Electrical stimulation directs engineered cardiac tissue to an age-matched native phenotype

Abstract

Quantifying structural features of native myocardium in engineered tissue is essential for creating functional tissue that can serve as a surrogate for in vitro testing or the eventual replacement of diseased or injured myocardium. We applied three-dimensional confocal imaging and image analysis to quantitatively describe the features of native and engineered cardiac tissue. Quantitative analysis methods were developed and applied to test the hypothesis that environmental cues direct engineered tissue toward a phenotype resembling that of age-matched native myocardium. The analytical approach was applied to engineered cardiac tissue with and without the application of electrical stimulation as well as to age-matched and adult native tissue. Individual myocytes were segmented from confocal image stacks and assigned a coordinate system from which measures of cell geometry and connexin-43 spatial distribution were calculated. The data were collected from 9 nonstimulated and 12 electrically stimulated engineered tissue constructs and 5 postnatal day 12 and 7 adult hearts. The myocyte volume fraction was nearly double in stimulated engineered tissue compared to nonstimulated engineered tissue (0.34 ± 0.14 vs 0.18 ± 0.06) but less than half of the native postnatal day 12 (0.90 ± 0.06) and adult (0.91 ± 0.04) myocardium. The myocytes under electrical stimulation were more elongated compared to nonstimulated myocytes and exhibited similar lengths, widths, and heights as in age-matched myocardium. Furthermore, the percentage of connexin-43-positive membrane staining was similar in the electrically stimulated, postnatal day 12, and adult myocytes, whereas it was significantly lower in the nonstimulated myocytes. Connexin-43 was found to be primarily located at cell ends for adult myocytes and irregularly but densely clustered over the membranes of nonstimulated, stimulated, and postnatal day 12 myocytes. These findings support our hypothesis and reveal that the application of environmental cues produces tissue with structural features more representative of age-matched native myocardium than adult myocardium. We suggest that the presented approach can be applied to quantitatively characterize developmental processes and mechanisms in engineered tissue.

Full text

Journal of Tissue Engineering
3(1) 2041731412455354
© The Author(s) 2012
Reprints and permission: sagepub.
co.uk/journalsPermissions.nav
DOI: 10.1177/2041731412455354
tej.sagepub.com
Introduction
Establishing hallmarks of the native myocardium in engi-
neered tissue is essential for creating functional tissue that
can serve as a surrogate for in vitro testing or the eventual
replacement of diseased or injured myocardium.1
Quantitative measures of structural and functional tissue
characteristics form a technical cornerstone for the devel-
opment and testing of engineered cardiac tissue. Native tis-
sue is complex and exhibits a three-dimensional (3D)
multicellular structure and function. This 3D microenviron-
ment has profound effects on the properties, behaviors, and
functions of resident cells.1–3 Furthermore, native tissue
exhibits astonishing variation in the quantity, density, and
morphology of cardiac cells during development, among
species, between tissue types and in disease states.4–6 Most
Electrical stimulation directs engineered cardiac
tissue to an age-matched native phenotype
Richard A Lasher1, Aric Q Pahnke1, Jeffrey M Johnson1,
Frank B Sachse1,2 and Robert W Hitchcock1
Abstract
Quantifying structural features of native myocardium in engineered tissue is essential for creating functional tissue that
can serve as a surrogate for in vitro testing or the eventual replacement of diseased or injured myocardium. We applied
three-dimensional confocal imaging and image analysis to quantitatively describe the features of native and engineered
cardiac tissue. Quantitative analysis methods were developed and applied to test the hypothesis that environmental
cues direct engineered tissue toward a phenotype resembling that of age-matched native myocardium. The analytical
approach was applied to engineered cardiac tissue with and without the application of electrical stimulation as well as
to age-matched and adult native tissue. Individual myocytes were segmented from confocal image stacks and assigned
a coordinate system from which measures of cell geometry and connexin-43 spatial distribution were calculated. The
data were collected from 9 nonstimulated and 12 electrically stimulated engineered tissue constructs and 5 postnatal
day 12 and 7 adult hearts. The myocyte volume fraction was nearly double in stimulated engineered tissue compared
to nonstimulated engineered tissue (0.34 ± 0.14 vs 0.18 ± 0.06) but less than half of the native postnatal day 12 (0.90 ±
0.06) and adult (0.91 ± 0.04) myocardium. The myocytes under electrical stimulation were more elongated compared to
nonstimulated myocytes and exhibited similar lengths, widths, and heights as in age-matched myocardium. Furthermore,
the percentage of connexin-43-positive membrane staining was similar in the electrically stimulated, postnatal day
12, and adult myocytes, whereas it was significantly lower in the nonstimulated myocytes. Connexin-43 was found
to be primarily located at cell ends for adult myocytes and irregularly but densely clustered over the membranes
of nonstimulated, stimulated, and postnatal day 12 myocytes. These findings support our hypothesis and reveal that
the application of environmental cues produces tissue with structural features more representative of age-matched
native myocardium than adult myocardium. We suggest that the presented approach can be applied to quantitatively
characterize developmental processes and mechanisms in engineered tissue.
Keywords
Tissue engineering, confocal microscopy, structural modeling, cardiac muscle, cardiac cell
1 Department of Bioengineering, University of Utah, Salt Lake City, UT,
USA
2 Nora Eccles Harrison Cardiovascular Research and Training Institute,
University of Utah, Salt Lake City, UT, USA
Corresponding author:
Robert W Hitchcock, Department of Bioengineering, University of
Utah, Building (SMBB) Room 4509, 36 S. Wasatch Drive, Salt Lake City,
UT 84112, USA.
Email: r.hitchcock@utah.edu
455354TEJ3110.1177/2041731412455354Journal of Tissue EngineeringLasher et al.
2012
Article

