Biophysically defined and cytocompatible covalently adaptable networks as viscoelastic 3d cell culture systems

Article abstract of DOI:10.1002/adma.201303680

Presented here is a cytocompatible covalently adaptable hydrogel uniquely capable of mimicking the complex biophysical properties of native tissue and enabling natural cell functions without matrix degradation. Demonstrated is both the ability to control elastic modulus and stress relaxation time constants by more than an order of magnitude while predicting these values based on fundamental theoretical understanding and the simulation of muscle tissue and the encapsulation of myoblasts.

Full text of DOI:10.1002/adma.201303680

www.advmat.de
www.MaterialsViews.com
Biophysically Defined and Cytocompatible Covalently
Adaptable Networks as Viscoelastic 3D Cell Culture Systems
Daniel D. McKinnon, Dylan W. Domaille, Jennifer N. Cha, and Kristi S. Anseth*
Covalently crosslinked synthetic hydrogels are especially suit-
able as tissue engineering scaffolds due to their well-defined
and easily tunable biochemical and biophysical properties. In
order to enable complex cell functions like ECM deposition,
motility, and spreading, a mechanism for crosslink degradation
must be engineered into the material; however, the presence
of a degradation trigger can complicate the cellular biophys-
ical microenvironment. Furthermore, covalently crosslinked
polymers typically produce a predominantly elastic material,
while native tissues are complex viscoelastic structures. Here,
we present a step-growth poly(ethylene glycol) (PEG) hydrogel
crosslinked by reversible hydrazone bonds. The macromer
components are readily synthesized from commercially avail-
able precursors, and the resulting gels form rapidly under
physiological conditions and provide a non-toxic matrix that is
suitable for cell culture. This material is capable of mimicking
aspects of the viscoelastic properties of native tissues, and the
dynamic stress relaxing crosslinks permit complex cellular
functions to occur while retaining the benefits of traditional
covalently crosslinked hydrogels. Taken together, these attrib-
utes make hydrazone crosslinked hydrogels a unique tool for
designing viscoelastic scaffolds and studying cellular responses
to scaffold elasticity.
culture environments, early approaches to hydrogel design
often incorporated hydrolytically degradable crosslinks, engi-
neered to degrade at a rate that was matched by the rate of
the cellular function under study.[4–6] This strategy allowed for
uniform ECM elaboration, but in many instances, only local
changes in gel properties are desired (e.g., to promote cell
spreading), and pre-engineering the degradation rates is often
a complex optimization process. Thus, hampered by artificial
constraints of bulk property changes at pre-determined rates,
researchers developed complementary strategies based on pro-
cesses that occur in native ECM, where cells bind, degrade and
respond to their local proteinaceous microenvironment. For
example, Hubbell et al. pioneered the development of hydro-
gels crosslinked with cell-degradable and integrin binding
peptide sequences, which mimicked critical aspects of the bio-
chemical function of ECM.[7,8] Such polymer-peptide gels have
permitted researchers to observe and characterize the dynamics
of cell spreading,[9–11] migration,[11,12] axon extension,[13,14] pro-
liferation,[15,16] and differentiation in a synthetic three-dimen-
sional matrices.[16,17] However, unlike hydrolytically degradable
gels, the cellular microenvironment in cellularly remodeled
gels can be dramatically different from that of synthetic and
hydrolytically degradable materials, which makes it difficult to
characterize and draw conclusions about the role of the local
biophysical environment on cell function.
In contrast, a covalently adaptable network responds to
cell-induced stress by rapidly breaking and reforming elasti-
cally active crosslinks while maintaining stable material prop-
erties over time. This allows the material to respond to cell
stresses yet provides a uniform scaffold elasticity to enable
the elucidation of the effect of biophysical microenvironment
on cell function. Given the increasing use of progenitor cells
in biomaterials research and the increasing understanding
of the role of biomechanical signaling on cell fate,[18–21] such
a material should enable the observation of cell behavior in
a biophysically defined scaffold. While dynamic bonds have
previously been used in biomaterial scaffolds,[22–26] these
linkages are supplemented by additional interactions, which
precludes the precise and predictable biophysical control
of the hydrogel. We sought to use a linkage that was stable
enough to maintain the integrity of a crosslinked PEG net-
work for several weeks of cell culture, but that would be labile
enough to relax in response to cell-applied stress and allow
for cell spreading, process extension, migration, or axon
outgrowth to occur. Furthermore, few dynamic chemistries
are compatible with cell encapsulation, and previously pub-
lished covalently adaptable networks have either relied on
high temperatures, low pH, or UV light to trigger covalent
adaptability.[27–34]
Culturing cells in hydrogels for the purpose of regener-
ating tissue requires that the polymer matrix accommodate
numerous cellular functions, beginning with matrix interac-
tions and morphological changes. This initial adaptation allows
encapsulated cells to subsequently engage in an external-
internal exchange of information that influences critical pro-
cesses including proliferation, motility, and extracellular matrix
(ECM) deposition.[1–3] To facilitate improved three-dimensional
D. D. McKinnon, D. W. Domaille, J. N. Cha, K. S. Anseth
Department of Chemical and Biological Engineering
University of Colorado Boulder
Jennie Smoly Caruthers Biotechnology Building
3415 Colorado Ave, 596 UCB, Boulder, CO, 80303
D. D. McKinnon, K. S. Anseth
BioFrontiers Institute
University of Colorado Boulder
Jennie Smoly Caruthers Biotechnology Building
3415 Colorado Ave, 596 UCB, Boulder, CO, 80303
K. S. Anseth
Howard Hughes Medical Institute University of Colorado Boulder
Jennie Smoly Caruthers Biotechnology Building
3415 Colorado Ave, 596 UCB, Boulder, CO, 80303
E-mail: kristi.anseth@colorado.edu
Tel: 303–735–5336, Fax: 303–492–4341
DOI: 10.1002/adma.201303680
©
wileyonlinelibrary.com
Adv. Mater. 2014, 26, 865–872
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
865

www.advmat.de
www.MaterialsViews.com
Figure 1. a) Table showing the relevant kinetic and thermodynamic information for reactions between N-methyl hydrazine and butyraldehyde or
4-nitrobenzaldehyde at 25 °C in pH 7.4 buffer containing 1% DMSO. b) Chemical structures of 4-H and 4-AA showing reversible gelation. c) Rheological
time sweep plot showing rates of gelation. Evolution of 90% of G∞ occurs in 5 minutes for 4-H:4-AA (thick solid line, G′; thick dashed line G″) and ca.
