.

662 T. Zaglia et al. / Journal of Molecular and Cellular Cardiology 46 (2009) 653–662

[29] Watt AJ, Battle MA, Li J, Duncan SA. GATA4 is essential for formation of theproepicardium and regulates cardiogenesis. Proc Natl Acad Sci U S A 2004;101:12573–8.

[30] Roberts DJ, Smith DM, Goff DJ, Tabin CJ. Epithelial–mesenchymal signalingduring the regionalization of the chick gut. Development 1998;125:2791–801.

[31] Zhou B, von Gise A, Ma Q, Rivera-Feliciano J, Pu WT. Nkx2–5- and Isl1-expressingcardiac progenitors contribute to proepicardium. Biochem Biophys Res Commun2008;375:450–3.

[32] Bisping E, Ikeda S, Kong SW, Tarnavski O, Bodyak N, McMullen JR, et al. Procl NatlAcad Sci U S A 2006;103:14471–6.

[33] Rojas A, Kong SW, Agarwal P, Gilliss B, Pu WT, Black BL. GATA4 is a directtranscriptional activator of cyclin D2 and Cdk4 and is required for cardiomyocyteproliferation in anterior heart field-derived myocardium. Mol Cell Biol 2008;28:5420–31.

[34] Dellavalle A, Sampaolesi M, Tonlorenzi R, Tagliafico E, Sacchetti B, Perani L, et al.Pericytes of human skeletal muscle are myogenic precursors distinct from satellitecells. Nat Cell Biol 2007;9:255–67.

[35] Galvez BG, Sampaolesi M, Barbuti A, Crespi A, Covarello D, Brunelli S, et al.Cardiac mesoangioblasts are committed, self-renewable progenitors, associatedwith small vessels of juvenile mouse ventricle. Cell Death Differ 2008;15(9):1417–28.

[36] El-Helou V, Beguin PC, Assimakopoulos J, Clement R, Gosselin H, Brugada R, et al.The rat heart contains a neural stem cell population; Role in sympatheticsprouting and angiogenesis. J Mol Cell Cardiol 2008.

[37] Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouseembryonic and adult fibroblast cultures by defined factors. Cell 2006;126:663–76.

[38] Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, Tomoda K, et al. Inductionof pluripotent stem cells from adult human fibroblasts by defined factors. Cell2007;131:861–72.

[39] Dispersyn GD, Geuens E, Ver Donck L, Ramaekers FC, Borgers M. Adult rabbitcardiomyocytes undergo hibernation-like dedifferentiation when co-culturedwith cardiac fibroblasts. Cardiovasc Res 2001;51:230–40.

[40] Rücker-Martin C, Pecker F, Godreau D, Hatem SN. Dedifferentiation of atrialmyocytes during atrial fibrillation: role of fibroblast proliferation in vitro.Cardiovasc Res 2002;55:38–52.

[41] Driesen RB, Verheyen FK, Dispersyn GD, Thoné F, Lenders MH, Ramaekers FC, et al.Structural adaptation in adult rabbit ventricular myocytes: influence of dynamicphysical interaction with fibroblasts. Cell Biochem Biophys 2006;44:119–28.

[42] Heusch G, Schulz R. The biology of myocardial hibernation. Trends Cardiovasc Med2000;10:108–14.

[43] Ausma J, Borgers M. Dedifferentiation of atrial cardiomyocytes: from in vivo to invitro. Cardiovasc Res 2002;55:9–12.

[44] Driesen RB, Verheyen FK, Dijkstra P, Thoné F, Cleutjens JP, Lenders MH, et al.Structural remodelling of cardiomyocytes in the border zone of infarcted rabbitheart. Mol Cell Biochem 2007;302:225–32.

[45] Bettencourt-Dias M, Mittnacht S, Brockes JP. Heterogeneous proliferative potentialin regenerative adult newt cardiomyocytes. J Cell Sci 2003;116:4001–9.

[46] Ahuja P, Perriard E, Perriard JC, Ehler E. Sequential myofibrillar breakdownaccompanies mitotic division of mammalian cardiomyocytes. J Cell Sci 2004;117:3295–306.

[47] Pasumarthi KB, Field LJ. Cardiomyocyte cell cycle regulation. Circ Res 2002;90:1044–54.[48] Nakajima H, Nakajima HO, Tsai SC, Field LJ. Expression of mutant p193 and p53

permits cardiomyocyte cell cycle reentry after myocardial infarction in transgenicmice. Circ Res 2004;94:1606–14.