2 Journal of Tissue Engineering 3(1)
engineered cardiac tissue aims to replicate left ventricular
myocardium, which is heterogeneous and composed of
densely packed myocytes, fibroblasts, and other cell types.
Fibroblasts account for the majority of the cells in the
heart and play important roles in normal cardiac function
and disease.7,8 Although myocytes only account for 20%–
40% of the cells that make up cardiac tissue, they occupy
approximately 80%–90% of the tissue volume and are the
contractile cells solely responsible for pump function.9,10
Alterations in myocyte geometry and structure are known
to occur during development and in disease states.11–13
Myocyte structures that are critical for cardiac function
include sarcomeres and gap junctions. Sarcomeres, the fun-
damental unit of contraction, occupy a large fraction of the
intracellular volume and are highly aligned in healthy myo-
cytes. Gap junctions allow for rapid electrical signaling
between myocytes necessary for synchronous cardiac con-
traction. Connexin-43 (Cx43), the predominant isoform of
gap junction channels in ventricular myocytes,14,15 has a
half-life of 2 h. The continuous turnover allows Cx43 to
redistribute along the cell surface in response to environ-
mental conditions.16,17 The distribution of Cx43 is known to
vary during development and in disease states.18,19 For
example, in rat cardiac tissue, Cx43 redistributes in
response to tissue maturity. In neonatal tissue, Cx43 clus-
ters are found to be distributed over the myocyte mem-
brane. As the tissue matures, Cx43 slowly becomes
organized and at approximately 90 days after birth concen-
trates at the cell ends (i.e. polarized).18 Gap junctions also
remodel due to disease. For example, as human cardiac
hypertrophy progresses into heart failure, Cx43 expression
decreases and accumulates at the lateral sides of the myo-
cytes instead of the ends (i.e. lateralized).4,14,20 Gap junc-
tions can be coerced to rearrange in vitro. A recent study in
2D monolayers of neonatal rat myocytes indicated polari-
zation of Cx43 localization by stretching.21 The functional
importance and dynamic nature of Cx43 makes it a target
for analysis, and these types of responses may indicate
some level of control over engineered cardiac tissue.
Several approaches have been developed to produce 3D
engineered cardiac tissue, including seeding preformed
scaffold materials with cells,22 entrapping cells in a 3D
environment,23 stacking cell sheets,24 and decellularizing
and recellularizing tissue25 (reviewed in detail in Refs26,27).
The application of electrical stimulation,22,28,29 mechanical
stimulation,30–32 or perfusion33 has been shown to aid in the
tissue development. To investigate the structure of these
engineered tissues, most reported methods rely on qualita-
tive interpretation of the 2D images. A more comprehen-
sive analysis of structure can be accomplished through 3D
confocal microscopy.34,35 Confocal microscopy is based on
fluorescent labeling and has the ability to control the depth
of field (slice resolution of <1 µm), reject out-of-focus
light, and collect sequential optical sections from thick
specimens.36,37 The application of 3D confocal imaging to
quantitatively characterize structure has not been widely
performed on engineered tissue.
The hypothesis of this study is that the application of
environmental cues directs engineered tissue toward a phe-
notype resembling that of age-matched native myocardium.
The hypothesis was tested by applying 3D confocal imag-
ing and image analysis to characterize hallmarks of cardiac
tissue, including myocyte geometry and spatial distribution
of Cx43, in engineered cardiac tissue with and without the
application of electrical stimulation. The results of the
study support our hypothesis and reveal that the application
of environmental cues produces tissue with structural fea-
tures resembling age-matched native myocardium as
opposed to adult tissue.
Methods
Cell isolation
All animal procedures were performed in accordance with
an approved protocol by the University of Utah Institutional
Animal Use and Care Committee. Ventricular cardiac cells
were harvested from 1-day old Sprague Dawley rats
(Charles River, MA) using a protocol and supplies from
Worthington Biochemical (Lakewood, NJ). Briefly, hearts
were aseptically removed and collected in calcium- and
magnesium-free Hank’s balanced salt solution. Atria were
removed, and the ventricles were finely minced and
digested in 50 µg/mL trypsin at 4°C overnight. Further
digestion was performed the following day with colla-
genase (1500 units) in Leibovitz L-15 media. Cell suspen-
sions were triturated, filtered, centrifuged, and resuspended
in culture medium. Culture medium was made following
the methods described by Hansen et al.38 using Dulbecco’s
modified Eagle’s medium (DMEM) F12 (Thermo Fisher
Scientific, Waltham, MA), 10% equine serum (Thermo
Fisher Scientific), 2% chick embryo extract (Gemini
Bioproducts, West Sacramento, CA), 50 µg/mL human
insulin (Sigma–Aldrich, St. Louis, MO), 2 mM l-glutamine
(Thermo Fisher Scientific), 20 U/mL penicillin (MP
Biomedicals, Solon, OH), 50 µg/mL streptomycin (MP
Biomedicals), 63 µg/mL tranexamic acid (Sigma–Aldrich),
and 33 µg/mL aprotinin (Sigma–Aldrich).
Sample preparation and culture
Fibrin-based engineered tissue samples were fabricated
using methods described by Hansen et al.38 Briefly, a recon-
stitution mixture was prepared on ice comprising of 4.1 ×
106 cells/mL, 5 mg/mL bovine fibrinogen (Sigma–Aldrich),
and 100 µL/mL Matrigel (BD Biosciences, San Jose, CA).
For each sample, 485 µL of reconstitution mixture was
mixed with 15 µL thrombin (100 U/mL; Sigma–Aldrich)
and transferred to a custom mold (Figure 1). The custom
mold was contained in a Petri dish and consisted of a

Lasher et al. 3
Delrin® (McMaster-Carr, Los Angeles, CA) housing and
base, each containing two neodymium magnets (Applied
Magnets, Plano, TX), which allowed for easy coupling and
uncoupling of the mold and base. The housing had a center
channel 4.8 mm in width and 20 mm in length with 6.35
mm holes centered with the silicone posts, and contained
cylinder-shaped (1.6 mm diameter × 6.4 mm length) mag-
nets. The base was 34 mm × 20 mm and contained disk-
shaped (4.8 mm diameter × 1.6 mm thick) magnets that
aligned with the housing. Rectangular frames (34 × 12 mm)
were cut from 0.30-mm-thick polyester sheets (Mylar®;
Fralock, Valencia, CA) using a cutting plotter (Graphtech
FC7000, Irvine, CA) and AutoCAD (San Rafael, CA), and
sandwiched between the housing and base. Frames had a
rectangular center (10 mm × 4.8 mm) and two 4-mm
through holes spaced 26 mm apart (center to center).
Silicone rods (2 mm diameter × 7 mm length) were fabri-
cated from a platinum-cured silicone elastomer (VST-50;
Factor II, Lakeside, AZ) and attached to either side of the
frame window (spaced 12 mm center to center). The sili-
cone posts served to suspend the fibrin-based gel (Figure
1(d)). Samples were allowed to polymerize at 37°C for 90
min. After 30 min of polymerization, 500 µL of the culture
medium was added to keep the sample hydrated and to aid
in removal of the mold from the tissue sample. The frame
was cut on both sides, and the sample was elongated by
40% and secured with nylon screws into a custom bioreac-
tor consisted of two Petri dishes outfitted with carbon rods
spaced 2 cm apart for electrical stimulation (Figure 1(c)).39
Engineered tissue samples were precultured for 3 days
before onset of electrical stimulation.22 Following precul-
ture, samples were subjected to electrical field stimulation
(2 ms symmetric biphasic square pulses, 4 V/cm, 1 Hz) for
9 days. Nonstimulated samples served as controls for
stimulated samples. Bright field images of central regions
of the engineered tissue samples were obtained at days 3, 6,
9, and 12 of culture. The diameter was measured, and the
cross-sectional area was estimated assuming a cylindrical
cross section. The percent decrease in sample size was cal-
culated normalized to the start of the stimulation, that is,
day 3 of culture. At the end of culture, samples were fixed
with 4% paraformaldehyde and stored in phosphate-buff-
ered saline (PBS) at 4°C.
Excitation threshold and maximum
capture rate
The excitation threshold (ET) and maximum capture rate
(MCR) were measured at days 6, 9, and 12 of culture and
for postnatal day 3 (P3) rat hearts following methods
described previously.29,33 ET was defined as the minimum
voltage required to elicit synchronous contractions over the
entire sample and MCR as the maximum frequency for
synchronous contractions at 150% of the ET. For engi-
neered tissue samples, measurements were made following
30 min of media exchange. For P3 hearts, rats (n = 4) were
anesthetized with isofluorane inhalation. Following thora-
cotomy, hearts were quickly excised and placed in a modi-
fied oxygenated Tyrode’s solution (126 mM NaCl, 11 mM
dextrose, 0.1 mM CaCl2, 13.2 mM KCl, 1 mM MgCl2, 12.9
mM NaOH, and 24 mM 4-(2-hydroxyethyl)-1-pipera-
zineethanesulfonic acid (HEPES)) at room temperature.
Strips of left ventricular myocardium (≈2 mm × 2 mm × 4
mm) were excised and placed in the same bioreactors used
for tissue culture. For all samples, ET was measured by
applying square-wave monophasic pulses of 2 ms starting
at 0 V/cm and then increasing until the sample was observed
to beat synchronously. MCR was measured by setting the
voltage to 150% of the ET and increasing the frequency
until the contractions became asynchronous, irregular, or
ceased.
Native tissue preparation and sectioning
P12 and adult rat hearts were used for comparison to the
engineered tissue samples. Tissue was processed as previ-
ously described.35 Briefly, rats were anesthetized through
methoxyflurane, and hearts were quickly removed. Hearts
were perfused with a zero calcium Tyrode’s solution for 5
min followed by 2% paraformaldehyde for 15 min for fixa-
tion using the retrograde Langendorff method.40 Whole
hearts and engineered tissue samples were stored in 30%
sucrose in preparation for sectioning. For adult hearts,
biopsies were performed with a 5-mm-diameter biopsy
punch through the left ventricular wall. P12 hearts were
maintained as whole hearts. The biopsied adult hearts,
whole P12 hearts, and engineered tissue samples were fro-
zen in tissue freezing medium (Triangle Biomedical
Sciences, Durham, NC) and sectioned using a cryostat
Figure 1. Sample preparation and bioreactor. (a) Exploded and
(b) assembled view of mold for producing tissue samples, (c)
bioreactor consisted of two Petri dishes and carbon rods for
electrical stimulation, and (d) tissue sample.