1 hour for 4-H:4-BA (thin solid line, G′; thin dashed line G″).
In an effort to identify hydrazine and aldehyde constituents
that would react on a timescale compatible with hydrogel forma-
tion and cell encapsulation, we first investigated the kinetic and
thermodynamic properties of small-molecule hydrazones that
result from the reaction of a model aliphatic hydrazine, meth-
ylhydrazine, with two model aldehydes: an aliphatic aldehyde,
butyraldehyde, and an aryl aldehyde, 4-nitrobenzaldehyde.
Hydrazone formation was monitored by UV-vis spectroscopy,
and the data were fitted to a bimolecular reversible kinetic
model.[38] We assume equimolar starting concentrations
of reactants and that the concentration of the hydrazone is
simply the initial concentration of one reactant less the current
concentration of that reactant. These assumptions simplify the
rate equation to Equation (1),
In order to identify a cytocompatible chemistry that would
crosslink hydrogels yet remain dynamic under physiologi-
cally relevant conditions, we looked to the field of dynamic
covalent chemistry, which is comprised of reversible reactions
under thermodynamic control, and include but are not lim-
ited to transesterification, acetal exchange, and hydrazone/
oxime exchange.[35] We were especially drawn to the hydrazone
transimination reaction because it has been used in in vitro
biological systems to assemble enzyme substrates[36] and to per-
form bioconjugations.[37–43] Further, the reaction rate between
the hydrazine/alkoxyamine nucleophiles and aldehyde elec-
trophiles and the stability of the resulting hydrazone/oxime
are highly dependent on the precise chemical structure of the
nucleophilic and electrophilic components; stability half-life
values range from minutes to months at pH 7.0, depending on
the exact functionality.[39,42] Thus, we hypothesized that a hydra-
zone crosslinked hydrogel would show covalently adaptable
behavior, characterized by a frequency-dependent modulus and
stress relaxation, under physiological conditions.
The material presented here is a PEG hydrogel formed
through the ligation of an aliphatic hydrazine-terminated multi-
arm PEG macromer with either a benzaldehyde- or an aliphatic
aldehyde-terminated multi-arm PEG macromer. Gel forma-
tion occurs rapidly at physiological pH and temperature. The
resulting cytocompatible covalently adaptable hydrogel allows
a large degree of tunability of the biochemical and biophysical
microenvironment, captures both the modulus and adaptability
of native tissue, and enables the encapsulation of a diverse
array of cell types. We believe that this material system will
provide a unique platform to culture cells in three-dimensions,
allow for improved reproduction of the stress relaxation found
in native ECM, and permit the determination of how dynamic
changes in the local biophysical environment influence cellular
functions.
d[A]
= −k1[A]2 + k [A]0 − [A]
(
)
−1
(1)
dt
where A is the concentration of either one of the reactants. The
solution to the equation is derived in Dirksen et al. and allows
the calculation of k1, k−1, and K by monitoring the evolution of
the hydrazone bond or the disappearance of one of the reac-
tants.[38] The reaction between methylhydrazine (25 μM) with
butyraldehyde (25 μM) was rapid and reached equilibrium
in less than an hour in pH 7.4 buffer (Figure Supporting
Information, SI 1a). In contrast, the reaction of methylhydra-
zine (25 μM) and 4-nitrobenzaldehyde (25 μM) took several
days to reach equilibrium (Figure SI 1b). The calculated forward
rate constants (k1), back rate constants (k−1), and equilibrium
constants (Keq) are summarized in Figure 1a.
