Biomaterials 19 (1998) 1343 — 1352

Light-induced tailoring of PEG-hydrogel properties

Fotios M. Andreopoulos!,", Eric J. Beckman!,", Alan J. Russell!,",*
! Department of Chemical Engineering, University of Pittsburgh, Pittsburgh, PA 15261, USA
" Center for Biotechnology and Bioengineering, University of Pittsburgh, Pittsburgh, PA 15261, USA
Received 14 July 1997; accepted 16 October 1997

Abstract

We have previously reported (Andreopoulos et al. J Am Chem Soc 118 (1996) 6235—6240) the synthesis of hydrogels via the
photopolymerization of water-soluble PEG molecules. In this paper, PEG-hydrogel membranes were prepared by the irradiation
('300 nm) of aqueous solutions of photosensitive 4-armed PEG (nominal molecular weight of 20 000), in the absence of photo-
initiators. The hydroxyl termini of the PEG’s were functionalized with cinnamylidene acetate groups to form photosensitive PEG
macromers (PEG-CA), which upon irradiation ('300 nm) formed crosslinks between adjacent cinnamylidene groups resulting in
highly crosslinked networks (hydrogels) (Andreopoulos et al. J Am Chem Soc 118 (1996) 6235—6240). The hydrogel membranes were
highly swellable with equilibrium volume fractions ranging from 0.02 to 0.05. Their swellability was a function of irradiation light
('300 nm) and degree of modification of the PEG molecules. The effect of light on the permeation fluxes of myoglobin (Mb),
hemoglobin (Hb), and lactate dehydrogenase-L (LDH) through PEG membranes was also assessed and the diffusion coefficients of the
proteins were determined accordingly. The PEG-CA membranes exhibited photoscissive behavior upon exposure to UV irradiation
(254 nm). Therefore, UV light was used as a trigger to control the mesh size of the membranes, and thereby the permeation fluxes of
Mb, Hb, and LDH. Equilibrium swelling experiments with membranes prepared under different irradiation conditions were
performed, and the Flory—Huggins model was utilized to determine the mesh size and the average molecular weight between
crosslinks of the synthesized hydrogels. ( 1998 Elsevier Science Ltd. All rights reserved

Keywords: Hydrogels; Diffusion coefficients; Photoscission; Proteins

1. Introduction
In the last two decades there has been an increas-
ing interest in the use of hydrogel membranes in bio-
separation processes, hybrid-type artificial organs, and
controlled release systems for drug delivery [1—5]. The
composition of the hydrogel, the crosslinking density,
polymerization mechanism, as well as the type and size of
the analyte to be released have a significant impact on the
permeative characteristics of hydrogel membranes. Bio-
logically active agents can often be immobilized within
a hydrogel matrix [5—9], and a device can be designed to
release the agent at a predetermined rate if the diffusive
characteristics of the membrane are known.
Environmentally sensitive hydrogels are promising
materials for separation and controlled drug delivery.
Various stimuli such as pH [1—3, 10], ionic strength
* Corresponding author. Tel: 412 624 9631; fax: 412 624 9639; e-mail:
ajrche#@pitt.edu
[1, 4, 7], temperature [5, 6, 8, 9], electric field [4], and
magnetic field [11, 12] have been utilized as triggers for
drug release. Peppas [2—5, 10] and coworkers have
studied the pH-triggered release of small drugs through
polymethacrylic-g-ethylene glycol (PMA-g-EG) copoly-
mers [2], polyvinyl alcohol (PVA) polymeric networks
[3, 13] or polyhydroxyethylmethacrylate (PHEMA) hy-
drogels [10, 14]. Hoffman et al. [6, 9] have investigated
thermally reversible hydrogels as drug delivery materials,
where poly-N-isopropylacrylamide (P-NIPA) gels were
used as controlled matrices for the delivery of drugs or
as immobilizing support for enzymes upon temperature
changes [6]. These thermally reversible gels exhibit
sharp volume-phase transition near their phase
transition temperature (LCST), such that as the temper-
ature is raised above the LCST the gel collapses releasing
water. As the temperature is lowered below the LCST the
gel reswells [6]. Gehrke [1] has also investigated P-
NIPA gels that can respond to temperature changes
faster than the conventional temperature-sensitive
hydrogels.