4 Journal of Tissue Engineering 3(1)
(Leica CM1950; Wetzlar, Germany). For adult heart biop-
sies, tissues were sectioned parallel to the epicardial sur-
face and for P12 hearts from the top of the ventricles to
approximately 2 mm from the apex to produce 80- to
100-µm-thick sections. Longitudinal and transverse cross
sections with a thickness of 100 µm were produced for
engineered tissue samples.
Fluorescent labeling
Fluorescent labeling was performed before sectioning
for engineered tissue and after sectioning for native tis-
sue samples. Tissue samples were labeled as described
previously.35 Samples were either quad-labeled with
wheat germ agglutinin (WGA) to identify cell borders,
α-sarcomeric actinin to identify myocytes, Cx43 to iden-
tify gap junction channels, and 4′,6-diamidino-2-phe-
nylindole dihydrochloride (DAPI) to identify nuclei or
tri-labeled with α-sarcomeric actinin to identify myo-
cytes, vimentin to identify nonmyocytes (mostly fibro-
blasts), and with DAPI to identify nuclei.
All labeling was performed on a laboratory platform
rocker at room temperature (Thermo Fisher Scientific). The
antibodies were diluted in blocking solution consisting of
4% goat serum (Invitrogen, Carlsbad, CA) and 0.5% Triton
X-100 (Fisher Scientific, Pittsburgh, PA) diluted in PBS.
Rinsing was performed between all incubation steps and
included three 15-min rinses. For quad-labeling, samples
were incubated for 16 h with WGA-conjugated CF488 (20–
40 µg/mL in PBS; 29022; Biotium, Hayward, CA), 16 h
with mouse IgG1 anti-α-sarcomeric actinin (1:100; ab9465;
Abcam, Cambridge, MA) followed by 6 h with goat anti-
mouse IgG1-conjugated Alexa Fluor 633 (1:200; A21126;
Invitrogen), 1 h with Image-iT® FX signal enhancer (Alexa
Fluor 555 Goat Anti-Rabbit SFX Kit, A31630; Invitrogen)
to block nonspecific antibody binding, 16 h with rabbit
anti-GJA1 (1:100; SAB4300504; Sigma–Aldrich) followed
by 6 h with goat anti-rabbit IgG-conjugated Alexa Fluor
555 (1:200; A31630; Invitrogen), and 3 h with DAPI
(1:500; Sigma–Aldrich). For tri-labeling, samples were
incubated for 16 h with mouse IgG1 anti-α-sarcomeric
actinin (1:100; ab9465; Abcam) followed by 6 h with goat
anti-mouse IgG1-conjugated Alexa Fluor 633 (1:200;
A21126; Invitrogen), 16 h with mouse monoclonal anti-
vimentin-conjugated Cy3 (1:50; C9080; Sigma–Aldrich),
and 3 h with DAPI (1:500; Sigma–Aldrich). Tissue samples
were stored in PBS.
Confocal imaging
The 3D image stacks were acquired for samples labeled
with WGA, α-sarcomeric actinin, Cx43, and DAPI on a
Zeiss LSM 5 Duo confocal microscope (Carl Zeiss, Jena,
Germany) using a 40× oil-immersion objective lens with a
numerical aperture of 1.3.35 The sectioned tissue samples
were placed on a glass slide and surrounded by 15–30 µL of
Fluoromount-G™ Slide Mounting Medium (Electron
Microscopy Sciences, Hatfield, PA). The tissue sample was
covered with a coverslip (no. 0) and placed on the imaging
stage. The x-axis of the image stack was aligned with the
long axis of the myocytes by visual inspection and adjust-
ment of the scan direction. For engineered tissue samples,
sections were briefly scanned using a 10× objective lens to
locate dense regions of myocytes. Only regions with high
cell density were imaged in this study.
The image stacks were acquired with a voxel size of 200
nm × 200 nm × 200 nm and a typical field of view of 1024
× 768 × 200 voxels using a multitrack protocol for quasi-
simultaneous imaging of fluorophores in each 2D image
slice. Laser lines with a wavelength of 364, 488, 543, and
633 nm were alternately applied to excite their associated
fluorophores and collected using long-pass 385 nm, band-
pass 505–555 nm, long-pass 560 nm, and band-pass 630–
650 nm filters, respectively. The dwell time was typically
1.3–1.5 µs/pixel resulting in a total imaging time of approx-
imately 1 h per image stack. Signal-to-noise ratio (SNR) of
each image stack was measured as described previously.35
The image stacks with a SNR below 3 were rejected. For
whole sample examination of engineered tissue, 2D images
were acquired using a 10× objective of central transverse
and longitudinal tissue sections stained with α-sarcomeric
actinin, vimentin, and DAPI. Higher magnification (40×)
2D images were also acquired for engineered and native
tissue samples stained with α-sarcomeric actinin, vimentin,
and DAPI.
Image processing
The image stacks were processed to improve image quality
as previously described.34,35 In brief, the image stacks
were processed to remove background, correct for depth-
dependent attenuation, and deconvolved using the iterative
Richardson–Lucy algorithm with measured point spread
functions. Cross-reactivity was corrected in image proto-
cols where a primary antibody reacted with two secondary
antibodies. The cross-reactivity was characterized by
colocalization of Cy3-associated and α-sarcomeric actinin–
associated signal and removed by subtraction of Cy3-
associated intensities. Individual myocytes were segmented
using a manual deformable triangle mesh fitted in three
image planes (XY, XZ, and YZ) using the WGA,
α-sarcomeric actinin, Cx43, and DAPI image data.34,35,41
The manual segmentation was refined using the WGA
image data. A principal component analysis was performed
for each segmented myocyte to yield eigenvectors e1, e2,
and e3. A bounding box was created for each segmented
myocyte using the coordinate system spanned by the eigen-
vectors. Length, width, and height were determined from
the dimensions of the bounding box. Myocyte volume was
defined as the volume of voxels within the segmented

