To bridge the gap between two-dimensional cell culture and tissue, various three-dimensional (3-D) cell culture approaches have been developed for the investigation of cardiac myocytes (CMs) and cardiac fibroblasts (CFs). However, several limitations still exist. This study was designed to develop a cardiac 3-D culture model with a scaffold-free technology that can easily and inexpensively generate large numbers of microtissues with cellular distribution and functional behavior similar to cardiac tissue. Using micromolded nonadhesive agarose hydrogels containing 822 concave recesses (800 μm deep × 400 μm wide), we demonstrated that neonatal rat ventricular CMs and CFs alone or in combination self-assembled into viable (Live/Dead stain) spherical-shaped microtissues. Importantly, when seeded simultaneously or sequentially, CMs and CFs self-sorted to be interspersed, reminiscent of their myocardial distribution, as shown by cell type-specific CellTracker or antibody labeling. Microelectrode recordings and optical mapping revealed characteristic triangular action potentials (APs) with a resting membrane potential of −66 ± 7 mV (n = 4) in spontaneously contracting CM microtissues. Under pacing, optically mapped AP duration at 90% repolarization and conduction velocity were 100 ± 30 ms and 18.0 ± 1.9 cm/s, respectively (n = 5 each). The presence of CFs led to a twofold AP prolongation in heterogenous microtissues (CM-to-CF ratio of 1:1). Importantly, Ba2+-sensitive inward rectifier K+ currents and Ca2+-handling proteins, including sarco(endo)plasmic reticulum Ca2+-ATPase 2a, were detected in CM-containing microtissues. Furthermore, cell type-specific adenoviral gene transfer was achieved, with no impact on microtissue formation or cell viability. In conclusion, we developed a novel scaffold-free cardiac 3-D culture model with several advancements for the investigation of CM and CF function and cross-regulation.
- three-dimensional culture model
- optical mapping
cardiac myocytes (CMs) and cardiac fibroblasts (CFs) determine the structural, mechanical, and electrical characteristics of the myocardium (8). Both cell types play important roles in cardiac remodeling by altering their size, morphology, signaling, and function. For example, in response to changes in mechanical, chemical, and/or electrical signals, CMs hypertrophy and CFs produce excess extracellular matrix (ECM) (51). Progressive remodeling often leads to cardiac arrhythmias, sudden cardiac death, and heart failure and is a major risk factor for cardiovascular mortality (5, 31).
Investigations of CMs and CFs in two-dimensional (2-D) cell cultures have yielded invaluable information on their biology and regulation. However, cells grown on flat rigid surfaces are exposed to a different environment than in tissue; this can affect cell behavior, as indicated by differences in morphology, differentiation, and cell-cell and cell-matrix interactions compared with cells growing in a more physiological three-dimensional (3-D) environment (56). 3-D cell culture approaches, which can promote the maintenance of molecular properties and facilitate more in vivo-like gene expression, have been developed to bridge the gap between 2-D cell culture and tissue (1, 35).
Spontaneous self-assembly of heart cells into spontaneously beating 3-D structures in culture was first reported in 1971 (21). Since then, a number of 3-D tissue culture approaches have been developed to achieve self-organization, cell-cell interactions, electrical coupling, cell-matrix interactions, and tissue-like morphology in a 3-D environment (for reviews, see Refs. 19 and 39): scaffold-based approaches provide structural templates/surfaces using synthetic polymers and/or natural matrixes (e.g., ECM) for the cells, whereas scaffold-free approaches promote and depend on cellular self-assembly and organization in the absence of external cues. While interactions between cells and external cues dominate in scaffold-based cultures, cell-cell interactions dominate in scaffold-free 3-D models, in which the cells themselves create specific environments (including ECM).
Scaffold-free approaches range from layering of cell sheets formed on temperature-sensitive polymer surfaces (47) to rotational cultures (2), gravity-enforced assembly in hanging drops (28), and nonadhesive surface culture systems (3, 17) for the generation of spherical cardiac microtissues. For the latter systems, there has been little functional characterization to date, and their broad practical application is limited in part due to restrictions in the spheroid numbers that can be generated and in experimental control. Furthermore, the distribution of CMs and CFs when cocultured does not necessarily resemble that in the myocardium (27). Magnetic levitation of cells was recently introduced as a novel approach (48) (albeit not in cardiac cells): it requires a hydrogel consisting of gold, magnetic iron oxide nanoparticles and a filamentous bacteriophage, which introduces nonphysiological components as potential confounding factors.
The goal of the present study was to develop a scaffold-free cardiac culture that could easily and inexpensively create many uniform cardiac microtissues, which mimic the cellular distribution and functional behavior of CMs and CFs in tissue. To that end, we took advantage of a nonadhesive micromolded hydrogel system we recently developed (33, 34): it enables the production of hundreds of self-assembled microtissues, allows for fine control of microtissue size and media exchange, and is suitable for visualization and quantitation. We used neonatal rat ventricular CMs and CFs, alone or in different ratios, to mimic the different cell composition in the myocardium (4). Here, we demonstrate that CMs and CFs self-sort in hydrogels to be interspersed reminiscent of their distribution in ventricular tissue and that the resulting microtissues, depending on their cellular composition, express ECM and Ca2+-handling proteins, have typical inward rectifier K+ current (IK1) density, form functional cell-cell connections [as evidenced by spontaneous action potentials (APs), contractions, and connexin (Cx)43 expression], and are amenable to cell type-selective gene transfer.
