Mechanical changes in the rat right ventricle with decellularization

Colleen Witzenburg

, Ramesh Raghupathy

, Stefan M. Kren

, Doris A. Taylor

, Victor H. Barocas

Department of Mechanical Engineering, University of Minnesota, Minneapolis, MN 55455, USA

Center for Cardiovascular Repair, Department of Integrative Biology and Physiology, University of Minnesota, Minneapolis, MN 55455, USA

Department of Biomedical Engineering, University of Minnesota, 7-105 Nils Hasselmo Hall, 312 Church St SE, Minneapolis, MN 55455, USA

article info

Article history:

Accepted 30 September 2011

Keywords:

Perfusion-decellularization

Heart ventricle

Heterogeneous

Biaxial testing

abstract

The stiffness, anisotropy, and hete rogeneity of freshly dissected (control) and perfusion-decellularized

rat right ventricles were compared using an anisotropic inverse mechanics method. Cruciform tissue

samples were speckled and then tested under a series of different biaxial loading conﬁgurations with

simultaneous force measurement on all four arms and displacement mapping via image correlation.

Based on the displacement and force data, the sample was segmented into piecewise homogeneous

partitions. Tissue stiffness and anisotropy were characterized for each partition using a large-

deformation extension of the general linear elastic model. The perfusion-decellularized tissue had

signiﬁcantly higher stiffness than the control, suggesting that the cellular contribution to stiffness, at

least under the conditions used, was relatively small. Neither anisotropy nor heterogeneity (measured

by the partition standard deviation of the m odulus and anisotropy) varied signiﬁcantly between control

and decellularized samples. We thus conclude that although decellularization produces quantitative

differences in modulus, decellularized tissue can provide a useful model of the native tissue

extracellular matrix. Further, the large-deformation inverse method presented herein can be used to

characterize complex soft tissue behaviors.

1. Introduction

A bioartiﬁcial heart is a possible alternative to transplantation

or a mechanical left ventricular assist device. Whole-organ

perfusion-decellularization has been established as a potential

tool to produce an intact three-dimensional scaffold for bioarti-

ﬁcial hearts (Ott et al., 2008; Wainwright et al., 2010; Weymann

et al., 2011). Separating the cells from the extracellular matrix

also provides the opportunity to study cardiac wall mechanics in

a simpliﬁed system, and attribute myocardial tissue properties to

the respective cellular/extracellular constituents. For instance,

localized matrix injury or scarring that occurs in myocardial

infarction (Fomovsky and Holmes, 2010; Chen et al., 2003) could

be studied on the extracellular matrix with decellularized tissue.

Previous studies of healthy and/or infarcted myocardium have

focused on constitutive modeling of small samples cut from

different locations to assess heterogeneity while assuming homo-

geneity in each sample (e.g. Sacks and Chuong, 1993; Emery et al.,

1997). Assessing heterogeneity by means of multiple samples is

impossible, however, if the entire tissue is too small. For example,

the ventricle of the Sprague-Dawley (SD) rat, a popular choice

for myocardial decellularization experiments (Ott et al., 2008), is

only about 10 mm across and not amenable to assessment of

heterogeneity with samples cut from different locations. Although

indentation (e.g. Cox et al., 2010) could be applied regionally, it is

less relevant to in vivo function. An experiment in which hetero-

geneous properties could be extracted from a single sample

during biaxial loading would be far more valuable.

The gen eralized anisotropic inverse mechanics (GAIM) m ethod

(Raghupathy and Barocas, 2010; Raghupathy et al., 2011)enables

study of heterogeneous, anisotropic tissues, like the ventricle, by

dissecting small samples computationally rather than physically.

GAIM directly solves the ﬁnite element representation of the stress

balance in the tissue for the unknown components of the two-

dimensional general linear elasticity tensor. Heterogen eity within the

sample is managed by partitioning the sample into many subdo-

mains, each assumed to have uniform properties. In this study, we

utilized multiple biaxial tests and an extended version of GAIM to

determine the mechanical properties of freshly dissected (control)

and decellularized right ventricles from SD rats. We focused on the

right ventricle due to the prohibitively large thicknesses of control left

ventricle samples.