These data provided convincing evidence that both the ali-
phatic and aryl hydrazone bonds would form rapidly enough
to crosslink a cell-laden hydrogel, as the rate constants for
aliphatic and aryl hydrazone formation fall close to the range of
©
866 wileyonlinelibrary.com
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2014, 26, 865–872

www.advmat.de
www.MaterialsViews.com
Figure 2. a) Hydrogel modulus can be controlled by incorporating crosslinkers of different functionalities or varying the stoichiometry of functional
groups (dark bars, on stoichiometry; light bars 50% excess hydrazine, Student’s t-test, p < 0.05). b) 4-H:4-AA shows frequency-dependent crossover
below 0.03 rad/s, which indicates that it behaves as a Maxwellian viscoelastic fluid (filled squares, G′; empty squares, G″; solid line, G′ Maxwell fit;
dashed line, G″ Maxwell fit).[43] c) Stress relaxation is strongly dependent on the ratio of aliphatic to aryl aldehyde crosslinker, with relaxation times
ranging from tens of seconds to tens of hours (bars, 100:0 4-AA:4-BA; triangles, 80:20; circles, 0:100). When the mole fraction of 4-BA crosslinker
crosses the percolation threshold of the system, the stress relaxation behavior collapses to that of 0:100. Stress relaxation of 4-H:4-AA and 4-H:4-BA
are in good agreement with the Maxwell model (dashed lines). Covalent adaptability can be abolished upon treatment by sodium cyanoborohydride,
which reduces the hydrazone bond to the corresponding secondary hydrazine (asterisks). d) Stress relaxation can be shown macroscopically through
molding. 4-H:4-AA and 4-H:4-BA were pressed into square and triangular molds for 60 s, with 4-H:4-AA adopting the shape of the mold as a viscoelastic
fluid and 4-H:4-BA retaining its shape as a more elastic solid (scale bar = 3 mm).
reported rate constants for other hydrogel crosslinking chem-
istry, such as those for dibenzylcyclooctyne-azide (2.1 M−1 s−1)
and vinyl sulfone-thiol couplings (0.17 M−1 s−1).[7,44–47] Thus,
we functionalized 4-arm 20 kDa and 8-arm 10 kDa PEG
macromers with aliphatic hydrazine end groups (4-H, 8-H) and
4-arm 10 kDa and 8-arm 10 kDa PEG macromers with either
benzaldehyde (4-BA, 8-BA) or aliphatic aldehyde (4-AA, 8-AA)
end groups. Gram-scale quantities of the macromers are syn-
thetically available in 1–2 steps from inexpensive commercially
available precursors, which renders this system accessible to a
wide range of biomedical, chemical, and polymer engineering
labs. Mixing 18 mM (72 mM functional groups) solutions of
either 4-H and 4-AA (Figure 1b) or 4-H and 4-BA resulted in
the formation of a hydrogel at room temperature and pH 7.4.
The rheological properties of the gels were investigated under
shear, and the results are shown in Figures 1 and 2. 4-H:4-AA
reaches its crossover point in tens of seconds. Evolution of 90%
of its final elastic modulus is achieved in 5 minutes, which can
be predicted by fitting modulus evolution data to Equation (2),
In addition to exerting control over the stress relaxation
characteristics of these materials, we can also control equilib-
rium modulus by varying the size and functionality of each
macromer. Furthermore, because the crosslinking density is
linearly related to modulus through Equation (3),
G = kT Q−1/ 3Dx
(3)
where k is Boltzmann constant, T is the temperature, Q is
the volume swelling ratio, and ρx is the crosslinking den-
sity, the final material modulus can be systematically varied
to target a tissue or magnitude of interest. In Figure 2a, the
swollen shear modulus varied from ∼600 Pa (4-H:4-AA, 2:1)
to ∼9 kPa (8-H:8-AA), and all of the values are in reasonable
agreement with that predicted by rubber elasticity theory. While
non-idealities certainly exist, Equations (4) and (7) enable the
rational design of a polymeric viscoelastic material that could
serve as a biophysical mimic for complex tissues.
Frequency sweeps of 4-H:4-AA (Figure 2b) and 4-H:4-BA
(Figure SI 2b) indicate that the 32-fold difference in k−1 dra-
matically changes the ability of the materials to adapt to strain.
4-H:4-AA exhibits a frequency-dependent modulus that is char-
acteristic of a single-mode Maxwell viscoelastic fluid, whereas
4-H:4-BA exhibits a frequency-independent modulus indica-
tive of a Hookean solid. Presumably, the angular frequency
at which G″ crosses G′ is lower than could be accurately
G(t) = G∞ − Ae−Bt
(2)
where G∞ is the plateau modulus and A and B are fit para-
meters. 4-H:4-BA reaches its final modulus in ca. 1 hour at
pH 7.4, but this process is highly pH dependent (Figure 1c,
Figure SI 2a).
©
wileyonlinelibrary.com
Adv. Mater. 2014, 26, 865–872
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
867

www.advmat.de
www.MaterialsViews.com
measured. Using Equations (4) and (5) for a Maxwellian viscoe-
lastic fluid,[48]
be applied, which precluded accurate measurement of the time
constant. Mixed crosslink gels containing a fraction of benzal-
dehyde crosslinkers below the percolation threshold, calculated
using Equations (7) and (8),[50–52]
G
T2J2
∞
ꢀ ( )
,
G T =
(4)
1 + T2 J2
1
pc = ꢀ
,
(7)
(8)
(
)(
−1
)
G TJ
f
f
−1
A
B
∞
ꢀꢀ ( )
G
T =
r
(5)
1 + T2 J2
where G∞ is the equilibrium modulus, ω is the angular
frequency, and τ is the relaxation time; we find τ = 34 s
(Figure 2b). This agreement with the Maxwell model indicates
that these gels should relax an applied stress on a time scale
on the order of tens of seconds and that by varying the ratio
of benzaldehyde to aliphatic aldehyde crosslinker, one should
retain a great deal of control over the adaptation properties of
the material.
fBnB
fAnA
r =
which equals 0.33 for 4-H:4-AA and 4-H:4-BA, are modeled rea-
sonably well in their early and late stages by summing both the
aliphatic aldehyde and the benzaldehyde modes (Figure SI 4).