MATERIALS AND METHODS
Isolations and Culture of Neonatal CMs and CFs
All animal studies conformed with the Guiding Principles in the Care and Use of Vertebrate Animals in Research and Training, and all experimental protocols involving animals were approved by the Institutional Animal Care and Use Committee of Rhode Island Hospital. CMs and CFs were isolated from 2-day-old Sprague-Dawley rats as previously described (58) with minor modifications. In brief, after an overnight digestion at 4°C with 1 mg/ml trypsin (in HBSS, Sigma, St. Louis, MO), 10 ml DMEM-F-12 (Invitrogen) supplemented with 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin (“serum medium”) was added, and serial digestion was performed with 0.6 mg/ml collagenase type 2 (Worthington, Lakewood, NJ) at 37°C. The cell suspension was filtered through a cell strainer (40 μm, BD Falcon, San Jose, CA) and centrifuged at 870 g for 5 min. Discontinuous Percoll gradient centrifugation of the cell pellet (resuspended in 1× HBSS) yielded CM- and CF-enriched cell fractions that were resuspended in serum medium, counted, and plated in either conventional 2-D cultures (5 × 105 cells/6-well plate) or 3-D cultures (see below). Average yields per pup were 2.3 ± 0.8 × 106 CMs/pup and 2.8 ± 0.7 × 106 CFs/pup (n = 35), with ∼95% of CMs and ∼90% of CFs viable based on trypan blue exclusion. The purity of CM and CF fractions (n = 19 each) was 88 ± 3% and 99 ± 1%, respectively, as determined with cell type-specific markers after 24 h in serum medium in 2-D culture (see below).
In experiments where CMs and CFs were precultured separately at 3 × 106 cells/100-mm dish for 1–2 days in 2-D, a standard trypsinization protocol was used for cell detachment before 3-D culture. Compared with the original cell numbers seeded, 57.7 ± 3.3% of CMs and 50.1 ± 3.0% of CFs (n = 15 each) were recovered after trypsinization. The fraction of viable (trypan blue negative) CMs and CFs was comparable (86 ± 0.4% and 87 ± 0.6%, respectively).
Fabrication of Hydrogels
Fabrication of hydrogels was performed as previously described (34). In brief, computer-assisted design (Solid Works, Concord, MA) was used to create a template of the desired gel features, including a cell seeding chamber, 822 recesses with hemispherical bottoms (800 μm deep × 400 μm wide), and media exchange ports (Fig. 1A). A wax mold was generated with a ThermoJet rapid prototyping machine and then used to generate a negative replicate with Reprorubber, a fast-curing polydimethysiloxane (PDMS) elastomer (Flexbar, Islandia, NY). The negatives were filled with Sylgard 184 PDMS (Dow Corning, Midland, MI) to produce positive replicates, which were washed with 70% ethanol, rinsed with distilled water, and autoclaved before use. Sterilized 2% agarose solution was pipetted into the PDMS molds, and the gel-containing mold was placed in a −20°C freezer for 5 min to harden (but not freeze) the hydrogel, which was separated from the mold, transferred into single wells of six-well plates with 5 ml serum medium, and equilibrated overnight at 37°C in a humidified incubator with a 5% CO2-95% air atmosphere.
Cell Seeding and 3-D Cultures
Before cell seeding, air bubbles that had formed in the recesses of the hydrogel during equilibration were removed using a vacuum chamber (Lindberg/Blue M, Thermo Scientific, Waltham, MA), and medium was aspirated. The desired cells in suspension (1 × 106 cells in 200 μl) were added drop wise to the center of the seeding chamber. Cells were allowed to settle into the individual recesses for 30–45 min, after which 3 ml serum medium was added to each well. Seeded cells were cultured at 37°C for indicated times. Medium was exchanged every other day.
Cell Viability Assessments
Cell viability assessments were performed using the Live/Dead Viability/Cytotoxicity kit (Invitrogen, Carlsbad, CA). Cells were rinsed with PBS and stained with 2 μM calcein-AM and 4 μM ethidium homodimer-1 (in 300 μl PBS) at 37°C for 30 min, followed by image acquisition.
Live Cell Fluorescent Labeling
CMs and CFs were labeled with CellTracker (CT; Invitrogen) orange CMRA (10 μM) and CT green CMFDA (5 μM), respectively, in DMEM/F-12 medium supplemented with 1× insulin-transferrin-sodium selenite (Sigma), 100 U/ml penicillin, and 100 mg/ml streptomycin (termed as serum-free medium) for 30 min at 37°C. After a media change and incubation for additional 2 h, cells were centrifuged at 870 g for 5 min and resuspended in serum medium for subsequent experiments. Both dyes could be monitored for 7 days, and neither one transferred to neighboring cells (data not shown).
Cells in 2-D culture.