Two hypotheses were tested. First, since the extracellular

matrix, particularly collagen, plays a large role in tissue mechan-

ical behavior (Fomovsky et al., 2010), and since decellularization

reduces tissue volume considerably with negligible collagen loss

(Ott et al., 2008), we hypothesized that decellularized samples

would have larger stiffness values for the right ventricle as measured

by our methods. We expected this increase in stiffness to be

proportional to the thickness reduction, as observed previously

Contents lists available at SciVerse ScienceDirect

journal homepage: www.elsevier.com/locate/jbiomech

www.JBiomech.com

Journal of Biomechanics

doi:10.1016/j.jbiomech.2011.11.025

Corresponding author. Tel.: þ1 612 626 5572; fax: þ 1 612 626 6583.

E-mail address: [email protected] (V.H. Barocas).

Journal of Biomechanics 45 (2012) 842–849

(Ott et al., 2008). Second, based on minimal changes in the

collagen structure following decellularization, we hypothesized

that decellularized samples would not exhibit a change in degree of

mechanical anisotropy or heterogeneity.

2. Methods

2.1. Sample Preparation

Whole cadaveric rat hearts were removed from ten adult female SD rats (9–13

weeks) for testing. All experiments were performed in accordance with US Animal

Welfare Act and were approved by the Institutional Animal Care and Use

Committee at the University of Minnesota. Five hearts underwent decellulariza-

tion prior to ventricle dissection and ﬁve immediately underwent ventricle

dissection (control). One sample from each group was discarded due to damage

during experimentation; therefore, four samples from each group were analyzed.

Measurements of live rat body weight prior to dissection showed no signiﬁcant

difference (p¼0.28) between groups. Unless otherwise stated, all p-values

reported refer to unpaired two-tailed t-tests.

For decellularization, hearts were placed in a modiﬁed Langendorff apparatus

and perfused through the coronary arteries with 1% sodium dodecyl sulfate (SDS)

in water for 20 h, as described previously (Ott et al., 2008). Each heart was then

perfused with 60 ml deionized water, 60 ml 1% Triton-X100 in water, and 60 ml

additional deionized water to ﬂush the detergent. The hearts were then perfused

with 500 ml phosphate buffered saline (PBS) in 5 aliquots, leaving only decel-

lularized matrix behind. Fig. 1 shows a rat heart during decellularization and

depicts where samples were cut from the right ventricle. Sample geometry is

shown in Fig. 1b.

The right ventricle was dissected from each heart and laid ﬂat. A punch was

used to cut each sample into a cruciform shape for biaxial testing. Anatomical

orientation was marked on the endocardial surface as shown in Fig. 2a. Samples

were stored in 1% PBS at 4 1C when not undergoing biaxial testing or thickness

measurement. Sample thickness was measured by laser micrometer 5 min after

removal from PBS.

2.2. Biaxial testing

Samples were tested within 48 h of dissection. Verhoeff’s stain was used to

texture the epicardial surface for optical strain tracking (Fig. 2b). The textured

sample was attached to an Instron biaxial tester with four 5 N load cells via a

custom rig that allowed easy sample mounting and immersion in 1% PBS at room

temperature for the duration of testing (Fig. 2c).

A series of biaxial protocols was applied to each sample. First, a slight preload,

0.01 N, was applied to each cruciform arm. The grip forces were zeroed with

respect to the preload. The sample was preconditioned with nine equibiaxial

extensions; each of the four arms was extended 0.75 mm. Each loading and

unloading cycle lasted 10 s. After preconditioning, ﬁfteen separate displacement-

controlled biaxial experimental extensions were performed (Fig. 3). Unlike

traditional mechanical tests, which are designed to maximize homogeneity of

the strain ﬁeld, our protocol was designed to produce many heterogeneous strain

ﬁelds and to vary that heterogeneity over different tests. Heterogeneous, variable

strain ﬁelds are necessary to provide detailed deformation data for GAIM, and

using numerous ﬁelds improves the accuracy of parameters. An additional

Fig. 1. Sample geometry. (a) Rat heart during decellularization. Labels indicate

aorta (Ao), right atrium (RA), left atrium (LA), right ventricle (RV), and left

ventricle (LV). (b) Sketch of sample geometry overlaid on fully decellularized rat

heart, showing the circumferential and longitudinal directions.

