Heart disease is one of the major causes of death in the industrial countries, with great socioeconomic relevance. Successful treatment strategies for acute myocardial infarction led (somewhat paradoxically) to an increase in the number of patients with congestive heart failure (Takeishi and Walsh, 2001). Because most of the adult cardiomyocyte population has left the cell cycle—resulting in a very limited self-regeneration potential—much effort concerns the therapeutic restoration of sufficient myocardial function. Besides numerous pharmacological and surgical approaches, cell therapy may offer attractive possibilities for the long-term regeneration of viable de novo myocardium. Cellular cardiomyoplasty—meaning the direct implantation of isolated cells into myocardial lesions—has actually been performed in more than 150 patients worldwide (Chachques et al., 2004) with promising preliminary results. As an alternative approach, a lot of research is directed toward the creation of whole implantable myocardial grafts. Despite impressive advances in related technologies in the recent years, unresolved issues remain, with basic questions about cell source, cellular differentiation, angiogenesis, and optimal functional and electrical integration requiring to be addressed. One of the crucial points is the identification of the microenvironmental mechanisms leading to the differentiation of precursor cells into functional heart cell types. Factors modulating straightforward differentiation (e.g., extracellular matrix, growth factors, and cell–cell interactions) are of particular interest. Therefore, in addition to animal experiments, cardiomyocyte cell cultures are important in vitro models for analysis and manipulation of cellular differentiation processes.
Most in vitro experiments have been performed using confluent monolayers. These monolayers enable cell metabolism analysis (Zoref-Shani et al., 1988), hypoxia experiments (Hiebert and Ping, 1997), pharmacological treatment (Clark et al., 1991; Ponsard et al., 1999; Carrier et al., 2002), and gene manipulation (Gojo et al., 1996). However, this culture technique has inherent disadvantages. Long-term cultivation may result in a selection of specific cellular subtypes and in significant changes of structure and gene expression toward a dedifferentiation (Eppenberger-Eberhardt et al., 1990; Li et al., 1996; Bird et al., 2003). Furthermore, the lack of three-dimensional orientation of cardiomyocyte monolayers is not suitable for investigations of cell–cell interaction in an in vivo–like environment.
Three-dimensional cardiomyocyte cultures in combination with recent molecular biology techniques may offer new possibilities for the examination of the growth potential, analysis of time-specific gene expression, and cellular arrangement in an organotypical environment (Eschenhagen et al., 1997; Akins et al., 1999; Zimmermann et al., 2002; Baar et al., 2005). Numerous different tissue-like culture models are established by using various culture techniques, biological and nonbiological scaffolds, culture media, and cell sources (Zimmermann et al., 2000; Shimizu et al., 2003; Shin et al., 2003; Kelm et al., 2004; Kofidis et al., 2004). The aim of our study was to develop an easy-to-produce three-dimensional scaffold-free culture model with homogenously distributed cardiomyocytes that could serve as a robust test system for basic research in vitro and in vivo.
In this study, we proposed a two-step culture technique for the generation of three-dimensional cell aggregates (Fig. 1). This approach requires a viable single cell suspension prepared from fetal heart ventricles (E13). Therefore, we initially tested different digestion protocols to obtain a single cell suspension. Thus, prepared ventricles were incubated with papain (0.1%, 0.2%) or trypsin (0.025%, 0.1%, 0.25%) solutions for 30 and 60 min at 37°C.
The highest mean number of vital cells after dissociation estimated by Trypan Blue staining was observed by using 0.2% papain solution for 30 min (Fig. 2A). Although there were no significant differences between the number of vital cells following papain digestion for 30 or 60 min and trypsin treatment for 60 min, the percentage of cultured cardiomyocytes analyzed by alpha-actinin immunocytochemistry staining was significantly diminished in the trypsin group (Fig. 2B). The use of papain instead of trypsin in the dissociation procedure resulted in a higher percentage of alpha-actinin–positive myocytes (approximately 78%) after 5 days in culture. Therefore, the treatment of 0.2% papain for 30 min was used as the standard digestion protocol for all subsequent experiments in this study. Figure 3A demonstrates alpha-actinin immunofluorescence of cardiomyocytes isolated with papain after 7 days in vitro.