However, once the gel is comprised of more than 33% 4-BA
crosslinks, the stress relaxation behavior of these materials
collapses to that of 4-H:4-BA. Finally, when the gel is treated
with sodium cyanoborohydride, the hydrazone bond is reduced
to the corresponding hydrazine, thus abolishing the dynamic
nature of the bond and, consequently, the relaxation behavior
(Figure 2c).
Thus, we next polymerized six gels of varying compositions,
ranging from 100% 4-AA crosslinks to 100% 4-BA crosslinks
and strained them 100% on the rheometer (Figure 2c). Gels
crosslinked by 100% aliphatic aldehyde macromers relaxed
100% of their stress within tens of seconds, while those
crosslinked by 100% benzaldehyde macromers took closer to
14 hours to relax 75% of their stress, although it is likely there
is some error associated with these measurements due to
instrument slippage (Figure SI 3a). We observed that the stress
relaxation profile can be adjusted by changing the ratio of ali-
phatic aldehyde to benzaldehyde macromer: as the proportion of
benzaldehyde macromer increases, the hydrogel relaxes stress
more slowly, until a critical point, the percolation threshold (vide
infra), is reached (Figure 2c). Thus, by tuning these characteris-
tics — the number of arms on the PEG polymer, the choice of
either aliphatic aldehyde or benzaldehyde end groups, the ratio
of the aliphatic to aryl aldehyde-derivatized macromers, and the
relative stoichiometry of hydrazine and aldehyde components
— it is possible to precisely control the physiological modulus
and stress relaxation characteristics of the gel to tailor the gel
characteristics specifically to the tissue under study.
The self-healing, dynamic characteristics of 4-H:4AA and
static nature of 4-H:4-BA were also visualized macroscopi-
cally. Figure 2d shows two disk-shaped hydrogels formed using
a syringe mold 5 mm in diameter. Both hydrogels were then
pressed into a 4 mm × 4 mm square mold for 60 s, re-imaged,
pressed into a 6 mm × 7 mm right triangular mold for 60 s,
and imaged again. These qualitative images clearly demon-
strate that 4-H:4-AA has the capacity to rearrange its macro-
scopic network structure to relax stresses applied by the mold
within 60 s, allowing it to be re-processed into a new macro-
scopic shape. On the contrary, 4-H:4-BA is immutable on this
time scale and retains its original disk shape throughout the
processing. Though the aryl hydrazone bonds can relax stress,
albeit on a much longer time scale than the aliphatic hydra-
zone bonds, we observed no shape change, even after 1 hour.
Figure SI 5 further explores these macroscopic demonstrations
of stress relaxation and shows the fusion of three gels into a
larger shape, their molding into other shapes, and finally their
injection through an 18 G needle to form a long extruded fiber.
We next sought to directly compare the viscoelastic prop-
erties of these hydrogel networks to that of a complex tissue,
namely a mouse gastrocnemius muscle. We selected muscle for
comparison, as it is widely known to relax stress in two separate
modes, with a fast λ varying from 0.03 s to 8.4 s and a slow λ var-
ying from 2.2 s to 93.8 s, depending on the species.[53] Further,
muscle has a Young's modulus around 10 kPa, which is well
within the range of the moduli for the PEG networks that were
synthesized here (1.8 – 27 kPa).[18,53–55] A mouse gastrocnemius
muscle was isolated and characterized using shear rheology.
The results show a highly frequency-dependent elastic modulus
between 2–3 kPa and relaxation of 80% of an applied stress
within 600 seconds (Figure 3a, 3b). Consistent with mechan-
ical theories of fiber containing composite materials,[56] we
hypothesized that the steep reduction in modulus below 1 rad/s
shear was due primarily to the fibrous structure of skeletal
muscle, rather than formation and reformation of physical
crosslinks. Interestingly, the stress relaxing hydrogels capture
Next, we fit the relaxation data to Equation (6), a standard
Maxwell model for viscoelastic fluids,[49]
t
N
−
ꢀ
8k
0k
G (t) =
e
+ G
∞
(6)
8k
k= 1
where G∞ is the elastic modulus, η is the viscosity, and λ
is the time constant of relaxation. While all adaptable gels
should have a polymer relaxation mode and a bond relaxation
mode, the rapid relaxation of the polymer chains (typical λ =
0.05 s) was not captured in the rheology. The 4-arm-co-4-arm
(4-H:4-AA) aliphatic gel fit the model well with a single mode
of relaxation (λ = 10.7 s), as did the 8-arm-co-8-arm (8-H:8-AA)
aliphatic gel (λ = 91.0 s). These numbers are consistent with
the fact that at equal weight percent of PEG, the 8-arm-co-
8-arm gels contain 8-fold more crosslinks, leading to a roughly
8-fold slower time constant of relaxation. The benzaldehyde
crosslinked gels did not fit the model as well, which is likely
a result of slow relaxation competing with instrument slip.