Fig. 2. Sample dissection and clamping. (a) Schematic showing the anatomical position of the tissue during punching. (b) Epicardial surface of a representative

decellularized right rat ventricle after clamping and speckling. (c) Clamping conﬁguration during biaxial testing. The sample is submerged in 1% phosphate buffered saline

at room temperature.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849 843

equibiaxial extension was performed at the end of testing to conﬁrm that minimal

damage had occurred during the experiment. During testing, images of the

epicardial surface of the tissue and the forces at each grip were recorded.

2.3. Image analysis and strain tracking

Digital video of ventricle deformation was obtained at 24 fps, 1080p HD

resolution, and spatial resolution of 96 pixels/mm. The video was synchronized

and downsampled to construct grayscale image sequences corresponding to the

loading curves of each test. The image at the end of preconditioning was used as

the reference conﬁguration. The tissue boundary was sketched on top of the

reference image in Abaqus

, and meshed with quadrilateral elements. Successive

pairs of images were correlated to track the movement of the mesh nodes

throughout the loading sequence. This image correlation method was used

successfully in Raghupathy et al. (2011). Displacement ﬁelds were constructed

from movement of locations corresponding to the mesh nodes and smoothed to

reduce noise. Green strains were computed from cumulative displacement ﬁelds

by standard ﬁnite element theory for bilinear quadrilateral elements. We esti-

mated the accuracy of our displacement calculations by digitally deforming the

image of Fig. 2b and comparing the displacements obtained from image correla-

tion with analytical values. Displacements were accurate to 7 0.06 pixels and the

strains to 7 0.0026 (95% CI).

2.4. Generalized anisotropic inverse mechanics (GAIM)

Regional mechanical properties were determined by an extension of our GAIM

method (Raghupathy and Barocas, 2010; Raghupathy et al., 2011). The method,

presented in more detail in our previous work, involves solving the direct inverse

problem for a general linear elastic solid. That is, given the displacement data and

forces on the sample grips, the method uses ﬁnite elements to solve the linear

elastic stress balance (C

ijkl

)

¼0 for the components of C. The domain is

partitioned into regions, over each of which C is taken to be a constant. The

partitions are optimized so as to minimize error in the force balance while

maintaining tight conﬁdence intervals on the components of C ( Raghupathy and

Barocas, 2010). All experiments within the protocol are evaluated concurrently to

produce a large, overdetermined system of linear equations for C.

For highly deformable tissues such as myocardium, a formulation that

accounted for large-deformation kinematics but maintained the generality,

anisotropy, and linearity of the regression problem from our previous work was

desired. We modiﬁed our original form to

¼ K

ijkl

ð1Þ

where S is the 2nd Piola–Kirchhoff stress, E the Green strain, and K an elasticity

tensor analogous to C in linear elasticity and has the same major and minor

symmetries. The stress balance is

ðF

¼ðF

ijkl

¼ 0 ð2Þ

where F is the deformation tensor, and differentiation is with respect to the

undeformed coordinates. Eq. (2), although nonlinear in the displacements, is linear

with respect to the components of K, so the problem can be treated with linear

regression theory (Raghupathy and Barocas, 2010). Thus, the nonlinear kinematics

are incorporated into the analysis, but the ﬁtting problem remains linear.

