Cellular and genetic approaches to myocardial regeneration Tuyn, J. van

(1)

Cellular and genetic approaches to myocardial regeneration

Tuyn, J. van

Citation

Tuyn, J. van. (2008, January 9). Cellular and genetic approaches to myocardial

regeneration. Department of Cardiology and Department of Molecular Cell Biology (MCB), Faculty of Medicine, Leiden University Medical Center (LUMC), Leiden University.

Retrieved from https://hdl.handle.net/1887/12548

Version: Corrected Publisher’s Version

License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden

Downloaded from: https://hdl.handle.net/1887/12548

Note: To cite this publication please use the final published version (if applicable).

(2)

Chapter

4

John van Tuyn Daniël A. Pijnappels Antoine A.F. de Vries Ingrid de Vries Ietje van der Velde-van Dijke Shoshan Knaän-Shanzer Arnoud van der Laarse Martin J. Schalij Douwe E. Atsma

the FASEB Journal 2007;21:3369-3379

Fibroblasts from human post-myocardial

infarction scars acquire properties of

cardiomyocytes after transduction with a

recombinant myocardin gene

(3)

82 Abstract

Myocardial scar formation impairs heart function by inducing cardiac remodeling, decreasing myocardial compliance, and compromising normal electrical conduction. Conversion of ventricular scar fibroblasts (VSFs) into (functional) cardiomyocytes may be an effective treatment alternative to limit loss of cardiac performance following myocardial injury. In this study, we investigated whether the phenotype of VSFs can be modified by gene transfer into cells with properties of cardiomyocytes. To this end, fibroblasts from post-myocardial infarction scars of human left ventricles were isolated and characterized by cell biological, immunological, and molecular biological assays. Cultured human VSFs express GATA4 and connexin 43 and display adipogenic differentiation potential. Infection of human VSFs with a lentivirus vector encoding the potent cardiogenic transcription factor myocardin renders them positive for a wide variety of cardiomyocyte-specific proteins including sarcomeric components, transcription factors, and ion channels, and induces the expression of several smooth muscle marker genes. Forced myocardin expression also endowed human VSFs with the ability to transmit an action potential and to repair an artificially created conduction block in cardiomyocyte cultures. These findings indicate that in vivo myocardin gene transfer may potentially limit cardiomyocyte loss, myocardial fibrosis, and disturbances in electrical conduction caused by myocardial infarction.

(4)

83

Chapter 4Fibroblasts from human post-myocardial infarction scars acquire properties of cardiomyocytes after transduction with a recombinant myocardin gene

Introduction

After myocardial infarction (MI), the cardiomyocytes in the infarcted myocardium are replaced by fibrotic scar tissue composed of (myo)fibroblasts and a collagen- rich extracellular matrix¹ (ECM). These cells persist and remain metabolically active even in areas of long-standing fibrosis². Myocardial scars contribute to deleterious remodeling and increased mechanical stiffness of the heart resulting in impaired systolic and diastolic function which may lead to symptoms of heart failure. In addition, excessive deposition of ECM components reduces electrical coupling between cardiomyocytes thereby impairing impulse propagation³. Currently, stem cell transplantation is explored as a new therapeutic modality to regenerate damaged myocardium⁴. Thus far, most animal and clinical studies have demonstrated only a modest and sometimes even transient improvement of cardiac function after infusion of somatic stem cells in infarcted myocardium^5,

6. In addition, true myocardial regeneration via the in vivo (trans)differentiation of somatic stem cells into cardiomyocytes seems to be at best a rare event⁷. Therefore, alternative treatment strategies targeting cardiac fibroblasts have been proposed, including (i) inhibition of ECM synthesis by these cells, (ii) their conversion into (cardio)myocytes, and (iii) the use of these cells as a local source of cardioregenerative and/or angiogenic factors⁸. Introduction of a gene encoding a developmental regulator of (cardio)myogenesis in myocardial scar fibroblasts (MSFs) may induce their transdifferentiation into (cardiac) muscle cells. Illustrating the feasibility of this concept, in vitro and in vivo experiments have shown that transfer of the gene coding for the skeletal muscle-specific transcription factor (TF) MYOD1 leads to skeletal muscle differentiation in a variety of non-muscle cell types including cardiac fibroblasts^9-11. Given the differences in ion channel composition between skeletal muscle cells and cardiomyocytes¹² it is, however, unlikely that MYOD1-transduced fibroblasts can produce a cardiac action potential. It would therefore be preferable to use a gene encoding a TF involved in cardiomyogenesis to induce heart muscle differentiation in MSFs and to endow them with the desired electrical properties.

Several TFs including Nkx2.5, GATA-binding protein family members (e.g.

GATA4), T box-containing protein family members (e.g. Tbx5), the heart- and neural crest derivatives-expressed (Hand) 1 and 2 proteins, polypeptide C of the MADS box transcription enhancer factor 2 (Mef2) group of proteins (e.g.

Mef2C), and serum response factor (SRF) have been found essential for proper heart development during embryogenesis (for reviews see¹³ and¹⁴). Each of these cardiac TFs is directly involved in the transcriptional regulation of genes encoding the specialized structural components and channel proteins of cardiomyocytes, by binding to well-established DNA sequences in the promoter/enhancer regions of these genes. Recently, muscle development in the roundworm Caenorhabditis elegans was found to be regulated by three functionally redundant genes, i.e. the MyoD family homologue hlh-1, the SRF/Mef2 homologue unc-120 and the HAND homologue hnd-1¹⁵. In vertebrates the corresponding genes have adopted more speciﬁc roles in muscle development with the MyoD family members regulating skeletal muscle formation while the Hand genes are involved in both smooth and cardiac muscle development and SRF and the Mef2 family members play an

(5)

84

important role in the specification of all three muscle lineages^13-15. However, forced expression of the genes encoding the aforementioned TFs in somatic mammalian cells of non-muscle origin has not yet been reported to result in the formation of cardiomyocytes. Recently, a new cardiac and smooth muscle-restricted TF designated myocardin has been identified¹⁶. This protein, which exerts most of its effects by forming a complex with serum response factor (SRF) and Mef2c, appears to be a key regulator of cardiac and smooth muscle differentiation^16-18. Infection with a myocardin-encoding adenovirus vector caused ±10% of human dermal fibroblasts (hDFs) and human mesenchymal stem cells (hMSCs) and ±100% of human epicardial cells to adopt a cardiomyocyte-like phenotype through activation of genes specifying a wide variety of cardiomyocyte-specific polypeptides including sarcomeric components and heart muscle-specific TFs^{19, 20}.

In this study, we used reporter gene assays to compare the ability of the cardiogenic TFs (cTFs) GATA-binding protein 4 (GATA4), heart- and neural crest derivatives-expressed 1 (Hand1, also known as eHand), MADS box transcription enhancer factor 2C (Mef2C), myocardin (both cardiomyocyte-enriched isoforms), Nkx2.5, and Tbx5 to transactivate the promoters of the human natriuretic peptide precursor A (ANF), myosin heavy chain 6 (

α

MHC), and gap junction protein

α

¹

(Cx43) genes. As myocardin outperformed all other cTFs as a transcriptional activator, we generated a self-inactivating (SIN) lentivirus vector (LV.CMV.myocL- HA) directing the synthesis of the largest isoform of human myocardin (myocL).