Cells in 2-D culture were fixed with 4% formaldehyde for 15 min after two brief PBS washes and permeabilized with 0.1% Triton X-100 in PBS for 15 min. Nonspecific binding sites were blocked with Image-iT FX Signal Enhancer (Invitrogen) for 30 min, followed by an incubation with primary antibodies against α-sarcomeric actinin (α-SA; 1:1,600, Sigma) or vimentin (Vim; 1:100, Sigma) for 1 h and secondary antibodies conjugated to Alexa fluor 488 or Alexa fluor 594 (1:200, Invitrogen) for 1.5 h. Coverslips were mounted with ProLong Gold antifade reagent containing 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). All steps were performed at room temperature.
Microtissues in 3-D culture.
Microtissues in 3-D culture were harvested by gently spinning inverted hydrogels at 500 rpm (63 g) for 5 min, followed by a transfer into a 1.7-ml tube, a wash in PBS, and embedding in OCT compound. Frozen sections (7 μm) were air dried for 10 min and fixed with 4% paraformaldehyde for 10 min, followed by three 5-min washes with 1× wash buffer (DAKO, Carpinteria, CA). Permeabilization (0.1% Triton X-100 in wash buffer, 10 min) and blocking [5% normal goat serum (Sigma) in wash buffer, 60 min] were followed by 1-h incubations each with antibodies directed against α-SA (1:1,500, Sigma, or 1:1,500, Abcam, for double staining), Vim (1:100), and/or Cx43 (1:100, Millipore) and Alexa fluor-conjugated secondary antibodies (see above). Coverslips were mounted with ProLong Gold antifade reagent containing DAPI (Invitrogen). All steps were performed at room temperature. Frozen sections (7 μm) were also stained with hematoxylin and eosin as well as sirius red using standard procedures.
Western Blot Analysis
Cell lysates were obtained from harvested microtissues of the indicated CM and CF composition (see above) by sonicating them for 10 s on ice in 1× lysis buffer (Cell Signaling Technology, Danvers, MA) containing a protease inhibitor cocktail (Roche, Indianapolis, IN), followed by a 5-min centrifugation at 420 g (2,000 rpm) at 4°C. Cell lysates from adult rat ventricular CMs and CFs as well as ventricular homogenates from rat ventricular tissue were used for comparison. Protein was measured using the Bradford microassay (Bio-Rad) with BSA as the standard. Equal amounts of protein (10 μg/lane) were separated by SDS-PAGE (Tris/glycine). After a transfer to nitrocellulose membranes (Protran, Whatman, Piscataway, NJ), gels were stained with GelCode Blue Stain Reagent (Thermo Scientific) to verify complete protein transfer, and the nitrocellulose membranes were stained with Ponceau S to confirm equal loading and even transfer efficiency. Immunoblots were performed using goat polyclonal antibodies against procollagen 1 (sc-8787, 1:500, Santa Cruz Biotechnology) and sarco(endo)plasmic reticulum Ca2+-ATPase 2a (SERCA2a; sc-8094, 1:100, Santa Cruz Biotechnology), a mouse monoclonal antibody against phospholamban (MA3-922, 1:500, Pierce), or rabbit polyclonal antibodies against laminin (BT-594, 1:3,000, Biomedical Technologies), fibronectin (F3648, 1:1,200, Sigma), the L-type Ca2+ channel α1c-subunit (ACC-013, 1:200, Alomone Labs), and calsequestrin (PA1-913, 1:2,500, Pierce). Proteins of interest were visualized by chemiluminescence (SuperSignal West Pico or Femto substrate, Pierce) using horseradish-conjugated secondary antibodies (ImmunoPure, Pierce). Film exposure times varied depending on the abundance of the protein and were adjusted so that the signal stayed within the linear range.
Adenoviral infection could be performed either before or 24 h after cell seeding in the hydrogels. Adenovirus (Ad) encoding green fluorescent protein (Ad-GFP) at a multiplicity of infection (MOI) of 10 was used to visualize gene transfer. For infection before seeding (data not shown), cells in suspension or precultured in 2-D were incubated with the appropriate amount of Ad-GFP in serum-free medium for 2 h or overnight, respectively, after which cells were collected, resuspended in serum medium, and added to the hydrogel seeding chamber. Alternatively, the appropriate amount of Ad in 200 μl serum-free medium was added to the microtissue-containing seeding chamber immediately after medium aspiration, followed by the addition of 3 ml serum medium/well after 2 h. 2-D cultures plated and infected at the same time as 3-D cultures served as controls. Medium was changed every other day. All cultures were imaged daily at the same exposure times for 5 days to visualize GFP expression. On day 5, cell viability was determined with ethidium homodimer-1 from the Live/Dead Viability/Cytotoxicity kit, as described above. For cell type-specific infections in heterotypic microtissues, precultured CMs or CFs were infected with Ad-GFP at a MOI of 10 overnight. The next day, they were trypsinized, counted, and coseeded with cells from the other cell type that had been labeled with CT orange CMRA.