In our earlier work, the tensor C was used to identify the major features of the

tissue. Speciﬁcally, the eigentensors of C represent principal states of stress and

strain (Cowin and Mehrabadi, 1995; Thomson, 1856), and the corresponding

eigenvalues (Kelvin moduli) provide measures of material stiffness. We have

found (Raghupathy et al., 2011) that the largest Kelvin modulus is a useful

measure and was comparable to Young’s modulus for polydimethylsiloxane. A

second important measure comes from the eigenvectors of the eigentensor

corresponding to the largest Kelvin modulus. These eigenvectors describe the

anisotropy of the material, with the direction corresponding to the larger

eigenvalue aligning with the tissue ﬁber direction in theoretical studies

(Raghupathy and Barocas, 2010), on a simple ﬁber model (Raghupathy and

Barocas, 2009), and in experiments on cell-compacted collagen gels

(Raghupathy et al., 2011).

In the current work, we computed large-deformation analogs of the Kelvin

moduli (henceforth simply ‘‘Kelvin moduli’’ for brevity) and corresponding

principal directions which give the preferred stiffness direction (

y). In addition,

the eigenvalues of the principal eigentensor of K,

, and l

were converted into an

anisotropy index (r), indicating strength of the alignment. It varies from 0 for an

isotropic sample to 1 for a perfectly aligned sample:

r ¼

99l

9þ9l

ð3Þ

In addition to stiffness and alignment, we quantiﬁed tissue heterogeneity by

calculating the standard deviation of each variable over all partitions within the

center of the sample. That is, each partition was treated as an independent

measurement, and the standard deviation was calculated. We refer to this value as

the ‘‘partition standard deviation’’ to avoid confusion with measures of sample-to-

sample variability.

3. Results

Right ventricle wall thickness was 21387 527

m (mean7

95% CI) for controls and 3817 157

m for the decellularization

group (po 10

4

). This was expected since cellular material

comprises much of the tissue volume. When the ﬁrst and ﬁnal

equibiaxial protocol strain maps were compared they showed

extremely similar patterns and values. There was no signiﬁcant

difference between the protocols in peak force for any arm of any

sample (p4 0.1 for all samples, paired two-tailed t-test).

Loading behavior for a representative sample of decellula-

rized tissue (Fig. 2b) is shown in Fig. 4. Fig. 4b and c highlight a

displacement-controlled equibiaxial test and single-arm extension

test. For equibiaxial loading, the circumferential arms, right and left,

had higher loads than the longitudinal arms, top and bottom. Shear

forces were responsible for the discrepancy in force between

Fig. 3. Protocol used for the biaxial testing of each sample. Dashed lines indicate ﬁxed arms. Equibiaxial: each arm moved 0.75 mm. Adjacent two-arm stretch: extended

arms moved 1.0 mm. Single–arm stretch: extended arm moved 1.0 mm. Strip biaxial: extended arms moved 0.75 mm. Three-arm stretch: middle extended arm moved

1.0 mm, other extended arms moved 0.75 mm.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849844

opposite arms. The single-arm extension involved displacement of

the right arm, as indicated by large forces on that arm and the

reduced response of the left arm. Forces on the top and bottom arms

were small in compari son.

For the same sample, the Green strains – E

, E

and E

–at

peak displacement for both equibiaxial and single-arm extension

tests are shown in Fig. 5. For the equibiaxial test, strain was larger

in the vertical direction than in the horizontal direction in the

central region of the sample. This effect was reversed for the

single-arm extension test. Also, E

showed a strong horizontal

gradient across the central region of the sample for the single-arm

extension ( Fig. 5e). Thus, as desired, strain ﬁelds were hetero-

geneous and the strain differed within the same region of the

sample from one protocol to another. For both tests, however,

there was relatively little shear strain.

Alignment and stiffness maps were calculated for each sample

using the extended GAIM method. Fig. 6 shows maps for the

sample pictured in Fig. 2b. The arms of the sample present as

artiﬁcially aligned in the pull direction as a result of our inability

to measure transverse forces on the grips and minimal transverse

displacement in the arms. However, the central region of the

sample is well-speciﬁed, so our analysis focuses only on the

central region (Fig. 6b, d).