This lentivirus vector was subsequently used to impose a cardiomyocyte-like gene expression program on human myocardial scar fibroblats isolated from the left ventricle (hVSFs) as evidenced by reverse transcription-polymerase chain reaction (RT-PCR) analyses and immunofluorescence microscopy (IFM). Furthermore, electrophysiological studies using multi-electrode array (MEA)-containing culture dishes revealed that hVSFs transduced with LV.CMV.myocL-HA were highly effective at repairing an artificially created conduction block in monolayers of neonatal rat cardiomyocytes (rCMs).

Materials and Methods

Isolation and culture of human myocardial scar ﬁbroblasts

hVSFs were isolated from multiple samples of human ventricular scar tissue obtained during surgical reconstruction of the left cardiac ventricle. Each sample was cut into small pieces, which were transferred to porcine gelatin (Sigma- Aldrich)-coated 10-cm² culture dishes, and cultured in Dulbecco’s modiﬁed Eagle’s medium (DMEM) containing 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% fetal bovine serum (FBS; all from Invitrogen). A glass coverslip was placed on top of the tissue pieces to prevent them from ﬂoating. The culture medium was refreshed every 3 days. Outgrowth of cells was visible 2 days after culture initiation. Three days later the coverslip and tissue pieces were removed, the hVSFs were detached with trypsin-EDTA solution (Invitrogen), and reseeded in new culture dishes. When the cultures reached 3-6 × 10⁶cells, aliquots of 2 × 10⁵ hVSFs were frozen to -80°C at -1°C per min in DMEM containing 40% FBS

(6)

85

and 10% dimethylsulfoxide (Sigma-Aldrich) and subsequently stored in nitrogen vapor.

Cell culture and in vitro differentiation assays

HeLa cells were propagated in DMEM supplemented with 10% FBS. hDFs were cultured as described before¹⁹ and the isolation and culture of rCMs has been speciﬁed elsewhere²¹. The induction and analysis of in vitro adipogenesis and osteogenesis has been carried out as previously described¹⁹.

Immunophenotyping

The surface antigen expression proﬁle of hVSFs was determined by ﬂow cytometry as described previously for bone marrow-derived hMSCs¹⁹.

Promoter assays

For comparing the transactivating activity of different cTFs, we generated the luciferase-encoding reporter constructs pANF.LUC, pMHC.LUC, and pCX43.LUC containing 2241-, 5505-, and 3009-bp fragments of the human ANF,

α

^MHC,

and Cx43 promoters, respectively. The nucleotide sequences of these pGL3- basic (Promega) derivatives have been deposited in the GenBank database under accession numbers EF186080, EF186081, and EF186082.

In the expression plasmids pU.CAG.hrGFP, pU.CAG.myocL, pU.CAG.myocS, pU.CAG.Hand1, pU.CAG.Nkx2.5, pU.CAG.GATA4, pU.CAG.Mef2C, and pU.CAG.

Tbx5, the coding sequences of the Renilla reniformis green ﬂuorescent protein (Stratagene), myocL, the small cardiomyocyte-enriched isoform of human myocardin (myocS), and the human versions of Hand1, Nkx2.5, GATA4, Mef2C, and Tbx5 are placed under the transcriptional control of the CAG promoter²² and rabbit β globin gene polyadenylation signal. For the nucleotide sequences of these constructs see GenBank accession numbers EF186083, EF186084, EF208955, EF208956, EF186085, EF186086, EF186087, and EF186088, respectively.

One day before transfection, 3 × 10⁴HeLa cells were seeded per well of a 24- well plate (Greiner). The next day, the cells were transfected using ExGen 500 (Fermentas) with 0.2 μg pMHC.LUC DNA or the molar equivalent of pANF.LUC or pCX43.LUC supplemented with pUCBM21 (Boehringer Mannheim) to a total of 0.2 μg DNA and 0.2 μg pU.CAG.myocL DNA or the molar equivalent of pU.CAG.

myocS, pU.CAG.Hand1, pU.CAG.Nkx2.5, pU.CAG.GATA4, pU.CAG.Mef2C, pU.CAG.

Tbx5, or pU.CAG.hrGFP again supplemented with pUCBM21 to a total of 0.2 μg DNA. Sixteen hours later, the transfection medium was replaced by fresh culture medium. At 48 h post-transfection, the luciferase activity in each cell sample was measured using Steady-Glo (Promega) and a Lumat LB 9507 luminometer (EG &

G Berthold). The induction factors of the various TFs were calculated for all 3 heart muscle-speciﬁc promoters by dividing the luciferase activity observed after co- transfection of 1 of the 3 reporter constructs and a TF-encoding plasmid by that found after co-transfection of pU.CAG.hrGFP and the same reporter construct.

Each pU.CAG plasmid was tested in quadruplo in 3 independent experiments to

(7)

86

compensate for possible differences in transfection efﬁciency.

Lentivirus vector production

To generate SIN human immunodeficiency virus type I vectors encoding a recombinant myocL protein carrying an epitope of the human influenza virus A hemagglutinin (HA) at its C terminus (myocL-HA) or specifiying a nuclear-targeted version of Escherichia coli β-galactosidase (nls-bGal), the shuttle plasmids pLV.CMV.

myocL-HA and pLV.C-PGK.nls-bGal were made on the basis of construct pLV-CMV- IRES-eGFP²³. The expression of myocL-HA from pLV.CMV.myocL-HA is controlled by the human cytomegalovirus immediate-early gene (CMV-IE) promoter while pLV.C-PGK.nls-bGal contains a hybrid promoter consisting of the enhancer of the CMV-IE fused to the human 3’-phosphoglycerate kinase gene promoter to drive transgene expression. Furthermore, to enhance transgene expression in both lentivirus vector shuttle plasmids the 3’ long terminal repeat is preceded by the woodchuck hepatitis virus posttranscriptional regulatory element²⁴. The nucleotide sequences of pLV.CMV.myocL-HA and pLV.C-PGK.nls-bGal have been deposited in the GenBank database under accession numbers EF186078 and EF186079, respectively. The lentivirus vector shuttle plasmid pLV.CMV.myocL.IRES.eGFP and pLV.CMV.bGal.IRES.eGFP were also derived from pLV-CMV-IRES-eGFP and specify bicistronic mRNAs encoding the enhanced green ﬂuorescent protein and myocL or bGal, respectively. For the nucleotide sequences of these constructs see GenBank accession numbers EF205035 and EF205034.