Image Acquisition and Processing
Phase-contrast and epifluorescent images were acquired with a Nikon TE2000-U (Nikon, Mellville, NY) with a cooled Black and White/Color Digital Camera (MicroVideo Instruments, Avon, MA). Confocal images were acquired with a Nikon C1si confocal (Nikon) using diode lasers at 488 and 561 nm. Serial optical sections (12–13 μm) were performed with EZ-C1 computer software (Nikon). Z-series sections were collected at 0.35 μm with a ×20 PlanApo lens and a scan zoom of 2. Each wavelength was acquired separately by invoking frame lambda. Deconvolution and projections were performed using NIS Elements (Nikon) computer software. Time lapse experiments were performed in a humidified chamber with temperature (37°C) and CO2 (5%) control (Zeiss). Bright-field images were acquired using a Carl Zeiss Axio Observer Z1 equipped with an AxioCam MRm camera (Carl Zeiss MicroImaging, Thornwood, NY) and Axiovision software. Images were obtained from the ×2.5 objective with the focal plane manually adjusted every 12 h for maximum clarity. Images were collected at 6-h intervals over 72 h using Axiovision software.
APs from CM microtissues or whole cell IK1 in single dissociated CMs were obtained with an Axopatch-200B amplifier (Axon Instruments, Union City, CA) in Tyrode solution containing (in mM) 140 NaCl, 5.4 KCl, 0.33 NaH2PO4, 1 MgCl2, 1 CaCl2, 5 HEPES, and 7.5 glucose (pH 7.4 at ∼36°C). Standard microelectrode techniques were used for AP recordings; both recording and reference microelectrodes were filled with 3 M KCl. The tip resistance was 15–25 MΩ. IK1 was recorded with a 2-s voltage ramp from −120 to 60 mV with standard whole cell patch-clamp techniques. Whole cell pipette resistances were 2–4 MΩ when filled with the intracellular solution, which contained (in mM) 120 KCl, 5 MgCl2, 0.36 CaCl2, 5 EGTA, 5 HEPES, 5 glucose, 5 K2-ATP, 5 Na2-CrP, and 0.39 Tris-GTP (pH 7.2). IK1 was isolated as the difference between the currents before and after application of 100 μM Ba2+ and normalized to the capacitance of the CM.
Optical Mapping of Membrane Potential
3-D microtissues were collected via centrifugation of overturned hydrogels and incubated with 7–10 μM of voltage-sensitive RH237 dye (Invitrogen) for 10 min at 37°C. Glass-bottom 35-mm culture dishes (to accommodate the short working distance requirements of high numerical aperture Nikon fluor ×20 or ×40 objectives) were mounted on a temperature-controlled chamber (Dual Automatic Temperature Controller TC-344B, Warner Instrument) to maintain 35 ± 1°C and constant perfusion with (in mM) 140 NaCl, 5.1 KCl, 1 MgCl2, 1 CaCl2, 0.33 NaH2PO411). The local conduction velocity (CV) was measured by calculating the local gradient of activation time (using 7 × 7 Gaussian windows), and the mean value of local CVs was used to represent the CV of the spheroid (44). Fluorescence could be recorded up to 8 s without further deterioration of the signal amplitudes and phototoxic damages.Vm) polarities during stimulation for optimization (typically 5–13 V/cm, 4-ms duration, 1 Hz). Fluorescence images were acquired on a Nikon Eclipse Ti-S microscope connected to a xenon arc lamp (Cairn) through a μStep filter (Ionoptix) and a high-speed CMOS camera (100 × 100 pixels, Ultima-L, Scimedia) with a frame rate of up to 5,000 frames/s at 2.5 × 2.5-μm2 pixel resolution with a ×40 objective. A custom filter set (Omega Optical) was used to maximize excitation and emission light collection. The activation time points at each site were determined from fluorescence signals by calculating the maximal change in fluorescence over time. AP durations (APDs) at the indicated times of relaxation were measured in each pixel, and APD maps were generated using a custom-built software package (
Data from representative assays are shown and are expressed as means ± SD for n determinations unless otherwise indicated. Statistical differences were assessed by an unpaired two-tailed Student's t-test and two-way ANOVA for comparison of individual means. P values of <0.05 were considered statistically significant.
The goal of the present study was to develop a cardiac 3-D culture model that mimics both the cellular distribution and functional behavior of CMs and CFs in tissue and can be easily generated. To that end, we used a nonadhesive micromolded hydrogel system, in which cell recesses were located 1.2 mm above the bottom of the agarose hydrogel, allowing imaging with an inverted microscope (33, 34). We (13) have previously shown that the recess size and shape chosen for this study (800 × 400 μm, with a hemispherical bottom) allow for sufficient diffusion of nutrients. Upon seeding, cells randomly distributed in the seeding chamber, settled into the recesses by gravity, and coalesced into spherical microtissues (also called spheroids) at the bottom of each recess without attaching to the hydrogel.
Self-Assembly of CMs and CFs
CMs or CFs (seeded at 1 × 106) aggregated over several days; the spheroids reached a minimum diameter within 4–5 days (Fig. 2A,a). Whereas CM spheroids were mostly rounded with few excluded cells, CF spheroids were less rounded with more cells excluded. Spheroids containing a mixture of CMs and CFs at varying ratios (3:1, 1:1, and 1:3) showed a gradual transition in overall morphology (Fig. 2A,b). Importantly, Live/Dead staining indicated that the vast majority of cells remained viable after 7 days in cultures across all conditions, as shown by epifluorescence (Fig. 2A,b) and confocal microscopy (Fig. 2A,c).