Fig. 6a shows strong circumferential (i.e., left–right) alignment

in the central region. The average value of the anisotropy index

within the central region (Fig. 6b) was 0.71, indicating strong

anisotropy (roughly ﬁve times stiffer in the preferred, i.e. circum-

ferential, direction). The partition standard deviation for the

anisotropy index was low, 0.084. The average value for the

preferred stiffness direction within the central region was 7.661

from the horizontal, and its partition standard deviation was

7.551. Fig. 6c shows the maximum Kelvin modulus within each

region. The average value of the maximum Kelvin modulus within

the central region (Fig. 6d) was 400 kPa, and its partition standard

deviation was 107 kPa.

Fig. 7 summarizes the GAIM results for this study. Fig. 7a–c

show the average values of maximum Kelvin modulus, /E

anisotropy index, /rS, and preferred stiffness direction, /

over the central region. There was signiﬁcant difference between

S for the control and decellularized groups (p¼0.0003),

indicating that the central region of the decellularized tissue

was stiffer than that of the control. However, there was no

signiﬁcant difference between /rS or /

S for the control and

decellularized groups (p¼0.06, 0.31), indicating no change in

strength or direction of anisotropy. All samples were stiffer in

the circumferential than the longitudinal direction.

Fig. 7d–f show the normalized partition standard deviations

of the maximum Kelvin modulus,

//E

S, anisotropy index,

//rS, and preferred stiffness direction,

S, over the central

region. These values were used to quantify the degree of

Fig. 4. Representative load data for the decellularized sample shown in Fig. 2b. (a) Three of nine preconditioning cycles are shown followed by the equibiaxial

protocol, four adjacent two-arm extension protocols, four single-arm extension protocols, two strip biaxial protocols and one three-arm extension protocol.

Other protocols not shown. (b) During equibiaxial loading, signiﬁcant forces were generated in all four arms, with higher forces in the circumferential (left–right)

direction because of tissue anisotropy. (c) When only the right arm was displaced, all forces were much smaller, and circumferential forces were much larger than

longitudinal forces.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849 845

heterogeneity of the parameters. For all three measures of

heterogeneity (associated with stiffness, degree of anisotropy,

and preferred direction) the control and decellularized samples

gave statistically indistinguishable results (p¼0.22, 0.76, 0.11),

indicating no signiﬁcant change in heterogeneity between the

decellularized and control groups.

Fig. 5. Representative strain results for the decellularized sample shown in Fig. 2b. Strains are overlaid on the deformed sample shape at maximum deformation. During

equibiaxial loading (a, b), the vertical strain (E

) was larger than the horizontal strain (E

) in the central region of the sample. During single-arm loading (d, e), E

was

larger in the central region of the sample. Also, there was a gradient in the central region of the sample in the map for E

. In both cases, there was relatively little shear

strain (c, f).

Fig. 6. Representative alignment and stiffness results from GAIM for the decellularized sample shown in Fig. 2b. Results are overlaid on the undeformed shape. For the

alignment maps (a, b), the color shows the anisotropy index, and the vectors indicate the preferred stiffness direction, with vector lengths corresponding to the anisotropy

index. For the stiffness maps (c, d) the contour shows the maximum Kelvin modulus. (a) Alignment map shows roughly uniform alignment in the center of the sample.

(b) There was strong circumferential alignment in the central region of the tissue, /rS¼0.71, with low partition standard deviation

¼0.084, and /yS¼7.661, with a low

partition standard deviation of

¼7.551. Black blocks indicate partition domain. (c) Stiffness map shows roughly uniform E

in the center of the sample. (d) The average

value of the maximum Kelvin modulus within the central region of the sample, /E

S¼400 kPa, and the partition standard deviation,

¼ 107 kPa. Black blocks indicate

partition domain.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849846

4. Discussion

The major technical advance of this work was introduction of a

nonlinear kinematic framework to the GAIM technique. The new

formulation, which relates the second Piola–Kirchhoff stress to

the Green strain linearly, maintains the efﬁciency of a linear

regression model (i.e., the model is linear in the coefﬁcients) but

is suitable to large deformations often experienced by soft tissues.

