Acta Biomaterialia 8 (2012) 18381848

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Photopolymerization of cell-encapsulating hydrogels: Crosslinking efciency

versus cytotoxicity
Iris Mironi-Harpaz a, Dennis Yingquan Wang b, Subbu Venkatraman b, Dror Seliktar a,
a
Faculty of Biomedical Engineering, Technion Israel Institute of Technology, Haifa 32000, Israel
b
School of Materials Science and Engineering, Nanyang Technological University, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: Cell-encapsulating hydrogels used in regenerative medicine are designed to undergo a rapid liquid-to-
Received 30 March 2011 solid phase transition in the presence of cells and tissues so as to maximize crosslinking and minimize
Received in revised form 17 December 2011 cell toxicity. Light-activated free-radical crosslinking (photopolymerization) is of particular interest in
Accepted 30 December 2011
this regard because it can provide rapid reaction rates that result in uniform hydrogel properties with
Available online 13 January 2012
excellent temporal and spatial control features. Among the many initiator systems available for photopo-
lymerization, only a few have been identied as suitable for cell-based hydrogel formation owing to their
Keywords:
water solubility, crosslinking properties and non-toxic reaction conditions. In this study, three long-wave
Scaffold
Biocompatibility
ultraviolet (UV) light-activtied photoinitiators (PIs) were comparatively tested in terms of cytotoxicity,
Toxicity crosslinking efciency and crosslinking kinetics of cell-encapsulating hydrogels. The hydrogels were pho-
Tissue engineering topolymerized from poly(ethylene glycol) (PEG) diacrylate or PEGbrinogen precursors using Irgacure
Mechanical properties PIs I2959, I184 and I651, as well as with a chemical initiator/accelerator (APS/TEMED). The study specif-
ically evaluated the PI type, PI concentration and UV light intensity, and how these affected the mechan-
ical properties of the hydrogel (i.e. maximum storage modulus), the crosslinking reaction times and the
reactions cytotoxicity to encapsulated cells. Only two initiators (I2959 and I184) were identied as being
suitable for achieving both high cell viability and efcient crosslinking of the cell-encapsulating hydro-
gels during the photopolymerization reaction. Optimization of PI concentration or irradiation intensity
was particularly important for achieving maximum mechanical properties; a sub-optimal choice of PI
concentration or irradiation intensity resulted in a substantial reduction in hydrogel modulus.
Cytocompatibility may be compromised by unnecessarily prolonging exposure to cytotoxic free radi-
cals or inadvertently enhancing the instantaneous dose of radicals in solution, both of which are depen-
dent on the PI type/concentration and irradiation intensity. In the absence of a radical initiator, the short
exposures to long-wave UV light irradiation (up to 5 min, 20 mW cm2, 365 nm) did not prove to be cyto-
toxic to cells. Therefore, it is important to understand the relationship between PIs, light irradiation con-
ditions and crosslinking when attempting to identify a suitable hydrogel formation process for cell
encapsulating hydrogels.
2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction structural support for cells, enable proper diffusion of metabolites

Hydrogels are hydrophilic polymeric networks that are widely
used in a variety of applications [1]. Their high water content mim-
ics the permeability of the extracellular matrix (ECM) for optimal
transport of oxygen, nutrients and waste products, making them
ideal for medical applications. Depending on their composition,
hydrogels can be designed with cellular biocompatibility and thus
can be utilized as cell carriers in cell therapy [2], cell scaffolds for
tissue engineering [3], delivery vehicles for drug molecules [4]
and conduits for guided in vivo tissue regeneration [5]. The three-
dimensional (3-D) hydrogels in cell therapy, for example, provide
Corresponding author. Tel.: +972 4 829 4805; fax: +972 4 829 4599.
E-mail address: dror@bm.technion.ac.il (D. Seliktar).
and can offer immunoisolation or local protection from host
inammation [6].
Hydrogels are usually classied into categories depending on
their macromolecular precursors, and the crosslinking reaction
that forms the network. Hydrogels can be formed from biological,
synthetic and semi-synthetic hybrid macromolecules that are
physically or chemically crosslinked in the presence of water [7].
Biological hydrogels are inherently bioactive, but have limitations
in controlling physical properties, which is essential for many bio-
medical applications. Synthetic polymer hydrogels, which do offer
control over physical properties, are devoid of biological signaling
for cells and tissues in the body, rendering them biologically inert.
The integration of biological and synthetic polymers to form a
biosynthetic hydrogel network permits both controlled physical