We next examined whether the number of excluded and nonviable cells could be reduced by culturing the cells in 2-D culture, removing the unattached cells with a medium exchange after 24 h, and then trypsinizing, counting, and seeding the cells into hydrogels (“preculturing”). Precultured CMs and CFs showed accelerated microtissue formation compared with freshly isolated cells, as evident from representative images (Fig. 2B) and quantitative analysis of the decrease in spheroid area over time (Fig. 2C). The number of excluded cells (particularly in CF-containing cultures) was markedly reduced. Preculturing did not affect the viability of resulting spheroids, and, when plated in 2-D for control, cells showed normal morphology and attached well to the polystyrene substrate (Fig. 2B,b). Furthermore, preculturing improved the shape of spheroids, as indicated by an increase in circularity, which was most pronounced for CF-containing microtissues (Fig. 2D). In addition, the ability to maintain cell populations in 2-D before seeding them in 3-D cultures enabled subsequent investigations of the plasticity of microtissues in sequential self-sorting experiments (see below). Figure 3 shows that hundreds of microtissues that were uniform in size and shape could be generated within each hydrogel.
CM and CF Distribution in Microtissues
Figure 4A shows hematoxylin-eosin and sirius red staining and immunostaining for CM- and CF-specific proteins (i.e., α-SA and Vim, respectively) in representative microtissues that were generated by seeding of the indicated cell fractions. As expected, α-SA staining was predominantly observed in microtissues generated from CMs (either seeded alone or in combination with CFs). Very few CMs were detected in CF microtissues, consistent with a very high purity of the CF fraction (see materials and methods). Consistent with a slightly lower purity of CM fractions, a few more Vim-positive cells were observed in CM microtissues.
We then asked whether the two cell types evenly mingle in heterotypic microtissues or segregate toward different areas (e.g., the center vs. the outer rim) when seeded simultaneously (1:1 ratio). The expression pattern of α-SA and Vim (Fig. 4A) suggested a distribution of both CMs and CFs throughout the microtissue, which was further demonstrated by fluorescent double staining of cryosections (Fig. 4B). The observed spatial intermingling of both cell types was reminiscent of their distribution in the myocardium (8, 30).
Rearrangement of CMs and CFs in Microtissues
We next examined self-sorting of CMs and CFs when they were seeded sequentially (i.e., the second cell type was seeded on preformed microtissues composed of the first cell type). Labeling of CM and CF fractions with different CT dyes enabled monitoring of their respective locations within spheroids over time. First, CT orange-labeled CMs were allowed to form microtissues for 2 days. CFs, which had been cultured in conventional 2-D culture in parallel for 2 days, were then trypsinized, labeled with CT green, and then added to the seeding chamber (Fig. 5A). Images taken within 30 min after cell seeding showed loosely arranged CT green-labeled CFs surrounding a core of CT orange-labeled CM. Within another 2 days, CFs did not surround a CM core, as one could have predicted, but both cell types were highly intermingled (see confocal images of a microtissue midsection; Fig. 5A, right). This observation suggests that CMs and CFs can change their location within the microtissue during culture. Trypsinized CT green-labeled CFs plated alone as controls (see representative image in Fig. 5A, left), along with nonlabeled trypsinized CFs (data not shown), showed that CT labeling did not affect cell aggregation. Cell viability was also preserved, as shown by the absence of Live/Dead staining (red) in CT green-labeled CF microtissues (data not shown).
The reverse experiments, in which trypsinized CT orange-labeled CMs were seeded 2 days after microtissues composed of CT green-labeled CFs had formed, showed similar results: CMs and CFs were fully interspersed 2 days after CF addition (Fig. 5B), and CT labeling did not affect CM aggregation (or viability; data not shown). Together, these findings suggest that microtissues retain the ability to reassemble after initial spheroid formation, indicating a high degree of plasticity. Similar results were obtained when microtissues were allowed to form for 4 days before the second cell type was seeded (data not shown).
ECM and Ca2+-Handling Protein Expression in Cardiac Microtissues
While no exogenous ECM proteins were added in this model, laminin, fibronectin, and procollagen-1 were detected in both CM and CF microtissues (Fig. 6, top). As expected, their expression was less pronounced in CM microtissues compared with CF microtissues and enhanced in microtissues containing both cell types. Figure 6, middle, shows that microtissues assembled in our hydrogel system express Ca2+-handling proteins, including SERCA2a, phospholamban, calsequestrin, and the α1c-subunit of the L-type calcium channel.
Functionality of Cardiac Microtissues
Spontaneous APs could be recorded from beating CM microtissues using microelectrodes (Fig. 7A); the maximum diastolic Vm was −66 ± 7 mV (n = 4). Patch-clamping experiments demonstrated the functional expression of Ba2+-sensitive IK1 in single CMs dissociated from CM microtissues 2 days after assembly (Fig. 7B). The amount of Ba2+-sensitive IK1 was variable among different CMs, with an average of 15.0 ± 10.9 pA/pF at −120 mV (n = 17 of 22 cells; the remaining 5 cells tested showed little or no IK1).