By doing so, we extended the potential applicability of the

method to a wider range of soft-tissue applications. The general-

ized model is not sufﬁcient to capture many soft-tissue behaviors,

including the large toe region followed by a sharp rise in stiffness,

but it provides a qualitative estimate of the material properties

and, most importantly, an assessment of material anisotropy in

different regions of an intact sample.

The major conclusions of this work as to the tissues tested

were as follows: (1) The stiffness of right ventricular tissue, as

measured by the average of the largest Kelvin modulus over the

sample’s center, increased by a factor of 6.7 with decellulariza-

tion. This change was consistent with densiﬁcation of the tissue

(average 5.6-fold reduction in thickness) and was similar to that

reported by others for decellularized cardiac tissue (Ott et al.,

2008; Wang et al., 2010). Although other factors (e.g., chemical

interactions between decellularizing agents and the extracellular

matrix) could be important, we attribute the slightly larger

increase in stiffness than decrease in thickness to structural

changes, speciﬁcally rotation of collagen ﬁbers into the plane of

testing. (2) The mechanical anisotropy of the tissue, which we

have shown previously (Raghupathy et al., 2011) to correlate to

structural anisotropy, was largely unchanged by the decellular-

ization process. (3) Finally, a key goal of this work was to assess

tissue heterogeneity. We found that the spatial variation in tissue

properties (stiffness, degree of anisotropy, and preferred stiffness

direction) over the central region of the sample was largely

unchanged by the decellularization process. Based on these three

observations, we conclude that decellularized tissue can be used

as a model for studying mechanical changes to the extracellular

matrix in the heart. The decellularized model is particularly

attractive for studies of rat left ventricle, which is sufﬁciently

thick (4 mm) relative to the other tissue dimensions (10 mm)

that planar biaxial tests would be of questionable validity. In the

decellularized sample, the thickness is 5–10% of the in-plane

dimensions, making the assumptions underlying planar biaxial

tests much more acceptable. Of course, the decellularized model

Fig. 7. Stiffness, alignment and direction results for full study. (a) There was a signiﬁcant difference between the control and decellularized groups in /E

S, which

indicates overall stiffening of the tissue with decellularization. Units of /E

S are kPa. (b, c) No signiﬁcant difference in /rS or /yS was found between the control and

decellularized samples, indicating minimal change in anisotropy. Units of /

yS are degrees and /rS is unitless. (d, e, f) No signiﬁcant difference was found between the

normalized partition standard deviation of the maximum Kelvin modulus, anisotropy index or preferred stiffness direction. This indicates that the effect of

decellularization is roughly uniform. All plots show mean7 95% CI, n¼4.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849 847

does not allow for the study of the cells, but the combination of

the decellularized system and GAIM analysis could allow assess-

ment of mechanical consequences of ventricular remodeling (e.g.,

scar formation) without isolation of the remodeled tissue.

Another potential advantage of the decellularized model is that,

by virtue of being thinner, it would be less prone to artifacts from

the ﬂattening prior to planar biaxial tests.

The signiﬁcant anisotropy of the right ventricular tissue is

consistent with previous work (Ott et al., 2008) showing that both

control and decellularized left ventricle were stiffer in the

circumferential than longitudinal direction in equibiaxial tests

and with previous studies of right ventricular properties (e.g.

Ghaemi et al., 2009). Ott et al. (2008) performed biaxial tests on

samples cut from cadaveric vs. decellularized left ventricle.

Decellularized samples had signiﬁcantly larger tangent moduli

than their intact cadaveric counterparts. When adjusted for

thickness, the signiﬁcant difference in tangent moduli vanished,

similar to our observations. Ott did not attempt to assess

differences in anisotropy or heterogeneity. Figure 11 of Ghaemi

et al. (2009) appears to have a ratio of about 10:1 in stress

between the two directions at 20% equibiaxial strain of bovine

right ventricular wall. The Ghaemi study used hooks rather than

clamps, but a 10:1 stress ratio would correspond to a very large

value of r in our analysis, roughly 0.8 (the exact value would

depend on the results of non-equibiaxial tests). Our calculated

value of /rS¼0.52 for the cadaveric right ventricle samples is

much lower. This difference is most likely due to our use of full-

thickness samples in comparison with the use of samples

extracted from the mid-one-third section of the heart by Ghaemi.