Lasher et al. 5
myocyte, and surface area was estimated from the surface
area of the triangle mesh.
Cx43 analysis
The percentage of the membrane stained positive with Cx43
was calculated for each segmented myocyte using projec-
tions of Cx43 intensities onto the myocyte surface. An illus-
tration of this method is shown in Figure 2. The membrane
was approximated by surface voxels around the perimeter of
the segmented myocyte. A 3D distance map was calculated
from both the inside and outside of the membrane. Gradient
vectors were calculated from the distance map. Cx43 inten-
sities within 1 µm of the membrane were projected onto the
membrane using the calculated distance map and vectors.
The percentage of the membrane positive for Cx43
(MemCx43Pos) was calculated for each myocyte
Memnv
nv
Cx43Pos
Mem,Cx43>0
Mem
=
with the number of membrane voxels (nvMem) and the
number of membrane voxels with nonzero Cx43 intensity
(nvMem,Cx43>0).
The spatial distribution of Cx43 was characterized
through projections of Cx43 intensities on the eigenvectors
of the myocyte.35 Profiles were normalized with respect to
total intensities, and the range of arguments was trans-
formed to [−1, 1] (i.e. centered with respect to the respec-
tive bounding box dimension). For each eigenvector,
polarization (Pol25%) was characterized through summation
of Cx43 intensities from 25% of either end of the myocyte.
The minimal polarization (Pol25%min), maximum polariza-
tion (Pol25%max), and sum of Pol25%min and Pol25%max
(Pol25%total) were reported. Uniform Cx43 distributions for a
profile would lead to Pol25%total of 50%. Higher-order statis-
tical moments, skewness (γ1) and kurtosis (γ2), were calcu-
lated for the Cx43 intensity profiles. Skewness and kurtosis
are measures of asymmetry and peakedness, respectively. A
skewness of zero indicates that intensities are evenly dis-
tributed on both sides of the mean, whereas positive and
negative values of skewness indicate that intensities are
concentrated in the negative (x < 0) and positive (x > 0)
domains, respectively. The kurtosis of a normal and uni-
form distributions is 0 and −1.2, respectively.
Myocyte volume fraction
The myocyte volume fraction (MVF) was calculated by
down-sampling the processed 3D image data for the
α-sarcomeric actinin labeling. Original voxels with dimen-
sions of 0.2 µm × 0.2 µm × 0.2 µm were resampled to 1.6
µm × 1.6 µm × 1.6 µm using the maximum value in a
26-voxel neighborhood relation.42 This effectively “blurred”
the sarcomeres and filled gaps between adjacent z-disks.
Histograms of voxel intensities associated with actinin-pos-
itive regions were generated, and thresholds were defined as
mode intensity minus one standard deviation. Voxels above
the threshold were considered actinin positive. MVF was
defined as the sum of actinin-positive voxels divided by the
sum of all voxels within the image stack.
Statistical analysis
The data were reported as mean ± standard deviations.
Statistical significance was determined with a one-way
analysis of variance (ANOVA) for each measure, followed
by post hoc Tukey–Kramer tests with an α = 0.05. Where
appropriate, F-tests were performed to determine differ-
ences in variances with an α = 0.05.
Results
Visual inspection of engineered tissue
preparations
Bright field images of the engineered tissue samples showed
that samples progressively condensed during culture (Figure
3). Engineered tissue sample cross-sectional area estimated
from the measured diameter was found to decrease to 17% ±
3% and 16% ± 5% for nonstimulated and stimulated samples
at the end of culture from the onset of stimulation. No signifi-
cant differences in cross-sectional area were observed between
the nonstimulated and stimulated samples. Central transverse
and longitudinal cross sections of whole tissue samples exhib-
ited dense regions of aligned myocytes and fibroblasts (Figure
4). Although nuclei appeared to be homogeneously distributed
through the sample thickness, elongated myocytes were
located approximately 200 µm from the sample periphery.
Higher magnification confocal images showed that fibroblasts
were in close spatial proximity to myocytes; however, P12 and
adult native tissue samples exhibited a higher density of fibro-
blasts and myocytes (Figure 5).
Figure 2. Simplified schematic representation for calculating
percentage of membrane positive for Cx43. Voxels are
represented on a grid with different colours indicating
membrane and Cx43 staining. The integer values represent
distance in voxels from the membrane. Gradient vectors were
calculated from the distance map, and Cx43 intensities were
mapped to the membrane as shown in (b).
Cx43: connexin-43.

6 Journal of Tissue Engineering 3(1)
Functional analysis
ET and MCR were measured at days 6, 9, and 12 of culture
and for isolated strips of P3 left ventricular myocardium.
ET and MCR were not measurable at day 3 of culture as the
samples did not respond to pacing. The ET decreased as a
function of time in culture for both nonstimulated and
stimulated samples, and the stimulated samples nearly
approached the ET of P3 rat myocardium (Figure 6).
Stimulated samples had significantly lower ET at days 6
(2.79 ± 0.15 vs 3.85 ± 0.29 V/cm), 9 (1.78 ± 0.13 vs 2.93 ±
0.13 V/cm), and 12 (1.00 ± 0.12 vs 2.46 ± 0.08 V/cm) of
culture compared to nonstimulated samples (p < 0.01).
MCR increased as a function of time in the culture for the
Figure 6. (a) Excitation threshold and (b) maximum capture
rate at days 6, 9, and 12 of culture and P3 rat hearts. Excitation
threshold progressively decreased and maximum capture rate
progressively increased as a function of time in culture. The
symbol “*” denotes statistical difference in means determined by
post hoc Tukey–Kramer tests (α = 0.05).
NS: nonstimulated; S: stimulated; P3: postnatal day 3.
Figure 3. Engineered tissue size. (a) Images showing
progression of engineered cardiac tissue size during
culture (nonstimulated (n = 9) and stimulated (n = 12)).
Scale bar = 1 mm. (b) Cross-sectional area normalized to
start of electrical stimulation (day 3). Error bars denote
standard deviation. Engineered cardiac tissue progressively
decreases in volume as a function of time in culture;
however, there is no statistically significant difference
between nonstimulated and stimulated engineered
cardiac tissues.
Figure 4. Typical central confocal images of nonstimulated (a
and c) and stimulated (b and d) engineered tissue stained with
actinin to identify myocytes, vimentin to identify fibroblasts,
and DAPI to identify nuclei. (a and b) Transverse and (c and d)
longitudinal cross sections. Dense regions of myocytes were
found approximately 200 µm from the periphery of the sample.
Fibroblasts were found in close spatial proximity to myocytes.
Scale bar: (a) 200 µm applies to all.
DAPI: 4′,6-diamidino-2-phenylindole dihydrochloride.
Figure 5. Typical confocal images of cardiac tissue samples
stained with actinin to identify myocytes, vimentin to identify
fibroblasts, and DAPI to identify nuclei. Engineered tissue from
(a) nonstimulated and (b) stimulated samples cultured for 12
days. Left ventricular myocardium from (c) P12 and (d) adult rats.
Myocytes and fibroblasts are in close spatial proximity. Visually,
the (b) stimulated sample has more densely packed myocytes
compared to the (a) nonstimulated sample. P12 and adult rat
myocardium have densely packed myocytes and fibroblasts. Scale
bar: (a) 50 µm applies to all.
DAPI: 4′,6-diamidino-2-phenylindole dihydrochloride.