Spontaneous coordinated contractions in CM microtissues (see the Supplemental Material, Supplemental Movie S1, for an example)1 further suggested the formation of functional cell-cell connections. We therefore examined Cx43 expression in CM and CF microtissues that were double stained with α-SA or Vim antibodies: Cx43 staining was observed in CM microtissues as early as 1 day after seeding, and a comparable expression pattern was observed 5 days after seeding (Fig. 8A). Cx43 and α-SA colocalized, as evident from the merged images. In contrast, CFs were not Cx43 positive, as shown by the virtual absence of Cx43 staining in CF microtissues at 1 or 5 days after seeding (Fig. 8B). The very few Cx43-positive cells detected in CF microtissues are likely CMs, as suggested by the colocalization of α-SA and Cx43 staining in CF microtissue (Fig. 8A).
We next used optical mapping to characterize APs, including the rate of rise, duration, dispersion, and its propagation, as important functional parameters of cardiac microtissues in more detail. Pacing cycle lengths of up to 2.5 Hz at 37°C in some instances were observed in spontaneously contracting microtissues and often changed during experiments. We selected spheroids contracting slower than 1 Hz and typically paced spheroids at 1 Hz in the culture dish, so that spheroids could be paced at the same cycle length for data analysis. Figure 9B shows representative traces from a spontaneously beating (left) and an electrically paced CM microtissue (right): whereas AP shape and duration differed slightly (see the slow vs. sharp rise of the AP upstroke, respectively), the triangular shape was overall highly characteristic of rat ventricular CMs. Under pacing, APD at 50% and 90% repolarization were 33 ± 7 and 100 ± 30 ms, and CV was 18.0 ± 1.9 cm/s (n = 5 each). Figure 9C and Supplemental Movie S2 show a phase-contrast image of a representative microtissue with a nearby microtip iridium electrode and AP propagation upon electrical stimulation, respectively. The higher sampling rate (5,000 frames/s) and signal quality allowed detailed maps of activation (Fig. 9D) and APD (Fig. 9E). During field stimulation, current flow through the intracellular and extracellular space caused hyperpolarization near the anode and depolarization near the cathode, which can be seen in the intact heart and is known as virtual electrode phenomenon (15) (Fig. 9F).
When we compared microtissues comprised of a mixture of CMs:CFs (1:1) or CMs, we observed that the presence of CFs was associated with AP prolongation in CMs, as evident by a 1.6-fold rise in APD at 75% repolarization (Fig. 9G). This rise in APD was further enhanced (2.2-fold) when the microtissues were subjected to continuous electrical field stimulation (CES) for 24 h before optical mapping, along with a prolongation of APD in CM microtissues only.
Effective and Cell Type-Selective Gene Transfer
Ads are widely used reagents for gene transfer, particularly in cell types that cannot easily be transfected with lipid-based plasmid DNA transfection techniques. We asked if Ads can be used for efficient, uniform, and cell type-specific gene transfer in this model. GFP expression was detectable within 1 day in CM, CM:CF, and CF microtissues and reached maximum expression levels 2 days after infection (Fig. 10A). Spheroid formation and morphology were not altered compared with uninfected control cultures from the same isolation (data not shown). Ad-mediated gene transfer could also be achieved by infecting cardiac cells before seeding (data not shown). Ethidium homodimer-1 staining, 4 days after infection, showed very few nonviable cells regardless of the time of infection (data not shown), indicating high cell viability. Importantly, cell type-selective gene transfer could be achieved, as demonstrated by the lack of colocalization when CT orange-labeled CMs were coseeded with Ad-GFP-infected CFs (and vice versa; Fig. 10B).
Using nonadhesive micromolded hydrogels, we developed a cardiac 3-D culture model, in which >800 homotypic microtissues (i.e., CMs or CFs plated separately) or heterotypic microtissues (i.e., CM:CF mixture) can be formed in a six-well plate with ease, analyzed functionally, and subjected to cell type-specific gene manipulation. The major findings are as follows. First, both freshly isolated and precultured neonatal ventricular cells self-assembled (albeit with a different time course) into viable spherical microtissues that expressed endogenously produced ECM proteins. Second, CMs and CFs self-sorted to be highly interspersed when they were seeded both simultaneously or sequentially, indicating a high degree of plasticity of the cardiac microtissues. Third, Ba2+-sensitive IK1 and Ca2+-handling proteins, including SERCA2a, were detected in CM-containing microtissues. Fourth, AP initiation and propagation as well as rhythmic contractions of CM-containing microtissues indicated the formation of functional cell-cell connections. Fifth, Cx43 expression was detectable in CMs but not CFs as early as 1 day after cell seeding. Sixth, the presence of CFs led to a prolongation of APD, which was more pronounced in paced microtissues. Finally, cell type-specific gene expression uniformly across the microtissue was achieved via Ad-mediated gene transfer, paving the way for future in vitro investigations of cardiac cell behavior and their interactions in a 3-D environment.
CM and CF Self-Assembly in Nonadhesive Hydrogels
The system we used creates hundreds of uniform cardiac microtissues, whose size was determined by recess dimensions and seeding density (data not shown). The microtissues generated were compatible with diffusion limits for O2 and nutrients [∼150 μm (12, 39)] when plated at up to ∼1,800 cells/recess (∼1.5 × 106 cells/gel; data not shown). Consistent with previous studies (17, 27), ECM proteins (such as fibronectin, laminin, and procollagen-1) were expressed in cardiac microtissues. Enhanced expression in microtissues containing both cell types (compared with CM or CF microtissues) support the notion of cross-regulation between the two cell types with regard to ECM production [for example, from CMs to CFs (45)].