Although the decellularized tissue model is a simpler than the

intact tissue, there remain questions to be considered. Variation

in ﬁber orientation through the thickness of the myocardium

(along with any contribution from the epicardium and endocar-

dium) is particularly important. The ﬁber angle changes continu-

ously transmurally through the myocardium, with the average

through-thickness ﬁber orientation of both ventricles in the

circumferential direction. In addition, both the endocardium and

the epicardium exhibit higher stiffness in the circumferential

direction (Kang et al., 1996; Humphrey et al., 1990). Due to the

whole-organ perfusion-decellularization technique used we could

not excise tissue from the midwall, as is standard for biaxial

testing of ventricular tissue, for the decellularized sample group.

Therefore, we tested full thickness samples from both groups, and

our results must be considered as mean through-thickness

descriptions. In our planar analysis, we assumed that the dis-

placement was uniform through the tissue thickness, a reasonable

but not necessarily correct assumption; further work is underway

to track motion through the thickness as well as on the epicardial

surface. Again, the properties calculated in this study must be

seen as a transmural average of the properties of the decellular-

ized tissue since there is no way to assess contribution from

different layers.

To summarize, the current work has demonstrated that, in the

right ventricle, decellularized tissue can provide a useful model of

the native tissue extracellular matrix. Decellularization causes an

increase in the metric describing stiffness, but when adjusted for

the thickness decrease this effect is greatly reduced. There was no

signiﬁcant change in anisotropy or heterogeneity. A natural next

step would be to apply the methods to the left ventricle (which is

similar, (Humphrey, 2002; Costa et al., 2001)), where scarring and

remodeling are of great interest. The GAIM technique is able to

assess property variation within a tissue analog (Raghupathy

et al., 2011) and, with the improved kinematic framework devel-

oped herein, it can be applied to more complex problems

of ventricular tissue mechanical characterization, especially

in small-animal models, for which it may be difﬁcult if not

impossible to isolate a homogeneous sample of sufﬁcient size

for biaxial testing.

Conﬂict of interest statement

DAT holds a ﬁnancial interest in Miromatrix, Inc. and is

entitled to sales royalty through the University of Minnesota for

products related to the research described in this paper. This

relationship has been reviewed and managed by the University in

accordance with its conﬂict of interest policies.

Acknowledgments

This work was supported in part by the National Institutes of

Health (R21-EB-009788), the Medtronic Foundation, the National

Heart Lung and Blood Institute’s Progenitor Cell Biology Consor-

tium (#1U01HL100407-1), and the American Heart Association’s

Jon Hold DeHaan Cardiac Myogenesis Research Center

(#AHA09070499N). The technical assistance of Erich Boldt in

preparing and testing samples is gratefully acknowledged. We

also thank the Minnesota Supercomputing Institute for the

computing resources.

References

Chen, J., Song, S.K., Liu, W., McLean, M., Allen, J.S., Tan, J., Wickline, S.A., Yu, X.,

2003. Remodeling of cardiac ﬁber structure after infarction in rats quantiﬁed

with diffusion tensor MRI. American Journal of Physiology Heart and Circula-

tory Physiology 285 (3), H946–H954.

Costa, K.D., Holmes, J.W., McCulloch, A.D., 2001. Modelling cardiac mechanical

properties in three dimensions. Philosophical Transactions: Mathematical,

Physical and Engineering Sciences, 359(1783, The Integrated Heart: Modelling

Cardiac Structure and Function), 1233–1250.

Cowin, S.C., Mehrabadi, M.M., 1995. Anisotropic symmetries of linear elasticity.

Applied Mechanics Reviews 48 (5) 247-285.

Cox, M.A., Kortsmit, J., Driessen, N., Bouten, C.V., Baaijens, F.P., 2010. Tissue-

engineered heart valves develop native-like collagen ﬁber architecture. Tissue

Engineering. Part A 16 (5), 1527–1537.