However, the 4-H:4-BA gel relaxed with a time constant of λ =
2370 s, while the 8-H:8-BA gel tore before 100% strain could
©
868 wileyonlinelibrary.com
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2014, 26, 865–872

www.advmat.de
www.MaterialsViews.com
Figure 3. The rheological properties of complex tissue can be emulated using covalently adaptable hydrogels. a) The elastic modulus of mouse muscle
(thin solid line, G′; thin dashed line, G″) begins to sharply decrease below 1 rad/s, possibly due to the fibrous nature of the tissue, while the 8-H:8-AA
(thick solid line, G′; thick dashed line, G″) exhibits this property below 0.05 rad/s. However, G∞ of the mouse muscle can be exactly matched. b) The
mouse muscle (triangles) demonstrates significant stress relaxation over 10 minutes and fits a two-mode Maxwell model (solid line). 8-H:8-AA (dia-
monds) comes close to reproducing the initial and final stress with just one mode of relaxation. 8-H:8-BA (squares) relaxes much more slowly.
aspects of this behavior through breaking and reforming of
the material crosslinks, and we observed two discrete modes of
relaxation and fit them to a two-mode Maxwell model.[53,54,57]
This behavior is readily seen in Figure 3b, where the applied
stress initially decays rapidly and then slows. In contrast, the
trace of the 100% 8-H:8-AA gel displays a single mode of relaxa-
tion. When fit to the Maxwell model, muscle relaxes with time
constants of λ = 7.5 s and λ = 100 s with a ca. 55% weighting
of the slow-relaxing mode and a ca. 45% weighting of the fast-
relaxing mode. These values are consistent with published
measurements of tensile muscle relaxation,[53,54] and they fall
within the range of the relaxation time constants of the 4-arm
(λ = 10.7 s) and 8-arm aliphatic gels (λ = 91.0 s). Collectively,
these results suggest that covalently adaptable networks impart
unique viscoelastic properties that mimic aspects of native
tissue materials that are difficult to capture with traditional
synthetic hydrogels. Further, by tuning the initial macromer
composition and concentration (8-H:8-AA, 2:1 ratio), gels were
formulated with initial and time dependent properties very
similar to those of the gastrocnemius.
We next sought to demonstrate the cytocompatibility of the
reaction and encapsulation processes and examine how cells
would respond to longer-term culture in these covalently adapt-
able networks. Because of our characterization of mouse skel-
etal muscle, this compelled us to examine a cell line, C2C12s,
a common source of myoblasts that fuse into myotubes under
confluent conditions. For cell encapsulation studies, we formu-
lated three hydrogels using the hydrazine- and aldehyde-termi-
nated PEG macromers in combination with a benzaldehyde-
terminated KGRGDS peptide, which provides sites for integrin
binding and promotes cell survival in PEG gels.[58–61] Specifi-
cally, one gel was synthesized with 100% aliphatic crosslinks
(8-H:8-AA), another with 80% aliphatic crosslinks and 20% aryl
aldehyde crosslinks (8-H:(80% 8-AA 20% 8-BA)), and a final
gel with 100% aryl aldehyde crosslinks (8-H:8-BA). All gels
contained 3 mM KGRGDS peptide. We selected these specific
compositions to form highly adaptable gels that would remain
stable for several weeks of cell culture (Figure SI 3b) yet allow a
significant variation in stress relaxation properties, from highly
relaxing to minimally relaxing. C2C12 myoblasts were encap-
sulated at a concentration of 106 cells/mL, and the cells were
stained and imaged for viability after 24, 72, and 240 hours
(Figure 4a, 4b, Figure SI 6) and cytoskeletal organization at 24,
96, 144, and 264 hours (Figure 4c, Figure SI 7).
First, we note high levels of cell survival after encapsulation,
thus validating the applicability of this crosslinking chem-
istry for the conditions studied (Figure 4a, 4b). Calcein-AM
and ethidium homodimer staining indicated that viability was
higher than 90% in all three conditions after 72 h (Figure 4b),
which confirms that this hydrogel platform is cytocom-
patible and appropriate for studying cellular responses to
dynamic, tissue-mimicking biophysical environments. Fur-
thermore, the significant drop in cell viability in 8-H:8-BA
after 10 days in culture implies that matrix dynamism could
play an important role in long-term cell viability (Figure 4b).