Lasher et al. 7
stimulated group and exceeded that of P3 native myocar-
dium by the end of culture (p < 0.01). Nonstimulated sam-
ples exhibited an increase in MCR between days 6 and 9 (p
< 0.01) but not between days 9 and 12 (p > 0.05).
Furthermore, the stimulated samples had significantly
higher MCR at days 6 (374 ± 51 vs 273 ± 25 beats/min), 9
(569 ± 40 vs 379 ± 33 beats/min), and 12 (645 ± 39 vs 393
± 18 beats/min) of culture compared to nonstimulated sam-
ples (p < 0.01).
3D confocal imaging
The 3D confocal imaging and image analysis were applied
to 9 nonstimulated and 12 electrically stimulated engineered
tissue constructs and 5 P12 hearts and 7 adult hearts. The
approach was applied to preparations stained with WGA,
α-sarcomeric actinin, Cx43, and DAPI. Seventy-one image
stacks from the four experimental groups were obtained.
The image stacks with low SNR or motion artifact were
removed from further analysis. Final data were obtained
from 7 nonstimulated samples (n = 11 image stacks), 7 stim-
ulated samples (n = 13 stacks), 5 P12 hearts (n = 8 image
stacks), and 7 adult hearts (n = 13 image stacks). Raw image
data for engineered tissue samples are presented in Figure 7.
These stacks originate from ~1 µm outside the tissue surface
and extend ~50 µm into the tissue sample.
The processed image stacks from all groups confirmed
that myocytes exhibited an elongated morphology (Figure
8). Marked differences between the nonstimulated and
stimulated samples were visually noticeable in the 3D
image stacks. The stimulated group exhibited more densely
packed myocytes with a more pronounced elongated mor-
phology (Figure 8(a) and (e)), aligned sarcomeres in regis-
try (Figure 8(c) and (g)), and more Cx43 plaque formation
on the myocyte membrane (Figure 8(d) and (h)). Marked
differences between P12 and adult tissues were also appar-
ent by visual observation (Figure 8). P12 myocytes
appeared smaller in size (Figure 8(i) and (l)) and had Cx43
plaque formation around the lateral sarcolemma (Figure
8(l)), whereas adult myocytes had Cx43 plaque formation
primarily at cell ends (Figure 8(p)).
MVF
MVF was quantified by down-sampling the processed 3D
image data for the α-sarcomeric actinin labeling (Figure 9).
Down-sampling of the original images (Figure 9(a) and (b))
resulted in “blurring” of the actinin-associated intensities
(Figure 9(c) and (d)). Thresholding of the down-sampled
images resulted in identification of the intracellular space
of myocytes (Figure 9(e) and (f)). The MVF was nearly
double for the stimulated engineered tissue compared to the
nonstimulated engineered tissue (0.34 ± 0.14 vs 0.18 ±
0.06, p < 0.01). However, the MVF for both nonstimulated
(0.18 ± 0.06) and stimulated (0.34 ± 0.14) engineered tissue
was significantly lower than that of P12 (0.90 ± 0.06) and
adult (0.91 ± 0.04) myocardium (p < 0.01).
Myocyte segmentation and Cx43 analysis
Myocyte geometry was quantified through segmentation of
individual cells from the 3D image stacks. The segmenta-
tion process is shown in Figure 10 with example myocytes
from the four experimental groups. The manual manipula-
tion of 3D triangle meshes and thresholding of the WGA
channel were used to create 3D reconstructions of myo-
cytes. Central cross sections of the reconstructed myocytes
(Figure 10(a), (d), (g), and (j)) served for masking the WGA
and Cx43 image data (Figure 10(b), (e), (h), and (k)). The
3D visualizations of the segmented myocytes and associ-
ated Cx43 labeling are shown in Figure 10(c), (f), (i), and
(l). Myocyte geometry was calculated from the segmented
cells (Table 1). Adult myocytes were significantly larger in
length, width, height, surface area, and volume compared
Figure 7. Raw 3D image data of (a–d) nonstimulated and (e–h) stimulated engineered cardiac tissue. (a and e) DAPI, (b and f) WGA,
(c and g) Cx43, and (d and h) α-sarcomeric actinin. Scale bar: (a) 50 µm applies to all.
3D: three-dimensional; DAPI: 4′,6-diamidino-2-phenylindole dihydrochloride; WGA: wheat germ agglutinin; Cx43: connexin-43.

8 Journal of Tissue Engineering 3(1)
to nonstimulated and stimulated engineered tissue and P12
native rat myocardium. Length, width, height, surface area,
and volume were not statistically different between myo-
cytes from electrically stimulated tissue samples and P12
native myocardium. However, nonstimulated myocytes
had more often a rounded morphology as indicated by a
smaller mean length compared to stimulated and P12 myo-
cytes and higher widths and heights compared to stimulated
myocytes.
The spatial distribution of Cx43 was characterized
through projections of Cx43 intensities on myocyte eigen-
vectors e1, e2, and e3 and measures of polarization and
higher-order statistical moments. Figure 11 shows the pro-
file projections for the segmented example cells in Figure
10. In the nonstimulated myocyte, there was little Cx43
plaque formation indicated by the low percent membrane
positive for Cx43 (Figure 11(d)), and a large plaque domi-
nated the profiles as indicated by a sharp peak in the Cx43
projection profiles (Figure 11(a) to (c)). The stimulated
myocyte had the majority of Cx43 plaque formation on
one end of the cell as can be seen in the profile on eigen-
vector e1 (Figure 11(e)) and the large difference between
Pol25%e1min and Pol25%e1max and strong negative skewness
(γ1e1) (Figure 11(h)). The P12 myocyte had an approxi-
mately uniform distribution of Cx43 around the lateral
membrane as can be seen in the profile for eigenvector e1
(Figure 11(i)). The distribution had a skewness (γ1e1) near
0 and a kurtosis near −1.2, which indicates a uniform dis-
tribution (Figure 11(l)). Furthermore, the profile for eigen-
vector e3 (Figure 11(k)) for the P12 myocyte showed a
Figure 8. Processed 3D image data of (a–d) nonstimulated engineered cardiac tissue, (e–h) electrically stimulated engineered
cardiac tissue, (i–l) P12 native left ventricular cardiac tissue, and (m–p) adult native left ventricular cardiac tissue. (a, e, i, and m) DAPI,
α-sarcomeric actinin, and Cx43. (b, f, j, and n) DAPI, WGA, and Cx43. (c, d, g, h, k, l, o, and p) Zoomed regions of (a), (b), (e), (f), (i),
(j), (m), and (n), respectively, indicated by box. Scale bar: (a) 50 µm also applies to (b), (e), (f), (i), (j), (m), and (n), and (c) 10 µm also
applies to (d), (g), (h), (k), (l), (o), and (p).
3D: three-dimensional; DAPI: 4′,6-diamidino-2-phenylindole dihydrochloride; WGA: wheat germ agglutinin; Cx43: connexin-43.

Lasher et al. 9
bimodal distribution, which indicates that Cx43 plaques
were concentrated on the lateral sarcolemma as opposed to
cell ends as seen in the adult myocyte. The adult myocyte
had the majority of Cx43-associated intensities at cell
ends, which can be seen from projections for eigenvector
e1 (Figure 11(m)) and a Pol25%e1total greater than 50%. The
Cx43 distribution was weakly asymmetric as indicated by
a small difference in Pol25%e1min and Pol25%e1max and a small
positive skewness (γ1e1).
The extent of Cx43 plaque formation was assessed by
calculating the percentage of membrane positive for Cx43
staining on the segmented myocytes. Nonstimulated
engineered tissue had a significantly lower percentage of
the membrane area stained positive for Cx43 (3.5% ± 3.4%)
compared to stimulated engineered tissue (6.9% ± 3.8%)
and that of P12 (7.1% ± 2.3%) and adult (8.3% ± 4.8%) rat
myocardium (Figure 12) (p < 0.01).
Statistical results of Cx43 profiles for all segmented
cells are presented in Figure 13. Myocytes from nonstimu-
lated and stimulated engineered tissue and P12 native myo-
cardium exhibited no polarization of Cx43, whereas adult
myocytes had the majority of their Cx43 concentrated at
cell ends (Figure 13(a)). Nonstimulated myocytes had a
large difference in Pol25%e1min and Pol25%e1max (Figure 13(a))
Figure 9. Calculating MVF in the 3D image stacks. (a and b) Actinin labeling is (c and d) resampled at 1.6 µm resolution and (e and
f) thresholded to estimate MVF. (a, c, and e) Nonstimulated and (b, d, and f) stimulated. Scale bar : (a) 50 µm applies to (b–f).
3D: three-dimensional; MVF: myocyte volume fraction.