Emery, J.L., Omens, J.H., McCulloch, A.D., 1997. Biaxial mechanics of the passively

overstretched left ventricle. The American Journal of Physiology 272 (5 Pt 2),

H2299–H2305.

Fomovsky, G.M., Holmes, J.W., 2010. Evolution of scar structure, mechanics, and

ventricular function after myocardial infarction in the rat. American Journal of

Physiology-Heart and Circulatory Physiology 298 (1), H221–H228.

Fomovsky, G.M., Thomopoulos, S., Holmes, J.W., 2010. Contribution of extracellular

matrix to the mechanical properties of the heart. Journal of Molecular and

Cellular Cardiology 48 (3), 490–496.

Ghaemi, H., Behdinan, K., Spence, A.D., 2009. In vitro technique in estimation of

passive mechanical properties of bovine heart part I. Experimental techniques

and data. Medical Engineering & Physics 31 (1), 76–82.

Humphrey, J.D., 2002. Passive Myocardium. In: Humphrey, J.D. (Ed.), Cardiovas-

cular Solid Mechanics; Cells, Tissues, and Organs. Springer, New York,

pp. 668–685.

Humphrey, J.D., Strumpf, R.K., Yin, F.C., 1990. Biaxial mechanical behavior of

excised ventricular epicardium. The American Journal of Physiology 259 (1 Pt

2), H101–H108.

Kang, T., Humphrey, J.D., Yin, F.C., 1996. Comparison of biaxial mechanical

properties of excised endocardium and epicardium. The American Journal of

Physiology 270 (6 Pt 2), H2169–H2176.

Ott, H.C., Matthiesen, T.S., Goh, S.K., Black, L.D., Kren, S.M., Netoff, T.I., Taylor, D.A.,

2008. Perfusion-decellularized matrix: using nature’s platform to engineer a

bioartiﬁcial heart. Nature Medicine 14 (2), 213–221.

Raghupathy, R., Barocas, V.H., 2009. A closed-form structural model of planar

ﬁbrous tissue mechanics. Journal of Biomechanics 42 (10), 1424–1428.

Raghupathy, R., Barocas, V.H., 2010. Generalized anisotropic inverse mechanics for

soft tissues. Journal of Biomechanical Engineering 132 (8), 081006.

Raghupathy, R., Witzenburg, C., Lake, S.P., Sander, E.A., & Barocas, V.H., 2011.

Identiﬁcation of regional mechanical anisotropy in soft tissue analogs. Journal

of Biomechanical Engineering 133 (9), 091011.

Sacks, M.S., Chuong, C.J., 1993. Biaxial mechanical properties of passive right

ventricular free wall myocardium. Journal of Biomechanical Engineering 115

(2), 202–205.

Thomson, W., 1856. Elements of a mathematical theory of elasticity. Philosophical

Transactions of the Royal Society of London 146, 481–498.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849848

Wainwright, J.M., Czajka, C.A., Patel, U.B., Freytes, D.O., Tobita, K., Gilbert, T.W.,

Badylak, S.F., 2010. Preparation of cardiac extracellular matrix from an intact

porcine heart. Tissue Engineering.Part C, Methods 16 (3), 525–532.

Wang, B., Borazjani, A., Tahai, M., Curry, A.L., Simionescu, D.T., Guan, J., To, F., Elder,

S.H., Liao, J., 2010. Fabrication of cardiac patch with decellularized porcine

myocardial scaffold and bone marrow mononuclear cells. Journal of Biomedi-

cal Materials Research. Part A 94 (4), 1100–1110.

Weymann, A., Loganathan, S., Takahashi, H., Schies, C., Claus, B., Hirschberg, K.,

Soos, P., Korkmaz, S., Schmack, B., Szabo, G., 2011. Development and evalua-

tion of a perfusion decellularization porcine heart model—generation of

3-dimensional myocardial neoscaffolds. Circulation Journal: Ofﬁcial Journal

of the Japanese Circulation Society 75 (4), 852–860.

C. Witzenburg et al. / Journal of Biomechanics 45 (2012) 842–849 849