Enhanced Green Fluorescent Protein Gene Transfer Into Cardiomyocytes
Adenoviral infection was performed to analyze the efficiency of gene transfer into fetal cardiomyocytes. After papain digestion, dissociated cardiocytes were cultured for 7 days before adenoviral infection with the eGFP vector (Fig. 3B–G) or with the enhanced green fluorescent protein (eGFP) -actin fusion protein vector (Fig. 3H–N). First, eGFP signals in autonomously beating cells were observed 12 hr after infection. Seven days after adenoviral infection with eGFP vector, 77% of the alpha-actinin–positive cardiomyocytes (76.9 ± 5.1; mean ± SEM; n = 6) expressed GFP.
Newly synthesized actin stress fibers in fibroblast-like cells (Fig. 3H) as well as actin fibers in cardiomyocytes (Fig. 3I–N) were analyzed 7 days after adenoviral infection with the eGFP–actin fusion protein vector. A total of 68% of the alpha-actinin–positive cardiomyocytes (68.2 ± 4.3; mean ± SEM; n = 10) expressed the eGFP-actin fusion protein. However, due to the adenoviral infection, appearance of alpha-actinin aggregation could be observed in cardiomyocytes by immunocytochemistry (Fig. 3E,M).
Three-Dimensional Tissue Culture
Homogenous three-dimensional aggregates were generated by using a two-step culture technique (Figs. 1, 4). Some of the aggregates could be cultured up to 8 months without loss of autonomous beating. For histological analysis and pharmacological experiments, scaffold-free aggregates with diameters of 250 to 600 μm were kept in culture for 4 to 12 weeks. The three-dimensional cultures possessed more than 25 cell layers without central necrosis shown by immunohistochemistry for the cardiomyocyte-specific protein alpha-actinin and 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI) nuclei staining after 8 weeks in vitro (Fig. 5A–C,E). In contrast to alpha-actinin distribution, expression of fibronectin and vimentin were predominantly observed at the border, indicating that fibroblast-like cells have preferentially located themselves near the tissue surface (Fig. 5C,D,J,K,L). As shown by immunohistochemistry, desmosomal cadherins and connexin 43 (a specific protein of gap junctions) were homogeneously expressed throughout the entire tissue culture (Fig. 5F–I).
Scanning and transmission electron microscopy were performed to analyze the organization and ultrastructure of cultured cells. Myocardium-specific filaments, bands, and cellular contact zones could be clearly identified (Fig. 6A,B). However, in comparison to cardiac tissue in vivo, the immunohistochemical and electron microscopic views indicate that, in the tissue culture model, the cardiomyocytes were organized in a circular manner.
Pharmacological treatment using the beta-adrenergic receptor agonist isoprenaline and the antagonist propranolol modulated the beating frequency of the cell aggregate in a dose-dependent manner (Fig. 7A,B; movie 1, the movie can be viewed at http://www.anatomie.uni-tuebingen.de/anatue/forsch/cardio/). The application of isoprenaline resulted in a dose-related increase of the beating rate, up to 2.5-fold, at a concentration of 10−5 M (Fig. 7B). This effect could be inhibited by the beta-adrenergic antagonist propranolol in a dose-related manner as well (Fig. 7C). However, in contrast to the percentage effect of isoprenaline the absolute baseline beat frequency of cultures could be markedly different between individual aggregates as shown in Figure 7A (aggregate 1 vs. aggregate 2). These pharmacological effects confirm the physiological status of the beta-adrenergic system in the myocyte tissue cultures.