1742-7061/$ - see front matter 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actbio.2011.12.034
 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848
properties and bioactivity for various medical uses [8]. It is the
unique combination of control and bioactivity that makes biosyn-
thetic hydrogels particularly suitable for cell therapy and as scaf-
folds in tissue engineering, by enabling the encapsulation of cells
within the hydrogel network during the crosslinking reaction [9].
There are several methods to crosslink liquid macromolecular
precursors into a hydrogel, including ionic crosslinking [10],
enzyme-triggered gelation [11], thermally induced gelation [12]
and covalent crosslinking [13]. Light-activated free-radical cross-
linking of hydrogels is of particular interest in biomedical applica-
tions requiring in situ, non-toxic gelation. Here, free radicals lead
to the formation of covalent bonds between macromolecular
precursor molecules in a rapid reaction that results in the forma-
tion of a polymer network with uniform and replicable physical
properties [14]. Long-wave ultraviolet (UV) light-activated
polymerization is one of the most common methods of forming
biomedical hydrogels because of the advantages of temporal and
spatial control of the reaction, low energy requirements, and
clinically acceptable curing times [9,1517]. Photopolymerization
can be easily carried out at physiological temperature and pH,
in vitro or in vivo; the latter enabling the use of photopolymerized
hydrogels with minimally invasive surgical procedures.
Photopolymerization of biomedical hydrogels is typically
carried out using a non-toxic photoinitiator (PI) and irradiation
with the appropriate wavelength, whereby free radicals dissociate
from the PI upon irradiation [18]. These radicals attack vinyl groups
on precursor macromolecules with high efciency, creating cova-
lent bonds that crosslink the hydrogel network within seconds to
minutes of UV light exposure. There are a couple of shortcoming
to PIs used with biomedical hydrogels, including limited water sol-
ubility and sub-optimal crosslinking efciency when applied with
safe light spectra (i.e. P350 nm) [19]. Despite these shortcoming
PIs have been applied successfully for the rapid in situ gelation of
hydrogels, including constructs with spatial anisotropy and pre-de-
ned gradients [2023].
A rapid chemical crosslinking reaction is a particularly impor-
tant consideration when forming a hydrogel that is to be used for
cell encapsulation in tissue engineering and cell therapy. Although
the high efciency and homogeneity of free-radical crosslinking
allows for greater ne-tuning over the porosity and the crosslink-
ing density of the hydrogel, the radical species present during
photo-initiation may adversely interact with cells, and may be det-
rimental to cell survival [24]. In addition to cell-radical interac-
tions, the use of toxic solvents and photoinitiators, intense UV
light irradiation and high temperatures resulting from a photopo-
lymerization process can potentially cause cell damage [9]. Bryant
et al. investigated the viability of NIH/3T3 cells with different PIs
and PI concentrations, and concluded that one photoinitiating
system in particular, 2-hydroxy-1-[4-(hydroxyethoxy)phenyl]-2-
methyl 1-propanone (Irgacure 2959) and long-wave UV light,
showed promise as a minimally toxic system for cell encapsulation
in hydrogels [25]. Williams et al. conrmed that the same PI sys-
tem was cytocompatible with several different cell types [26]. In
these studies, cell viability was tested by exposing the cells for sev-
eral minutes to direct low intensity long-wave UV light in the pres-
ence of the PI in solution.
It is important that cytotoxicity of PIs and crosslinking reactions
are investigated in the context of the hydrogel network formation.
An optimal crosslinking reaction produces a hydrogel that exhibits
good physical properties via adequate crosslinking, while ensuring
maximum cell survival by minimizing exposure to toxic reaction
products. In the photopolymerization of cell-encapsulating hydro-
gels, this balance can be achieved through a better understanding
of the interplay between the radical crosslinking efciency,
hydrogel network formation and cytotoxicity. Therefore, the
current study seeks to outline a relationship between hydrogel
1839
mechanical properties and cell viability after cell encapsulation
by photopolymerization. For this purpose, a biosynthetic hybrid
hydrogel synthesized from functionalized poly(ethylene glycol)
(PEG) conjugated to brinogen was used to encapsulate bovine
aortic smooth muscle cells (SMCs) by photopolymerization [27].
Using three different free-radical PIs with two different solvents
that were chosen based on water solubility, crosslinking properties
and non-toxic reaction conditions, we sought to identify the conse-
quence of the photopolymerization reaction on the hydrogels
maximum storage modulus after crosslinking, and the viability of
the encapsulated SMCs. We further compare the photopolymeriza-
tion results with a conventional non-light-activated radical
crosslinking reaction.
2. Materials and methods
2.1. PEG-diacrylate synthesis
PEG-diacrylate (PEG-DA) was synthesized from linear PEG-OH
(Fluka, Buchs, Switzerland) with an average molecular weight of
10 kDA similar to what has been described elsewhere [28]. Briey,
a solution of PEG-OH in dichloromethane was reacted under argon
with acryloyl chloride (Merck, Darstadt, Germany) and triethyl-
amine (Fluka) at a acryloyl chloride:OH molar ratio of 1.5:1. The
product was precipitated in ice-cold diethyl ether and dried under
vacuum for 2 days. The nal product was characterized by proton
nuclear magnetic resonance imaging to evaluate its purity and to
calculate the per cent conversion of functional acrylate groups on
the polymer chains.
2.2. Fibrinogen PEGylation
The hydrogels used in the present study are based on protein
polymer adducts that are formed according to protocols described
previously [29]. PEGylation of the brinogen protein was per-
formed according to a published protocol whereby a solution of
7 mg ml1 brinogen (MP Biomedical, OH, USA) in 150 mM phos-
phate-buffered saline (PBS) with 8 M urea was prepared [29].
Tris(2-carboxyethyl) phosphine hydrochloride (TCEP-HCl) (Sigma)
was added to the solution at a molar ratio of 1.5:1 TCEP to brin-
ogen cysteines. PEG-DA is then added to the brinogen solution at
a molar ratio of 4:1 to brinogen cysteines with an incubation
period of 3 h in the dark at room temperature (pH 8). Next, the
solution was diluted with an equal volume of PBS with 8 M urea
and precipitated with excess acetone. The PEGylated brinogen
(PF) precipitate was redissolved in 150 mM PBS with 8 M urea
and dialyzed (1214 kDa MW cutoff; Spectrum, Los Angeles, CA)
against 150 mM PBS for 24 h at 4 C with several changes of PBS.
The brinogen concentration was determined using a standard
BCA protein assay kit (Pierce Biotechnology, Rockford, IL).
2.3. Photoinitiatior system
Three different UV light PIs were used in this study, along with a
non-light-activated initiator/accelerator as a control group (Fig. 1).
The PIs included 2-hydroxy-1-[4-(hydroxyethoxy)phenyl]-2-
methyl-1-propanone (Irgacure 2959, I2959; Ciba Specialty Chemi-
cals, Basel, Switzerland), 1-hydroxycyclohexyl phenyl ketone
(Irgacure 184, I184; Sigma) and 2,2-dimethoxy-1,2-diphenyle-
than-1-one (Irgacure 651, I651; Ciba Specialty Chemicals). A
non-light-activated initiator/accelerator system comprising a
mixture of ammonium persulfate (APS) and tetramethylethylenedi-
amine (TEMED) was also used; the APS/TEMED accelerates the poly-
merization when forming a hydrogel from a precursor solution, but
does so without UV irradiation [30]. The three PIs were dissolved in
1840 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848