0142-9612/98/$ — See front matter ( 1998 Elsevier Science Ltd. All rights reserved.
PII S 0 1 4 2 - 9 6 1 2 ( 9 7 ) 0 0 2 1 9 - 6
1344 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
In the work described herein, we examine the diffusive
characteristics of light sensitive PEG-based hydrogel
membranes [15] by monitoring the permeation rates of
model proteins through the membranes. Briefly, the end
groups of star-polyethylene glycol molecules are modi-
fied with cinnamylidene acetate groups which crosslink
upon exposure to light. Photopolymerization of PEG-
CA monomers in water provides a straightforward route
to prepare hydrogel membranes in the absence of poten-
tially toxic photosensitizers and/or photoinitiators.
These hydrogels exhibit photoreversible behavior, and
upon exposure to UV light (254 nm) the gels photode-
grade. Because of the unique photoscissive behavior of
the PEG-CA hydrogels, light may be used as a trigger for
the release of drugs or other pharmacological agents
from a hydrogel matrix.
2. Materials and methods
2.1. Materials
‘Four-armed’ polyethylene glycol (PEG) with nominal
molecular weight of 20 000 (PDI"1.026, by GPC-Poly-
mer Lab. Gel mixed D column) was purchased from
Shearwater Polymer Inc. According to the distributor,
the branched PEG was synthesized by ethoxylation of
polyols (derived by glycerol condensation) and contained
4 arms. No linear PEG was utilized since they would not
form crosslinks as needed. All other materials were pur-
chased either from Aldrich or Sigma (St. Louis, MO) and
used without further purification.
2.2. Synthesis of photosensitive PEG monomers
A detailed synthetic procedure is described in a pre-
vious paper [15]. Typically, PEG (4.5 g, four armed,
MW"20 000) was dissolved in toluene (100 ml) and
dried via azeotropic distillation at 65°C. The dried PEG
solution was transferred to a 500 ml, three-neck flask,
and triethylamine (1.8 ml) was added to act as a scaven-
ger for the acid byproduct. The mixture was then stirred
and purged with nitrogen (99% purity) for 15 min. Next,
cinnamylidene acetyl chloride [15] (13 g) dissolved in
tetrahydrofuran (30 ml, 99#% purity), was transferred
dropwise through a septum into the reaction flask. The
reaction mixture was stirred continuously at 80°C for
17 h. The resulting product was purified by first remov-
ing the solvent and then precipitation in ether and finally,
in acetone. The modified PEG (4.3 g, 92% yield) was
collected and stored in the dark and under vacuum. The
number average molecular weight of the modified PEG
molecules was determined using Matrix Assisted Laser
Desorption Time of Flight mass spectroscopy (MALDI-
TOF) (M "22 342 for the 85% modified molecule. The
/ 4
M of the unmodified PEG molecule was 21 567). Specifi-
/
cally, 2 ll of PEG-CA (10 mg ml~1) were combined with
18 ll of 2,5-dihydroxybenzoic acid (10 mg ml~1 in H O).
2
Approximately, 1 ll of this solution was applied to a tar-
get that had previously been spotted with sodium iodide
(0.1 mg ml~1 in H O). The mass spectrometer was oper-
2
ated in linear mode with a high mass hybrid detector, and
external calibration was performed using a mixture of
cytochrome-C, myoglobin, trypsinogen (CMT) with
sinapinic acid as the matrix.
2.3. Photogelation and photoscission of PEG-CA hydrogel
membranes
Typically, 15.4 w/v% aqueous solution of PEG-CA
was prepared by dissolving PEG-CA monomers in de-
ionized water at room temperature. A sample of the
solution was poured into a silinazed quartz mold
(2]1.5]0.08 cm3, Fig. 1) and irradiated with a 450 W
medium pressure mercury lamp at a distance of 10 cm.
A Pyrex filter surrounded the light source in order to cut
off wavelengths lower that 300 nm. For the photoscission
experiments an 8 W (254 nm) reactor was used to de-
polymerize the membranes at room temperature. After
irradiation, the hydrogel membranes were equilibrated in
Tris-HCl (50 mM, pH"7.5) solution for 12 h prior to
immediate use. The thickness of the equilibrated mem-
brane was measured with a laser micrometer at three
different places and an average was taken. The standard