Based on our findings of long-term contractility, specific expression of multiple cardiac-associated markers, and appropriate response of the differentiated cells to cardioactive drugs, the culture technique presented in this article seems to be useful for studying the developmental capacity and functional properties of cultured cardiomyocytes in a three-dimensional arrangement. To obtain cell aggregates, the suspension of dissociated heart cells was shaken on a rocker platform. Under these conditions, homogenous cell aggregates could be generated. The induction of reaggregation by shaking or rotating, which prevents surface adhesion of the isolated cells, is an established culture method (Moscona, 1968; Hermersdorfer and Schulze, 1976). We have combined this technique with a more recent membrane culture method by transferring the aggregates onto Millipore membrane inserts. This membrane culture system has been shown to improve the long-term culture of three-dimensional tissues under organotypical conditions. Cells or tissues are cultured at the interface between culture medium and gas environment to ensure adequate in vitro nutrition and oxygen supply (Stoppini et al., 1991; Just et al., 1996). The used uncoated inserts (millicell-CM, Millipore) strongly reduced cellular attachment on the surface and cell outgrowth from tissue culture. This finding is due to the polytetrafluoroethylene membrane of culture inserts. Therefore, the membrane characteristics are responsible for the maintenance of three-dimensional aggregate morphology even after 12 weeks in vitro.
Watzka and coworkers (2000) have generated scaffold-free aggregate cultures with a two-step process similar to our technique. Neonatal and postnatal murine heart cells were aggregated in a 25-ml Erlenmeyer spinner flask for 24 hr. Subsequently, the aggregates were cultured for 1 week on Falcon membrane inserts with a pore diameter of 1 μm (in contrast to the smaller 0.4-μm pore size we used). This study demonstrated that the capacity for aggregate formation and autonomous contraction was inversely related to the postnatal developmental stage of prepared cardiac tissue. In contrast to our fetal cultures, the spontaneous contractile activity of neonatal cardiomyocytes completely disappeared after 144 hr in vitro (Watzka et al., 2000). Cell viability and function in culture substantially decrease with increasing developmental stage of cardiac tissue. Additionally, older cells aggravate the known aging processes of long-term cell cultures (Terman et al., 2004). In our culture model, we isolated cardiocytes from embryonic day 13. Using fetal hearts offers the advantage of higher numbers of intact cells and a higher percentage of progenitors obtained after digestion (Das et al., 2004). This fact may explain why, even after long-term cultivation of up to 8 months, a functional loss of autonomous beating did not occur. On the other hand, this increases technical demands with respect to tissue acquisition in comparison to the widely used neonatal rats (Akins et al., 1997; Bursac et al., 2002; Shimizu et al., 2002; Zimmermann et al., 2004; Baar et al., 2005).
In contrast to the majority of culture protocols, our basic cell culture medium was supplemented with adult horse serum. In comparison to adult horse serum, fetal calf serum is a more mitotic culture medium supplement. However, for the cultivation of three-dimensional organotypical tissue cultures, horse serum was often added to the culture medium to induce a more pronounced cell differentiation and to reduce strong cell proliferation (Stoppini et al., 1991; Just et al., 1996). Because each cell type shows different proliferation rates, a change in cell population distribution during long-term cultivation could be expected, dependent on the serum used.