Fig. 1. Free-radical polymerization of PEG-DA hydrogels using several different initiator systems. (A) Mechanism of radical dissociation from Irgacure 2959, Irgacure 651 and
Irgacure 184 (upon irradiation by ultraviolet light), and APS/TEMED (a non-light-activated initiator/accelerator). (B) Radical initiation whereby the initiator reacts with a PEG-
DA monomer, generating an active center. (C) Hydrogel network formation involving radical propagation and termination.

one of two solvents (100% methanol or 70% ethanol) at weight per soluble in 70% ethanol. The PI stock solutions were stored in the dark
volume ratios of 0.01, 0.05, 0.1 and 0.2 g ml1; the I651 was not at 25 C and diluted in a PF precursor solution just prior to their use
 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848
for studying hydrogel shear modulus, crosslinking kinetics, and cell
viability, upon hydrogel formation. A range of concentrations of APS
and TEMED were also tested when measuring hydrogel properties,
including 0.10.8% w/v APS and 0.10.4% w/v TEMED. For cell
viability, concentrations of 0.2% w/v APS and 0.10.2% w/v TEMED
were evaluated.
2.4. Rheological measurements
The in situ PF hydrogel gelation, mechanical properties and
crosslinking kinetics were characterized using an AR-G2 rheometer
(TA Instruments) equipped with a UV curing attachment and a
20 mm parallel plate geometry. Time-sweep oscillatory tests were
performed at 25 C at a sinusoidal 2% strain rate and a 3 rad s1
angular frequency (within the linear viscoelastic region as deter-
mined previously [29]). Multi-wave experiments were also
performed to verify that the plateau G0 values from the time-sweep
data represent the modulus at the end of the cross-linking reaction
(not shown). The PF precursor solution (200 ll at 6, 8 and
10 mg ml1) with added PI (nal PI concentrations: 0.01, 0.05, 0.1
and 0.2% w/v) was loaded into the rheometer, yielding a gap size
of 0.6 mm. The small gap setting minimized the optical path length
and allowed nearly uniform UV light penetration throughout the
hydrogel sample during photopolymerization. The storage and loss
modulus values were continuously recorded using the Rheological
Advantage Instrument Control software. After 1 min of equilibra-
tion, the precursor solution was exposed to UV light from an
Omnicure Series 1000 UV light source (Exfo, Canada) having a band
pass lter with a cutoff of 320500 nm. Two different UV light
intensities, 5 and 20 mW cm2, were used for the crosslinking reac-
tion of the precursor. The reported maximum shear modulus was
taken as the real part (i.e. the storage modulus) of the complex
00
shear modulus G G0 iG , based on the rationale that the G00
was negligible with respect to G0 value for each test.
2.5. Cell viability and morphology
Bovine aortic SMCs were purchased from Genlantis (San Diego,
USA) and cultured according to standard protocol. The cells were
cultured in Dulbeccos modied Eagles medium (Gibco, Paisley,
UK) containing 10% fetal bovine serum (Hyclone, Cramlington,
UK), 1% penicillinstreptomycin (Biological Industries, Israel), 1%
non-essential amino acids (Biological Industries, Israel) and 0.2%
2-mercaptoethanol (Gibco, UK). In order to prepare cell-encapsu-
lated hydrogel constructs, the SMCs (cell density of 2.5
million cells ml1) were mixed in 80 ll of 8 mg ml1 PF precursor
solution with PI and crosslinked in 5 mm diameter silicone tubes.
The PF underwent UV light-initiated polymerization for 5 min at
5 or 20 mW cm2 intensity. The cell-laden PF constructs were de-
graded either immediately or after 24 h in cell culture medium.
The constructs were degraded in a solution of 0.4 mg ml1 collage-
nase type 1A solution (Sigma) for 4 h at room temperature with
mild agitation. The cells in solution were collected by centrifugation
and stained with Ethidium Homodimer I (red nuclear stain for dead
cells) and Hoechst 33342 (blue nuclear stain for all cells). The
stained cells were observed using a Nikon TE2000 uorescence
microscope using a 10 objective and digitally imaged using a Pro-
gRes C10 camera (Jenoptik, Germany). The blue Hoechst-stained
nuclei were counted as the total number of cells; the red stained
nuclei were counted as the total number of dead cells in the sample.
The images were superimposed and the stains were analyzed to
yield viability results normalized to a control suspension of non-
encapsulated collagenase-treated trypsinized cells in solution. Cell
morphology was evaluated by labeling the cytoskeleton of cells
within the constructs using a lamentous actin (f-actin) uores-
cence stain. Speccially, the F-actin of cells was labeled with
1841
Phalloidin (Sigma Aldrich) and the cell nuclei were counterstained
with Hoechst33342 (SigmaAldrich). The uorescently labeled
cells in the gels were visualized and imaged on a Nikon TE2000
microscope equipped with a Prosilica CV640 digital CCD camera
(Prosilica Inc., B.C., Canada).
2.6. Statistical analysis
Each experiment was performed in triplicate, using at least
three different batches of either cells or materials. Statistical anal-
ysis was performed using Microsoft Excel software. Data from
independent experiments were quantied and analyzed for each
variable. Comparisons between two treatments were made using
Students t-test (two tail, unequal variance) and comparisons be-
tween multiple treatments were made with analysis of variance
(ANOVA). A p-value of <0.05 was considered to be statistically
signicant.
3. Results
3.1. Shear modulus and crosslinking kinetics
Rheological time-sweep tests were performed at 25 C using a
UV light attachment for in situ gelation to investigate the effects
of the various PIs, PI concentrations, UV light intensities and hydro-
gel precursor concentrations on the crosslinking kinetics and
rheological properties of the hydrogels. The rheological data
demonstrated that the representative hydrogel shear modulus
may be obtained from the mean peak value of the storage modulus
G0 ; the loss shear modulus G00 value was determined to be negligible
in comparison (Fig. 2). The mean peak G0 values of the materials
using several different frequencies in the time-sweep measurement
(i.e. multi-wave characterization) did not reveal signicant differ-
ence in modulus, suggesting that the reported modulus values rep-
resented the fully cross-linked state of the gels. Therefore, the
maximum storage modulus, which was dened as the mean peak
value of the shear storage modulus, was used for characterizing
the nal elastic properties of the hydrogels. The total crosslinking
time was dened as the time taken to reach 95% of the maximum
storage modulus following the radical initiation.
An initial screening helped identify concentrations of PI (0.01,
0.05, 0.1 and 0.2% w/v) that were used in the experimental design
because they yielded rapid crosslinking rates of the PF hydrogels at
the UV light intensities reported (5 and 20 mW cm2). The APS/
TEMED system proved effective for crosslinking the hydrogel pre-
cursor solution (Fig. 2a), albeit with slower reaction rates in com-
parison to the PIs (Fig. 2b). The crosslinking time of the APS/TEMED
system was nearly an order of magnitude larger than the I2959, at
each concentration that was tested. Because the APS/TEMED was
highly cytotoxic to cells, this initiator was not reported to the same
extent as the PIs.
Altering the concentration of the PI in the precursor solution al-
tered both the maximum storage modulus of the hydrogel and its
crosslinking time (Table 1). The precise effects of PI I2959 concen-
tration on shear modulus and crosslinking time are shown in
Fig. 3a and b for hydrogels made from 10 mg ml1 PF and 5% w/v
PEG-DA, respectively. Consequently, this relationship proved con-
sistent with the other two PIs that were tested (Table 1). The min-
imum PI concentration for the PEG-DA control gels was 0.02% w/v,
as crosslinking did not occur at the lowest PI concentration of
0.01% w/v. The shear modulus of both hydrogel types reached a
maximum when the PI concentration was 0.05% w/v. Upon further
addition of PI, the maximum storage modulus decreased. For both
materials, there was a direct relationship between the PI concen-
tration and the crosslinking time; the precursor was crosslinked
1842 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848