deviation of the thickness average was less than
5]10~4 mm indicating that the hydrogel membranes
were homogeneous with uniform thickness.
2.4. Swelling experiments
Equilibrium swelling experiments were used to deter-
mine the structural characteristics of the PEG-CA hydro-
gels. Immediately following irradiation the hydrogel
membranes were removed from the molds and their
volume (» ) was determined by measuring the displaced
3
volume of these gels in a nonsolvent (heptane). The sam-
ples were then allowed to swell to equilibrium in Tris-
HCl (pH"7.5, 50 mM) for 12 h, and again their volume
(» ) was measured. The weights of the hydrogels after
4
irradiation (¼ ) and after swelling (¼ ) were measured in
3 4
air. Finally, the hydrogel membranes were dried under
vacuum at 30°C for 24 h and their weights and volumes
were measured according to the above procedure.
The polymer volume fractions of the hydrogels before
(u ) and after (u ) swelling were calculated from the
2,3 2,4
following equations [16, 18]:
u "» /» , (1)
2,3 $ 3
u "» /» , (2)
2,4 $ 4
where » is the volume of the dry hydrogel, » the volume
$ 3
of the hydrogel following irradiation and » the volume
of the hydrogel after equilibrium swelling.
 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
Fig. 1. A diagrammatic representation of the procedure followed to prepare PEG-CA membranes, and the diffusion apparatus used for the permeation
studies.
The degree of swelling (DS) of the hydrogel mem-
branes was determined according to the equation:
DS"(¼ !¼ )/¼ , (3)
4 $ $ (0.25 mg/ml). The receptor cell was filled with Tris-HCl
where ¼ is the weight of the gel in air after swelling, and
4
¼ the weight of the dry gel.
$
2.5. Diffusion experiments
Equine myoglobin (MW"17 600; Diameter"44 A_ ),
bovine hemoglobin (MW"64 500; diameter"77 A_ ),
and bovine lactate dehydrogenase-L (MW"140 000;
diameter"172 A_ ) were used as model diffusants to study
the diffusive characteristics of the PEG-CA hydrogel
membranes. For the permeation studies a horizontal
side-by-side diffusion cell apparatus with a defined com-
partment volume (3 ml), and diffusion cross-section area
was used (ID"9 mm, Perme Gear Inc., Fig. 1). A con-
stant temperature (37$1°C) was maintained through-
out the permeation studies by surrounding the diffusion
cells with a water jacket. Stirring bars were placed in each
compartment to maintain constant mixing, hence elimin-
ating any mass transport resistance.
1345
Typically, a preswollen membrane in Tris-HCl (50 mM,
pH"7.5), was placed between the two diffusion cells,
and the donor cell was filled with a protein solution
(50 mM, pH"7.5), and periodically, samples from the
receptor cell were collected and analyzed using a UV—Vis
spectrophotometer. Aliquots from the receptor cell were
replaced with fresh buffer in order to maintain the high
concentration gradient between the donor and the recep-
tor cell. The concentrations of the myoglobin and hemo-
globin were evaluated from a corresponding protein cal-
ibration curve. An activity assay (Diagnostics Lactate
Dehydrogenase reagent) based on the oxidation of lac-
tate to pyruvate was used to measure the LDH activity in
the receptor cell.
The permeability coefficients of the proteins were de-
termined using the equation derived by Colton [17],
lnM(C !2C )/C N"!2APt/», (4)
0 5 0
where C is the concentration of the protein in the donor
0
cell (mg ml~1), C the concentration of the protein
5
(mg ml~1) in the receptor cell at time t (min), A is the area
of permeation (cm2), » the volume of each cell (ml), and
1346 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
t the time of permeation (min). A plot of the ln(C) versus
time has a slope of !2APt/», from which the permeabil-
ity coefficient for each solute was calculated (cm min~1).
The diffusion coefficient for the solute (cm2 min~1)
through the membrane was calculated from:
D"Pl/K, (5)
where P is the permeability coefficient (cm min~1), l the
thickness of the swollen membrane (cm), and K the
dimensionless partition coefficient of the solute through
the membrane.