Cells maintain a rounded morphology when encapsulated
in covalently crosslinked hydrogels, so we next sought to
examine how cells would behave in hydrogels in which the
crosslinks dynamically break and re-form.[10,14] Interestingly,
we observed that after only 24 hours in culture, a minority
of the cells encapsulated in the 8-H:8-AA gels extended small
filopodia and lamellipodia, while those in the 8-H:(80% 8-AA
20% 8-BA) and 8-H:8-BA gels retained their circular mor-
phology. By 72 hours, however, nearly all cells in the 8-H:8-AA
gels began to extend processes, while those in the 8-H:8-BA
gels remained rounded; the 8-H:(80% 8-AA 20% 8-BA) gels
contained cells of both phenotypes (Figure SI 6). At later
time points, the filopodia and lamellipodia became more
pronounced, and essentially all cells in both the 8-H:8-AA
and 8-H:(80% 8-AA 20% 8-BA) gels had begun to spread. By
10 days in culture, the cells in the 8-H:8-AA gels demonstrated
the spread myoblast phenotype (Figure 4c, Figure SI 7), and
some cells with close neighbors began to fuse into multinu-
cleated structures demonstrating a myotube-like morphology
with continuous actin filaments spanning the entire mass
(Figure 4d). This myoblast fusion suggests the beginning
©
wileyonlinelibrary.com
Adv. Mater. 2014, 26, 865–872
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
869

www.advmat.de
www.MaterialsViews.com
Figure 4. a) Images of encapsulated C2C12 myoblasts 72 hours post-encapsulation stained with calcium-AM (green, live) and ethidium homodimer
(red, dead; scale bar 200 μm). b) Cell viability quantified at 24 hours (dark bars), 72 hours (medium bars), and 240 hours (light bars). c) Representative
images of cells stained for f-actin (red) and the nucleus (blue) are shown (scale bar 20 μm). After 10 days in culture, the cells encapsulated in 8-H:8-AA
exhibit significant spreading and robust actin fiber formation. Cells encapsulated in 8-H:(80% 8-AA,20% 8-BA) extend lamellipodia and filopodia and
show actin filaments but do not deviate as far from rounded. Cells encapsulated in 8-H:8-BA remain rounded. d) Some cells in the 8-H:8-AA gels fuse
into multinuclear structures. e) Quantification of cell spreading (dark bars, left axis) and fraction of cells extending processes (light bars, right axis)
by 10 days in culture. Cells in 8-H:8-AA and 8-H:(80% 8-AA,20% 8-BA) both displayed significantly more spreading, as measured by projected cell area
and process extension, than those in 8-H:8-BA (Student’s t-test, p<0.05).
stages of differentiation; this process did not happen to cells
in 8-H:8-BA. C2C12s in 8-H:(80% 8-AA 20% 8-BA) displayed
significant process extension, but while their mean projected
cell area was not significantly different from the cells in
8-H:8-AA, they retained a more rounded morphology. Fur-
thermore, we observed fewer fused multinuclear structures.
Cells in the less-dynamic 8-H:8-BA gels remained rounded,
even after 10 days with less than 30% of cells extending pro-
cesses into the material (Figure 4c). These results were quan-
tified by measuring the mean projected cell area and counting
the number of cells extending processes into the material. We
found that cells in 8-H:8-AA and 8-H:(80% 8-AA 20% 8-BA)
both displayed significantly higher mean cell projected areas
and a higher fraction of cells extending processes than cells
in 8-H:8-BA (Figure 4e). Based on these three-dimensional
observations, we propose that the dynamic nature of the ali-
phatic hydrazone linked hydrogel allows the gel to rearrange
on a timescale that is compatible with filopodia and lamel-
lipodia extension, and thus, the material can accommodate
cytoskeletal outgrowth. In contrast, the more slowly relaxing
gels, such as 8-H:8-BA, produce a much more static matrix
that constrains the cells to a circular morphology throughout
the course of the experiment. Rheological experiments on
cell-laden gels show that the incorporation of cells lowers
the equilibrium modulus but does not significantly change
the stress relaxation properties (Figure SI 8a and 8b). SEM
imaging reveals that the hydrogels are amorphous solids
lacking pores or ordered structure (Figure SI 9). Given the
constancy of the stress relaxing properties and the lack of
pores or order, it is likely that encapsulated cells apply stress,
which leads to a dynamic rearrangement of the hydrazone
bonds and allows them to spread out in the material.
Here, we examined the simulation of muscle and encapsu-
lation of myoblasts, but this cytocompatible covalently adapt-
able hydrogel could be applied to other cellular systems where
viscoelasticity or careful control of the scaffold biophysical
properties is desirable. For instance, it is known that reduced
crosslinking in the pericellular environment induces increased
proliferation in myofibroblasts,[15] and this covalently adaptable
hydrogel could be used to quantify the degree of stress neces-
sary to inhibit proliferation and the kinetics of the associated
signaling pathways. Scaffold mechanics and cell shape have
both been implicated in progenitor cell differentiation,[18,62,63]
but decoupling these two phenomena is challenging in cell
degradable gels. Covalently adaptable gels allow the study of
progenitor cells in a defined biophysical environment while
simultaneously allowing cells to adopt their preferred geom-
etries. Finally traditional covalent gels force secreted ECM
molecules into a confined geometry,[5] preluding the assembly
of micron-size features like collagen fibers. Covalently adapt-
able networks may prove a useful system to respond and
accommodate the assembly of secreted ECM and subsequent
collagen fiber formation.
We have developed a cytocompatible covalent adaptable
hydrogel capable of mimicking the modulus and stress relaxa-
tion properties of many complex biological tissues. Here, we
have described a material that simulates some of the proper-
ties of native mouse muscle and demonstrated its utility as a
©
870 wileyonlinelibrary.com
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2014, 26, 865–872

www.advmat.de
www.MaterialsViews.com
[16] T. P. Kraehenbuehl, P. Zammaretti, A. J. van der Vlies, R. G. Schoen-
makers, M. P. Lutolf, M. E. Jaconi, J. A. Hubbell, Biomaterials 2008,
29, 2757–2766.