a and the crosslinking time, as summarized in Fig. 3c for UV intensi-

ties of 5 and 20 mW cm2. Interestingly, the maximum storage
modulus values at 20 mW cm2 are somewhat lower, and the
crosslinking times are signicantly shortened as compared with
5 mW cm2.
Different PIs were also examined in terms of the hydrogel cross-
linking reaction (i.e. shear storage modulus and crosslinking kinet-
ics) (Fig. 4, Table 1). Among the different PIs tested, the PF
precursor crosslinked using I2959 initiator reached the highest
storage modulus after UV light exposure, followed by the one
crosslinked using I184 initiator (Table 1). The precursor crosslinked
using I651 initiator reached a much lower modulus, as seen in
Fig. 4a. The crosslinking rate was fastest when using I2959 initiator
(Fig. 4, inset), followed by I184 and I651. Fig. 4a underscores that
the crosslinking time needed to reach a maximum modulus when
using I651 initiator is more than one order of magnitude higher
b than for that of the other two PIs. Fig. 4b summarizes the shear
storage modulus and crosslinking times for the various PIs that
were investigated (crosslinking time for I651 was excluded). The
higher storage modulus and shorter crosslinking time for I2959
are evident when comparing these data (Table 1).

3.2. Biocompatibility experiments

The use of a viability assay based on cell counting with uores-

Fig. 2. The crosslinking kinetics of PF with Irgacure 2959 or APS/TEMED. (a) The
time-sweep shear modulus (G0 , G00 ) of PF materials in the presence of 0.1% TEMED
and 0.4% APS showing a gradual increase in both storage modulus (G0 ) and loss
modulus (G00 ). (b) The time-sweep shear modulus (G0 , G00 ) of PF materials with
Irgacure 2959 is shown upon the application of continuous 365 nm UV light at an
intensity of 5 mW cm2; the UV light was turned on after 60 s. The intersections of
the storage and loss moduli corresponds to the gelation points of the materials.
faster as the PI concentration increased. At 5 mW cm2, the
10 mg ml1 PF precursor required about 600 s to fully cure at
0.01% w/v PI, compared to less than 100 s for a PI concentration
of 0.2% w/v. A similar trend was observed for the PEG-DA control
hydrogels.
The effect of curing intensities on the crosslinking time and
shear storage modulus was also investigated. Increasing the UV
light intensity from 5 to 20 mW cm2 resulted in a decrease of both
the maximum storage modulus and the crosslinking times in both
PF precursor and PEG-DA hydrogels, irrespective of PI concentra-
tion. Altering the concentration of the PF precursor solution
resulted in alterations to both the maximum storage modulus
cence microscopy provided a quantitative assessment of cell viabil-
ity as a measure of the biocompatibility of the crosslinking
reactions. In this regard, SMCs were used because they are more
sensitive than broblasts to chemical perturbations [31]. Cell
viability was evaluated both immediately after encapsulation by
radical polymerization and 1 day in hydrogel culture. The viability
assay involved degrading the gel and staining it with a membrane
integrity assay marker, ethidium homodimer I, which stained non-
viable cell nuclei in red. A blue nuclear counterstain, Hoechst
33342, was used to identify the total number of cell nuclei, as seen
in Fig. 5a. The viability assay was performed with all three PIs sol-
ubilized in methanol because I651 was not soluble in 70% ethanol.
At the highest PI concentration of 0.2 wt.% shown in Fig. 5b, both
I2959 and I184 were shown to be relatively cytocompatible, with
survival rates of 76 4 and 88 12%, respectively, immediately
after crosslinking. In contrast, I651 was signicantly more toxic,
with a relative survival rate of 53 27%. The cell viability in hydro-
gels cultured for 1 day after crosslinking with I2959 and I184 did
not exhibit any adverse change in viability, whereas hydrogels
crosslinked with I651 exhibited a further drop in cell survival rate
to 7 4%. In order to evaluate the effects of the solvent, I2959 and
I184 were also dissolved in 70% ethanol and evaluated for cell
viability relative to the methanol data. The differences in viability