2.6. Partition studies
The partition coefficient describes the equilibrium ra-
tio of the saturation concentration of the solute in the
membrane over the concentration in the surrounding
media:
K"C /C ,
. %
(6)
where C is the concentration of the solute in the mem-
. n
brane after equilibrium has been reached, and C the
% backbone (1.48 A_ ), a the linear deformation of the end-
equilibrium solute concentration of the surrounding me-
dia. The partition coefficient was measured by solute
uptake experiments where the change of solute concen-
tration in the surrounding media was monitored spectro-
photometrically. Specifically, preswollen hydrogel
membranes in Tris-HCl (pH"7.5, 50 mM) were placed
in the corresponding protein solution (5 ml,
0.25 mg ml~1). The change in concentration of the sur-
rounding solution was measured as a function of time
using a UV—Vis spectrophotometer. The equilibrium
concentration, C , of the protein solution was used to
% (M of PEG"44). a is proportional to equilibrium
calculate the partition coefficient using the equation:
» (C !C )
K" 4 * %, (7)
» *C
. %
where » is the volume of the protein solution (ml), C is
4 *
the initial concentration of the protein solution, » the
.
volume (ml) of the hydrogel membrane, and C the con-
%
centration of the surrounding protein solution after
membrane equilibration.
2.7. Molecular weight between crosslinks and average
mesh size
The number average molecular weight between cross-
links (M ) for PEG-CA hydrogel networks was evaluated
# .!9
via swelling experiments in water. An expression de-
veloped by Flory [18], and later modified by Merrill and
Bray [19], was used to calculate the experimental values
of M :
# monomer concentration, the concentration of the at-
1 2 (uN /» ) Mln(1!u )#u #su2
" ! 1 2,4 2,4 2,4 , (8)
M M u M(u /u )1@3!1 (u /u )N
# / 2,3 2,4 2,3 2 2,4 2,3
where M is the number average molecular weight be-
#
tween two crosslink points, uN is the specific volume of the
PEG-CA monomers (0.861 cm3 g~1) before crosslinking,
» the molar volume of water (18.1 cm3 mol~1), u is
1 2,4
the equilibrium polymer volume fraction (is given by the
ratio of the volume of the polymer over the volume of the
swollen gel), u is the relaxed polymer volume fraction
2,3
(given by the ratio of the volume of the polymer over the
volume of the gel immediately after crosslinking and
before swelling), and s is the polymer—solvent interaction
parameter for PEG-CA—water. For the present experi-
mental work s was assumed constant over the range of
u values used in the swelling experiments (0.022—0.04),
2,4
and a s value determined by Merrill [20, 21] for PEO in
water was used throughout this study (0.426).
The average mesh size of highly swollen hydrogels
where crosslinks are introduced in the solution state was
calculated with the following equation:
m"a (C nl2)1@2 (9)
4 n
where C is the characteristic ratio of the polymer (C for
n
PEG+4.0) [20], l the bond length of the gel’s polymeric
4
to-end distance between two crosslinks for the network
chains in water with respect to the end-to-end distance of
the unperturbed (unswollen) state, and n the number
of bonds between two crosslinks and can be evaluated
by
3M
n" # , (10)
M
3
where M is the molecular weight of the repeat unit
3
3 4
degree of swelling,
a "Q1@3"u~1@3 .
4 . 2,4
(11)
3. Results and discussion
3.1. Hydrogel membrane characterization
Hydrogel membranes were prepared by the irradiation
of aqueous solutions of PEG-CA monomers with a 450
W Hg lamp. The functionality of the PEG-CA monomers
ranged from 68% to 85% modification as measured by
UV spectroscopy at the absorption maximum of PEG-
CA monomers in dichloromethane (j "313 nm). As
the initial concentration of the PEG monomers and the
time of exposure to light ('300 nm) increase, the hydro-
gel membranes become more densely crosslinked and
their degree of swelling decreases. With an increase in
tached cinnamylidene groups will also increase, and con-
sequently the distance between available double bonds
 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352 1347