Vesicular stomatitis virus G protein-pseudotyped lentivirus vector stocks were produced by seeding six 175-cm² culture ﬂasks with 6.6 × 10⁴ 293T per cm² and transfecting these producer cells the next day with 83 ng/cm² of either pLV.CMV.

myocL-HA and pLV.C-PGK.nls-bGal together with 76 ng/cm² psPAX2 (Addgene) and 41 ng/cm² pLP/VSVG (Invitrogen) using the calcium phosphate-DNA co- precipitation method²⁵. Sixteen hours later, the transfection medium was replaced by new culture medium. At 48 h post-transfection, the culture fluid was collected and freed of cellular debris by centrifugation at room temperature for 10 min at 800 × g and filtration through a 0.45-μm-pore-size cellulose acetate filter (PALL Corporation). To concentrate the lentivirus vector particles, 31 ml of the cleared culture medium was loaded on a 4-ml cushion of 20% sucrose in phosphate- buffered saline (PBS) and centrifuged for 90 min at 25,000 rpm and 10°C in an SW 28 rotor (Beckman Coulter). Next, the supernatant was discarded and the pellets with the vector particles were suspended in 200 μl PBS-10% FBS by gentle rocking overnight at 4°C.

An aliquot of both vector stocks was serially diluted in culture medium containing 8 μg/ml polybrene and used to infect HeLa cells. Four hours later, the inoculum was exchanged with fresh culture medium. At 2 days post-infection, the functional titer (in terms of Hela cell-transducing units [HTUs]) of the LV.C-PGK.

nls-bGal preparation was determined by 5-bromo-4-chloro-3-indolyl-beta-D- galactopyranoside staining of formaldehyde-ﬁxed cells and that of the LV.CMV.

myocL-HA stock by immunostaining of ethanol-ﬁxed cells with a 1:1000 dilution of monoclonal antibody (mAb) HA.11 (Covance) directed against the epitope tag.

(8)

87

Lentivirus vector transduction

hVSFs were seeded at a density of 2 × 10⁴ cells per cm² in 10-cm² culture dishes with (for IFM) or without (for RT-PCR analysis) glass coverslips. The next day, the culture ﬂuid was replaced by 1 ml fresh growth medium containing 20 μg/ml diethylaminoethyl-dextran sulfate (GE Healthcare) and 1 HTU of LV.CMV.myocL- HA or LV.C-PGK.nls-bGal per cell. Four hours later, the cells were washed 3 times with PBS and the inoculum was replaced by fresh culture medium. One week after transduction, the cells were analyzed by IFM or RT-PCR analysis.

Reverse transcription-polymerase chain reaction analysis

Semi-quantitative RT-PCR analyses were performed as previously reported²⁰ with the addition of primer pairs targeting transcripts encoding the

α

^{1C and}

α

^1D

subunits of the voltage-dependent L-type calcium channels (CACNA1C or Cav1.2; 5’- AGGAGCAGTTTTTGGGGTTT-3’ and 5’-TGGAGCTGACTGTGGAGATG-3’ and CACNA1D or Cav1.3; 5’-GCAAGATGACGAGCCTGAG-3’ and 5’-ATGGTTATGATGGTTATGACAC- 3’, respectively), the cardiac pacemaker channel (hyperpolarization-activated cyclic nucleotide-gated potassium channel 4; HCN4; 5’-CGCCTCATTCGATATATTCAC- 3’ and 5’-CGCGTAGGAGTACTGCTTC-3’), and the inwardly rectifying potassium channel J3 (KCNJ3 also known as Kir3.1; 5’-TCCCCTTGACCAACTTGAACT-3’ and 5’- ACGACATGAGAAGCATTTCCTC-3’). For these new primer pairs 35 PCR cycles and an annealing temperature of 60°C were applied. As internal controls for the quantity and quality of the RNA specimens, RT-PCR ampliﬁcations targeting transcripts of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were performed in parallel. PCRs carried out on cDNA derived from total RNA samples of human atrium, ventricle, vascular smooth muscle, and skeletal muscle served as positive controls and PCRs in which the cDNA was replaced by water were included as negative controls. For comparing relative mRNA levels only PCR samples from within the linear range of ampliﬁcation were used. Quantitative RT- PCR (qPCR) was performed in triplicate using the qPCR-&GO kit (Q-Biogene) and the LightCycler 480 real-time PCR system (Roche), according to manufacturers’

speciﬁcations. To determine PCR efﬁciencies, standard curves were generated by 2- fold serial dilution of the cDNA template derived from VSFs treated with myocardin (series of 8). Both GAPDH and the ß-actin (ACTB) gene were used for normalization of transcription levels. Melting curve analysis was carried out to verify that only a single product was formed during the qPCR. Normalization and statistical analysis was performed using the Relative Expression Software Tool (REST)²⁶. Additional primers used in the qPCR were 5’-AGAAGGATTCCTATGTGGGCG-3’ and 5’-CATGTCGTCCCAGTTGGTGAC-3’ for ACTB, 5’-TGTACAATGCCGTGTCCAAC-3’ and 5’-TCTTCATTCGGCTCACTGAG-3’ for ANF, 5’-TCTGAGTGCCTGAACTTGC-3’ and 5’- ACTGACAGCCACACCTTCC-3’ for Cx43, 5’-CCTGTTCGACAGTCAGCCG-3’ and 5’- CGACCAAATCCGTTGACTCC-3’ for GAPDH, and 5’-AACCTGAGGGAGCGGTACTTC-3’

for smMHC.

(9)

88

Immunoﬂuorescence microscopy

Immunoﬂuorescent labeling of cells was performed as previously described²⁰. Nuclei were stained with Hoechst 33342 (Molecular Probes). HA-tagged myocardin was stained with a 1:10,000 dilution of mAb HA.11 in PBS-5% FBS. The mAb directed against sarcomeric tropomyosin (clone CH1; IgG1) and the rabbit polyclonal Ab recognizing Cx43 were obtained from Sigma-Aldrich. Goat polyclonal Abs speciﬁc for gap junction proteins

α

5 (Cx40) and

α

7 (Cx45), CACNA1C and the

α

^subunit

of the voltage-gated sodium channel type V (SCN5a also known as Nav1.5) were obtained from Santa-Cruz and were used at a 100-fold dilution. For the properties of the other cardiac and smooth muscle protein-specific Abs see^{19, 20}. Of each channel protein staining 10 randomly distributed recordings were made at 200-fold magnification and signal intensities were quantified using Image-Pro Plus (Version 4.1.0.0, Media Cybernetics). All samples were stained at the same time and processed equally. The mean intensities of samples were considered significantly different when p<0.05 using Student’s t-test.

Induction and restoration of artiﬁcial conduction block

Experimental conduction blocks were created in monolayers of rCMs as described before²¹. In brief, 2 × 10⁶ freshly isolated rCMs were seeded on pre-treated MEA culture dishes (Multi Channel Systems) with sixty titanium nitride electrodes (electrode diameter: 30 μm, inter-electrode distance: 200 μm) to generate conﬂuent monolayers of synchronously contracting cells. After 2 days of culture, a 330-μm gap spanning the entire diameter of the culture dish was created by a laser dissection microscope (P.A.L.M. Microlaser Technologies) which was either left free of cells (n=10) or ﬁlled with 5 × 10⁴ rCMs (positive control; n=9), LV.CMV.

bGal.IRES.eGFP-infected hVSFs (negative control; n=10), or LV.CMV.myocL.IRES.

eGFP-transduced hVSFs (n=9). The infection of hVSFs with LV.CMV.bGal.IRES.

eGFP or LV.CMV.myocL.IRES.eGFP took place 2 days before application to the channel. After 7 days of culture, electrograms were recorded and color-coded activation maps were generated.