10 Journal of Tissue Engineering 3(1)
and a high standard deviation of skewness (Figure 13(b)),
indicating that most cells had Cx43 plaques concentrated
on one side of the myocyte. Furthermore, the measured
skewness (Figure 13(b), (e), and (h)) and kurtosis (Figure
13(c), (f), and (i)) were highly variable for the nonstimu-
lated group compared to all other groups for all three eigen-
vector profiles.
Discussion
In this study, we applied 3D confocal imaging and analysis
to test the hypothesis that environmental cues direct engi-
neered tissue toward a phenotype resembling that of age-
matched native myocardium. We characterized effects of
electrical stimulation on myocyte geometry and the spatial
distribution of Cx43 in engineered cardiac tissue, and
applied the same techniques for characterization to age-
matched and adult native tissue. The results of the study
support the hypothesis that electrical stimulation directs
engineered cardiac tissue toward a phenotype resembling
that of age-matched native myocardium. Adult myocytes
were found to have significantly different geometries and
Cx43 distributions compared to both engineered tissue
constructs and P12 myocardium (Table 1 and Figures 12
and 13). This suggests that age-matched native tissue
serves as a more realistic target for both analytical com-
parison and development of design specifications for engi-
neered tissue samples.
Although other studies have elucidated some of the
effects of electrical stimulation on the development of
Figure 10. Myocyte segmentation and visualization. (a–c) Nonstimulated, (d–f) stimulated, (g–i) P12 rat, and (j–l) adult rat. (a, d,
g, and j) Central XZ-section of segmented myocyte from the image stack. (b, e, h, and k) Corresponding central XZ-section of the
image stack with WGA and Cx43. (c, f, i, l) 3D visualization of segmented cells and Cx43. Scale bar: (a) 30 µm applies to all.
3D: three-dimensional; WGA: wheat germ agglutinin; Cx43: connexin-43; P12: postnatal day 12.
Table 1. Myocyte geometry and volume fraction.
Length (µm) Width (µm) Height (µm) Volume (µm3) Surface area (µm2) Myocyte volume
fraction
NS: samples, n = 7;
segmented cells, n = 64
58.8 ± 21.8a,b,c 13.0 ± 2.7a,c 10.1 ± 2.5a,c 2647 ± 790c1400 ± 381c0.18 ± 0.06a,b,c
S: samples, n = 7;
segmented cells, n = 58
81.5 ± 19.7c11.3 ± 2.0c8.6 ± 1.6c2968 ± 1296c1775 ± 585c0.34 ± 0.14b,c
P12: rats, n = 5;
segmented cells, n = 41
72.0 ± 10.9c11.5 ± 1.6c9.1 ± 1.3c3167 ± 783c1732 ± 344c0.90 ± 0.06
Adult: rats, n = 7;
segmented cells, n = 51
120.1 ± 31.3 29.4 ± 5.9 19.6 ± 3.3 26,916 ± 11,550 7431 ± 2555 0.91 ± 0.04
NS: nonstimulated; S: stimulated; P12: postnatal day 12.
Values shown are mean ± standard deviation. Statistical difference in means determined by post hoc Tukey–Kramer tests (α = 0.05).
aDifference from S.
bDifference from P12.
cDifference from adult.

Lasher et al. 11
engineered cardiac tissue,22,28,29 a novelty of our study is in
the comprehensive 3D imaging and analysis approach and
quantitative comparison to native tissue. A key feature of
the approach is the ability to extract individual myocytes
from the image data and assign a reliable coordinate system
from which measures of geometry and Cx43 are computed.
The common approach of 2D imaging produces a single
cross section through a cell, which increases the chances of
misinterpreting or overlooking data and introducing
variability.
The approach for fabricating tissue samples was based
on methods described by Hansen et al.38 and was selected
because of its success in producing tissue samples with
densely packed, aligned myocytes. Both nonstimulated and
stimulated engineered tissue samples showed dense regions
of myocytes and fibroblasts, the two most abundant cell
types in the heart (Figure 4). Our observations of tissue
sample development are in agreement with Hansen et al.38
Samples condensed to approximately 20% of their initial
cross-sectional area (Figure 3), and this process was inde-
pendent of electrical stimulation. Cells appeared to have a
rounded morphology and appeared to be homogeneously
distributed through the sample at the beginning of culture.
Figure 11. Cx43 intensity profiles from cells in Figure 10. (a–d) Nonstimulated, (e–h) stimulated, (i–l) P12 rat, and (m–p) adult rat.
Profiles were produced by projection of Cx43 intensities on the principal axes (a, e, i, and m) e1, (b, f, j, and n) e2 and (c, g, k, and o) e3.
Respective quantitative results from example cells are shown in (d), (h), (l), and (p).
Cx43: connexin-43; P12: postnatal day 12.
Figure 12. Percentage of membrane stained positive for
Cx43. Statistical difference in means determined by post hoc
Tukey–Kramer tests (α = 0.05). Symbols denote difference from
*S, #P12, and &adult.
NS: nonstimulated; S: stimulated; Cx43: connexin-43; P12: postnatal
day 12.

12 Journal of Tissue Engineering 3(1)
After 3–6 days in culture, the cells began to elongate, align,
and contract in isolated regions of the sample. Contracting
regions increased in size between 6 and 9 days of culture,
and whole samples were macroscopically observed to con-
tract by the end of culture (day 12).
In this study, electrical stimulation had a high impact on
MVF. Although the MVF for the electrically stimulated
group was less than half of that of P12 and adult myocar-
dium, it was nearly double compared to nonstimulated
samples. We suggest that the increase in MVF in the electri-
cally stimulated group is caused by more myocytes devel-
oping and maturing. This notion is supported by the fact
that both the nonstimulated and stimulated samples started
with the same number of cells, construct sizes were not sta-
tistically different at the end of culture (Figure 3), and myo-
cytes did not differ in volume (Table 1).
Myocytes subjected to electrical stimulation had geom-
etries and Cx43 distributions that more closely matched
P12 myocardium compared to nonstimulated myocytes and
native adult myocytes. Under electrical stimulation, myo-
cytes were found to assume an elongated morphology as
opposed to their nonstimulated counterparts, which often
had a rounded morphology. This is consistent with previous
studies.22 Furthermore, sarcomeres were often disorganized
in nonstimulated myocytes (Figure 8(c)) compared to stim-
ulated samples, P12, and adult tissue (Figure 8(g), (k), and
(o), respectively). Moreover, electrical stimulation was
found to increase the percentage of membrane positive for
Cx43 plaques compared to nonstimulated samples (Figure
12). In fact, the percentage of membrane positive for Cx43
in the electrically stimulated group was not statistically dif-
ferent from P12 or adult rat myocardium.
Previous studies have reported the spatiotemporal
dynamics of Cx43 in postnatal development of rat18 and
human cardiac tissue.19 In neonatal rat cardiac tissue, Cx43
plaques were found to be uniformly distributed over the
myocyte membrane and remodeled to become concentrated
at cell ends (i.e. polarization) at approximately 90 days. In
adult rat myocytes, Cx43 is mostly located at cell ends,
which was found in our previous35 and other studies.15,43
Thus, our findings in this study are in agreement as P12
myocytes had no preference for Cx43 plaques at cell ends
Figure 13. Quantitative results of Cx43 analysis. (a, d, and g) Polarization (Pol25%), (b, e, and h) skewness, (c, f, and i) and kurtosis for
principal axes (a–c) e1, (d–f) e2, (g–i) and e3. Statistical difference in means of total Cx43 (a, d, and g) polarization (Pol25%) determined
by post hoc Tukey–Kramer tests (α = 0.05) and variance of (b, e, and h) skewness and (c, f, and i) kurtosis determined by F-tests (α =
0.05). Symbols denote difference from *S, #P12, and &adult.
NS: nonstimulated; S: stimulated; Cx43: connexin-43; P12: postnatal day 12.