The generation of cell aggregates requires a viable cell suspension. Therefore, we initially tested different dissociation protocols to optimize the digestion process. Treatment with papain resulted in a high yield of viable cells immediately after cell dissociation and in a high percentage of viable fetal cardiomyocytes available during culturing. However, our satisfactory results with papain should not be regarded as a general recommendation, because the optimal choice of digestion enzyme and enzymatic incubation duration depends on the developmental stage of heart tissue. The protocols for cardiomyocyte isolation and dissociation significantly differ in the literature. Two enzymes, trypsin and collagenase are usually applied to dissociate minced pieces of heart tissue (1–2 mm3) into isolated cells. The reported concentrations of trypsin, the most commonly used enzyme, range from 0.05 to 0.25% in the literature. Collagenase is used at lower concentrations, typically 0.05–0.1%. Some protocols use combinations of trypsin and collagenase, or collagenase and pancreatin. A detailed methods review has been published recently by Chlopcíková et al. (2001). We decided to use a scaffold-free culture technique, because insufficient and nonhomogenous cell distribution and migration into the scaffolds are known problems of solid scaffolds that can have an impact on cell–cell interactions in the cultured tissue. The reconstitution of sufficiently strong contracting heart muscle constructs (considered to be the capacity to generate a systolic force of 1–3 mN/ mm2 (Zimmermann et al., 2002) has proven difficult using solid scaffolds (Li et al., 1999). Liquid matrices provided improved functional properties of the constructs (Zimmermann et al., 2000). Scaffold-free cell sheet stacking after cultivation of monolayer patches on a thermosensitive culturing membrane showed promising mechanical properties (Shimizu et al., 2003). Recently, the scaffold-free generation of beating three-dimensional “myocardial microtissues” was reported by Kelm and coworkers (Kelm et al., 2004). A cultivation technique as hanging drops resulted in aggregation of dissociated rat heart cells to three-dimensional spherical clusters up to 200 μm in diameter. Beyond this size (above 10,000 cells per hanging drop), central necrosis could be observed. Characterization of the aggregates revealed key tissue properties closely comparable to those of our cultures.
Gene transfer plays a pivotal role for the entire technology, because it is thought that successful cardiac tissue engineering will require efficient gene transfer to express desired therapeutic or phenotype-modulating transgenes. The induction and study of capillary angiogenesis is of particular interest. The successful application of growth factors such as basic fibroblast growth factor and vascular endothelial growth factor (VEGF) to spontaneously beating “heart-like” tissue from murine and human pluripotent embryonic stem by using specific culture conditions could be demonstrated by different groups (Yamamoto et al., 2003; Yau et al., 2004). In this study, we could demonstrate a high efficiency of gene transfer into fetal cardiomyocytes by adenoviral infection. However, with increasing cultivation time, appearance of alpha-actinin aggregation could be observed in beating cardiomyocytes after adenoviral infection.
In comparison to native cardiac tissue, our culture model revealed a circular spatial fiber organization. This arrangement would have limitations for transplantation experiments. However, this strategy was not the aim of our study. The use of fetal myocardial cells resulted in strong spontaneous contractions of the spheroids, even without the application of mechanical stretching or rotation. To reach the requirements of a potentially implantable graft, a parallel fiber direction will be necessary that can be obtained by mechanical stretching techniques.
The proposed cell culture system offers possibilities for in vitro studies of basic processes that play major roles in cardiac cell differentiation and disease-like apoptosis, cellular signaling, cell cycle regulation, calcium handling, electrophysiological behavior, or gene expression under different culture conditions, including hypoxia, hyperglycemia, and treatment with drugs, cytokines, or growth factors. As a result of the early developmental stage of the fetal heart cells, our proposed model could be suitable for comparative investigations with in vitro differentiated cardiomyocyte-like cells derived from embryonic stem cells and for future tissue engineering and biocompatibility studies.
Timed pregnant rats (Sprague-Dawley), anesthetized with ketamine/xylazine were killed 13 days after coitus (E13), and the embryos were removed. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (NIH Publication No. 85-23, revised 1996). The embryonic hearts were prepared and collected in Hanks' balanced salt solution buffer (Gibco/Invitrogen, Karlsruhe, Germany). Ventricles were dissected and transferred into papain (Sigma, Taufkirchen, Germany) or trypsin (Gibco/Invitrogen) digestion solutions containing 0.05% (w/v) DNAse I (Sigma) for 30 and 60 min at 37°C. Trypsin was used within a concentration range of between 0.025 and 0.25% (w/v) in Ca2+ and Mg2+ free phosphate buffered saline (PBS) buffer (Gibco/Invitrogen). Papain (0.1 and 0.2% [w/v]) was solubilized in MEM medium containing 25 mM HEPES and 0.35 g/L bicarbonate. During incubation, tissues were carefully triturated every 15 min with a fire-polished blue tip. After completion of the incubation period, horse serum was added to give a final concentration of 10% in the enzyme solution. Afterward, ventricular tissue was triturated with a fire-polished yellow tip to obtain a single cell suspension. This cell suspension was centrifuged at 220 × g for 5 min. The cell pellet was resuspended in culture medium containing DMEM/F12 medium, 15 mM HEPES, and 10% horse serum (all from Gibco/Invitrogen). Cell number and viability were determined in a hemocytometer by means of Trypan Blue staining. To investigate the survival of cardiomyocytes under each of the isolation protocols, cells were seeded onto uncoated wells of a 24-well plate (Falcon/Becton Dickinson, Heidelberg, Germany) at a density of 10,000 cells per cm2 and cultured up to 7 days before immunocytochemical analysis. Under these culture conditions, cardiomyocytes formed small cell clusters on the culture plates.