Results

Transactivation of heart muscle-speciﬁc promoters by cardiogenic transcription factors

To identify the most suitable cTF for endowing hVSFs with properties of cardiomyocytes, expression plasmids were generated encoding human GATA4, Hand1, Mef2C, the 2 myocardin splice variants myocL and myocS, Nkx2.5, and Tbx5 and transient transfection experiments in HeLa cells were performed to test their ability to transactivate the promoters of the human genes specifying (i) the atrial peptide hormone ANF, (ii) the cardiac sarcomeric protein

α

MHC, and (iii) the gap junction protein Cx43, which is abundantly expressed in cardiomyocytes.

Although many promoter activation studies focus on proximal promoter elements,

(10)

89

we also included distal promoter elements in our reporter gene constructs to more closely mimic endogenous gene regulation. Functional GATA4, Hand1-E12 complex, Mef2, and Tbx5 binding sites are present in all three promoters, while only the ANF and Cx43 promoter regions contain known Nkx2.5 recognition sequences. CARG boxes, which are required for DNA binding of the myocardin- SRF complex, are present only in the ANF and

α

MHC promoters.

As shown in Figure 1, both myocardin isoforms strongly stimulated the activity of all 3 heart muscle-speciﬁc promoters. GATA4 failed to induce ANF promoter activity, but did activate the

α

MHC and Cx43 promoters albeit to a lesser extent than myocardin. All other cTFs signiﬁcantly activated only the Cx43 promoter, with Hand1 and Tbx5 showing the lowest transactivating activity toward this promoter.

Based on these screenings myocardin was used in the rest of the study.

Isolation and characterization of human ventricular scar ﬁbroblasts

In vitro culture of human myocardial scar tissue resulted in the outgrowth of cells with a spindle-shaped morphology (Figure 2). After 1 round of subculturing the cells stained negative for sarcomeric

α

-actinin, sarcomeric MHC, and smooth muscle MHC (smMHC), indicating the cultures were free from contaminating cardiac and smooth muscle cells. The surface antigen profile of the hVSFs as determined by fluorescence-activated flow cytometry was similar to that of hDFs (Table 1). Both cell types expressed at their surface large amounts of the hyaluronate receptor (CD44), the major T-cell antigen (Thy-1; CD90), endoglin (CD105), and human leukocyte class I antigens (HLA-ABC) and low numbers of membrane cofactor protein (MCP also known as CD46), a component of the complement system. In contrast, typical MSC markers such as neural cell adhesion molecule (NCAM;CD56), β3 integrin (CD61), P-selectin (CD62P), transferrin receptor (CD71), and vascular cell adhesion molecule 1 (VCAM-1; CD106), were not present at their plasma membranes. The surface of hVSFs and hDFs also stained negative for the hematopoietic markers CD11A, CD14, CD15, CD19, CD34, and CD45 as well as for the endothelial markers platelet-endothelial cell adhesion molecule 1 (PECAM- 1; CD31), vascular endothelial growth factor receptor 2 (VEGFR-2; Flk-1), and VE-cadherin. In addition, neither cell type expressed the CXC motif chemokine receptor 4 (CXCR-4; fusin), human leukocyte class II subtype DR antigens (HLA-

ANF

GAT A4

Hand1 Mef2C

MyocL Myoc

S Nkx2.5

Tbx5 0

1000 2000 3000 4000

induction factor

AMHC

GATA4 Hand1

Mef2CMyocL MyocS

Nkx2.5 Tbx5 0

100 200 300

Cx43

GAT A4

Hand1 Mef2C

MyocL My

ocS Nkx

2.5 Tbx5 0

20 40 60 80

Figure 1. ANF, AMHC, and Cx43 promoter activation by the cTFs GATA4, Hand1, Mef2C, myocL, myocS, Nkx2.5, and Tbx5 in HeLa cells. The average induction factors and standard errors are shown (n=12).

(11)

90

DR), the Coxsackie and adenovirus receptor (CAR) and the human and murine stem cell markers CD133 and Sca-1 at their surface or contained the cardiac progenitor cell marker Isl-1. Finally, in contrast to hVSFs, hDFs displayed neutral endopeptidase (NEP; MME; CD10) at their plasma membrane.

Semi-quantitative RT-PCR analyses and IFM were used to investigate the expression of cardiac, smooth, and skeletal muscle genes in hVSFs. These experiments revealed that hVSFs naturally express the GATA4 and Mef2C genes at low levels (Figure 3 and Figure 5A, lane U). Interestingly, in the majority of hVSFs the GATA4 protein is localized in the cytoplasm, while in rCMs this cTF is exclusively found in the nucleus (Figure 3). In addition, hVSFs abundantly express Cx43 (Figure 3 and Figure 5A, lane U). In these cells, the Cx43 protein is mainly localized around the nucleus with small amounts being present at intercellular borders, while in rCMs Cx43 is largely conﬁned to areas of cell-cell contacts (Figure 3). Interestingly, hVSFs also contain low amounts of CACNA1C- and SCN5a-speciﬁc mRNAs (Figure 5A, lane U).

Human ventricular scar ﬁbroblasts can undergo adipogenesis but not osteogenesis

To investigate whether hVSFs have characteristics of hMSCs, their capacity to differentiate into osteoblasts and adipocytes was tested. Under conditions that induced osteogenesis in hMSCs, hVSFs as well as hDFs failed to differentiate into osteoblasts (Figure 4). When exposed to adipogenic differentiation medium hMSCs and hVSFs underwent adipogenesis in contrast to hDFs (Figure 4).

Table 1. Surface marker proﬁle of hDFs and hVSFs.

hDFs hVSFs

CAR - -

CD 10 (NEP; MME) + -

CD11A (LFA-1; integrin αL) - -

CD14 (LPS-R) - -

CD15 - -

CD19 - -

CD31 (PECAM-1) - -

CD34 (sialomucin) - -

CD44 (HCAM) ++ ++

CD45 (LCA) - -

CD46 (MCP) + +

CD56 (NCAM) - -

CD61 (integrin ß3) - -

CD62P (P-selectin) - -

CD71 (transferrin receptor) - -

CD90 (Thy-1) ++ ++

CD105 (endoglin) ++ ++

CD106 (VCAM-1) - -

CD133 - -

CXCR4 (fusin) - -

HLA-ABC ++ ++

HLA DR - -

Sca-1 (Ly-6A/E) - -

VE-cadherin - -

VEGFR2 (Flk-1) - -

(12)

91

Forced myocardin expression results in the activation of cardiomyocyte-speciﬁc genes in human ventricular scar ﬁbroblasts