results with the two solvents were not found to be statistically
signicant (Fig. 6).

Table 1
UV light intensity and photoinitiator type/concentration effect on PEG-brinogen hydrogel formation as indicated by the mean peak storage modulus (G0 in Pa) and crosslinking
time (in seconds).

PI concentration (% w/v) 0.01 0.05 0.1 0.2

Exposure intensity (mW cm2) 5 20 5 20 5 20 5 20
I2959 G0 134 11 134 8 164 2 137 2 161 2 123 2 153 2 110 1
Time 609 32 151 11 151 8 33 1 95 8 22 1 68 3 39 1
I184 G0 82 1 53 5 135 4 132 4 148 1 125 2 147 4 124 4
Time 2068 320 466 19 306 11 51 3 186 6 44 5 168 3 39 1
I651 G0 41 16 N/A 95 7 N/A 94 15 92 4 110 7 90 1
Time 14,041 3234 N/A 2066 531 N/A 1472 487 1436 63 2684 184 1858 578
 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848 1843

Fig. 3. Effects of UV light intensity on PEGbrinogen hydrogel formation with Irgacure 2959 PI, as indicated by the mean peak storage modulus (left) and crosslinking time
(right). Two levels of UV light intensity were applied: 5 (solid line) and 20 mW cm2 (dotted line). Modulus and cross-linking times for 10 mg ml1 PF precursors (a) and 5% w/
v PEG-DA (b) using various I2959 PI concentrations. (c) Modulus and cross-linking times for various concentrations of PF precursor with 0.1% w/v I2959. Statistically
signicant differences are noted between the two treatments, except where noted by (p < 0.05, n > 6).
1844 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848

Fig. 4. Effects of PI type on PF hydrogel formation as indicated by the oscillatory shear storage modulus, mean peak storage modulus values (G0 ) and crosslinking times. (a)
The storage modulus time-sweep data for PF crosslinking (10 mg ml1) with I2959, I184 or I651 PI (0.1% w/v). The irradiation with continuous UV light (5 mW cm2) after 60 s
demonstrates rapid reactions with I2959 and I184 and a less rapid one with I651 (the inset shows the rate of modulus change obtained by the rst derivative of the time-
sweep G0 curves). (b) Storage modulus (left) and cross-linking times (right) of PF (10 mg ml1) using three different PIs and concentrations, exposed to a UV light intensity of
5 mW cm2 (I651 crosslinking times are not shown). Statistically signicant differences are noted between the treatments, except where noted by (p < 0.05, n > 6).

The viability of cells was also conferred from observations of
cell spreading within the encapsulating PF hydrogels. The cells
began to invade the dense matrix through cellular lamellipodia
after several hours following their encapsulation (Fig. 7). The
initially rounded cells (Fig. 7A) formed protrusions after 16 h in
the 3-D PF hydrogels (Fig. 7B) and became highly spindled by
24 h in culture (Fig. 7C). Fluorescence staining of f-actin within
the encapsulated cells showed a highly spindled organization of
the cell cytoskeleton after 48 h in 3-D culture (Fig. 7D).
4. Discussion
Biomedical hydrogels used in regenerative medicine are often
designed to undergo a rapid liquid-to-solid phase transition
(gelation) in the presence of cells and tissues so as to maximize cell
viability during crosslinking. When chemical crosslinking is
involved in the gelation process, there are a number of important
factors to consider when deciding on a specic chemistry, particu-
larly in light of the possible cytotoxic consequences of such reac-
tions. Therefore, the crosslinking kinetics, efciency and
cytotoxicity should be thoroughly evaluated for a given material
and its intended application. In the context of cell-encapsulating
biomaterials, these factors are particularly crucial given the vulner-
ability of the cells to the reactive species and solvents present in
the hydrogel precursor solution prior to and during the chemical
crosslinking reaction. Rapid chemical reactions involving radicals
(i.e. free-radical polymerization reactions) are often favored
because of their relatively fast and efcient crosslinking kinetics,
but also because they can crosslink polymer-bound functional
groups in a protein-rich environment [28]. In this regard, rapid
 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848 1845