for crosslinking mechanism should be reduced. There-

fore, the likelihood of forming cyclobutane rings between
adjacent cinnamylidene groups increases with an increas-
ing monomer concentration, and one would expect
decreased swellability with increasing monomer concen-
tration, until the maximum achievable degree of cross-
linking is reached. Fig. 2 depicts the predicted decrease in
the degree of swelling as a function of monomer concen-
tration. Fig. 3 shows the relationship between irradiation
time and equilibrium volume fraction. As the irradiation
time ('300 nm) increased from 2 h to 12 h, u changed
2,4
from 0.022 to 0.04. It is therefore evident, that by control-
ling the initial PEG-CA monomer concentration and
irradiation time we can tailor the crosslinking density of
the membranes, and thus the physical properties of the
hydrogel.
Equilibrium swelling studies were used to determine
changes in the mesh size of the PEG-CA hydrogel mem-
branes. The molecular weight between crosslinks, M ,
#
was evaluated using the expression developed by Merrill
and Bray [19] for swollen networks produced by cross- Fig. 3. The equlibrium volume fraction, u , as a function of light
2,4
linking in solution. The linear average mesh size, m, of the exposure ('300 nm). The initial PEG-CA monomer concentration in
gel networks was calculated using Eq. (9). Figures 4 and 5 water was 15.4 w/v%, and the degree of modification of the PEG-CA
represent the change of m as a function of irradiation time molecules with cinnamylidene acetate groups was 85%. A 450 W Hg
lamp was used as the light source.