Using a combination of RT-PCR analyses and IFM, we investigated the presence of cardiac, smooth, and skeletal muscle-speciﬁc transcripts in hVSFs infected with the SIN lentivirus vectors LV.C-PGK.nls-bGal or LV.CMV.myocL-HA. After transduction of the cells with the control vector LV.C-PGK.nls-bGal no changes in the expression of the analyzed genes were observed as compared to uninfected hVSFs (compare columns U and L in Figure 5A). In contrast, infection of hVSFs with the myocardin- encoding SIN lentivirus vector LV.CMV.myocL-HA led to activation of the genes encoding the cTFs Hand1 and Hand2 and caused an increase in GATA4 (Figure 5A and 6C) and Mef2C (Figure 5A) expression. Nkx2.5-speciﬁc transcripts were not detected in LV.CMV.myocL-HA-transduced hVSFs (Figure 5A, lane M). Furthermore, myocardin induced the expression in hVSFs of the genes encoding the cardiac sarcomeric components, cardiac troponin T (cTnT), atrial and ventricular myosin light chain 2 (Mlc2a and Mlc2v), the

α

and β isoforms of cardiac myosin heavy

Figure 2. Phase contrast image of hVSFs after 30 days in culture. Scale bar represents 50 μm.

Figure 3. GATA4 (green) and Cx43 (red) staining of hVSFs and rCMs. Nuclei were stained with the blue-emitting ﬂuorochrome Hoechst 33342.

nuclei GATA4 merge nuclei Cx43 merge

hVSFsrCMs

(13)

92

chain (

α

MHC and βMHC; Figure 5),

α

-actinin, and tropomyosin (Figure 6C) as well as of the ANF gene (Figure 5 and 6C). LV.CMV.myocL-HA infection of hVSFs did not result in the accumulation of detectable amounts of cardiac troponin I (cTnI)-speciﬁc mRNA, which is normally present only in mature cardiomyocytes, but did induce expression of slow-twitch skeletal troponin I (ssTnI), which encodes the major isoform of troponin I in immature heart muscle cells²⁷ (Figure 5A and 6A). Forced myocardin expression also led to the expression in hVSFs of genes specifying the sarcoplasmic reticulum Ca(2+)-ATPase 2a (SERCA2a) and the cardiac ion channel proteins CACNA1D, HCN4, and KCNJ3 and caused a large increase in the intracellular levels of the gap junction proteins Cx40 and Cx45 (Figure 6A, B). In addition, myocardin upregulated the expression of CACNA1C and of the gene encoding SCN5a at both the mRNA and protein level (Figure 5 and 6A, B). These effects were seen in essentially all LV.CMV.myocL-HA-transduced hVSFs as judged by IFM (Figure 6). Besides activating cardiac genes, myocardin also induced the expression of the smooth muscle genes encoding aortic smooth muscle actin (ASMA), calponin h1 (CNN1), transgelin (SM22), and smMHC (Figure 5 and 6A). However, no synthesis of the skeletal muscle-speciﬁc transcripts for fast-twitch skeletal troponin I (fsTnI) and skeletal myosin heavy chain (skMHC) was observed after infection of hVSFs with LV.CMV.myocL-HA. Importantly, the presence of cardiac and smooth muscle markers in hVSFs was always associated with myocardin gene expression as detected by immunostaining for the HA tag attached to myocardin (Figure 6A). Double labeling for smMHC and either ANF or sarcomeric

α

-actinin, conﬁrmed that cardiac and smooth muscle marker genes were expressed in the same cells (Figure 6D). Spontaneous beating of LV.CMV.

myocL-HA-transduced VSFs was never observed consistent with the absence of properly organized sarcomeres as evidenced by the lack of striation following immunostaining with Abs directed against sarcomeric components (Figure 6A, C, D).

Figure 4. Adipogenesis and osteogenesis of hDFs, hMSCs, and hVSFs. Red stain labels calcium deposits in the osteogenesis assay and fatty inclusions in the adipogenesis assay.

hDFs

regular mediumdifferentiation medium

osteogenesis adipogenesis osteogenesis adipogenesis

hMSCs

+ +

osteogenesis adipogenesis hVSFs

+

(14)

93

Forced myocardin expression endows human ventricular scar ﬁbroblasts with the capacity to repair an artiﬁcially created conduction block in vitro

Using MEA culture dishes (Figure 7A), we recorded extracellular electrograms throughout a spontaneously and synchronously beating monolayer of rCMs²¹. Electrical coupling between the lower and upper part of the culture dishes was lost after establishment of a 330±20 μm-wide cell-free gap, resulting in 2 fields of asynchronously beating rCMs (n=10; Figure 7B, C). Filling the channel with LV.CMV.bGal.IRES.eGFP-transduced hVSFs (n=10) did not restore electrical conduction between both rCM fields. In contrast, seeding of rCMs in the gap (n=9) restored action potential transmission across the channel to normal levels (conduction velocity 19±0.8 cm/s), thereby resynchronizing the 2 rCM fields (Figure 7B, C). Importantly, application of LV.CMV.myocL.IRES.eGFP-transduced hVSFs into the gap (n=9) also resulted in restoration of conduction across the channel and consequently in resynchronization of both rCM fields (Figure 7B, C).

The conduction velocity across the hVSFs was only ~30% lower than that across the rCMs (13±1.1 cm/s and 19±0.8 cm/s, respectively. p<0.01).

Figure 5. (A) Semi-quantitative RT-PCR analysis of smooth, cardiac, and skeletal muscle marker gene expression in untreated hVSFs (U) and in hVSFs infected with either LV.C- PGK.nls-bGal (L) or LV.CMV.myocL-HA (M). PCRs in which the cDNA was replaced by water were included to demonstrate absence of cross-contamination (-). All PCR samples shown are from within the linear range of amplification. (B) Quantification of the relative levels of transcripts encoding CACNA1C, cTnT, Cx43, ANF, and smMHC in LV.CMV.myocL-HA- versus LV.C-PGK.nls-bGal-treated VSFs, using qPCR analysis. No significant differences were observed between LV.C-PGK.nls-bGal- and mock-transduced VSFs (data not shown). Values are normalized to both GAPDH and ACTB transcripts; p values are indicated for significantly upregulated genes (p<0.05).

U L M - U L M - U L M - U L M -

GAPDH MyocL cTnI cTnT Mlc2a Mlc2v AMHC

BMHC Hand1 Hand2 GATA4 Mef2C Nkx2.5 ANF

CACNA1C CACNA1D Cx43 HCN4 KCNJ3 SCN5a SERCA2a

fsTnI skMHC ssTnI ASMA CNN1 SM22 smMHC

ACTB GAPDH

CA CNA

1CcTnT Cx4

3 ANF

smMH C 0

20 40 60 80 100 500 1000 1500 2000

fold induction

p<0.001

p<0.001 p<0.001

p=0.031

A

B

(15)

94

Discussion

Key findings of this study are (i) myocL gene transfer leads to the induction of a (partial) cardiac phenotype in hVSFs from adult patients by activating genes encoding among others cardiomyocyte-specific TFs, sarcomeric components, and channel proteins involved in cell-to-cell communication and action potential propagation between heart muscle cells; and (ii) myocL-transduced hVSFs are able to conduct an electrical impulse over considerable distances at relatively high velocities thereby restoring an artificially created conduction block in cardiomyocyte cultures.