Fig. 5. SMC viability after enzymatic recovery from encapsulating PF hydrogels (8 mg ml1); the cells were encapsulated by photopolymerization with I2959, I184 or I651 PI
(0.2% w/v). (a) Fluorescence images of recovered cells taken 1 day after encapsulation, showing Hoechst 33342 staining all cell nuclei in blue and ethidium homodimer I
staining non-viable cell nuclei in red. (b) Quantitative viability data shows recovered cell populations immediately and 1 day after encapsulation, in comparison to a control
cell population in PBS solution exposed to 5 mW cm2 UV light (with 0.2% w/v PI). Statistically signicant differences between treatment and PBS+UV are observed, where
noted by  (p < 0.05, n > 5).

crosslinking reactions may also enhance the protective effects to

Fig. 6. Effect of PI solvent on SMC viability after encapsulation in PF hydrogels
(8 mg ml1) using photopolymerization with I2959 or I184 PI (0.2% w/v). The
initiators were solvated in either 70 vol.% ethanol (E) or 100 vol.% methanol (M). The
relative survival rates of cells recovered immediately and after 1 day of encapsu-
lation are shown. No statistically signicant differences were noted between
treatments (p < 0.05, n > 4).
resident cells by favoring crosslink formation between polymer
groups rather than protein-bound vinyl groups.
In this study, the cytotoxicity, crosslinking efciency and cross-
linking kinetics associated with photoinitiated and chemically ini-
tiated free-radical crosslinking of cell-encapsulating hydrogels
were comparatively evaluated. The rheological properties of the
hydrogels were used to determine the crosslinking reaction
efciency and kinetics based on the assumption that the degree
of crosslinking is proportional to the maximum shear storage
modulus of the hydrogel [32]. The factors affecting the maximum
storage modulus of the hydrogels were found to be the PI type/con-
centration, the precursor concentration and the UV light intensity
to which the hydrogel was exposed during photopolymerization.
At any given PI concentration and light intensity, for example, an
increase in precursor concentration resulted in higher maximum
storage modulus values and also increased gelation times, most
likely owing to the consequent increase in the number of function-
alized acrylate groups in solution. When the precursor concentra-
tion was held constant and the PI concentration or light intensity
1846 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848
Fig. 7. Encapsulated SMCs in PF hydrogels express a spindled morphology in 3-D culture. The SMC morphology after 2 (A), 16 (B) and 24 h (C) in 3-D culture was visualized by
phase-contrast microscopy. The f-actin of the cells in the PF hydrogels was also visualized in situ by uorescence microscopy after 48 h (D). Scale bars represent 25 lm.
was increased, the storage modulus exhibited a biphasic behavior.
Initially, the initiation rate and radical generation rate increased,
causing the crosslinking efciency to increase and form stiffer gels.
However, if the radical generation rate was too fast, as was the case
with the high concentrations of initiator, the crosslinking efciency
was reduced. It is speculated that factors such as cyclization of
pendant PEG-acrylate groups on the PF backbone or higher termi-
nation during the free-radical chain transfer contributed to this
effect [33], although no experimental evidence of these phenom-
ena were provided. Nevertheless, the polymerization crosslinking
kinetics data suggested that, beyond a certain amount of initiator
(>0.05% w/v PI), the PF hydrogel network formation culminated
with a less efcient crosslinked conguration. Accordingly, higher
PI concentrations or light intensities resulted in lower maximum
storage modulus values in shorter crosslinking times.
When comparing the crosslinking efciency and kinetics of the
three different PIs with PF and PEG-DA hydrogels, it became appar-
ent that I651 resulted in a much lower storage modulus than I2959
or I184. This may be due to precipitation of the non-polar I651 (and
to a lesser extent I184) upon addition to the hydrogel precursor
solution, causing an effective PI concentration that is below the
critical concentration required for polymerization. Alternatively,
the I651 has a much higher molar extinction coefcient, which
may explain the inefciency in crosslinking; the extinction coef-
cient of I651 is 15 times larger compared to that of the two other
PIs [25]. A high molar extinction coefcient entails that I651 forms
radicals much more rapidly, which may result in a more
pronounced cyclization of PEG-acrylate groups within the PF
hydrogel network. Evidence to support this speculation would
require further characterization of the hydrogel network structure
[34], but this is beyond the scope of this investigation.
As noted previously, the crosslinking reaction required that
hydrogel formation must not culminate in such a way so as to
compromise the network integrity via premature termination or
functional end group cyclization. On the other hand, the hydrogel
formation must be sufciently short when encapsulating cells/tis-
sues in situ so as not to expose cells to toxic by-products of the
chemical reaction. Therefore, to minimize cell cytotoxicity, the PI
type/concentration should target a photopolymerization process
that results in hydrogel formation within a couple of minutes.
The data revealed that utilizing PIs I2959 and I184 in excess of
0.05 vol.% yields a crosslinking time of less than 5 min for the
hydrogels to reach their maximum storage modulus value without
compromising the high crosslinking efciency of the reaction.
These results indicated that both I2959 and I184 yielded rapid