Freshly isolated CMs and CFs self-assembled into viable microtissues within days, whereas precultured cells, in particular of CFs, showed markedly accelerated cell aggregation and more regular microtissue shapes. Importantly, microtissues formed from precultured cells did not show any diminution in cell viability, formation of cell-cell connections, or functional behavior in this study. AP recordings could be obtained from the entire field of view for the mapping of conduction and functional experiments.
CM and CF Self-Sorting in 3-D Microtissues
A key finding of this study is that CMs and CFs fully interspersed when seeded simultaneously, because this interspersion is reminiscent of their distribution in the myocardium (8, 30). This was not the only possible outcome. For example, when human fibroblasts and umbilical vein endothelial cells are seeded together, they self-sort into multilayered spherical microtissues, with a fibroblast core enveloped by an endothelial cell layer (33). Interestingly, when hanging drops were used to form cardiac microtissues, CFs formed a core, whereas CMs moved toward the outside (27), which is different from the interspersion we observed. The reason is not clear but may include differences in the experimental approach (hanging drops vs. hydromolds). Importantly, CMs and CFs also spatially intermingled when they were plated sequentially in our hydrogels, indicating plasticity of the microtissue assembly.
Functionality of Cardiac Microtissues
AP and rhythmic contractions indicate functional cell-cell connections and Ca2+ handling. Cardiac microtissues that were generated in hanging drops have been reported to have little or no SERCA2a expression (3), which could limit subsequent functional investigations. In contrast, we observed robust expression of SERCA2a and demonstrated the expression of other Ca2+-handling proteins in our model. Cx43 was detected in CM microtissues as early as 1 day after cell seeding, which is not surprising because Cx43 has been widely reported to form gap junctions between CMs in conventional 2-D culture and myocardial tissue (for a review, see Ref. 25) as well as other cardiac 3-D culture models (7). Interestingly, Cx43 was virtually not detectable in CF microtissues and did not colocalize with Vim-positive cells that were interspersed at low levels in CM microtissues, suggesting that Cx43 does not participate in gap junction formation between CF-CFs or CF-CMs in our 3-D microtissues. This observation contrasts with previous reports from conventional and micropatterned 2-D cultures (18, 32, 36, 43) but is consistent with observations in the intact myocardium, where, with the exception of the sinoatrial node (9), gap junctional coupling of CMs and CFs is generally not observed (14).
The microtissues generated in this study provided a good substrate for cell-cell interactions in 3-D, as indicated by the fact that AP shape, duration, and conduction were overall comparable with those of rat ventricular tissue using complementary approaches. Microelectrodes were used to record APs by impalement of single cells on the spheroid. The resting Vm was consistent with neonatal rat myocytes that had been either freshly isolated (20), cultured in 2-D monolayers (7, 42), cultured as scaffolded 3-D constructs (7, 59), or present in a tissue slice (7). The patch-clamp technique was used to measure IK1 from CMs dissociated from a CM microtissue. The amount of Ba2+-sensitive IK1 was comparable with previously reported IK1 in freshly isolated neonatal rat CMs (29, 54).
Optical mapping of APD and AP propagation with a voltage-sensitive dye has superior signal quality in 3-D compared with monolayer cultures. Unlike 2-D micropatterned cultures, 3-D spheroids provide several layers of cells. In addition, fluorescence recordings can give a high signal-to-noise ratio with relatively low excitation light, reducing phototoxic damage and allowing a longer time recording and high sampling rate. In this study, fluorescence signals from a 3-D spheroid could be achieved with >5,000 frames/s, which is critical to measure virtual electrode phenomena during field stimulation and AP propagation right after the end of the stimulation. CV was measured for wavefronts arising from unipolar electrode stimulation. The CV we measured in CM microtissues exceeded the CV reported for a 3-D neonatal rat CM construct on a collagen scaffold [8–14 cm/s (38)], was within the range reported for anisotropic 2-D CM cultures [10–20 cm/s (6)], and slightly lower than in the neonatal intact ventricle [25-cm/s range (50)]. The APD was within the range reported for a neonatal rat ventricular tissue slice [APD at 90% repolarization of ∼105 ms (7)] and slightly shorter than in monolayer 2-D cultures [APD at 80% repolarization of ∼118 ms (24)] and scaffolded 3-D cultures [APD at 90% repolarization of ∼148 ms (59) or ∼220 ms (7)]. APD can shorten with time in culture (42), which was reduced in our model to 2–3 days by the incorporation of a brief 2-D preculture step.
Effect of CFs on APD
The presence of CFs was associated with AP prolongation in CMs. This prolongation was further enhanced by CES before optical mapping, which exposes the cells to more in vivo-like conditions (10). APD values can increase or decrease depending on CF resting Vm, density, and degree of electrotonic coupling to CMs (53). During early repolarization, it is possible that CF coupling might dissipate transient outward K+ current, causing slow repolarization, thereby prolonging APD as observed (55). A paracrine effect may contribute as well, since conditioned CF medium has been shown to markedly prolong APD (271%) as well (37). Alternatively, APD can be modulated by CFs through mechanoelectric feedback (52). Furthermore, CES has been shown to improve the structure and function of scaffold-based 3-D cultures of neonatal CMs (38, 40) and to counteract intrinsic APD shortening in unpaced conventional 2-D cultures (46).