Fig. 4. The average mesh size of the PEG-CA hydrogel membrane as

Fig. 2. The degree of swelling of PEG-CA hydrogel membranes as
a function of initial monomer concentration in water. Two different
molecular weight PEG molecules (10 000 and 20 000 respectively) with
similar degree of modification were used as the starting materials for the
hydrogel preparation. Aqueous solutions of PEG-CA were irradiated
with a 450 W Hg lamp for 2 h ('300 nm).
a function of light exposure ('300 nm) based on swelling experiments.
The initial monomer concentration in water was 15.4 w/v% and the
degree of the modification was 85%. A 450 W Hg lamp was used as the
light source.
and monomer concentration, respectively. As expected,
longer exposure to irradiation leads to more dense net-
works with smaller molecular weight between crosslinks,
and decreased swellabilities.
1348 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
Fig. 5. The average mesh size of the PEG-CA membrane as a function
of initial monomer concentration in water. The irradiation time was 2 h
('300 nm) and the degree of modification of the PEG-CA monomers
was 78% as measured by UV spectroscopy. A 450 W Hg lamp was used
as the light source.
3.2. Partition studies
In order to calculate the diffusion coefficients of myo-
globin, hemoglobin, and lactate dehydrogenase-L in the
membranes, their partition coefficients into the hydrogels
had to be evaluated. Eq. (7) was used to calculate the
experimental values of K. Table 1 summarizes the cal-
culated partition coefficients of the proteins under study.
For myoglobin and lactate dehydrogenase-L the parti-
tion coefficients were close to unity indicating that there
was no interaction between the crosslinked structure of
the hydrogel membranes and the solutes. On the other
hand, the partition coefficients of hemoglobin were found
to be higher than unity, indicating positive interactions of
hemoglobin with the membrane. Over the range of ir-
radiation times assessed, the PEG-CA membranes
showed no significant differences in the values of the
partition coefficients. There was also no systematic trend
in K within the PEG-CA hydrogels versus either the
molecular size of the permeating solute or exposure to
photoscissive light.
3.3. Diffusion of proteins through PEG-CA membranes
Three proteins of varying molecular weights (myo-
globin: MW"17 600, hemoglobin: MW"64 500, and
lactate dehydrogenase-L: MW"140 000) were used to
determine the effect of light on the permeative properties
of the PEG-CA hydrogel membranes. Diffusion coeffi-
Table 1
The partition coefficients of the proteins under study are shown as
a function of light exposure ('300 nm)
Proteins Time of light Partition
exposure coefficients
(h'300 nm) (K)
Myoglobin 2 0.760
3 0.694
6 0.854
12 0.850
Hemoglobin 6 1.93
12 2.3
21 2.2
Lactate dehydrogenase-L 12 1.27
12#45 min of 1.18
254 nm irradiation
cients for the proteins in PEG-CA membranes with vary-
ing crosslinking density were determined using Eqs. (4)
and (5). We have described above how light can tailor
membrane properties. Our prediction would be that as
mesh size and swellability decrease, the rate of protein
diffusion would also decrease as long as the mesh size was
close to that of the protein diameter. As the protein
diameter approached the mesh size one would predict
significant effects of light-induced mesh destruction.
Table 2 summarizes the data obtained experimentally.
We shall first consider how changes in irradiation time
('300 nm) affect the permeation rates of myoglobin
through the PEG-CA membranes (Fig. 6). Clearly, as the
irradiation time increases the mesh size of the hydrogel
membranes decreases resulting in a decrease in the value
of Mb’s diffusion coefficient (Fig. 7). Given the large mesh
size relative to the protein’s diameter (44 A_ ), the per-
meation of myoglobin is likely to be only partially affec-
ted by changes in mesh size. As the number of crosslinks
increases the observed protein’s permeation rates were
decreased, indicating that the crosslinking impeded to
a greater extent some of the protein’s movement. This
phenomenon was more apparent for larger molecules
such as hemoglobin and lactate dehydrogenase-L. In
Fig. 8, the permeation rates of hemoglobin through mem-
branes irradiated for 6 and 12 h ('300 nm), respectively,
are shown (PEG-CA "85% modification). For hemo-
globin there is a 56% decrease in the permeation rate
through the 12 h membrane (m"154 A_ ) compared to the
6 h membrane (m"181 A_ ). The above results clearly
show that the hydrogel’s crosslinked structure has
a screening effect on the permeation rate of proteins. As
the molecular weight of the permeating protein increases
the screening effect also increases, resulting in the de-
creased permeation of large molecules through the
PEG-CA membrane. Thus, light can be used to alter the
crosslinking structure of the PEG-CA membranes, there-
by controlling release of biomolecules.
 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352 1349