reaction kinetics for cell encapsulation. Consequently, another
water-soluble photoinitiaor recently tested with cell-encapsulat-
ing hydrogels, lithium phenyl-2,4,6-trimethylbenzoylphosphinate,
may provide an even faster crosslinking reaction that is minimally
cytotoxic [19]. In contrast, the PI I651 was not able to crosslink the
hydrogel to the maximum extent possible within this time-frame,
as indicated by the signicantly lower maximum storage modulus
values obtained.
Having a PI that provides sufcient crosslinking for optimal
network formation with minimal radical exposure does not pre-
clude the need to verify cytocompatibility with cells (and polymer)
present during the reaction. A pivotal benchmark test to demon-
strate in situ cytotoxicity of the radical polymerization process is
the quantication of the number of live and dead cells following
the encapsulating reaction. This test was performed with cell-
laden hydrogels that were crosslinked from a PF precursor solution
containing SMCs and PI. The I2959 and I184 supported high viabil-
ity with exposure to 5 min of UV irradiation, whereas the I651
exhibited cytotoxic effects that were manifested as a substantial
reduction in cell viability. This nding is in agreement with a
recent report by Patel et al. that documented the cytotoxicity of
I651 and UV irradiation with preadipocytes in culture [35]. It has
been proposed that I651 is more cytotoxic to cells because of its
chemical structure, which is more hydrophobic than the other
two PIs (I2959 and I184), enabling the molecule to diffuse more
 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848
easily through the phospholipid bilayer membrane of cells [26].
Alternatively, the cytotoxicity of I651 may be associated with the
potent radical formation and by-products produced by its dissoci-
ation [25]. Consequently, the cell viability of SMCs demonstrates a
dramatic decrease 1 day after UV irradiation with I651, suggesting
that some toxic remnants are present in the hydrogel well after the
photopolymerization reaction. Although additional experiments
would be required to isolate the specic origins of toxicity associ-
ated with this initiator, it is interesting to note that the initiator
alone (without UV) does not pose the same extent of toxicity on
a control culture of SMCs (data not shown). Moreover, the use of
UV light alone without I651 on a control population of SMCs did
not prove to be nearly as detrimental to cell survival (not shown),
which suggests that 5 min of UV irradiation at 365 nm and
20 mW cm2 was minimally cytotoxic. Consequently, the type of
solvent used with I2959 and I184 did not appear to signicantly
alter the SMC viability results.
In contrast to photochemical crosslinking with I2959 or I184,
the use of a non-light-activated chemical initiator such APS/TEMED
proved highly toxic to cells, as indicated by the <20% cell survival
rate after 24 h (not shown). This toxic effect was likely owing to
the extremely slower reaction kinetics associated with the APS/
TEMED. Much like with PIs, the APS/TEMED catalyzes the polymer-
ization of the precursor solution with free-radical formation; how-
ever, the APS/TEMED reaction involves a long-lived exposure of
cells to toxic free-radicals during polymerization that can last more
than 30 min. It is assumed that this sustained exposure to even low
doses of radical proves too strenuous for SMCs in the solution/
hydrogel and, as a result, most cells are dead or dying by the time
the polymerization reaction has culminated in the formation of a
polymer network.
5. Conclusion
This investigation focused on the importance of understanding
the curing kinetics of hydrogels undergoing photopolymerization
in terms of the formation of covalent chemical crosslinks and their
effects on cell viability and network properties. Two PIs, Irgacure
2959 and Irgacure 184, were identied as being more suitable for
achieving high cell viability during the photopolymerization reac-
tion of cell-encapsulating hydrogels starting from a PF precursor
solution. Moreover, these two initiators provided high photocross-
linking efciency for bridging acrylate groups on the PF precursor,
thus achieving adequate mechanical properties within a relatively
short UV exposure time. Optimization of PI concentration or irradi-
ation intensity was important for achieving maximum cell viability
and mechanical properties. A sub-optimal choice of PI concentra-
tion or irradiation intensity can hamper the polymer network for-
mation, leading to an unexpected reduction in hydrogel modulus.
Likewise, cytocompatibility may be compromised by unnecessarily
prolonging exposure to cytotoxic free radicals or inadvertently
enhancing the instantaneous dose of radicals in solution, both of
which are dependent on the PI type/concentration and irradiation
intensity. Therefore, it is important to optimize the various PI con-
ditions of the polymer system to achieve the desired crosslinking
outcome that is most suitable for cell encapsulation purposes.
Acknowledgements
The authors are grateful for the nancial support of the Eco-
nomic Development Board (Singapore) through the STRAT pro-
gram, and the Singapore National Research Foundation CREATE
grant. Additional support for this research was provided by Grant
No. 2007366 from the United StatesIsrael Binational Science
Foundation.
1847
Appendix A. Figures with essential color discrimination
Certain gures in this article, particularly Figs. 27, are difcult
to interpret in black and white. The full color images can be found
in the on-line version, at doi:10.1016/j.actbio.2011.12.034.
References
[1] Hoffman AS. Hydrogels for biomedical applications. Adv Drug Deliv Rev
2002;54:312.
[2] Nakaji-Hirabayashi T, Kato K, Iwata H. Hyaluronic acid hydrogel loaded with