Aggregation of single cell suspensions was induced by adding 300,000 cells in 1-ml culture medium to a 15-ml Falcon tube located on a culture rocker platform (Fig. 1). Under shaking conditions, cells were cultured for 3 days in a humidified incubator at 37°C and 5% CO2. After this incubation period, the resultant aggregates were transferred onto uncoated membrane inserts (Millicell CM, 0.4 μm pore size, Millipore, Eschborn, Germany) that were positioned in the wells of a six-well plate containing 1.2 ml of culture medium. We cultivated up to four aggregates per insert in a humidified incubator at 37°C and 5% CO2. The culture medium was renewed every 3 days. For histological analysis and pharmacological experiments, scaffold-free aggregates were cultured in this way for 4 to 12 weeks. Some aggregates could be kept in culture up to 8 months without loss of autonomous beating.
Cultured heart cells and cryostat sections (cross-sections, 12 μm) of membrane cultures were fixed with 4% phosphate buffered paraformaldehyde for 20 min. After three washes with PBS (0.1 mM, pH 7.4) for 5 min, cells were blocked with PBS containing 0.3% Triton X-100 and 4% normal goat serum for 30 min. Sections and cells were then incubated with primary antibodies overnight at 4°C. The following primary antibodies were used: alpha-actinin (1:1,000, monoclonal, mouse, Sigma), pan-cadherin (1:500, monoclonal, mouse, Sigma), connexin 43 (1:100, monoclonal, mouse; Babco, Richmond, CA), fibronectin (1:800, polyclonal, rabbit, DAKO), and vimentin (1:100, monoclonal, mouse, DAKO). After four washes with PBS, cells were incubated with DAPI and fluorochrome-linked secondary antibody for 1 hr at room temperature. The following secondary antibodies were used: goat anti-mouse Alexa 488, 1:200 (Mobitec, Göttingen, Germany); goat anti-rabbit Alexa 488, 1:200 (Mobitec); goat anti-mouse Cy3, 1:300 (Jackson ImmunoResearch, West Grove, PA); goat anti-rabbit Cy3, 1:300 (Jackson). For fluorescence detection, sections and cells were washed again four times for 15 min with PBS and then briefly rinsed in distilled water, dried, and covered with Kaiser's solution (Merck, Darmstadt, Germany).
Scanning Electron Microscopy
Cardiomyocyte aggregate cultures were fixed with a solution containing 2% glutaraldehyde and 3% formaldehyde in cacodylate buffer (0.1 M cacodylate, 0.09 M sucrose, 0.01 M MgCl2 and 0.01 M CaCl2, pH 6.9) for 1 hr on ice and washed with cacodylate buffer. After washing five times in TE buffer (0.02 M TRIS, 1 mM ethylenediaminetetraacetic acid, pH 6.9), samples were dehydrated with a graded series of acetone (10, 30, 50, 70, 90, 100%) on ice, each step 15 min, followed by critical-point drying with liquid CO2. Samples were sputter coated with a gold film of approximately 10-nm thickness before examination with a Zeiss field emission scanning electron microscope DSM982 Gemini at an acceleration voltage of 5 kV using the Everhart Thornley SE detector and the inlens-SE detector in a 50:50 ratio.