In vitro characterization of hVSFs

The in vitro characterization of hVSFs performed in this study revealed that these cells can transdifferentiate into adipocytes in vitro. This finding may have clinical significance since substitution of compact myocardial scar tissue by fat tissue is regularly observed in human patients at late stages of heart failure^{28, 29}. The ability of hVSFs to undergo adipogenesis in culture could be used to develop an in vitro model system to investigate the molecular mechanisms underlying the lipomatous changes occurring in diseased hearts. We also found that hVSFs are positive for GATA4 but that in most cells this cTF is found outside of the nucleus precluding a direct role in transcription activation. In addition, ex vivo cultured hVSFs contained small quantities of the cardiac ion channel proteins SCN5a and CACNA1C (Figure 5, and 6A, B) and large amounts of the gap junction component Cx43 (Figure 3, 5, and 6A, B). The expression of SCN5a and CACNA1C by cardiac scar fibroblasts has not been described before. The presence of Cx43 in hVSFs is consistent with earlier studies showing that in vitro culture and myocardial damage both lead to activation of the Cx43 gene in cardiac fibroblasts while the interstitial cells in healthy ventricular myocardium do not possess significant amounts of Cx43 (reviewed by³⁰).

Transactivating capacity of cardiogenic transcription factors

The comparative analysis of the ability of several human cTFs to transactivate 3 different human heart muscle-specific promoters corroborated previous experiments demonstrating that myocardin is an exceptionally strong transcription activator acting directly or indirectly on a wide variety of different cardiac genes¹⁶. In fact, none of the other cTFs were able to induce the human ANF, αMHC, or Cx43 promoters to a similar degree as either myocS or myocL (Figure 1). Furthermore, in this first-time comparison of the 2 main cardiomyocyte-enriched isoforms of myocardin, we failed to detect significant differences in their transactivating capacity, leaving the biological role of the amino acid sequence contributed by myocardin exon 11 undefined. The activation of pANF.LUC and pMHC.LUC by myocardin is compatible with the presence in the ANF and

α

MHC promoters of SRF binding sites. Although the Cx43 promoter lacks SRF binding sites it does contain recognition sequences for activating protein–1 (AP-1) family members such as c-Fos and c-Jun. Transcription of the c-Fos and c-Jun genes is activated by SRF,

(16)

95

sarcomeric A-actininANFGATA4 myocardin control

SERCA2atropomyosin

smMHC + HAssTnI + HA

myocardin control

HA

sarcomeric A-actinin smMHC merge merge

smMHC merge merge

ANF

myocardin control

CACNA1C + HACx40 + HACx43 + HACx45 + HASCN5a + HA

CACNA1C SCN5a Cx40 Cx43 Cx45

myocardin control 0

100 200 4000 8000

expression (%)

*

myocardin control 0

100 200 4000

8000

*

myocardin control 0

100

200 #

myocardin control 0

100 200 4000

8000

*

myocardin control 0

100 200 300

400

*

Figure 6. Analysis by IFM of cardiac and smooth muscle marker gene expression in hVSFs transduced with either LV.C-PGK.nls-bGal (control) or LV.CMV.myocL-HA (myocardin). (A) Co-labeling of LV.CMV.myocL-HA (left panel)- and LV.C-PGK.nls-bGal (right panel)-treated hVSFs for HA-tagged myocardin (green) and smMHC, ssTnI, CACNA1C, Cx40, Cx45, or SCN5a (red). (B) Quantiﬁcation of channel protein levels following immunostaining of LV.CMV.

myocL-HA- or LV.C-PGK.nls-bGal-transduced hVSFs. Values are expressed as mean±SEM relative to control samples (n=10). * indicates significant difference (p<0.001), # indicates no significant difference (p=0.23). (C) Immunostaining of LV.CMV.myocL-HA (left column)- or LV.C-PGK.nls-bGal (right column)-transduced hVSFs with antibodies specific for sarcomeric α-actinin, ANF, GATA4, SERCA2a, or tropomyosin. (D) Co-labeling of LV.CMV.myocL-HA (3 leftmost columns)- and LV.C-PGK.nls-bGal (right columns)-transduced cells for sarcomeric α-actinin and smMHC (upper row) or ANF (lower row).

A B

C

D

thereby indirectly regulating Cx43 gene expression³¹. It is likely that regulation of the Cx43 gene by SRF-myocardin complexes is indirect as well.

(17)

96

Figure 7. Electrical conduction across monolayers of rCMs after establishment of a cell- free gap and its ﬁlling with different cell populations. (A) Arrangement of the MEA in the cell culture dishes. The electrodes in rows 2 and 3 cover the area in which a cell-free gap is created, while those in rows 1 and 4 are positioned underneath bordering cardiomyocytes.

(B) Typical electrograms from electrodes bordering and spanning the gap region in rCM cultures in which the channel was left open or filled with either LV.CMV.bGal.IRES.eGFP- transduced hVSFs (hVSFs + control), LV.CMV.myocL.IRES.eGFP-transduced hVSFs, (hVSFs + myocardin) or rCMs. Electrograms were recorded 7 days after the cell-free gap was made and filled with cells. (C) Isochronous color-coded activation maps showing 2 electrically separated rCM fields after generation of the cell-free gap or after the establishment of a cellular conduction block using LV.CMV.bGal.IRES.eGFP-transduced hVSFs. Restoration of electrical conduction across the gap and consequently electrical synchronization of the culture occurred after filling the channel with LV.CMV.myocL.IRES.eGFP-transduced hVSFs or control rCMs.

107

107 100

107 100 100

107

100

103

105 100

1 2 3 4

no cells

hVSFs + control

hVSFs +

myocardin rCMs

1 2 3 4

gap

103

108 109

100

107

A

B

C

(18)

97

Similarities between myocardin-transduced human ventricular scar ﬁbroblasts and embryonic cardiomyocytes

In spite of the induction of numerous cardiac genes upon transduction with the myocardin-encoding lentivirus vector, hVSFs did not acquire the phenotype of mature cardiomyocytes, as evidenced by the absence of Nkx2.5 and cTnI, the co-expression of smooth muscle marker genes, and the absence of properly structured sarcomeres. However, the observed phenotype does resemble that of early cardiomyocytes. Like neonatal cardiomyocytes, myocL-transduced hVSFs display ssTnI (Figure 5A and 6A) instead of cTnI expression (Figure 5A)^{32, 33}. Furthermore, the expression in reprogrammed hVSFs of HCN4 and CACNA1D (Figure 5A), both required for cardiac pacemaker activity, is consistent with the phenotype of immature cardiomyocytes^{34, 35}, which are spontaneously active. It should be noted, however, that automaticity in myocL-transduced hVSFs was never observed. Other features shared by myocL-transduced hVSFs and cardiomyocyte precursor cells are the presence of heart muscle-specific sarcomeric components that are not organized into sarcomeres³⁶ and the co-expression of cardiac and smooth muscle genes^37-39 (Figure 6D). In accordance with these observations, Wu et al. recently identified a common precursor for cardiac and smooth muscle cells in the mammalian heart, which initially possesses both cardiac and smooth muscle markers⁴⁰. It should be noted, however, that expression of genes encoding definitive cardiac and smooth muscle marker genes like cTnT and smMHC was not observed until the bipotential precursor cells started to differentiate into heart or smooth muscle cells.