genetically-engineered brain-derived neurotrophic factor as a neural cell
carrier. Biomaterials 2009;30:45819.
[3] Lee KY, Mooney DJ. Hydrogels for tissue engineering. Chem Rev
2001;101:186979.
[4] Hoare TR, Kohane DS. Hydrogels in drug delivery: progress and challenges.
Polymer 2008;49:19932007.
[5] Sarig-Nadir O, Seliktar D. Compositional alterations of brin-based materials
for regulating in vitro neural outgrowth. Tissue Eng Part A 2008;14:40111.
[6] Nicodemus GD, Bryant SJ. Cell encapsulation in biodegradable hydrogels for
tissue engineering applications. Tissue Eng Part B Rev 2008;14:14965.
[7] Anseth KS, Metters AT, Bryant SJ, Martens PJ, Elisseeff JH, Bowman CN. In situ
forming degradable networks and their application in tissue engineering and
drug delivery. J Control Release 2002;78:199209.
[8] Krishna OD, Kiick KL. Protein- and peptide-modied synthetic polymeric
biomaterials. Biopolymers 2010;94:3248.
[9] Nguyen KT, West JL. Photopolymerizable hydrogels for tissue engineering
applications. Biomaterials 2002;23:430714.
[10] Kuo CK, Ma PX. Ionically crosslinked alginate hydrogels as scaffolds for tissue
engineering. Part 1. Structure, gelation rate and mechanical properties.
Biomaterials 2001;22:51121.
[11] Hu BH, Messersmith PB. Rational design of transglutaminase substrate
peptides for rapid enzymatic formation of hydrogels. J Am Chem Soc
2003;125:142989.
[12] Chen G, Hoffman AS. Graft copolymers that exhibit temperature-induced
phase transitions over a wide range of pH. Nature 1995;373:4952.
[13] Hennink WE, van Nostrum CF. Novel crosslinking methods to design
hydrogels. Adv Drug Deliv Rev 2002;54:1336.
[14] Sawhney AS, Pathak CP, Hubbell JA. Interfacial photopolymerization of
poly(ethylene glycol)-based hydrogels upon alginate-poly(L-lysine)
microcapsules for enhanced biocompatibility. Biomaterials 1993;14:100816.
[15] Bryant SJ, Anseth KS. Photopolymerization of hydrogel scaffolds. In: Ma PX,
Elisseeff JH, editors. Scaffolding tissue engineering. Boca Raton, FL: CRC Press;
2005. p. 7190.
[16] Baroli B. Photopolymerization of biomaterials: issues and potentialities in drug
delivery, tissue engineering, and cell encapsulation applications. J Chem
Technol Biotechnol 2006;81:4919.
[17] Ifkovits JL, Burdick JA. Review: photopolymerizable and degradable
biomaterials for tissue engineering applications. Tissue Eng 2007;13:236985.
[18] Bowman CN, Kloxin CJ. Toward an enhanced understanding and
implementation of photopolymerization reactions. Aiche J 2008;54:277595.
[19] Fairbanks BD, Schwartz MP, Bowman CN, Anseth KS. Photoinitiated
polymerization of PEG-diacrylate with lithium phenyl-2,4,6-trimethylbenzoyl
phosphinate: polymerization rate and cytocompatibility. Biomaterials
2009;30:67027.
[20] Cheung YK, Azeloglu EU, Shiovitz DA, Costa KD, Seliktar D, Sia SK. Microscale
control of stiffness in a cell-adhesive substrate using microuidics-based
lithography. Angew Chem Int Ed Engl 2009;48:718892.
[21] Kaehr B, Shear JB. Multiphoton fabrication of chemically responsive protein
hydrogels for microactuation. Proc Natl Acad Sci USA 2008;105:88504.
[22] Khetan S, Burdick JA. Patterning network structure to spatially control cellular
remodeling and stem cell fate within 3-dimensional hydrogels. Biomaterials
2010;31:822834.
[23] Kloxin AM, Kasko AM, Salinas CN, Anseth KS. Photodegradable hydrogels for
dynamic tuning of physical and chemical properties. Science 2009;324:5963.
[24] Fedorovich NE, Oudshoorn MH, van Geemen D, Hennink WE, Alblas J, Dhert
WJA. The effect of photopolymerization on stem cells embedded in hydrogels.
Biomaterials 2009;30:34453.
[25] Bryant SJ, Nuttelman CR, Anseth KS. Cytocompatibility of UV and visible light
photoinitiating systems on cultured NIH/3T3 broblasts in vitro. J Biomater Sci
Polym Ed 2000;11:43957.
[26] Williams CG, Malik AN, Kim TK, Manson PN, Elisseeff JH. Variable
cytocompatibility of six cell lines with photoinitiators used for polymerizing
hydrogels and cell encapsulation. Biomaterials 2005;26:12118.
[27] Almany L, Seliktar D. Biosynthetic hydrogel scaffolds made from brinogen
and polyethylene glycol for 3D cell cultures. Biomaterials 2005;26:246777.
[28] Halstenberg S, Panitch A, Rizzi S, Hall H, Hubbell JA. Biologically engineered
protein-graft-poly(ethylene glycol) hydrogels: a cell adhesive and plasmin-
degradable biosynthetic material for tissue repair. Biomacromolecules
2002;3:71023.
[29] Dikovsky D, Bianco-Peled H, Seliktar D. Dening the role of matrix compliance
and proteolysis in three-dimensional cell spreading and remodeling. Biophys J
2008;94:291425.
1848 I. Mironi-Harpaz et al. / Acta Biomaterialia 8 (2012) 18381848
[30] Gotoh T, Nakatani Y, Sakohara S. Novel synthesis of thermosensitive porous
hydrogels. J Appl Polym Sci 1998;69:895906.
[31] Peuster M, Fink C, Wohlsein P, Bruegmann M, Gunther A, Kaese V, et al.
Degradation of tungsten coils implanted into the subclavian artery of New
Zealand white rabbits is not associated with local or systemic toxicity.
Biomaterials 2003;24:3939.
[32] Van den Bulcke AI, Bogdanov B, De Rooze N, Schacht EH, Cornelissen M,
Berghmans H. Structural and rheological properties of methacrylamide
modied gelatin hydrogels. Biomacromolecules 2000;1:318.
[33] Naghash HJ, Okay O, Yagci Y. Gel formation by chain-crosslinking
photopolymerization of methyl methacrylate and ethylene glycol
dimethacrylate. Polymer 1997;38:118796.
[34] Frisman I, Orbach R, Seliktar D, Bianco-Peled H. Structural investigation of
PEGbrinogen conjugates. J Mater Sci Mater Med 2010;21:7380.
[35] Patel PN, Gobin AS, West JL, Patrick Jr CW. Poly(ethylene glycol) hydrogel
system supports preadipocyte viability, adhesion, and proliferation. Tissue Eng
2005;11:1498505.