Table 2
The diffusion coefficients of Mb, Hb, and LDH-L are shown as a function of light exposure, and degree of functionality. The thickness of the gels was
measured using an electron micrometer and an average between different experiments was taken. Each membrane was measured at three different
places to assess its uniformity and an average was taken

Proteins Thickness of Time of light Average Diffusion PEG-CA

gel membrane exposure mesh size coefficients modification
(mm$SD) (h'300 nm) (As ) (cm2 min~1)

Myoglobin 1.62$0.01 2 242 7.0e-5

(MW"17600) 1.58$0.02 3 200 6.20e-5 85%
1.52$0.08 12 154 3.70e-5
Hemoglobin 1.57$0.01 6 181 8.0e-6 85%
(MW"64 500) 1.52$0.08 12 154 2.76e-6 85%
1.50$0.01 21 171 1.05e-5 68%
1.50$0.01 21 142 9.70e-6 68%
1.30$0.02 21#1 h of 254 nm 155 1.12e-5 68%
irradiation
Lactate 1.59$0.05 12 165 2.25e-6 68%
dehydrogenase-L 1.50$0.01 12#45 min of 254 nm 216 4.62e-6 68%
(MW"140 000) irradiation

Fig. 7. The diffusion coefficient of Mb through PEG-CA membrane as

Fig. 6. The permeation of myoglobin (Mb) through hydrogel mem-
branes irradiated at different time intervals ('300 nm). The degree of
modification of the PEG-CA monomers was 85% and the concentra-
tion of the aqueous PEG-CA solution prior to gelation was 15.4 w/v%.
A side-by-side diffusion apparatus described in the text was used for the
permeation experiments.
As described in a previous paper [1] an alternative
method to vary mesh size is to alter the functionality of
the monomer. One would expect decreasing initial func-
tionality to increase the mesh size, and mirror the effect of
reducing irradiation time ('300 nm). Table 2 clearly
a function of mesh size. PEG-CA membranes were prepared by irra-
diating ('300 nm) aqueous PEG-CA solutions (15.4 w/v%, 85% modi-
fication) for 2, 3, and 12 h, respectively. A 450 W Hg lamp was used as
the light source.
shows that changing the initial functionality from 85% to
68% increases the diffusion of Hb through the gel mem-
branes. There is a 73% increase in the value of the
diffusion coefficient of Hb through 12 h membranes pre-
pared with a 17% change in functionality of their initial
PEG-CA monomers. This 17% change in functionality
accomplishes the same effect in the permeation rate as an
approximately 7.5 h difference in irradiation time.
1350 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352

3.4. Photoscission lobutane crosslinks degrade to their monomeric form,

In a previous paper we demonstrated that photoscis-
sion of PEG-CA hydrogels occurs upon exposure to
irradiation at 254 nm [1]. During photoscission the cyc-
Fig. 8. The permeation of hemoglobin (Hb) through PEG-CA hydrogel
membranes prepared under 6 and 12 h UV exposure ('300 nm). The
monomer concentration was 15.4 w/v% and the degree of modification
was 85%.
resulting in the scission of the hydrogel networks. Figure 9
demonstrates the potential of this chemistry. Here, ultra-
violet irradiation was used to rationally alter the mesh
size of PEG-CA hydrogels, and consequently control the
release of fluorescein-isothiocynate dextran (MW"
140 000) through the membrane’s network. The gel
matrix (Fig. 9a) was almost completely dissociated upon
irradiation (254 nm) for 40 min, allowing the entrapped
dextran to be released in the water solution (Fig. 9b). The
permeation of Hb and LDH-L through PEG-CA mem-
branes which had been irradiated with 254 nm irradia-
tion were also investigated. The effect of photoscission on
the permeation rate is given in Figs. 10 and 11. In Fig. 10,
the permeation rate of hemoglobin through a PEG-CA
membrane (functionality"68%, 21 h of'300 nm ir-
radiation) was increased by 18% when exposed for 1 h to
254 nm light. The effect of UV irradiation was more
striking on the permeation of lactate dehydrogenase
through a hydrogel membrane (Fig. 11). A 48% increase
in the permeation rate of LDH was observed after 45 min
of irradiation at 254 nm. Once again, the result of the
experiment is predictable. Since the photoscission light
changes the mesh size from 165 A_ for the 12 h membrane
to 216 A_ (30% increase) for the same membrane irra-
diated for 45 min with a 254 nm light, one would predict
an increase in the permeation rate and a more significant
effect on the larger proteins.
We have shown that there is a direct relationship
between mesh size, size of permeating protein, and light
exposure. As the size of the proteins is increased the