Electrical properties of (myocardin-transduced) human ventricular scar ﬁbroblasts

In our in vitro electrophysiological studies, hVSFs transduced with a control lentivirus vector were unable to conduct an electrical impulse over a distance of 330±20 μm (Figure 7) despite the fact that they contained high levels of Cx43 (Figure 3, 5, and 6A, B). The rapid decline of the electrical signal (Figure 7B) is consistent with previous in vitro experiments showing that the maximum distance over which cardiac ﬁbroblasts can conduct an electrical impulse from neighboring cardiomyocytes is approximately 300 μm⁴¹. Conduction under these circumstances is characterized by its low speed (4.6±1.8 mm/s) and electrotonic nature. Because ﬁbroblasts do not generate an action potential, they function as an electrical load which reduces the speed of action potential propagation in bordering cardiomyocytes^{42, 43}.

Following transduction with a lentivirus vector encoding myocardin, hVSFs abundantly expressed the genes for the gap junction components Cx40, Cx43, and Cx45 and the ion channel proteins CACNA1C, CACNA1D, HCN4, KCNJ3, and SCN5a (Figure 5, and 6A, B). These data suggest that myocardin-transduced hVSFs obtained components essential for impulse propagation¹². Although the functionality of the aforementioned channel proteins has not been tested, myocardin-transduced hVSFs were able to resynchronize in vitro 2 ﬁelds of rCMs spaced more than 300 μm apart (Figure 7). The conduction velocity across the

(19)

98

cell-ﬁlled gap (13±1.1 cm/s) approached that of rCMs (19±0.8 cm/s), which makes purely electrotonic conduction unlikely. However, as yet it cannot be ruled out completely that increased Cx40 and Cx45 levels are important contributors to the improved conduction velocity of these cells.

Study limitations

In our in vitro electrophysiological studies neonatal rat cardiomyocytes were used.

Although the use of human heart muscle cells is preferable, the limited availability of this cell type and their tendency to dedifferentiate during culture prevented the use of human cardiomyocytes. Furthermore, although forced myocardin expression endows hVSFs with (i) the 3 major connexins characteristic for rCMs (Figure 6A), (ii) cardiac ion channel proteins necessary for the propagation of Na⁺, K⁺, and Ca²⁺ currents (Figure 5 and 6A), and (iii) the ability to conduct electricity at high velocities and to repair an artiﬁcially created conduction block in vitro (Figure 7), an in depth electrophysiological characterization of these cells is required to prove that they truly become excitable.

Perspectives

Genetic modification of VSFs may provide a means of repairing conduction defects in damaged hearts. However, the transduction of VSFs with proper doses of all genes needed to produce a cardiac action potential represents a formidable technical challenge. Overexpression of myocardin activates an extensive cardiac gene expression program in hVSFs leading to the production of key sarcomeric and electrical components of cardiomyocytes. This coincides with the appearance of electrical conduction in these cells and the restoration of an artificially created cellular conduction block in vitro. With further refinements this technology holds the promise to transdifferentiate VSFs into cardiomyocyte-like cells in situ, thereby potentially improving myocardial mechanical performance and repairing electrical conduction defects.

Acknowledgements

The authors are indebted to Dirk Ypey for critical reading of the manuscript and Binie Klein for supplying the hDFs. We also thank Robert Adelstein, Rob Hoeben, Jun-ichi Miyazaki, Didier Trono, and Tetsuya Yamagata for supplying the smMHC-speciﬁc rabbit antiserum, pLV-CMV-IRES-eGFP, pCAGGS, psPAX2, and pSSR

α

-GATA-4, respectively. Part of this work was ﬁnanced by the Netherlands Organization for Health Research and Development (ZonMw; Grant MKG5942) and the Interuniversity Cardiology Institute of the Netherlands (ICIN).

(20)

99

References

1. Frangogiannis NG. The mechanistic basis of infarct healing. Antioxid Redox Signal. 2006;8(11-12):1907-1939.

2. Sun Y, Kiani MF, Postlethwaite AE, Weber KT. Infarct scar as living tissue.

Basic Res Cardiol. 2002;97(5):343-347.

3. Swynghedauw B. Molecular mechanisms of myocardial remodeling. Physiol Rev. 1999;79(1):215-262.

4. Dimmeler S, Zeiher AM, Schneider MD. Unchain my heart: the scientiﬁc foundations of cardiac repair. J Clin Invest. 2005;115(3):572-583.

5. de Muinck ED, Thompson C, Simons M. Progress and prospects: cell based regenerative therapy for cardiovascular disease. Gene Ther.

2005;13(8):659-671.

6. Chien KR. Lost and found: cardiac stem cell therapy revisited. J Clin Invest.

2006;116(7):1838-1840.

7. Murry CE, Reinecke H, Pabon LM. Regeneration gaps. Observations on stem cells and cardiac repair. J Am Coll Cardiol. 2006;47(9):1777-1785.

8. Brown RD, Ambler SK, Mitchell MD, Long CS. The cardiac ﬁbroblast:

therapeutic target in myocardial remodeling and failure. Annu Rev Pharmacol Toxicol. 2005;45:657-687.

9. Etzion S, Barbash IM, Feinberg MS, Zarin P, Miller L, Guetta E, Holbova R, Kloner RA, Kedes LH, Leor J. Cellular cardiomyoplasty of cardiac ﬁbroblasts by adenoviral delivery of MyoD ex vivo: an unlimited source of cells for myocardial repair. Circulation. 2002;106(12 Suppl 1):I125-130.

10. Kizana E, Ginn SL, Allen DG, Ross DL, Alexander IE. Fibroblasts can be genetically modiﬁed to produce excitable cells capable of electrical coupling.

Circulation. 2005;111(4):394-398.

11. Murry CE, Kay MA, Bartosek T, Hauschka SD, Schwartz SM. Muscle differentiation during repair of myocardial necrosis in rats via gene transfer with MyoD. J Clin Invest. 1996;98(10):2209-2217.

12. Hille B. Ion channels of excitable membranes. Third ed. Massachusetts, USA: Sinauer Associates, Inc; 2001.

13. Dunwoodie SL. Combinatorial signaling in the heart orchestrates cardiac induction, lineage speciﬁcation and chamber formation. Semin Cell Dev Biol. 2007;18(1):54-66.

14. Srivastava D. Making or breaking the heart: from lineage determination to morphogenesis. Cell. 2006;126(6):1037-1048.