Transmission Electron Microscopy
Aggregate cultures grown on membranes were fixed as described above, washed with cacodylate buffer, contrasted with 1% aqueous osmium tetroxide for 2 hr at room temperature and subsequently washed with cacodylate buffer. Samples were then dehydrated with a graded series of acetone and embedded in an epoxy resin according to the Spurr formula (Spurr, 1969). Ultrathin sections were cut with glass knives, collected onto Formvar-coated copper grids and counterstained with uranyl acetate and lead citrate before examination in a Zeiss transmission electron microscope EM910 at an acceleration voltage of 80 kV.
To examine the effect of adrenergic agonist and antagonist, cardiomyocyte aggregates cultured for 5 to 12 weeks were incubated in culture medium with various concentrations of isoprenaline (10−10 M to 10−5 M) and propranolol (10−5 M to 10−4 M, both from Sigma). Determination of the respective beating rates was performed using a Zeiss Axiovert microscope with an Axio Cam HRc camera (Zeiss, Jena, Germany). Culture dishes were warmed on a heating plate to ensure a constant temperature at 37°C during measurement. Ten minutes after the addition of drugs, the frequency of the contractions per minute were counted 3 times for 1 min.
Adenoviral Gene Transfer
Recombinant adenoviruses were derived from E1/E3 deleted adenovirus type 5 cosmid vectors by cloning an expression cassette consisting of CMVie-promoter, coding region, and SV40 poly-A site, into an adeno–cosmid vector. The expression cassette for enhanced green fluorescent protein (eGFP) was amplified by polymerase chain reaction (PCR) from peGFP-C1 (Clontech Laboratories, Palo Alto, CA) using primers with additional SwaI and Psp1406I sites and ClaI and XbaI sites to the 5′- and 3′-ends, respectively. The PCR fragment was cloned into pGEM-T (Invitrogen, Carlsbad, CA) to yield pGEMeGFP. Excision of the expression cassette with Psp1640I/XbaI and insertion into ClaI/XbaI-digested pAdcos45 vector yielded pAdcos45eGC1.
In an analogous manner, the coding region for the eGFP–actin fusion protein (Ballestrem et al., 1998) was inserted into pGEMTO to yield pGEMTOeGFP–actin. The expression cassette was excised by –I and –I and cloned into the single XbaI site of adenocosmid vectors, which had been filled in using T4-DNA polymerase, to yield pAdcos45TOeGFPactinE3TetR. This adenoviral vector allows efficient expression of eGFP–actin in the presence of tetracycline (1 μg/ml). Recombinant adenovirus was produced after transfection of cosmid DNA into 293LP cells (Microbix Biosystems, Ontario, Canada) as described previously (Wiethe et al., 2003). The adenoviruses were propagated and purified according to previously published methods (Graham and Prevec, 1992), with minor modifications. Horse serum was replaced by fetal calf serum. Virus particles were harvested from freeze–thaw lysates digested with Benzonase (Merck, West Point, PA) to reduce viscosity. Purified virus was isolated by two rounds of CsCl density gradient centrifugation and stored at 4°C. The titers of the virus stocks were determined by plaque assay on the 293LP cells (Mittereder et al., 1996).
Before adenoviral infection with the eGFP vector or with the eGFP–actin fusion protein vector, papain digested cardiocytes were directly cultured on culture plates for 7 days. Seven days after infection, cells were fixed and cardiomyocytes were stained for alpha-actinin. The eGFP-, eGFP-actin–, and alpha-actinin–expressing cells were monitored by fluorescence microscopy.
Quantification and Statistical Analysis
Data are expressed as mean and the standard error of the mean (± SEM). The statistical significance of the differences were analyzed by analysis of variance and the post hoc Fisher's Protected Least Significant Difference test. A P value less than 0.05 was considered as statistically significant.
We thank Mr. Jan Tinius and Mrs. Renate Bonewald for expert technical assistance.