[17] S. T. Lee, J. I. Yun, Y. S. Jo, M. Mochizuki, A. J. van der Vlies,
S. Kontos, J. E. Ihm, J. M. Lim, J. A. Hubbell, Biomaterials 2010, 31,
1219–1226.
scaffold to culture mouse myoblasts. The modular nature of
hydrogel construction and the large number of easily tuned
variables provide access to gels with a wide range of modulus
and stress relaxation characteristics; the same approach could
potentially be applied to nearly any tissue in the body. C2C12
cell encapsulation studies demonstrate that these covalently
adaptable hydrogels allow for the development of physiologi-
cally relevant morphologies, whereas static, non-adaptable
gels prevent cytoskeletal rearrangement and extension. Taken
together, these studies show that hydrazone linked hydrogels
offer unique advantages in terms of dynamic tunability and
should serve as a valuable complement to existing hydrogel
technologies.
[18] A. Engler, S. Sen, H. Sweeney, D. E. Discher, Cell 2006, 126,
677–689.
[19] A. M. Kloxin, J. A. Benton, K. S. Anseth, Biomaterials 2010, 31, 1–8.
[20] J. Zoldan, E. D. Karagiannis, C. Y. Lee, D. G. Anderson, R. Langer,
S. Levenberg, Biomaterials 2011, 32, 9612–9621.
[21] M. Guvendiren, J. A. Burdick, Nat. Commun. 2012, 3, 792.
[22] L. Weng, A. Romanov, J. Rooney, W. Chen, Biomaterials 2008, 29,
3905–3913.
[23] L. Weng, X. Chen, W. Chen, Biomacromolecules 2007, 8, 1109–1115.
[24] H. Zhang, A. Qadeer, W. Chen, Biomacromolecules 2011, 12,
1428–1437.
Supporting Information
[25] S. K. Bhatia, S. D. Arthur, H. K. Chenault, G. K. Kodokian,
Biotechnol. Lett. 2007, 29, 1645–1649.
[26] J. Dahlmann, A. Krause, L. Möller, G. Kensah, M. Möwes,
A. Diekmann, U. Martin, A. Kirschning, I. Gruh, G. Dräger, Biomate-
rials 2013, 34, 940–951.
Supporting Information is available from the Wiley Online Library or
from the author.
[27] X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt, K. Sheran,
F. Wudl, Science 2002, 295, 1698–1702.
[28] C. N. Bowman, C. J. Kloxin, Angew. Chem. Int. Ed. 2012, 4272–4274.
[29] G. Deng, C. Tang, F. Li, H. Jiang, Y. Chen, Macromolecules 2010, 43,
1191–1194.
[30] C. J. Kloxin, T. F. Scott, B. J. Adzima, C. N. Bowman, Macromolecules
2010, 43, 2643–2653.
[31] B. J. Adzima, C. J. Kloxin, C. N. Bowman, Adv. Mater. 2010, 22,
2784–2787.
[32] T. F. Scott, A. D. Schneider, W. D. Cook, C. N. Bowman, Science
2005, 308, 1615–1617.
[33] C. J. Kloxin, T. F. Scott, H. Y. Park, C. N. Bowman, Adv. Mater. 2011,
23, 1977–1981.
[34] M. C. Roberts, M. C. Hanson, A. P. Massey, E. A. Karren, P. F. Kiser,
Adv. Mater. 2007, 19, 2503–2507.
Acknowledgements
The authors acknowledge Martin Guess in Prof. Leslie Leinwand’s lab
for the generous gift of the C2C12 cells and the mouse gastrocnemius
muscle, Drs. Malar Azagarsamy and Daniel Alge for helpful discussions
related to chemistry and material properties, and Dr. Kelly Schultz for
her invaluable input regarding rheological characterization. These
studies were supported by a grant from the NSF (DMR 1006711), the
NIH Pharmaceutical Biotechnology Training Program (5 T32 GM 8732),
Office of Naval Research (N00014–13–1–0283), and the Howard Hughes
Medical Institute.
Received: August 7, 2013
Revised: August 23, 2013
Published online: October 11, 2013
[35] S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders,
J. F. Stoddart, Angew. Chem. Int. Ed. 2002, 41, 898–952.
[36] V. T. Bhat, A. M. Caniard, T. Luksch, R. Brenk, D. J. Campopiano,
M. F. Greaney, Nat. Chem. 2010, 2, 490–497.
[37] G. Iyer, F. Pinaud, J. Xu, Y. Ebenstein, J. Li, J. Chang, D. Maxime,
S. Weiss, Bioconjugate Chem. 2011, 22, 1006–1011.
[38] A. Dirksen, S. Dirksen, T. M. Hackeng, P. E. Dawson, J. Am. Chem.
Soc. 2006, 128, 15602–15603.
[39] A. Dirksen, S. Yegneswaran, P. E. Dawson, Angew. Chem. Int. Ed.
2010, 49, 2023–2027.
[40] A. Dirksen, P. E. Dawson, Bioconjugate Chem. 2008, 19, 2543–2548.
[41] A. R. Blanden, K. Mukherjee, O. Dilek, M. Loew, S. L. Bane, Biocon-
jugate Chem. 2011, 22, 1954–1961.