15. Fukushige T, Brodigan TM, Schriefer LA, Waterston RH, Krause M. Deﬁning the transcriptional redundancy of early bodywall muscle development in C. elegans: evidence for a uniﬁed theory of animal muscle development.

Genes Dev. 2006;20(24):3395-3406.

16. Wang D-Z, Chang PS, Wang Z, Sutherland L, Richardson JA, Small E, Krieg PA, Olson EN. Activation of cardiac gene expression by myocardin, a transcriptional cofactor for serum response factor. Cell. 2001;105(7):851- 862.

(21)

100

17. Chen J, Kitchen CM, Streb JW, Miano JM. Myocardin: a component of a molecular switch for smooth muscle differentiation. J. Mol. Cell Cardiol.

2002;34(10):1345-1356.

18. Creemers EE, Sutherland LB, Oh J, Barbosa AC, Olson EN. Coactivation of MEF2 by the SAP domain proteins myocardin and MASTR. Mol Cell.

2006;23(1):83-96.

19. van Tuyn J, Knaan-Shanzer S, van de Watering MJ, de Graaf M, van der Laarse A, Schalij MJ, van der Wall EE, de Vries AA, Atsma DE. Activation of cardiac and smooth muscle-speciﬁc genes in primary human cells after forced expression of human myocardin. Cardiovasc Res. 2005;67(2):245- 255.

20. van Tuyn J, Atsma DE, Winter EM, van der Velde-van Dijke I, Pijnappels DA, Bax NA, Knaan-Shanzer S, Gittenberger-de Groot AC, Poelmann RE, van der Laarse A, van der Wall EE, Schalij MJ, de Vries AA. Epicardial cells of human adults can undergo an epithelial-to-mesenchymal transition and obtain characteristics of smooth muscle cells in vitro. Stem Cells.

2006;25(2):271-278.

21. Pijnappels DA, Schalij MJ, van Tuyn J, Ypey DL, de Vries AA, van der Wall EE, van der Laarse A, Atsma DE. Progressive increase in conduction velocity across human mesenchymal stem cells is mediated by enhanced electrical coupling. Cardiovasc Res. 2006;72(2):282-291.

22. Niwa H, Yamamura K, Miyazaki J. Efﬁcient selection for high-expression transfectants with a novel eukaryotic vector. Gene. 1991;108(2):193- 199.

23. Vellinga J, Uil TG, de Vrij J, Rabelink MJ, Lindholm L, Hoeben RC. A system for efﬁcient generation of adenovirus protein IX-producing helper cell lines.

J Gene Med. 2006;8(2):147-154.

24. Donello JE, Loeb JE, Hope TJ. Woodchuck hepatitis virus contains a tripartite posttranscriptional regulatory element. J Virol. 1998;72(6):5085-5092.

25. Graham FL, van der Eb AJ. A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology. 1973;52(2):456-467.

26. Pfafﬂ MW, Horgan GW, Dempﬂe L. Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Res. 2002;30(9):e36.

27. Parmacek MS, Solaro RJ. Biology of the troponin complex in cardiac myocytes. Prog Cardiovasc Dis. 2004;47(3):159-176.

28. Baroldi G, Silver MD, De Maria R, Parodi O, Pellegrini A. Lipomatous metaplasia in left ventricular scar. Can J Cardiol. 1997;13(1):65-71.

29. Winer-Muram HT, Tann M, Aisen AM, Ford L, Jennings SG, Bretz R. Computed tomography demonstration of lipomatous metaplasia of the left ventricle following myocardial infarction. J Comput Assist Tomogr. 2004;28(4):455- 458.

30. Camelliti P, Green CR, Kohl P. Structural and functional coupling of cardiac myocytes and ﬁbroblasts. Adv Cardiol. 2006;42:132-149.

31. Mitchell JA, Shynlova O, Langille BL, Lye SJ. Mechanical stretch and progesterone differentially regulate activator protein-1 transcription factors

(22)

101

in primary rat myometrial smooth muscle cells. Am J Physiol Endocrinol Metab. 2004;287(3):E439-445.

32. Reiser PJ, Westfall MV, Schiafﬁno S, Solaro RJ. Tension production and thin-ﬁlament protein isoforms in developing rat myocardium. Am J Physiol.

1994;267(4 Pt 2):H1589-1596.

33. Huang XP, Pi YQ, Lee KJ, Henkel AS, Gregg RG, Powers PA, Walker JW.

Cardiac troponin I gene knockout: a mouse model of myocardial troponin I deﬁciency. Circ. Res. 1999;84(1):1-8.

34. Takemura H, Yasui K, Opthof T, Niwa N, Horiba M, Shimizu A, Lee JK, Honjo H, Kamiya K, Ueda Y, Kodama I. Subtype switching of L-Type Ca 2+ channel from Cav1.3 to Cav1.2 in embryonic murine ventricle. Circ J.

2005;69(11):1405-1411.

35. Yasui K, Liu W, Opthof T, Kada K, Lee JK, Kamiya K, Kodama I. I(f) current and spontaneous activity in mouse embryonic ventricular myocytes. Circ Res. 2001;88(5):536-542.

36. Gonzalez-Sanchez A, Bader D. In vitro analysis of cardiac progenitor cell differentiation. Dev Biol. 1990;139(1):197-209.

37. Jain MK, Kashiki S, Hsieh CM, Layne MD, Yet SF, Sibinga NE, Chin MT, Feinberg MW, Woo I, Maas RL, Haber E, Lee ME. Embryonic expression suggests an important role for CRP2/SmLIM in the developing cardiovascular system. Circ Res. 1998;83(10):980-985.

38. Li L, Miano JM, Cserjesi P, Olson EN. SM22 alpha, a marker of adult smooth muscle, is expressed in multiple myogenic lineages during embryogenesis.

Circ Res. 1996;78(2):188-195.

39. Samaha FF, Ip HS, Morrisey EE, Seltzer J, Tang Z, Solway J, Parmacek MS. Developmental pattern of expression and genomic organization of the calponin-h1 gene. A contractile smooth muscle cell marker. J Biol Chem.

1996;271(1):395-403.

40. Wu SM, Fujiwara Y, Cibulsky SM, Clapham DE, Lien CL, Schultheiss TM, Orkin SH. Developmental origin of a bipotential myocardial and smooth muscle cell precursor in the mammalian heart. Cell. 2006;127(6):1137- 1150.

41. Gaudesius G, Miragoli M, Thomas SP, Rohr S. Coupling of cardiac electrical activity over extended distances by ﬁbroblasts of cardiac origin. Circ Res.

2003;93(5):421-428.

42. Miragoli M, Gaudesius G, Rohr S. Electrotonic modulation of cardiac impulse conduction by myoﬁbroblasts. Circ Res. 2006;98(6):801-810.

43. Kizana E, Ginn SL, Smyth CM, Boyd A, Thomas SP, Allen DG, Ross DL, Alexander IE. Fibroblasts modulate cardiomyocyte excitability: implications for cardiac gene therapy. Gene Ther. 2006;13(22):1611-1615.

(23)