Manipulation of Gene Expression Using Ad
Mechanistic and therapeutic experiments frequently involve gain and loss of function approaches, which require efficient gene or short interfering RNA transfer. In conventional 2-D cultures, Ad-mediated gene transfer has been widely used in CMs and CFs to modulate short-term gene expression [e.g., CMs (22, 58) and CFs (57)]. As a proof of concept, we demonstrate in this study that GFP can be delivered and expressed when the cells are transduced before or after spheroid formation, without an impact on cell viability, aggregation, or morphology. GFP expression was observed throughout the spheroid when Ad was added to microtissues, which is an advancement to an existing report (27) in which lentiviral transduction of cardiac microtissues in hanging drops led to GFP gene transfer that was restricted to the periphery of the spheroids. Importantly, we demonstrate, for the first time, cell type-selective gene transfer in cardiac microtissues, which opens the door for future mechanistic studies (see below).
Benefits of Cardiac Microtissues Created in Nonadhesive Hydrogels
We have developed a novel cardiac 3-D model using nonadhesive micromolded agarose gels, in which monodispersed neonatal heart cells self-assemble to form 3-D microtissues in a scaffold-free environment. In contrast to scaffold-based systems, in which cell-ECM interactions dominate (39), cell-cell interactions are maximized in self-assembling scaffold-free cultures. Our approach shares features with, and has several advantages over, existing approaches, such as rotational cultures, gravity-enforced assembly in hanging drops, and nonadhesive surface culture systems. The key advantages are 1) the ability to create up >800 uniform microtissues simultaneously, which far exceeds the numbers reported for nonadhesive cultures (17) and hanging drops (27); 2) the ability to control spheroid size by varying recess size and cell density (rotational cultures rely on the random collision of cells, which is much more difficult to control); 3) the facilitation of media exchange, which is very difficult in the early stages of hanging drops and rotational cultures; and 4) the ability to obtain high-quality images of microtissues over the entire time in culture, which is not possible in hanging drops and rotational cultures.
Conclusions and Potential Future Applications
Taken together, nonadhesive micromolded hydrogels can be used to easily generate large numbers of viable cardiac microtissues of the desired cell composition and in vivo-like CM and CF interspersion and functional behavior. The microtissues respond to electrical stimulation and can be subjected to gene transfer and easily harvested for subsequent optical mapping, electrophysiological, molecular, biochemical, and morphological/pathological analyses. This model will be useful for mechanistic and pharmacological investigations of heart cell behavior and their interactions in a controllable, tissue-like 3-D environment. Some examples are as follows. First, seeding of different CM and CF ratios can mimic the cell composition in the myocardium, which varies among species and with age (4). Second, sandwiching a CF microtissue between CM microtissues mimics a larger area of CFs (such as in scar formation) that separates CMs. Successful fusions have been reported for microtissues composed of noncardiac cells (41) as well as cardiac cells (27). Third, cell type-selective gene transfer or silencing can be used to modulate signaling pathways known to play a pathophysiological role in cardiac remodeling. Fourth, mouse heart cells, which are capable of forming 3-D microtissues (27), can be used from genetically modified models to conduct mechanistic studies in the absence of Ad-mediated gene transfer or silencing. Fifth, incorporation of cells from embryonic and adult animals could probe for maturation-dependent effects (23). Finally, CFs from diseased hearts, which maintain their phenotype in culture for several passages (16, 26, 49), can be used to investigate cellular mechanisms leading to reactive scar formation after myocardial infarction or reactive interstitial fibrosis in response to pressure overload and other stimuli in a 3-D environment.
This work was supported by National Heart, Lung, and Blood Institute Grant HL-80127 (to U. Mende), by American Heart Association (AHA) Grant 0740098N (to U. Mende), by undergraduate teaching and research assistant awards from Brown University (to B. R. Desroches and N. Nath), and by an AHA undergraduate summer internship (to N. Nath).
J. R. Morgan has an equity interest in MicroTissues, Incorporated. The other authors have no conflicts of interest to declare.
Author contributions: B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., A.R., G.L., N.N., and K.M.H. performed experiments; B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., G.L., B.Y., and U.M. analyzed data; B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., G.L., and U.M. interpreted results of experiments; B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., G.L., B.Y., and U.M. prepared figures; B.R.D., P.Z., B.-R.C., and U.M. drafted manuscript; B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., G.L., G.K., J.R.M., and U.M. edited and revised manuscript; B.R.D., P.Z., B.-R.C., M.E.K., A.E.M., W.L., A.R., G.L., N.N., K.M.H., B.Y., G.K., J.R.M., and U.M. approved final version of manuscript; P.Z., B.-R.C., J.R.M., and U.M. conception and design of research.
The authors are grateful to Virginia Hovanesian from the Core Research Laboratories at Lifespan (Director: Dr. Paul N. McMillan) for the acquisition of the confocal images shown.
↵1 Supplemental Material for this article is available at the American Journal of Physiology-Heart and Circulatory Physiology website.
- Copyright © 2012 the American Physiological Society