[42] J. Kalia, R. T. Raines, Angew. Chem. Int. Ed. 2008, 47, 7523–7526.
[43] D. W. Domaille, J. H. Lee, J. N. Cha, Chem. Commun. 2013, 49,
1759–1761.
[44] M. Karver, R. Weissleder, Angew. Chem. Int. Ed. 2012, 51, 920–922.
[45] J. W. Bae, E. Lee, K. M. Park, K. D. Park, Macromolecules
2009.
[1] B. V. Slaughter, S. S. Khurshid, O. Z. Fisher, A. Khademhosseini,
N. A. Peppas, Adv. Mater. 2009, 21, 3307–3329.
[2] J. L. Drury, D. J. Mooney, Biomaterials 2003, 24, 4337–4351.
[3] O. Guillame-Gentil, O. Semenov, A. S. Roca, T. Groth, R. Zahn,
J. Vörös, M. Zenobi-Wong, Adv. Mater. 2010, 22, 5443–5462.
[4] S. J. Bryant, K. S. Anseth, J. Biomed. Mater. Res. 2001, 59, 63–72.
[5] S. J. Bryant, K. S. Anseth, J. Biomed. Mater. Res. 2002, 64A, 70–79.
[6] S. J. Bryant, K. L. Durand, K. S. Anseth, J. Biomed. Mater. Res. 2003,
67A, 1430–1436.
[7] M. P. Lutolf, J. A. Hubbell, Biomacromolecules 2003, 4, 713–722.
[8] B. D. Fairbanks, M. P. Schwartz, A. E. Halevi, C. R. Nuttelman,
C. N. Bowman, K. S. Anseth, Adv. Mater. 2009, 21, 5005–5010.
[9] J. Benton, B. Fairbanks, K. S. Anseth, Biomaterials 2009, 30,
6593–6603.
[10] S. B. Anderson, C. C. Lin, D. V. Kuntzler, K. S. Anseth, Biomaterials
2011, 3564–3574.
[46] J. Xu, T. M. Filion, F. Prifti, J. Song, Chem. Asian J. 2011, 6,
2730–2737.
[11] K. A. Kyburz, K. S. Anseth, Acta Biomater. 2013, DOI 10.1016/j.
actbio.2013.01.026.
[47] J. Zheng, L. A. Smith Callahan, J. Hao, K. Guo, C. Wesdemiotis,
R. A. Weiss, M. L. Becker, ACS Macro Lett. 2012, 1, 1071–1073.
[48] T. Annable, J. Rheol. 1993, 37, 695–726.
[49] F. A. Morrison, Understanding Rheology, Oxford University Press,
USA 2001.
[12] M. P. Schwartz, B. D. Fairbanks, R. E. Rogers, R. Rangarajan,
M. H. Zaman, K. S. Anseth, Integr. Biol. 2010.
[13] J. Shepard, A. Stevans, S. Holland, C. E. Wang, A. Shikanov,
L. D. Shea, Biotechnol. Bioeng. 2012, 109, 830–839.
[14] D. D. McKinnon, A. M. Kloxin, K. S. Anseth, Biomaterials Science
2013, 1, 460–469.
[15] J. Patterson, J. A. Hubbell, Biomaterials 2010, 31, 7836–7845.
[50] P. J. Flory, J. Am. Chem. Soc. 1941, 63, 3083–3090.
[51] P. J. Flory, J. Phys. Chem. 1942, 46, 132–140.
©
wileyonlinelibrary.com
Adv. Mater. 2014, 26, 865–872
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
871

www.advmat.de
www.MaterialsViews.com
[60] C. R. Nuttelman, M. C. Tripodi, K. S. Anseth, Matrix Biol. 2005, 24,
208–218.
[61] E. A. Phelps, N. O. Enemchukwu, V. F. Fiore, J. C. Sy, N. Murthy,
T. A. Sulchek, T. H. Barker, A. J. García, Adv. Mater. 2011, 24,
64–70.
[52] W. H. Stockmayer, J. Chem. Phys. 1943, 11, 45.
[53] B. C. Abbott, J. Lowy, Proc. R. Soc. B 1957, 146, 281–288.
[54] A. Magid, D. J. Law, Science 1985.
[55] A. J. Engler, M. A. Griffin, S. Sen, C. G. Bönnemann, H. L. Sweeney,
D. E. Discher, J. Cell Biol. 2004, 166, 877–887.
[62] R. McBeath, D. M. Pirone, C. M. Nelson, K. Bhadriraju, C. S. Chen,
Dev. Cell 2004, 6, 483–495.
[63] K. A. Kilian, B. Bugarija, B. T. Lahn, M. Mrksich, Proc. Natl. Acad. of
Sci. 2010, 107, 4872–4877.
[56] S. Onogi, K. Ui, Journal of Colloid Science 1956, 11, 214–225.
[57] K. W. Ranatunga, J. Muscle Res. Cell Motil. 2001, 22, 399–414.
[58] M. D. Pierschbacher, E. Ruoslahti, Nature 1984, 309, 30–33.
[59] J. A. Burdick, K. S. Anseth, Biomaterials 2002, 23, 4315–4323.
©
872 wileyonlinelibrary.com
2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2014, 26, 865–872