Fig. 9. The release of fluorescein—isothiocyanate dextran (MW"145 000) from a PEG-CA hydrogel (degree of modification"70%) in water (a), upon
irradiation with a 254 nm light for 40 min (b). 0.02 g of dextran-FTIC were dissolved in aqueous PEG-CA solution (25 w/v%) and the sample
irradiated with light ('300 nm) at the side of a quartz vial for 2 h. Following gelation the gel let equilibrated in water for 1 day.
 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
Fig. 10. The permeation of Hb through PEG-CA hydrogel membranes
(68% modification). Both hydrogel membranes were prepared upon
irradiation ('300 nm) of aqueous PEG-CA solutions (15.4 w/v%) for
21 h. One of these membranes was then irradiated with 254 nm light for
1 h at room temperature.
crosslinked structure of the hydrogel exhibits a screening
effect on the protein’s permeation rate, a relationship
which is correlated to the calculated mesh size. We stress
that the determination of the mesh size of PEG-CA
hydrogels using swelling experiments depends strongly
on the polymer—solvent interaction parameter of the sys-
tem. Since it would be impossible to measure s at every
molecular weight and functionality, we assumed a s value
for PEG in water as given by Merrill [20, 21]. In using
this s value one can expect some deviation from the
actual mesh size, but, we believe that since the small
cinnamylidene acetate groups do not change significantly
the phase behavior of the PEG chains in water, the
relative changes in the mesh size as a function of light
exposure, wavelength, and initial monomer concentra-
tion should not be changed by a small alteration in s.
Currently, we are evaluating the use of light scattering
spectroscopy to determine s for PEG-CA in water.
In summary, the diffusion coefficients given in Table 2
depend on the size of the permeating protein, the degree
of modification of the PEG-CA monomers, irradiation
time, and irradiation wavelength. The combined effect of
light intensity, degree of modification, and type of ir-
radiation can be thus used to control the selective exclu-
sion of different size analytes in the PEG-CA hydrogel
membranes.
1351
Fig. 11. The permeation of lactate dehydrogenase-L through PEG-CA
hydrogel membranes (68% modification). Both hydrogel membranes
were prepared upon irradiation ('300 nm) of PEG-CA aliquots
(15.4 w/v%) for 12 h. One of the membrane was then irradiated at
254 nm for 45 min.
4. Conclusions
We have developed a new generation of hydrogel
membranes based on photosensitive PEG monomers.
These hydrogel membranes exhibit a unique photoscis-
sive behavior upon exposure to UV light at 254 nm.
Light exposure, initial concentration of PEG monomers,
and light wavelength were used as triggers to control the
mesh size of the membranes and consequently, the per-
meation rates of various proteins. Hydrogel’s mesh size
was increased with increasing UV exposure at 254 nm
and decreasing initial PEG-CA concentration. The diffu-
sion coefficients of myoglobin, hemoglobin, and lactate
dehydrogenase-L in PEG-CA membranes, were deter-
mined and showed a decreasing trend upon increasing
the crosslinking irradiation, increasing the degree of
modification of PEG-CA monomers, and increasing sol-
ute molecular size.
We have shown that PEG-based hydrogel membranes
can be synthesized using light to initiate gelation in the
absence of initiators. UV light can be used to control the
permeative characteristics of the hydrogel membranes
and selectively separate high molecular weight solutes.
PEG-CA hydrogels appear to be promising materials for
the selective release of drugs and other pharmacological
agents in controlled delivery devices.
1352 F.M. Andreopoulos et al. / Biomaterials 19 (1998) 1343—1352
Acknowledgements
We greatly appreciate all the valuable thoughts of
Dr. Nicholas Peppas and Tony Lowman at the Univer-
sity of Purdue in completing this paper. We would also
like to thank Glenn Critchley at Micromass Inc., Man-
chester, UK, Billie J. Kline and Rebecca Rodney at the
University of Pittsburgh for their help in performing the
MALDI-TOF experiments. The work was funded by the
National Science Foundation (NSF-AMPP-BES
9408438).
References
[1] Gehrke SH. Mass transfer in pH-sensitive hydrogels. Chem Eng
Sci 1989;44:559—66.
[2] Bell CL, Peppas NA. Water, solute and protein diffusion in
physiologically responsive hydrogels of poly(methacrylic acid-g-
ethylene glycol). Biomaterials 1996;17:1203—18.
[3] Gudeman LF, Peppas NA. pH-sensitive membranes from
poly(vinyl alcohol)/poly(acrylic acid) interpenetrating networks.
J Membrane Sci 1995;107:239—48.
[4] Peppas NA, Khare AR. Preparation, structure and diffusional
behavior of hydrogels in controlled release. Adv Drug Delivery
Rev 1993;11:1—35.
[5] Brazel CS, Peppas NA. Temperature and pH-sensitive hydrogels
for controlled release of antithrombogenic agents. Mat Res Soc
Symp Proc 1994;331:211—16.
[6] Hoffman Alan S. ‘Intelligent’ polymers in medicine and biotech-
nology. Artificial Organs 1995;19:458—67.
[7] Peppas NA, Hariharan D, Khare AR. Bioresponsive hydrogels
for controlled release of solutes. Polym Mater Sci Eng
1992;66:83—4.
[8] Galaev IY, Mattiasson B. Thermoreactive water-soluble poly-
mers, nonionic surfactants, and hydrogels as reagents in biotech-
nology. Enzyme Microb Technol 1993;15:354—66.
[9] Park TG, Hoffman AS. Thermal cycling effects on the bioreactor
performances of immobilized b-galactosidase in temperature-sen-
sitive hydrogel beads. Enzyme Microb Technol 1993;15:476—82.
[10] Brannon-Peppas L, Peppas NA. Equilibrium swelling behavior of
pH-sensitive hydrogels. Chem Eng Sci 1991;46:715—22.
[11] Langer R, Peppas N. Chemical and physical structure of polymers
as carriers for controlled release of bioactive agents: a review.
JMS—Rev Macromol Chem Phys 1983;c23:61—126.
[12] Langer R et al. Control of release kinetics of macromolecules from
polyesters. J Membrane Sci 1980;7:333—50.
[13] Peppas NA, Merrill EW. Crosslinked poly(vinyl alcohol) hydro-
gels as swollen elastic networks. J Appl Poly Sci 1977;21:1763—70.
[14] Peppas NA et al. The structure of highly crosslinked poly(2-
hydroxyethyl methacrylate) hydrogels. J Biomed Mater Res 1985;
19:397—411.
[15] Andreopoulos, FM et al. Photoscissable hydrogel synthesis via
rapid photopolymerization of novel PEG-based polymers in the
absence of photoinitiators. J Am Chem Soc 1996;118:6235—40.
[16] Reinhart CT, Peppas NA. Solute diffusion in swollen membranes.
Part II. Influence of crosslinking on diffusive properties. J Mem-
brane Sci 1984;18:227—39.
[17] Colton CK et al. Permeability studies with cellulosic membranes.
J Biomed Mater Res 1971;5:459—88.
[18] Flory PJ. Principles of polymer chemistry. Ithaka: Cornell Uni-
versity Press, 1953:476—82.
[19] Bray JC, Merrill EW. Poly(vinyl alcohol) hydrogels. Formation
by electron beam irradiation of aqueous solutions and subsequent
crystallization. J Polym Sci 1973;17:3779—94.
[20] Merrill EW et al. Partitioning and diffusion of solutes in hydro-
gels of poly(ethylene oxide). Biomaterials 1993;14:1117—26.
[21] Kofinas P et al. Hydrogels prepared by electron irradiation of
poly(ethylene oxide) in water solution: unexpected dependence of
cross-link density and protein diffusion coefficients on initial PEO
molecular weight. Biomaterials 1996;17:1547—50.