Lasher et al. 13
(Pol25% = 40% ± 9%) compared to adult myocytes, which
showed that the majority of Cx43 plaques were found at
cell ends (Pol25% = 63% ± 13%). Myocytes from both engi-
neered tissue groups were similar to P12 myocytes and had
no significant Cx43 polarization. The Cx43 profiles showed
that myocytes from all groups had significant variability
with respect to symmetry and polarization. Projections of
Cx43 intensities on eigenvector e1 indicated that the non-
stimulated myocytes had the largest difference between
Pol25%min and Pol25%max, which was reflected in the large
standard deviation of the skewness. Projections on eigen-
vectors e2 and e3 showed that the P12 myocytes had signifi-
cant polarization, indicating that Cx43 plaques were located
along the lateral sarcolemma as opposed to centralized
regions of cell ends as in adult tissue. The nonstimulated
myocytes had the highest standard deviation in both meas-
ures of skewness and kurtosis for intensity profiles on all
axes, suggesting high variability with respect to Cx43
distributions.
Functional measures of ET and MCR were found to be
influenced by electrical stimulation (Figure 6). The lower
ET and higher MCR found in the electrically stimulated
group are in agreement with other studies, which have
applied electrical stimulation.22,29,33 The measured ET of
0.63 ± 0.05 V/cm and MCR of 541 ± 75 beats/min for P3
neonatal ventricles measured in this study were in close
agreement with other studies, which ranged from 0.74 ± 0.2
to 1.6 ± 0.1 V/cm for ET and 413 ± 7 to 475 ± 25 beats/min
for MCR.33,44
Applications of developed approaches
An application of the developed approach is to define spec-
ifications for tissue engineering. Specifications are para-
mount to the engineering paradigm, and tissue engineering
is no exception. In this study, specifications were derived
from normal age-matched and adult left ventricular myo-
cardium of rat since a central goal of tissue engineering is
reestablishing features of the native myocardium. However,
specifications can be derived for any requirement, for
example, diseased cardiac tissue where the focus may be to
understand the effects of pharmaceutical agents.1,38
Another application of the imaging approach and analy-
sis is to characterize structural features of stem cell–based
engineered tissue samples. Induced pluripotent human stem
cells have the potential to differentiate into any cell type45,46
and have been specifically differentiated to cardiomyo-
cytes.47 However, their application in developing 3D tissue
constructs is in its infancy.48
The imaging approach can also be applied to character-
ize other structural features of cardiac tissue. Sarcomeric
actinin staining revealed that many myocytes in the non-
stimulated group had sarcomeres that appeared to be disor-
ganized compared to the stimulated, P12 and adult
myocytes, which had well-defined sarcomeres in registry.
Fibroblasts, the majority of cells in the heart, play an impor-
tant role in normal cardiac function such as maintaining the
extracellular matrix (ECM), paracrine signaling, and cell–
cell communication with myocytes and other fibro-
blasts.7,8,49 Quantifying sarcomere organization and the
spatial relationship of myocytes and fibroblast is of interest
for future studies. Furthermore, measurements of function
could be used in conjunction with the structural features
quantified in this study to elucidate the complex structure–
function relationships found in cardiac tissue. Moreover,
the application of other environmental conditions, such as
the combinational effects of electrical and mechanical stim-
ulation, is of interest for future work.
Limitations
Limitations relating to the preparation of native cardiac tis-
sue are described in our previous study.35 Our approach for
structural characterization did not determine if the Cx43
labeling resulted in functional gap junctions. Assessment of
phosphorylation50 and colocalization with N-cadherin51,52
can offer insight into the potential functionality but were
not performed in this study. Furthermore, the total Cx43
expression was not quantified in this study. Instead meas-
ures of the percent membrane positive for Cx43 were char-
acterized, which can serve as indirect measure of Cx43
expression. Furthermore, the phenotype of the fibroblasts
found in our engineered tissue constructs was not charac-
terized. Fibroblasts can differentiate into myofibroblasts in
culture, and the myofibroblastic phenotype can be present
in injured myocardium.53 Myofibroblasts are responsible
for remodeling the ECM and paracrine signaling; however,
their effects on engineered tissue have not been studied.54
Functional analyses were limited to measures of ET and
MCR. Although these measures are well established in the
field,22,29,33 other functional measures of, for instance, elec-
trical conduction and excitation–contraction coupling,
would be beneficial for comprehensive assessment of tissue
constructs. However, the focus of this study was on charac-
terizing structure through 3D confocal imaging and not on
functional analyses. Measures of cardiac structure are not
limited to myocyte geometry and Cx43 distributions.
However, those were selected because they are known to
influence functional properties and undergo significant
changes during development and diseased states.11–13
Some of our image data required cross-talk correction,
which is an a posteriori method based on detailed investi-
gation of signal intensities. Images costained with
α-sarcomeric actinin and vimentin exhibited cross-reac-
tivity of vimentin secondary (Cy3) with α-sarcomeric
actinin primary, that is, actinin exhibited both Alexa Fluor
633 and Cy3 fluorophores. The cross-reactivity is due to
the same species and isotype (mouse IgG1) of the antibod-
ies. Vimentin antibodies raised in different species do
exist and were tried in this study (e.g. anti-vimentin,

14 Journal of Tissue Engineering 3(1)
C-terminal antibody produced in rabbit; SAB4503083;
Sigma–Aldrich), however, without success.
The described method for myocyte segmentation
requires 3D confocal imaging and manual manipulation
of triangle meshes, both of which are inherently time-
consuming and tedious. Automated methods for image
acquisition and myocyte segmentation are a possible
solution to this issue and will be addressed in future
work. A further limitation of the presented approach is
related to the volume of imaged regions. Each image
stack spans approximately 200 µm × 150 µm × 50 µm of
the sample volume. To overcome the relatively small vol-
ume of the image stack, several image stacks were
obtained, and only regions dense with myocytes were
imaged. Dense regions were identified by scanning the
sample with a 10× objective lens.
Funding
Funding for this work was provided through the Department of
Bioengineering and Richard A. and Nora Eccles Fund for
Cardiovascular Research and awards from the Nora Eccles
Treadwell Foundation.
Acknowledgements
We gratefully acknowledge Mr. Gustavo Lenis and Ms. Bettina
Schwab for assistance with the presented study.
Conflict of interest
The authors declare that there is no conflict of interest.
References
1. Elliott NT and Yuan F. A review of three-dimensional in
vitro tissue models for drug discovery and transport studies.
J Pharm Sci 2011; 100: 59–74.
2. Griffith LG and Swartz MA. Capturing complex 3D tissue
physiology in vitro. Nat Rev Mol Cell Biol 2006; 7: 211–224.
3. Mikos AG, Herring SW, Ochareon P, et al. Engineering
complex tissues. Tissue Eng 2006; 12: 3307–3339.
4. Hill JA and Olson EN. Cardiac plasticity. N Engl J Med
2008; 358: 1370–1380.
5. Ahuja P, Sdek P and MacLellan WR. Cardiac myocyte cell
cycle control in development, disease, and regeneration.
Physiol Rev 2007; 87: 521–544.
6. Banerjee I, Fuseler JW, Price RL, et al. Determination of cell
types and numbers during cardiac development in the neona-
tal and adult rat and mouse. Am J Physiol Heart Circ Physiol
2007; 293: H1883–H1891.
7. Souders CA, Bowers SLK and Baudino TA. Cardiac fibro-
blast: the renaissance cell. Circ Res 2009; 105: 1164–1176.
8. Kakkar R and Lee RT. Intramyocardial fibroblast myocyte
communication. Circ Res 2010; 106: 47–57.
9. Nag AC. Study of non-muscle cells of the adult mammalian
heart: a fine structural analysis and distribution. Cytobios
1980; 28: 41–61.
10. Camelliti P, Borg TK and Kohl P. Structural and functional
characterisation of cardiac fibroblasts. Cardiovasc Res 2005;
65: 40–51.
11. Vliegen HW, Laarse A, Huysman JAN, et al. Morphometric
quantification of myocyte dimensions validated in normal
growing rat hearts and applied to hypertrophic human hearts.
Cardiovasc Res 1987; 21: 352–357.
12. Campbell SE, Korecky B and Rakusan K. Remodeling of
myocyte dimensions in hypertrophic and atrophic rat hearts.
Circ Res 1991; 68: 984–996.
13. Swynghedauw B. Molecular mechanisms of myocardial
remodeling. Physiol Rev 1999; 79: 215–262.
14. Teunissen BEJ, Jongsma HJ and Bierhuizen MFA. Regulation
of myocardial connexins during hypertrophic remodelling.
Eur Heart J 2004; 25: 1979–1989.
15. Severs NJ, Dupont E, Coppen SR, et al. Remodelling of gap
junctions and connexin expression in heart disease. Biochim
Biophys Acta 2004; 1662: 138–148.
16. Fozzard HA. Gap junctions and liminal length in hyper-
trophy: something old and something new. J Cardiovasc
Electrophysiol 2001; 12: 836–837.
17. Beardslee MA, Laing JG, Beyer EC, et al. Rapid turnover
of connexin43 in the adult rat heart. Circ Res 1998; 83:
629–635.
18. Angst BD, Khan LU, Severs NJ, et al. Dissociated spatial
patterning of gap junctions and cell adhesion junctions dur-
ing postnatal differentiation of ventricular myocardium. Circ
Res 1997; 80: 88–94.
19. Peters NS, Severs NJ, Rothery SM, et al. Spatiotemporal
relation between gap junctions and fascia adherens junctions
during postnatal development of human ventricular myocar-
dium. Circulation 1994; 90: 713–725.
20. Kostin S, Dammer S, Hein S, et al. Connexin 43 expression
and distribution in compensated and decompensated cardiac
hypertrophy in patients with aortic stenosis. Cardiovasc Res
2004; 62: 426–436.
21. Salameh A, Wustmann A, Karl S, et al. Cyclic mechanical
stretch induces cardiomyocyte orientation and polarization
of the gap junction protein connexin43. Circ Res 2010; 106:
1592–1602.
22. Radisic M, Park H, Shing H, et al. Functional assembly of
engineered myocardium by electrical stimulation of cardiac
myocytes cultured on scaffolds. Proc Natl Acad Sci U S A
2004; 101: 18129–18134.
23. Eschenhagen T, Fink C, Remmers U, et al. Three-dimensional
reconstitution of embryonic cardiomyocytes in a collagen
matrix: a new heart muscle model system. FASEB J 1997;
11: 683–694.
24. Shimizu T, Yamato M, Isoi Y, et al. Fabrication of pulsatile
cardiac tissue grafts using a novel 3-dimensional cell sheet
manipulation technique and temperature-responsive cell cul-
ture surfaces. Circ Res 2002; 90: e40.

Lasher et al. 15
25. Ott HC, Matthiesen TS, Goh SK, et al. Perfusion-
decellularized matrix: using nature’s platform to engineer a
bioartificial heart. Nat Med 2008; 14: 213–221.
26. Zimmermann WH and Cesnjevar R. Cardiac tissue engineer-
ing: implications for pediatric heart surgery. Pediatr Cardiol
2009; 30: 716–723.
27. Vunjak-Novakovic G, Lui KO, Tandon N, et al.
Bioengineering heart muscle: a paradigm for regenerative
medicine. Annu Rev Biomed Eng 2011; 13: 245–267.
28. Tandon N, Marsano A, Maidhof R, et al. Optimization of
electrical stimulation parameters for cardiac tissue engineer-
ing. J Tissue Eng Regen Med 2011; 5: e115–e125.
29. Chiu LL, Iyer RK, King JP, et al. Biphasic electrical field
stimulation aids in tissue engineering of multicell-type car-
diac organoids. Tissue Eng Part A 2008; 17: 1465–1477.
30. Fink C, Ergun S, Kralisch D, et al. Chronic stretch of engi-
neered heart tissue induces hypertrophy and functional
improvement. FASEB J 2000; 14: 669–679.
31. Akhyari P, Fedak PW, Weisel RD, et al. Mechanical stretch
regimen enhances the formation of bioengineered autologous
cardiac muscle grafts. Circulation 2002; 106: I137–I142.
32. Zimmermann WH, Schneiderbanger K, Schubert P, et al.
Tissue engineering of a differentiated cardiac muscle con-
struct. Circ Res 2002; 90: 223–230.
33. Radisic M, Yang L, Boublik J, et al. Medium perfusion ena-
bles engineering of compact and contractile cardiac tissue.
Am J Physiol Heart Circ Physiol 2004; 286: H507–H516.
34. Lasher RA, Hitchcock RW and Sachse FB. Towards mod-
eling of cardiac micro-structure with catheter-based confocal
microscopy: a novel approach for dye delivery and tissue char-
acterization. IEEE Trans Med Imaging 2009; 28: 1156–1164.
35. Lackey DP, Carruth ED, Lasher RA, et al. Three-
dimensional modeling and quantitative analysis of gap junc-
tion distributions in cardiac tissue. Ann Biomed Eng 2011;
39: 2683–2694.
36. Conchello JA and Lichtman JW. Optical sectioning micros-
copy. Nat Methods 2005; 2: 920–931.
37. Diaspro A. Confocal and two-photon microscopy: founda-
tions, applications, and advances. New York: Wiley, 2002,
p. 567.
38. Hansen A, Eder A, Bonstrup M, et al. Development of a drug
screening platform based on engineered heart tissue. Circ
Res 2010; 107: 35–44.
39. Tandon N, Cannizzaro C, Chao PH, et al. Electrical stimula-
tion systems for cardiac tissue engineering. Nat Protoc 2009;
4: 155–173.
40. Langendorff O. Untersuchungen am überlebenden
Säugetierherzen. Pflügers Arch 1895; 61: 291–332.
41. Savio-Galimberti E, Frank J, Inoue M, et al. Novel features
of the rabbit transverse tubular system revealed by quantita-
tive analysis of three-dimensional reconstructions from con-
focal images. Biophys J 2008; 95: 2053–2062.
42. Jahne B. Digital image processing: concepts, algorithms,
and scientific applications. Secaucus, NJ: Springer-Verlag
New York, Inc., 1993, pp. 1–383.
43. Dolber PC, Beyer EC, Junker JL, et al. Distribution of gap
junctions in dog and rat ventricle studied with a double-label
technique. J Mol Cell Cardiol 1992; 24: 1443–1457.
44. Bursac N, Papadaki M, Cohen RJ, et al. Cardiac muscle tis-
sue engineering: toward an in vitro model for electrophysi-
ological studies. Am J Physiol 1999; 277: H433–H444.
45. Shiba Y, Hauch KD and Laflamme MA. Cardiac applica-
tions for human pluripotent stem cells. Curr Pharm Des
2009; 15: 2791–2806.
46. Nunes SS, Song H, Chiang CK, et al. Stem cell-based car-
diac tissue engineering. J Cardiovasc Transl Res 2011; 4:
592–602.
47. Zhang J, Wilson GF, Soerens AG, et al. Functional cardio-
myocytes derived from human induced pluripotent stem
cells. Circ Res 2009; 104: e30–e41.
48. Liau B, Christoforou N, Leong KW, et al. Pluripotent stem
cell-derived cardiac tissue patch with advanced structure and
function. Biomaterials 2011; 32: 9180–9187.
49. Zlochiver S, Muñoz V, Vikstrom KL, et al. Electrotonic
myofibroblast-to-myocyte coupling increases propensity to
reentrant arrhythmias in two-dimensional cardiac monolay-
ers. Biophys J 2008; 95: 4469–4480.
50. Beardslee MA, Lerner DL, Tadros PN, et al.
Dephosphorylation and intracellular redistribution of ven-
tricular connexin43 during electrical uncoupling induced by
ischemia. Circ Res 2000; 87: 656–662.
51. Kostetskii I, Li J, Xiong Y, et al. Induced deletion of the
N-cadherin gene in the heart leads to dissolution of the inter-
calated disc structure. Circ Res 2005; 96: 346–354.
52. Li J, Levin MD, Xiong Y, et al. N-cadherin haploinsuffi-
ciency affects cardiac gap junctions and arrhythmic suscepti-
bility. J Mol Cell Cardiol 2008; 44: 597–606.
53. Rohr S. Cardiac fibroblasts in cell culture systems: myofibro-
blasts all along? J Cardiovasc Pharmacol 2011; 57: 389–399.
54. Tomasek JJ, Gabbiani G, Hinz B, et al. Myofibroblasts and
mechano-regulation of connective tissue remodelling. Nat
Rev Mol Cell Biol 2002; 3: 349–363.